From d72b4d3918f6a51c1da7580a932983e66acc2dff Mon Sep 17 00:00:00 2001 From: Konstantin Belousov Date: Mon, 9 Jan 2017 20:14:18 +0000 Subject: [PATCH 01/79] Forcibly remove the cached items from pseudofs vncache on module unload. If some process' nodes were accessed using procfs and the process cannot exit properly at the time modunload event is reported to the pseudofs-backed filesystem, the assertion in pfs_vncache_unload() is triggered. Assertion is correct, the cache should be cleaned. Approved by: des (pseudofs maintainer) Reported and tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 1 week --- sys/fs/pseudofs/pseudofs_vncache.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) diff --git a/sys/fs/pseudofs/pseudofs_vncache.c b/sys/fs/pseudofs/pseudofs_vncache.c index 1bec5a4c885c..093d80565eaa 100644 --- a/sys/fs/pseudofs/pseudofs_vncache.c +++ b/sys/fs/pseudofs/pseudofs_vncache.c @@ -51,6 +51,7 @@ static struct mtx pfs_vncache_mutex; static struct pfs_vdata *pfs_vncache; static eventhandler_tag pfs_exit_tag; static void pfs_exit(void *arg, struct proc *p); +static void pfs_purge_locked(struct pfs_node *pn, bool force); static SYSCTL_NODE(_vfs_pfs, OID_AUTO, vncache, CTLFLAG_RW, 0, "pseudofs vnode cache"); @@ -97,6 +98,9 @@ pfs_vncache_unload(void) { EVENTHANDLER_DEREGISTER(process_exit, pfs_exit_tag); + mtx_lock(&pfs_vncache_mutex); + pfs_purge_locked(NULL, true); + mtx_unlock(&pfs_vncache_mutex); KASSERT(pfs_vncache_entries == 0, ("%d vncache entries remaining", pfs_vncache_entries)); mtx_destroy(&pfs_vncache_mutex); @@ -272,7 +276,7 @@ pfs_vncache_free(struct vnode *vp) * used to implement the cache. */ static void -pfs_purge_locked(struct pfs_node *pn) +pfs_purge_locked(struct pfs_node *pn, bool force) { struct pfs_vdata *pvd; struct vnode *vnp; @@ -280,7 +284,8 @@ pfs_purge_locked(struct pfs_node *pn) mtx_assert(&pfs_vncache_mutex, MA_OWNED); pvd = pfs_vncache; while (pvd != NULL) { - if (pvd->pvd_dead || (pn != NULL && pvd->pvd_pn == pn)) { + if (force || pvd->pvd_dead || + (pn != NULL && pvd->pvd_pn == pn)) { vnp = pvd->pvd_vnode; vhold(vnp); mtx_unlock(&pfs_vncache_mutex); @@ -301,7 +306,7 @@ pfs_purge(struct pfs_node *pn) { mtx_lock(&pfs_vncache_mutex); - pfs_purge_locked(pn); + pfs_purge_locked(pn, false); mtx_unlock(&pfs_vncache_mutex); } @@ -321,6 +326,6 @@ pfs_exit(void *arg, struct proc *p) if (pvd->pvd_pid == p->p_pid) dead = pvd->pvd_dead = 1; if (dead) - pfs_purge_locked(NULL); + pfs_purge_locked(NULL, false); mtx_unlock(&pfs_vncache_mutex); } From 0ac43d9728a7916453a4d8f1cf09e5f47c7038fd Mon Sep 17 00:00:00 2001 From: Marius Strobl Date: Mon, 9 Jan 2017 20:51:51 +0000 Subject: [PATCH 02/79] In dummynet(4), random chunks of memory are casted to struct dn_*, potentially leading to fatal unaligned accesses on architectures with strict alignment requirements. This change fixes dummynet(4) as far as accesses to 64-bit members of struct dn_* are concerned, tripping up on sparc64 with accesses to 32-bit members happening to be correctly aligned there. In other words, this only fixes the tip of the iceberg; larger parts of dummynet(4) still need to be rewritten in order to properly work on all of !x86. In principle, considering the amount of code in dummynet(4) that needs this erroneous pattern corrected, an acceptable workaround would be to declare all struct dn_* packed, forcing compilers to do byte-accesses as a side-effect. However, given that the structs in question aren't laid out well either, this would break ABI/KBI. While at it, replace all existing bcopy(9) calls with memcpy(9) for performance reasons, as there is no need to check for overlap in these cases. PR: 189219 MFC after: 5 days --- sys/netpfil/ipfw/ip_dummynet.c | 178 +++++++++++++++++++++------------ 1 file changed, 114 insertions(+), 64 deletions(-) diff --git a/sys/netpfil/ipfw/ip_dummynet.c b/sys/netpfil/ipfw/ip_dummynet.c index f6d9c28237f5..7240a99e5c44 100644 --- a/sys/netpfil/ipfw/ip_dummynet.c +++ b/sys/netpfil/ipfw/ip_dummynet.c @@ -931,29 +931,35 @@ delete_schk(int i) static int copy_obj(char **start, char *end, void *_o, const char *msg, int i) { - struct dn_id *o = _o; + struct dn_id o; + union { + struct dn_link l; + struct dn_schk s; + } dn; int have = end - *start; - if (have < o->len || o->len == 0 || o->type == 0) { + memcpy(&o, _o, sizeof(o)); + if (have < o.len || o.len == 0 || o.type == 0) { D("(WARN) type %d %s %d have %d need %d", - o->type, msg, i, have, o->len); + o.type, msg, i, have, o.len); return 1; } - ND("type %d %s %d len %d", o->type, msg, i, o->len); - bcopy(_o, *start, o->len); - if (o->type == DN_LINK) { + ND("type %d %s %d len %d", o.type, msg, i, o.len); + if (o.type == DN_LINK) { + memcpy(&dn.l, _o, sizeof(dn.l)); /* Adjust burst parameter for link */ - struct dn_link *l = (struct dn_link *)*start; - l->burst = div64(l->burst, 8 * hz); - l->delay = l->delay * 1000 / hz; - } else if (o->type == DN_SCH) { - /* Set id->id to the number of instances */ - struct dn_schk *s = _o; - struct dn_id *id = (struct dn_id *)(*start); - id->id = (s->sch.flags & DN_HAVE_MASK) ? - dn_ht_entries(s->siht) : (s->siht ? 1 : 0); - } - *start += o->len; + dn.l.burst = div64(dn.l.burst, 8 * hz); + dn.l.delay = dn.l.delay * 1000 / hz; + memcpy(*start, &dn.l, sizeof(dn.l)); + } else if (o.type == DN_SCH) { + /* Set dn.s.sch.oid.id to the number of instances */ + memcpy(&dn.s, _o, sizeof(dn.s)); + dn.s.sch.oid.id = (dn.s.sch.flags & DN_HAVE_MASK) ? + dn_ht_entries(dn.s.siht) : (dn.s.siht ? 1 : 0); + memcpy(*start, &dn.s, sizeof(dn.s)); + } else + memcpy(*start, _o, o.len); + *start += o.len; return 0; } @@ -974,7 +980,7 @@ copy_obj_q(char **start, char *end, void *_o, const char *msg, int i) return 1; } ND("type %d %s %d len %d", o->type, msg, i, len); - bcopy(_o, *start, len); + memcpy(*start, _o, len); ((struct dn_id*)(*start))->len = len; *start += len; return 0; @@ -1022,7 +1028,7 @@ copy_profile(struct copy_args *a, struct dn_profile *p) D("error have %d need %d", have, profile_len); return 1; } - bcopy(p, *a->start, profile_len); + memcpy(*a->start, p, profile_len); ((struct dn_id *)(*a->start))->len = profile_len; *a->start += profile_len; return 0; @@ -1584,6 +1590,9 @@ config_fs(struct dn_fs *nfs, struct dn_id *arg, int locked) { int i; struct dn_fsk *fs; +#ifdef NEW_AQM + struct dn_extra_parms *ep; +#endif if (nfs->oid.len != sizeof(*nfs)) { D("invalid flowset len %d", nfs->oid.len); @@ -1592,6 +1601,15 @@ config_fs(struct dn_fs *nfs, struct dn_id *arg, int locked) i = nfs->fs_nr; if (i <= 0 || i >= 3*DN_MAX_ID) return NULL; +#ifdef NEW_AQM + ep = NULL; + if (arg != NULL) { + ep = malloc(sizeof(*ep), M_TEMP, locked ? M_NOWAIT : M_WAITOK); + if (ep == NULL) + return (NULL); + memcpy(ep, arg, sizeof(*ep)); + } +#endif ND("flowset %d", i); /* XXX other sanity checks */ if (nfs->flags & DN_QSIZE_BYTES) { @@ -1630,12 +1648,15 @@ config_fs(struct dn_fs *nfs, struct dn_id *arg, int locked) if (bcmp(&fs->fs, nfs, sizeof(*nfs)) == 0) { ND("flowset %d unchanged", i); #ifdef NEW_AQM - /* reconfigure AQM as the parameters can be changed. - * we consider the flowsetis busy if it has scheduler instance(s) - */ - s = locate_scheduler(nfs->sched_nr); - config_aqm(fs, (struct dn_extra_parms *) arg, - s != NULL && s->siht != NULL); + if (ep != NULL) { + /* + * Reconfigure AQM as the parameters can be changed. + * We consider the flowset as busy if it has scheduler + * instance(s). + */ + s = locate_scheduler(nfs->sched_nr); + config_aqm(fs, ep, s != NULL && s->siht != NULL); + } #endif break; /* no change, nothing to do */ } @@ -1657,13 +1678,19 @@ config_fs(struct dn_fs *nfs, struct dn_id *arg, int locked) fs->fs = *nfs; /* copy configuration */ #ifdef NEW_AQM fs->aqmfp = NULL; - config_aqm(fs, (struct dn_extra_parms *) arg, s != NULL && s->siht != NULL); + if (ep != NULL) + config_aqm(fs, ep, s != NULL && + s->siht != NULL); #endif if (s != NULL) fsk_attach(fs, s); } while (0); if (!locked) DN_BH_WUNLOCK(); +#ifdef NEW_AQM + if (ep != NULL) + free(ep, M_TEMP); +#endif return fs; } @@ -1773,7 +1800,7 @@ config_sched(struct dn_sch *_nsch, struct dn_id *arg) D("cannot allocate profile"); goto error; //XXX } - bcopy(pf, s->profile, sizeof(*pf)); + memcpy(s->profile, pf, sizeof(*pf)); } } p.link_nr = 0; @@ -1795,7 +1822,7 @@ config_sched(struct dn_sch *_nsch, struct dn_id *arg) pf = malloc(sizeof(*pf), M_DUMMYNET, M_NOWAIT | M_ZERO); if (pf) /* XXX should issue a warning otherwise */ - bcopy(s->profile, pf, sizeof(*pf)); + memcpy(pf, s->profile, sizeof(*pf)); } /* remove from the hash */ dn_ht_find(dn_cfg.schedhash, i, DNHT_REMOVE, NULL); @@ -1917,7 +1944,7 @@ config_profile(struct dn_profile *pf, struct dn_id *arg) olen = s->profile->oid.len; if (olen < pf->oid.len) olen = pf->oid.len; - bcopy(pf, s->profile, pf->oid.len); + memcpy(s->profile, pf, pf->oid.len); s->profile->oid.len = olen; } DN_BH_WUNLOCK(); @@ -1953,30 +1980,35 @@ dummynet_flush(void) int do_config(void *p, int l) { - struct dn_id *next, *o; - int err = 0, err2 = 0; - struct dn_id *arg = NULL; - uintptr_t *a; + struct dn_id o; + union { + struct dn_profile profile; + struct dn_fs fs; + struct dn_link link; + struct dn_sch sched; + } *dn; + struct dn_id *arg; + uintptr_t a; + int err, err2, off; - o = p; - if (o->id != DN_API_VERSION) { - D("invalid api version got %d need %d", - o->id, DN_API_VERSION); + memcpy(&o, p, sizeof(o)); + if (o.id != DN_API_VERSION) { + D("invalid api version got %d need %d", o.id, DN_API_VERSION); return EINVAL; } - for (; l >= sizeof(*o); o = next) { - struct dn_id *prev = arg; - if (o->len < sizeof(*o) || l < o->len) { - D("bad len o->len %d len %d", o->len, l); + arg = NULL; + dn = NULL; + for (off = 0; l >= sizeof(o); memcpy(&o, (char *)p + off, sizeof(o))) { + if (o.len < sizeof(o) || l < o.len) { + D("bad len o.len %d len %d", o.len, l); err = EINVAL; break; } - l -= o->len; - next = (struct dn_id *)((char *)o + o->len); + l -= o.len; err = 0; - switch (o->type) { + switch (o.type) { default: - D("cmd %d not implemented", o->type); + D("cmd %d not implemented", o.type); break; #ifdef EMULATE_SYSCTL @@ -1994,31 +2026,30 @@ do_config(void *p, int l) case DN_CMD_DELETE: /* the argument is in the first uintptr_t after o */ - a = (uintptr_t *)(o+1); - if (o->len < sizeof(*o) + sizeof(*a)) { + if (o.len < sizeof(o) + sizeof(a)) { err = EINVAL; break; } - switch (o->subtype) { + memcpy(&a, (char *)p + off + sizeof(o), sizeof(a)); + switch (o.subtype) { case DN_LINK: /* delete base and derived schedulers */ DN_BH_WLOCK(); - err = delete_schk(*a); - err2 = delete_schk(*a + DN_MAX_ID); + err = delete_schk(a); + err2 = delete_schk(a + DN_MAX_ID); DN_BH_WUNLOCK(); if (!err) err = err2; break; default: - D("invalid delete type %d", - o->subtype); + D("invalid delete type %d", o.subtype); err = EINVAL; break; case DN_FS: - err = (*a <1 || *a >= DN_MAX_ID) ? - EINVAL : delete_fs(*a, 0) ; + err = (a < 1 || a >= DN_MAX_ID) ? + EINVAL : delete_fs(a, 0) ; break; } break; @@ -2028,28 +2059,47 @@ do_config(void *p, int l) dummynet_flush(); DN_BH_WUNLOCK(); break; - case DN_TEXT: /* store argument the next block */ - prev = NULL; - arg = o; + case DN_TEXT: /* store argument of next block */ + if (arg != NULL) + free(arg, M_TEMP); + arg = malloc(o.len, M_TEMP, M_WAITOK); + memcpy(arg, (char *)p + off, o.len); break; case DN_LINK: - err = config_link((struct dn_link *)o, arg); + if (dn == NULL) + dn = malloc(sizeof(*dn), M_TEMP, M_WAITOK); + memcpy(&dn->link, (char *)p + off, sizeof(dn->link)); + err = config_link(&dn->link, arg); break; case DN_PROFILE: - err = config_profile((struct dn_profile *)o, arg); + if (dn == NULL) + dn = malloc(sizeof(*dn), M_TEMP, M_WAITOK); + memcpy(&dn->profile, (char *)p + off, + sizeof(dn->profile)); + err = config_profile(&dn->profile, arg); break; case DN_SCH: - err = config_sched((struct dn_sch *)o, arg); + if (dn == NULL) + dn = malloc(sizeof(*dn), M_TEMP, M_WAITOK); + memcpy(&dn->sched, (char *)p + off, + sizeof(dn->sched)); + err = config_sched(&dn->sched, arg); break; case DN_FS: - err = (NULL==config_fs((struct dn_fs *)o, arg, 0)); + if (dn == NULL) + dn = malloc(sizeof(*dn), M_TEMP, M_WAITOK); + memcpy(&dn->fs, (char *)p + off, sizeof(dn->fs)); + err = (NULL == config_fs(&dn->fs, arg, 0)); break; } - if (prev) - arg = NULL; if (err != 0) break; + off += o.len; } + if (arg != NULL) + free(arg, M_TEMP); + if (dn != NULL) + free(dn, M_TEMP); return err; } @@ -2261,7 +2311,7 @@ dummynet_get(struct sockopt *sopt, void **compat) a.type = cmd->subtype; if (compat == NULL) { - bcopy(cmd, start, sizeof(*cmd)); + memcpy(start, cmd, sizeof(*cmd)); ((struct dn_id*)(start))->len = sizeof(struct dn_id); buf = start + sizeof(*cmd); } else From 129adb62dc426aec3d46c177882ddd1c27b41fa8 Mon Sep 17 00:00:00 2001 From: Adrian Chadd Date: Mon, 9 Jan 2017 21:46:24 +0000 Subject: [PATCH 03/79] [rsu] add support for the "green" rsu NICs. They're still a 1T2R NIC, so reuse the same rfconfig and nstream configuration. Submitted by: Idwer Vollering --- sys/dev/usb/wlan/if_rsu.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/sys/dev/usb/wlan/if_rsu.c b/sys/dev/usb/wlan/if_rsu.c index 9fe82eea8cab..abab5b5f6004 100644 --- a/sys/dev/usb/wlan/if_rsu.c +++ b/sys/dev/usb/wlan/if_rsu.c @@ -523,6 +523,12 @@ rsu_attach(device_t self) sc->sc_ntxstream = 2; rft = "2T2R"; break; + case 0x3: /* "green" NIC */ + sc->sc_rftype = RTL8712_RFCONFIG_1T2R; + sc->sc_nrxstream = 2; + sc->sc_ntxstream = 1; + rft = "1T2R ('green')"; + break; default: device_printf(sc->sc_dev, "%s: unknown board type (rfconfig=0x%02x)\n", From f50c49cca7362d5fd8cb57ec6bc7e843f301e300 Mon Sep 17 00:00:00 2001 From: Navdeep Parhar Date: Mon, 9 Jan 2017 22:18:08 +0000 Subject: [PATCH 04/79] cxgbe(4): The wraparound logic in start_wrq_wr() should not get involved in work requests that end at the end of the descriptor ring, even though the pidx wraps around to 0. MFC after: 3 days --- sys/dev/cxgbe/t4_sge.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sys/dev/cxgbe/t4_sge.c b/sys/dev/cxgbe/t4_sge.c index bdfaa4db2b00..31ec87f4c1dd 100644 --- a/sys/dev/cxgbe/t4_sge.c +++ b/sys/dev/cxgbe/t4_sge.c @@ -2298,7 +2298,7 @@ start_wrq_wr(struct sge_wrq *wrq, int len16, struct wrq_cookie *cookie) w = &eq->desc[eq->pidx]; IDXINCR(eq->pidx, ndesc, eq->sidx); - if (__predict_false(eq->pidx < ndesc - 1)) { + if (__predict_false(cookie->pidx + ndesc > eq->sidx)) { w = &wrq->ss[0]; wrq->ss_pidx = cookie->pidx; wrq->ss_len = len16 * 16; From 87b027ba698be7a143e49aad4ac76f4958bd7d8a Mon Sep 17 00:00:00 2001 From: Navdeep Parhar Date: Mon, 9 Jan 2017 22:20:09 +0000 Subject: [PATCH 05/79] cxgbe(4): Enable automatic cidx flush for all control queues. MFC after: 3 days --- sys/dev/cxgbe/t4_sge.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/sys/dev/cxgbe/t4_sge.c b/sys/dev/cxgbe/t4_sge.c index 31ec87f4c1dd..aa0a8a3b804e 100644 --- a/sys/dev/cxgbe/t4_sge.c +++ b/sys/dev/cxgbe/t4_sge.c @@ -3305,12 +3305,13 @@ ctrl_eq_alloc(struct adapter *sc, struct sge_eq *eq) c.cmpliqid_eqid = htonl(V_FW_EQ_CTRL_CMD_CMPLIQID(eq->iqid)); c.physeqid_pkd = htobe32(0); c.fetchszm_to_iqid = - htobe32(V_FW_EQ_CTRL_CMD_HOSTFCMODE(X_HOSTFCMODE_NONE) | + htobe32(V_FW_EQ_CTRL_CMD_HOSTFCMODE(X_HOSTFCMODE_STATUS_PAGE) | V_FW_EQ_CTRL_CMD_PCIECHN(eq->tx_chan) | F_FW_EQ_CTRL_CMD_FETCHRO | V_FW_EQ_CTRL_CMD_IQID(eq->iqid)); c.dcaen_to_eqsize = htobe32(V_FW_EQ_CTRL_CMD_FBMIN(X_FETCHBURSTMIN_64B) | V_FW_EQ_CTRL_CMD_FBMAX(X_FETCHBURSTMAX_512B) | + V_FW_EQ_CTRL_CMD_CIDXFTHRESH(X_CIDXFLUSHTHRESH_32) | V_FW_EQ_CTRL_CMD_EQSIZE(qsize)); c.eqaddr = htobe64(eq->ba); From 5b51fcfc7ab714d7313b6649fd3764052084dc91 Mon Sep 17 00:00:00 2001 From: Sean Bruno Date: Mon, 9 Jan 2017 23:41:10 +0000 Subject: [PATCH 06/79] Remove unused mtx_held() macro. --- sys/net/iflib.c | 4 ---- 1 file changed, 4 deletions(-) diff --git a/sys/net/iflib.c b/sys/net/iflib.c index beeddcf6fabb..f1aee37ecd78 100644 --- a/sys/net/iflib.c +++ b/sys/net/iflib.c @@ -447,10 +447,6 @@ struct iflib_rxq { static int enable_msix = 1; -#define mtx_held(m) (((m)->mtx_lock & ~MTX_FLAGMASK) != (uintptr_t)0) - - - #define CTX_ACTIVE(ctx) ((if_getdrvflags((ctx)->ifc_ifp) & IFF_DRV_RUNNING)) #define CTX_LOCK_INIT(_sc, _name) mtx_init(&(_sc)->ifc_mtx, _name, "iflib ctx lock", MTX_DEF) From 5c06728ccb4d0a99dcc5e183780a6b9a7b333a7e Mon Sep 17 00:00:00 2001 From: Andriy Voskoboinyk Date: Mon, 9 Jan 2017 23:42:02 +0000 Subject: [PATCH 07/79] rtwn_usb(4): do not try to modify global static structure. Use a local copy for modifications instead. Tested with RTL8821AU (AP) + RTL8188EU (STA). Reported by: hselasky --- sys/dev/rtwn/usb/rtwn_usb_ep.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/sys/dev/rtwn/usb/rtwn_usb_ep.c b/sys/dev/rtwn/usb/rtwn_usb_ep.c index 328e5bd36e1a..52c628d5a9a7 100644 --- a/sys/dev/rtwn/usb/rtwn_usb_ep.c +++ b/sys/dev/rtwn/usb/rtwn_usb_ep.c @@ -58,7 +58,7 @@ __FBSDID("$FreeBSD$"); #include -static struct usb_config rtwn_config[RTWN_N_TRANSFER] = { +static const struct usb_config rtwn_config_common[RTWN_N_TRANSFER] = { [RTWN_BULK_RX] = { .type = UE_BULK, .endpoint = UE_ADDR_ANY, @@ -161,6 +161,7 @@ rtwn_usb_setup_queues(struct rtwn_usb_softc *uc) int rtwn_usb_setup_endpoints(struct rtwn_usb_softc *uc) { + struct usb_config *rtwn_config; struct rtwn_softc *sc = &uc->uc_sc; const uint8_t iface_index = RTWN_IFACE_INDEX; struct usb_endpoint *ep, *ep_end; @@ -197,6 +198,9 @@ rtwn_usb_setup_endpoints(struct rtwn_usb_softc *uc) return (EINVAL); } + rtwn_config = malloc(sizeof(rtwn_config_common), M_TEMP, M_WAITOK); + memcpy(rtwn_config, rtwn_config_common, sizeof(rtwn_config_common)); + /* NB: keep in sync with rtwn_dma_init(). */ rtwn_config[RTWN_BULK_TX_VO].endpoint = addr[0]; switch (uc->ntx) { @@ -224,6 +228,8 @@ rtwn_usb_setup_endpoints(struct rtwn_usb_softc *uc) rtwn_config[RTWN_BULK_RX].bufsize = sc->rx_dma_size + 1024; error = usbd_transfer_setup(uc->uc_udev, &iface_index, uc->uc_xfer, rtwn_config, RTWN_N_TRANSFER, uc, &sc->sc_mtx); + free(rtwn_config, M_TEMP); + if (error) { device_printf(sc->sc_dev, "could not allocate USB transfers, " "err=%s\n", usbd_errstr(error)); From 46aa753364634c7ef17d4a66245da18c2ffe0367 Mon Sep 17 00:00:00 2001 From: Sean Bruno Date: Mon, 9 Jan 2017 23:45:40 +0000 Subject: [PATCH 08/79] White space cleanup from an cut-n-paste. Submitted by: mmacy@nextbsd.org --- sys/conf/files | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/sys/conf/files b/sys/conf/files index 3abe344cfa78..ce9042a7b196 100644 --- a/sys/conf/files +++ b/sys/conf/files @@ -3905,9 +3905,9 @@ net/if_tun.c optional tun net/if_tap.c optional tap net/if_vlan.c optional vlan net/if_vxlan.c optional vxlan inet | vxlan inet6 -net/ifdi_if.m optional ether pci -net/iflib.c optional ether pci -net/mp_ring.c optional ether +net/ifdi_if.m optional ether pci +net/iflib.c optional ether pci +net/mp_ring.c optional ether net/mppcc.c optional netgraph_mppc_compression net/mppcd.c optional netgraph_mppc_compression net/netisr.c standard From d5fadb019e0226e1a617d83e9eec320926d65de5 Mon Sep 17 00:00:00 2001 From: Conrad Meyer Date: Mon, 9 Jan 2017 23:51:31 +0000 Subject: [PATCH 09/79] cd9660: Add a prototype for cd9660_vfs_hash_cmp GCC warns (and errors, with -Werror) about it otherwise. Clang doesn't care. Introduced in r311665. Reported by: np@ --- sys/fs/cd9660/cd9660_vfsops.c | 1 + 1 file changed, 1 insertion(+) diff --git a/sys/fs/cd9660/cd9660_vfsops.c b/sys/fs/cd9660/cd9660_vfsops.c index 7473f97dceef..2d97029a6642 100644 --- a/sys/fs/cd9660/cd9660_vfsops.c +++ b/sys/fs/cd9660/cd9660_vfsops.c @@ -88,6 +88,7 @@ static struct vfsops cd9660_vfsops = { VFS_SET(cd9660_vfsops, cd9660, VFCF_READONLY); MODULE_VERSION(cd9660, 1); +static int cd9660_vfs_hash_cmp(struct vnode *vp, cd_ino_t *pino); static int iso_mountfs(struct vnode *devvp, struct mount *mp); /* From f7da1444fd0e0bae1b16e07281f5c668b372b268 Mon Sep 17 00:00:00 2001 From: Conrad Meyer Date: Mon, 9 Jan 2017 23:56:45 +0000 Subject: [PATCH 10/79] cd9660: typedef cd_ino_t in preference to #define Suggested by: kib@ --- sys/fs/cd9660/iso.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sys/fs/cd9660/iso.h b/sys/fs/cd9660/iso.h index 53f005c870f0..0d3e46c9a739 100644 --- a/sys/fs/cd9660/iso.h +++ b/sys/fs/cd9660/iso.h @@ -222,7 +222,7 @@ enum ISO_FTYPE { ISO_FTYPE_DEFAULT, ISO_FTYPE_9660, ISO_FTYPE_RRIP, /* * When ino_t becomes 64-bit, we can remove this definition in favor of ino_t. */ -#define cd_ino_t uint64_t +typedef __uint64_t cd_ino_t; struct iso_mnt { uint64_t im_flags; From cc2166783be0b1cf010b480d98613bc61aea9126 Mon Sep 17 00:00:00 2001 From: Conrad Meyer Date: Tue, 10 Jan 2017 00:03:43 +0000 Subject: [PATCH 11/79] Adjust ACPI_EXPORT_SYMBOL for AcpiGetTableWithSize Suggested by: jbeich@ --- sys/contrib/dev/acpica/components/tables/tbxface.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sys/contrib/dev/acpica/components/tables/tbxface.c b/sys/contrib/dev/acpica/components/tables/tbxface.c index 254958bd4924..07e2fe846e61 100644 --- a/sys/contrib/dev/acpica/components/tables/tbxface.c +++ b/sys/contrib/dev/acpica/components/tables/tbxface.c @@ -386,7 +386,7 @@ AcpiGetTableWithSize ( return (Status); } -ACPI_EXPORT_SYMBOL (AcpiGetTable) +ACPI_EXPORT_SYMBOL (AcpiGetTableWithSize) /******************************************************************************* From 5ca4ae8bb7f474cd6d3b1346393ebfb03cf7dd8a Mon Sep 17 00:00:00 2001 From: Bruce M Simpson Date: Tue, 10 Jan 2017 00:28:01 +0000 Subject: [PATCH 12/79] Add PID for Belkin F5U258 "Windows Easy Transfer Cable", a udbp-like device. --- sys/dev/usb/usbdevs | 1 + 1 file changed, 1 insertion(+) diff --git a/sys/dev/usb/usbdevs b/sys/dev/usb/usbdevs index a625c2bf4b2d..797077e16f80 100644 --- a/sys/dev/usb/usbdevs +++ b/sys/dev/usb/usbdevs @@ -1318,6 +1318,7 @@ product BELKIN RTL8188CU 0x1102 RTL8188CU Wireless Adapter product BELKIN F9L1103 0x1103 F9L1103 Wireless Adapter product BELKIN RTL8192CU 0x2102 RTL8192CU Wireless Adapter product BELKIN F7D2102 0x2103 F7D2102 Wireless Adapter +product BELKIN F5U258 0x258A F5U258 Host to Host cable product BELKIN ZD1211B 0x4050 ZD1211B product BELKIN F5D5055 0x5055 F5D5055 product BELKIN F5D7050 0x7050 F5D7050 Wireless Adapter From ae60d856a31f28c45f814421765e716ec1bd4278 Mon Sep 17 00:00:00 2001 From: Andriy Voskoboinyk Date: Tue, 10 Jan 2017 01:09:39 +0000 Subject: [PATCH 13/79] rtwn_pci(4): fix possible race while accessing 'matched_chip' variable. --- sys/dev/rtwn/pci/rtwn_pci_attach.c | 34 ++++++++++++++++++++---------- 1 file changed, 23 insertions(+), 11 deletions(-) diff --git a/sys/dev/rtwn/pci/rtwn_pci_attach.c b/sys/dev/rtwn/pci/rtwn_pci_attach.c index e021c401ad3c..3af8d176f790 100644 --- a/sys/dev/rtwn/pci/rtwn_pci_attach.c +++ b/sys/dev/rtwn/pci/rtwn_pci_attach.c @@ -94,20 +94,31 @@ static void rtwn_pci_beacon_update_end(struct rtwn_softc *, static void rtwn_pci_attach_methods(struct rtwn_softc *); -static int matched_chip = RTWN_CHIP_MAX_PCI; +static const struct rtwn_pci_ident * +rtwn_pci_probe_sub(device_t dev) +{ + const struct rtwn_pci_ident *ident; + int vendor_id, device_id; + + vendor_id = pci_get_vendor(dev); + device_id = pci_get_device(dev); + + for (ident = rtwn_pci_ident_table; ident->name != NULL; ident++) + if (vendor_id == ident->vendor && device_id == ident->device) + return (ident); + + return (NULL); +} static int rtwn_pci_probe(device_t dev) { const struct rtwn_pci_ident *ident; - for (ident = rtwn_pci_ident_table; ident->name != NULL; ident++) { - if (pci_get_vendor(dev) == ident->vendor && - pci_get_device(dev) == ident->device) { - matched_chip = ident->chip; - device_set_desc(dev, ident->name); - return (BUS_PROBE_DEFAULT); - } + ident = rtwn_pci_probe_sub(dev); + if (ident != NULL) { + device_set_desc(dev, ident->name); + return (BUS_PROBE_DEFAULT); } return (ENXIO); } @@ -591,13 +602,15 @@ rtwn_pci_attach_methods(struct rtwn_softc *sc) static int rtwn_pci_attach(device_t dev) { + const struct rtwn_pci_ident *ident; struct rtwn_pci_softc *pc = device_get_softc(dev); struct rtwn_softc *sc = &pc->pc_sc; struct ieee80211com *ic = &sc->sc_ic; uint32_t lcsr; int cap_off, i, error, rid; - if (matched_chip >= RTWN_CHIP_MAX_PCI) + ident = rtwn_pci_probe_sub(dev); + if (ident == NULL) return (ENXIO); /* @@ -649,8 +662,7 @@ rtwn_pci_attach(device_t dev) mtx_init(&sc->sc_mtx, ic->ic_name, MTX_NETWORK_LOCK, MTX_DEF); rtwn_pci_attach_methods(sc); - /* XXX something similar to USB_GET_DRIVER_INFO() */ - rtwn_pci_attach_private(pc, matched_chip); + rtwn_pci_attach_private(pc, ident->chip); /* Allocate Tx/Rx buffers. */ error = rtwn_pci_alloc_rx_list(sc); From b91f227f60cfd9ad39599ffc994df0ad1baf9e91 Mon Sep 17 00:00:00 2001 From: Navdeep Parhar Date: Tue, 10 Jan 2017 01:30:41 +0000 Subject: [PATCH 14/79] cxgbe(4): Refresh t4_msg.h, mainly for definitions related to the crypto engine. Obtained from: Chelsio Communications MFC after: 2 weeks Sponsored by: Chelsio Communications --- sys/dev/cxgbe/common/t4_msg.h | 595 +++++++++++++++++++++++++++++++++- 1 file changed, 594 insertions(+), 1 deletion(-) diff --git a/sys/dev/cxgbe/common/t4_msg.h b/sys/dev/cxgbe/common/t4_msg.h index 830743b91a8f..f15c3a89a56a 100644 --- a/sys/dev/cxgbe/common/t4_msg.h +++ b/sys/dev/cxgbe/common/t4_msg.h @@ -106,6 +106,7 @@ enum { CPL_RX_FCOE_DIF = 0x4A, CPL_RX_DATA_DIF = 0x4B, CPL_ERR_NOTIFY = 0x4D, + CPL_RX_TLS_CMP = 0x4E, CPL_RDMA_READ_REQ = 0x60, CPL_RX_ISCSI_DIF = 0x60, @@ -113,6 +114,11 @@ enum { CPL_SET_LE_REQ = 0x80, CPL_PASS_OPEN_REQ6 = 0x81, CPL_ACT_OPEN_REQ6 = 0x83, + CPL_TX_TLS_PDU = 0x88, + CPL_TX_TLS_SFO = 0x89, + + CPL_TX_SEC_PDU = 0x8A, + CPL_TX_TLS_ACK = 0x8B, CPL_RDMA_TERMINATE = 0xA2, CPL_RDMA_WRITE = 0xA4, @@ -129,6 +135,7 @@ enum { CPL_TRACE_PKT = 0xB0, CPL_RX2TX_DATA = 0xB1, + CPL_TLS_DATA = 0xB1, CPL_ISCSI_DATA = 0xB2, CPL_FCOE_DATA = 0xB3, @@ -136,6 +143,7 @@ enum { CPL_FW4_PLD = 0xC1, CPL_FW4_ACK = 0xC3, CPL_SRQ_TABLE_RPL = 0xCC, + CPL_RX_PHYS_DSGL = 0xD0, CPL_FW6_MSG = 0xE0, CPL_FW6_PLD = 0xE1, @@ -200,6 +208,7 @@ enum { ULP_MODE_RDMA = 4, ULP_MODE_TCPDDP = 5, ULP_MODE_FCOE = 6, + ULP_MODE_TLS = 8, }; enum { @@ -993,6 +1002,23 @@ struct cpl_abort_req_rss { __u8 status; }; +struct cpl_abort_req_rss6 { + RSS_HDR + union opcode_tid ot; + __u32 srqidx_status; +}; + +#define S_ABORT_RSS_STATUS 0 +#define M_ABORT_RSS_STATUS 0xff +#define V_ABORT_RSS_STATUS(x) ((x) << S_ABORT_RSS_STATUS) +#define G_ABORT_RSS_STATUS(x) (((x) >> S_ABORT_RSS_STATUS) & M_ABORT_RSS_STATUS) + +#define S_ABORT_RSS_SRQIDX 8 +#define M_ABORT_RSS_SRQIDX 0xffffff +#define V_ABORT_RSS_SRQIDX(x) ((x) << S_ABORT_RSS_SRQIDX) +#define G_ABORT_RSS_SRQIDX(x) (((x) >> S_ABORT_RSS_SRQIDX) & M_ABORT_RSS_SRQIDX) + + /* cpl_abort_req status command code in case of T6, * bit[0] specifies whether to send RST (0) to remote peer or suppress it (1) * bit[1] indicates ABORT_REQ was sent after a CLOSE_CON_REQ @@ -1014,6 +1040,12 @@ struct cpl_abort_rpl_rss { __u8 status; }; +struct cpl_abort_rpl_rss6 { + RSS_HDR + union opcode_tid ot; + __u32 srqidx_status; +}; + struct cpl_abort_rpl { WR_HDR; union opcode_tid ot; @@ -2612,6 +2644,7 @@ enum { FW_TYPE_RSSCPL = 4, FW_TYPE_WRERR_RPL = 5, FW_TYPE_PI_ERR = 6, + FW_TYPE_TLS_KEY = 7, }; struct cpl_fw2_pld { @@ -2712,7 +2745,8 @@ enum { ULP_TX_SC_IMM = 0x81, ULP_TX_SC_DSGL = 0x82, ULP_TX_SC_ISGL = 0x83, - ULP_TX_SC_PICTRL = 0x84 + ULP_TX_SC_PICTRL = 0x84, + ULP_TX_SC_MEMRD = 0x86 }; #define S_ULPTX_CMD 24 @@ -2763,6 +2797,12 @@ struct ulptx_idata { #define S_ULPTX_NSGE 0 #define M_ULPTX_NSGE 0xFFFF #define V_ULPTX_NSGE(x) ((x) << S_ULPTX_NSGE) +#define G_ULPTX_NSGE(x) (((x) >> S_ULPTX_NSGE) & M_ULPTX_NSGE) + +struct ulptx_sc_memrd { + __be32 cmd_to_len; + __be32 addr; +}; struct ulp_mem_io { WR_HDR; @@ -2816,6 +2856,21 @@ struct ulp_txpkt { __be32 len; }; +/* ulp_txpkt.cmd_dest fields */ +#define S_ULP_TXPKT_DATAMODIFY 23 +#define M_ULP_TXPKT_DATAMODIFY 0x1 +#define V_ULP_TXPKT_DATAMODIFY(x) ((x) << S_ULP_TXPKT_DATAMODIFY) +#define G_ULP_TXPKT_DATAMODIFY(x) \ + (((x) >> S_ULP_TXPKT_DATAMODIFY) & M_ULP_TXPKT_DATAMODIFY_) +#define F_ULP_TXPKT_DATAMODIFY V_ULP_TXPKT_DATAMODIFY(1U) + +#define S_ULP_TXPKT_CHANNELID 22 +#define M_ULP_TXPKT_CHANNELID 0x1 +#define V_ULP_TXPKT_CHANNELID(x) ((x) << S_ULP_TXPKT_CHANNELID) +#define G_ULP_TXPKT_CHANNELID(x) \ + (((x) >> S_ULP_TXPKT_CHANNELID) & M_ULP_TXPKT_CHANNELID) +#define F_ULP_TXPKT_CHANNELID V_ULP_TXPKT_CHANNELID(1U) + /* ulp_txpkt.cmd_dest fields */ #define S_ULP_TXPKT_DEST 16 #define M_ULP_TXPKT_DEST 0x3 @@ -3044,4 +3099,542 @@ struct cpl_rx_mps_pkt { #define X_CPL_RX_MPS_PKT_TYPE_QFC (1 << 2) #define X_CPL_RX_MPS_PKT_TYPE_PTP (1 << 3) +struct cpl_tx_tls_sfo { + __be32 op_to_seg_len; + __be32 pld_len; + __be64 rsvd; + __be32 seqno_numivs; + __be32 ivgen_hdrlen; + __be64 scmd1; +}; + +/* cpl_tx_tls_sfo macros */ +#define S_CPL_TX_TLS_SFO_OPCODE 24 +#define M_CPL_TX_TLS_SFO_OPCODE 0xff +#define V_CPL_TX_TLS_SFO_OPCODE(x) ((x) << S_CPL_TX_TLS_SFO_OPCODE) +#define G_CPL_TX_TLS_SFO_OPCODE(x) \ + (((x) >> S_CPL_TX_TLS_SFO_OPCODE) & M_CPL_TX_TLS_SFO_OPCODE) + +#define S_CPL_TX_TLS_SFO_DATA_TYPE 20 +#define M_CPL_TX_TLS_SFO_DATA_TYPE 0xf +#define V_CPL_TX_TLS_SFO_DATA_TYPE(x) ((x) << S_CPL_TX_TLS_SFO_DATA_TYPE) +#define G_CPL_TX_TLS_SFO_DATA_TYPE(x) \ + (((x) >> S_CPL_TX_TLS_SFO_DATA_TYPE) & M_CPL_TX_TLS_SFO_DATA_TYPE) + +#define S_CPL_TX_TLS_SFO_CPL_LEN 16 +#define M_CPL_TX_TLS_SFO_CPL_LEN 0xf +#define V_CPL_TX_TLS_SFO_CPL_LEN(x) ((x) << S_CPL_TX_TLS_SFO_CPL_LEN) +#define G_CPL_TX_TLS_SFO_CPL_LEN(x) \ + (((x) >> S_CPL_TX_TLS_SFO_CPL_LEN) & M_CPL_TX_TLS_SFO_CPL_LEN) +#define S_CPL_TX_TLS_SFO_SEG_LEN 0 +#define M_CPL_TX_TLS_SFO_SEG_LEN 0xffff +#define V_CPL_TX_TLS_SFO_SEG_LEN(x) ((x) << S_CPL_TX_TLS_SFO_SEG_LEN) +#define G_CPL_TX_TLS_SFO_SEG_LEN(x) \ + (((x) >> S_CPL_TX_TLS_SFO_SEG_LEN) & M_CPL_TX_TLS_SFO_SEG_LEN) + +struct cpl_tls_data { + RSS_HDR + __be32 op_tid; + __be32 length_pkd; + __be32 seq; + __be32 r1; +}; + +#define S_CPL_TLS_DATA_OPCODE 24 +#define M_CPL_TLS_DATA_OPCODE 0xff +#define V_CPL_TLS_DATA_OPCODE(x) ((x) << S_CPL_TLS_DATA_OPCODE) +#define G_CPL_TLS_DATA_OPCODE(x) \ + (((x) >> S_CPL_TLS_DATA_OPCODE) & M_CPL_TLS_DATA_OPCODE) + +#define S_CPL_TLS_DATA_TID 0 +#define M_CPL_TLS_DATA_TID 0xffffff +#define V_CPL_TLS_DATA_TID(x) ((x) << S_CPL_TLS_DATA_TID) +#define G_CPL_TLS_DATA_TID(x) \ + (((x) >> S_CPL_TLS_DATA_TID) & M_CPL_TLS_DATA_TID) + +#define S_CPL_TLS_DATA_LENGTH 0 +#define M_CPL_TLS_DATA_LENGTH 0xffff +#define V_CPL_TLS_DATA_LENGTH(x) ((x) << S_CPL_TLS_DATA_LENGTH) +#define G_CPL_TLS_DATA_LENGTH(x) \ + (((x) >> S_CPL_TLS_DATA_LENGTH) & M_CPL_TLS_DATA_LENGTH) + +struct cpl_rx_tls_cmp { + RSS_HDR + __be32 op_tid; + __be32 pdulength_length; + __be32 seq; + __be32 ddp_report; + __be32 r; + __be32 ddp_valid; +}; + +#define S_CPL_RX_TLS_CMP_OPCODE 24 +#define M_CPL_RX_TLS_CMP_OPCODE 0xff +#define V_CPL_RX_TLS_CMP_OPCODE(x) ((x) << S_CPL_RX_TLS_CMP_OPCODE) +#define G_CPL_RX_TLS_CMP_OPCODE(x) \ + (((x) >> S_CPL_RX_TLS_CMP_OPCODE) & M_CPL_RX_TLS_CMP_OPCODE) + +#define S_CPL_RX_TLS_CMP_TID 0 +#define M_CPL_RX_TLS_CMP_TID 0xffffff +#define V_CPL_RX_TLS_CMP_TID(x) ((x) << S_CPL_RX_TLS_CMP_TID) +#define G_CPL_RX_TLS_CMP_TID(x) \ + (((x) >> S_CPL_RX_TLS_CMP_TID) & M_CPL_RX_TLS_CMP_TID) + +#define S_CPL_RX_TLS_CMP_PDULENGTH 16 +#define M_CPL_RX_TLS_CMP_PDULENGTH 0xffff +#define V_CPL_RX_TLS_CMP_PDULENGTH(x) ((x) << S_CPL_RX_TLS_CMP_PDULENGTH) +#define G_CPL_RX_TLS_CMP_PDULENGTH(x) \ + (((x) >> S_CPL_RX_TLS_CMP_PDULENGTH) & M_CPL_RX_TLS_CMP_PDULENGTH) + +#define S_CPL_RX_TLS_CMP_LENGTH 0 +#define M_CPL_RX_TLS_CMP_LENGTH 0xffff +#define V_CPL_RX_TLS_CMP_LENGTH(x) ((x) << S_CPL_RX_TLS_CMP_LENGTH) +#define G_CPL_RX_TLS_CMP_LENGTH(x) \ + (((x) >> S_CPL_RX_TLS_CMP_LENGTH) & M_CPL_RX_TLS_CMP_LENGTH) + +#define S_SCMD_SEQ_NO_CTRL 29 +#define M_SCMD_SEQ_NO_CTRL 0x3 +#define V_SCMD_SEQ_NO_CTRL(x) ((x) << S_SCMD_SEQ_NO_CTRL) +#define G_SCMD_SEQ_NO_CTRL(x) \ + (((x) >> S_SCMD_SEQ_NO_CTRL) & M_SCMD_SEQ_NO_CTRL) + +/* StsFieldPrsnt- Status field at the end of the TLS PDU */ +#define S_SCMD_STATUS_PRESENT 28 +#define M_SCMD_STATUS_PRESENT 0x1 +#define V_SCMD_STATUS_PRESENT(x) ((x) << S_SCMD_STATUS_PRESENT) +#define G_SCMD_STATUS_PRESENT(x) \ + (((x) >> S_SCMD_STATUS_PRESENT) & M_SCMD_STATUS_PRESENT) +#define F_SCMD_STATUS_PRESENT V_SCMD_STATUS_PRESENT(1U) + +/* ProtoVersion - Protocol Version 0: 1.2, 1:1.1, 2:DTLS, 3:Generic, + * 3-15: Reserved. */ +#define S_SCMD_PROTO_VERSION 24 +#define M_SCMD_PROTO_VERSION 0xf +#define V_SCMD_PROTO_VERSION(x) ((x) << S_SCMD_PROTO_VERSION) +#define G_SCMD_PROTO_VERSION(x) \ + (((x) >> S_SCMD_PROTO_VERSION) & M_SCMD_PROTO_VERSION) + +/* EncDecCtrl - Encryption/Decryption Control. 0: Encrypt, 1: Decrypt */ +#define S_SCMD_ENC_DEC_CTRL 23 +#define M_SCMD_ENC_DEC_CTRL 0x1 +#define V_SCMD_ENC_DEC_CTRL(x) ((x) << S_SCMD_ENC_DEC_CTRL) +#define G_SCMD_ENC_DEC_CTRL(x) \ + (((x) >> S_SCMD_ENC_DEC_CTRL) & M_SCMD_ENC_DEC_CTRL) +#define F_SCMD_ENC_DEC_CTRL V_SCMD_ENC_DEC_CTRL(1U) + +/* CipherAuthSeqCtrl - Cipher Authentication Sequence Control. */ +#define S_SCMD_CIPH_AUTH_SEQ_CTRL 22 +#define M_SCMD_CIPH_AUTH_SEQ_CTRL 0x1 +#define V_SCMD_CIPH_AUTH_SEQ_CTRL(x) \ + ((x) << S_SCMD_CIPH_AUTH_SEQ_CTRL) +#define G_SCMD_CIPH_AUTH_SEQ_CTRL(x) \ + (((x) >> S_SCMD_CIPH_AUTH_SEQ_CTRL) & M_SCMD_CIPH_AUTH_SEQ_CTRL) +#define F_SCMD_CIPH_AUTH_SEQ_CTRL V_SCMD_CIPH_AUTH_SEQ_CTRL(1U) + +/* CiphMode - Cipher Mode. 0: NOP, 1:AES-CBC, 2:AES-GCM, 3:AES-CTR, + * 4:Generic-AES, 5-15: Reserved. */ +#define S_SCMD_CIPH_MODE 18 +#define M_SCMD_CIPH_MODE 0xf +#define V_SCMD_CIPH_MODE(x) ((x) << S_SCMD_CIPH_MODE) +#define G_SCMD_CIPH_MODE(x) \ + (((x) >> S_SCMD_CIPH_MODE) & M_SCMD_CIPH_MODE) + +/* AuthMode - Auth Mode. 0: NOP, 1:SHA1, 2:SHA2-224, 3:SHA2-256 + * 4-15: Reserved */ +#define S_SCMD_AUTH_MODE 14 +#define M_SCMD_AUTH_MODE 0xf +#define V_SCMD_AUTH_MODE(x) ((x) << S_SCMD_AUTH_MODE) +#define G_SCMD_AUTH_MODE(x) \ + (((x) >> S_SCMD_AUTH_MODE) & M_SCMD_AUTH_MODE) + +/* HmacCtrl - HMAC Control. 0:NOP, 1:No truncation, 2:Support HMAC Truncation + * per RFC 4366, 3:IPSec 96 bits, 4-7:Reserved + */ +#define S_SCMD_HMAC_CTRL 11 +#define M_SCMD_HMAC_CTRL 0x7 +#define V_SCMD_HMAC_CTRL(x) ((x) << S_SCMD_HMAC_CTRL) +#define G_SCMD_HMAC_CTRL(x) \ + (((x) >> S_SCMD_HMAC_CTRL) & M_SCMD_HMAC_CTRL) + +/* IvSize - IV size in units of 2 bytes */ +#define S_SCMD_IV_SIZE 7 +#define M_SCMD_IV_SIZE 0xf +#define V_SCMD_IV_SIZE(x) ((x) << S_SCMD_IV_SIZE) +#define G_SCMD_IV_SIZE(x) \ + (((x) >> S_SCMD_IV_SIZE) & M_SCMD_IV_SIZE) + +/* NumIVs - Number of IVs */ +#define S_SCMD_NUM_IVS 0 +#define M_SCMD_NUM_IVS 0x7f +#define V_SCMD_NUM_IVS(x) ((x) << S_SCMD_NUM_IVS) +#define G_SCMD_NUM_IVS(x) \ + (((x) >> S_SCMD_NUM_IVS) & M_SCMD_NUM_IVS) + +/* EnbDbgId - If this is enabled upper 20 (63:44) bits if SeqNumber + * (below) are used as Cid (connection id for debug status), these + * bits are padded to zero for forming the 64 bit + * sequence number for TLS + */ +#define S_SCMD_ENB_DBGID 31 +#define M_SCMD_ENB_DBGID 0x1 +#define V_SCMD_ENB_DBGID(x) ((x) << S_SCMD_ENB_DBGID) +#define G_SCMD_ENB_DBGID(x) \ + (((x) >> S_SCMD_ENB_DBGID) & M_SCMD_ENB_DBGID) + +/* IV generation in SW. */ +#define S_SCMD_IV_GEN_CTRL 30 +#define M_SCMD_IV_GEN_CTRL 0x1 +#define V_SCMD_IV_GEN_CTRL(x) ((x) << S_SCMD_IV_GEN_CTRL) +#define G_SCMD_IV_GEN_CTRL(x) \ + (((x) >> S_SCMD_IV_GEN_CTRL) & M_SCMD_IV_GEN_CTRL) +#define F_SCMD_IV_GEN_CTRL V_SCMD_IV_GEN_CTRL(1U) + +/* More frags */ +#define S_SCMD_MORE_FRAGS 20 +#define M_SCMD_MORE_FRAGS 0x1 +#define V_SCMD_MORE_FRAGS(x) ((x) << S_SCMD_MORE_FRAGS) +#define G_SCMD_MORE_FRAGS(x) (((x) >> S_SCMD_MORE_FRAGS) & M_SCMD_MORE_FRAGS) + +/*last frag */ +#define S_SCMD_LAST_FRAG 19 +#define M_SCMD_LAST_FRAG 0x1 +#define V_SCMD_LAST_FRAG(x) ((x) << S_SCMD_LAST_FRAG) +#define G_SCMD_LAST_FRAG(x) (((x) >> S_SCMD_LAST_FRAG) & M_SCMD_LAST_FRAG) + +/* TlsCompPdu */ +#define S_SCMD_TLS_COMPPDU 18 +#define M_SCMD_TLS_COMPPDU 0x1 +#define V_SCMD_TLS_COMPPDU(x) ((x) << S_SCMD_TLS_COMPPDU) +#define G_SCMD_TLS_COMPPDU(x) (((x) >> S_SCMD_TLS_COMPPDU) & M_SCMD_TLS_COMPPDU) + +/* KeyCntxtInline - Key context inline after the scmd OR PayloadOnly*/ +#define S_SCMD_KEY_CTX_INLINE 17 +#define M_SCMD_KEY_CTX_INLINE 0x1 +#define V_SCMD_KEY_CTX_INLINE(x) ((x) << S_SCMD_KEY_CTX_INLINE) +#define G_SCMD_KEY_CTX_INLINE(x) \ + (((x) >> S_SCMD_KEY_CTX_INLINE) & M_SCMD_KEY_CTX_INLINE) +#define F_SCMD_KEY_CTX_INLINE V_SCMD_KEY_CTX_INLINE(1U) + +/* TLSFragEnable - 0: Host created TLS PDUs, 1: TLS Framgmentation in ASIC */ +#define S_SCMD_TLS_FRAG_ENABLE 16 +#define M_SCMD_TLS_FRAG_ENABLE 0x1 +#define V_SCMD_TLS_FRAG_ENABLE(x) ((x) << S_SCMD_TLS_FRAG_ENABLE) +#define G_SCMD_TLS_FRAG_ENABLE(x) \ + (((x) >> S_SCMD_TLS_FRAG_ENABLE) & M_SCMD_TLS_FRAG_ENABLE) +#define F_SCMD_TLS_FRAG_ENABLE V_SCMD_TLS_FRAG_ENABLE(1U) + +/* MacOnly - Only send the MAC and discard PDU. This is valid for hash only + * modes, in this case TLS_TX will drop the PDU and only + * send back the MAC bytes. */ +#define S_SCMD_MAC_ONLY 15 +#define M_SCMD_MAC_ONLY 0x1 +#define V_SCMD_MAC_ONLY(x) ((x) << S_SCMD_MAC_ONLY) +#define G_SCMD_MAC_ONLY(x) \ + (((x) >> S_SCMD_MAC_ONLY) & M_SCMD_MAC_ONLY) +#define F_SCMD_MAC_ONLY V_SCMD_MAC_ONLY(1U) + +/* AadIVDrop - Drop the AAD and IV fields. Useful in protocols + * which have complex AAD and IV formations Eg:AES-CCM + */ +#define S_SCMD_AADIVDROP 14 +#define M_SCMD_AADIVDROP 0x1 +#define V_SCMD_AADIVDROP(x) ((x) << S_SCMD_AADIVDROP) +#define G_SCMD_AADIVDROP(x) \ + (((x) >> S_SCMD_AADIVDROP) & M_SCMD_AADIVDROP) +#define F_SCMD_AADIVDROP V_SCMD_AADIVDROP(1U) + +/* HdrLength - Length of all headers excluding TLS header + * present before start of crypto PDU/payload. */ +#define S_SCMD_HDR_LEN 0 +#define M_SCMD_HDR_LEN 0x3fff +#define V_SCMD_HDR_LEN(x) ((x) << S_SCMD_HDR_LEN) +#define G_SCMD_HDR_LEN(x) \ + (((x) >> S_SCMD_HDR_LEN) & M_SCMD_HDR_LEN) + +struct cpl_tx_sec_pdu { + __be32 op_ivinsrtofst; + __be32 pldlen; + __be32 aadstart_cipherstop_hi; + __be32 cipherstop_lo_authinsert; + __be32 seqno_numivs; + __be32 ivgen_hdrlen; + __be64 scmd1; +}; + +#define S_CPL_TX_SEC_PDU_OPCODE 24 +#define M_CPL_TX_SEC_PDU_OPCODE 0xff +#define V_CPL_TX_SEC_PDU_OPCODE(x) ((x) << S_CPL_TX_SEC_PDU_OPCODE) +#define G_CPL_TX_SEC_PDU_OPCODE(x) \ + (((x) >> S_CPL_TX_SEC_PDU_OPCODE) & M_CPL_TX_SEC_PDU_OPCODE) + +/* RX Channel Id */ +#define S_CPL_TX_SEC_PDU_RXCHID 22 +#define M_CPL_TX_SEC_PDU_RXCHID 0x1 +#define V_CPL_TX_SEC_PDU_RXCHID(x) ((x) << S_CPL_TX_SEC_PDU_RXCHID) +#define G_CPL_TX_SEC_PDU_RXCHID(x) \ +(((x) >> S_CPL_TX_SEC_PDU_RXCHID) & M_CPL_TX_SEC_PDU_RXCHID) +#define F_CPL_TX_SEC_PDU_RXCHID V_CPL_TX_SEC_PDU_RXCHID(1U) + +/* Ack Follows */ +#define S_CPL_TX_SEC_PDU_ACKFOLLOWS 21 +#define M_CPL_TX_SEC_PDU_ACKFOLLOWS 0x1 +#define V_CPL_TX_SEC_PDU_ACKFOLLOWS(x) ((x) << S_CPL_TX_SEC_PDU_ACKFOLLOWS) +#define G_CPL_TX_SEC_PDU_ACKFOLLOWS(x) \ +(((x) >> S_CPL_TX_SEC_PDU_ACKFOLLOWS) & M_CPL_TX_SEC_PDU_ACKFOLLOWS) +#define F_CPL_TX_SEC_PDU_ACKFOLLOWS V_CPL_TX_SEC_PDU_ACKFOLLOWS(1U) + +/* Loopback bit in cpl_tx_sec_pdu */ +#define S_CPL_TX_SEC_PDU_ULPTXLPBK 20 +#define M_CPL_TX_SEC_PDU_ULPTXLPBK 0x1 +#define V_CPL_TX_SEC_PDU_ULPTXLPBK(x) ((x) << S_CPL_TX_SEC_PDU_ULPTXLPBK) +#define G_CPL_TX_SEC_PDU_ULPTXLPBK(x) \ +(((x) >> S_CPL_TX_SEC_PDU_ULPTXLPBK) & M_CPL_TX_SEC_PDU_ULPTXLPBK) +#define F_CPL_TX_SEC_PDU_ULPTXLPBK V_CPL_TX_SEC_PDU_ULPTXLPBK(1U) + +/* Length of cpl header encapsulated */ +#define S_CPL_TX_SEC_PDU_CPLLEN 16 +#define M_CPL_TX_SEC_PDU_CPLLEN 0xf +#define V_CPL_TX_SEC_PDU_CPLLEN(x) ((x) << S_CPL_TX_SEC_PDU_CPLLEN) +#define G_CPL_TX_SEC_PDU_CPLLEN(x) \ + (((x) >> S_CPL_TX_SEC_PDU_CPLLEN) & M_CPL_TX_SEC_PDU_CPLLEN) + +/* PlaceHolder */ +#define S_CPL_TX_SEC_PDU_PLACEHOLDER 10 +#define M_CPL_TX_SEC_PDU_PLACEHOLDER 0x1 +#define V_CPL_TX_SEC_PDU_PLACEHOLDER(x) ((x) << S_CPL_TX_SEC_PDU_PLACEHOLDER) +#define G_CPL_TX_SEC_PDU_PLACEHOLDER(x) \ + (((x) >> S_CPL_TX_SEC_PDU_PLACEHOLDER) & \ + M_CPL_TX_SEC_PDU_PLACEHOLDER) + +/* IvInsrtOffset: Insertion location for IV */ +#define S_CPL_TX_SEC_PDU_IVINSRTOFST 0 +#define M_CPL_TX_SEC_PDU_IVINSRTOFST 0x3ff +#define V_CPL_TX_SEC_PDU_IVINSRTOFST(x) ((x) << S_CPL_TX_SEC_PDU_IVINSRTOFST) +#define G_CPL_TX_SEC_PDU_IVINSRTOFST(x) \ + (((x) >> S_CPL_TX_SEC_PDU_IVINSRTOFST) & \ + M_CPL_TX_SEC_PDU_IVINSRTOFST) + +/* AadStartOffset: Offset in bytes for AAD start from + * the first byte following + * the pkt headers (0-255 + * bytes) */ +#define S_CPL_TX_SEC_PDU_AADSTART 24 +#define M_CPL_TX_SEC_PDU_AADSTART 0xff +#define V_CPL_TX_SEC_PDU_AADSTART(x) ((x) << S_CPL_TX_SEC_PDU_AADSTART) +#define G_CPL_TX_SEC_PDU_AADSTART(x) \ + (((x) >> S_CPL_TX_SEC_PDU_AADSTART) & \ + M_CPL_TX_SEC_PDU_AADSTART) + +/* AadStopOffset: offset in bytes for AAD stop/end from the first byte following + * the pkt headers (0-511 bytes) */ +#define S_CPL_TX_SEC_PDU_AADSTOP 15 +#define M_CPL_TX_SEC_PDU_AADSTOP 0x1ff +#define V_CPL_TX_SEC_PDU_AADSTOP(x) ((x) << S_CPL_TX_SEC_PDU_AADSTOP) +#define G_CPL_TX_SEC_PDU_AADSTOP(x) \ + (((x) >> S_CPL_TX_SEC_PDU_AADSTOP) & M_CPL_TX_SEC_PDU_AADSTOP) + +/* CipherStartOffset: offset in bytes for encryption/decryption start from the + * first byte following the pkt headers (0-1023 + * bytes) */ +#define S_CPL_TX_SEC_PDU_CIPHERSTART 5 +#define M_CPL_TX_SEC_PDU_CIPHERSTART 0x3ff +#define V_CPL_TX_SEC_PDU_CIPHERSTART(x) ((x) << S_CPL_TX_SEC_PDU_CIPHERSTART) +#define G_CPL_TX_SEC_PDU_CIPHERSTART(x) \ + (((x) >> S_CPL_TX_SEC_PDU_CIPHERSTART) & \ + M_CPL_TX_SEC_PDU_CIPHERSTART) + +/* CipherStopOffset: offset in bytes for encryption/decryption end + * from end of the payload of this command (0-511 bytes) */ +#define S_CPL_TX_SEC_PDU_CIPHERSTOP_HI 0 +#define M_CPL_TX_SEC_PDU_CIPHERSTOP_HI 0x1f +#define V_CPL_TX_SEC_PDU_CIPHERSTOP_HI(x) \ + ((x) << S_CPL_TX_SEC_PDU_CIPHERSTOP_HI) +#define G_CPL_TX_SEC_PDU_CIPHERSTOP_HI(x) \ + (((x) >> S_CPL_TX_SEC_PDU_CIPHERSTOP_HI) & \ + M_CPL_TX_SEC_PDU_CIPHERSTOP_HI) + +#define S_CPL_TX_SEC_PDU_CIPHERSTOP_LO 28 +#define M_CPL_TX_SEC_PDU_CIPHERSTOP_LO 0xf +#define V_CPL_TX_SEC_PDU_CIPHERSTOP_LO(x) \ + ((x) << S_CPL_TX_SEC_PDU_CIPHERSTOP_LO) +#define G_CPL_TX_SEC_PDU_CIPHERSTOP_LO(x) \ + (((x) >> S_CPL_TX_SEC_PDU_CIPHERSTOP_LO) & \ + M_CPL_TX_SEC_PDU_CIPHERSTOP_LO) + +/* AuthStartOffset: offset in bytes for authentication start from + * the first byte following the pkt headers (0-1023) + * */ +#define S_CPL_TX_SEC_PDU_AUTHSTART 18 +#define M_CPL_TX_SEC_PDU_AUTHSTART 0x3ff +#define V_CPL_TX_SEC_PDU_AUTHSTART(x) ((x) << S_CPL_TX_SEC_PDU_AUTHSTART) +#define G_CPL_TX_SEC_PDU_AUTHSTART(x) \ + (((x) >> S_CPL_TX_SEC_PDU_AUTHSTART) & \ + M_CPL_TX_SEC_PDU_AUTHSTART) + +/* AuthStopOffset: offset in bytes for authentication + * end from end of the payload of this command (0-511 Bytes) */ +#define S_CPL_TX_SEC_PDU_AUTHSTOP 9 +#define M_CPL_TX_SEC_PDU_AUTHSTOP 0x1ff +#define V_CPL_TX_SEC_PDU_AUTHSTOP(x) ((x) << S_CPL_TX_SEC_PDU_AUTHSTOP) +#define G_CPL_TX_SEC_PDU_AUTHSTOP(x) \ + (((x) >> S_CPL_TX_SEC_PDU_AUTHSTOP) & \ + M_CPL_TX_SEC_PDU_AUTHSTOP) + +/* AuthInsrtOffset: offset in bytes for authentication insertion + * from end of the payload of this command (0-511 bytes) */ +#define S_CPL_TX_SEC_PDU_AUTHINSERT 0 +#define M_CPL_TX_SEC_PDU_AUTHINSERT 0x1ff +#define V_CPL_TX_SEC_PDU_AUTHINSERT(x) ((x) << S_CPL_TX_SEC_PDU_AUTHINSERT) +#define G_CPL_TX_SEC_PDU_AUTHINSERT(x) \ + (((x) >> S_CPL_TX_SEC_PDU_AUTHINSERT) & \ + M_CPL_TX_SEC_PDU_AUTHINSERT) + +struct cpl_rx_phys_dsgl { + __be32 op_to_tid; + __be32 pcirlxorder_to_noofsgentr; + struct rss_header rss_hdr_int; +}; + +#define S_CPL_RX_PHYS_DSGL_OPCODE 24 +#define M_CPL_RX_PHYS_DSGL_OPCODE 0xff +#define V_CPL_RX_PHYS_DSGL_OPCODE(x) ((x) << S_CPL_RX_PHYS_DSGL_OPCODE) +#define G_CPL_RX_PHYS_DSGL_OPCODE(x) \ + (((x) >> S_CPL_RX_PHYS_DSGL_OPCODE) & M_CPL_RX_PHYS_DSGL_OPCODE) + +#define S_CPL_RX_PHYS_DSGL_ISRDMA 23 +#define M_CPL_RX_PHYS_DSGL_ISRDMA 0x1 +#define V_CPL_RX_PHYS_DSGL_ISRDMA(x) ((x) << S_CPL_RX_PHYS_DSGL_ISRDMA) +#define G_CPL_RX_PHYS_DSGL_ISRDMA(x) \ + (((x) >> S_CPL_RX_PHYS_DSGL_ISRDMA) & M_CPL_RX_PHYS_DSGL_ISRDMA) +#define F_CPL_RX_PHYS_DSGL_ISRDMA V_CPL_RX_PHYS_DSGL_ISRDMA(1U) + +#define S_CPL_RX_PHYS_DSGL_RSVD1 20 +#define M_CPL_RX_PHYS_DSGL_RSVD1 0x7 +#define V_CPL_RX_PHYS_DSGL_RSVD1(x) ((x) << S_CPL_RX_PHYS_DSGL_RSVD1) +#define G_CPL_RX_PHYS_DSGL_RSVD1(x) \ + (((x) >> S_CPL_RX_PHYS_DSGL_RSVD1) & M_CPL_RX_PHYS_DSGL_RSVD1) + +#define S_CPL_RX_PHYS_DSGL_PCIRLXORDER 31 +#define M_CPL_RX_PHYS_DSGL_PCIRLXORDER 0x1 +#define V_CPL_RX_PHYS_DSGL_PCIRLXORDER(x) \ + ((x) << S_CPL_RX_PHYS_DSGL_PCIRLXORDER) +#define G_CPL_RX_PHYS_DSGL_PCIRLXORDER(x) \ + (((x) >> S_CPL_RX_PHYS_DSGL_PCIRLXORDER) & \ + M_CPL_RX_PHYS_DSGL_PCIRLXORDER) +#define F_CPL_RX_PHYS_DSGL_PCIRLXORDER V_CPL_RX_PHYS_DSGL_PCIRLXORDER(1U) + +#define S_CPL_RX_PHYS_DSGL_PCINOSNOOP 30 +#define M_CPL_RX_PHYS_DSGL_PCINOSNOOP 0x1 +#define V_CPL_RX_PHYS_DSGL_PCINOSNOOP(x) \ + ((x) << S_CPL_RX_PHYS_DSGL_PCINOSNOOP) +#define G_CPL_RX_PHYS_DSGL_PCINOSNOOP(x) \ + (((x) >> S_CPL_RX_PHYS_DSGL_PCINOSNOOP) & \ + M_CPL_RX_PHYS_DSGL_PCINOSNOOP) +#define F_CPL_RX_PHYS_DSGL_PCINOSNOOP V_CPL_RX_PHYS_DSGL_PCINOSNOOP(1U) + +#define S_CPL_RX_PHYS_DSGL_PCITPHNTENB 29 +#define M_CPL_RX_PHYS_DSGL_PCITPHNTENB 0x1 +#define V_CPL_RX_PHYS_DSGL_PCITPHNTENB(x) \ + ((x) << S_CPL_RX_PHYS_DSGL_PCITPHNTENB) +#define G_CPL_RX_PHYS_DSGL_PCITPHNTENB(x) \ + (((x) >> S_CPL_RX_PHYS_DSGL_PCITPHNTENB) & \ + M_CPL_RX_PHYS_DSGL_PCITPHNTENB) +#define F_CPL_RX_PHYS_DSGL_PCITPHNTENB V_CPL_RX_PHYS_DSGL_PCITPHNTENB(1U) + +#define S_CPL_RX_PHYS_DSGL_PCITPHNT 27 +#define M_CPL_RX_PHYS_DSGL_PCITPHNT 0x3 +#define V_CPL_RX_PHYS_DSGL_PCITPHNT(x) ((x) << S_CPL_RX_PHYS_DSGL_PCITPHNT) +#define G_CPL_RX_PHYS_DSGL_PCITPHNT(x) \ + (((x) >> S_CPL_RX_PHYS_DSGL_PCITPHNT) & \ + M_CPL_RX_PHYS_DSGL_PCITPHNT) + +#define S_CPL_RX_PHYS_DSGL_DCAID 16 +#define M_CPL_RX_PHYS_DSGL_DCAID 0x7ff +#define V_CPL_RX_PHYS_DSGL_DCAID(x) ((x) << S_CPL_RX_PHYS_DSGL_DCAID) +#define G_CPL_RX_PHYS_DSGL_DCAID(x) \ + (((x) >> S_CPL_RX_PHYS_DSGL_DCAID) & \ + M_CPL_RX_PHYS_DSGL_DCAID) + +#define S_CPL_RX_PHYS_DSGL_NOOFSGENTR 0 +#define M_CPL_RX_PHYS_DSGL_NOOFSGENTR 0xffff +#define V_CPL_RX_PHYS_DSGL_NOOFSGENTR(x) \ + ((x) << S_CPL_RX_PHYS_DSGL_NOOFSGENTR) +#define G_CPL_RX_PHYS_DSGL_NOOFSGENTR(x) \ + (((x) >> S_CPL_RX_PHYS_DSGL_NOOFSGENTR) & \ + M_CPL_RX_PHYS_DSGL_NOOFSGENTR) + +/* CPL_TX_TLS_ACK */ +struct cpl_tx_tls_ack { + __be32 op_to_Rsvd2; + __be32 PldLen; + __be64 Rsvd3; +}; + +#define S_CPL_TX_TLS_ACK_OPCODE 24 +#define M_CPL_TX_TLS_ACK_OPCODE 0xff +#define V_CPL_TX_TLS_ACK_OPCODE(x) ((x) << S_CPL_TX_TLS_ACK_OPCODE) +#define G_CPL_TX_TLS_ACK_OPCODE(x) \ + (((x) >> S_CPL_TX_TLS_ACK_OPCODE) & M_CPL_TX_TLS_ACK_OPCODE) + +#define S_CPL_TX_TLS_ACK_RSVD1 23 +#define M_CPL_TX_TLS_ACK_RSVD1 0x1 +#define V_CPL_TX_TLS_ACK_RSVD1(x) ((x) << S_CPL_TX_TLS_ACK_RSVD1) +#define G_CPL_TX_TLS_ACK_RSVD1(x) \ + (((x) >> S_CPL_TX_TLS_ACK_RSVD1) & M_CPL_TX_TLS_ACK_RSVD1) +#define F_CPL_TX_TLS_ACK_RSVD1 V_CPL_TX_TLS_ACK_RSVD1(1U) + +#define S_CPL_TX_TLS_ACK_RXCHID 22 +#define M_CPL_TX_TLS_ACK_RXCHID 0x1 +#define V_CPL_TX_TLS_ACK_RXCHID(x) ((x) << S_CPL_TX_TLS_ACK_RXCHID) +#define G_CPL_TX_TLS_ACK_RXCHID(x) \ + (((x) >> S_CPL_TX_TLS_ACK_RXCHID) & M_CPL_TX_TLS_ACK_RXCHID) +#define F_CPL_TX_TLS_ACK_RXCHID V_CPL_TX_TLS_ACK_RXCHID(1U) + +#define S_CPL_TX_TLS_ACK_FWMSG 21 +#define M_CPL_TX_TLS_ACK_FWMSG 0x1 +#define V_CPL_TX_TLS_ACK_FWMSG(x) ((x) << S_CPL_TX_TLS_ACK_FWMSG) +#define G_CPL_TX_TLS_ACK_FWMSG(x) \ + (((x) >> S_CPL_TX_TLS_ACK_FWMSG) & M_CPL_TX_TLS_ACK_FWMSG) +#define F_CPL_TX_TLS_ACK_FWMSG V_CPL_TX_TLS_ACK_FWMSG(1U) + +#define S_CPL_TX_TLS_ACK_ULPTXLPBK 20 +#define M_CPL_TX_TLS_ACK_ULPTXLPBK 0x1 +#define V_CPL_TX_TLS_ACK_ULPTXLPBK(x) ((x) << S_CPL_TX_TLS_ACK_ULPTXLPBK) +#define G_CPL_TX_TLS_ACK_ULPTXLPBK(x) \ + (((x) >> S_CPL_TX_TLS_ACK_ULPTXLPBK) & M_CPL_TX_TLS_ACK_ULPTXLPBK) +#define F_CPL_TX_TLS_ACK_ULPTXLPBK V_CPL_TX_TLS_ACK_ULPTXLPBK(1U) + +#define S_CPL_TX_TLS_ACK_CPLLEN 16 +#define M_CPL_TX_TLS_ACK_CPLLEN 0xf +#define V_CPL_TX_TLS_ACK_CPLLEN(x) ((x) << S_CPL_TX_TLS_ACK_CPLLEN) +#define G_CPL_TX_TLS_ACK_CPLLEN(x) \ + (((x) >> S_CPL_TX_TLS_ACK_CPLLEN) & M_CPL_TX_TLS_ACK_CPLLEN) + +#define S_CPL_TX_TLS_ACK_COMPLONERR 15 +#define M_CPL_TX_TLS_ACK_COMPLONERR 0x1 +#define V_CPL_TX_TLS_ACK_COMPLONERR(x) ((x) << S_CPL_TX_TLS_ACK_COMPLONERR) +#define G_CPL_TX_TLS_ACK_COMPLONERR(x) \ + (((x) >> S_CPL_TX_TLS_ACK_COMPLONERR) & M_CPL_TX_TLS_ACK_COMPLONERR) +#define F_CPL_TX_TLS_ACK_COMPLONERR V_CPL_TX_TLS_ACK_COMPLONERR(1U) + +#define S_CPL_TX_TLS_ACK_LCB 14 +#define M_CPL_TX_TLS_ACK_LCB 0x1 +#define V_CPL_TX_TLS_ACK_LCB(x) ((x) << S_CPL_TX_TLS_ACK_LCB) +#define G_CPL_TX_TLS_ACK_LCB(x) \ + (((x) >> S_CPL_TX_TLS_ACK_LCB) & M_CPL_TX_TLS_ACK_LCB) +#define F_CPL_TX_TLS_ACK_LCB V_CPL_TX_TLS_ACK_LCB(1U) + +#define S_CPL_TX_TLS_ACK_PHASH 13 +#define M_CPL_TX_TLS_ACK_PHASH 0x1 +#define V_CPL_TX_TLS_ACK_PHASH(x) ((x) << S_CPL_TX_TLS_ACK_PHASH) +#define G_CPL_TX_TLS_ACK_PHASH(x) \ + (((x) >> S_CPL_TX_TLS_ACK_PHASH) & M_CPL_TX_TLS_ACK_PHASH) +#define F_CPL_TX_TLS_ACK_PHASH V_CPL_TX_TLS_ACK_PHASH(1U) + +#define S_CPL_TX_TLS_ACK_RSVD2 0 +#define M_CPL_TX_TLS_ACK_RSVD2 0x1fff +#define V_CPL_TX_TLS_ACK_RSVD2(x) ((x) << S_CPL_TX_TLS_ACK_RSVD2) +#define G_CPL_TX_TLS_ACK_RSVD2(x) \ + (((x) >> S_CPL_TX_TLS_ACK_RSVD2) & M_CPL_TX_TLS_ACK_RSVD2) + #endif /* T4_MSG_H */ From ae9fd9cccd6fbb4697069db649bf0aba3a63413c Mon Sep 17 00:00:00 2001 From: Bruce M Simpson Date: Tue, 10 Jan 2017 01:31:26 +0000 Subject: [PATCH 15/79] Allow udbp(4) to claim Belkin "Windows Easy Transfer Cable" for Netgraph use. These have been tested back-to-back with Linux 3.x and a similar attachment at the other end; a CDC EEM-like encapsulation can be used for emulated Ethernet over udbp(4) with ng_ether. --- sys/dev/usb/misc/udbp.c | 1 + 1 file changed, 1 insertion(+) diff --git a/sys/dev/usb/misc/udbp.c b/sys/dev/usb/misc/udbp.c index 3c07931bc068..f8f7a4b6f44a 100644 --- a/sys/dev/usb/misc/udbp.c +++ b/sys/dev/usb/misc/udbp.c @@ -259,6 +259,7 @@ static driver_t udbp_driver = { }; static const STRUCT_USB_HOST_ID udbp_devs[] = { + {USB_VPI(USB_VENDOR_BELKIN, USB_PRODUCT_BELKIN_F5U258, 0)}, {USB_VPI(USB_VENDOR_NETCHIP, USB_PRODUCT_NETCHIP_TURBOCONNECT, 0)}, {USB_VPI(USB_VENDOR_NETCHIP, USB_PRODUCT_NETCHIP_GADGETZERO, 0)}, {USB_VPI(USB_VENDOR_PROLIFIC, USB_PRODUCT_PROLIFIC_PL2301, 0)}, From e8257dbe43f221ece15dff805a2efe3b719bc513 Mon Sep 17 00:00:00 2001 From: Navdeep Parhar Date: Tue, 10 Jan 2017 01:36:50 +0000 Subject: [PATCH 16/79] cxgbe(4): Attach to the 2x25 debug card. This is for internal use only. MFC after: 3 days --- sys/dev/cxgbe/t4_main.c | 1 + 1 file changed, 1 insertion(+) diff --git a/sys/dev/cxgbe/t4_main.c b/sys/dev/cxgbe/t4_main.c index 1569b7439ff8..d4048cd4ce3e 100644 --- a/sys/dev/cxgbe/t4_main.c +++ b/sys/dev/cxgbe/t4_main.c @@ -623,6 +623,7 @@ struct { #endif }, t6_pciids[] = { {0xc006, "Chelsio Terminator 6 FPGA"}, /* T6 PE10K6 FPGA (PF0) */ + {0x6400, "Chelsio T6225-DBG"}, /* 2 x 10/25G, debug */ {0x6401, "Chelsio T6225-CR"}, /* 2 x 10/25G */ {0x6402, "Chelsio T6225-SO-CR"}, /* 2 x 10/25G, nomem */ {0x6407, "Chelsio T62100-LP-CR"}, /* 2 x 40/50/100G */ From f2d6ace4a684fcb98293983758c73d703338a78b Mon Sep 17 00:00:00 2001 From: Sean Bruno Date: Tue, 10 Jan 2017 03:23:22 +0000 Subject: [PATCH 17/79] Migrate e1000 to the IFLIB framework: - em(4) igb(4) and lem(4) - deprecate the igb device from kernel configurations - create a symbolic link in /boot/kernel from if_em.ko to if_igb.ko Devices tested: - 82574L - I218-LM - 82546GB - 82579LM - I350 - I217 Please report problems to freebsd-net@freebsd.org Partial review from jhb and suggestions on how to *not* brick folks who originally would have lost their igbX device. Submitted by: mmacy@nextbsd.org MFC after: 2 weeks Relnotes: yes Sponsored by: Limelight Networks and Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D8299 --- UPDATING | 5 + sys/amd64/conf/GENERIC | 1 - sys/arm64/conf/GENERIC | 1 - sys/conf/NOTES | 2 - sys/conf/files | 38 +- sys/conf/makeLINT.mk | 1 - sys/dev/e1000/em_txrx.c | 720 ++++ sys/dev/e1000/if_em.c | 5251 +++++++++-------------------- sys/dev/e1000/if_em.h | 295 +- sys/dev/e1000/if_igb.c | 6450 ------------------------------------ sys/dev/e1000/if_igb.h | 634 ---- sys/dev/e1000/if_lem.c | 4730 -------------------------- sys/dev/e1000/if_lem.h | 519 --- sys/dev/e1000/igb_txrx.c | 594 ++++ sys/i386/conf/GENERIC | 1 - sys/mips/conf/OCTEON1 | 1 - sys/modules/Makefile | 4 - sys/modules/em/Makefile | 13 +- sys/modules/igb/Makefile | 24 - sys/powerpc/conf/GENERIC64 | 1 - 20 files changed, 3115 insertions(+), 16170 deletions(-) create mode 100644 sys/dev/e1000/em_txrx.c delete mode 100644 sys/dev/e1000/if_igb.c delete mode 100644 sys/dev/e1000/if_igb.h delete mode 100644 sys/dev/e1000/if_lem.c delete mode 100644 sys/dev/e1000/if_lem.h create mode 100644 sys/dev/e1000/igb_txrx.c delete mode 100644 sys/modules/igb/Makefile diff --git a/UPDATING b/UPDATING index e5586009ff69..d604e4289431 100644 --- a/UPDATING +++ b/UPDATING @@ -51,6 +51,11 @@ NOTE TO PEOPLE WHO THINK THAT FreeBSD 12.x IS SLOW: ****************************** SPECIAL WARNING: ****************************** +20170109: + The igb(4), em(4) and lem(4) ethernet drivers are now implemented via + IFLIB. If you have a custom kernel configuration that excludes em(4) + but you use igb(4), you need to re-add em(4) to your custom configuration. + 20161217: Clang, llvm, lldb, compiler-rt and libc++ have been upgraded to 3.9.1. Please see the 20141231 entry below for information about prerequisites diff --git a/sys/amd64/conf/GENERIC b/sys/amd64/conf/GENERIC index db8d07d10b3c..ba5bfe688fc0 100644 --- a/sys/amd64/conf/GENERIC +++ b/sys/amd64/conf/GENERIC @@ -230,7 +230,6 @@ device puc # Multi I/O cards and multi-channel UARTs device bxe # Broadcom NetXtreme II BCM5771X/BCM578XX 10GbE device de # DEC/Intel DC21x4x (``Tulip'') device em # Intel PRO/1000 Gigabit Ethernet Family -device igb # Intel PRO/1000 PCIE Server Gigabit Family device ix # Intel PRO/10GbE PCIE PF Ethernet device ixv # Intel PRO/10GbE PCIE VF Ethernet device ixl # Intel XL710 40Gbe PCIE Ethernet diff --git a/sys/arm64/conf/GENERIC b/sys/arm64/conf/GENERIC index ee4bc49f3e64..790c02a2d369 100644 --- a/sys/arm64/conf/GENERIC +++ b/sys/arm64/conf/GENERIC @@ -120,7 +120,6 @@ device mii device miibus # MII bus support device awg # Allwinner EMAC Gigabit Ethernet device em # Intel PRO/1000 Gigabit Ethernet Family -device igb # Intel PRO/1000 PCIE Server Gigabit Family device ix # Intel 10Gb Ethernet Family device msk # Marvell/SysKonnect Yukon II Gigabit Ethernet device smc # SMSC LAN91C111 diff --git a/sys/conf/NOTES b/sys/conf/NOTES index 0ade0ad97011..711c9a88de4a 100644 --- a/sys/conf/NOTES +++ b/sys/conf/NOTES @@ -1972,7 +1972,6 @@ device xmphy # XaQti XMAC II # KNE110TX. # de: Digital Equipment DC21040 # em: Intel Pro/1000 Gigabit Ethernet 82542, 82543, 82544 based adapters. -# igb: Intel Pro/1000 PCI Express Gigabit Ethernet: 82575 and later adapters. # ep: 3Com 3C509, 3C529, 3C556, 3C562D, 3C563D, 3C572, 3C574X, 3C579, 3C589 # and PC Card devices using these chipsets. # ex: Intel EtherExpress Pro/10 and other i82595-based adapters, @@ -2145,7 +2144,6 @@ device cxgbe # Chelsio T4-T6 1/10/25/40/100 Gigabit Ethernet device cxgbev # Chelsio T4-T6 Virtual Functions device de # DEC/Intel DC21x4x (``Tulip'') device em # Intel Pro/1000 Gigabit Ethernet -device igb # Intel Pro/1000 PCIE Gigabit Ethernet device ixgb # Intel Pro/10Gbe PCI-X Ethernet device ix # Intel Pro/10Gbe PCIE Ethernet device ixv # Intel Pro/10Gbe PCIE Ethernet VF diff --git a/sys/conf/files b/sys/conf/files index ce9042a7b196..d9b7972be69f 100644 --- a/sys/conf/files +++ b/sys/conf/files @@ -1572,43 +1572,43 @@ dev/eisa/eisa_if.m standard dev/eisa/eisaconf.c optional eisa dev/e1000/if_em.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/if_lem.c optional em \ +dev/e1000/em_txrx.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/if_igb.c optional igb \ +dev/e1000/igb_txrx.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_80003es2lan.c optional em | igb \ +dev/e1000/e1000_80003es2lan.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_82540.c optional em | igb \ +dev/e1000/e1000_82540.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_82541.c optional em | igb \ +dev/e1000/e1000_82541.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_82542.c optional em | igb \ +dev/e1000/e1000_82542.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_82543.c optional em | igb \ +dev/e1000/e1000_82543.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_82571.c optional em | igb \ +dev/e1000/e1000_82571.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_82575.c optional em | igb \ +dev/e1000/e1000_82575.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_ich8lan.c optional em | igb \ +dev/e1000/e1000_ich8lan.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_i210.c optional em | igb \ +dev/e1000/e1000_i210.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_api.c optional em | igb \ +dev/e1000/e1000_api.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_mac.c optional em | igb \ +dev/e1000/e1000_mac.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_manage.c optional em | igb \ +dev/e1000/e1000_manage.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_nvm.c optional em | igb \ +dev/e1000/e1000_nvm.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_phy.c optional em | igb \ +dev/e1000/e1000_phy.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_vf.c optional em | igb \ +dev/e1000/e1000_vf.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_mbx.c optional em | igb \ +dev/e1000/e1000_mbx.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" -dev/e1000/e1000_osdep.c optional em | igb \ +dev/e1000/e1000_osdep.c optional em \ compile-with "${NORMAL_C} -I$S/dev/e1000" dev/et/if_et.c optional et dev/en/if_en_pci.c optional en pci diff --git a/sys/conf/makeLINT.mk b/sys/conf/makeLINT.mk index e2c7e5e00aff..04dfcee05b9c 100644 --- a/sys/conf/makeLINT.mk +++ b/sys/conf/makeLINT.mk @@ -38,7 +38,6 @@ LINT: ${NOTES} ../../conf/makeLINT.sed echo "nodevice bxe" >> ${.TARGET}-NOIP echo "nodevice em" >> ${.TARGET}-NOIP echo "nodevice fxp" >> ${.TARGET}-NOIP - echo "nodevice igb" >> ${.TARGET}-NOIP echo "nodevice jme" >> ${.TARGET}-NOIP echo "nodevice msk" >> ${.TARGET}-NOIP echo "nodevice mxge" >> ${.TARGET}-NOIP diff --git a/sys/dev/e1000/em_txrx.c b/sys/dev/e1000/em_txrx.c new file mode 100644 index 000000000000..cf52656c41c6 --- /dev/null +++ b/sys/dev/e1000/em_txrx.c @@ -0,0 +1,720 @@ +/* $FreeBSD$ */ +#include "if_em.h" + +#ifdef RSS +#include +#include +#endif + +#ifdef VERBOSE_DEBUG +#define DPRINTF device_printf +#else +#define DPRINTF(...) +#endif + +/********************************************************************* + * Local Function prototypes + *********************************************************************/ +static int em_tso_setup(struct adapter *adapter, if_pkt_info_t pi, u32 *txd_upper, u32 *txd_lower); +static int em_transmit_checksum_setup(struct adapter *adapter, if_pkt_info_t pi, u32 *txd_upper, u32 *txd_lower); +static int em_isc_txd_encap(void *arg, if_pkt_info_t pi); +static void em_isc_txd_flush(void *arg, uint16_t txqid, uint32_t pidx); +static int em_isc_txd_credits_update(void *arg, uint16_t txqid, uint32_t cidx_init, bool clear); +static void em_isc_rxd_refill(void *arg, uint16_t rxqid, uint8_t flid __unused, + uint32_t pidx, uint64_t *paddrs, caddr_t *vaddrs __unused, uint16_t count, uint16_t buflen __unused); +static void em_isc_rxd_flush(void *arg, uint16_t rxqid, uint8_t flid __unused, uint32_t pidx); +static int em_isc_rxd_available(void *arg, uint16_t rxqid, uint32_t idx, + int budget); +static int em_isc_rxd_pkt_get(void *arg, if_rxd_info_t ri); + +static void lem_isc_rxd_refill(void *arg, uint16_t rxqid, uint8_t flid __unused, + uint32_t pidx, uint64_t *paddrs, caddr_t *vaddrs __unused, uint16_t count, uint16_t buflen __unused); + +static int lem_isc_rxd_available(void *arg, uint16_t rxqid, uint32_t idx, + int budget); +static int lem_isc_rxd_pkt_get(void *arg, if_rxd_info_t ri); + +static void lem_receive_checksum(int status, int errors, if_rxd_info_t ri); +static void em_receive_checksum(uint32_t status, if_rxd_info_t ri); +extern int em_intr(void *arg); + +struct if_txrx em_txrx = { + em_isc_txd_encap, + em_isc_txd_flush, + em_isc_txd_credits_update, + em_isc_rxd_available, + em_isc_rxd_pkt_get, + em_isc_rxd_refill, + em_isc_rxd_flush, + em_intr +}; + +struct if_txrx lem_txrx = { + em_isc_txd_encap, + em_isc_txd_flush, + em_isc_txd_credits_update, + lem_isc_rxd_available, + lem_isc_rxd_pkt_get, + lem_isc_rxd_refill, + em_isc_rxd_flush, + em_intr +}; + +extern if_shared_ctx_t em_sctx; + +/********************************************************************** + * + * Setup work for hardware segmentation offload (TSO) on + * adapters using advanced tx descriptors + * + **********************************************************************/ +static int +em_tso_setup(struct adapter *adapter, if_pkt_info_t pi, u32 *txd_upper, u32 *txd_lower) +{ + if_softc_ctx_t scctx = adapter->shared; + struct em_tx_queue *que = &adapter->tx_queues[pi->ipi_qsidx]; + struct tx_ring *txr = &que->txr; + struct e1000_context_desc *TXD; + struct em_txbuffer *tx_buffer; + int cur, hdr_len; + + hdr_len = pi->ipi_ehdrlen + pi->ipi_ip_hlen + pi->ipi_tcp_hlen; + *txd_lower = (E1000_TXD_CMD_DEXT | /* Extended descr type */ + E1000_TXD_DTYP_D | /* Data descr type */ + E1000_TXD_CMD_TSE); /* Do TSE on this packet */ + + /* IP and/or TCP header checksum calculation and insertion. */ + *txd_upper = (E1000_TXD_POPTS_IXSM | E1000_TXD_POPTS_TXSM) << 8; + + cur = pi->ipi_pidx; + TXD = (struct e1000_context_desc *)&txr->tx_base[cur]; + tx_buffer = &txr->tx_buffers[cur]; + + /* + * Start offset for header checksum calculation. + * End offset for header checksum calculation. + * Offset of place put the checksum. + */ + TXD->lower_setup.ip_fields.ipcss = pi->ipi_ehdrlen; + TXD->lower_setup.ip_fields.ipcse = + htole16(pi->ipi_ehdrlen + pi->ipi_ip_hlen - 1); + TXD->lower_setup.ip_fields.ipcso = pi->ipi_ehdrlen + offsetof(struct ip, ip_sum); + + /* + * Start offset for payload checksum calculation. + * End offset for payload checksum calculation. + * Offset of place to put the checksum. + */ + TXD->upper_setup.tcp_fields.tucss = pi->ipi_ehdrlen + pi->ipi_ip_hlen; + TXD->upper_setup.tcp_fields.tucse = 0; + TXD->upper_setup.tcp_fields.tucso = + pi->ipi_ehdrlen + pi->ipi_ip_hlen + offsetof(struct tcphdr, th_sum); + + /* + * Payload size per packet w/o any headers. + * Length of all headers up to payload. + */ + TXD->tcp_seg_setup.fields.mss = htole16(pi->ipi_tso_segsz); + TXD->tcp_seg_setup.fields.hdr_len = hdr_len; + + TXD->cmd_and_length = htole32(adapter->txd_cmd | + E1000_TXD_CMD_DEXT | /* Extended descr */ + E1000_TXD_CMD_TSE | /* TSE context */ + E1000_TXD_CMD_IP | /* Do IP csum */ + E1000_TXD_CMD_TCP | /* Do TCP checksum */ + (pi->ipi_len - hdr_len)); /* Total len */ + tx_buffer->eop = -1; + txr->tx_tso = TRUE; + + if (++cur == scctx->isc_ntxd[0]) { + cur = 0; + } + DPRINTF(iflib_get_dev(adapter->ctx), "%s: pidx: %d cur: %d\n", __FUNCTION__, pi->ipi_pidx, cur); + return (cur); +} + +#define TSO_WORKAROUND 4 +#define DONT_FORCE_CTX 1 + + +/********************************************************************* + * The offload context is protocol specific (TCP/UDP) and thus + * only needs to be set when the protocol changes. The occasion + * of a context change can be a performance detriment, and + * might be better just disabled. The reason arises in the way + * in which the controller supports pipelined requests from the + * Tx data DMA. Up to four requests can be pipelined, and they may + * belong to the same packet or to multiple packets. However all + * requests for one packet are issued before a request is issued + * for a subsequent packet and if a request for the next packet + * requires a context change, that request will be stalled + * until the previous request completes. This means setting up + * a new context effectively disables pipelined Tx data DMA which + * in turn greatly slow down performance to send small sized + * frames. + **********************************************************************/ + +static int +em_transmit_checksum_setup(struct adapter *adapter, if_pkt_info_t pi, u32 *txd_upper, u32 *txd_lower) +{ + struct e1000_context_desc *TXD = NULL; + if_softc_ctx_t scctx = adapter->shared; + struct em_tx_queue *que = &adapter->tx_queues[pi->ipi_qsidx]; + struct tx_ring *txr = &que->txr; + struct em_txbuffer *tx_buffer; + int csum_flags = pi->ipi_csum_flags; + int cur, hdr_len; + u32 cmd; + + cur = pi->ipi_pidx; + hdr_len = pi->ipi_ehdrlen + pi->ipi_ip_hlen; + cmd = adapter->txd_cmd; + + /* + * The 82574L can only remember the *last* context used + * regardless of queue that it was use for. We cannot reuse + * contexts on this hardware platform and must generate a new + * context every time. 82574L hardware spec, section 7.2.6, + * second note. + */ + if (DONT_FORCE_CTX && + adapter->tx_num_queues == 1 && + txr->csum_lhlen == pi->ipi_ehdrlen && + txr->csum_iphlen == pi->ipi_ip_hlen && + txr->csum_flags == csum_flags) { + /* + * Same csum offload context as the previous packets; + * just return. + */ + *txd_upper = txr->csum_txd_upper; + *txd_lower = txr->csum_txd_lower; + return (cur); + } + + TXD = (struct e1000_context_desc *)&txr->tx_base[cur]; + if (csum_flags & CSUM_IP) { + *txd_upper |= E1000_TXD_POPTS_IXSM << 8; + /* + * Start offset for header checksum calculation. + * End offset for header checksum calculation. + * Offset of place to put the checksum. + */ + TXD->lower_setup.ip_fields.ipcss = pi->ipi_ehdrlen; + TXD->lower_setup.ip_fields.ipcse = htole16(hdr_len); + TXD->lower_setup.ip_fields.ipcso = pi->ipi_ehdrlen + offsetof(struct ip, ip_sum); + cmd |= E1000_TXD_CMD_IP; + } + + if (csum_flags & (CSUM_TCP|CSUM_UDP)) { + uint8_t tucso; + + *txd_upper |= E1000_TXD_POPTS_TXSM << 8; + *txd_lower = E1000_TXD_CMD_DEXT | E1000_TXD_DTYP_D; + + if (csum_flags & CSUM_TCP) { + tucso = hdr_len + offsetof(struct tcphdr, th_sum); + cmd |= E1000_TXD_CMD_TCP; + } else + tucso = hdr_len + offsetof(struct udphdr, uh_sum); + TXD->upper_setup.tcp_fields.tucss = hdr_len; + TXD->upper_setup.tcp_fields.tucse = htole16(0); + TXD->upper_setup.tcp_fields.tucso = tucso; + } + + txr->csum_lhlen = pi->ipi_ehdrlen; + txr->csum_iphlen = pi->ipi_ip_hlen; + txr->csum_flags = csum_flags; + txr->csum_txd_upper = *txd_upper; + txr->csum_txd_lower = *txd_lower; + + TXD->tcp_seg_setup.data = htole32(0); + TXD->cmd_and_length = + htole32(E1000_TXD_CMD_IFCS | E1000_TXD_CMD_DEXT | cmd); + + tx_buffer = &txr->tx_buffers[cur]; + tx_buffer->eop = -1; + + if (++cur == scctx->isc_ntxd[0]) { + cur = 0; + } + DPRINTF(iflib_get_dev(adapter->ctx), "checksum_setup csum_flags=%x txd_upper=%x txd_lower=%x hdr_len=%d cmd=%x\n", + csum_flags, *txd_upper, *txd_lower, hdr_len, cmd); + return (cur); +} + +static int +em_isc_txd_encap(void *arg, if_pkt_info_t pi) +{ + struct adapter *sc = arg; + if_softc_ctx_t scctx = sc->shared; + struct em_tx_queue *que = &sc->tx_queues[pi->ipi_qsidx]; + struct tx_ring *txr = &que->txr; + bus_dma_segment_t *segs = pi->ipi_segs; + int nsegs = pi->ipi_nsegs; + int csum_flags = pi->ipi_csum_flags; + int i, j, first, pidx_last; + u32 txd_upper = 0, txd_lower = 0; + + struct em_txbuffer *tx_buffer; + struct e1000_tx_desc *ctxd = NULL; + bool do_tso, tso_desc; + + i = first = pi->ipi_pidx; + do_tso = (csum_flags & CSUM_TSO); + tso_desc = FALSE; + /* + * TSO Hardware workaround, if this packet is not + * TSO, and is only a single descriptor long, and + * it follows a TSO burst, then we need to add a + * sentinel descriptor to prevent premature writeback. + */ + if ((!do_tso) && (txr->tx_tso == TRUE)) { + if (nsegs == 1) + tso_desc = TRUE; + txr->tx_tso = FALSE; + } + + /* Do hardware assists */ + if (do_tso) { + i = em_tso_setup(sc, pi, &txd_upper, &txd_lower); + tso_desc = TRUE; + } else if (csum_flags & CSUM_OFFLOAD) { + i = em_transmit_checksum_setup(sc, pi, &txd_upper, &txd_lower); + } + + if (pi->ipi_mflags & M_VLANTAG) { + /* Set the vlan id. */ + txd_upper |= htole16(pi->ipi_vtag) << 16; + /* Tell hardware to add tag */ + txd_lower |= htole32(E1000_TXD_CMD_VLE); + } + + DPRINTF(iflib_get_dev(sc->ctx), "encap: set up tx: nsegs=%d first=%d i=%d\n", nsegs, first, i); + /* XXX adapter->pcix_82544 -- lem_fill_descriptors */ + + /* Set up our transmit descriptors */ + for (j = 0; j < nsegs; j++) { + bus_size_t seg_len; + bus_addr_t seg_addr; + uint32_t cmd; + + ctxd = &txr->tx_base[i]; + tx_buffer = &txr->tx_buffers[i]; + seg_addr = segs[j].ds_addr; + seg_len = segs[j].ds_len; + cmd = E1000_TXD_CMD_IFCS | sc->txd_cmd; + + /* + ** TSO Workaround: + ** If this is the last descriptor, we want to + ** split it so we have a small final sentinel + */ + if (tso_desc && (j == (nsegs - 1)) && (seg_len > 8)) { + seg_len -= TSO_WORKAROUND; + ctxd->buffer_addr = htole64(seg_addr); + ctxd->lower.data = htole32(cmd | txd_lower | seg_len); + ctxd->upper.data = htole32(txd_upper); + + if (++i == scctx->isc_ntxd[0]) + i = 0; + + /* Now make the sentinel */ + ctxd = &txr->tx_base[i]; + tx_buffer = &txr->tx_buffers[i]; + ctxd->buffer_addr = htole64(seg_addr + seg_len); + ctxd->lower.data = htole32(cmd | txd_lower | TSO_WORKAROUND); + ctxd->upper.data = htole32(txd_upper); + pidx_last = i; + if (++i == scctx->isc_ntxd[0]) + i = 0; + DPRINTF(iflib_get_dev(sc->ctx), "TSO path pidx_last=%d i=%d ntxd[0]=%d\n", pidx_last, i, scctx->isc_ntxd[0]); + } else { + ctxd->buffer_addr = htole64(seg_addr); + ctxd->lower.data = htole32(cmd | txd_lower | seg_len); + ctxd->upper.data = htole32(txd_upper); + pidx_last = i; + if (++i == scctx->isc_ntxd[0]) + i = 0; + DPRINTF(iflib_get_dev(sc->ctx), "pidx_last=%d i=%d ntxd[0]=%d\n", pidx_last, i, scctx->isc_ntxd[0]); + } + tx_buffer->eop = -1; + } + + /* + * Last Descriptor of Packet + * needs End Of Packet (EOP) + * and Report Status (RS) + */ + ctxd->lower.data |= + htole32(E1000_TXD_CMD_EOP | E1000_TXD_CMD_RS); + + tx_buffer = &txr->tx_buffers[first]; + tx_buffer->eop = pidx_last; + DPRINTF(iflib_get_dev(sc->ctx), "tx_buffers[%d]->eop = %d ipi_new_pidx=%d\n", first, pidx_last, i); + pi->ipi_new_pidx = i; + + return (0); +} + +static void +em_isc_txd_flush(void *arg, uint16_t txqid, uint32_t pidx) +{ + struct adapter *adapter = arg; + struct em_tx_queue *que = &adapter->tx_queues[txqid]; + struct tx_ring *txr = &que->txr; + + E1000_WRITE_REG(&adapter->hw, E1000_TDT(txr->me), pidx); +} + +static int +em_isc_txd_credits_update(void *arg, uint16_t txqid, uint32_t cidx_init, bool clear) +{ + struct adapter *adapter = arg; + if_softc_ctx_t scctx = adapter->shared; + struct em_tx_queue *que = &adapter->tx_queues[txqid]; + struct tx_ring *txr = &que->txr; + + u32 cidx, processed = 0; + int last, done; + struct em_txbuffer *buf; + struct e1000_tx_desc *tx_desc, *eop_desc; + + cidx = cidx_init; + buf = &txr->tx_buffers[cidx]; + tx_desc = &txr->tx_base[cidx]; + last = buf->eop; + eop_desc = &txr->tx_base[last]; + + DPRINTF(iflib_get_dev(adapter->ctx), "credits_update: cidx_init=%d clear=%d last=%d\n", + cidx_init, clear, last); + /* + * What this does is get the index of the + * first descriptor AFTER the EOP of the + * first packet, that way we can do the + * simple comparison on the inner while loop. + */ + if (++last == scctx->isc_ntxd[0]) + last = 0; + done = last; + + + while (eop_desc->upper.fields.status & E1000_TXD_STAT_DD) { + /* We clean the range of the packet */ + while (cidx != done) { + if (clear) { + tx_desc->upper.data = 0; + tx_desc->lower.data = 0; + tx_desc->buffer_addr = 0; + buf->eop = -1; + } + tx_desc++; + buf++; + processed++; + + /* wrap the ring ? */ + if (++cidx == scctx->isc_ntxd[0]) { + cidx = 0; + } + buf = &txr->tx_buffers[cidx]; + tx_desc = &txr->tx_base[cidx]; + } + /* See if we can continue to the next packet */ + last = buf->eop; + if (last == -1) + break; + eop_desc = &txr->tx_base[last]; + /* Get new done point */ + if (++last == scctx->isc_ntxd[0]) + last = 0; + done = last; + } + + DPRINTF(iflib_get_dev(adapter->ctx), "Processed %d credits update\n", processed); + return(processed); +} + +static void +lem_isc_rxd_refill(void *arg, uint16_t rxqid, uint8_t flid __unused, + uint32_t pidx, uint64_t *paddrs, caddr_t *vaddrs __unused, uint16_t count, uint16_t buflen __unused) +{ + struct adapter *sc = arg; + if_softc_ctx_t scctx = sc->shared; + struct em_rx_queue *que = &sc->rx_queues[rxqid]; + struct rx_ring *rxr = &que->rxr; + struct e1000_rx_desc *rxd; + int i; + uint32_t next_pidx; + + for (i = 0, next_pidx = pidx; i < count; i++) { + rxd = (struct e1000_rx_desc *)&rxr->rx_base[next_pidx]; + rxd->buffer_addr = htole64(paddrs[i]); + /* status bits must be cleared */ + rxd->status = 0; + + if (++next_pidx == scctx->isc_nrxd[0]) + next_pidx = 0; + } +} + +static void +em_isc_rxd_refill(void *arg, uint16_t rxqid, uint8_t flid __unused, + uint32_t pidx, uint64_t *paddrs, caddr_t *vaddrs __unused, uint16_t count, uint16_t buflen __unused) +{ + struct adapter *sc = arg; + if_softc_ctx_t scctx = sc->shared; + struct em_rx_queue *que = &sc->rx_queues[rxqid]; + struct rx_ring *rxr = &que->rxr; + union e1000_rx_desc_extended *rxd; + int i; + uint32_t next_pidx; + + for (i = 0, next_pidx = pidx; i < count; i++) { + rxd = &rxr->rx_base[next_pidx]; + rxd->read.buffer_addr = htole64(paddrs[i]); + /* DD bits must be cleared */ + rxd->wb.upper.status_error = 0; + + if (++next_pidx == scctx->isc_nrxd[0]) + next_pidx = 0; + } +} + +static void +em_isc_rxd_flush(void *arg, uint16_t rxqid, uint8_t flid __unused, uint32_t pidx) +{ + struct adapter *sc = arg; + struct em_rx_queue *que = &sc->rx_queues[rxqid]; + struct rx_ring *rxr = &que->rxr; + + E1000_WRITE_REG(&sc->hw, E1000_RDT(rxr->me), pidx); +} + +static int +lem_isc_rxd_available(void *arg, uint16_t rxqid, uint32_t idx, int budget) +{ + struct adapter *sc = arg; + if_softc_ctx_t scctx = sc->shared; + struct em_rx_queue *que = &sc->rx_queues[rxqid]; + struct rx_ring *rxr = &que->rxr; + struct e1000_rx_desc *rxd; + u32 staterr = 0; + int cnt, i; + + for (cnt = 0, i = idx; cnt < scctx->isc_nrxd[0] && cnt <= budget;) { + rxd = (struct e1000_rx_desc *)&rxr->rx_base[i]; + staterr = rxd->status; + + if ((staterr & E1000_RXD_STAT_DD) == 0) + break; + + if (++i == scctx->isc_nrxd[0]) + i = 0; + + if (staterr & E1000_RXD_STAT_EOP) + cnt++; + } + return (cnt); +} + +static int +em_isc_rxd_available(void *arg, uint16_t rxqid, uint32_t idx, int budget) +{ + struct adapter *sc = arg; + if_softc_ctx_t scctx = sc->shared; + struct em_rx_queue *que = &sc->rx_queues[rxqid]; + struct rx_ring *rxr = &que->rxr; + union e1000_rx_desc_extended *rxd; + u32 staterr = 0; + int cnt, i; + + for (cnt = 0, i = idx; cnt < scctx->isc_nrxd[0] && cnt <= budget;) { + rxd = &rxr->rx_base[i]; + staterr = le32toh(rxd->wb.upper.status_error); + + if ((staterr & E1000_RXD_STAT_DD) == 0) + break; + + if (++i == scctx->isc_nrxd[0]) { + i = 0; + } + + if (staterr & E1000_RXD_STAT_EOP) + cnt++; + + } + return (cnt); +} + +static int +lem_isc_rxd_pkt_get(void *arg, if_rxd_info_t ri) +{ + struct adapter *adapter = arg; + if_softc_ctx_t scctx = adapter->shared; + struct em_rx_queue *que = &adapter->rx_queues[ri->iri_qsidx]; + struct rx_ring *rxr = &que->rxr; + struct e1000_rx_desc *rxd; + u16 len; + u32 status, errors; + bool eop; + int i, cidx; + + status = errors = i = 0; + cidx = ri->iri_cidx; + + do { + rxd = (struct e1000_rx_desc *)&rxr->rx_base[cidx]; + status = rxd->status; + errors = rxd->errors; + + /* Error Checking then decrement count */ + MPASS ((status & E1000_RXD_STAT_DD) != 0); + + len = le16toh(rxd->length); + ri->iri_len += len; + + eop = (status & E1000_RXD_STAT_EOP) != 0; + + /* Make sure bad packets are discarded */ + if (errors & E1000_RXD_ERR_FRAME_ERR_MASK) { + adapter->dropped_pkts++; + /* XXX fixup if common */ + return (EBADMSG); + } + + ri->iri_frags[i].irf_flid = 0; + ri->iri_frags[i].irf_idx = cidx; + ri->iri_frags[i].irf_len = len; + /* Zero out the receive descriptors status. */ + rxd->status = 0; + + if (++cidx == scctx->isc_nrxd[0]) + cidx = 0; + i++; + } while (!eop); + + /* XXX add a faster way to look this up */ + if (adapter->hw.mac.type >= e1000_82543 && !(status & E1000_RXD_STAT_IXSM)) + lem_receive_checksum(status, errors, ri); + + if (status & E1000_RXD_STAT_VP) { + ri->iri_vtag = le16toh(rxd->special); + ri->iri_flags |= M_VLANTAG; + } + + ri->iri_nfrags = i; + + return (0); +} + +static int +em_isc_rxd_pkt_get(void *arg, if_rxd_info_t ri) +{ + struct adapter *adapter = arg; + if_softc_ctx_t scctx = adapter->shared; + struct em_rx_queue *que = &adapter->rx_queues[ri->iri_qsidx]; + struct rx_ring *rxr = &que->rxr; + union e1000_rx_desc_extended *rxd; + + u16 len; + u32 staterr = 0; + bool eop; + int i, cidx, vtag; + + i = vtag = 0; + cidx = ri->iri_cidx; + + do { + rxd = &rxr->rx_base[cidx]; + staterr = le32toh(rxd->wb.upper.status_error); + + /* Error Checking then decrement count */ + MPASS ((staterr & E1000_RXD_STAT_DD) != 0); + + len = le16toh(rxd->wb.upper.length); + ri->iri_len += len; + + eop = (staterr & E1000_RXD_STAT_EOP) != 0; + + /* Make sure bad packets are discarded */ + if (staterr & E1000_RXDEXT_ERR_FRAME_ERR_MASK) { + adapter->dropped_pkts++; + return EBADMSG; + } + + ri->iri_frags[i].irf_flid = 0; + ri->iri_frags[i].irf_idx = cidx; + ri->iri_frags[i].irf_len = len; + /* Zero out the receive descriptors status. */ + rxd->wb.upper.status_error &= htole32(~0xFF); + + if (++cidx == scctx->isc_nrxd[0]) + cidx = 0; + i++; + } while (!eop); + + /* XXX add a faster way to look this up */ + if (adapter->hw.mac.type >= e1000_82543) + em_receive_checksum(staterr, ri); + + if (staterr & E1000_RXD_STAT_VP) { + vtag = le16toh(rxd->wb.upper.vlan); + } + + ri->iri_vtag = vtag; + ri->iri_nfrags = i; + if (vtag) + ri->iri_flags |= M_VLANTAG; + + return (0); +} + +/********************************************************************* + * + * Verify that the hardware indicated that the checksum is valid. + * Inform the stack about the status of checksum so that stack + * doesn't spend time verifying the checksum. + * + *********************************************************************/ +static void +lem_receive_checksum(int status, int errors, if_rxd_info_t ri) +{ + /* Did it pass? */ + if (status & E1000_RXD_STAT_IPCS && !(errors & E1000_RXD_ERR_IPE)) + ri->iri_csum_flags = (CSUM_IP_CHECKED|CSUM_IP_VALID); + + if (status & E1000_RXD_STAT_TCPCS) { + /* Did it pass? */ + if (!(errors & E1000_RXD_ERR_TCPE)) { + ri->iri_csum_flags |= + (CSUM_DATA_VALID | CSUM_PSEUDO_HDR); + ri->iri_csum_data = htons(0xffff); + } + } +} + +static void +em_receive_checksum(uint32_t status, if_rxd_info_t ri) +{ + ri->iri_csum_flags = 0; + + /* Ignore Checksum bit is set */ + if (status & E1000_RXD_STAT_IXSM) + return; + + /* If the IP checksum exists and there is no IP Checksum error */ + if ((status & (E1000_RXD_STAT_IPCS | E1000_RXDEXT_STATERR_IPE)) == + E1000_RXD_STAT_IPCS) { + ri->iri_csum_flags = (CSUM_IP_CHECKED | CSUM_IP_VALID); + } + + /* TCP or UDP checksum */ + if ((status & (E1000_RXD_STAT_TCPCS | E1000_RXDEXT_STATERR_TCPE)) == + E1000_RXD_STAT_TCPCS) { + ri->iri_csum_flags |= (CSUM_DATA_VALID | CSUM_PSEUDO_HDR); + ri->iri_csum_data = htons(0xffff); + } + if (status & E1000_RXD_STAT_UDPCS) { + ri->iri_csum_flags |= (CSUM_DATA_VALID | CSUM_PSEUDO_HDR); + ri->iri_csum_data = htons(0xffff); + } +} diff --git a/sys/dev/e1000/if_em.c b/sys/dev/e1000/if_em.c index b1bf35b5e464..a49f578682c4 100644 --- a/sys/dev/e1000/if_em.c +++ b/sys/dev/e1000/if_em.c @@ -1,99 +1,10 @@ -/****************************************************************************** - - Copyright (c) 2001-2015, Intel Corporation - All rights reserved. - - Redistribution and use in source and binary forms, with or without - modification, are permitted provided that the following conditions are met: - - 1. Redistributions of source code must retain the above copyright notice, - this list of conditions and the following disclaimer. - - 2. Redistributions in binary form must reproduce the above copyright - notice, this list of conditions and the following disclaimer in the - documentation and/or other materials provided with the distribution. - - 3. Neither the name of the Intel Corporation nor the names of its - contributors may be used to endorse or promote products derived from - this software without specific prior written permission. - - THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" - AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE - LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR - CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF - SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS - INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN - CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) - ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE - POSSIBILITY OF SUCH DAMAGE. - -******************************************************************************/ -/*$FreeBSD$*/ - -#include "opt_em.h" -#include "opt_ddb.h" -#include "opt_inet.h" -#include "opt_inet6.h" - -#ifdef HAVE_KERNEL_OPTION_HEADERS -#include "opt_device_polling.h" -#endif - -#include -#include -#ifdef DDB -#include -#include -#endif -#if __FreeBSD_version >= 800000 -#include -#endif -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include -#include -#include -#include -#include - -#include -#include - -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include -#include - -#include "e1000_api.h" -#include "e1000_82571.h" +/* $FreeBSD$ */ #include "if_em.h" +#include +#include + +#define em_mac_min e1000_82547 +#define igb_mac_min e1000_82575 /********************************************************************* * Driver version: @@ -110,184 +21,213 @@ char em_driver_version[] = "7.6.1-k"; * { Vendor ID, Device ID, SubVendor ID, SubDevice ID, String Index } *********************************************************************/ -static em_vendor_info_t em_vendor_info_array[] = +static pci_vendor_info_t em_vendor_info_array[] = { - /* Intel(R) PRO/1000 Network Connection */ - { 0x8086, E1000_DEV_ID_82571EB_COPPER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82571EB_FIBER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82571EB_SERDES, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82571EB_SERDES_DUAL, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82571EB_SERDES_QUAD, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82571EB_QUAD_COPPER, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82571EB_QUAD_COPPER_LP, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82571EB_QUAD_FIBER, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82571PT_QUAD_COPPER, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82572EI_COPPER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82572EI_FIBER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82572EI_SERDES, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82572EI, PCI_ANY_ID, PCI_ANY_ID, 0}, + /* Intel(R) PRO/1000 Network Connection - Legacy em*/ + PVID(0x8086, E1000_DEV_ID_82540EM, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82540EM_LOM, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82540EP, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82540EP_LOM, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82540EP_LP, "Intel(R) PRO/1000 Network Connection"), - { 0x8086, E1000_DEV_ID_82573E, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82573E_IAMT, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82573L, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82583V, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_80003ES2LAN_COPPER_SPT, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_80003ES2LAN_SERDES_SPT, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_80003ES2LAN_COPPER_DPT, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_80003ES2LAN_SERDES_DPT, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH8_IGP_M_AMT, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH8_IGP_AMT, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH8_IGP_C, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH8_IFE, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH8_IFE_GT, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH8_IFE_G, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH8_IGP_M, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH8_82567V_3, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH9_IGP_M_AMT, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH9_IGP_AMT, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH9_IGP_C, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH9_IGP_M, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH9_IGP_M_V, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH9_IFE, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH9_IFE_GT, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH9_IFE_G, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH9_BM, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82574L, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82574LA, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH10_R_BM_LM, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH10_R_BM_LF, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH10_R_BM_V, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH10_D_BM_LM, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH10_D_BM_LF, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_ICH10_D_BM_V, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_M_HV_LM, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_M_HV_LC, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_D_HV_DM, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_D_HV_DC, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH2_LV_LM, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH2_LV_V, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_LPT_I217_LM, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_LPT_I217_V, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_LPTLP_I218_LM, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_LPTLP_I218_V, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_I218_LM2, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_I218_V2, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_I218_LM3, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_I218_V3, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_SPT_I219_LM, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_SPT_I219_V, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_SPT_I219_LM2, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_SPT_I219_V2, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_LBG_I219_LM3, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_SPT_I219_LM4, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_SPT_I219_V4, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_SPT_I219_LM5, - PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_PCH_SPT_I219_V5, PCI_ANY_ID, PCI_ANY_ID, 0}, + PVID(0x8086, E1000_DEV_ID_82541EI, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82541ER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82541ER_LOM, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82541EI_MOBILE, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82541GI, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82541GI_LF, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82541GI_MOBILE, "Intel(R) PRO/1000 Network Connection"), + + PVID(0x8086, E1000_DEV_ID_82542, "Intel(R) PRO/1000 Network Connection"), + + PVID(0x8086, E1000_DEV_ID_82543GC_FIBER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82543GC_COPPER, "Intel(R) PRO/1000 Network Connection"), + + PVID(0x8086, E1000_DEV_ID_82544EI_COPPER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82544EI_FIBER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82544GC_COPPER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82544GC_LOM, "Intel(R) PRO/1000 Network Connection"), + + PVID(0x8086, E1000_DEV_ID_82545EM_COPPER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82545EM_FIBER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82545GM_COPPER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82545GM_FIBER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82545GM_SERDES, "Intel(R) PRO/1000 Network Connection"), + + PVID(0x8086, E1000_DEV_ID_82546EB_COPPER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82546EB_FIBER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82546EB_QUAD_COPPER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82546GB_COPPER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82546GB_FIBER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82546GB_SERDES, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82546GB_PCIE, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82546GB_QUAD_COPPER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82546GB_QUAD_COPPER_KSP3, "Intel(R) PRO/1000 Network Connection"), + + PVID(0x8086, E1000_DEV_ID_82547EI, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82547EI_MOBILE, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82547GI, "Intel(R) PRO/1000 Network Connection"), + + /* Intel(R) PRO/1000 Network Connection - em */ + PVID(0x8086, E1000_DEV_ID_82571EB_COPPER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82571EB_FIBER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82571EB_SERDES, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82571EB_SERDES_DUAL, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82571EB_SERDES_QUAD, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82571EB_QUAD_COPPER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82571EB_QUAD_COPPER_LP, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82571EB_QUAD_FIBER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82571PT_QUAD_COPPER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82572EI_COPPER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82572EI_FIBER, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82572EI_SERDES, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82573E, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82573E_IAMT, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82573L, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82583V, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_80003ES2LAN_COPPER_SPT, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_80003ES2LAN_SERDES_SPT, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_80003ES2LAN_COPPER_DPT, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_80003ES2LAN_SERDES_DPT, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH8_IGP_M_AMT, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH8_IGP_AMT, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH8_IGP_C, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH8_IFE, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH8_IFE_GT, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH8_IFE_G, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH8_IGP_M, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH8_82567V_3, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH9_IGP_M_AMT, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH9_IGP_AMT, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH9_IGP_C, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH9_IGP_M, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH9_IGP_M_V, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH9_IFE, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH9_IFE_GT, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH9_IFE_G, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH9_BM, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82574L, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_82574LA, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH10_R_BM_LM, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH10_R_BM_LF, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH10_R_BM_V, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH10_D_BM_LM, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH10_D_BM_LF, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_ICH10_D_BM_V, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_M_HV_LM, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_M_HV_LC, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_D_HV_DM, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_D_HV_DC, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH2_LV_LM, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH2_LV_V, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_LPT_I217_LM, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_LPT_I217_V, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_LPTLP_I218_LM, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_LPTLP_I218_V, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_I218_LM2, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_I218_V2, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_I218_LM3, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_I218_V3, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_SPT_I219_LM, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_SPT_I219_V, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_SPT_I219_LM2, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_SPT_I219_V2, "Intel(R) PRO/1000 Network Connection"), + PVID(0x8086, E1000_DEV_ID_PCH_LBG_I219_LM3, "Intel(R) PRO/1000 Network Connection"), /* required last entry */ - { 0, 0, 0, 0, 0} + PVID_END }; -/********************************************************************* - * Table of branding strings for all supported NICs. - *********************************************************************/ - -static char *em_strings[] = { - "Intel(R) PRO/1000 Network Connection" +static pci_vendor_info_t igb_vendor_info_array[] = +{ + /* Intel(R) PRO/1000 Network Connection - em */ + PVID(0x8086, E1000_DEV_ID_82575EB_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82575EB_FIBER_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82575GB_QUAD_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82576, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82576_NS, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82576_NS_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82576_FIBER, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82576_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82576_SERDES_QUAD, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82576_QUAD_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82576_QUAD_COPPER_ET2, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82576_VF, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82580_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82580_FIBER, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82580_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82580_SGMII, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82580_COPPER_DUAL, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_82580_QUAD_FIBER, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_DH89XXCC_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_DH89XXCC_SGMII, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_DH89XXCC_SFP, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_DH89XXCC_BACKPLANE, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I350_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I350_FIBER, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I350_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I350_SGMII, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I350_VF, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I210_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I210_COPPER_IT, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I210_COPPER_OEM1, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I210_COPPER_FLASHLESS, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I210_SERDES_FLASHLESS, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I210_FIBER, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I210_SERDES, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I210_SGMII, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I211_COPPER, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I354_BACKPLANE_1GBPS, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I354_BACKPLANE_2_5GBPS, "Intel(R) PRO/1000 PCI-Express Network Driver"), + PVID(0x8086, E1000_DEV_ID_I354_SGMII, "Intel(R) PRO/1000 PCI-Express Network Driver"), + /* required last entry */ + PVID_END }; /********************************************************************* * Function prototypes *********************************************************************/ -static int em_probe(device_t); -static int em_attach(device_t); -static int em_detach(device_t); -static int em_shutdown(device_t); -static int em_suspend(device_t); -static int em_resume(device_t); -#ifdef EM_MULTIQUEUE -static int em_mq_start(if_t, struct mbuf *); -static int em_mq_start_locked(if_t, - struct tx_ring *); -static void em_qflush(if_t); -#else -static void em_start(if_t); -static void em_start_locked(if_t, struct tx_ring *); -#endif -static int em_ioctl(if_t, u_long, caddr_t); -static uint64_t em_get_counter(if_t, ift_counter); -static void em_init(void *); -static void em_init_locked(struct adapter *); -static void em_stop(void *); -static void em_media_status(if_t, struct ifmediareq *); -static int em_media_change(if_t); -static void em_identify_hardware(struct adapter *); -static int em_allocate_pci_resources(struct adapter *); -static int em_allocate_legacy(struct adapter *); -static int em_allocate_msix(struct adapter *); -static int em_allocate_queues(struct adapter *); -static int em_setup_msix(struct adapter *); -static void em_free_pci_resources(struct adapter *); -static void em_local_timer(void *); -static void em_reset(struct adapter *); -static int em_setup_interface(device_t, struct adapter *); -static void em_flush_desc_rings(struct adapter *); +static void *em_register(device_t dev); +static void *igb_register(device_t dev); +static int em_if_attach_pre(if_ctx_t ctx); +static int em_if_attach_post(if_ctx_t ctx); +static int em_if_detach(if_ctx_t ctx); +static int em_if_shutdown(if_ctx_t ctx); +static int em_if_suspend(if_ctx_t ctx); +static int em_if_resume(if_ctx_t ctx); -static void em_setup_transmit_structures(struct adapter *); -static void em_initialize_transmit_unit(struct adapter *); -static int em_allocate_transmit_buffers(struct tx_ring *); -static void em_free_transmit_structures(struct adapter *); -static void em_free_transmit_buffers(struct tx_ring *); +static int em_if_tx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int ntxqs, int ntxqsets); +static int em_if_rx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int nrxqs, int nrxqsets); +static void em_if_queues_free(if_ctx_t ctx); -static int em_setup_receive_structures(struct adapter *); -static int em_allocate_receive_buffers(struct rx_ring *); -static void em_initialize_receive_unit(struct adapter *); -static void em_free_receive_structures(struct adapter *); -static void em_free_receive_buffers(struct rx_ring *); +static uint64_t em_if_get_counter(if_ctx_t, ift_counter); +static void em_if_init(if_ctx_t ctx); +static void em_if_stop(if_ctx_t ctx); +static void em_if_media_status(if_ctx_t, struct ifmediareq *); +static int em_if_media_change(if_ctx_t ctx); +static int em_if_mtu_set(if_ctx_t ctx, uint32_t mtu); +static void em_if_timer(if_ctx_t ctx, uint16_t qid); +static void em_if_vlan_register(if_ctx_t ctx, u16 vtag); +static void em_if_vlan_unregister(if_ctx_t ctx, u16 vtag); -static void em_enable_intr(struct adapter *); -static void em_disable_intr(struct adapter *); +static void em_identify_hardware(if_ctx_t ctx); +static int em_allocate_pci_resources(if_ctx_t ctx); +static void em_free_pci_resources(if_ctx_t ctx); +static void em_reset(if_ctx_t ctx); +static int em_setup_interface(if_ctx_t ctx); +static int em_setup_msix(if_ctx_t ctx); + +static void em_initialize_transmit_unit(if_ctx_t ctx); +static void em_initialize_receive_unit(if_ctx_t ctx); + +static void em_if_enable_intr(if_ctx_t ctx); +static void em_if_disable_intr(if_ctx_t ctx); +static int em_if_queue_intr_enable(if_ctx_t ctx, uint16_t rxqid); +static void em_if_multi_set(if_ctx_t ctx); +static void em_if_update_admin_status(if_ctx_t ctx); static void em_update_stats_counters(struct adapter *); static void em_add_hw_stats(struct adapter *adapter); -static void em_txeof(struct tx_ring *); -static bool em_rxeof(struct rx_ring *, int, int *); -#ifndef __NO_STRICT_ALIGNMENT -static int em_fixup_rx(struct rx_ring *); -#endif -static void em_setup_rxdesc(union e1000_rx_desc_extended *, - const struct em_rxbuffer *rxbuf); -static void em_receive_checksum(uint32_t status, struct mbuf *); -static void em_transmit_checksum_setup(struct tx_ring *, struct mbuf *, int, - struct ip *, u32 *, u32 *); -static void em_tso_setup(struct tx_ring *, struct mbuf *, int, struct ip *, - struct tcphdr *, u32 *, u32 *); -static void em_set_promisc(struct adapter *); -static void em_disable_promisc(struct adapter *); -static void em_set_multi(struct adapter *); -static void em_update_link_status(struct adapter *); -static void em_refresh_mbufs(struct rx_ring *, int); -static void em_register_vlan(void *, if_t, u16); -static void em_unregister_vlan(void *, if_t, u16); +static int em_if_set_promisc(if_ctx_t ctx, int flags); static void em_setup_vlan_hw_support(struct adapter *); -static int em_xmit(struct tx_ring *, struct mbuf **); -static int em_dma_malloc(struct adapter *, bus_size_t, - struct em_dma_alloc *, int); -static void em_dma_free(struct adapter *, struct em_dma_alloc *); static int em_sysctl_nvm_info(SYSCTL_HANDLER_ARGS); static void em_print_nvm_info(struct adapter *); static int em_sysctl_debug_info(SYSCTL_HANDLER_ARGS); @@ -301,63 +241,122 @@ static void em_init_manageability(struct adapter *); static void em_release_manageability(struct adapter *); static void em_get_hw_control(struct adapter *); static void em_release_hw_control(struct adapter *); -static void em_get_wakeup(device_t); -static void em_enable_wakeup(device_t); +static void em_get_wakeup(if_ctx_t ctx); +static void em_enable_wakeup(if_ctx_t ctx); static int em_enable_phy_wakeup(struct adapter *); -static void em_led_func(void *, int); static void em_disable_aspm(struct adapter *); -static int em_irq_fast(void *); +int em_intr(void *arg); +static void em_disable_promisc(if_ctx_t ctx); /* MSIX handlers */ -static void em_msix_tx(void *); -static void em_msix_rx(void *); -static void em_msix_link(void *); -static void em_handle_tx(void *context, int pending); -static void em_handle_rx(void *context, int pending); -static void em_handle_link(void *context, int pending); +static int em_if_msix_intr_assign(if_ctx_t, int); +static int em_msix_link(void *); +static void em_handle_link(void *context); -#ifdef EM_MULTIQUEUE -static void em_enable_vectors_82574(struct adapter *); -#endif +static void em_enable_vectors_82574(if_ctx_t); static void em_set_sysctl_value(struct adapter *, const char *, const char *, int *, int); static int em_set_flowcntl(SYSCTL_HANDLER_ARGS); static int em_sysctl_eee(SYSCTL_HANDLER_ARGS); +static void em_if_led_func(if_ctx_t ctx, int onoff); -static __inline void em_rx_discard(struct rx_ring *, int); +static void em_init_tx_ring(struct em_tx_queue *que); +static int em_get_regs(SYSCTL_HANDLER_ARGS); + +static void lem_smartspeed(struct adapter *adapter); +static void igb_configure_queues(struct adapter *adapter); -#ifdef DEVICE_POLLING -static poll_handler_t em_poll; -#endif /* POLLING */ /********************************************************************* * FreeBSD Device Interface Entry Points *********************************************************************/ - static device_method_t em_methods[] = { /* Device interface */ - DEVMETHOD(device_probe, em_probe), - DEVMETHOD(device_attach, em_attach), - DEVMETHOD(device_detach, em_detach), - DEVMETHOD(device_shutdown, em_shutdown), - DEVMETHOD(device_suspend, em_suspend), - DEVMETHOD(device_resume, em_resume), - DEVMETHOD_END + DEVMETHOD(device_register, em_register), + DEVMETHOD(device_probe, iflib_device_probe), + DEVMETHOD(device_attach, iflib_device_attach), + DEVMETHOD(device_detach, iflib_device_detach), + DEVMETHOD(device_shutdown, iflib_device_shutdown), + DEVMETHOD(device_suspend, iflib_device_suspend), + DEVMETHOD(device_resume, iflib_device_resume), + DEVMETHOD_END }; +static device_method_t igb_methods[] = { + /* Device interface */ + DEVMETHOD(device_register, igb_register), + DEVMETHOD(device_probe, iflib_device_probe), + DEVMETHOD(device_attach, iflib_device_attach), + DEVMETHOD(device_detach, iflib_device_detach), + DEVMETHOD(device_shutdown, iflib_device_shutdown), + DEVMETHOD(device_suspend, iflib_device_suspend), + DEVMETHOD(device_resume, iflib_device_resume), + DEVMETHOD_END +}; + + static driver_t em_driver = { "em", em_methods, sizeof(struct adapter), }; -devclass_t em_devclass; +static devclass_t em_devclass; DRIVER_MODULE(em, pci, em_driver, em_devclass, 0, 0); + MODULE_DEPEND(em, pci, 1, 1, 1); MODULE_DEPEND(em, ether, 1, 1, 1); -#ifdef DEV_NETMAP -MODULE_DEPEND(em, netmap, 1, 1, 1); -#endif /* DEV_NETMAP */ +MODULE_DEPEND(em, iflib, 1, 1, 1); + +static driver_t igb_driver = { + "igb", igb_methods, sizeof(struct adapter), +}; + +static devclass_t igb_devclass; +DRIVER_MODULE(igb, pci, igb_driver, igb_devclass, 0, 0); + +MODULE_DEPEND(igb, pci, 1, 1, 1); +MODULE_DEPEND(igb, ether, 1, 1, 1); +MODULE_DEPEND(igb, iflib, 1, 1, 1); + + +static device_method_t em_if_methods[] = { + DEVMETHOD(ifdi_attach_pre, em_if_attach_pre), + DEVMETHOD(ifdi_attach_post, em_if_attach_post), + DEVMETHOD(ifdi_detach, em_if_detach), + DEVMETHOD(ifdi_shutdown, em_if_shutdown), + DEVMETHOD(ifdi_suspend, em_if_suspend), + DEVMETHOD(ifdi_resume, em_if_resume), + DEVMETHOD(ifdi_init, em_if_init), + DEVMETHOD(ifdi_stop, em_if_stop), + DEVMETHOD(ifdi_msix_intr_assign, em_if_msix_intr_assign), + DEVMETHOD(ifdi_intr_enable, em_if_enable_intr), + DEVMETHOD(ifdi_intr_disable, em_if_disable_intr), + DEVMETHOD(ifdi_tx_queues_alloc, em_if_tx_queues_alloc), + DEVMETHOD(ifdi_rx_queues_alloc, em_if_rx_queues_alloc), + DEVMETHOD(ifdi_queues_free, em_if_queues_free), + DEVMETHOD(ifdi_update_admin_status, em_if_update_admin_status), + DEVMETHOD(ifdi_multi_set, em_if_multi_set), + DEVMETHOD(ifdi_media_status, em_if_media_status), + DEVMETHOD(ifdi_media_change, em_if_media_change), + DEVMETHOD(ifdi_mtu_set, em_if_mtu_set), + DEVMETHOD(ifdi_promisc_set, em_if_set_promisc), + DEVMETHOD(ifdi_timer, em_if_timer), + DEVMETHOD(ifdi_vlan_register, em_if_vlan_register), + DEVMETHOD(ifdi_vlan_unregister, em_if_vlan_unregister), + DEVMETHOD(ifdi_get_counter, em_if_get_counter), + DEVMETHOD(ifdi_led_func, em_if_led_func), + DEVMETHOD(ifdi_queue_intr_enable, em_if_queue_intr_enable), + DEVMETHOD_END +}; + + /* + * note that if (adapter->msix_mem) is replaced by: + * if (adapter->intr_type == IFLIB_INTR_MSIX) + */ +static driver_t em_if_driver = { + "em_if", em_if_methods, sizeof(struct adapter) +}; /********************************************************************* * Tunable default values. @@ -365,10 +364,16 @@ MODULE_DEPEND(em, netmap, 1, 1, 1); #define EM_TICKS_TO_USECS(ticks) ((1024 * (ticks) + 500) / 1000) #define EM_USECS_TO_TICKS(usecs) ((1000 * (usecs) + 512) / 1024) +#define M_TSO_LEN 66 #define MAX_INTS_PER_SEC 8000 #define DEFAULT_ITR (1000000000/(MAX_INTS_PER_SEC * 256)) +/* Allow common code without TSO */ +#ifndef CSUM_TSO +#define CSUM_TSO 0 +#endif + #define TSO_WORKAROUND 4 static SYSCTL_NODE(_hw, OID_AUTO, em, CTLFLAG_RD, 0, "EM driver parameters"); @@ -393,39 +398,15 @@ SYSCTL_INT(_hw_em, OID_AUTO, rx_abs_int_delay, CTLFLAG_RDTUN, &em_rx_abs_int_delay_dflt, 0, "Default receive interrupt delay limit in usecs"); -static int em_rxd = EM_DEFAULT_RXD; -static int em_txd = EM_DEFAULT_TXD; -SYSCTL_INT(_hw_em, OID_AUTO, rxd, CTLFLAG_RDTUN, &em_rxd, 0, - "Number of receive descriptors per queue"); -SYSCTL_INT(_hw_em, OID_AUTO, txd, CTLFLAG_RDTUN, &em_txd, 0, - "Number of transmit descriptors per queue"); - static int em_smart_pwr_down = FALSE; SYSCTL_INT(_hw_em, OID_AUTO, smart_pwr_down, CTLFLAG_RDTUN, &em_smart_pwr_down, 0, "Set to true to leave smart power down enabled on newer adapters"); /* Controls whether promiscuous also shows bad packets */ -static int em_debug_sbp = FALSE; +static int em_debug_sbp = TRUE; SYSCTL_INT(_hw_em, OID_AUTO, sbp, CTLFLAG_RDTUN, &em_debug_sbp, 0, "Show bad packets in promiscuous mode"); -static int em_enable_msix = TRUE; -SYSCTL_INT(_hw_em, OID_AUTO, enable_msix, CTLFLAG_RDTUN, &em_enable_msix, 0, - "Enable MSI-X interrupts"); - -#ifdef EM_MULTIQUEUE -static int em_num_queues = 1; -SYSCTL_INT(_hw_em, OID_AUTO, num_queues, CTLFLAG_RDTUN, &em_num_queues, 0, - "82574 only: Number of queues to configure, 0 indicates autoconfigure"); -#endif - -/* -** Global variable to store last used CPU when binding queues -** to CPUs in igb_allocate_msix. Starts at CPU_FIRST and increments when a -** queue is bound to a cpu. -*/ -static int em_last_bind_cpu = -1; - /* How many packets rxeof tries to clean at a time */ static int em_rx_process_limit = 100; SYSCTL_INT(_hw_em, OID_AUTO, rx_process_limit, CTLFLAG_RDTUN, @@ -438,64 +419,258 @@ static int eee_setting = 1; SYSCTL_INT(_hw_em, OID_AUTO, eee_setting, CTLFLAG_RDTUN, &eee_setting, 0, "Enable Energy Efficient Ethernet"); +/* +** Tuneable Interrupt rate +*/ +static int em_max_interrupt_rate = 8000; +SYSCTL_INT(_hw_em, OID_AUTO, max_interrupt_rate, CTLFLAG_RDTUN, + &em_max_interrupt_rate, 0, "Maximum interrupts per second"); + + + /* Global used in WOL setup with multiport cards */ static int global_quad_port_a = 0; -#ifdef DEV_NETMAP /* see ixgbe.c for details */ -#include -#endif /* DEV_NETMAP */ +extern struct if_txrx igb_txrx; +extern struct if_txrx em_txrx; +extern struct if_txrx lem_txrx; -/********************************************************************* - * Device identification routine - * - * em_probe determines if the driver should be loaded on - * adapter based on PCI vendor/device id of the adapter. - * - * return BUS_PROBE_DEFAULT on success, positive on failure - *********************************************************************/ +static struct if_shared_ctx em_sctx_init = { + .isc_magic = IFLIB_MAGIC, + .isc_q_align = PAGE_SIZE, + .isc_tx_maxsize = EM_TSO_SIZE, + .isc_tx_maxsegsize = PAGE_SIZE, + .isc_rx_maxsize = MJUM9BYTES, + .isc_rx_nsegments = 1, + .isc_rx_maxsegsize = MJUM9BYTES, + .isc_nfl = 1, + .isc_nrxqs = 1, + .isc_ntxqs = 1, + .isc_admin_intrcnt = 1, + .isc_vendor_info = em_vendor_info_array, + .isc_driver_version = em_driver_version, + .isc_driver = &em_if_driver, + .isc_flags = IFLIB_NEED_SCRATCH | IFLIB_TSO_INIT_IP, -static int -em_probe(device_t dev) + .isc_nrxd_min = {EM_MIN_RXD}, + .isc_ntxd_min = {EM_MIN_TXD}, + .isc_nrxd_max = {EM_MAX_RXD}, + .isc_ntxd_max = {EM_MAX_TXD}, + .isc_nrxd_default = {EM_DEFAULT_RXD}, + .isc_ntxd_default = {EM_DEFAULT_TXD}, +}; + +if_shared_ctx_t em_sctx = &em_sctx_init; + + +static struct if_shared_ctx igb_sctx_init = { + .isc_magic = IFLIB_MAGIC, + .isc_q_align = PAGE_SIZE, + .isc_tx_maxsize = EM_TSO_SIZE, + .isc_tx_maxsegsize = PAGE_SIZE, + .isc_rx_maxsize = MJUM9BYTES, + .isc_rx_nsegments = 1, + .isc_rx_maxsegsize = MJUM9BYTES, + .isc_nfl = 1, + .isc_nrxqs = 1, + .isc_ntxqs = 1, + .isc_admin_intrcnt = 1, + .isc_vendor_info = igb_vendor_info_array, + .isc_driver_version = em_driver_version, + .isc_driver = &em_if_driver, + .isc_flags = IFLIB_NEED_SCRATCH | IFLIB_TSO_INIT_IP, + + .isc_nrxd_min = {EM_MIN_RXD}, + .isc_ntxd_min = {EM_MIN_TXD}, + .isc_nrxd_max = {EM_MAX_RXD}, + .isc_ntxd_max = {EM_MAX_TXD}, + .isc_nrxd_default = {EM_DEFAULT_RXD}, + .isc_ntxd_default = {EM_DEFAULT_TXD}, +}; + +if_shared_ctx_t igb_sctx = &igb_sctx_init; + +/***************************************************************** + * + * Dump Registers + * + ****************************************************************/ +#define IGB_REGS_LEN 739 + +static int em_get_regs(SYSCTL_HANDLER_ARGS) { - char adapter_name[60]; - uint16_t pci_vendor_id = 0; - uint16_t pci_device_id = 0; - uint16_t pci_subvendor_id = 0; - uint16_t pci_subdevice_id = 0; - em_vendor_info_t *ent; + struct adapter *adapter = (struct adapter *)arg1; + struct e1000_hw *hw = &adapter->hw; - INIT_DEBUGOUT("em_probe: begin"); + struct sbuf *sb; + u32 *regs_buff = (u32 *)malloc(sizeof(u32) * IGB_REGS_LEN, M_DEVBUF, M_NOWAIT); + int rc; - pci_vendor_id = pci_get_vendor(dev); - if (pci_vendor_id != EM_VENDOR_ID) - return (ENXIO); + memset(regs_buff, 0, IGB_REGS_LEN * sizeof(u32)); - pci_device_id = pci_get_device(dev); - pci_subvendor_id = pci_get_subvendor(dev); - pci_subdevice_id = pci_get_subdevice(dev); + rc = sysctl_wire_old_buffer(req, 0); + MPASS(rc == 0); + if (rc != 0) + return (rc); - ent = em_vendor_info_array; - while (ent->vendor_id != 0) { - if ((pci_vendor_id == ent->vendor_id) && - (pci_device_id == ent->device_id) && + sb = sbuf_new_for_sysctl(NULL, NULL, 32*400, req); + MPASS(sb != NULL); + if (sb == NULL) + return (ENOMEM); - ((pci_subvendor_id == ent->subvendor_id) || - (ent->subvendor_id == PCI_ANY_ID)) && + /* General Registers */ + regs_buff[0] = E1000_READ_REG(hw, E1000_CTRL); + regs_buff[1] = E1000_READ_REG(hw, E1000_STATUS); + regs_buff[2] = E1000_READ_REG(hw, E1000_CTRL_EXT); + regs_buff[3] = E1000_READ_REG(hw, E1000_ICR); + regs_buff[4] = E1000_READ_REG(hw, E1000_RCTL); + regs_buff[5] = E1000_READ_REG(hw, E1000_RDLEN(0)); + regs_buff[6] = E1000_READ_REG(hw, E1000_RDH(0)); + regs_buff[7] = E1000_READ_REG(hw, E1000_RDT(0)); + regs_buff[8] = E1000_READ_REG(hw, E1000_RXDCTL(0)); + regs_buff[9] = E1000_READ_REG(hw, E1000_RDBAL(0)); + regs_buff[10] = E1000_READ_REG(hw, E1000_RDBAH(0)); + regs_buff[11] = E1000_READ_REG(hw, E1000_TCTL); + regs_buff[12] = E1000_READ_REG(hw, E1000_TDBAL(0)); + regs_buff[13] = E1000_READ_REG(hw, E1000_TDBAH(0)); + regs_buff[14] = E1000_READ_REG(hw, E1000_TDLEN(0)); + regs_buff[15] = E1000_READ_REG(hw, E1000_TDH(0)); + regs_buff[16] = E1000_READ_REG(hw, E1000_TDT(0)); + regs_buff[17] = E1000_READ_REG(hw, E1000_TXDCTL(0)); + regs_buff[18] = E1000_READ_REG(hw, E1000_TDFH); + regs_buff[19] = E1000_READ_REG(hw, E1000_TDFT); + regs_buff[20] = E1000_READ_REG(hw, E1000_TDFHS); + regs_buff[21] = E1000_READ_REG(hw, E1000_TDFPC); + + sbuf_printf(sb, "General Registers\n"); + sbuf_printf(sb, "\tCTRL\t %08x\n", regs_buff[0]); + sbuf_printf(sb, "\tSTATUS\t %08x\n", regs_buff[1]); + sbuf_printf(sb, "\tCTRL_EXIT\t %08x\n\n", regs_buff[2]); - ((pci_subdevice_id == ent->subdevice_id) || - (ent->subdevice_id == PCI_ANY_ID))) { - sprintf(adapter_name, "%s %s", - em_strings[ent->index], - em_driver_version); - device_set_desc_copy(dev, adapter_name); - return (BUS_PROBE_DEFAULT); - } - ent++; + sbuf_printf(sb, "Interrupt Registers\n"); + sbuf_printf(sb, "\tICR\t %08x\n\n", regs_buff[3]); + + sbuf_printf(sb, "RX Registers\n"); + sbuf_printf(sb, "\tRCTL\t %08x\n", regs_buff[4]); + sbuf_printf(sb, "\tRDLEN\t %08x\n", regs_buff[5]); + sbuf_printf(sb, "\tRDH\t %08x\n", regs_buff[6]); + sbuf_printf(sb, "\tRDT\t %08x\n", regs_buff[7]); + sbuf_printf(sb, "\tRXDCTL\t %08x\n", regs_buff[8]); + sbuf_printf(sb, "\tRDBAL\t %08x\n", regs_buff[9]); + sbuf_printf(sb, "\tRDBAH\t %08x\n\n", regs_buff[10]); + + sbuf_printf(sb, "TX Registers\n"); + sbuf_printf(sb, "\tTCTL\t %08x\n", regs_buff[11]); + sbuf_printf(sb, "\tTDBAL\t %08x\n", regs_buff[12]); + sbuf_printf(sb, "\tTDBAH\t %08x\n", regs_buff[13]); + sbuf_printf(sb, "\tTDLEN\t %08x\n", regs_buff[14]); + sbuf_printf(sb, "\tTDH\t %08x\n", regs_buff[15]); + sbuf_printf(sb, "\tTDT\t %08x\n", regs_buff[16]); + sbuf_printf(sb, "\tTXDCTL\t %08x\n", regs_buff[17]); + sbuf_printf(sb, "\tTDFH\t %08x\n", regs_buff[18]); + sbuf_printf(sb, "\tTDFT\t %08x\n", regs_buff[19]); + sbuf_printf(sb, "\tTDFHS\t %08x\n", regs_buff[20]); + sbuf_printf(sb, "\tTDFPC\t %08x\n\n", regs_buff[21]); + +#ifdef DUMP_DESCS + { + if_softc_ctx_t scctx = adapter->shared; + struct rx_ring *rxr = &rx_que->rxr; + struct tx_ring *txr = &tx_que->txr; + int ntxd = scctx->isc_ntxd[0]; + int nrxd = scctx->isc_nrxd[0]; + int j; + + for (j = 0; j < nrxd; j++) { + u32 staterr = le32toh(rxr->rx_base[j].wb.upper.status_error); + u32 length = le32toh(rxr->rx_base[j].wb.upper.length); + sbuf_printf(sb, "\tReceive Descriptor Address %d: %08" PRIx64 " Error:%d Length:%d\n", j, rxr->rx_base[j].read.buffer_addr, staterr, length); } - return (ENXIO); + for (j = 0; j < min(ntxd, 256); j++) { + struct em_txbuffer *buf = &txr->tx_buffers[j]; + unsigned int *ptr = (unsigned int *)&txr->tx_base[j]; + + sbuf_printf(sb, "\tTXD[%03d] [0]: %08x [1]: %08x [2]: %08x [3]: %08x eop: %d DD=%d\n", + j, ptr[0], ptr[1], ptr[2], ptr[3], buf->eop, + buf->eop != -1 ? txr->tx_base[buf->eop].upper.fields.status & E1000_TXD_STAT_DD : 0); + + } + } +#endif + + rc = sbuf_finish(sb); + sbuf_delete(sb); + return(rc); } +static void * +em_register(device_t dev) +{ + return (em_sctx); +} + +static void * +igb_register(device_t dev) +{ + return (igb_sctx); +} + +static void +em_init_tx_ring(struct em_tx_queue *que) +{ + struct adapter *sc = que->adapter; + if_softc_ctx_t scctx = sc->shared; + struct tx_ring *txr = &que->txr; + struct em_txbuffer *tx_buffer; + + tx_buffer = txr->tx_buffers; + for (int i = 0; i < scctx->isc_ntxd[0]; i++, tx_buffer++) { + tx_buffer->eop = -1; + } +} + +static int +em_set_num_queues(if_ctx_t ctx) +{ + struct adapter *adapter = iflib_get_softc(ctx); + int maxqueues; + + /* Sanity check based on HW */ + switch (adapter->hw.mac.type) { + case e1000_82576: + case e1000_82580: + case e1000_i350: + case e1000_i354: + maxqueues = 8; + break; + case e1000_i210: + case e1000_82575: + maxqueues = 4; + break; + case e1000_i211: + case e1000_82574: + maxqueues = 2; + break; + default: + maxqueues = 1; + break; + } + + return (maxqueues); +} + + +#define EM_CAPS \ + IFCAP_TSO4 | IFCAP_TXCSUM | IFCAP_LRO | IFCAP_RXCSUM | IFCAP_VLAN_HWFILTER | IFCAP_WOL_MAGIC | \ + IFCAP_WOL_MCAST | IFCAP_WOL | IFCAP_VLAN_HWTSO | IFCAP_HWCSUM | IFCAP_VLAN_HWTAGGING | \ + IFCAP_VLAN_HWCSUM | IFCAP_VLAN_HWTSO | IFCAP_VLAN_MTU; + +#define IGB_CAPS \ + IFCAP_TSO4 | IFCAP_TXCSUM | IFCAP_LRO | IFCAP_RXCSUM | IFCAP_VLAN_HWFILTER | IFCAP_WOL_MAGIC | \ + IFCAP_WOL_MCAST | IFCAP_WOL | IFCAP_VLAN_HWTSO | IFCAP_HWCSUM | IFCAP_VLAN_HWTAGGING | IFCAP_VLAN_HWCSUM | \ + IFCAP_VLAN_HWTSO | IFCAP_VLAN_MTU | IFCAP_TXCSUM_IPV6 | IFCAP_HWCSUM_IPV6 | IFCAP_JUMBO_MTU; + /********************************************************************* * Device initialization routine * @@ -507,23 +682,30 @@ em_probe(device_t dev) *********************************************************************/ static int -em_attach(device_t dev) +em_if_attach_pre(if_ctx_t ctx) { - struct adapter *adapter; + struct adapter *adapter; + if_softc_ctx_t scctx; + device_t dev; struct e1000_hw *hw; int error = 0; - INIT_DEBUGOUT("em_attach: begin"); + INIT_DEBUGOUT("em_if_attach_pre begin"); + dev = iflib_get_dev(ctx); + adapter = iflib_get_softc(ctx); if (resource_disabled("em", device_get_unit(dev))) { device_printf(dev, "Disabled by device hint\n"); return (ENXIO); } - adapter = device_get_softc(dev); + adapter->ctx = ctx; adapter->dev = adapter->osdep.dev = dev; - hw = &adapter->hw; - EM_CORE_LOCK_INIT(adapter, device_get_nameunit(dev)); + scctx = adapter->shared = iflib_get_softc_ctx(ctx); + adapter->media = iflib_get_media(ctx); + hw = &adapter->hw; + + adapter->tx_process_limit = scctx->isc_ntxd[0]; /* SYSCTL stuff */ SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), @@ -541,13 +723,56 @@ em_attach(device_t dev) OID_AUTO, "fc", CTLTYPE_INT|CTLFLAG_RW, adapter, 0, em_set_flowcntl, "I", "Flow Control"); - callout_init_mtx(&adapter->timer, &adapter->core_mtx, 0); + SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), + SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), + OID_AUTO, "reg_dump", CTLTYPE_STRING | CTLFLAG_RD, adapter, 0, + em_get_regs, "A", "Dump Registers"); /* Determine hardware and mac info */ - em_identify_hardware(adapter); + em_identify_hardware(ctx); + + /* Set isc_msix_bar */ + scctx->isc_msix_bar = PCIR_BAR(EM_MSIX_BAR); + scctx->isc_tx_nsegments = EM_MAX_SCATTER; + scctx->isc_tx_tso_segments_max = scctx->isc_tx_nsegments; + scctx->isc_tx_tso_size_max = EM_TSO_SIZE; + scctx->isc_tx_tso_segsize_max = EM_TSO_SEG_SIZE; + scctx->isc_nrxqsets_max = scctx->isc_ntxqsets_max = em_set_num_queues(ctx); + device_printf(dev, "attach_pre capping queues at %d\n", scctx->isc_ntxqsets_max); + + scctx->isc_tx_csum_flags = CSUM_TCP | CSUM_UDP | CSUM_IP_TSO; + + + if (adapter->hw.mac.type >= igb_mac_min) { + scctx->isc_txqsizes[0] = roundup2(scctx->isc_ntxd[0] * sizeof(union e1000_adv_tx_desc), EM_DBA_ALIGN); + scctx->isc_rxqsizes[0] = roundup2(scctx->isc_nrxd[0] * sizeof(union e1000_adv_rx_desc), EM_DBA_ALIGN); + scctx->isc_txrx = &igb_txrx; + scctx->isc_capenable = IGB_CAPS; + scctx->isc_tx_csum_flags = CSUM_TCP | CSUM_UDP | CSUM_TSO | CSUM_IP6_TCP \ + | CSUM_IP6_UDP | CSUM_IP6_TCP; + if (adapter->hw.mac.type != e1000_82575) + scctx->isc_tx_csum_flags |= CSUM_SCTP | CSUM_IP6_SCTP; + + } else if (adapter->hw.mac.type >= em_mac_min) { + scctx->isc_txqsizes[0] = roundup2(scctx->isc_ntxd[0]* sizeof(struct e1000_tx_desc), EM_DBA_ALIGN); + scctx->isc_rxqsizes[0] = roundup2(scctx->isc_nrxd[0] * sizeof(union e1000_rx_desc_extended), EM_DBA_ALIGN); + scctx->isc_txrx = &em_txrx; + scctx->isc_capenable = EM_CAPS; + scctx->isc_tx_csum_flags = CSUM_TCP | CSUM_UDP | CSUM_IP_TSO; + } else { + scctx->isc_txqsizes[0] = roundup2((scctx->isc_ntxd[0] + 1) * sizeof(struct e1000_tx_desc), EM_DBA_ALIGN); + scctx->isc_rxqsizes[0] = roundup2((scctx->isc_nrxd[0] + 1) * sizeof(struct e1000_rx_desc), EM_DBA_ALIGN); + scctx->isc_tx_csum_flags = CSUM_TCP | CSUM_UDP | CSUM_IP_TSO; + scctx->isc_txrx = &lem_txrx; + scctx->isc_capenable = EM_CAPS; + if (adapter->hw.mac.type < e1000_82543) + scctx->isc_capenable &= ~(IFCAP_HWCSUM|IFCAP_VLAN_HWCSUM); + scctx->isc_tx_csum_flags = CSUM_TCP | CSUM_UDP | CSUM_IP_TSO; + scctx->isc_msix_bar = 0; + } /* Setup PCI resources */ - if (em_allocate_pci_resources(adapter)) { + if (em_allocate_pci_resources(ctx)) { device_printf(dev, "Allocation of PCI resources failed\n"); error = ENXIO; goto err_pci; @@ -603,11 +828,7 @@ em_attach(device_t dev) goto err_pci; } - /* - * Setup MSI/X or MSI if PCI Express - */ - adapter->msix = em_setup_msix(adapter); - + em_setup_msix(ctx); e1000_get_bus_info(hw); /* Set up some sysctls for the tunable interrupt delays */ @@ -637,32 +858,15 @@ em_attach(device_t dev) em_set_sysctl_value(adapter, "rx_processing_limit", "max number of rx packets to process", &adapter->rx_process_limit, em_rx_process_limit); - - /* - * Validate number of transmit and receive descriptors. It - * must not exceed hardware maximum, and must be multiple - * of E1000_DBA_ALIGN. - */ - if (((em_txd * sizeof(struct e1000_tx_desc)) % EM_DBA_ALIGN) != 0 || - (em_txd > EM_MAX_TXD) || (em_txd < EM_MIN_TXD)) { - device_printf(dev, "Using %d TX descriptors instead of %d!\n", - EM_DEFAULT_TXD, em_txd); - adapter->num_tx_desc = EM_DEFAULT_TXD; - } else - adapter->num_tx_desc = em_txd; - - if (((em_rxd * sizeof(union e1000_rx_desc_extended)) % EM_DBA_ALIGN) != 0 || - (em_rxd > EM_MAX_RXD) || (em_rxd < EM_MIN_RXD)) { - device_printf(dev, "Using %d RX descriptors instead of %d!\n", - EM_DEFAULT_RXD, em_rxd); - adapter->num_rx_desc = EM_DEFAULT_RXD; - } else - adapter->num_rx_desc = em_rxd; - + hw->mac.autoneg = DO_AUTO_NEG; hw->phy.autoneg_wait_to_complete = FALSE; hw->phy.autoneg_advertised = AUTONEG_ADV_DEFAULT; + if (adapter->hw.mac.type < em_mac_min) { + e1000_init_script_state_82541(&adapter->hw, TRUE); + e1000_set_tbi_compatibility_82543(&adapter->hw, TRUE); + } /* Copper options */ if (hw->phy.media_type == e1000_media_type_copper) { hw->phy.mdix = AUTO_ALL_MODES; @@ -683,14 +887,6 @@ em_attach(device_t dev) */ hw->mac.report_tx_early = 1; - /* - ** Get queue/ring memory - */ - if (em_allocate_queues(adapter)) { - error = ENOMEM; - goto err_pci; - } - /* Allocate multicast array memory. */ adapter->mta = malloc(sizeof(u8) * ETH_ADDR_LEN * MAX_NUM_MULTICAST_ADDRESSES, M_DEVBUF, M_NOWAIT); @@ -720,7 +916,6 @@ em_attach(device_t dev) */ e1000_reset_hw(hw); - /* Make sure we have a good EEPROM before we read from it */ if (e1000_validate_nvm_checksum(hw) < 0) { /* @@ -753,68 +948,58 @@ em_attach(device_t dev) /* Disable ULP support */ e1000_disable_ulp_lpt_lp(hw, TRUE); - /* - ** Do interrupt configuration - */ - if (adapter->msix > 1) /* Do MSIX */ - error = em_allocate_msix(adapter); - else /* MSI or Legacy */ - error = em_allocate_legacy(adapter); - if (error) - goto err_late; - - /* + /* * Get Wake-on-Lan and Management info for later use */ - em_get_wakeup(dev); + em_get_wakeup(ctx); + iflib_set_mac(ctx, hw->mac.addr); + + return (0); + +err_late: + em_release_hw_control(adapter); +err_pci: + em_free_pci_resources(ctx); + free(adapter->mta, M_DEVBUF); + + return (error); +} + +static int +em_if_attach_post(if_ctx_t ctx) +{ + struct adapter *adapter = iflib_get_softc(ctx); + struct e1000_hw *hw = &adapter->hw; + int error = 0; + /* Setup OS specific network interface */ - if (em_setup_interface(dev, adapter) != 0) + error = em_setup_interface(ctx); + if (error != 0) { goto err_late; + } - em_reset(adapter); + em_reset(ctx); /* Initialize statistics */ em_update_stats_counters(adapter); - hw->mac.get_link_status = 1; - em_update_link_status(adapter); - - /* Register for VLAN events */ - adapter->vlan_attach = EVENTHANDLER_REGISTER(vlan_config, - em_register_vlan, adapter, EVENTHANDLER_PRI_FIRST); - adapter->vlan_detach = EVENTHANDLER_REGISTER(vlan_unconfig, - em_unregister_vlan, adapter, EVENTHANDLER_PRI_FIRST); - + em_if_update_admin_status(ctx); em_add_hw_stats(adapter); /* Non-AMT based hardware can now take control from firmware */ if (adapter->has_manage && !adapter->has_amt) em_get_hw_control(adapter); + + INIT_DEBUGOUT("em_if_attach_post: end"); - /* Tell the stack that the interface is not active */ - if_setdrvflagbits(adapter->ifp, IFF_DRV_OACTIVE, IFF_DRV_RUNNING); - - adapter->led_dev = led_create(em_led_func, adapter, - device_get_nameunit(dev)); -#ifdef DEV_NETMAP - em_netmap_attach(adapter); -#endif /* DEV_NETMAP */ - - INIT_DEBUGOUT("em_attach: end"); - - return (0); + return (error); err_late: - em_free_transmit_structures(adapter); - em_free_receive_structures(adapter); em_release_hw_control(adapter); - if (adapter->ifp != (void *)NULL) - if_free(adapter->ifp); -err_pci: - em_free_pci_resources(adapter); + em_free_pci_resources(ctx); + em_if_queues_free(ctx); free(adapter->mta, M_DEVBUF); - EM_CORE_LOCK_DESTROY(adapter); return (error); } @@ -830,60 +1015,17 @@ em_attach(device_t dev) *********************************************************************/ static int -em_detach(device_t dev) +em_if_detach(if_ctx_t ctx) { - struct adapter *adapter = device_get_softc(dev); - if_t ifp = adapter->ifp; + struct adapter *adapter = iflib_get_softc(ctx); INIT_DEBUGOUT("em_detach: begin"); - /* Make sure VLANS are not using driver */ - if (if_vlantrunkinuse(ifp)) { - device_printf(dev,"Vlan in use, detach first\n"); - return (EBUSY); - } - -#ifdef DEVICE_POLLING - if (if_getcapenable(ifp) & IFCAP_POLLING) - ether_poll_deregister(ifp); -#endif - - if (adapter->led_dev != NULL) - led_destroy(adapter->led_dev); - - EM_CORE_LOCK(adapter); - adapter->in_detach = 1; - em_stop(adapter); - EM_CORE_UNLOCK(adapter); - EM_CORE_LOCK_DESTROY(adapter); - e1000_phy_hw_reset(&adapter->hw); em_release_manageability(adapter); em_release_hw_control(adapter); - - /* Unregister VLAN events */ - if (adapter->vlan_attach != NULL) - EVENTHANDLER_DEREGISTER(vlan_config, adapter->vlan_attach); - if (adapter->vlan_detach != NULL) - EVENTHANDLER_DEREGISTER(vlan_unconfig, adapter->vlan_detach); - - ether_ifdetach(adapter->ifp); - callout_drain(&adapter->timer); - -#ifdef DEV_NETMAP - netmap_detach(ifp); -#endif /* DEV_NETMAP */ - - em_free_pci_resources(adapter); - bus_generic_detach(dev); - if_free(ifp); - - em_free_transmit_structures(adapter); - em_free_receive_structures(adapter); - - em_release_hw_control(adapter); - free(adapter->mta, M_DEVBUF); + em_free_pci_resources(ctx); return (0); } @@ -895,448 +1037,78 @@ em_detach(device_t dev) **********************************************************************/ static int -em_shutdown(device_t dev) +em_if_shutdown(if_ctx_t ctx) { - return em_suspend(dev); + return em_if_suspend(ctx); } /* * Suspend/resume device methods. */ static int -em_suspend(device_t dev) +em_if_suspend(if_ctx_t ctx) { - struct adapter *adapter = device_get_softc(dev); - - EM_CORE_LOCK(adapter); + struct adapter *adapter = iflib_get_softc(ctx); em_release_manageability(adapter); em_release_hw_control(adapter); - em_enable_wakeup(dev); - - EM_CORE_UNLOCK(adapter); - - return bus_generic_suspend(dev); -} - -static int -em_resume(device_t dev) -{ - struct adapter *adapter = device_get_softc(dev); - struct tx_ring *txr = adapter->tx_rings; - if_t ifp = adapter->ifp; - - EM_CORE_LOCK(adapter); - if (adapter->hw.mac.type == e1000_pch2lan) - e1000_resume_workarounds_pchlan(&adapter->hw); - em_init_locked(adapter); - em_init_manageability(adapter); - - if ((if_getflags(ifp) & IFF_UP) && - (if_getdrvflags(ifp) & IFF_DRV_RUNNING) && adapter->link_active) { - for (int i = 0; i < adapter->num_queues; i++, txr++) { - EM_TX_LOCK(txr); -#ifdef EM_MULTIQUEUE - if (!drbr_empty(ifp, txr->br)) - em_mq_start_locked(ifp, txr); -#else - if (!if_sendq_empty(ifp)) - em_start_locked(ifp, txr); -#endif - EM_TX_UNLOCK(txr); - } - } - EM_CORE_UNLOCK(adapter); - - return bus_generic_resume(dev); -} - - -#ifndef EM_MULTIQUEUE -static void -em_start_locked(if_t ifp, struct tx_ring *txr) -{ - struct adapter *adapter = if_getsoftc(ifp); - struct mbuf *m_head; - - EM_TX_LOCK_ASSERT(txr); - - if ((if_getdrvflags(ifp) & (IFF_DRV_RUNNING|IFF_DRV_OACTIVE)) != - IFF_DRV_RUNNING) - return; - - if (!adapter->link_active) - return; - - while (!if_sendq_empty(ifp)) { - /* Call cleanup if number of TX descriptors low */ - if (txr->tx_avail <= EM_TX_CLEANUP_THRESHOLD) - em_txeof(txr); - if (txr->tx_avail < EM_MAX_SCATTER) { - if_setdrvflagbits(ifp,IFF_DRV_OACTIVE, 0); - break; - } - m_head = if_dequeue(ifp); - if (m_head == NULL) - break; - /* - * Encapsulation can modify our pointer, and or make it - * NULL on failure. In that event, we can't requeue. - */ - if (em_xmit(txr, &m_head)) { - if (m_head == NULL) - break; - if_sendq_prepend(ifp, m_head); - break; - } - - /* Mark the queue as having work */ - if (txr->busy == EM_TX_IDLE) - txr->busy = EM_TX_BUSY; - - /* Send a copy of the frame to the BPF listener */ - ETHER_BPF_MTAP(ifp, m_head); - - } - - return; -} - -static void -em_start(if_t ifp) -{ - struct adapter *adapter = if_getsoftc(ifp); - struct tx_ring *txr = adapter->tx_rings; - - if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) { - EM_TX_LOCK(txr); - em_start_locked(ifp, txr); - EM_TX_UNLOCK(txr); - } - return; -} -#else /* EM_MULTIQUEUE */ -/********************************************************************* - * Multiqueue Transmit routines - * - * em_mq_start is called by the stack to initiate a transmit. - * however, if busy the driver can queue the request rather - * than do an immediate send. It is this that is an advantage - * in this driver, rather than also having multiple tx queues. - **********************************************************************/ -/* -** Multiqueue capable stack interface -*/ -static int -em_mq_start(if_t ifp, struct mbuf *m) -{ - struct adapter *adapter = if_getsoftc(ifp); - struct tx_ring *txr = adapter->tx_rings; - unsigned int i, error; - - if (M_HASHTYPE_GET(m) != M_HASHTYPE_NONE) - i = m->m_pkthdr.flowid % adapter->num_queues; - else - i = curcpu % adapter->num_queues; - - txr = &adapter->tx_rings[i]; - - error = drbr_enqueue(ifp, txr->br, m); - if (error) - return (error); - - if (EM_TX_TRYLOCK(txr)) { - em_mq_start_locked(ifp, txr); - EM_TX_UNLOCK(txr); - } else - taskqueue_enqueue(txr->tq, &txr->tx_task); - + em_enable_wakeup(ctx); return (0); } static int -em_mq_start_locked(if_t ifp, struct tx_ring *txr) +em_if_resume(if_ctx_t ctx) { - struct adapter *adapter = txr->adapter; - struct mbuf *next; - int err = 0, enq = 0; + struct adapter *adapter = iflib_get_softc(ctx); - EM_TX_LOCK_ASSERT(txr); + if (adapter->hw.mac.type == e1000_pch2lan) + e1000_resume_workarounds_pchlan(&adapter->hw); + em_if_init(ctx); + em_init_manageability(adapter); - if (((if_getdrvflags(ifp) & IFF_DRV_RUNNING) == 0) || - adapter->link_active == 0) { - return (ENETDOWN); - } - - /* Process the queue */ - while ((next = drbr_peek(ifp, txr->br)) != NULL) { - if ((err = em_xmit(txr, &next)) != 0) { - if (next == NULL) { - /* It was freed, move forward */ - drbr_advance(ifp, txr->br); - } else { - /* - * Still have one left, it may not be - * the same since the transmit function - * may have changed it. - */ - drbr_putback(ifp, txr->br, next); - } - break; - } - drbr_advance(ifp, txr->br); - enq++; - if_inc_counter(ifp, IFCOUNTER_OBYTES, next->m_pkthdr.len); - if (next->m_flags & M_MCAST) - if_inc_counter(ifp, IFCOUNTER_OMCASTS, 1); - ETHER_BPF_MTAP(ifp, next); - if ((if_getdrvflags(ifp) & IFF_DRV_RUNNING) == 0) - break; - } - - /* Mark the queue as having work */ - if ((enq > 0) && (txr->busy == EM_TX_IDLE)) - txr->busy = EM_TX_BUSY; - - if (txr->tx_avail < EM_MAX_SCATTER) - em_txeof(txr); - if (txr->tx_avail < EM_MAX_SCATTER) { - if_setdrvflagbits(ifp, IFF_DRV_OACTIVE,0); - } - return (err); + return(0); } -/* -** Flush all ring buffers -*/ -static void -em_qflush(if_t ifp) -{ - struct adapter *adapter = if_getsoftc(ifp); - struct tx_ring *txr = adapter->tx_rings; - struct mbuf *m; - - for (int i = 0; i < adapter->num_queues; i++, txr++) { - EM_TX_LOCK(txr); - while ((m = buf_ring_dequeue_sc(txr->br)) != NULL) - m_freem(m); - EM_TX_UNLOCK(txr); - } - if_qflush(ifp); -} -#endif /* EM_MULTIQUEUE */ - -/********************************************************************* - * Ioctl entry point - * - * em_ioctl is called when the user wants to configure the - * interface. - * - * return 0 on success, positive on failure - **********************************************************************/ - static int -em_ioctl(if_t ifp, u_long command, caddr_t data) +em_if_mtu_set(if_ctx_t ctx, uint32_t mtu) { - struct adapter *adapter = if_getsoftc(ifp); - struct ifreq *ifr = (struct ifreq *)data; -#if defined(INET) || defined(INET6) - struct ifaddr *ifa = (struct ifaddr *)data; -#endif - bool avoid_reset = FALSE; - int error = 0; - - if (adapter->in_detach) - return (error); - - switch (command) { - case SIOCSIFADDR: -#ifdef INET - if (ifa->ifa_addr->sa_family == AF_INET) - avoid_reset = TRUE; -#endif -#ifdef INET6 - if (ifa->ifa_addr->sa_family == AF_INET6) - avoid_reset = TRUE; -#endif - /* - ** Calling init results in link renegotiation, - ** so we avoid doing it when possible. - */ - if (avoid_reset) { - if_setflagbits(ifp,IFF_UP,0); - if (!(if_getdrvflags(ifp)& IFF_DRV_RUNNING)) - em_init(adapter); -#ifdef INET - if (!(if_getflags(ifp) & IFF_NOARP)) - arp_ifinit(ifp, ifa); -#endif - } else - error = ether_ioctl(ifp, command, data); - break; - case SIOCSIFMTU: - { - int max_frame_size; - - IOCTL_DEBUGOUT("ioctl rcv'd: SIOCSIFMTU (Set Interface MTU)"); - - EM_CORE_LOCK(adapter); - switch (adapter->hw.mac.type) { - case e1000_82571: - case e1000_82572: - case e1000_ich9lan: - case e1000_ich10lan: - case e1000_pch2lan: - case e1000_pch_lpt: - case e1000_pch_spt: - case e1000_82574: - case e1000_82583: - case e1000_80003es2lan: /* 9K Jumbo Frame size */ - max_frame_size = 9234; - break; - case e1000_pchlan: - max_frame_size = 4096; - break; - /* Adapters that do not support jumbo frames */ - case e1000_ich8lan: - max_frame_size = ETHER_MAX_LEN; - break; - default: - max_frame_size = MAX_JUMBO_FRAME_SIZE; - } - if (ifr->ifr_mtu > max_frame_size - ETHER_HDR_LEN - - ETHER_CRC_LEN) { - EM_CORE_UNLOCK(adapter); - error = EINVAL; - break; - } - - if_setmtu(ifp, ifr->ifr_mtu); - adapter->hw.mac.max_frame_size = - if_getmtu(ifp) + ETHER_HDR_LEN + ETHER_CRC_LEN; - if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) - em_init_locked(adapter); - EM_CORE_UNLOCK(adapter); - break; - } - case SIOCSIFFLAGS: - IOCTL_DEBUGOUT("ioctl rcv'd:\ - SIOCSIFFLAGS (Set Interface Flags)"); - EM_CORE_LOCK(adapter); - if (if_getflags(ifp) & IFF_UP) { - if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) { - if ((if_getflags(ifp) ^ adapter->if_flags) & - (IFF_PROMISC | IFF_ALLMULTI)) { - em_disable_promisc(adapter); - em_set_promisc(adapter); - } - } else - em_init_locked(adapter); - } else - if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) - em_stop(adapter); - adapter->if_flags = if_getflags(ifp); - EM_CORE_UNLOCK(adapter); - break; - case SIOCADDMULTI: - case SIOCDELMULTI: - IOCTL_DEBUGOUT("ioctl rcv'd: SIOC(ADD|DEL)MULTI"); - if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) { - EM_CORE_LOCK(adapter); - em_disable_intr(adapter); - em_set_multi(adapter); -#ifdef DEVICE_POLLING - if (!(if_getcapenable(ifp) & IFCAP_POLLING)) -#endif - em_enable_intr(adapter); - EM_CORE_UNLOCK(adapter); - } - break; - case SIOCSIFMEDIA: - /* Check SOL/IDER usage */ - EM_CORE_LOCK(adapter); - if (e1000_check_reset_block(&adapter->hw)) { - EM_CORE_UNLOCK(adapter); - device_printf(adapter->dev, "Media change is" - " blocked due to SOL/IDER session.\n"); - break; - } - EM_CORE_UNLOCK(adapter); - /* falls thru */ - case SIOCGIFMEDIA: - IOCTL_DEBUGOUT("ioctl rcv'd: \ - SIOCxIFMEDIA (Get/Set Interface Media)"); - error = ifmedia_ioctl(ifp, ifr, &adapter->media, command); - break; - case SIOCSIFCAP: - { - int mask, reinit; - - IOCTL_DEBUGOUT("ioctl rcv'd: SIOCSIFCAP (Set Capabilities)"); - reinit = 0; - mask = ifr->ifr_reqcap ^ if_getcapenable(ifp); -#ifdef DEVICE_POLLING - if (mask & IFCAP_POLLING) { - if (ifr->ifr_reqcap & IFCAP_POLLING) { - error = ether_poll_register(em_poll, ifp); - if (error) - return (error); - EM_CORE_LOCK(adapter); - em_disable_intr(adapter); - if_setcapenablebit(ifp, IFCAP_POLLING, 0); - EM_CORE_UNLOCK(adapter); - } else { - error = ether_poll_deregister(ifp); - /* Enable interrupt even in error case */ - EM_CORE_LOCK(adapter); - em_enable_intr(adapter); - if_setcapenablebit(ifp, 0, IFCAP_POLLING); - EM_CORE_UNLOCK(adapter); - } - } -#endif - if (mask & IFCAP_HWCSUM) { - if_togglecapenable(ifp,IFCAP_HWCSUM); - reinit = 1; - } - if (mask & IFCAP_TSO4) { - if_togglecapenable(ifp,IFCAP_TSO4); - reinit = 1; - } - if (mask & IFCAP_VLAN_HWTAGGING) { - if_togglecapenable(ifp,IFCAP_VLAN_HWTAGGING); - reinit = 1; - } - if (mask & IFCAP_VLAN_HWFILTER) { - if_togglecapenable(ifp, IFCAP_VLAN_HWFILTER); - reinit = 1; - } - if (mask & IFCAP_VLAN_HWTSO) { - if_togglecapenable(ifp, IFCAP_VLAN_HWTSO); - reinit = 1; - } - if ((mask & IFCAP_WOL) && - (if_getcapabilities(ifp) & IFCAP_WOL) != 0) { - if (mask & IFCAP_WOL_MCAST) - if_togglecapenable(ifp, IFCAP_WOL_MCAST); - if (mask & IFCAP_WOL_MAGIC) - if_togglecapenable(ifp, IFCAP_WOL_MAGIC); - } - if (reinit && (if_getdrvflags(ifp) & IFF_DRV_RUNNING)) - em_init(adapter); - if_vlancap(ifp); - break; - } - - default: - error = ether_ioctl(ifp, command, data); - break; - } - - return (error); + int max_frame_size; + struct adapter *adapter = iflib_get_softc(ctx); + struct ifnet *ifp = iflib_get_ifp(ctx); + + IOCTL_DEBUGOUT("ioctl rcv'd: SIOCSIFMTU (Set Interface MTU)"); + + switch (adapter->hw.mac.type) { + case e1000_82571: + case e1000_82572: + case e1000_ich9lan: + case e1000_ich10lan: + case e1000_pch2lan: + case e1000_pch_lpt: + case e1000_pch_spt: + case e1000_82574: + case e1000_82583: + case e1000_80003es2lan: /* 9K Jumbo Frame size */ + max_frame_size = 9234; + break; + case e1000_pchlan: + max_frame_size = 4096; + break; + /* Adapters that do not support jumbo frames */ + case e1000_ich8lan: + max_frame_size = ETHER_MAX_LEN; + break; + default: + max_frame_size = MAX_JUMBO_FRAME_SIZE; + } + if (mtu > max_frame_size - ETHER_HDR_LEN - ETHER_CRC_LEN) { + return (EINVAL); + } + + adapter->hw.mac.max_frame_size = if_getmtu(ifp) + ETHER_HDR_LEN + ETHER_CRC_LEN; + return (0); } - /********************************************************************* * Init entry point * @@ -1349,20 +1121,15 @@ em_ioctl(if_t ifp, u_long command, caddr_t data) **********************************************************************/ static void -em_init_locked(struct adapter *adapter) +em_if_init(if_ctx_t ctx) { - if_t ifp = adapter->ifp; - device_t dev = adapter->dev; + struct adapter *adapter = iflib_get_softc(ctx); + struct ifnet *ifp = iflib_get_ifp(ctx); - INIT_DEBUGOUT("em_init: begin"); - - EM_CORE_LOCK_ASSERT(adapter); - - em_disable_intr(adapter); - callout_stop(&adapter->timer); + INIT_DEBUGOUT("em_if_init: begin"); /* Get the latest mac address, User can use a LAA */ - bcopy(if_getlladdr(adapter->ifp), adapter->hw.mac.addr, + bcopy(if_getlladdr(ifp), adapter->hw.mac.addr, ETHER_ADDR_LEN); /* Put the address into the Receive Address Array */ @@ -1381,29 +1148,24 @@ em_init_locked(struct adapter *adapter) } /* Initialize the hardware */ - em_reset(adapter); - em_update_link_status(adapter); + em_reset(ctx); + em_if_update_admin_status(ctx); /* Setup VLAN support, basic and offload if available */ E1000_WRITE_REG(&adapter->hw, E1000_VET, ETHERTYPE_VLAN); - /* Set hardware offload abilities */ - if_clearhwassist(ifp); - if (if_getcapenable(ifp) & IFCAP_TXCSUM) - if_sethwassistbits(ifp, CSUM_TCP | CSUM_UDP, 0); - - if (if_getcapenable(ifp) & IFCAP_TSO4) - if_sethwassistbits(ifp, CSUM_TSO, 0); + /* Clear bad data from Rx FIFOs */ + if (adapter->hw.mac.type >= igb_mac_min) + e1000_rx_fifo_flush_82575(&adapter->hw); /* Configure for OS presence */ em_init_manageability(adapter); /* Prepare transmit descriptors and buffers */ - em_setup_transmit_structures(adapter); - em_initialize_transmit_unit(adapter); + em_initialize_transmit_unit(ctx); /* Setup Multicast table */ - em_set_multi(adapter); + em_if_multi_set(ctx); /* ** Figure out the desired mbuf @@ -1416,13 +1178,7 @@ em_init_locked(struct adapter *adapter) else adapter->rx_mbuf_sz = MJUM9BYTES; - /* Prepare receive descriptors and buffers */ - if (em_setup_receive_structures(adapter)) { - device_printf(dev, "Could not setup receive structures\n"); - em_stop(adapter); - return; - } - em_initialize_receive_unit(adapter); + em_initialize_receive_unit(ctx); /* Use real VLAN Filter support? */ if (if_getcapenable(ifp) & IFCAP_VLAN_HWTAGGING) { @@ -1438,118 +1194,54 @@ em_init_locked(struct adapter *adapter) } /* Don't lose promiscuous settings */ - em_set_promisc(adapter); - - /* Set the interface as ACTIVE */ - if_setdrvflagbits(ifp, IFF_DRV_RUNNING, IFF_DRV_OACTIVE); - - callout_reset(&adapter->timer, hz, em_local_timer, adapter); + em_if_set_promisc(ctx, IFF_PROMISC); e1000_clear_hw_cntrs_base_generic(&adapter->hw); /* MSI/X configuration for 82574 */ if (adapter->hw.mac.type == e1000_82574) { - int tmp; - tmp = E1000_READ_REG(&adapter->hw, E1000_CTRL_EXT); + int tmp = E1000_READ_REG(&adapter->hw, E1000_CTRL_EXT); + tmp |= E1000_CTRL_EXT_PBA_CLR; E1000_WRITE_REG(&adapter->hw, E1000_CTRL_EXT, tmp); /* Set the IVAR - interrupt vector routing. */ E1000_WRITE_REG(&adapter->hw, E1000_IVAR, adapter->ivars); - } + } else if (adapter->intr_type == IFLIB_INTR_MSIX) /* Set up queue routing */ + igb_configure_queues(adapter); -#ifdef DEVICE_POLLING - /* - * Only enable interrupts if we are not polling, make sure - * they are off otherwise. - */ - if (if_getcapenable(ifp) & IFCAP_POLLING) - em_disable_intr(adapter); - else -#endif /* DEVICE_POLLING */ - em_enable_intr(adapter); + /* this clears any pending interrupts */ + E1000_READ_REG(&adapter->hw, E1000_ICR); + E1000_WRITE_REG(&adapter->hw, E1000_ICS, E1000_ICS_LSC); /* AMT based hardware can now take control from firmware */ if (adapter->has_manage && adapter->has_amt) em_get_hw_control(adapter); -} -static void -em_init(void *arg) -{ - struct adapter *adapter = arg; - - EM_CORE_LOCK(adapter); - em_init_locked(adapter); - EM_CORE_UNLOCK(adapter); -} - - -#ifdef DEVICE_POLLING -/********************************************************************* - * - * Legacy polling routine: note this only works with single queue - * - *********************************************************************/ -static int -em_poll(if_t ifp, enum poll_cmd cmd, int count) -{ - struct adapter *adapter = if_getsoftc(ifp); - struct tx_ring *txr = adapter->tx_rings; - struct rx_ring *rxr = adapter->rx_rings; - u32 reg_icr; - int rx_done; - - EM_CORE_LOCK(adapter); - if ((if_getdrvflags(ifp) & IFF_DRV_RUNNING) == 0) { - EM_CORE_UNLOCK(adapter); - return (0); + /* Set Energy Efficient Ethernet */ + if (adapter->hw.mac.type >= igb_mac_min && + adapter->hw.phy.media_type == e1000_media_type_copper) { + if (adapter->hw.mac.type == e1000_i354) + e1000_set_eee_i354(&adapter->hw, TRUE, TRUE); + else + e1000_set_eee_i350(&adapter->hw, TRUE, TRUE); } - - if (cmd == POLL_AND_CHECK_STATUS) { - reg_icr = E1000_READ_REG(&adapter->hw, E1000_ICR); - if (reg_icr & (E1000_ICR_RXSEQ | E1000_ICR_LSC)) { - callout_stop(&adapter->timer); - adapter->hw.mac.get_link_status = 1; - em_update_link_status(adapter); - callout_reset(&adapter->timer, hz, - em_local_timer, adapter); - } - } - EM_CORE_UNLOCK(adapter); - - em_rxeof(rxr, count, &rx_done); - - EM_TX_LOCK(txr); - em_txeof(txr); -#ifdef EM_MULTIQUEUE - if (!drbr_empty(ifp, txr->br)) - em_mq_start_locked(ifp, txr); -#else - if (!if_sendq_empty(ifp)) - em_start_locked(ifp, txr); -#endif - EM_TX_UNLOCK(txr); - - return (rx_done); } -#endif /* DEVICE_POLLING */ - /********************************************************************* * * Fast Legacy/MSI Combined Interrupt Service routine * *********************************************************************/ -static int -em_irq_fast(void *arg) +int +em_intr(void *arg) { - struct adapter *adapter = arg; - if_t ifp; + struct adapter *adapter = arg; + if_ctx_t ctx = adapter->ctx; u32 reg_icr; - ifp = adapter->ifp; - reg_icr = E1000_READ_REG(&adapter->hw, E1000_ICR); + if (adapter->intr_type != IFLIB_INTR_LEGACY) + goto skip_stray; /* Hot eject? */ if (reg_icr == 0xffffffff) return FILTER_STRAY; @@ -1566,80 +1258,42 @@ em_irq_fast(void *arg) (reg_icr & E1000_ICR_INT_ASSERTED) == 0) return FILTER_STRAY; - em_disable_intr(adapter); - taskqueue_enqueue(adapter->tq, &adapter->que_task); - +skip_stray: /* Link status change */ if (reg_icr & (E1000_ICR_RXSEQ | E1000_ICR_LSC)) { adapter->hw.mac.get_link_status = 1; - taskqueue_enqueue(taskqueue_fast, &adapter->link_task); + iflib_admin_intr_deferred(ctx); } if (reg_icr & E1000_ICR_RXO) adapter->rx_overruns++; - return FILTER_HANDLED; + + return (FILTER_SCHEDULE_THREAD); } -/* Combined RX/TX handler, used by Legacy and MSI */ static void -em_handle_que(void *context, int pending) +igb_enable_queue(struct adapter *adapter, struct em_rx_queue *rxq) { - struct adapter *adapter = context; - if_t ifp = adapter->ifp; - struct tx_ring *txr = adapter->tx_rings; - struct rx_ring *rxr = adapter->rx_rings; - - if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) { - bool more = em_rxeof(rxr, adapter->rx_process_limit, NULL); - - EM_TX_LOCK(txr); - em_txeof(txr); -#ifdef EM_MULTIQUEUE - if (!drbr_empty(ifp, txr->br)) - em_mq_start_locked(ifp, txr); -#else - if (!if_sendq_empty(ifp)) - em_start_locked(ifp, txr); -#endif - EM_TX_UNLOCK(txr); - if (more) { - taskqueue_enqueue(adapter->tq, &adapter->que_task); - return; - } - } - - em_enable_intr(adapter); - return; + E1000_WRITE_REG(&adapter->hw, E1000_EIMS, rxq->eims); } - -/********************************************************************* - * - * MSIX Interrupt Service Routines - * - **********************************************************************/ static void -em_msix_tx(void *arg) +em_enable_queue(struct adapter *adapter, struct em_rx_queue *rxq) { - struct tx_ring *txr = arg; - struct adapter *adapter = txr->adapter; - if_t ifp = adapter->ifp; + E1000_WRITE_REG(&adapter->hw, E1000_IMS, rxq->eims); +} - ++txr->tx_irq; - EM_TX_LOCK(txr); - em_txeof(txr); -#ifdef EM_MULTIQUEUE - if (!drbr_empty(ifp, txr->br)) - em_mq_start_locked(ifp, txr); -#else - if (!if_sendq_empty(ifp)) - em_start_locked(ifp, txr); -#endif - - /* Reenable this interrupt */ - E1000_WRITE_REG(&adapter->hw, E1000_IMS, txr->ims); - EM_TX_UNLOCK(txr); - return; +static int +em_if_queue_intr_enable(if_ctx_t ctx, uint16_t rxqid) +{ + struct adapter *adapter = iflib_get_softc(ctx); + struct em_rx_queue *rxq = &adapter->rx_queues[rxqid]; + + if (adapter->hw.mac.type >= igb_mac_min) + igb_enable_queue(adapter, rxq); + else + em_enable_queue(adapter, rxq); + return (0); } /********************************************************************* @@ -1647,25 +1301,14 @@ em_msix_tx(void *arg) * MSIX RX Interrupt Service routine * **********************************************************************/ - -static void -em_msix_rx(void *arg) +static int +em_msix_que(void *arg) { - struct rx_ring *rxr = arg; - struct adapter *adapter = rxr->adapter; - bool more; - - ++rxr->rx_irq; - if (!(if_getdrvflags(adapter->ifp) & IFF_DRV_RUNNING)) - return; - more = em_rxeof(rxr, adapter->rx_process_limit, NULL); - if (more) - taskqueue_enqueue(rxr->tq, &rxr->rx_task); - else { - /* Reenable this interrupt */ - E1000_WRITE_REG(&adapter->hw, E1000_IMS, rxr->ims); - } - return; + struct em_rx_queue *que = arg; + + ++que->irqs; + + return (FILTER_SCHEDULE_THREAD); } /********************************************************************* @@ -1673,103 +1316,51 @@ em_msix_rx(void *arg) * MSIX Link Fast Interrupt Service routine * **********************************************************************/ -static void +static int em_msix_link(void *arg) { struct adapter *adapter = arg; u32 reg_icr; ++adapter->link_irq; + MPASS(adapter->hw.back != NULL); reg_icr = E1000_READ_REG(&adapter->hw, E1000_ICR); if (reg_icr & E1000_ICR_RXO) adapter->rx_overruns++; if (reg_icr & (E1000_ICR_RXSEQ | E1000_ICR_LSC)) { - adapter->hw.mac.get_link_status = 1; - em_handle_link(adapter, 0); - } else + em_handle_link(adapter->ctx); + } else { E1000_WRITE_REG(&adapter->hw, E1000_IMS, - EM_MSIX_LINK | E1000_IMS_LSC); + EM_MSIX_LINK | E1000_IMS_LSC); + if (adapter->hw.mac.type >= igb_mac_min) + E1000_WRITE_REG(&adapter->hw, E1000_EIMS, adapter->link_mask); + + } + /* ** Because we must read the ICR for this interrupt ** it may clear other causes using autoclear, for ** this reason we simply create a soft interrupt ** for all these vectors. */ - if (reg_icr) { + if (reg_icr && adapter->hw.mac.type < igb_mac_min) { E1000_WRITE_REG(&adapter->hw, E1000_ICS, adapter->ims); } - return; + + return (FILTER_HANDLED); } static void -em_handle_rx(void *context, int pending) +em_handle_link(void *context) { - struct rx_ring *rxr = context; - struct adapter *adapter = rxr->adapter; - bool more; + if_ctx_t ctx = context; + struct adapter *adapter = iflib_get_softc(ctx); - more = em_rxeof(rxr, adapter->rx_process_limit, NULL); - if (more) - taskqueue_enqueue(rxr->tq, &rxr->rx_task); - else { - /* Reenable this interrupt */ - E1000_WRITE_REG(&adapter->hw, E1000_IMS, rxr->ims); - } -} - -static void -em_handle_tx(void *context, int pending) -{ - struct tx_ring *txr = context; - struct adapter *adapter = txr->adapter; - if_t ifp = adapter->ifp; - - EM_TX_LOCK(txr); - em_txeof(txr); -#ifdef EM_MULTIQUEUE - if (!drbr_empty(ifp, txr->br)) - em_mq_start_locked(ifp, txr); -#else - if (!if_sendq_empty(ifp)) - em_start_locked(ifp, txr); -#endif - E1000_WRITE_REG(&adapter->hw, E1000_IMS, txr->ims); - EM_TX_UNLOCK(txr); -} - -static void -em_handle_link(void *context, int pending) -{ - struct adapter *adapter = context; - struct tx_ring *txr = adapter->tx_rings; - if_t ifp = adapter->ifp; - - if (!(if_getdrvflags(ifp) & IFF_DRV_RUNNING)) - return; - - EM_CORE_LOCK(adapter); - callout_stop(&adapter->timer); - em_update_link_status(adapter); - callout_reset(&adapter->timer, hz, em_local_timer, adapter); - E1000_WRITE_REG(&adapter->hw, E1000_IMS, - EM_MSIX_LINK | E1000_IMS_LSC); - if (adapter->link_active) { - for (int i = 0; i < adapter->num_queues; i++, txr++) { - EM_TX_LOCK(txr); -#ifdef EM_MULTIQUEUE - if (!drbr_empty(ifp, txr->br)) - em_mq_start_locked(ifp, txr); -#else - if (if_sendq_empty(ifp)) - em_start_locked(ifp, txr); -#endif - EM_TX_UNLOCK(txr); - } - } - EM_CORE_UNLOCK(adapter); + adapter->hw.mac.get_link_status = 1; + iflib_admin_intr_deferred(ctx); } @@ -1782,21 +1373,19 @@ em_handle_link(void *context, int pending) * **********************************************************************/ static void -em_media_status(if_t ifp, struct ifmediareq *ifmr) +em_if_media_status(if_ctx_t ctx, struct ifmediareq *ifmr) { - struct adapter *adapter = if_getsoftc(ifp); - u_char fiber_type = IFM_1000_SX; + struct adapter *adapter = iflib_get_softc(ctx); + u_char fiber_type = IFM_1000_SX; + + INIT_DEBUGOUT("em_if_media_status: begin"); - INIT_DEBUGOUT("em_media_status: begin"); - - EM_CORE_LOCK(adapter); - em_update_link_status(adapter); + iflib_admin_intr_deferred(ctx); ifmr->ifm_status = IFM_AVALID; ifmr->ifm_active = IFM_ETHER; if (!adapter->link_active) { - EM_CORE_UNLOCK(adapter); return; } @@ -1804,6 +1393,8 @@ em_media_status(if_t ifp, struct ifmediareq *ifmr) if ((adapter->hw.phy.media_type == e1000_media_type_fiber) || (adapter->hw.phy.media_type == e1000_media_type_internal_serdes)) { + if (adapter->hw.mac.type == e1000_82545) + fiber_type = IFM_1000_LX; ifmr->ifm_active |= fiber_type | IFM_FDX; } else { switch (adapter->link_speed) { @@ -1822,7 +1413,6 @@ em_media_status(if_t ifp, struct ifmediareq *ifmr) else ifmr->ifm_active |= IFM_HDX; } - EM_CORE_UNLOCK(adapter); } /********************************************************************* @@ -1834,17 +1424,16 @@ em_media_status(if_t ifp, struct ifmediareq *ifmr) * **********************************************************************/ static int -em_media_change(if_t ifp) +em_if_media_change(if_ctx_t ctx) { - struct adapter *adapter = if_getsoftc(ifp); - struct ifmedia *ifm = &adapter->media; + struct adapter *adapter = iflib_get_softc(ctx); + struct ifmedia *ifm = iflib_get_media(ctx); - INIT_DEBUGOUT("em_media_change: begin"); + INIT_DEBUGOUT("em_if_media_change: begin"); if (IFM_TYPE(ifm->ifm_media) != IFM_ETHER) return (EINVAL); - EM_CORE_LOCK(adapter); switch (IFM_SUBTYPE(ifm->ifm_media)) { case IFM_AUTO: adapter->hw.mac.autoneg = DO_AUTO_NEG; @@ -1876,356 +1465,40 @@ em_media_change(if_t ifp) device_printf(adapter->dev, "Unsupported media type\n"); } - em_init_locked(adapter); - EM_CORE_UNLOCK(adapter); + em_if_init(ctx); return (0); } -/********************************************************************* - * - * This routine maps the mbufs to tx descriptors. - * - * return 0 on success, positive on failure - **********************************************************************/ - static int -em_xmit(struct tx_ring *txr, struct mbuf **m_headp) +em_if_set_promisc(if_ctx_t ctx, int flags) { - struct adapter *adapter = txr->adapter; - bus_dma_segment_t segs[EM_MAX_SCATTER]; - bus_dmamap_t map; - struct em_txbuffer *tx_buffer, *tx_buffer_mapped; - struct e1000_tx_desc *ctxd = NULL; - struct mbuf *m_head; - struct ether_header *eh; - struct ip *ip = NULL; - struct tcphdr *tp = NULL; - u32 txd_upper = 0, txd_lower = 0; - int ip_off, poff; - int nsegs, i, j, first, last = 0; - int error; - bool do_tso, tso_desc, remap = TRUE; - - m_head = *m_headp; - do_tso = (m_head->m_pkthdr.csum_flags & CSUM_TSO); - tso_desc = FALSE; - ip_off = poff = 0; - - /* - * Intel recommends entire IP/TCP header length reside in a single - * buffer. If multiple descriptors are used to describe the IP and - * TCP header, each descriptor should describe one or more - * complete headers; descriptors referencing only parts of headers - * are not supported. If all layer headers are not coalesced into - * a single buffer, each buffer should not cross a 4KB boundary, - * or be larger than the maximum read request size. - * Controller also requires modifing IP/TCP header to make TSO work - * so we firstly get a writable mbuf chain then coalesce ethernet/ - * IP/TCP header into a single buffer to meet the requirement of - * controller. This also simplifies IP/TCP/UDP checksum offloading - * which also has similar restrictions. - */ - if (do_tso || m_head->m_pkthdr.csum_flags & CSUM_OFFLOAD) { - if (do_tso || (m_head->m_next != NULL && - m_head->m_pkthdr.csum_flags & CSUM_OFFLOAD)) { - if (M_WRITABLE(*m_headp) == 0) { - m_head = m_dup(*m_headp, M_NOWAIT); - m_freem(*m_headp); - if (m_head == NULL) { - *m_headp = NULL; - return (ENOBUFS); - } - *m_headp = m_head; - } - } - /* - * XXX - * Assume IPv4, we don't have TSO/checksum offload support - * for IPv6 yet. - */ - ip_off = sizeof(struct ether_header); - if (m_head->m_len < ip_off) { - m_head = m_pullup(m_head, ip_off); - if (m_head == NULL) { - *m_headp = NULL; - return (ENOBUFS); - } - } - eh = mtod(m_head, struct ether_header *); - if (eh->ether_type == htons(ETHERTYPE_VLAN)) { - ip_off = sizeof(struct ether_vlan_header); - if (m_head->m_len < ip_off) { - m_head = m_pullup(m_head, ip_off); - if (m_head == NULL) { - *m_headp = NULL; - return (ENOBUFS); - } - } - } - if (m_head->m_len < ip_off + sizeof(struct ip)) { - m_head = m_pullup(m_head, ip_off + sizeof(struct ip)); - if (m_head == NULL) { - *m_headp = NULL; - return (ENOBUFS); - } - } - ip = (struct ip *)(mtod(m_head, char *) + ip_off); - poff = ip_off + (ip->ip_hl << 2); - - if (do_tso || (m_head->m_pkthdr.csum_flags & CSUM_TCP)) { - if (m_head->m_len < poff + sizeof(struct tcphdr)) { - m_head = m_pullup(m_head, poff + - sizeof(struct tcphdr)); - if (m_head == NULL) { - *m_headp = NULL; - return (ENOBUFS); - } - } - tp = (struct tcphdr *)(mtod(m_head, char *) + poff); - /* - * TSO workaround: - * pull 4 more bytes of data into it. - */ - if (m_head->m_len < poff + (tp->th_off << 2)) { - m_head = m_pullup(m_head, poff + - (tp->th_off << 2) + - TSO_WORKAROUND); - if (m_head == NULL) { - *m_headp = NULL; - return (ENOBUFS); - } - } - ip = (struct ip *)(mtod(m_head, char *) + ip_off); - tp = (struct tcphdr *)(mtod(m_head, char *) + poff); - if (do_tso) { - ip->ip_len = htons(m_head->m_pkthdr.tso_segsz + - (ip->ip_hl << 2) + - (tp->th_off << 2)); - ip->ip_sum = 0; - /* - * The pseudo TCP checksum does not include TCP - * payload length so driver should recompute - * the checksum here what hardware expect to - * see. This is adherence of Microsoft's Large - * Send specification. - */ - tp->th_sum = in_pseudo(ip->ip_src.s_addr, - ip->ip_dst.s_addr, htons(IPPROTO_TCP)); - } - } else if (m_head->m_pkthdr.csum_flags & CSUM_UDP) { - if (m_head->m_len < poff + sizeof(struct udphdr)) { - m_head = m_pullup(m_head, poff + - sizeof(struct udphdr)); - if (m_head == NULL) { - *m_headp = NULL; - return (ENOBUFS); - } - } - ip = (struct ip *)(mtod(m_head, char *) + ip_off); - } - *m_headp = m_head; - } - - /* - * Map the packet for DMA - * - * Capture the first descriptor index, - * this descriptor will have the index - * of the EOP which is the only one that - * now gets a DONE bit writeback. - */ - first = txr->next_avail_desc; - tx_buffer = &txr->tx_buffers[first]; - tx_buffer_mapped = tx_buffer; - map = tx_buffer->map; - -retry: - error = bus_dmamap_load_mbuf_sg(txr->txtag, map, - *m_headp, segs, &nsegs, BUS_DMA_NOWAIT); - - /* - * There are two types of errors we can (try) to handle: - * - EFBIG means the mbuf chain was too long and bus_dma ran - * out of segments. Defragment the mbuf chain and try again. - * - ENOMEM means bus_dma could not obtain enough bounce buffers - * at this point in time. Defer sending and try again later. - * All other errors, in particular EINVAL, are fatal and prevent the - * mbuf chain from ever going through. Drop it and report error. - */ - if (error == EFBIG && remap) { - struct mbuf *m; - - m = m_collapse(*m_headp, M_NOWAIT, EM_MAX_SCATTER); - if (m == NULL) { - adapter->mbuf_defrag_failed++; - m_freem(*m_headp); - *m_headp = NULL; - return (ENOBUFS); - } - *m_headp = m; - - /* Try it again, but only once */ - remap = FALSE; - goto retry; - } else if (error != 0) { - adapter->no_tx_dma_setup++; - m_freem(*m_headp); - *m_headp = NULL; - return (error); - } - - /* - * TSO Hardware workaround, if this packet is not - * TSO, and is only a single descriptor long, and - * it follows a TSO burst, then we need to add a - * sentinel descriptor to prevent premature writeback. - */ - if ((!do_tso) && (txr->tx_tso == TRUE)) { - if (nsegs == 1) - tso_desc = TRUE; - txr->tx_tso = FALSE; - } - - if (txr->tx_avail < (nsegs + EM_MAX_SCATTER)) { - txr->no_desc_avail++; - bus_dmamap_unload(txr->txtag, map); - return (ENOBUFS); - } - m_head = *m_headp; - - /* Do hardware assists */ - if (m_head->m_pkthdr.csum_flags & CSUM_TSO) { - em_tso_setup(txr, m_head, ip_off, ip, tp, - &txd_upper, &txd_lower); - /* we need to make a final sentinel transmit desc */ - tso_desc = TRUE; - } else if (m_head->m_pkthdr.csum_flags & CSUM_OFFLOAD) - em_transmit_checksum_setup(txr, m_head, - ip_off, ip, &txd_upper, &txd_lower); - - if (m_head->m_flags & M_VLANTAG) { - /* Set the vlan id. */ - txd_upper |= htole16(if_getvtag(m_head)) << 16; - /* Tell hardware to add tag */ - txd_lower |= htole32(E1000_TXD_CMD_VLE); - } - - i = txr->next_avail_desc; - - /* Set up our transmit descriptors */ - for (j = 0; j < nsegs; j++) { - bus_size_t seg_len; - bus_addr_t seg_addr; - - tx_buffer = &txr->tx_buffers[i]; - ctxd = &txr->tx_base[i]; - seg_addr = segs[j].ds_addr; - seg_len = segs[j].ds_len; - /* - ** TSO Workaround: - ** If this is the last descriptor, we want to - ** split it so we have a small final sentinel - */ - if (tso_desc && (j == (nsegs - 1)) && (seg_len > 8)) { - seg_len -= TSO_WORKAROUND; - ctxd->buffer_addr = htole64(seg_addr); - ctxd->lower.data = htole32( - adapter->txd_cmd | txd_lower | seg_len); - ctxd->upper.data = htole32(txd_upper); - if (++i == adapter->num_tx_desc) - i = 0; - - /* Now make the sentinel */ - txr->tx_avail--; - ctxd = &txr->tx_base[i]; - tx_buffer = &txr->tx_buffers[i]; - ctxd->buffer_addr = - htole64(seg_addr + seg_len); - ctxd->lower.data = htole32( - adapter->txd_cmd | txd_lower | TSO_WORKAROUND); - ctxd->upper.data = - htole32(txd_upper); - last = i; - if (++i == adapter->num_tx_desc) - i = 0; - } else { - ctxd->buffer_addr = htole64(seg_addr); - ctxd->lower.data = htole32( - adapter->txd_cmd | txd_lower | seg_len); - ctxd->upper.data = htole32(txd_upper); - last = i; - if (++i == adapter->num_tx_desc) - i = 0; - } - tx_buffer->m_head = NULL; - tx_buffer->next_eop = -1; - } - - txr->next_avail_desc = i; - txr->tx_avail -= nsegs; - - tx_buffer->m_head = m_head; - /* - ** Here we swap the map so the last descriptor, - ** which gets the completion interrupt has the - ** real map, and the first descriptor gets the - ** unused map from this descriptor. - */ - tx_buffer_mapped->map = tx_buffer->map; - tx_buffer->map = map; - bus_dmamap_sync(txr->txtag, map, BUS_DMASYNC_PREWRITE); - - /* - * Last Descriptor of Packet - * needs End Of Packet (EOP) - * and Report Status (RS) - */ - ctxd->lower.data |= - htole32(E1000_TXD_CMD_EOP | E1000_TXD_CMD_RS); - /* - * Keep track in the first buffer which - * descriptor will be written back - */ - tx_buffer = &txr->tx_buffers[first]; - tx_buffer->next_eop = last; - - /* - * Advance the Transmit Descriptor Tail (TDT), this tells the E1000 - * that this frame is available to transmit. - */ - bus_dmamap_sync(txr->txdma.dma_tag, txr->txdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - E1000_WRITE_REG(&adapter->hw, E1000_TDT(txr->me), i); - - return (0); -} - -static void -em_set_promisc(struct adapter *adapter) -{ - if_t ifp = adapter->ifp; + struct adapter *adapter = iflib_get_softc(ctx); u32 reg_rctl; + em_disable_promisc(ctx); + reg_rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); - if (if_getflags(ifp) & IFF_PROMISC) { + if (flags & IFF_PROMISC) { reg_rctl |= (E1000_RCTL_UPE | E1000_RCTL_MPE); /* Turn this on if you want to see bad packets */ if (em_debug_sbp) reg_rctl |= E1000_RCTL_SBP; E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); - } else if (if_getflags(ifp) & IFF_ALLMULTI) { + } else if (flags & IFF_ALLMULTI) { reg_rctl |= E1000_RCTL_MPE; reg_rctl &= ~E1000_RCTL_UPE; E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); } + return (0); } static void -em_disable_promisc(struct adapter *adapter) +em_disable_promisc(if_ctx_t ctx) { - if_t ifp = adapter->ifp; + struct adapter *adapter = iflib_get_softc(ctx); + struct ifnet *ifp = iflib_get_ifp(ctx); u32 reg_rctl; int mcnt = 0; @@ -2251,9 +1524,10 @@ em_disable_promisc(struct adapter *adapter) **********************************************************************/ static void -em_set_multi(struct adapter *adapter) +em_if_multi_set(if_ctx_t ctx) { - if_t ifp = adapter->ifp; + struct adapter *adapter = iflib_get_softc(ctx); + struct ifnet *ifp = iflib_get_ifp(ctx); u32 reg_rctl = 0; u8 *mta; /* Multicast array memory */ int mcnt = 0; @@ -2302,17 +1576,14 @@ em_set_multi(struct adapter *adapter) **********************************************************************/ static void -em_local_timer(void *arg) +em_if_timer(if_ctx_t ctx, uint16_t qid) { - struct adapter *adapter = arg; - if_t ifp = adapter->ifp; - struct tx_ring *txr = adapter->tx_rings; - struct rx_ring *rxr = adapter->rx_rings; - u32 trigger = 0; + struct adapter *adapter = iflib_get_softc(ctx); + struct em_rx_queue *que; + int i; + int trigger = 0; - EM_CORE_LOCK_ASSERT(adapter); - - em_update_link_status(adapter); + em_if_update_admin_status(ctx); em_update_stats_counters(adapter); /* Reset LAA into RAR[0] on 82571 */ @@ -2320,53 +1591,26 @@ em_local_timer(void *arg) e1000_get_laa_state_82571(&adapter->hw)) e1000_rar_set(&adapter->hw, adapter->hw.mac.addr, 0); - /* Mask to use in the irq trigger */ - if (adapter->msix_mem) { - for (int i = 0; i < adapter->num_queues; i++, rxr++) - trigger |= rxr->ims; - rxr = adapter->rx_rings; - } else - trigger = E1000_ICS_RXDMT0; + if (adapter->hw.mac.type < em_mac_min) + lem_smartspeed(adapter); - /* - ** Check on the state of the TX queue(s), this - ** can be done without the lock because its RO - ** and the HUNG state will be static if set. - */ - for (int i = 0; i < adapter->num_queues; i++, txr++) { - if (txr->busy == EM_TX_HUNG) - goto hung; - if (txr->busy >= EM_TX_MAXTRIES) - txr->busy = EM_TX_HUNG; - /* Schedule a TX tasklet if needed */ - if (txr->tx_avail <= EM_MAX_SCATTER) - taskqueue_enqueue(txr->tq, &txr->tx_task); + /* Mask to use in the irq trigger */ + if (adapter->intr_type == IFLIB_INTR_MSIX) { + for (i = 0, que = adapter->rx_queues; i < adapter->rx_num_queues; i++, que++) + trigger |= que->eims; + } else { + trigger = E1000_ICS_RXDMT0; } - - callout_reset(&adapter->timer, hz, em_local_timer, adapter); -#ifndef DEVICE_POLLING - /* Trigger an RX interrupt to guarantee mbuf refresh */ - E1000_WRITE_REG(&adapter->hw, E1000_ICS, trigger); -#endif - return; -hung: - /* Looks like we're hung */ - device_printf(adapter->dev, "Watchdog timeout Queue[%d]-- resetting\n", - txr->me); - em_print_debug_info(adapter); - if_setdrvflagbits(ifp, 0, IFF_DRV_RUNNING); - adapter->watchdog_events++; - em_init_locked(adapter); } static void -em_update_link_status(struct adapter *adapter) +em_if_update_admin_status(if_ctx_t ctx) { + struct adapter *adapter = iflib_get_softc(ctx); struct e1000_hw *hw = &adapter->hw; - if_t ifp = adapter->ifp; - device_t dev = adapter->dev; - struct tx_ring *txr = adapter->tx_rings; + struct ifnet *ifp = iflib_get_ifp(ctx); + device_t dev = iflib_get_dev(ctx); u32 link_check = 0; /* Get the cached link value or read phy for real */ @@ -2380,8 +1624,9 @@ em_update_link_status(struct adapter *adapter) link_check = !hw->mac.get_link_status; if (link_check) /* ESB2 fix */ e1000_cfg_on_link_up(hw); - } else + } else { link_check = TRUE; + } break; case e1000_media_type_fiber: e1000_check_for_link(hw); @@ -2401,18 +1646,6 @@ em_update_link_status(struct adapter *adapter) if (link_check && (adapter->link_active == 0)) { e1000_get_speed_and_duplex(hw, &adapter->link_speed, &adapter->link_duplex); - /* - ** There have proven to be problems with TSO when not - ** at full gigabit speed, so disable the assist automatically - ** when at lower speeds. -jfv - */ - if (adapter->link_speed != SPEED_1000) { - if_sethwassistbits(ifp, 0, CSUM_TSO); - if_setcapenablebit(ifp, 0, IFCAP_TSO4); - if_setcapabilitiesbit(ifp, 0, IFCAP_TSO4); - - } - /* Check if we must disable SPEED_MODE bit on PCI-E */ if ((adapter->link_speed != SPEED_1000) && ((hw->mac.type == e1000_82571) || @@ -2430,7 +1663,8 @@ em_update_link_status(struct adapter *adapter) adapter->link_active = 1; adapter->smartspeed = 0; if_setbaudrate(ifp, adapter->link_speed * 1000000); - if_link_state_change(ifp, LINK_STATE_UP); + iflib_link_state_change(ctx, LINK_STATE_UP, ifp->if_baudrate); + printf("Link state changed to up\n"); } else if (!link_check && (adapter->link_active == 1)) { if_setbaudrate(ifp, 0); adapter->link_speed = 0; @@ -2438,11 +1672,11 @@ em_update_link_status(struct adapter *adapter) if (bootverbose) device_printf(dev, "Link is Down\n"); adapter->link_active = 0; - /* Link down, disable hang detection */ - for (int i = 0; i < adapter->num_queues; i++, txr++) - txr->busy = EM_TX_IDLE; - if_link_state_change(ifp, LINK_STATE_DOWN); + iflib_link_state_change(ctx, LINK_STATE_DOWN, ifp->if_baudrate); + printf("link state changed to down\n"); } + + E1000_WRITE_REG(&adapter->hw, E1000_IMS, EM_MSIX_LINK | E1000_IMS_LSC); } /********************************************************************* @@ -2455,35 +1689,15 @@ em_update_link_status(struct adapter *adapter) **********************************************************************/ static void -em_stop(void *arg) +em_if_stop(if_ctx_t ctx) { - struct adapter *adapter = arg; - if_t ifp = adapter->ifp; - struct tx_ring *txr = adapter->tx_rings; - - EM_CORE_LOCK_ASSERT(adapter); + struct adapter *adapter = iflib_get_softc(ctx); INIT_DEBUGOUT("em_stop: begin"); - - em_disable_intr(adapter); - callout_stop(&adapter->timer); - - /* Tell the stack that the interface is no longer active */ - if_setdrvflagbits(ifp, IFF_DRV_OACTIVE, IFF_DRV_RUNNING); - - /* Disarm Hang Detection. */ - for (int i = 0; i < adapter->num_queues; i++, txr++) { - EM_TX_LOCK(txr); - txr->busy = EM_TX_IDLE; - EM_TX_UNLOCK(txr); - } - - /* I219 needs some special flushing to avoid hangs */ - if (adapter->hw.mac.type == e1000_pch_spt) - em_flush_desc_rings(adapter); - + e1000_reset_hw(&adapter->hw); - E1000_WRITE_REG(&adapter->hw, E1000_WUC, 0); + if (adapter->hw.mac.type >= e1000_82544) + E1000_WRITE_REG(&adapter->hw, E1000_WUC, 0); e1000_led_off(&adapter->hw); e1000_cleanup_led(&adapter->hw); @@ -2496,12 +1710,12 @@ em_stop(void *arg) * **********************************************************************/ static void -em_identify_hardware(struct adapter *adapter) +em_identify_hardware(if_ctx_t ctx) { - device_t dev = adapter->dev; - + device_t dev = iflib_get_dev(ctx); + struct adapter *adapter = iflib_get_softc(ctx); + /* Make sure our PCI config space has the necessary stuff set */ - pci_enable_busmaster(dev); adapter->hw.bus.pci_cmd_word = pci_read_config(dev, PCIR_COMMAND, 2); /* Save off the information about this board */ @@ -2521,10 +1735,11 @@ em_identify_hardware(struct adapter *adapter) } static int -em_allocate_pci_resources(struct adapter *adapter) +em_allocate_pci_resources(if_ctx_t ctx) { - device_t dev = adapter->dev; - int rid; + struct adapter *adapter = iflib_get_softc(ctx); + device_t dev = iflib_get_dev(ctx); + int rid, val; rid = PCIR_BAR(0); adapter->memory = bus_alloc_resource_any(dev, SYS_RES_MEMORY, @@ -2539,489 +1754,388 @@ em_allocate_pci_resources(struct adapter *adapter) rman_get_bushandle(adapter->memory); adapter->hw.hw_addr = (u8 *)&adapter->osdep.mem_bus_space_handle; + /* Only older adapters use IO mapping */ + if (adapter->hw.mac.type < em_mac_min && + adapter->hw.mac.type > e1000_82543) { + /* Figure our where our IO BAR is ? */ + for (rid = PCIR_BAR(0); rid < PCIR_CIS;) { + val = pci_read_config(dev, rid, 4); + if (EM_BAR_TYPE(val) == EM_BAR_TYPE_IO) { + adapter->io_rid = rid; + break; + } + rid += 4; + /* check for 64bit BAR */ + if (EM_BAR_MEM_TYPE(val) == EM_BAR_MEM_TYPE_64BIT) + rid += 4; + } + if (rid >= PCIR_CIS) { + device_printf(dev, "Unable to locate IO BAR\n"); + return (ENXIO); + } + adapter->ioport = bus_alloc_resource_any(dev, + SYS_RES_IOPORT, &adapter->io_rid, RF_ACTIVE); + if (adapter->ioport == NULL) { + device_printf(dev, "Unable to allocate bus resource: " + "ioport\n"); + return (ENXIO); + } + adapter->hw.io_base = 0; + adapter->osdep.io_bus_space_tag = + rman_get_bustag(adapter->ioport); + adapter->osdep.io_bus_space_handle = + rman_get_bushandle(adapter->ioport); + } + adapter->hw.back = &adapter->osdep; return (0); } -/********************************************************************* - * - * Setup the Legacy or MSI Interrupt handler - * - **********************************************************************/ -int -em_allocate_legacy(struct adapter *adapter) -{ - device_t dev = adapter->dev; - struct tx_ring *txr = adapter->tx_rings; - int error, rid = 0; - - /* Manually turn off all interrupts */ - E1000_WRITE_REG(&adapter->hw, E1000_IMC, 0xffffffff); - - if (adapter->msix == 1) /* using MSI */ - rid = 1; - /* We allocate a single interrupt resource */ - adapter->res = bus_alloc_resource_any(dev, - SYS_RES_IRQ, &rid, RF_SHAREABLE | RF_ACTIVE); - if (adapter->res == NULL) { - device_printf(dev, "Unable to allocate bus resource: " - "interrupt\n"); - return (ENXIO); - } - - /* - * Allocate a fast interrupt and the associated - * deferred processing contexts. - */ - TASK_INIT(&adapter->que_task, 0, em_handle_que, adapter); - adapter->tq = taskqueue_create_fast("em_taskq", M_NOWAIT, - taskqueue_thread_enqueue, &adapter->tq); - taskqueue_start_threads(&adapter->tq, 1, PI_NET, "%s que", - device_get_nameunit(adapter->dev)); - /* Use a TX only tasklet for local timer */ - TASK_INIT(&txr->tx_task, 0, em_handle_tx, txr); - txr->tq = taskqueue_create_fast("em_txq", M_NOWAIT, - taskqueue_thread_enqueue, &txr->tq); - taskqueue_start_threads(&txr->tq, 1, PI_NET, "%s txq", - device_get_nameunit(adapter->dev)); - TASK_INIT(&adapter->link_task, 0, em_handle_link, adapter); - if ((error = bus_setup_intr(dev, adapter->res, INTR_TYPE_NET, - em_irq_fast, NULL, adapter, &adapter->tag)) != 0) { - device_printf(dev, "Failed to register fast interrupt " - "handler: %d\n", error); - taskqueue_free(adapter->tq); - adapter->tq = NULL; - return (error); - } - - return (0); -} - /********************************************************************* * * Setup the MSIX Interrupt handlers - * This is not really Multiqueue, rather - * its just separate interrupt vectors - * for TX, RX, and Link. * **********************************************************************/ -int -em_allocate_msix(struct adapter *adapter) +static int +em_if_msix_intr_assign(if_ctx_t ctx, int msix) { - device_t dev = adapter->dev; - struct tx_ring *txr = adapter->tx_rings; - struct rx_ring *rxr = adapter->rx_rings; - int error, rid, vector = 0; - int cpu_id = 0; - - - /* Make sure all interrupts are disabled */ - E1000_WRITE_REG(&adapter->hw, E1000_IMC, 0xffffffff); + struct adapter *adapter = iflib_get_softc(ctx); + struct em_rx_queue *rx_que = adapter->rx_queues; + struct em_tx_queue *tx_que = adapter->tx_queues; + int error, rid, i, vector = 0; + char buf[16]; /* First set up ring resources */ - for (int i = 0; i < adapter->num_queues; i++, rxr++, vector++) { - - /* RX ring */ - rid = vector + 1; - - rxr->res = bus_alloc_resource_any(dev, - SYS_RES_IRQ, &rid, RF_ACTIVE); - if (rxr->res == NULL) { - device_printf(dev, - "Unable to allocate bus resource: " - "RX MSIX Interrupt %d\n", i); - return (ENXIO); + for (i = 0; i < adapter->rx_num_queues; i++, rx_que++, vector++) { + rid = vector +1; + snprintf(buf, sizeof(buf), "rxq%d", i); + error = iflib_irq_alloc_generic(ctx, &rx_que->que_irq, rid, IFLIB_INTR_RX, em_msix_que, rx_que, rx_que->me, buf); + if (error) { + device_printf(iflib_get_dev(ctx), "Failed to allocate que int %d err: %d", i, error); + adapter->rx_num_queues = i + 1; + goto fail; } - if ((error = bus_setup_intr(dev, rxr->res, - INTR_TYPE_NET | INTR_MPSAFE, NULL, em_msix_rx, - rxr, &rxr->tag)) != 0) { - device_printf(dev, "Failed to register RX handler"); - return (error); - } -#if __FreeBSD_version >= 800504 - bus_describe_intr(dev, rxr->res, rxr->tag, "rx%d", i); -#endif - rxr->msix = vector; - if (em_last_bind_cpu < 0) - em_last_bind_cpu = CPU_FIRST(); - cpu_id = em_last_bind_cpu; - bus_bind_intr(dev, rxr->res, cpu_id); - - TASK_INIT(&rxr->rx_task, 0, em_handle_rx, rxr); - rxr->tq = taskqueue_create_fast("em_rxq", M_NOWAIT, - taskqueue_thread_enqueue, &rxr->tq); - taskqueue_start_threads(&rxr->tq, 1, PI_NET, "%s rxq (cpuid %d)", - device_get_nameunit(adapter->dev), cpu_id); + rx_que->msix = vector; + /* ** Set the bit to enable interrupt ** in E1000_IMS -- bits 20 and 21 ** are for RX0 and RX1, note this has ** NOTHING to do with the MSIX vector */ - rxr->ims = 1 << (20 + i); - adapter->ims |= rxr->ims; - adapter->ivars |= (8 | rxr->msix) << (i * 4); - - em_last_bind_cpu = CPU_NEXT(em_last_bind_cpu); + if (adapter->hw.mac.type == e1000_82574) { + rx_que->eims = 1 << (20 + i); + adapter->ims |= rx_que->eims; + adapter->ivars |= (8 | rx_que->msix) << (i * 4); + } else if (adapter->hw.mac.type == e1000_82575) + rx_que->eims = E1000_EICR_TX_QUEUE0 << vector; + else + rx_que->eims = 1 << vector; } - for (int i = 0; i < adapter->num_queues; i++, txr++, vector++) { - /* TX ring */ + for (i = 0; i < adapter->tx_num_queues; i++, tx_que++) { rid = vector + 1; - txr->res = bus_alloc_resource_any(dev, - SYS_RES_IRQ, &rid, RF_ACTIVE); - if (txr->res == NULL) { - device_printf(dev, - "Unable to allocate bus resource: " - "TX MSIX Interrupt %d\n", i); - return (ENXIO); - } - if ((error = bus_setup_intr(dev, txr->res, - INTR_TYPE_NET | INTR_MPSAFE, NULL, em_msix_tx, - txr, &txr->tag)) != 0) { - device_printf(dev, "Failed to register TX handler"); - return (error); - } -#if __FreeBSD_version >= 800504 - bus_describe_intr(dev, txr->res, txr->tag, "tx%d", i); -#endif - txr->msix = vector; + snprintf(buf, sizeof(buf), "txq%d", i); + tx_que = &adapter->tx_queues[i]; + iflib_softirq_alloc_generic(ctx, rid, IFLIB_INTR_TX, tx_que, tx_que->me, buf); - if (em_last_bind_cpu < 0) - em_last_bind_cpu = CPU_FIRST(); - cpu_id = em_last_bind_cpu; - bus_bind_intr(dev, txr->res, cpu_id); + tx_que->msix = vector; - TASK_INIT(&txr->tx_task, 0, em_handle_tx, txr); - txr->tq = taskqueue_create_fast("em_txq", M_NOWAIT, - taskqueue_thread_enqueue, &txr->tq); - taskqueue_start_threads(&txr->tq, 1, PI_NET, "%s txq (cpuid %d)", - device_get_nameunit(adapter->dev), cpu_id); - /* + /* ** Set the bit to enable interrupt ** in E1000_IMS -- bits 22 and 23 ** are for TX0 and TX1, note this has ** NOTHING to do with the MSIX vector */ - txr->ims = 1 << (22 + i); - adapter->ims |= txr->ims; - adapter->ivars |= (8 | txr->msix) << (8 + (i * 4)); - - em_last_bind_cpu = CPU_NEXT(em_last_bind_cpu); + if (adapter->hw.mac.type < igb_mac_min) { + tx_que->eims = 1 << (22 + i); + adapter->ims |= tx_que->eims; + adapter->ivars |= (8 | tx_que->msix) << (8 + (i * 4)); + } if (adapter->hw.mac.type == e1000_82575) + tx_que->eims = E1000_EICR_TX_QUEUE0 << (i % adapter->tx_num_queues); + else + tx_que->eims = 1 << (i % adapter->tx_num_queues); } - + /* Link interrupt */ rid = vector + 1; - adapter->res = bus_alloc_resource_any(dev, - SYS_RES_IRQ, &rid, RF_SHAREABLE | RF_ACTIVE); - if (!adapter->res) { - device_printf(dev,"Unable to allocate " - "bus resource: Link interrupt [%d]\n", rid); - return (ENXIO); - } - /* Set the link handler function */ - error = bus_setup_intr(dev, adapter->res, - INTR_TYPE_NET | INTR_MPSAFE, NULL, - em_msix_link, adapter, &adapter->tag); + error = iflib_irq_alloc_generic(ctx, &adapter->irq, rid, IFLIB_INTR_ADMIN, em_msix_link, adapter, 0, "aq"); + if (error) { - adapter->res = NULL; - device_printf(dev, "Failed to register LINK handler"); - return (error); + device_printf(iflib_get_dev(ctx), "Failed to register admin handler"); + goto fail; } -#if __FreeBSD_version >= 800504 - bus_describe_intr(dev, adapter->res, adapter->tag, "link"); -#endif adapter->linkvec = vector; - adapter->ivars |= (8 | vector) << 16; - adapter->ivars |= 0x80000000; - + if (adapter->hw.mac.type < igb_mac_min) { + adapter->ivars |= (8 | vector) << 16; + adapter->ivars |= 0x80000000; + } return (0); + fail: + iflib_irq_free(ctx, &adapter->irq); + rx_que = adapter->rx_queues; + for (int i = 0; i < adapter->rx_num_queues; i++, rx_que++) + iflib_irq_free(ctx, &rx_que->que_irq); + return (error); } - static void -em_free_pci_resources(struct adapter *adapter) +igb_configure_queues(struct adapter *adapter) { - device_t dev = adapter->dev; - struct tx_ring *txr; - struct rx_ring *rxr; - int rid; + struct e1000_hw *hw = &adapter->hw; + struct em_rx_queue *rx_que; + struct em_tx_queue *tx_que; + u32 tmp, ivar = 0, newitr = 0; + /* First turn on RSS capability */ + if (adapter->hw.mac.type != e1000_82575) + E1000_WRITE_REG(hw, E1000_GPIE, + E1000_GPIE_MSIX_MODE | E1000_GPIE_EIAME | + E1000_GPIE_PBA | E1000_GPIE_NSICR); - /* - ** Release all the queue interrupt resources: - */ - for (int i = 0; i < adapter->num_queues; i++) { - txr = &adapter->tx_rings[i]; - /* an early abort? */ - if (txr == NULL) - break; - rid = txr->msix +1; - if (txr->tag != NULL) { - bus_teardown_intr(dev, txr->res, txr->tag); - txr->tag = NULL; + /* Turn on MSIX */ + switch (adapter->hw.mac.type) { + case e1000_82580: + case e1000_i350: + case e1000_i354: + case e1000_i210: + case e1000_i211: + case e1000_vfadapt: + case e1000_vfadapt_i350: + /* RX entries */ + for (int i = 0; i < adapter->rx_num_queues; i++) { + u32 index = i >> 1; + ivar = E1000_READ_REG_ARRAY(hw, E1000_IVAR0, index); + rx_que = &adapter->rx_queues[i]; + if (i & 1) { + ivar &= 0xFF00FFFF; + ivar |= (rx_que->msix | E1000_IVAR_VALID) << 16; + } else { + ivar &= 0xFFFFFF00; + ivar |= rx_que->msix | E1000_IVAR_VALID; + } + E1000_WRITE_REG_ARRAY(hw, E1000_IVAR0, index, ivar); } - if (txr->res != NULL) - bus_release_resource(dev, SYS_RES_IRQ, - rid, txr->res); - - rxr = &adapter->rx_rings[i]; - /* an early abort? */ - if (rxr == NULL) - break; - rid = rxr->msix +1; - if (rxr->tag != NULL) { - bus_teardown_intr(dev, rxr->res, rxr->tag); - rxr->tag = NULL; + /* TX entries */ + for (int i = 0; i < adapter->tx_num_queues; i++) { + u32 index = i >> 1; + ivar = E1000_READ_REG_ARRAY(hw, E1000_IVAR0, index); + tx_que = &adapter->tx_queues[i]; + if (i & 1) { + ivar &= 0x00FFFFFF; + ivar |= (tx_que->msix | E1000_IVAR_VALID) << 24; + } else { + ivar &= 0xFFFF00FF; + ivar |= (tx_que->msix | E1000_IVAR_VALID) << 8; + } + E1000_WRITE_REG_ARRAY(hw, E1000_IVAR0, index, ivar); + adapter->que_mask |= tx_que->eims; } - if (rxr->res != NULL) - bus_release_resource(dev, SYS_RES_IRQ, - rid, rxr->res); + + /* And for the link interrupt */ + ivar = (adapter->linkvec | E1000_IVAR_VALID) << 8; + adapter->link_mask = 1 << adapter->linkvec; + E1000_WRITE_REG(hw, E1000_IVAR_MISC, ivar); + break; + case e1000_82576: + /* RX entries */ + for (int i = 0; i < adapter->rx_num_queues; i++) { + u32 index = i & 0x7; /* Each IVAR has two entries */ + ivar = E1000_READ_REG_ARRAY(hw, E1000_IVAR0, index); + rx_que = &adapter->rx_queues[i]; + if (i < 8) { + ivar &= 0xFFFFFF00; + ivar |= rx_que->msix | E1000_IVAR_VALID; + } else { + ivar &= 0xFF00FFFF; + ivar |= (rx_que->msix | E1000_IVAR_VALID) << 16; + } + E1000_WRITE_REG_ARRAY(hw, E1000_IVAR0, index, ivar); + adapter->que_mask |= rx_que->eims; + } + /* TX entries */ + for (int i = 0; i < adapter->tx_num_queues; i++) { + u32 index = i & 0x7; /* Each IVAR has two entries */ + ivar = E1000_READ_REG_ARRAY(hw, E1000_IVAR0, index); + tx_que = &adapter->tx_queues[i]; + if (i < 8) { + ivar &= 0xFFFF00FF; + ivar |= (tx_que->msix | E1000_IVAR_VALID) << 8; + } else { + ivar &= 0x00FFFFFF; + ivar |= (tx_que->msix | E1000_IVAR_VALID) << 24; + } + E1000_WRITE_REG_ARRAY(hw, E1000_IVAR0, index, ivar); + adapter->que_mask |= tx_que->eims; + } + + /* And for the link interrupt */ + ivar = (adapter->linkvec | E1000_IVAR_VALID) << 8; + adapter->link_mask = 1 << adapter->linkvec; + E1000_WRITE_REG(hw, E1000_IVAR_MISC, ivar); + break; + + case e1000_82575: + /* enable MSI-X support*/ + tmp = E1000_READ_REG(hw, E1000_CTRL_EXT); + tmp |= E1000_CTRL_EXT_PBA_CLR; + /* Auto-Mask interrupts upon ICR read. */ + tmp |= E1000_CTRL_EXT_EIAME; + tmp |= E1000_CTRL_EXT_IRCA; + E1000_WRITE_REG(hw, E1000_CTRL_EXT, tmp); + + /* Queues */ + for (int i = 0; i < adapter->rx_num_queues; i++) { + rx_que = &adapter->rx_queues[i]; + tmp = E1000_EICR_RX_QUEUE0 << i; + tmp |= E1000_EICR_TX_QUEUE0 << i; + rx_que->eims = tmp; + E1000_WRITE_REG_ARRAY(hw, E1000_MSIXBM(0), + i, rx_que->eims); + adapter->que_mask |= rx_que->eims; + } + + /* Link */ + E1000_WRITE_REG(hw, E1000_MSIXBM(adapter->linkvec), + E1000_EIMS_OTHER); + adapter->link_mask |= E1000_EIMS_OTHER; + default: + break; } - if (adapter->linkvec) /* we are doing MSIX */ - rid = adapter->linkvec + 1; + /* Set the starting interrupt rate */ + if (em_max_interrupt_rate > 0) + newitr = (4000000 / em_max_interrupt_rate) & 0x7FFC; + + if (hw->mac.type == e1000_82575) + newitr |= newitr << 16; else - (adapter->msix != 0) ? (rid = 1):(rid = 0); + newitr |= E1000_EITR_CNT_IGNR; - if (adapter->tag != NULL) { - bus_teardown_intr(dev, adapter->res, adapter->tag); - adapter->tag = NULL; + for (int i = 0; i < adapter->rx_num_queues; i++) { + rx_que = &adapter->rx_queues[i]; + E1000_WRITE_REG(hw, E1000_EITR(rx_que->msix), newitr); } - if (adapter->res != NULL) - bus_release_resource(dev, SYS_RES_IRQ, rid, adapter->res); - - - if (adapter->msix) - pci_release_msi(dev); - - if (adapter->msix_mem != NULL) - bus_release_resource(dev, SYS_RES_MEMORY, - PCIR_BAR(EM_MSIX_BAR), adapter->msix_mem); - - if (adapter->memory != NULL) - bus_release_resource(dev, SYS_RES_MEMORY, - PCIR_BAR(0), adapter->memory); - - if (adapter->flash != NULL) - bus_release_resource(dev, SYS_RES_MEMORY, - EM_FLASH, adapter->flash); + return; } -/* - * Setup MSI or MSI/X - */ -static int -em_setup_msix(struct adapter *adapter) +static void +em_free_pci_resources(if_ctx_t ctx) { - device_t dev = adapter->dev; - int val; + struct adapter *adapter = iflib_get_softc(ctx); + struct em_rx_queue *que = adapter->rx_queues; + device_t dev = iflib_get_dev(ctx); - /* Nearly always going to use one queue */ - adapter->num_queues = 1; + /* Release all msix queue resources */ + if (adapter->intr_type == IFLIB_INTR_MSIX) + iflib_irq_free(ctx, &adapter->irq); - /* - ** Try using MSI-X for Hartwell adapters - */ - if ((adapter->hw.mac.type == e1000_82574) && - (em_enable_msix == TRUE)) { -#ifdef EM_MULTIQUEUE - adapter->num_queues = (em_num_queues == 1) ? 1 : 2; - if (adapter->num_queues > 1) - em_enable_vectors_82574(adapter); -#endif - /* Map the MSIX BAR */ - int rid = PCIR_BAR(EM_MSIX_BAR); - adapter->msix_mem = bus_alloc_resource_any(dev, - SYS_RES_MEMORY, &rid, RF_ACTIVE); - if (adapter->msix_mem == NULL) { - /* May not be enabled */ - device_printf(adapter->dev, - "Unable to map MSIX table \n"); - goto msi; - } - val = pci_msix_count(dev); - -#ifdef EM_MULTIQUEUE - /* We need 5 vectors in the multiqueue case */ - if (adapter->num_queues > 1 ) { - if (val >= 5) - val = 5; - else { - adapter->num_queues = 1; - device_printf(adapter->dev, - "Insufficient MSIX vectors for >1 queue, " - "using single queue...\n"); - goto msix_one; - } - } else { -msix_one: -#endif - if (val >= 3) - val = 3; - else { - device_printf(adapter->dev, - "Insufficient MSIX vectors, using MSI\n"); - goto msi; - } -#ifdef EM_MULTIQUEUE - } -#endif - - if ((pci_alloc_msix(dev, &val) == 0)) { - device_printf(adapter->dev, - "Using MSIX interrupts " - "with %d vectors\n", val); - return (val); - } - - /* - ** If MSIX alloc failed or provided us with - ** less than needed, free and fall through to MSI - */ - pci_release_msi(dev); + for (int i = 0; i < adapter->rx_num_queues; i++, que++) { + iflib_irq_free(ctx, &que->que_irq); } -msi: - if (adapter->msix_mem != NULL) { + + + /* First release all the interrupt resources */ + if (adapter->memory != NULL) { bus_release_resource(dev, SYS_RES_MEMORY, - PCIR_BAR(EM_MSIX_BAR), adapter->msix_mem); - adapter->msix_mem = NULL; + PCIR_BAR(0), adapter->memory); + adapter->memory = NULL; + } + + if (adapter->flash != NULL) { + bus_release_resource(dev, SYS_RES_MEMORY, + EM_FLASH, adapter->flash); + adapter->flash = NULL; + } + if (adapter->ioport != NULL) + bus_release_resource(dev, SYS_RES_IOPORT, + adapter->io_rid, adapter->ioport); +} + +/* Setup MSI or MSI/X */ +static int +em_setup_msix(if_ctx_t ctx) +{ + struct adapter *adapter = iflib_get_softc(ctx); + + if (adapter->hw.mac.type == e1000_82574) { + em_enable_vectors_82574(ctx); } - val = 1; - if (pci_alloc_msi(dev, &val) == 0) { - device_printf(adapter->dev, "Using an MSI interrupt\n"); - return (val); - } - /* Should only happen due to manual configuration */ - device_printf(adapter->dev,"No MSI/MSIX using a Legacy IRQ\n"); return (0); } - -/* -** The 3 following flush routines are used as a workaround in the -** I219 client parts and only for them. -** -** em_flush_tx_ring - remove all descriptors from the tx_ring -** -** We want to clear all pending descriptors from the TX ring. -** zeroing happens when the HW reads the regs. We assign the ring itself as -** the data of the next descriptor. We don't care about the data we are about -** to reset the HW. -*/ -static void -em_flush_tx_ring(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - struct tx_ring *txr = adapter->tx_rings; - struct e1000_tx_desc *txd; - u32 tctl, txd_lower = E1000_TXD_CMD_IFCS; - u16 size = 512; - - tctl = E1000_READ_REG(hw, E1000_TCTL); - E1000_WRITE_REG(hw, E1000_TCTL, tctl | E1000_TCTL_EN); - - txd = &txr->tx_base[txr->next_avail_desc++]; - if (txr->next_avail_desc == adapter->num_tx_desc) - txr->next_avail_desc = 0; - - /* Just use the ring as a dummy buffer addr */ - txd->buffer_addr = txr->txdma.dma_paddr; - txd->lower.data = htole32(txd_lower | size); - txd->upper.data = 0; - - /* flush descriptors to memory before notifying the HW */ - wmb(); - - E1000_WRITE_REG(hw, E1000_TDT(0), txr->next_avail_desc); - mb(); - usec_delay(250); -} - -/* -** em_flush_rx_ring - remove all descriptors from the rx_ring -** -** Mark all descriptors in the RX ring as consumed and disable the rx ring -*/ -static void -em_flush_rx_ring(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - u32 rctl, rxdctl; - - rctl = E1000_READ_REG(hw, E1000_RCTL); - E1000_WRITE_REG(hw, E1000_RCTL, rctl & ~E1000_RCTL_EN); - E1000_WRITE_FLUSH(hw); - usec_delay(150); - - rxdctl = E1000_READ_REG(hw, E1000_RXDCTL(0)); - /* zero the lower 14 bits (prefetch and host thresholds) */ - rxdctl &= 0xffffc000; - /* - * update thresholds: prefetch threshold to 31, host threshold to 1 - * and make sure the granularity is "descriptors" and not "cache lines" - */ - rxdctl |= (0x1F | (1 << 8) | E1000_RXDCTL_THRESH_UNIT_DESC); - E1000_WRITE_REG(hw, E1000_RXDCTL(0), rxdctl); - - /* momentarily enable the RX ring for the changes to take effect */ - E1000_WRITE_REG(hw, E1000_RCTL, rctl | E1000_RCTL_EN); - E1000_WRITE_FLUSH(hw); - usec_delay(150); - E1000_WRITE_REG(hw, E1000_RCTL, rctl & ~E1000_RCTL_EN); -} - -/* -** em_flush_desc_rings - remove all descriptors from the descriptor rings -** -** In i219, the descriptor rings must be emptied before resetting the HW -** or before changing the device state to D3 during runtime (runtime PM). -** -** Failure to do this will cause the HW to enter a unit hang state which can -** only be released by PCI reset on the device -** -*/ -static void -em_flush_desc_rings(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - device_t dev = adapter->dev; - u16 hang_state; - u32 fext_nvm11, tdlen; - - /* First, disable MULR fix in FEXTNVM11 */ - fext_nvm11 = E1000_READ_REG(hw, E1000_FEXTNVM11); - fext_nvm11 |= E1000_FEXTNVM11_DISABLE_MULR_FIX; - E1000_WRITE_REG(hw, E1000_FEXTNVM11, fext_nvm11); - - /* do nothing if we're not in faulty state, or if the queue is empty */ - tdlen = E1000_READ_REG(hw, E1000_TDLEN(0)); - hang_state = pci_read_config(dev, PCICFG_DESC_RING_STATUS, 2); - if (!(hang_state & FLUSH_DESC_REQUIRED) || !tdlen) - return; - em_flush_tx_ring(adapter); - - /* recheck, maybe the fault is caused by the rx ring */ - hang_state = pci_read_config(dev, PCICFG_DESC_RING_STATUS, 2); - if (hang_state & FLUSH_DESC_REQUIRED) - em_flush_rx_ring(adapter); -} - - /********************************************************************* * * Initialize the hardware to a configuration * as specified by the adapter structure. * **********************************************************************/ + static void -em_reset(struct adapter *adapter) +lem_smartspeed(struct adapter *adapter) { - device_t dev = adapter->dev; - if_t ifp = adapter->ifp; + u16 phy_tmp; + + if (adapter->link_active || (adapter->hw.phy.type != e1000_phy_igp) || + adapter->hw.mac.autoneg == 0 || + (adapter->hw.phy.autoneg_advertised & ADVERTISE_1000_FULL) == 0) + return; + + if (adapter->smartspeed == 0) { + /* If Master/Slave config fault is asserted twice, + * we assume back-to-back */ + e1000_read_phy_reg(&adapter->hw, PHY_1000T_STATUS, &phy_tmp); + if (!(phy_tmp & SR_1000T_MS_CONFIG_FAULT)) + return; + e1000_read_phy_reg(&adapter->hw, PHY_1000T_STATUS, &phy_tmp); + if (phy_tmp & SR_1000T_MS_CONFIG_FAULT) { + e1000_read_phy_reg(&adapter->hw, + PHY_1000T_CTRL, &phy_tmp); + if(phy_tmp & CR_1000T_MS_ENABLE) { + phy_tmp &= ~CR_1000T_MS_ENABLE; + e1000_write_phy_reg(&adapter->hw, + PHY_1000T_CTRL, phy_tmp); + adapter->smartspeed++; + if(adapter->hw.mac.autoneg && + !e1000_copper_link_autoneg(&adapter->hw) && + !e1000_read_phy_reg(&adapter->hw, + PHY_CONTROL, &phy_tmp)) { + phy_tmp |= (MII_CR_AUTO_NEG_EN | + MII_CR_RESTART_AUTO_NEG); + e1000_write_phy_reg(&adapter->hw, + PHY_CONTROL, phy_tmp); + } + } + } + return; + } else if(adapter->smartspeed == EM_SMARTSPEED_DOWNSHIFT) { + /* If still no link, perhaps using 2/3 pair cable */ + e1000_read_phy_reg(&adapter->hw, PHY_1000T_CTRL, &phy_tmp); + phy_tmp |= CR_1000T_MS_ENABLE; + e1000_write_phy_reg(&adapter->hw, PHY_1000T_CTRL, phy_tmp); + if(adapter->hw.mac.autoneg && + !e1000_copper_link_autoneg(&adapter->hw) && + !e1000_read_phy_reg(&adapter->hw, PHY_CONTROL, &phy_tmp)) { + phy_tmp |= (MII_CR_AUTO_NEG_EN | + MII_CR_RESTART_AUTO_NEG); + e1000_write_phy_reg(&adapter->hw, PHY_CONTROL, phy_tmp); + } + } + /* Restart process after EM_SMARTSPEED_MAX iterations */ + if(adapter->smartspeed++ == EM_SMARTSPEED_MAX) + adapter->smartspeed = 0; +} + + +static void +em_reset(if_ctx_t ctx) +{ + device_t dev = iflib_get_dev(ctx); + struct adapter *adapter = iflib_get_softc(ctx); + struct ifnet *ifp = iflib_get_ifp(ctx); struct e1000_hw *hw = &adapter->hw; u16 rx_buffer_size; u32 pba; @@ -3156,10 +2270,6 @@ em_reset(struct adapter *adapter) break; } - /* I219 needs some special flushing to avoid hangs */ - if (hw->mac.type == e1000_pch_spt) - em_flush_desc_rings(adapter); - /* Issue a global reset */ e1000_reset_hw(hw); E1000_WRITE_REG(hw, E1000_WUC, 0); @@ -3173,7 +2283,145 @@ em_reset(struct adapter *adapter) E1000_WRITE_REG(hw, E1000_VET, ETHERTYPE_VLAN); e1000_get_phy_info(hw); e1000_check_for_link(hw); - return; +} + +#define RSSKEYLEN 10 +static void +em_initialize_rss_mapping(struct adapter *adapter) +{ + uint8_t rss_key[4 * RSSKEYLEN]; + uint32_t reta = 0; + struct e1000_hw *hw = &adapter->hw; + int i; + + /* + * Configure RSS key + */ + arc4rand(rss_key, sizeof(rss_key), 0); + for (i = 0; i < RSSKEYLEN; ++i) { + uint32_t rssrk = 0; + + rssrk = EM_RSSRK_VAL(rss_key, i); + E1000_WRITE_REG(hw,E1000_RSSRK(i), rssrk); + } + + /* + * Configure RSS redirect table in following fashion: + * (hash & ring_cnt_mask) == rdr_table[(hash & rdr_table_mask)] + */ + for (i = 0; i < sizeof(reta); ++i) { + uint32_t q; + + q = (i % adapter->rx_num_queues) << 7; + reta |= q << (8 * i); + } + + for (i = 0; i < 32; ++i) + E1000_WRITE_REG(hw, E1000_RETA(i), reta); + + E1000_WRITE_REG(hw, E1000_MRQC, E1000_MRQC_RSS_ENABLE_2Q | + E1000_MRQC_RSS_FIELD_IPV4_TCP | + E1000_MRQC_RSS_FIELD_IPV4 | + E1000_MRQC_RSS_FIELD_IPV6_TCP_EX | + E1000_MRQC_RSS_FIELD_IPV6_EX | + E1000_MRQC_RSS_FIELD_IPV6); + +} + +static void +igb_initialize_rss_mapping(struct adapter *adapter) +{ + struct e1000_hw *hw = &adapter->hw; + int i; + int queue_id; + u32 reta; + u32 rss_key[10], mrqc, shift = 0; + + /* XXX? */ + if (adapter->hw.mac.type == e1000_82575) + shift = 6; + + /* + * The redirection table controls which destination + * queue each bucket redirects traffic to. + * Each DWORD represents four queues, with the LSB + * being the first queue in the DWORD. + * + * This just allocates buckets to queues using round-robin + * allocation. + * + * NOTE: It Just Happens to line up with the default + * RSS allocation method. + */ + + /* Warning FM follows */ + reta = 0; + for (i = 0; i < 128; i++) { +#ifdef RSS + queue_id = rss_get_indirection_to_bucket(i); + /* + * If we have more queues than buckets, we'll + * end up mapping buckets to a subset of the + * queues. + * + * If we have more buckets than queues, we'll + * end up instead assigning multiple buckets + * to queues. + * + * Both are suboptimal, but we need to handle + * the case so we don't go out of bounds + * indexing arrays and such. + */ + queue_id = queue_id % adapter->rx_num_queues; +#else + queue_id = (i % adapter->rx_num_queues); +#endif + /* Adjust if required */ + queue_id = queue_id << shift; + + /* + * The low 8 bits are for hash value (n+0); + * The next 8 bits are for hash value (n+1), etc. + */ + reta = reta >> 8; + reta = reta | ( ((uint32_t) queue_id) << 24); + if ((i & 3) == 3) { + E1000_WRITE_REG(hw, E1000_RETA(i >> 2), reta); + reta = 0; + } + } + + /* Now fill in hash table */ + + /* + * MRQC: Multiple Receive Queues Command + * Set queuing to RSS control, number depends on the device. + */ + mrqc = E1000_MRQC_ENABLE_RSS_8Q; + +#ifdef RSS + /* XXX ew typecasting */ + rss_getkey((uint8_t *) &rss_key); +#else + arc4rand(&rss_key, sizeof(rss_key), 0); +#endif + for (i = 0; i < 10; i++) + E1000_WRITE_REG_ARRAY(hw, + E1000_RSSRK(0), i, rss_key[i]); + + /* + * Configure the RSS fields to hash upon. + */ + mrqc |= (E1000_MRQC_RSS_FIELD_IPV4 | + E1000_MRQC_RSS_FIELD_IPV4_TCP); + mrqc |= (E1000_MRQC_RSS_FIELD_IPV6 | + E1000_MRQC_RSS_FIELD_IPV6_TCP); + mrqc |=( E1000_MRQC_RSS_FIELD_IPV4_UDP | + E1000_MRQC_RSS_FIELD_IPV6_UDP); + mrqc |=( E1000_MRQC_RSS_FIELD_IPV6_UDP_EX | + E1000_MRQC_RSS_FIELD_IPV6_TCP_EX); + + E1000_WRITE_REG(hw, E1000_MRQC, mrqc); } /********************************************************************* @@ -3182,57 +2430,36 @@ em_reset(struct adapter *adapter) * **********************************************************************/ static int -em_setup_interface(device_t dev, struct adapter *adapter) +em_setup_interface(if_ctx_t ctx) { - if_t ifp; - + struct ifnet *ifp = iflib_get_ifp(ctx); + struct adapter *adapter = iflib_get_softc(ctx); + if_softc_ctx_t scctx = adapter->shared; + uint64_t cap = 0; + INIT_DEBUGOUT("em_setup_interface: begin"); - ifp = adapter->ifp = if_gethandle(IFT_ETHER); - if (ifp == 0) { - device_printf(dev, "can not allocate ifnet structure\n"); - return (-1); - } - if_initname(ifp, device_get_name(dev), device_get_unit(dev)); - if_setdev(ifp, dev); - if_setinitfn(ifp, em_init); - if_setsoftc(ifp, adapter); - if_setflags(ifp, IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST); - if_setioctlfn(ifp, em_ioctl); - if_setgetcounterfn(ifp, em_get_counter); - /* TSO parameters */ ifp->if_hw_tsomax = IP_MAXPACKET; /* Take m_pullup(9)'s in em_xmit() w/ TSO into acount. */ ifp->if_hw_tsomaxsegcount = EM_MAX_SCATTER - 5; ifp->if_hw_tsomaxsegsize = EM_TSO_SEG_SIZE; -#ifdef EM_MULTIQUEUE - /* Multiqueue stack interface */ - if_settransmitfn(ifp, em_mq_start); - if_setqflushfn(ifp, em_qflush); -#else - if_setstartfn(ifp, em_start); - if_setsendqlen(ifp, adapter->num_tx_desc - 1); - if_setsendqready(ifp); -#endif + /* Single Queue */ + if (adapter->tx_num_queues == 1) { + if_setsendqlen(ifp, scctx->isc_ntxd[0] - 1); + if_setsendqready(ifp); + } - ether_ifattach(ifp, adapter->hw.mac.addr); + cap = IFCAP_HWCSUM | IFCAP_VLAN_HWCSUM | IFCAP_TSO4; + cap |= IFCAP_VLAN_HWTAGGING | IFCAP_VLAN_HWTSO | IFCAP_VLAN_MTU; - if_setcapabilities(ifp, 0); - if_setcapenable(ifp, 0); - - - if_setcapabilitiesbit(ifp, IFCAP_HWCSUM | IFCAP_VLAN_HWCSUM | - IFCAP_TSO4, 0); /* * Tell the upper layer(s) we * support full VLAN capability */ if_setifheaderlen(ifp, sizeof(struct ether_vlan_header)); - if_setcapabilitiesbit(ifp, IFCAP_VLAN_HWTAGGING | IFCAP_VLAN_HWTSO | - IFCAP_VLAN_MTU, 0); - if_setcapenable(ifp, if_getcapabilities(ifp)); + if_setcapabilitiesbit(ifp, cap, 0); /* ** Don't turn this on by default, if vlans are @@ -3244,10 +2471,6 @@ em_setup_interface(device_t dev, struct adapter *adapter) */ if_setcapabilitiesbit(ifp, IFCAP_VLAN_HWFILTER,0); -#ifdef DEVICE_POLLING - if_setcapabilitiesbit(ifp, IFCAP_POLLING,0); -#endif - /* Enable only WOL MAGIC by default */ if (adapter->wol) { if_setcapabilitiesbit(ifp, IFCAP_WOL, 0); @@ -3258,396 +2481,146 @@ em_setup_interface(device_t dev, struct adapter *adapter) * Specify the media types supported by this adapter and register * callbacks to update media and link information */ - ifmedia_init(&adapter->media, IFM_IMASK, - em_media_change, em_media_status); if ((adapter->hw.phy.media_type == e1000_media_type_fiber) || (adapter->hw.phy.media_type == e1000_media_type_internal_serdes)) { u_char fiber_type = IFM_1000_SX; /* default type */ - ifmedia_add(&adapter->media, IFM_ETHER | fiber_type | IFM_FDX, - 0, NULL); - ifmedia_add(&adapter->media, IFM_ETHER | fiber_type, 0, NULL); + if (adapter->hw.mac.type == e1000_82545) + fiber_type = IFM_1000_LX; + ifmedia_add(adapter->media, IFM_ETHER | fiber_type | IFM_FDX, 0, NULL); + ifmedia_add(adapter->media, IFM_ETHER | fiber_type, 0, NULL); } else { - ifmedia_add(&adapter->media, IFM_ETHER | IFM_10_T, 0, NULL); - ifmedia_add(&adapter->media, IFM_ETHER | IFM_10_T | IFM_FDX, - 0, NULL); - ifmedia_add(&adapter->media, IFM_ETHER | IFM_100_TX, - 0, NULL); - ifmedia_add(&adapter->media, IFM_ETHER | IFM_100_TX | IFM_FDX, - 0, NULL); + ifmedia_add(adapter->media, IFM_ETHER | IFM_10_T, 0, NULL); + ifmedia_add(adapter->media, IFM_ETHER | IFM_10_T | IFM_FDX, 0, NULL); + ifmedia_add(adapter->media, IFM_ETHER | IFM_100_TX, 0, NULL); + ifmedia_add(adapter->media, IFM_ETHER | IFM_100_TX | IFM_FDX, 0, NULL); if (adapter->hw.phy.type != e1000_phy_ife) { - ifmedia_add(&adapter->media, - IFM_ETHER | IFM_1000_T | IFM_FDX, 0, NULL); - ifmedia_add(&adapter->media, - IFM_ETHER | IFM_1000_T, 0, NULL); + ifmedia_add(adapter->media, IFM_ETHER | IFM_1000_T | IFM_FDX, 0, NULL); + ifmedia_add(adapter->media, IFM_ETHER | IFM_1000_T, 0, NULL); } } - ifmedia_add(&adapter->media, IFM_ETHER | IFM_AUTO, 0, NULL); - ifmedia_set(&adapter->media, IFM_ETHER | IFM_AUTO); + ifmedia_add(adapter->media, IFM_ETHER | IFM_AUTO, 0, NULL); + ifmedia_set(adapter->media, IFM_ETHER | IFM_AUTO); return (0); } - -/* - * Manage DMA'able memory. - */ -static void -em_dmamap_cb(void *arg, bus_dma_segment_t *segs, int nseg, int error) -{ - if (error) - return; - *(bus_addr_t *) arg = segs[0].ds_addr; -} - static int -em_dma_malloc(struct adapter *adapter, bus_size_t size, - struct em_dma_alloc *dma, int mapflags) +em_if_tx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int ntxqs, int ntxqsets) { - int error; + struct adapter *adapter = iflib_get_softc(ctx); + if_softc_ctx_t scctx = adapter->shared; + int error = E1000_SUCCESS; + struct em_tx_queue *que; + int i; - error = bus_dma_tag_create(bus_get_dma_tag(adapter->dev), /* parent */ - EM_DBA_ALIGN, 0, /* alignment, bounds */ - BUS_SPACE_MAXADDR, /* lowaddr */ - BUS_SPACE_MAXADDR, /* highaddr */ - NULL, NULL, /* filter, filterarg */ - size, /* maxsize */ - 1, /* nsegments */ - size, /* maxsegsize */ - 0, /* flags */ - NULL, /* lockfunc */ - NULL, /* lockarg */ - &dma->dma_tag); - if (error) { - device_printf(adapter->dev, - "%s: bus_dma_tag_create failed: %d\n", - __func__, error); - goto fail_0; + MPASS(adapter->tx_num_queues > 0); + MPASS(adapter->tx_num_queues == ntxqsets); + + /* First allocate the top level queue structs */ + if (!(adapter->tx_queues = + (struct em_tx_queue *) malloc(sizeof(struct em_tx_queue) * + adapter->tx_num_queues, M_DEVBUF, M_NOWAIT | M_ZERO))) { + device_printf(iflib_get_dev(ctx), "Unable to allocate queue memory\n"); + return(ENOMEM); } - error = bus_dmamem_alloc(dma->dma_tag, (void**) &dma->dma_vaddr, - BUS_DMA_NOWAIT | BUS_DMA_COHERENT, &dma->dma_map); - if (error) { - device_printf(adapter->dev, - "%s: bus_dmamem_alloc(%ju) failed: %d\n", - __func__, (uintmax_t)size, error); - goto fail_2; - } + for (i = 0, que = adapter->tx_queues; i < adapter->tx_num_queues; i++, que++) { + /* Set up some basics */ + struct tx_ring *txr = &que->txr; + txr->adapter = que->adapter = adapter; + txr->que = que; + que->me = txr->me = i; - dma->dma_paddr = 0; - error = bus_dmamap_load(dma->dma_tag, dma->dma_map, dma->dma_vaddr, - size, em_dmamap_cb, &dma->dma_paddr, mapflags | BUS_DMA_NOWAIT); - if (error || dma->dma_paddr == 0) { - device_printf(adapter->dev, - "%s: bus_dmamap_load failed: %d\n", - __func__, error); - goto fail_3; - } + /* Allocate transmit buffer memory */ + if (!(txr->tx_buffers = (struct em_txbuffer *) malloc(sizeof(struct em_txbuffer) * scctx->isc_ntxd[0], M_DEVBUF, M_NOWAIT | M_ZERO))) { + device_printf(iflib_get_dev(ctx), "failed to allocate tx_buffer memory\n"); + error = ENOMEM; + goto fail; + } + /* get the virtual and physical address of the hardware queues */ + txr->tx_base = (struct e1000_tx_desc *)vaddrs[i*ntxqs]; + txr->tx_paddr = paddrs[i*ntxqs]; + + } + + device_printf(iflib_get_dev(ctx), "allocated for %d tx_queues\n", adapter->tx_num_queues); return (0); - -fail_3: - bus_dmamap_unload(dma->dma_tag, dma->dma_map); -fail_2: - bus_dmamem_free(dma->dma_tag, dma->dma_vaddr, dma->dma_map); - bus_dma_tag_destroy(dma->dma_tag); -fail_0: - dma->dma_tag = NULL; - + fail: + em_if_queues_free(ctx); return (error); } -static void -em_dma_free(struct adapter *adapter, struct em_dma_alloc *dma) -{ - if (dma->dma_tag == NULL) - return; - if (dma->dma_paddr != 0) { - bus_dmamap_sync(dma->dma_tag, dma->dma_map, - BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); - bus_dmamap_unload(dma->dma_tag, dma->dma_map); - dma->dma_paddr = 0; - } - if (dma->dma_vaddr != NULL) { - bus_dmamem_free(dma->dma_tag, dma->dma_vaddr, dma->dma_map); - dma->dma_vaddr = NULL; - } - bus_dma_tag_destroy(dma->dma_tag); - dma->dma_tag = NULL; -} - - -/********************************************************************* - * - * Allocate memory for the transmit and receive rings, and then - * the descriptors associated with each, called only once at attach. - * - **********************************************************************/ static int -em_allocate_queues(struct adapter *adapter) +em_if_rx_queues_alloc(if_ctx_t ctx, caddr_t *vaddrs, uint64_t *paddrs, int nrxqs, int nrxqsets) { - device_t dev = adapter->dev; - struct tx_ring *txr = NULL; - struct rx_ring *rxr = NULL; - int rsize, tsize, error = E1000_SUCCESS; - int txconf = 0, rxconf = 0; + struct adapter *adapter = iflib_get_softc(ctx); + int error = E1000_SUCCESS; + struct em_rx_queue *que; + int i; + MPASS(adapter->rx_num_queues > 0); + MPASS(adapter->rx_num_queues == nrxqsets); - /* Allocate the TX ring struct memory */ - if (!(adapter->tx_rings = - (struct tx_ring *) malloc(sizeof(struct tx_ring) * - adapter->num_queues, M_DEVBUF, M_NOWAIT | M_ZERO))) { - device_printf(dev, "Unable to allocate TX ring memory\n"); + /* First allocate the top level queue structs */ + if (!(adapter->rx_queues = + (struct em_rx_queue *) malloc(sizeof(struct em_rx_queue) * + adapter->rx_num_queues, M_DEVBUF, M_NOWAIT | M_ZERO))) { + device_printf(iflib_get_dev(ctx), "Unable to allocate queue memory\n"); error = ENOMEM; - goto fail; + goto fail; } - /* Now allocate the RX */ - if (!(adapter->rx_rings = - (struct rx_ring *) malloc(sizeof(struct rx_ring) * - adapter->num_queues, M_DEVBUF, M_NOWAIT | M_ZERO))) { - device_printf(dev, "Unable to allocate RX ring memory\n"); - error = ENOMEM; - goto rx_fail; - } - - tsize = roundup2(adapter->num_tx_desc * - sizeof(struct e1000_tx_desc), EM_DBA_ALIGN); - /* - * Now set up the TX queues, txconf is needed to handle the - * possibility that things fail midcourse and we need to - * undo memory gracefully - */ - for (int i = 0; i < adapter->num_queues; i++, txconf++) { + for (i = 0, que = adapter->rx_queues; i < nrxqsets; i++, que++) { /* Set up some basics */ - txr = &adapter->tx_rings[i]; - txr->adapter = adapter; - txr->me = i; + struct rx_ring *rxr = &que->rxr; + rxr->adapter = que->adapter = adapter; + rxr->que = que; + que->me = rxr->me = i; - /* Initialize the TX lock */ - snprintf(txr->mtx_name, sizeof(txr->mtx_name), "%s:tx(%d)", - device_get_nameunit(dev), txr->me); - mtx_init(&txr->tx_mtx, txr->mtx_name, NULL, MTX_DEF); - - if (em_dma_malloc(adapter, tsize, - &txr->txdma, BUS_DMA_NOWAIT)) { - device_printf(dev, - "Unable to allocate TX Descriptor memory\n"); - error = ENOMEM; - goto err_tx_desc; - } - txr->tx_base = (struct e1000_tx_desc *)txr->txdma.dma_vaddr; - bzero((void *)txr->tx_base, tsize); - - if (em_allocate_transmit_buffers(txr)) { - device_printf(dev, - "Critical Failure setting up transmit buffers\n"); - error = ENOMEM; - goto err_tx_desc; - } -#if __FreeBSD_version >= 800000 - /* Allocate a buf ring */ - txr->br = buf_ring_alloc(4096, M_DEVBUF, - M_WAITOK, &txr->tx_mtx); -#endif + /* get the virtual and physical address of the hardware queues */ + rxr->rx_base = (union e1000_rx_desc_extended *)vaddrs[i*nrxqs]; + rxr->rx_paddr = paddrs[i*nrxqs]; } - /* - * Next the RX queues... - */ - rsize = roundup2(adapter->num_rx_desc * - sizeof(union e1000_rx_desc_extended), EM_DBA_ALIGN); - for (int i = 0; i < adapter->num_queues; i++, rxconf++) { - rxr = &adapter->rx_rings[i]; - rxr->adapter = adapter; - rxr->me = i; - - /* Initialize the RX lock */ - snprintf(rxr->mtx_name, sizeof(rxr->mtx_name), "%s:rx(%d)", - device_get_nameunit(dev), txr->me); - mtx_init(&rxr->rx_mtx, rxr->mtx_name, NULL, MTX_DEF); - - if (em_dma_malloc(adapter, rsize, - &rxr->rxdma, BUS_DMA_NOWAIT)) { - device_printf(dev, - "Unable to allocate RxDescriptor memory\n"); - error = ENOMEM; - goto err_rx_desc; - } - rxr->rx_base = (union e1000_rx_desc_extended *)rxr->rxdma.dma_vaddr; - bzero((void *)rxr->rx_base, rsize); - - /* Allocate receive buffers for the ring*/ - if (em_allocate_receive_buffers(rxr)) { - device_printf(dev, - "Critical Failure setting up receive buffers\n"); - error = ENOMEM; - goto err_rx_desc; - } - } + device_printf(iflib_get_dev(ctx), "allocated for %d rx_queues\n", adapter->rx_num_queues); return (0); - -err_rx_desc: - for (rxr = adapter->rx_rings; rxconf > 0; rxr++, rxconf--) - em_dma_free(adapter, &rxr->rxdma); -err_tx_desc: - for (txr = adapter->tx_rings; txconf > 0; txr++, txconf--) - em_dma_free(adapter, &txr->txdma); - free(adapter->rx_rings, M_DEVBUF); -rx_fail: -#if __FreeBSD_version >= 800000 - buf_ring_free(txr->br, M_DEVBUF); -#endif - free(adapter->tx_rings, M_DEVBUF); fail: + em_if_queues_free(ctx); return (error); } - -/********************************************************************* - * - * Allocate memory for tx_buffer structures. The tx_buffer stores all - * the information needed to transmit a packet on the wire. This is - * called only once at attach, setup is done every reset. - * - **********************************************************************/ -static int -em_allocate_transmit_buffers(struct tx_ring *txr) -{ - struct adapter *adapter = txr->adapter; - device_t dev = adapter->dev; - struct em_txbuffer *txbuf; - int error, i; - - /* - * Setup DMA descriptor areas. - */ - if ((error = bus_dma_tag_create(bus_get_dma_tag(dev), - 1, 0, /* alignment, bounds */ - BUS_SPACE_MAXADDR, /* lowaddr */ - BUS_SPACE_MAXADDR, /* highaddr */ - NULL, NULL, /* filter, filterarg */ - EM_TSO_SIZE, /* maxsize */ - EM_MAX_SCATTER, /* nsegments */ - PAGE_SIZE, /* maxsegsize */ - 0, /* flags */ - NULL, /* lockfunc */ - NULL, /* lockfuncarg */ - &txr->txtag))) { - device_printf(dev,"Unable to allocate TX DMA tag\n"); - goto fail; - } - - if (!(txr->tx_buffers = - (struct em_txbuffer *) malloc(sizeof(struct em_txbuffer) * - adapter->num_tx_desc, M_DEVBUF, M_NOWAIT | M_ZERO))) { - device_printf(dev, "Unable to allocate tx_buffer memory\n"); - error = ENOMEM; - goto fail; - } - - /* Create the descriptor buffer dma maps */ - txbuf = txr->tx_buffers; - for (i = 0; i < adapter->num_tx_desc; i++, txbuf++) { - error = bus_dmamap_create(txr->txtag, 0, &txbuf->map); - if (error != 0) { - device_printf(dev, "Unable to create TX DMA map\n"); - goto fail; - } - } - - return 0; -fail: - /* We free all, it handles case where we are in the middle */ - em_free_transmit_structures(adapter); - return (error); -} - -/********************************************************************* - * - * Initialize a transmit ring. - * - **********************************************************************/ static void -em_setup_transmit_ring(struct tx_ring *txr) +em_if_queues_free(if_ctx_t ctx) { - struct adapter *adapter = txr->adapter; - struct em_txbuffer *txbuf; - int i; -#ifdef DEV_NETMAP - struct netmap_slot *slot; - struct netmap_adapter *na = netmap_getna(adapter->ifp); -#endif /* DEV_NETMAP */ + struct adapter *adapter = iflib_get_softc(ctx); + struct em_tx_queue *tx_que = adapter->tx_queues; + struct em_rx_queue *rx_que = adapter->rx_queues; - /* Clear the old descriptor contents */ - EM_TX_LOCK(txr); -#ifdef DEV_NETMAP - slot = netmap_reset(na, NR_TX, txr->me, 0); -#endif /* DEV_NETMAP */ + if (tx_que != NULL) { + for (int i = 0; i < adapter->tx_num_queues; i++, tx_que++) { + struct tx_ring *txr = &tx_que->txr; + if (txr->tx_buffers == NULL) + break; - bzero((void *)txr->tx_base, - (sizeof(struct e1000_tx_desc)) * adapter->num_tx_desc); - /* Reset indices */ - txr->next_avail_desc = 0; - txr->next_to_clean = 0; + free(txr->tx_buffers, M_DEVBUF); + txr->tx_buffers = NULL; + } + free(adapter->tx_queues, M_DEVBUF); + adapter->tx_queues = NULL; + } - /* Free any existing tx buffers. */ - txbuf = txr->tx_buffers; - for (i = 0; i < adapter->num_tx_desc; i++, txbuf++) { - if (txbuf->m_head != NULL) { - bus_dmamap_sync(txr->txtag, txbuf->map, - BUS_DMASYNC_POSTWRITE); - bus_dmamap_unload(txr->txtag, txbuf->map); - m_freem(txbuf->m_head); - txbuf->m_head = NULL; - } -#ifdef DEV_NETMAP - if (slot) { - int si = netmap_idx_n2k(&na->tx_rings[txr->me], i); - uint64_t paddr; - void *addr; + if (rx_que != NULL) { + free(adapter->rx_queues, M_DEVBUF); + adapter->rx_queues = NULL; + } - addr = PNMB(na, slot + si, &paddr); - txr->tx_base[i].buffer_addr = htole64(paddr); - /* reload the map for netmap mode */ - netmap_load_map(na, txr->txtag, txbuf->map, addr); - } -#endif /* DEV_NETMAP */ + em_release_hw_control(adapter); - /* clear the watch index */ - txbuf->next_eop = -1; - } - - /* Set number of descriptors available */ - txr->tx_avail = adapter->num_tx_desc; - txr->busy = EM_TX_IDLE; - - /* Clear checksum offload context. */ - txr->last_hw_offload = 0; - txr->last_hw_ipcss = 0; - txr->last_hw_ipcso = 0; - txr->last_hw_tucss = 0; - txr->last_hw_tucso = 0; - - bus_dmamap_sync(txr->txdma.dma_tag, txr->txdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - EM_TX_UNLOCK(txr); -} - -/********************************************************************* - * - * Initialize all transmit rings. - * - **********************************************************************/ -static void -em_setup_transmit_structures(struct adapter *adapter) -{ - struct tx_ring *txr = adapter->tx_rings; - - for (int i = 0; i < adapter->num_queues; i++, txr++) - em_setup_transmit_ring(txr); - - return; + if (adapter->mta != NULL) { + free(adapter->mta, M_DEVBUF); + } } /********************************************************************* @@ -3656,19 +2629,36 @@ em_setup_transmit_structures(struct adapter *adapter) * **********************************************************************/ static void -em_initialize_transmit_unit(struct adapter *adapter) +em_initialize_transmit_unit(if_ctx_t ctx) { - struct tx_ring *txr = adapter->tx_rings; + struct adapter *adapter = iflib_get_softc(ctx); + if_softc_ctx_t scctx = adapter->shared; + struct em_tx_queue *que; + struct tx_ring *txr; struct e1000_hw *hw = &adapter->hw; - u32 tctl, txdctl = 0, tarc, tipg = 0; + u32 tctl, txdctl = 0, tarc, tipg = 0; INIT_DEBUGOUT("em_initialize_transmit_unit: begin"); - for (int i = 0; i < adapter->num_queues; i++, txr++) { - u64 bus_addr = txr->txdma.dma_paddr; + for (int i = 0; i < adapter->tx_num_queues; i++, txr++) { + u64 bus_addr; + caddr_t offp, endp; + + que = &adapter->tx_queues[i]; + txr = &que->txr; + bus_addr = txr->tx_paddr; + + /*Enable all queues */ + em_init_tx_ring(que); + + /* Clear checksum offload context. */ + offp = (caddr_t)&txr->csum_flags; + endp = (caddr_t)(txr + 1); + bzero(offp, endp - offp); + /* Base and Len of TX Ring */ E1000_WRITE_REG(hw, E1000_TDLEN(i), - adapter->num_tx_desc * sizeof(struct e1000_tx_desc)); + scctx->isc_ntxd[0] * sizeof(struct e1000_tx_desc)); E1000_WRITE_REG(hw, E1000_TDBAH(i), (u32)(bus_addr >> 32)); E1000_WRITE_REG(hw, E1000_TDBAL(i), @@ -3681,7 +2671,6 @@ em_initialize_transmit_unit(struct adapter *adapter) E1000_READ_REG(&adapter->hw, E1000_TDBAL(i)), E1000_READ_REG(&adapter->hw, E1000_TDLEN(i))); - txr->busy = EM_TX_IDLE; txdctl = 0; /* clear txdctl */ txdctl |= 0x1f; /* PTHRESH */ txdctl |= 1 << 8; /* HTHRESH */ @@ -3700,6 +2689,11 @@ em_initialize_transmit_unit(struct adapter *adapter) tipg |= DEFAULT_80003ES2LAN_TIPG_IPGR2 << E1000_TIPG_IPGR2_SHIFT; break; + case e1000_82542: + tipg = DEFAULT_82542_TIPG_IPGT; + tipg |= DEFAULT_82542_TIPG_IPGR1 << E1000_TIPG_IPGR1_SHIFT; + tipg |= DEFAULT_82542_TIPG_IPGR2 << E1000_TIPG_IPGR2_SHIFT; + break; default: if ((adapter->hw.phy.media_type == e1000_media_type_fiber) || (adapter->hw.phy.media_type == @@ -3734,7 +2728,7 @@ em_initialize_transmit_unit(struct adapter *adapter) } else if (adapter->hw.mac.type == e1000_82574) { tarc = E1000_READ_REG(&adapter->hw, E1000_TARC(0)); tarc |= TARC_ERRATA_BIT; - if ( adapter->num_queues > 1) { + if ( adapter->tx_num_queues > 1) { tarc |= (TARC_COMPENSATION_MODE | TARC_MQ_FIX); E1000_WRITE_REG(&adapter->hw, E1000_TARC(0), tarc); E1000_WRITE_REG(&adapter->hw, E1000_TARC(1), tarc); @@ -3742,7 +2736,6 @@ em_initialize_transmit_unit(struct adapter *adapter) E1000_WRITE_REG(&adapter->hw, E1000_TARC(0), tarc); } - adapter->txd_cmd = E1000_TXD_CMD_IFCS; if (adapter->tx_int_delay.value > 0) adapter->txd_cmd |= E1000_TXD_CMD_IDE; @@ -3769,780 +2762,6 @@ em_initialize_transmit_unit(struct adapter *adapter) } } - -/********************************************************************* - * - * Free all transmit rings. - * - **********************************************************************/ -static void -em_free_transmit_structures(struct adapter *adapter) -{ - struct tx_ring *txr = adapter->tx_rings; - - for (int i = 0; i < adapter->num_queues; i++, txr++) { - EM_TX_LOCK(txr); - em_free_transmit_buffers(txr); - em_dma_free(adapter, &txr->txdma); - EM_TX_UNLOCK(txr); - EM_TX_LOCK_DESTROY(txr); - } - - free(adapter->tx_rings, M_DEVBUF); -} - -/********************************************************************* - * - * Free transmit ring related data structures. - * - **********************************************************************/ -static void -em_free_transmit_buffers(struct tx_ring *txr) -{ - struct adapter *adapter = txr->adapter; - struct em_txbuffer *txbuf; - - INIT_DEBUGOUT("free_transmit_ring: begin"); - - if (txr->tx_buffers == NULL) - return; - - for (int i = 0; i < adapter->num_tx_desc; i++) { - txbuf = &txr->tx_buffers[i]; - if (txbuf->m_head != NULL) { - bus_dmamap_sync(txr->txtag, txbuf->map, - BUS_DMASYNC_POSTWRITE); - bus_dmamap_unload(txr->txtag, - txbuf->map); - m_freem(txbuf->m_head); - txbuf->m_head = NULL; - if (txbuf->map != NULL) { - bus_dmamap_destroy(txr->txtag, - txbuf->map); - txbuf->map = NULL; - } - } else if (txbuf->map != NULL) { - bus_dmamap_unload(txr->txtag, - txbuf->map); - bus_dmamap_destroy(txr->txtag, - txbuf->map); - txbuf->map = NULL; - } - } -#if __FreeBSD_version >= 800000 - if (txr->br != NULL) - buf_ring_free(txr->br, M_DEVBUF); -#endif - if (txr->tx_buffers != NULL) { - free(txr->tx_buffers, M_DEVBUF); - txr->tx_buffers = NULL; - } - if (txr->txtag != NULL) { - bus_dma_tag_destroy(txr->txtag); - txr->txtag = NULL; - } - return; -} - - -/********************************************************************* - * The offload context is protocol specific (TCP/UDP) and thus - * only needs to be set when the protocol changes. The occasion - * of a context change can be a performance detriment, and - * might be better just disabled. The reason arises in the way - * in which the controller supports pipelined requests from the - * Tx data DMA. Up to four requests can be pipelined, and they may - * belong to the same packet or to multiple packets. However all - * requests for one packet are issued before a request is issued - * for a subsequent packet and if a request for the next packet - * requires a context change, that request will be stalled - * until the previous request completes. This means setting up - * a new context effectively disables pipelined Tx data DMA which - * in turn greatly slow down performance to send small sized - * frames. - **********************************************************************/ -static void -em_transmit_checksum_setup(struct tx_ring *txr, struct mbuf *mp, int ip_off, - struct ip *ip, u32 *txd_upper, u32 *txd_lower) -{ - struct adapter *adapter = txr->adapter; - struct e1000_context_desc *TXD = NULL; - struct em_txbuffer *tx_buffer; - int cur, hdr_len; - u32 cmd = 0; - u16 offload = 0; - u8 ipcso, ipcss, tucso, tucss; - - ipcss = ipcso = tucss = tucso = 0; - hdr_len = ip_off + (ip->ip_hl << 2); - cur = txr->next_avail_desc; - - /* Setup of IP header checksum. */ - if (mp->m_pkthdr.csum_flags & CSUM_IP) { - *txd_upper |= E1000_TXD_POPTS_IXSM << 8; - offload |= CSUM_IP; - ipcss = ip_off; - ipcso = ip_off + offsetof(struct ip, ip_sum); - /* - * Start offset for header checksum calculation. - * End offset for header checksum calculation. - * Offset of place to put the checksum. - */ - TXD = (struct e1000_context_desc *)&txr->tx_base[cur]; - TXD->lower_setup.ip_fields.ipcss = ipcss; - TXD->lower_setup.ip_fields.ipcse = htole16(hdr_len); - TXD->lower_setup.ip_fields.ipcso = ipcso; - cmd |= E1000_TXD_CMD_IP; - } - - if (mp->m_pkthdr.csum_flags & CSUM_TCP) { - *txd_lower = E1000_TXD_CMD_DEXT | E1000_TXD_DTYP_D; - *txd_upper |= E1000_TXD_POPTS_TXSM << 8; - offload |= CSUM_TCP; - tucss = hdr_len; - tucso = hdr_len + offsetof(struct tcphdr, th_sum); - /* - * The 82574L can only remember the *last* context used - * regardless of queue that it was use for. We cannot reuse - * contexts on this hardware platform and must generate a new - * context every time. 82574L hardware spec, section 7.2.6, - * second note. - */ - if (adapter->num_queues < 2) { - /* - * Setting up new checksum offload context for every - * frames takes a lot of processing time for hardware. - * This also reduces performance a lot for small sized - * frames so avoid it if driver can use previously - * configured checksum offload context. - */ - if (txr->last_hw_offload == offload) { - if (offload & CSUM_IP) { - if (txr->last_hw_ipcss == ipcss && - txr->last_hw_ipcso == ipcso && - txr->last_hw_tucss == tucss && - txr->last_hw_tucso == tucso) - return; - } else { - if (txr->last_hw_tucss == tucss && - txr->last_hw_tucso == tucso) - return; - } - } - txr->last_hw_offload = offload; - txr->last_hw_tucss = tucss; - txr->last_hw_tucso = tucso; - } - /* - * Start offset for payload checksum calculation. - * End offset for payload checksum calculation. - * Offset of place to put the checksum. - */ - TXD = (struct e1000_context_desc *)&txr->tx_base[cur]; - TXD->upper_setup.tcp_fields.tucss = hdr_len; - TXD->upper_setup.tcp_fields.tucse = htole16(0); - TXD->upper_setup.tcp_fields.tucso = tucso; - cmd |= E1000_TXD_CMD_TCP; - } else if (mp->m_pkthdr.csum_flags & CSUM_UDP) { - *txd_lower = E1000_TXD_CMD_DEXT | E1000_TXD_DTYP_D; - *txd_upper |= E1000_TXD_POPTS_TXSM << 8; - tucss = hdr_len; - tucso = hdr_len + offsetof(struct udphdr, uh_sum); - /* - * The 82574L can only remember the *last* context used - * regardless of queue that it was use for. We cannot reuse - * contexts on this hardware platform and must generate a new - * context every time. 82574L hardware spec, section 7.2.6, - * second note. - */ - if (adapter->num_queues < 2) { - /* - * Setting up new checksum offload context for every - * frames takes a lot of processing time for hardware. - * This also reduces performance a lot for small sized - * frames so avoid it if driver can use previously - * configured checksum offload context. - */ - if (txr->last_hw_offload == offload) { - if (offload & CSUM_IP) { - if (txr->last_hw_ipcss == ipcss && - txr->last_hw_ipcso == ipcso && - txr->last_hw_tucss == tucss && - txr->last_hw_tucso == tucso) - return; - } else { - if (txr->last_hw_tucss == tucss && - txr->last_hw_tucso == tucso) - return; - } - } - txr->last_hw_offload = offload; - txr->last_hw_tucss = tucss; - txr->last_hw_tucso = tucso; - } - /* - * Start offset for header checksum calculation. - * End offset for header checksum calculation. - * Offset of place to put the checksum. - */ - TXD = (struct e1000_context_desc *)&txr->tx_base[cur]; - TXD->upper_setup.tcp_fields.tucss = tucss; - TXD->upper_setup.tcp_fields.tucse = htole16(0); - TXD->upper_setup.tcp_fields.tucso = tucso; - } - - if (offload & CSUM_IP) { - txr->last_hw_ipcss = ipcss; - txr->last_hw_ipcso = ipcso; - } - - TXD->tcp_seg_setup.data = htole32(0); - TXD->cmd_and_length = - htole32(adapter->txd_cmd | E1000_TXD_CMD_DEXT | cmd); - tx_buffer = &txr->tx_buffers[cur]; - tx_buffer->m_head = NULL; - tx_buffer->next_eop = -1; - - if (++cur == adapter->num_tx_desc) - cur = 0; - - txr->tx_avail--; - txr->next_avail_desc = cur; -} - - -/********************************************************************** - * - * Setup work for hardware segmentation offload (TSO) - * - **********************************************************************/ -static void -em_tso_setup(struct tx_ring *txr, struct mbuf *mp, int ip_off, - struct ip *ip, struct tcphdr *tp, u32 *txd_upper, u32 *txd_lower) -{ - struct adapter *adapter = txr->adapter; - struct e1000_context_desc *TXD; - struct em_txbuffer *tx_buffer; - int cur, hdr_len; - - /* - * In theory we can use the same TSO context if and only if - * frame is the same type(IP/TCP) and the same MSS. However - * checking whether a frame has the same IP/TCP structure is - * hard thing so just ignore that and always restablish a - * new TSO context. - */ - hdr_len = ip_off + (ip->ip_hl << 2) + (tp->th_off << 2); - *txd_lower = (E1000_TXD_CMD_DEXT | /* Extended descr type */ - E1000_TXD_DTYP_D | /* Data descr type */ - E1000_TXD_CMD_TSE); /* Do TSE on this packet */ - - /* IP and/or TCP header checksum calculation and insertion. */ - *txd_upper = (E1000_TXD_POPTS_IXSM | E1000_TXD_POPTS_TXSM) << 8; - - cur = txr->next_avail_desc; - tx_buffer = &txr->tx_buffers[cur]; - TXD = (struct e1000_context_desc *) &txr->tx_base[cur]; - - /* - * Start offset for header checksum calculation. - * End offset for header checksum calculation. - * Offset of place put the checksum. - */ - TXD->lower_setup.ip_fields.ipcss = ip_off; - TXD->lower_setup.ip_fields.ipcse = - htole16(ip_off + (ip->ip_hl << 2) - 1); - TXD->lower_setup.ip_fields.ipcso = ip_off + offsetof(struct ip, ip_sum); - /* - * Start offset for payload checksum calculation. - * End offset for payload checksum calculation. - * Offset of place to put the checksum. - */ - TXD->upper_setup.tcp_fields.tucss = ip_off + (ip->ip_hl << 2); - TXD->upper_setup.tcp_fields.tucse = 0; - TXD->upper_setup.tcp_fields.tucso = - ip_off + (ip->ip_hl << 2) + offsetof(struct tcphdr, th_sum); - /* - * Payload size per packet w/o any headers. - * Length of all headers up to payload. - */ - TXD->tcp_seg_setup.fields.mss = htole16(mp->m_pkthdr.tso_segsz); - TXD->tcp_seg_setup.fields.hdr_len = hdr_len; - - TXD->cmd_and_length = htole32(adapter->txd_cmd | - E1000_TXD_CMD_DEXT | /* Extended descr */ - E1000_TXD_CMD_TSE | /* TSE context */ - E1000_TXD_CMD_IP | /* Do IP csum */ - E1000_TXD_CMD_TCP | /* Do TCP checksum */ - (mp->m_pkthdr.len - (hdr_len))); /* Total len */ - - tx_buffer->m_head = NULL; - tx_buffer->next_eop = -1; - - if (++cur == adapter->num_tx_desc) - cur = 0; - - txr->tx_avail--; - txr->next_avail_desc = cur; - txr->tx_tso = TRUE; -} - - -/********************************************************************** - * - * Examine each tx_buffer in the used queue. If the hardware is done - * processing the packet then free associated resources. The - * tx_buffer is put back on the free queue. - * - **********************************************************************/ -static void -em_txeof(struct tx_ring *txr) -{ - struct adapter *adapter = txr->adapter; - int first, last, done, processed; - struct em_txbuffer *tx_buffer; - struct e1000_tx_desc *tx_desc, *eop_desc; - if_t ifp = adapter->ifp; - - EM_TX_LOCK_ASSERT(txr); -#ifdef DEV_NETMAP - if (netmap_tx_irq(ifp, txr->me)) - return; -#endif /* DEV_NETMAP */ - - /* No work, make sure hang detection is disabled */ - if (txr->tx_avail == adapter->num_tx_desc) { - txr->busy = EM_TX_IDLE; - return; - } - - processed = 0; - first = txr->next_to_clean; - tx_desc = &txr->tx_base[first]; - tx_buffer = &txr->tx_buffers[first]; - last = tx_buffer->next_eop; - eop_desc = &txr->tx_base[last]; - - /* - * What this does is get the index of the - * first descriptor AFTER the EOP of the - * first packet, that way we can do the - * simple comparison on the inner while loop. - */ - if (++last == adapter->num_tx_desc) - last = 0; - done = last; - - bus_dmamap_sync(txr->txdma.dma_tag, txr->txdma.dma_map, - BUS_DMASYNC_POSTREAD); - - while (eop_desc->upper.fields.status & E1000_TXD_STAT_DD) { - /* We clean the range of the packet */ - while (first != done) { - tx_desc->upper.data = 0; - tx_desc->lower.data = 0; - tx_desc->buffer_addr = 0; - ++txr->tx_avail; - ++processed; - - if (tx_buffer->m_head) { - bus_dmamap_sync(txr->txtag, - tx_buffer->map, - BUS_DMASYNC_POSTWRITE); - bus_dmamap_unload(txr->txtag, - tx_buffer->map); - m_freem(tx_buffer->m_head); - tx_buffer->m_head = NULL; - } - tx_buffer->next_eop = -1; - - if (++first == adapter->num_tx_desc) - first = 0; - - tx_buffer = &txr->tx_buffers[first]; - tx_desc = &txr->tx_base[first]; - } - if_inc_counter(ifp, IFCOUNTER_OPACKETS, 1); - /* See if we can continue to the next packet */ - last = tx_buffer->next_eop; - if (last != -1) { - eop_desc = &txr->tx_base[last]; - /* Get new done point */ - if (++last == adapter->num_tx_desc) last = 0; - done = last; - } else - break; - } - bus_dmamap_sync(txr->txdma.dma_tag, txr->txdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - - txr->next_to_clean = first; - - /* - ** Hang detection: we know there's work outstanding - ** or the entry return would have been taken, so no - ** descriptor processed here indicates a potential hang. - ** The local timer will examine this and do a reset if needed. - */ - if (processed == 0) { - if (txr->busy != EM_TX_HUNG) - ++txr->busy; - } else /* At least one descriptor was cleaned */ - txr->busy = EM_TX_BUSY; /* note this clears HUNG */ - - /* - * If we have a minimum free, clear IFF_DRV_OACTIVE - * to tell the stack that it is OK to send packets. - * Notice that all writes of OACTIVE happen under the - * TX lock which, with a single queue, guarantees - * sanity. - */ - if (txr->tx_avail >= EM_MAX_SCATTER) { - if_setdrvflagbits(ifp, 0, IFF_DRV_OACTIVE); - } - - /* Disable hang detection if all clean */ - if (txr->tx_avail == adapter->num_tx_desc) - txr->busy = EM_TX_IDLE; -} - -/********************************************************************* - * - * Refresh RX descriptor mbufs from system mbuf buffer pool. - * - **********************************************************************/ -static void -em_refresh_mbufs(struct rx_ring *rxr, int limit) -{ - struct adapter *adapter = rxr->adapter; - struct mbuf *m; - bus_dma_segment_t segs; - struct em_rxbuffer *rxbuf; - int i, j, error, nsegs; - bool cleaned = FALSE; - - i = j = rxr->next_to_refresh; - /* - ** Get one descriptor beyond - ** our work mark to control - ** the loop. - */ - if (++j == adapter->num_rx_desc) - j = 0; - - while (j != limit) { - rxbuf = &rxr->rx_buffers[i]; - if (rxbuf->m_head == NULL) { - m = m_getjcl(M_NOWAIT, MT_DATA, - M_PKTHDR, adapter->rx_mbuf_sz); - /* - ** If we have a temporary resource shortage - ** that causes a failure, just abort refresh - ** for now, we will return to this point when - ** reinvoked from em_rxeof. - */ - if (m == NULL) - goto update; - } else - m = rxbuf->m_head; - - m->m_len = m->m_pkthdr.len = adapter->rx_mbuf_sz; - m->m_flags |= M_PKTHDR; - m->m_data = m->m_ext.ext_buf; - - /* Use bus_dma machinery to setup the memory mapping */ - error = bus_dmamap_load_mbuf_sg(rxr->rxtag, rxbuf->map, - m, &segs, &nsegs, BUS_DMA_NOWAIT); - if (error != 0) { - printf("Refresh mbufs: hdr dmamap load" - " failure - %d\n", error); - m_free(m); - rxbuf->m_head = NULL; - goto update; - } - rxbuf->m_head = m; - rxbuf->paddr = segs.ds_addr; - bus_dmamap_sync(rxr->rxtag, - rxbuf->map, BUS_DMASYNC_PREREAD); - em_setup_rxdesc(&rxr->rx_base[i], rxbuf); - cleaned = TRUE; - - i = j; /* Next is precalulated for us */ - rxr->next_to_refresh = i; - /* Calculate next controlling index */ - if (++j == adapter->num_rx_desc) - j = 0; - } -update: - /* - ** Update the tail pointer only if, - ** and as far as we have refreshed. - */ - if (cleaned) - E1000_WRITE_REG(&adapter->hw, - E1000_RDT(rxr->me), rxr->next_to_refresh); - - return; -} - - -/********************************************************************* - * - * Allocate memory for rx_buffer structures. Since we use one - * rx_buffer per received packet, the maximum number of rx_buffer's - * that we'll need is equal to the number of receive descriptors - * that we've allocated. - * - **********************************************************************/ -static int -em_allocate_receive_buffers(struct rx_ring *rxr) -{ - struct adapter *adapter = rxr->adapter; - device_t dev = adapter->dev; - struct em_rxbuffer *rxbuf; - int error; - - rxr->rx_buffers = malloc(sizeof(struct em_rxbuffer) * - adapter->num_rx_desc, M_DEVBUF, M_NOWAIT | M_ZERO); - if (rxr->rx_buffers == NULL) { - device_printf(dev, "Unable to allocate rx_buffer memory\n"); - return (ENOMEM); - } - - error = bus_dma_tag_create(bus_get_dma_tag(dev), /* parent */ - 1, 0, /* alignment, bounds */ - BUS_SPACE_MAXADDR, /* lowaddr */ - BUS_SPACE_MAXADDR, /* highaddr */ - NULL, NULL, /* filter, filterarg */ - MJUM9BYTES, /* maxsize */ - 1, /* nsegments */ - MJUM9BYTES, /* maxsegsize */ - 0, /* flags */ - NULL, /* lockfunc */ - NULL, /* lockarg */ - &rxr->rxtag); - if (error) { - device_printf(dev, "%s: bus_dma_tag_create failed %d\n", - __func__, error); - goto fail; - } - - rxbuf = rxr->rx_buffers; - for (int i = 0; i < adapter->num_rx_desc; i++, rxbuf++) { - rxbuf = &rxr->rx_buffers[i]; - error = bus_dmamap_create(rxr->rxtag, 0, &rxbuf->map); - if (error) { - device_printf(dev, "%s: bus_dmamap_create failed: %d\n", - __func__, error); - goto fail; - } - } - - return (0); - -fail: - em_free_receive_structures(adapter); - return (error); -} - - -/********************************************************************* - * - * Initialize a receive ring and its buffers. - * - **********************************************************************/ -static int -em_setup_receive_ring(struct rx_ring *rxr) -{ - struct adapter *adapter = rxr->adapter; - struct em_rxbuffer *rxbuf; - bus_dma_segment_t seg[1]; - int rsize, nsegs, error = 0; -#ifdef DEV_NETMAP - struct netmap_slot *slot; - struct netmap_adapter *na = netmap_getna(adapter->ifp); -#endif - - - /* Clear the ring contents */ - EM_RX_LOCK(rxr); - rsize = roundup2(adapter->num_rx_desc * - sizeof(union e1000_rx_desc_extended), EM_DBA_ALIGN); - bzero((void *)rxr->rx_base, rsize); -#ifdef DEV_NETMAP - slot = netmap_reset(na, NR_RX, rxr->me, 0); -#endif - - /* - ** Free current RX buffer structs and their mbufs - */ - for (int i = 0; i < adapter->num_rx_desc; i++) { - rxbuf = &rxr->rx_buffers[i]; - if (rxbuf->m_head != NULL) { - bus_dmamap_sync(rxr->rxtag, rxbuf->map, - BUS_DMASYNC_POSTREAD); - bus_dmamap_unload(rxr->rxtag, rxbuf->map); - m_freem(rxbuf->m_head); - rxbuf->m_head = NULL; /* mark as freed */ - } - } - - /* Now replenish the mbufs */ - for (int j = 0; j != adapter->num_rx_desc; ++j) { - rxbuf = &rxr->rx_buffers[j]; -#ifdef DEV_NETMAP - if (slot) { - int si = netmap_idx_n2k(&na->rx_rings[rxr->me], j); - uint64_t paddr; - void *addr; - - addr = PNMB(na, slot + si, &paddr); - netmap_load_map(na, rxr->rxtag, rxbuf->map, addr); - rxbuf->paddr = paddr; - em_setup_rxdesc(&rxr->rx_base[j], rxbuf); - continue; - } -#endif /* DEV_NETMAP */ - rxbuf->m_head = m_getjcl(M_NOWAIT, MT_DATA, - M_PKTHDR, adapter->rx_mbuf_sz); - if (rxbuf->m_head == NULL) { - error = ENOBUFS; - goto fail; - } - rxbuf->m_head->m_len = adapter->rx_mbuf_sz; - rxbuf->m_head->m_flags &= ~M_HASFCS; /* we strip it */ - rxbuf->m_head->m_pkthdr.len = adapter->rx_mbuf_sz; - - /* Get the memory mapping */ - error = bus_dmamap_load_mbuf_sg(rxr->rxtag, - rxbuf->map, rxbuf->m_head, seg, - &nsegs, BUS_DMA_NOWAIT); - if (error != 0) { - m_freem(rxbuf->m_head); - rxbuf->m_head = NULL; - goto fail; - } - bus_dmamap_sync(rxr->rxtag, - rxbuf->map, BUS_DMASYNC_PREREAD); - - rxbuf->paddr = seg[0].ds_addr; - em_setup_rxdesc(&rxr->rx_base[j], rxbuf); - } - rxr->next_to_check = 0; - rxr->next_to_refresh = 0; - bus_dmamap_sync(rxr->rxdma.dma_tag, rxr->rxdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - -fail: - EM_RX_UNLOCK(rxr); - return (error); -} - -/********************************************************************* - * - * Initialize all receive rings. - * - **********************************************************************/ -static int -em_setup_receive_structures(struct adapter *adapter) -{ - struct rx_ring *rxr = adapter->rx_rings; - int q; - - for (q = 0; q < adapter->num_queues; q++, rxr++) - if (em_setup_receive_ring(rxr)) - goto fail; - - return (0); -fail: - /* - * Free RX buffers allocated so far, we will only handle - * the rings that completed, the failing case will have - * cleaned up for itself. 'q' failed, so its the terminus. - */ - for (int i = 0; i < q; ++i) { - rxr = &adapter->rx_rings[i]; - for (int n = 0; n < adapter->num_rx_desc; n++) { - struct em_rxbuffer *rxbuf; - rxbuf = &rxr->rx_buffers[n]; - if (rxbuf->m_head != NULL) { - bus_dmamap_sync(rxr->rxtag, rxbuf->map, - BUS_DMASYNC_POSTREAD); - bus_dmamap_unload(rxr->rxtag, rxbuf->map); - m_freem(rxbuf->m_head); - rxbuf->m_head = NULL; - } - } - rxr->next_to_check = 0; - rxr->next_to_refresh = 0; - } - - return (ENOBUFS); -} - -/********************************************************************* - * - * Free all receive rings. - * - **********************************************************************/ -static void -em_free_receive_structures(struct adapter *adapter) -{ - struct rx_ring *rxr = adapter->rx_rings; - - for (int i = 0; i < adapter->num_queues; i++, rxr++) { - em_free_receive_buffers(rxr); - /* Free the ring memory as well */ - em_dma_free(adapter, &rxr->rxdma); - EM_RX_LOCK_DESTROY(rxr); - } - - free(adapter->rx_rings, M_DEVBUF); -} - - -/********************************************************************* - * - * Free receive ring data structures - * - **********************************************************************/ -static void -em_free_receive_buffers(struct rx_ring *rxr) -{ - struct adapter *adapter = rxr->adapter; - struct em_rxbuffer *rxbuf = NULL; - - INIT_DEBUGOUT("free_receive_buffers: begin"); - - if (rxr->rx_buffers != NULL) { - for (int i = 0; i < adapter->num_rx_desc; i++) { - rxbuf = &rxr->rx_buffers[i]; - if (rxbuf->map != NULL) { - bus_dmamap_sync(rxr->rxtag, rxbuf->map, - BUS_DMASYNC_POSTREAD); - bus_dmamap_unload(rxr->rxtag, rxbuf->map); - bus_dmamap_destroy(rxr->rxtag, rxbuf->map); - } - if (rxbuf->m_head != NULL) { - m_freem(rxbuf->m_head); - rxbuf->m_head = NULL; - } - } - free(rxr->rx_buffers, M_DEVBUF); - rxr->rx_buffers = NULL; - rxr->next_to_check = 0; - rxr->next_to_refresh = 0; - } - - if (rxr->rxtag != NULL) { - bus_dma_tag_destroy(rxr->rxtag); - rxr->rxtag = NULL; - } - - return; -} - - /********************************************************************* * * Enable receive unit. @@ -4550,11 +2769,14 @@ em_free_receive_buffers(struct rx_ring *rxr) **********************************************************************/ static void -em_initialize_receive_unit(struct adapter *adapter) +em_initialize_receive_unit(if_ctx_t ctx) { - struct rx_ring *rxr = adapter->rx_rings; - if_t ifp = adapter->ifp; + struct adapter *adapter = iflib_get_softc(ctx); + if_softc_ctx_t scctx = adapter->shared; + struct ifnet *ifp = iflib_get_ifp(ctx); struct e1000_hw *hw = &adapter->hw; + struct em_rx_queue *que; + int i; u32 rctl, rxcsum, rfctl; INIT_DEBUGOUT("em_initialize_receive_units: begin"); @@ -4587,16 +2809,18 @@ em_initialize_receive_unit(struct adapter *adapter) if (!em_disable_crc_stripping) rctl |= E1000_RCTL_SECRC; - E1000_WRITE_REG(&adapter->hw, E1000_RADV, - adapter->rx_abs_int_delay.value); + if (adapter->hw.mac.type >= e1000_82540) { + E1000_WRITE_REG(&adapter->hw, E1000_RADV, + adapter->rx_abs_int_delay.value); + /* + * Set the interrupt throttling rate. Value is calculated + * as DEFAULT_ITR = 1/(MAX_INTS_PER_SEC * 256ns) + */ + E1000_WRITE_REG(hw, E1000_ITR, DEFAULT_ITR); + } E1000_WRITE_REG(&adapter->hw, E1000_RDTR, adapter->rx_int_delay.value); - /* - * Set the interrupt throttling rate. Value is calculated - * as DEFAULT_ITR = 1/(MAX_INTS_PER_SEC * 256ns) - */ - E1000_WRITE_REG(hw, E1000_ITR, DEFAULT_ITR); /* Use extended rx descriptor formats */ rfctl = E1000_READ_REG(hw, E1000_RFCTL); @@ -4615,60 +2839,37 @@ em_initialize_receive_unit(struct adapter *adapter) E1000_WRITE_REG(hw, E1000_RFCTL, rfctl); rxcsum = E1000_READ_REG(hw, E1000_RXCSUM); - if (if_getcapenable(ifp) & IFCAP_RXCSUM) { -#ifdef EM_MULTIQUEUE - rxcsum |= E1000_RXCSUM_TUOFL | - E1000_RXCSUM_IPOFL | - E1000_RXCSUM_PCSD; -#else - rxcsum |= E1000_RXCSUM_TUOFL; -#endif + if (if_getcapenable(ifp) & IFCAP_RXCSUM && + adapter->hw.mac.type >= e1000_82543) { + if (adapter->tx_num_queues > 1) { + if (adapter->hw.mac.type >= igb_mac_min) { + rxcsum |= E1000_RXCSUM_PCSD; + if (hw->mac.type != e1000_82575) + rxcsum |= E1000_RXCSUM_CRCOFL; + } else + rxcsum |= E1000_RXCSUM_TUOFL | + E1000_RXCSUM_IPOFL | + E1000_RXCSUM_PCSD; + } else { + if (adapter->hw.mac.type >= igb_mac_min) + rxcsum |= E1000_RXCSUM_IPPCSE; + else + rxcsum |= E1000_RXCSUM_TUOFL | E1000_RXCSUM_IPOFL; + if (adapter->hw.mac.type > e1000_82575) + rxcsum |= E1000_RXCSUM_CRCOFL; + } } else rxcsum &= ~E1000_RXCSUM_TUOFL; E1000_WRITE_REG(hw, E1000_RXCSUM, rxcsum); -#ifdef EM_MULTIQUEUE -#define RSSKEYLEN 10 - if (adapter->num_queues > 1) { - uint8_t rss_key[4 * RSSKEYLEN]; - uint32_t reta = 0; - int i; - - /* - * Configure RSS key - */ - arc4rand(rss_key, sizeof(rss_key), 0); - for (i = 0; i < RSSKEYLEN; ++i) { - uint32_t rssrk = 0; - - rssrk = EM_RSSRK_VAL(rss_key, i); - E1000_WRITE_REG(hw,E1000_RSSRK(i), rssrk); - } - - /* - * Configure RSS redirect table in following fashion: - * (hash & ring_cnt_mask) == rdr_table[(hash & rdr_table_mask)] - */ - for (i = 0; i < sizeof(reta); ++i) { - uint32_t q; - - q = (i % adapter->num_queues) << 7; - reta |= q << (8 * i); - } - - for (i = 0; i < 32; ++i) { - E1000_WRITE_REG(hw, E1000_RETA(i), reta); - } - - E1000_WRITE_REG(hw, E1000_MRQC, E1000_MRQC_RSS_ENABLE_2Q | - E1000_MRQC_RSS_FIELD_IPV4_TCP | - E1000_MRQC_RSS_FIELD_IPV4 | - E1000_MRQC_RSS_FIELD_IPV6_TCP_EX | - E1000_MRQC_RSS_FIELD_IPV6_EX | - E1000_MRQC_RSS_FIELD_IPV6); + if (adapter->rx_num_queues > 1) { + if (adapter->hw.mac.type >= igb_mac_min) + igb_initialize_rss_mapping(adapter); + else + em_initialize_rss_mapping(adapter); } -#endif + /* ** XXX TEMPORARY WORKAROUND: on some systems with 82573 ** long latencies are observed, like Lenovo X60. This @@ -4679,28 +2880,21 @@ em_initialize_receive_unit(struct adapter *adapter) if (hw->mac.type == e1000_82573) E1000_WRITE_REG(hw, E1000_RDTR, 0x20); - for (int i = 0; i < adapter->num_queues; i++, rxr++) { + for (i = 0, que = adapter->rx_queues; i < adapter->rx_num_queues; i++, que++) { + struct rx_ring *rxr = &que->rxr; /* Setup the Base and Length of the Rx Descriptor Ring */ - u64 bus_addr = rxr->rxdma.dma_paddr; - u32 rdt = adapter->num_rx_desc - 1; /* default */ + u64 bus_addr = rxr->rx_paddr; +#if 0 + u32 rdt = adapter->rx_num_queues -1; /* default */ +#endif E1000_WRITE_REG(hw, E1000_RDLEN(i), - adapter->num_rx_desc * sizeof(union e1000_rx_desc_extended)); + scctx->isc_nrxd[0] * sizeof(union e1000_rx_desc_extended)); E1000_WRITE_REG(hw, E1000_RDBAH(i), (u32)(bus_addr >> 32)); E1000_WRITE_REG(hw, E1000_RDBAL(i), (u32)bus_addr); /* Setup the Head and Tail Descriptor Pointers */ E1000_WRITE_REG(hw, E1000_RDH(i), 0); -#ifdef DEV_NETMAP - /* - * an init() while a netmap client is active must - * preserve the rx buffers passed to userspace. - */ - if (if_getcapenable(ifp) & IFCAP_NETMAP) { - struct netmap_adapter *na = netmap_getna(adapter->ifp); - rdt -= nm_kr_rxspace(&na->rx_rings[i]); - } -#endif /* DEV_NETMAP */ - E1000_WRITE_REG(hw, E1000_RDT(i), rdt); + E1000_WRITE_REG(hw, E1000_RDT(i), 0); } /* @@ -4710,6 +2904,7 @@ em_initialize_receive_unit(struct adapter *adapter) * Only write to RXDCTL(1) if there is a need for different * settings. */ + if (((adapter->hw.mac.type == e1000_ich9lan) || (adapter->hw.mac.type == e1000_pch2lan) || (adapter->hw.mac.type == e1000_ich10lan)) && @@ -4717,17 +2912,82 @@ em_initialize_receive_unit(struct adapter *adapter) u32 rxdctl = E1000_READ_REG(hw, E1000_RXDCTL(0)); E1000_WRITE_REG(hw, E1000_RXDCTL(0), rxdctl | 3); } else if (adapter->hw.mac.type == e1000_82574) { - for (int i = 0; i < adapter->num_queues; i++) { + for (int i = 0; i < adapter->rx_num_queues; i++) { u32 rxdctl = E1000_READ_REG(hw, E1000_RXDCTL(i)); - rxdctl |= 0x20; /* PTHRESH */ rxdctl |= 4 << 8; /* HTHRESH */ rxdctl |= 4 << 16;/* WTHRESH */ rxdctl |= 1 << 24; /* Switch to granularity */ E1000_WRITE_REG(hw, E1000_RXDCTL(i), rxdctl); } + } else if (adapter->hw.mac.type >= igb_mac_min) { + u32 psize, srrctl = 0; + + if (ifp->if_mtu > ETHERMTU) { + rctl |= E1000_RCTL_LPE; + + /* Set maximum packet len */ + psize = scctx->isc_max_frame_size; + if (psize <= 4096) { + srrctl |= 4096 >> E1000_SRRCTL_BSIZEPKT_SHIFT; + rctl |= E1000_RCTL_SZ_4096 | E1000_RCTL_BSEX; + } else if (psize > 4096) { + srrctl |= 8192 >> E1000_SRRCTL_BSIZEPKT_SHIFT; + rctl |= E1000_RCTL_SZ_8192 | E1000_RCTL_BSEX; + } + + /* are we on a vlan? */ + if (ifp->if_vlantrunk != NULL) + psize += VLAN_TAG_SIZE; + E1000_WRITE_REG(&adapter->hw, E1000_RLPML, psize); + } else { + rctl &= ~E1000_RCTL_LPE; + srrctl |= 2048 >> E1000_SRRCTL_BSIZEPKT_SHIFT; + rctl |= E1000_RCTL_SZ_2048; + } + + /* + * If TX flow control is disabled and there's >1 queue defined, + * enable DROP. + * + * This drops frames rather than hanging the RX MAC for all queues. + */ + if ((adapter->rx_num_queues > 1) && + (adapter->fc == e1000_fc_none || + adapter->fc == e1000_fc_rx_pause)) { + srrctl |= E1000_SRRCTL_DROP_EN; + } + /* Setup the Base and Length of the Rx Descriptor Rings */ + for (i = 0, que = adapter->rx_queues; i < adapter->rx_num_queues; i++, que++) { + struct rx_ring *rxr = &que->rxr; + u64 bus_addr = rxr->rx_paddr; + u32 rxdctl; + +#ifdef notyet + /* Configure for header split? -- ignore for now */ + rxr->hdr_split = igb_header_split; +#else + srrctl |= E1000_SRRCTL_DESCTYPE_ADV_ONEBUF; +#endif + + + E1000_WRITE_REG(hw, E1000_RDLEN(i), + scctx->isc_nrxd[0] * sizeof(struct e1000_rx_desc)); + E1000_WRITE_REG(hw, E1000_RDBAH(i), + (uint32_t)(bus_addr >> 32)); + E1000_WRITE_REG(hw, E1000_RDBAL(i), + (uint32_t)bus_addr); + E1000_WRITE_REG(hw, E1000_SRRCTL(i), srrctl); + /* Enable this Queue */ + rxdctl = E1000_READ_REG(hw, E1000_RXDCTL(i)); + rxdctl |= E1000_RXDCTL_QUEUE_ENABLE; + rxdctl &= 0xFFF00000; + rxdctl |= IGB_RX_PTHRESH; + rxdctl |= IGB_RX_HTHRESH << 8; + rxdctl |= IGB_RX_WTHRESH << 16; + E1000_WRITE_REG(hw, E1000_RXDCTL(i), rxdctl); + } } - if (adapter->hw.mac.type >= e1000_pch2lan) { if (if_getmtu(ifp) > ETHERMTU) e1000_lv_jumbo_workaround_ich8lan(hw, TRUE); @@ -4753,323 +3013,28 @@ em_initialize_receive_unit(struct adapter *adapter) return; } - -/********************************************************************* - * - * This routine executes in interrupt context. It replenishes - * the mbufs in the descriptor and sends data which has been - * dma'ed into host memory to upper layer. - * - * We loop at most count times if count is > 0, or until done if - * count < 0. - * - * For polling we also now return the number of cleaned packets - *********************************************************************/ -static bool -em_rxeof(struct rx_ring *rxr, int count, int *done) -{ - struct adapter *adapter = rxr->adapter; - if_t ifp = adapter->ifp; - struct mbuf *mp, *sendmp; - u32 status = 0; - u16 len; - int i, processed, rxdone = 0; - bool eop; - union e1000_rx_desc_extended *cur; - - EM_RX_LOCK(rxr); - - /* Sync the ring */ - bus_dmamap_sync(rxr->rxdma.dma_tag, rxr->rxdma.dma_map, - BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); - - -#ifdef DEV_NETMAP - if (netmap_rx_irq(ifp, rxr->me, &processed)) { - EM_RX_UNLOCK(rxr); - return (FALSE); - } -#endif /* DEV_NETMAP */ - - for (i = rxr->next_to_check, processed = 0; count != 0;) { - if ((if_getdrvflags(ifp) & IFF_DRV_RUNNING) == 0) - break; - - cur = &rxr->rx_base[i]; - status = le32toh(cur->wb.upper.status_error); - mp = sendmp = NULL; - - if ((status & E1000_RXD_STAT_DD) == 0) - break; - - len = le16toh(cur->wb.upper.length); - eop = (status & E1000_RXD_STAT_EOP) != 0; - - if ((status & E1000_RXDEXT_ERR_FRAME_ERR_MASK) || - (rxr->discard == TRUE)) { - adapter->dropped_pkts++; - ++rxr->rx_discarded; - if (!eop) /* Catch subsequent segs */ - rxr->discard = TRUE; - else - rxr->discard = FALSE; - em_rx_discard(rxr, i); - goto next_desc; - } - bus_dmamap_unload(rxr->rxtag, rxr->rx_buffers[i].map); - - /* Assign correct length to the current fragment */ - mp = rxr->rx_buffers[i].m_head; - mp->m_len = len; - - /* Trigger for refresh */ - rxr->rx_buffers[i].m_head = NULL; - - /* First segment? */ - if (rxr->fmp == NULL) { - mp->m_pkthdr.len = len; - rxr->fmp = rxr->lmp = mp; - } else { - /* Chain mbuf's together */ - mp->m_flags &= ~M_PKTHDR; - rxr->lmp->m_next = mp; - rxr->lmp = mp; - rxr->fmp->m_pkthdr.len += len; - } - - if (eop) { - --count; - sendmp = rxr->fmp; - if_setrcvif(sendmp, ifp); - if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1); - em_receive_checksum(status, sendmp); -#ifndef __NO_STRICT_ALIGNMENT - if (adapter->hw.mac.max_frame_size > - (MCLBYTES - ETHER_ALIGN) && - em_fixup_rx(rxr) != 0) - goto skip; -#endif - if (status & E1000_RXD_STAT_VP) { - if_setvtag(sendmp, - le16toh(cur->wb.upper.vlan)); - sendmp->m_flags |= M_VLANTAG; - } -#ifndef __NO_STRICT_ALIGNMENT -skip: -#endif - rxr->fmp = rxr->lmp = NULL; - } -next_desc: - /* Sync the ring */ - bus_dmamap_sync(rxr->rxdma.dma_tag, rxr->rxdma.dma_map, - BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); - - /* Zero out the receive descriptors status. */ - cur->wb.upper.status_error &= htole32(~0xFF); - ++rxdone; /* cumulative for POLL */ - ++processed; - - /* Advance our pointers to the next descriptor. */ - if (++i == adapter->num_rx_desc) - i = 0; - - /* Send to the stack */ - if (sendmp != NULL) { - rxr->next_to_check = i; - EM_RX_UNLOCK(rxr); - if_input(ifp, sendmp); - EM_RX_LOCK(rxr); - i = rxr->next_to_check; - } - - /* Only refresh mbufs every 8 descriptors */ - if (processed == 8) { - em_refresh_mbufs(rxr, i); - processed = 0; - } - } - - /* Catch any remaining refresh work */ - if (e1000_rx_unrefreshed(rxr)) - em_refresh_mbufs(rxr, i); - - rxr->next_to_check = i; - if (done != NULL) - *done = rxdone; - EM_RX_UNLOCK(rxr); - - return ((status & E1000_RXD_STAT_DD) ? TRUE : FALSE); -} - -static __inline void -em_rx_discard(struct rx_ring *rxr, int i) -{ - struct em_rxbuffer *rbuf; - - rbuf = &rxr->rx_buffers[i]; - bus_dmamap_unload(rxr->rxtag, rbuf->map); - - /* Free any previous pieces */ - if (rxr->fmp != NULL) { - rxr->fmp->m_flags |= M_PKTHDR; - m_freem(rxr->fmp); - rxr->fmp = NULL; - rxr->lmp = NULL; - } - /* - ** Free buffer and allow em_refresh_mbufs() - ** to clean up and recharge buffer. - */ - if (rbuf->m_head) { - m_free(rbuf->m_head); - rbuf->m_head = NULL; - } - return; -} - -#ifndef __NO_STRICT_ALIGNMENT -/* - * When jumbo frames are enabled we should realign entire payload on - * architecures with strict alignment. This is serious design mistake of 8254x - * as it nullifies DMA operations. 8254x just allows RX buffer size to be - * 2048/4096/8192/16384. What we really want is 2048 - ETHER_ALIGN to align its - * payload. On architecures without strict alignment restrictions 8254x still - * performs unaligned memory access which would reduce the performance too. - * To avoid copying over an entire frame to align, we allocate a new mbuf and - * copy ethernet header to the new mbuf. The new mbuf is prepended into the - * existing mbuf chain. - * - * Be aware, best performance of the 8254x is achived only when jumbo frame is - * not used at all on architectures with strict alignment. - */ -static int -em_fixup_rx(struct rx_ring *rxr) -{ - struct adapter *adapter = rxr->adapter; - struct mbuf *m, *n; - int error; - - error = 0; - m = rxr->fmp; - if (m->m_len <= (MCLBYTES - ETHER_HDR_LEN)) { - bcopy(m->m_data, m->m_data + ETHER_HDR_LEN, m->m_len); - m->m_data += ETHER_HDR_LEN; - } else { - MGETHDR(n, M_NOWAIT, MT_DATA); - if (n != NULL) { - bcopy(m->m_data, n->m_data, ETHER_HDR_LEN); - m->m_data += ETHER_HDR_LEN; - m->m_len -= ETHER_HDR_LEN; - n->m_len = ETHER_HDR_LEN; - M_MOVE_PKTHDR(n, m); - n->m_next = m; - rxr->fmp = n; - } else { - adapter->dropped_pkts++; - m_freem(rxr->fmp); - rxr->fmp = NULL; - error = ENOMEM; - } - } - - return (error); -} -#endif - static void -em_setup_rxdesc(union e1000_rx_desc_extended *rxd, const struct em_rxbuffer *rxbuf) +em_if_vlan_register(if_ctx_t ctx, u16 vtag) { - rxd->read.buffer_addr = htole64(rxbuf->paddr); - /* DD bits must be cleared */ - rxd->wb.upper.status_error= 0; -} - -/********************************************************************* - * - * Verify that the hardware indicated that the checksum is valid. - * Inform the stack about the status of checksum so that stack - * doesn't spend time verifying the checksum. - * - *********************************************************************/ -static void -em_receive_checksum(uint32_t status, struct mbuf *mp) -{ - mp->m_pkthdr.csum_flags = 0; - - /* Ignore Checksum bit is set */ - if (status & E1000_RXD_STAT_IXSM) - return; - - /* If the IP checksum exists and there is no IP Checksum error */ - if ((status & (E1000_RXD_STAT_IPCS | E1000_RXDEXT_STATERR_IPE)) == - E1000_RXD_STAT_IPCS) { - mp->m_pkthdr.csum_flags = (CSUM_IP_CHECKED | CSUM_IP_VALID); - } - - /* TCP or UDP checksum */ - if ((status & (E1000_RXD_STAT_TCPCS | E1000_RXDEXT_STATERR_TCPE)) == - E1000_RXD_STAT_TCPCS) { - mp->m_pkthdr.csum_flags |= (CSUM_DATA_VALID | CSUM_PSEUDO_HDR); - mp->m_pkthdr.csum_data = htons(0xffff); - } - if (status & E1000_RXD_STAT_UDPCS) { - mp->m_pkthdr.csum_flags |= (CSUM_DATA_VALID | CSUM_PSEUDO_HDR); - mp->m_pkthdr.csum_data = htons(0xffff); - } -} - -/* - * This routine is run via an vlan - * config EVENT - */ -static void -em_register_vlan(void *arg, if_t ifp, u16 vtag) -{ - struct adapter *adapter = if_getsoftc(ifp); + struct adapter *adapter = iflib_get_softc(ctx); u32 index, bit; - if ((void*)adapter != arg) /* Not our event */ - return; - - if ((vtag == 0) || (vtag > 4095)) /* Invalid ID */ - return; - - EM_CORE_LOCK(adapter); index = (vtag >> 5) & 0x7F; bit = vtag & 0x1F; adapter->shadow_vfta[index] |= (1 << bit); ++adapter->num_vlans; - /* Re-init to load the changes */ - if (if_getcapenable(ifp) & IFCAP_VLAN_HWFILTER) - em_init_locked(adapter); - EM_CORE_UNLOCK(adapter); } -/* - * This routine is run via an vlan - * unconfig EVENT - */ static void -em_unregister_vlan(void *arg, if_t ifp, u16 vtag) +em_if_vlan_unregister(if_ctx_t ctx, u16 vtag) { - struct adapter *adapter = if_getsoftc(ifp); + struct adapter *adapter = iflib_get_softc(ctx); u32 index, bit; - if (adapter != arg) - return; - - if ((vtag == 0) || (vtag > 4095)) /* Invalid */ - return; - - EM_CORE_LOCK(adapter); index = (vtag >> 5) & 0x7F; bit = vtag & 0x1F; adapter->shadow_vfta[index] &= ~(1 << bit); --adapter->num_vlans; - /* Re-init to load the changes */ - if (if_getcapenable(ifp) & IFCAP_VLAN_HWFILTER) - em_init_locked(adapter); - EM_CORE_UNLOCK(adapter); } static void @@ -5108,25 +3073,38 @@ em_setup_vlan_hw_support(struct adapter *adapter) } static void -em_enable_intr(struct adapter *adapter) +em_if_enable_intr(if_ctx_t ctx) { + struct adapter *adapter = iflib_get_softc(ctx); struct e1000_hw *hw = &adapter->hw; u32 ims_mask = IMS_ENABLE_MASK; if (hw->mac.type == e1000_82574) { E1000_WRITE_REG(hw, EM_EIAC, adapter->ims); ims_mask |= adapter->ims; - } + } if (adapter->intr_type == IFLIB_INTR_MSIX && hw->mac.type >= igb_mac_min) { + u32 mask = (adapter->que_mask | adapter->link_mask); + + E1000_WRITE_REG(&adapter->hw, E1000_EIAC, mask); + E1000_WRITE_REG(&adapter->hw, E1000_EIAM, mask); + E1000_WRITE_REG(&adapter->hw, E1000_EIMS, mask); + ims_mask = E1000_IMS_LSC; + } + E1000_WRITE_REG(hw, E1000_IMS, ims_mask); } static void -em_disable_intr(struct adapter *adapter) +em_if_disable_intr(if_ctx_t ctx) { + struct adapter *adapter = iflib_get_softc(ctx); struct e1000_hw *hw = &adapter->hw; - if (hw->mac.type == e1000_82574) - E1000_WRITE_REG(hw, EM_EIAC, 0); + if (adapter->intr_type == IFLIB_INTR_MSIX) { + if (hw->mac.type >= igb_mac_min) + E1000_WRITE_REG(&adapter->hw, E1000_EIMC, ~0); + E1000_WRITE_REG(&adapter->hw, E1000_EIAC, 0); + } E1000_WRITE_REG(&adapter->hw, E1000_IMC, 0xffffffff); } @@ -5245,15 +3223,34 @@ em_is_valid_ether_addr(u8 *addr) ** later use. */ static void -em_get_wakeup(device_t dev) +em_get_wakeup(if_ctx_t ctx) { - struct adapter *adapter = device_get_softc(dev); + struct adapter *adapter = iflib_get_softc(ctx); + device_t dev = iflib_get_dev(ctx); u16 eeprom_data = 0, device_id, apme_mask; adapter->has_manage = e1000_enable_mng_pass_thru(&adapter->hw); apme_mask = EM_EEPROM_APME; switch (adapter->hw.mac.type) { + case e1000_82542: + case e1000_82543: + break; + case e1000_82544: + e1000_read_nvm(&adapter->hw, + NVM_INIT_CONTROL2_REG, 1, &eeprom_data); + apme_mask = EM_82544_APME; + break; + case e1000_82546: + case e1000_82546_rev_3: + if (adapter->hw.bus.func == 1) { + e1000_read_nvm(&adapter->hw, + NVM_INIT_CONTROL3_PORT_B, 1, &eeprom_data); + break; + } else + e1000_read_nvm(&adapter->hw, + NVM_INIT_CONTROL3_PORT_A, 1, &eeprom_data); + break; case e1000_82573: case e1000_82583: adapter->has_amt = TRUE; @@ -5274,8 +3271,6 @@ em_get_wakeup(device_t dev) case e1000_ich10lan: case e1000_pchlan: case e1000_pch2lan: - case e1000_pch_lpt: - case e1000_pch_spt: apme_mask = E1000_WUC_APME; adapter->has_amt = TRUE; eeprom_data = E1000_READ_REG(&adapter->hw, E1000_WUC); @@ -5294,6 +3289,25 @@ em_get_wakeup(device_t dev) */ device_id = pci_get_device(dev); switch (device_id) { + case E1000_DEV_ID_82546GB_PCIE: + adapter->wol = 0; + break; + case E1000_DEV_ID_82546EB_FIBER: + case E1000_DEV_ID_82546GB_FIBER: + /* Wake events only supported on port A for dual fiber + * regardless of eeprom setting */ + if (E1000_READ_REG(&adapter->hw, E1000_STATUS) & + E1000_STATUS_FUNC_1) + adapter->wol = 0; + break; + case E1000_DEV_ID_82546GB_QUAD_COPPER_KSP3: + /* if quad port adapter, disable WoL on all but port A */ + if (global_quad_port_a != 0) + adapter->wol = 0; + /* Reset for multiple quad port adapters */ + if (++global_quad_port_a == 4) + global_quad_port_a = 0; + break; case E1000_DEV_ID_82571EB_FIBER: /* Wake events only supported on port A for dual fiber * regardless of eeprom setting */ @@ -5320,11 +3334,12 @@ em_get_wakeup(device_t dev) * Enable PCI Wake On Lan capability */ static void -em_enable_wakeup(device_t dev) +em_enable_wakeup(if_ctx_t ctx) { - struct adapter *adapter = device_get_softc(dev); - if_t ifp = adapter->ifp; - u32 pmc, ctrl, ctrl_ext, rctl, wuc; + struct adapter *adapter = iflib_get_softc(ctx); + device_t dev = iflib_get_dev(ctx); + if_t ifp = iflib_get_ifp(ctx); + u32 pmc, ctrl, ctrl_ext, rctl; u16 status; if ((pci_find_cap(dev, PCIY_PMG, &pmc) != 0)) @@ -5334,9 +3349,7 @@ em_enable_wakeup(device_t dev) ctrl = E1000_READ_REG(&adapter->hw, E1000_CTRL); ctrl |= (E1000_CTRL_SWDPIN2 | E1000_CTRL_SWDPIN3); E1000_WRITE_REG(&adapter->hw, E1000_CTRL, ctrl); - wuc = E1000_READ_REG(&adapter->hw, E1000_WUC); - wuc |= E1000_WUC_PME_EN; - E1000_WRITE_REG(&adapter->hw, E1000_WUC, wuc); + E1000_WRITE_REG(&adapter->hw, E1000_WUC, E1000_WUC_PME_EN); if ((adapter->hw.mac.type == e1000_ich8lan) || (adapter->hw.mac.type == e1000_pchlan) || @@ -5367,10 +3380,8 @@ em_enable_wakeup(device_t dev) E1000_WRITE_REG(&adapter->hw, E1000_RCTL, rctl); } - if ((adapter->hw.mac.type == e1000_pchlan) || - (adapter->hw.mac.type == e1000_pch2lan) || - (adapter->hw.mac.type == e1000_pch_lpt) || - (adapter->hw.mac.type == e1000_pch_spt)) { + if ((adapter->hw.mac.type == e1000_pchlan) || + (adapter->hw.mac.type == e1000_pch2lan)) { if (em_enable_phy_wakeup(adapter)) return; } else { @@ -5466,11 +3477,10 @@ em_enable_phy_wakeup(struct adapter *adapter) } static void -em_led_func(void *arg, int onoff) +em_if_led_func(if_ctx_t ctx, int onoff) { - struct adapter *adapter = arg; + struct adapter *adapter = iflib_get_softc(ctx); - EM_CORE_LOCK(adapter); if (onoff) { e1000_setup_led(&adapter->hw); e1000_led_on(&adapter->hw); @@ -5478,7 +3488,6 @@ em_led_func(void *arg, int onoff) e1000_led_off(&adapter->hw); e1000_cleanup_led(&adapter->hw); } - EM_CORE_UNLOCK(adapter); } /* @@ -5609,11 +3618,10 @@ em_update_stats_counters(struct adapter *adapter) } static uint64_t -em_get_counter(if_t ifp, ift_counter cnt) +em_if_get_counter(if_ctx_t ctx, ift_counter cnt) { - struct adapter *adapter; - - adapter = if_getsoftc(ifp); + struct adapter *adapter = iflib_get_softc(ctx); + struct ifnet *ifp = iflib_get_ifp(ctx); switch (cnt) { case IFCOUNTER_COLLISIONS: @@ -5649,11 +3657,10 @@ em_sysctl_reg_handler(SYSCTL_HANDLER_ARGS) static void em_add_hw_stats(struct adapter *adapter) { - device_t dev = adapter->dev; - - struct tx_ring *txr = adapter->tx_rings; - struct rx_ring *rxr = adapter->rx_rings; - + device_t dev = iflib_get_dev(adapter->ctx); + struct em_tx_queue *tx_que = adapter->tx_queues; + struct em_rx_queue *rx_que = adapter->rx_queues; + struct sysctl_ctx_list *ctx = device_get_sysctl_ctx(dev); struct sysctl_oid *tree = device_get_sysctl_tree(dev); struct sysctl_oid_list *child = SYSCTL_CHILDREN(tree); @@ -5700,7 +3707,8 @@ em_add_hw_stats(struct adapter *adapter) CTLFLAG_RD, &adapter->hw.fc.low_water, 0, "Flow Control Low Watermark"); - for (int i = 0; i < adapter->num_queues; i++, txr++, rxr++) { + for (int i = 0; i < adapter->tx_num_queues; i++, tx_que++) { + struct tx_ring *txr = &tx_que->txr; snprintf(namebuf, QUEUE_NAME_LEN, "queue_tx_%d", i); queue_node = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, namebuf, CTLFLAG_RD, NULL, "TX Queue Name"); @@ -5722,8 +3730,11 @@ em_add_hw_stats(struct adapter *adapter) SYSCTL_ADD_ULONG(ctx, queue_list, OID_AUTO, "no_desc_avail", CTLFLAG_RD, &txr->no_desc_avail, "Queue No Descriptor Available"); + } - snprintf(namebuf, QUEUE_NAME_LEN, "queue_rx_%d", i); + for (int j = 0; j < adapter->rx_num_queues; j++, rx_que++) { + struct rx_ring *rxr = &rx_que->rxr; + snprintf(namebuf, QUEUE_NAME_LEN, "queue_rx_%d", j); queue_node = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, namebuf, CTLFLAG_RD, NULL, "RX Queue Name"); queue_list = SYSCTL_CHILDREN(queue_node); @@ -6010,7 +4021,6 @@ em_sysctl_int_delay(SYSCTL_HANDLER_ARGS) adapter = info->adapter; - EM_CORE_LOCK(adapter); regval = E1000_READ_OFFSET(&adapter->hw, info->offset); regval = (regval & ~0xffff) | (ticks & 0xffff); /* Handle a few special cases. */ @@ -6027,7 +4037,6 @@ em_sysctl_int_delay(SYSCTL_HANDLER_ARGS) break; } E1000_WRITE_OFFSET(&adapter->hw, info->offset, regval); - EM_CORE_UNLOCK(adapter); return (0); } @@ -6112,10 +4121,9 @@ em_sysctl_eee(SYSCTL_HANDLER_ARGS) error = sysctl_handle_int(oidp, &value, 0, req); if (error || req->newptr == NULL) return (error); - EM_CORE_LOCK(adapter); adapter->hw.dev_spec.ich8lan.eee_disable = (value != 0); - em_init_locked(adapter); - EM_CORE_UNLOCK(adapter); + em_if_init(adapter->ctx); + return (0); } @@ -6148,8 +4156,8 @@ static void em_print_debug_info(struct adapter *adapter) { device_t dev = adapter->dev; - struct tx_ring *txr = adapter->tx_rings; - struct rx_ring *rxr = adapter->rx_rings; + struct tx_ring *txr = &adapter->tx_queues->txr; + struct rx_ring *rxr = &adapter->rx_queues->rxr; if (if_getdrvflags(adapter->ifp) & IFF_DRV_RUNNING) printf("Interface is RUNNING "); @@ -6161,38 +4169,33 @@ em_print_debug_info(struct adapter *adapter) else printf("and ACTIVE\n"); - for (int i = 0; i < adapter->num_queues; i++, txr++, rxr++) { + for (int i = 0; i < adapter->tx_num_queues; i++, txr++) { device_printf(dev, "TX Queue %d ------\n", i); device_printf(dev, "hw tdh = %d, hw tdt = %d\n", E1000_READ_REG(&adapter->hw, E1000_TDH(i)), E1000_READ_REG(&adapter->hw, E1000_TDT(i))); - device_printf(dev, "Tx Queue Status = %d\n", txr->busy); - device_printf(dev, "TX descriptors avail = %d\n", - txr->tx_avail); - device_printf(dev, "Tx Descriptors avail failure = %ld\n", - txr->no_desc_avail); - device_printf(dev, "RX Queue %d ------\n", i); + + } + for (int j=0; j < adapter->rx_num_queues; j++, rxr++) { + device_printf(dev, "RX Queue %d ------\n", j); device_printf(dev, "hw rdh = %d, hw rdt = %d\n", - E1000_READ_REG(&adapter->hw, E1000_RDH(i)), - E1000_READ_REG(&adapter->hw, E1000_RDT(i))); - device_printf(dev, "RX discarded packets = %ld\n", - rxr->rx_discarded); - device_printf(dev, "RX Next to Check = %d\n", rxr->next_to_check); - device_printf(dev, "RX Next to Refresh = %d\n", rxr->next_to_refresh); + E1000_READ_REG(&adapter->hw, E1000_RDH(j)), + E1000_READ_REG(&adapter->hw, E1000_RDT(j))); } } -#ifdef EM_MULTIQUEUE + /* * 82574 only: * Write a new value to the EEPROM increasing the number of MSIX * vectors from 3 to 5, for proper multiqueue support. */ static void -em_enable_vectors_82574(struct adapter *adapter) +em_enable_vectors_82574(if_ctx_t ctx) { + struct adapter *adapter = iflib_get_softc(ctx); struct e1000_hw *hw = &adapter->hw; - device_t dev = adapter->dev; + device_t dev = iflib_get_dev(ctx); u16 edata; e1000_read_nvm(hw, EM_NVM_PCIE_CTRL, 1, &edata); @@ -6207,7 +4210,7 @@ em_enable_vectors_82574(struct adapter *adapter) device_printf(dev, "Writing to eeprom: done\n"); } } -#endif + #ifdef DDB DB_COMMAND(em_reset_dev, em_ddb_reset_dev) @@ -6223,9 +4226,7 @@ DB_COMMAND(em_reset_dev, em_ddb_reset_dev) dev = devclass_get_device(dc, index); if (device_get_driver(dev) == &em_driver) { struct adapter *adapter = device_get_softc(dev); - EM_CORE_LOCK(adapter); - em_init_locked(adapter); - EM_CORE_UNLOCK(adapter); + em_if_init(adapter->ctx); } } } diff --git a/sys/dev/e1000/if_em.h b/sys/dev/e1000/if_em.h index 2a2bf2ccb7ab..f4cc2338c133 100644 --- a/sys/dev/e1000/if_em.h +++ b/sys/dev/e1000/if_em.h @@ -1,36 +1,67 @@ -/****************************************************************************** - - Copyright (c) 2001-2015, Intel Corporation - All rights reserved. - - Redistribution and use in source and binary forms, with or without - modification, are permitted provided that the following conditions are met: - - 1. Redistributions of source code must retain the above copyright notice, - this list of conditions and the following disclaimer. - - 2. Redistributions in binary form must reproduce the above copyright - notice, this list of conditions and the following disclaimer in the - documentation and/or other materials provided with the distribution. - - 3. Neither the name of the Intel Corporation nor the names of its - contributors may be used to endorse or promote products derived from - this software without specific prior written permission. - - THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" - AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE - LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR - CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF - SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS - INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN - CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) - ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE - POSSIBILITY OF SUCH DAMAGE. - -******************************************************************************/ /*$FreeBSD$*/ +#include "opt_em.h" +#include "opt_ddb.h" +#include "opt_inet.h" +#include "opt_inet6.h" + +#ifdef HAVE_KERNEL_OPTION_HEADERS +#include "opt_device_polling.h" +#endif + +#include +#include +#ifdef DDB +#include +#include +#endif +#if __FreeBSD_version >= 800000 +#include +#endif +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include + +#include "e1000_api.h" +#include "e1000_82571.h" +#include "ifdi_if.h" #ifndef _EM_H_DEFINED_ @@ -51,13 +82,10 @@ * desscriptors should meet the following condition. * (num_tx_desc * sizeof(struct e1000_tx_desc)) % 128 == 0 */ -#define EM_MIN_TXD 80 +#define EM_MIN_TXD 128 #define EM_MAX_TXD 4096 -#ifdef EM_MULTIQUEUE -#define EM_DEFAULT_TXD 4096 -#else -#define EM_DEFAULT_TXD 1024 -#endif +#define EM_DEFAULT_TXD 1024 +#define EM_DEFAULT_MULTI_TXD 4096 /* * EM_RXD - Maximum number of receive Descriptors @@ -72,13 +100,10 @@ * desscriptors should meet the following condition. * (num_tx_desc * sizeof(struct e1000_tx_desc)) % 128 == 0 */ -#define EM_MIN_RXD 80 +#define EM_MIN_RXD 128 #define EM_MAX_RXD 4096 -#ifdef EM_MULTIQUEUE -#define EM_DEFAULT_RXD 4096 -#else -#define EM_DEFAULT_RXD 1024 -#endif +#define EM_DEFAULT_RXD 1024 +#define EM_DEFAULT_MULTI_RXD 4096 /* * EM_TIDV - Transmit Interrupt Delay Value @@ -148,17 +173,6 @@ #define EM_RADV 64 #endif -/* - * This parameter controls the max duration of transmit watchdog. - */ -#define EM_WATCHDOG (10 * hz) - -/* - * This parameter controls when the driver calls the routine to reclaim - * transmit descriptors. - */ -#define EM_TX_CLEANUP_THRESHOLD (adapter->num_tx_desc / 8) - /* * This parameter controls whether or not autonegotation is enabled. * 0 - Disable autonegotiation @@ -221,6 +235,18 @@ #define PCICFG_DESC_RING_STATUS 0xe4 #define FLUSH_DESC_REQUIRED 0x100 + +#define IGB_RX_PTHRESH ((hw->mac.type == e1000_i354) ? 12 : \ + ((hw->mac.type <= e1000_82576) ? 16 : 8)) +#define IGB_RX_HTHRESH 8 +#define IGB_RX_WTHRESH ((hw->mac.type == e1000_82576 && \ + (adapter->intr_type == IFLIB_INTR_MSIX)) ? 1 : 4) + +#define IGB_TX_PTHRESH ((hw->mac.type == e1000_i354) ? 20 : 8) +#define IGB_TX_HTHRESH 1 +#define IGB_TX_WTHRESH ((hw->mac.type != e1000_82575 && \ + (adapter->intr_type == IFLIB_INTR_MSIX) ? 1 : 16) + /* * TDBA/RDBA should be aligned on 16 byte boundary. But TDLEN/RDLEN should be * multiple of 128 bytes. So we align TDBA/RDBA on 128 byte boundary. This will @@ -242,6 +268,7 @@ #define EM_BAR_TYPE(v) ((v) & EM_BAR_TYPE_MASK) #define EM_BAR_TYPE_MASK 0x00000001 #define EM_BAR_TYPE_MMEM 0x00000000 +#define EM_BAR_TYPE_IO 0x00000001 #define EM_BAR_TYPE_FLASH 0x0014 #define EM_BAR_MEM_TYPE(v) ((v) & EM_BAR_MEM_TYPE_MASK) #define EM_BAR_MEM_TYPE_MASK 0x00000006 @@ -279,6 +306,9 @@ #define ETH_ADDR_LEN 6 #define CSUM_OFFLOAD 7 /* Offload bits in mbuf flag */ +#define IGB_PKTTYPE_MASK 0x0000FFF0 +#define IGB_DMCTLX_DCFLUSH_DIS 0x80000000 /* Disable DMA Coalesce Flush */ + /* * 82574 has a nonstandard address for EIAC * and since its only used in MSIX, and in @@ -295,19 +325,6 @@ #define EM_NVM_MSIX_N_MASK (0x7 << EM_NVM_MSIX_N_SHIFT) #define EM_NVM_MSIX_N_SHIFT 7 -/* - * Bus dma allocation structure used by - * e1000_dma_malloc and e1000_dma_free. - */ -struct em_dma_alloc { - bus_addr_t dma_paddr; - caddr_t dma_vaddr; - bus_dma_tag_t dma_tag; - bus_dmamap_t dma_map; - bus_dma_segment_t dma_seg; - int dma_nseg; -}; - struct adapter; struct em_int_delay_info { @@ -321,35 +338,31 @@ struct em_int_delay_info { */ struct tx_ring { struct adapter *adapter; - struct mtx tx_mtx; - char mtx_name[16]; + struct em_tx_queue *que; u32 me; - u32 msix; - u32 ims; int busy; - struct em_dma_alloc txdma; struct e1000_tx_desc *tx_base; - struct task tx_task; - struct taskqueue *tq; - u32 next_avail_desc; - u32 next_to_clean; + uint64_t tx_paddr; struct em_txbuffer *tx_buffers; - volatile u16 tx_avail; u32 tx_tso; /* last tx was tso */ - u16 last_hw_offload; - u8 last_hw_ipcso; - u8 last_hw_ipcss; - u8 last_hw_tucso; - u8 last_hw_tucss; -#if __FreeBSD_version >= 800000 - struct buf_ring *br; -#endif + /* Interrupt resources */ - bus_dma_tag_t txtag; void *tag; struct resource *res; unsigned long tx_irq; unsigned long no_desc_avail; + + /* Saved csum offloading context information */ + int csum_flags; + int csum_lhlen; + int csum_iphlen; + + int csum_thlen; + int csum_mss; + int csum_pktlen; + + uint32_t csum_txd_upper; + uint32_t csum_txd_lower; /* last field */ }; /* @@ -357,26 +370,15 @@ struct tx_ring { */ struct rx_ring { struct adapter *adapter; + struct em_rx_queue *que; u32 me; - u32 msix; - u32 ims; - struct mtx rx_mtx; - char mtx_name[16]; u32 payload; - struct task rx_task; - struct taskqueue *tq; union e1000_rx_desc_extended *rx_base; - struct em_dma_alloc rxdma; - u32 next_to_refresh; - u32 next_to_check; - struct em_rxbuffer *rx_buffers; - struct mbuf *fmp; - struct mbuf *lmp; + uint64_t rx_paddr; /* Interrupt resources */ void *tag; struct resource *res; - bus_dma_tag_t rxtag; bool discard; /* Soft stats */ @@ -386,62 +388,68 @@ struct rx_ring { unsigned long rx_bytes; }; +struct em_tx_queue { + struct adapter *adapter; + u32 msix; + u32 eims; /* This queue's EIMS bit */ + u32 me; + struct tx_ring txr; +}; + +struct em_rx_queue { + struct adapter *adapter; + u32 me; + u32 msix; + u32 eims; + struct rx_ring rxr; + u64 irqs; + struct if_irq que_irq; +}; /* Our adapter structure */ struct adapter { - if_t ifp; + struct ifnet *ifp; struct e1000_hw hw; + if_softc_ctx_t shared; + if_ctx_t ctx; +#define tx_num_queues shared->isc_ntxqsets +#define rx_num_queues shared->isc_nrxqsets +#define intr_type shared->isc_intr /* FreeBSD operating-system-specific structures. */ struct e1000_osdep osdep; - device_t dev; + struct device *dev; struct cdev *led_dev; + struct em_tx_queue *tx_queues; + struct em_rx_queue *rx_queues; + struct if_irq irq; + struct resource *memory; struct resource *flash; - struct resource *msix_mem; + struct resource *ioport; + int io_rid; struct resource *res; void *tag; u32 linkvec; u32 ivars; - struct ifmedia media; - struct callout timer; + struct ifmedia *media; int msix; int if_flags; - int max_frame_size; int min_frame_size; - struct mtx core_mtx; int em_insert_vlan_header; u32 ims; bool in_detach; /* Task for FAST handling */ - struct task link_task; - struct task que_task; - struct taskqueue *tq; /* private task queue */ + struct grouptask link_task; - eventhandler_tag vlan_attach; - eventhandler_tag vlan_detach; - - u16 num_vlans; - u8 num_queues; - - /* - * Transmit rings: - * Allocated at run time, an array of rings. - */ - struct tx_ring *tx_rings; - int num_tx_desc; + u16 num_vlans; u32 txd_cmd; - /* - * Receive rings: - * Allocated at run time, an array of rings. - */ - struct rx_ring *rx_rings; - int num_rx_desc; + u32 tx_process_limit; u32 rx_process_limit; u32 rx_mbuf_sz; @@ -467,7 +475,12 @@ struct adapter { u16 link_speed; u16 link_duplex; u32 smartspeed; + u32 dmac; + int link_mask; + u64 que_mask; + + struct em_int_delay_info tx_int_delay; struct em_int_delay_info tx_abs_int_delay; struct em_int_delay_info rx_int_delay; @@ -502,33 +515,9 @@ typedef struct _em_vendor_info_t { } em_vendor_info_t; struct em_txbuffer { - int next_eop; /* Index of the desc to watch */ - struct mbuf *m_head; - bus_dmamap_t map; /* bus_dma map for packet */ + int eop; }; -struct em_rxbuffer { - int next_eop; /* Index of the desc to watch */ - struct mbuf *m_head; - bus_dmamap_t map; /* bus_dma map for packet */ - bus_addr_t paddr; -}; - - -/* -** Find the number of unrefreshed RX descriptors -*/ -static inline u16 -e1000_rx_unrefreshed(struct rx_ring *rxr) -{ - struct adapter *adapter = rxr->adapter; - - if (rxr->next_to_check > rxr->next_to_refresh) - return (rxr->next_to_check - rxr->next_to_refresh - 1); - else - return ((adapter->num_rx_desc + rxr->next_to_check) - - rxr->next_to_refresh - 1); -} #define EM_CORE_LOCK_INIT(_sc, _name) \ mtx_init(&(_sc)->core_mtx, _name, "EM Core Lock", MTX_DEF) diff --git a/sys/dev/e1000/if_igb.c b/sys/dev/e1000/if_igb.c deleted file mode 100644 index 3c9644d9b03b..000000000000 --- a/sys/dev/e1000/if_igb.c +++ /dev/null @@ -1,6450 +0,0 @@ -/****************************************************************************** - - Copyright (c) 2001-2015, Intel Corporation - All rights reserved. - - Redistribution and use in source and binary forms, with or without - modification, are permitted provided that the following conditions are met: - - 1. Redistributions of source code must retain the above copyright notice, - this list of conditions and the following disclaimer. - - 2. Redistributions in binary form must reproduce the above copyright - notice, this list of conditions and the following disclaimer in the - documentation and/or other materials provided with the distribution. - - 3. Neither the name of the Intel Corporation nor the names of its - contributors may be used to endorse or promote products derived from - this software without specific prior written permission. - - THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" - AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE - LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR - CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF - SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS - INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN - CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) - ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE - POSSIBILITY OF SUCH DAMAGE. - -******************************************************************************/ -/*$FreeBSD$*/ - - -#include "opt_inet.h" -#include "opt_inet6.h" -#include "opt_rss.h" - -#ifdef HAVE_KERNEL_OPTION_HEADERS -#include "opt_device_polling.h" -#include "opt_altq.h" -#endif - -#include "if_igb.h" - -/********************************************************************* - * Driver version: - *********************************************************************/ -char igb_driver_version[] = "2.5.3-k"; - - -/********************************************************************* - * PCI Device ID Table - * - * Used by probe to select devices to load on - * Last field stores an index into e1000_strings - * Last entry must be all 0s - * - * { Vendor ID, Device ID, SubVendor ID, SubDevice ID, String Index } - *********************************************************************/ - -static igb_vendor_info_t igb_vendor_info_array[] = -{ - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82575EB_COPPER, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82575EB_FIBER_SERDES, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82575GB_QUAD_COPPER, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82576, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82576_NS, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82576_NS_SERDES, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82576_FIBER, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82576_SERDES, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82576_SERDES_QUAD, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82576_QUAD_COPPER, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82576_QUAD_COPPER_ET2, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82576_VF, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82580_COPPER, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82580_FIBER, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82580_SERDES, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82580_SGMII, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82580_COPPER_DUAL, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_82580_QUAD_FIBER, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_DH89XXCC_SERDES, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_DH89XXCC_SGMII, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_DH89XXCC_SFP, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_DH89XXCC_BACKPLANE, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I350_COPPER, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I350_FIBER, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I350_SERDES, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I350_SGMII, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I350_VF, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I210_COPPER, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I210_COPPER_IT, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I210_COPPER_OEM1, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I210_COPPER_FLASHLESS, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I210_SERDES_FLASHLESS, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I210_FIBER, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I210_SERDES, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I210_SGMII, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I211_COPPER, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I354_BACKPLANE_1GBPS, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I354_BACKPLANE_2_5GBPS, 0, 0, 0}, - {IGB_INTEL_VENDOR_ID, E1000_DEV_ID_I354_SGMII, 0, 0, 0}, - /* required last entry */ - {0, 0, 0, 0, 0} -}; - -/********************************************************************* - * Table of branding strings for all supported NICs. - *********************************************************************/ - -static char *igb_strings[] = { - "Intel(R) PRO/1000 Network Connection" -}; - -/********************************************************************* - * Function prototypes - *********************************************************************/ -static int igb_probe(device_t); -static int igb_attach(device_t); -static int igb_detach(device_t); -static int igb_shutdown(device_t); -static int igb_suspend(device_t); -static int igb_resume(device_t); -#ifndef IGB_LEGACY_TX -static int igb_mq_start(struct ifnet *, struct mbuf *); -static int igb_mq_start_locked(struct ifnet *, struct tx_ring *); -static void igb_qflush(struct ifnet *); -static void igb_deferred_mq_start(void *, int); -#else -static void igb_start(struct ifnet *); -static void igb_start_locked(struct tx_ring *, struct ifnet *ifp); -#endif -static int igb_ioctl(struct ifnet *, u_long, caddr_t); -static uint64_t igb_get_counter(if_t, ift_counter); -static void igb_init(void *); -static void igb_init_locked(struct adapter *); -static void igb_stop(void *); -static void igb_media_status(struct ifnet *, struct ifmediareq *); -static int igb_media_change(struct ifnet *); -static void igb_identify_hardware(struct adapter *); -static int igb_allocate_pci_resources(struct adapter *); -static int igb_allocate_msix(struct adapter *); -static int igb_allocate_legacy(struct adapter *); -static int igb_setup_msix(struct adapter *); -static void igb_free_pci_resources(struct adapter *); -static void igb_local_timer(void *); -static void igb_reset(struct adapter *); -static int igb_setup_interface(device_t, struct adapter *); -static int igb_allocate_queues(struct adapter *); -static void igb_configure_queues(struct adapter *); - -static int igb_allocate_transmit_buffers(struct tx_ring *); -static void igb_setup_transmit_structures(struct adapter *); -static void igb_setup_transmit_ring(struct tx_ring *); -static void igb_initialize_transmit_units(struct adapter *); -static void igb_free_transmit_structures(struct adapter *); -static void igb_free_transmit_buffers(struct tx_ring *); - -static int igb_allocate_receive_buffers(struct rx_ring *); -static int igb_setup_receive_structures(struct adapter *); -static int igb_setup_receive_ring(struct rx_ring *); -static void igb_initialize_receive_units(struct adapter *); -static void igb_free_receive_structures(struct adapter *); -static void igb_free_receive_buffers(struct rx_ring *); -static void igb_free_receive_ring(struct rx_ring *); - -static void igb_enable_intr(struct adapter *); -static void igb_disable_intr(struct adapter *); -static void igb_update_stats_counters(struct adapter *); -static bool igb_txeof(struct tx_ring *); - -static __inline void igb_rx_discard(struct rx_ring *, int); -static __inline void igb_rx_input(struct rx_ring *, - struct ifnet *, struct mbuf *, u32); - -static bool igb_rxeof(struct igb_queue *, int, int *); -static void igb_rx_checksum(u32, struct mbuf *, u32); -static int igb_tx_ctx_setup(struct tx_ring *, - struct mbuf *, u32 *, u32 *); -static int igb_tso_setup(struct tx_ring *, - struct mbuf *, u32 *, u32 *); -static void igb_set_promisc(struct adapter *); -static void igb_disable_promisc(struct adapter *); -static void igb_set_multi(struct adapter *); -static void igb_update_link_status(struct adapter *); -static void igb_refresh_mbufs(struct rx_ring *, int); - -static void igb_register_vlan(void *, struct ifnet *, u16); -static void igb_unregister_vlan(void *, struct ifnet *, u16); -static void igb_setup_vlan_hw_support(struct adapter *); - -static int igb_xmit(struct tx_ring *, struct mbuf **); -static int igb_dma_malloc(struct adapter *, bus_size_t, - struct igb_dma_alloc *, int); -static void igb_dma_free(struct adapter *, struct igb_dma_alloc *); -static int igb_sysctl_nvm_info(SYSCTL_HANDLER_ARGS); -static void igb_print_nvm_info(struct adapter *); -static int igb_is_valid_ether_addr(u8 *); -static void igb_add_hw_stats(struct adapter *); - -static void igb_vf_init_stats(struct adapter *); -static void igb_update_vf_stats_counters(struct adapter *); - -/* Management and WOL Support */ -static void igb_init_manageability(struct adapter *); -static void igb_release_manageability(struct adapter *); -static void igb_get_hw_control(struct adapter *); -static void igb_release_hw_control(struct adapter *); -static void igb_enable_wakeup(device_t); -static void igb_led_func(void *, int); - -static int igb_irq_fast(void *); -static void igb_msix_que(void *); -static void igb_msix_link(void *); -static void igb_handle_que(void *context, int pending); -static void igb_handle_link(void *context, int pending); -static void igb_handle_link_locked(struct adapter *); - -static void igb_set_sysctl_value(struct adapter *, const char *, - const char *, int *, int); -static int igb_set_flowcntl(SYSCTL_HANDLER_ARGS); -static int igb_sysctl_dmac(SYSCTL_HANDLER_ARGS); -static int igb_sysctl_eee(SYSCTL_HANDLER_ARGS); - -#ifdef DEVICE_POLLING -static poll_handler_t igb_poll; -#endif /* POLLING */ - -/********************************************************************* - * FreeBSD Device Interface Entry Points - *********************************************************************/ - -static device_method_t igb_methods[] = { - /* Device interface */ - DEVMETHOD(device_probe, igb_probe), - DEVMETHOD(device_attach, igb_attach), - DEVMETHOD(device_detach, igb_detach), - DEVMETHOD(device_shutdown, igb_shutdown), - DEVMETHOD(device_suspend, igb_suspend), - DEVMETHOD(device_resume, igb_resume), - DEVMETHOD_END -}; - -static driver_t igb_driver = { - "igb", igb_methods, sizeof(struct adapter), -}; - -static devclass_t igb_devclass; -DRIVER_MODULE(igb, pci, igb_driver, igb_devclass, 0, 0); -MODULE_DEPEND(igb, pci, 1, 1, 1); -MODULE_DEPEND(igb, ether, 1, 1, 1); -#ifdef DEV_NETMAP -MODULE_DEPEND(igb, netmap, 1, 1, 1); -#endif /* DEV_NETMAP */ - -/********************************************************************* - * Tunable default values. - *********************************************************************/ - -static SYSCTL_NODE(_hw, OID_AUTO, igb, CTLFLAG_RD, 0, "IGB driver parameters"); - -/* Descriptor defaults */ -static int igb_rxd = IGB_DEFAULT_RXD; -static int igb_txd = IGB_DEFAULT_TXD; -SYSCTL_INT(_hw_igb, OID_AUTO, rxd, CTLFLAG_RDTUN, &igb_rxd, 0, - "Number of receive descriptors per queue"); -SYSCTL_INT(_hw_igb, OID_AUTO, txd, CTLFLAG_RDTUN, &igb_txd, 0, - "Number of transmit descriptors per queue"); - -/* -** AIM: Adaptive Interrupt Moderation -** which means that the interrupt rate -** is varied over time based on the -** traffic for that interrupt vector -*/ -static int igb_enable_aim = TRUE; -SYSCTL_INT(_hw_igb, OID_AUTO, enable_aim, CTLFLAG_RWTUN, &igb_enable_aim, 0, - "Enable adaptive interrupt moderation"); - -/* - * MSIX should be the default for best performance, - * but this allows it to be forced off for testing. - */ -static int igb_enable_msix = 1; -SYSCTL_INT(_hw_igb, OID_AUTO, enable_msix, CTLFLAG_RDTUN, &igb_enable_msix, 0, - "Enable MSI-X interrupts"); - -/* -** Tuneable Interrupt rate -*/ -static int igb_max_interrupt_rate = 8000; -SYSCTL_INT(_hw_igb, OID_AUTO, max_interrupt_rate, CTLFLAG_RDTUN, - &igb_max_interrupt_rate, 0, "Maximum interrupts per second"); - -#ifndef IGB_LEGACY_TX -/* -** Tuneable number of buffers in the buf-ring (drbr_xxx) -*/ -static int igb_buf_ring_size = IGB_BR_SIZE; -SYSCTL_INT(_hw_igb, OID_AUTO, buf_ring_size, CTLFLAG_RDTUN, - &igb_buf_ring_size, 0, "Size of the bufring"); -#endif - -/* -** Header split causes the packet header to -** be dma'd to a separate mbuf from the payload. -** this can have memory alignment benefits. But -** another plus is that small packets often fit -** into the header and thus use no cluster. Its -** a very workload dependent type feature. -*/ -static int igb_header_split = FALSE; -SYSCTL_INT(_hw_igb, OID_AUTO, header_split, CTLFLAG_RDTUN, &igb_header_split, 0, - "Enable receive mbuf header split"); - -/* -** This will autoconfigure based on the -** number of CPUs and max supported -** MSIX messages if left at 0. -*/ -static int igb_num_queues = 0; -SYSCTL_INT(_hw_igb, OID_AUTO, num_queues, CTLFLAG_RDTUN, &igb_num_queues, 0, - "Number of queues to configure, 0 indicates autoconfigure"); - -/* -** Global variable to store last used CPU when binding queues -** to CPUs in igb_allocate_msix. Starts at CPU_FIRST and increments when a -** queue is bound to a cpu. -*/ -static int igb_last_bind_cpu = -1; - -/* How many packets rxeof tries to clean at a time */ -static int igb_rx_process_limit = 100; -SYSCTL_INT(_hw_igb, OID_AUTO, rx_process_limit, CTLFLAG_RDTUN, - &igb_rx_process_limit, 0, - "Maximum number of received packets to process at a time, -1 means unlimited"); - -/* How many packets txeof tries to clean at a time */ -static int igb_tx_process_limit = -1; -SYSCTL_INT(_hw_igb, OID_AUTO, tx_process_limit, CTLFLAG_RDTUN, - &igb_tx_process_limit, 0, - "Maximum number of sent packets to process at a time, -1 means unlimited"); - -#ifdef DEV_NETMAP /* see ixgbe.c for details */ -#include -#endif /* DEV_NETMAP */ -/********************************************************************* - * Device identification routine - * - * igb_probe determines if the driver should be loaded on - * adapter based on PCI vendor/device id of the adapter. - * - * return BUS_PROBE_DEFAULT on success, positive on failure - *********************************************************************/ - -static int -igb_probe(device_t dev) -{ - char adapter_name[256]; - uint16_t pci_vendor_id = 0; - uint16_t pci_device_id = 0; - uint16_t pci_subvendor_id = 0; - uint16_t pci_subdevice_id = 0; - igb_vendor_info_t *ent; - - INIT_DEBUGOUT("igb_probe: begin"); - - pci_vendor_id = pci_get_vendor(dev); - if (pci_vendor_id != IGB_INTEL_VENDOR_ID) - return (ENXIO); - - pci_device_id = pci_get_device(dev); - pci_subvendor_id = pci_get_subvendor(dev); - pci_subdevice_id = pci_get_subdevice(dev); - - ent = igb_vendor_info_array; - while (ent->vendor_id != 0) { - if ((pci_vendor_id == ent->vendor_id) && - (pci_device_id == ent->device_id) && - - ((pci_subvendor_id == ent->subvendor_id) || - (ent->subvendor_id == 0)) && - - ((pci_subdevice_id == ent->subdevice_id) || - (ent->subdevice_id == 0))) { - sprintf(adapter_name, "%s, Version - %s", - igb_strings[ent->index], - igb_driver_version); - device_set_desc_copy(dev, adapter_name); - return (BUS_PROBE_DEFAULT); - } - ent++; - } - return (ENXIO); -} - -/********************************************************************* - * Device initialization routine - * - * The attach entry point is called when the driver is being loaded. - * This routine identifies the type of hardware, allocates all resources - * and initializes the hardware. - * - * return 0 on success, positive on failure - *********************************************************************/ - -static int -igb_attach(device_t dev) -{ - struct adapter *adapter; - int error = 0; - u16 eeprom_data; - - INIT_DEBUGOUT("igb_attach: begin"); - - if (resource_disabled("igb", device_get_unit(dev))) { - device_printf(dev, "Disabled by device hint\n"); - return (ENXIO); - } - - adapter = device_get_softc(dev); - adapter->dev = adapter->osdep.dev = dev; - IGB_CORE_LOCK_INIT(adapter, device_get_nameunit(dev)); - - /* SYSCTLs */ - SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), - SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), - OID_AUTO, "nvm", CTLTYPE_INT|CTLFLAG_RW, adapter, 0, - igb_sysctl_nvm_info, "I", "NVM Information"); - - igb_set_sysctl_value(adapter, "enable_aim", - "Interrupt Moderation", &adapter->enable_aim, - igb_enable_aim); - - SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), - SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), - OID_AUTO, "fc", CTLTYPE_INT|CTLFLAG_RW, - adapter, 0, igb_set_flowcntl, "I", "Flow Control"); - - callout_init_mtx(&adapter->timer, &adapter->core_mtx, 0); - - /* Determine hardware and mac info */ - igb_identify_hardware(adapter); - - /* Setup PCI resources */ - if (igb_allocate_pci_resources(adapter)) { - device_printf(dev, "Allocation of PCI resources failed\n"); - error = ENXIO; - goto err_pci; - } - - /* Do Shared Code initialization */ - if (e1000_setup_init_funcs(&adapter->hw, TRUE)) { - device_printf(dev, "Setup of Shared code failed\n"); - error = ENXIO; - goto err_pci; - } - - e1000_get_bus_info(&adapter->hw); - - /* Sysctls for limiting the amount of work done in the taskqueues */ - igb_set_sysctl_value(adapter, "rx_processing_limit", - "max number of rx packets to process", - &adapter->rx_process_limit, igb_rx_process_limit); - - igb_set_sysctl_value(adapter, "tx_processing_limit", - "max number of tx packets to process", - &adapter->tx_process_limit, igb_tx_process_limit); - - /* - * Validate number of transmit and receive descriptors. It - * must not exceed hardware maximum, and must be multiple - * of E1000_DBA_ALIGN. - */ - if (((igb_txd * sizeof(struct e1000_tx_desc)) % IGB_DBA_ALIGN) != 0 || - (igb_txd > IGB_MAX_TXD) || (igb_txd < IGB_MIN_TXD)) { - device_printf(dev, "Using %d TX descriptors instead of %d!\n", - IGB_DEFAULT_TXD, igb_txd); - adapter->num_tx_desc = IGB_DEFAULT_TXD; - } else - adapter->num_tx_desc = igb_txd; - if (((igb_rxd * sizeof(struct e1000_rx_desc)) % IGB_DBA_ALIGN) != 0 || - (igb_rxd > IGB_MAX_RXD) || (igb_rxd < IGB_MIN_RXD)) { - device_printf(dev, "Using %d RX descriptors instead of %d!\n", - IGB_DEFAULT_RXD, igb_rxd); - adapter->num_rx_desc = IGB_DEFAULT_RXD; - } else - adapter->num_rx_desc = igb_rxd; - - adapter->hw.mac.autoneg = DO_AUTO_NEG; - adapter->hw.phy.autoneg_wait_to_complete = FALSE; - adapter->hw.phy.autoneg_advertised = AUTONEG_ADV_DEFAULT; - - /* Copper options */ - if (adapter->hw.phy.media_type == e1000_media_type_copper) { - adapter->hw.phy.mdix = AUTO_ALL_MODES; - adapter->hw.phy.disable_polarity_correction = FALSE; - adapter->hw.phy.ms_type = IGB_MASTER_SLAVE; - } - - /* - * Set the frame limits assuming - * standard ethernet sized frames. - */ - adapter->max_frame_size = ETHERMTU + ETHER_HDR_LEN + ETHERNET_FCS_SIZE; - - /* - ** Allocate and Setup Queues - */ - if (igb_allocate_queues(adapter)) { - error = ENOMEM; - goto err_pci; - } - - /* Allocate the appropriate stats memory */ - if (adapter->vf_ifp) { - adapter->stats = - (struct e1000_vf_stats *)malloc(sizeof \ - (struct e1000_vf_stats), M_DEVBUF, M_NOWAIT | M_ZERO); - igb_vf_init_stats(adapter); - } else - adapter->stats = - (struct e1000_hw_stats *)malloc(sizeof \ - (struct e1000_hw_stats), M_DEVBUF, M_NOWAIT | M_ZERO); - if (adapter->stats == NULL) { - device_printf(dev, "Can not allocate stats memory\n"); - error = ENOMEM; - goto err_late; - } - - /* Allocate multicast array memory. */ - adapter->mta = malloc(sizeof(u8) * ETH_ADDR_LEN * - MAX_NUM_MULTICAST_ADDRESSES, M_DEVBUF, M_NOWAIT); - if (adapter->mta == NULL) { - device_printf(dev, "Can not allocate multicast setup array\n"); - error = ENOMEM; - goto err_late; - } - - /* Some adapter-specific advanced features */ - if (adapter->hw.mac.type >= e1000_i350) { - SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), - SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), - OID_AUTO, "dmac", CTLTYPE_INT|CTLFLAG_RW, - adapter, 0, igb_sysctl_dmac, "I", "DMA Coalesce"); - SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), - SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), - OID_AUTO, "eee_disabled", CTLTYPE_INT|CTLFLAG_RW, - adapter, 0, igb_sysctl_eee, "I", - "Disable Energy Efficient Ethernet"); - if (adapter->hw.phy.media_type == e1000_media_type_copper) { - if (adapter->hw.mac.type == e1000_i354) - e1000_set_eee_i354(&adapter->hw, TRUE, TRUE); - else - e1000_set_eee_i350(&adapter->hw, TRUE, TRUE); - } - } - - /* - ** Start from a known state, this is - ** important in reading the nvm and - ** mac from that. - */ - e1000_reset_hw(&adapter->hw); - - /* Make sure we have a good EEPROM before we read from it */ - if (((adapter->hw.mac.type != e1000_i210) && - (adapter->hw.mac.type != e1000_i211)) && - (e1000_validate_nvm_checksum(&adapter->hw) < 0)) { - /* - ** Some PCI-E parts fail the first check due to - ** the link being in sleep state, call it again, - ** if it fails a second time its a real issue. - */ - if (e1000_validate_nvm_checksum(&adapter->hw) < 0) { - device_printf(dev, - "The EEPROM Checksum Is Not Valid\n"); - error = EIO; - goto err_late; - } - } - - /* - ** Copy the permanent MAC address out of the EEPROM - */ - if (e1000_read_mac_addr(&adapter->hw) < 0) { - device_printf(dev, "EEPROM read error while reading MAC" - " address\n"); - error = EIO; - goto err_late; - } - - /* Check its sanity */ - if (!igb_is_valid_ether_addr(adapter->hw.mac.addr)) { - if (adapter->vf_ifp) { - u8 addr[ETHER_ADDR_LEN]; - arc4rand(&addr, sizeof(addr), 0); - addr[0] &= 0xFE; - addr[0] |= 0x02; - bcopy(addr, adapter->hw.mac.addr, sizeof(addr)); - } else { - device_printf(dev, "Invalid MAC address\n"); - error = EIO; - goto err_late; - } - } - - /* Setup OS specific network interface */ - if (igb_setup_interface(dev, adapter) != 0) - goto err_late; - - /* Now get a good starting state */ - igb_reset(adapter); - - /* Initialize statistics */ - igb_update_stats_counters(adapter); - - adapter->hw.mac.get_link_status = 1; - igb_update_link_status(adapter); - - /* Indicate SOL/IDER usage */ - if (e1000_check_reset_block(&adapter->hw)) - device_printf(dev, - "PHY reset is blocked due to SOL/IDER session.\n"); - - /* Determine if we have to control management hardware */ - adapter->has_manage = e1000_enable_mng_pass_thru(&adapter->hw); - - /* - * Setup Wake-on-Lan - */ - /* APME bit in EEPROM is mapped to WUC.APME */ - eeprom_data = E1000_READ_REG(&adapter->hw, E1000_WUC) & E1000_WUC_APME; - if (eeprom_data) - adapter->wol = E1000_WUFC_MAG; - - /* Register for VLAN events */ - adapter->vlan_attach = EVENTHANDLER_REGISTER(vlan_config, - igb_register_vlan, adapter, EVENTHANDLER_PRI_FIRST); - adapter->vlan_detach = EVENTHANDLER_REGISTER(vlan_unconfig, - igb_unregister_vlan, adapter, EVENTHANDLER_PRI_FIRST); - - igb_add_hw_stats(adapter); - - /* Tell the stack that the interface is not active */ - adapter->ifp->if_drv_flags &= ~IFF_DRV_RUNNING; - adapter->ifp->if_drv_flags |= IFF_DRV_OACTIVE; - - adapter->led_dev = led_create(igb_led_func, adapter, - device_get_nameunit(dev)); - - /* - ** Configure Interrupts - */ - if ((adapter->msix > 1) && (igb_enable_msix)) - error = igb_allocate_msix(adapter); - else /* MSI or Legacy */ - error = igb_allocate_legacy(adapter); - if (error) - goto err_late; - -#ifdef DEV_NETMAP - igb_netmap_attach(adapter); -#endif /* DEV_NETMAP */ - INIT_DEBUGOUT("igb_attach: end"); - - return (0); - -err_late: - if (igb_detach(dev) == 0) /* igb_detach() already did the cleanup */ - return(error); - igb_free_transmit_structures(adapter); - igb_free_receive_structures(adapter); - igb_release_hw_control(adapter); -err_pci: - igb_free_pci_resources(adapter); - if (adapter->ifp != NULL) - if_free(adapter->ifp); - free(adapter->mta, M_DEVBUF); - IGB_CORE_LOCK_DESTROY(adapter); - - return (error); -} - -/********************************************************************* - * Device removal routine - * - * The detach entry point is called when the driver is being removed. - * This routine stops the adapter and deallocates all the resources - * that were allocated for driver operation. - * - * return 0 on success, positive on failure - *********************************************************************/ - -static int -igb_detach(device_t dev) -{ - struct adapter *adapter = device_get_softc(dev); - struct ifnet *ifp = adapter->ifp; - - INIT_DEBUGOUT("igb_detach: begin"); - - /* Make sure VLANS are not using driver */ - if (adapter->ifp->if_vlantrunk != NULL) { - device_printf(dev,"Vlan in use, detach first\n"); - return (EBUSY); - } - - ether_ifdetach(adapter->ifp); - - if (adapter->led_dev != NULL) - led_destroy(adapter->led_dev); - -#ifdef DEVICE_POLLING - if (ifp->if_capenable & IFCAP_POLLING) - ether_poll_deregister(ifp); -#endif - - IGB_CORE_LOCK(adapter); - adapter->in_detach = 1; - igb_stop(adapter); - IGB_CORE_UNLOCK(adapter); - - e1000_phy_hw_reset(&adapter->hw); - - /* Give control back to firmware */ - igb_release_manageability(adapter); - igb_release_hw_control(adapter); - - if (adapter->wol) { - E1000_WRITE_REG(&adapter->hw, E1000_WUC, E1000_WUC_PME_EN); - E1000_WRITE_REG(&adapter->hw, E1000_WUFC, adapter->wol); - igb_enable_wakeup(dev); - } - - /* Unregister VLAN events */ - if (adapter->vlan_attach != NULL) - EVENTHANDLER_DEREGISTER(vlan_config, adapter->vlan_attach); - if (adapter->vlan_detach != NULL) - EVENTHANDLER_DEREGISTER(vlan_unconfig, adapter->vlan_detach); - - callout_drain(&adapter->timer); - -#ifdef DEV_NETMAP - netmap_detach(adapter->ifp); -#endif /* DEV_NETMAP */ - igb_free_pci_resources(adapter); - bus_generic_detach(dev); - if_free(ifp); - - igb_free_transmit_structures(adapter); - igb_free_receive_structures(adapter); - if (adapter->mta != NULL) - free(adapter->mta, M_DEVBUF); - - IGB_CORE_LOCK_DESTROY(adapter); - - return (0); -} - -/********************************************************************* - * - * Shutdown entry point - * - **********************************************************************/ - -static int -igb_shutdown(device_t dev) -{ - return igb_suspend(dev); -} - -/* - * Suspend/resume device methods. - */ -static int -igb_suspend(device_t dev) -{ - struct adapter *adapter = device_get_softc(dev); - - IGB_CORE_LOCK(adapter); - - igb_stop(adapter); - - igb_release_manageability(adapter); - igb_release_hw_control(adapter); - - if (adapter->wol) { - E1000_WRITE_REG(&adapter->hw, E1000_WUC, E1000_WUC_PME_EN); - E1000_WRITE_REG(&adapter->hw, E1000_WUFC, adapter->wol); - igb_enable_wakeup(dev); - } - - IGB_CORE_UNLOCK(adapter); - - return bus_generic_suspend(dev); -} - -static int -igb_resume(device_t dev) -{ - struct adapter *adapter = device_get_softc(dev); - struct tx_ring *txr = adapter->tx_rings; - struct ifnet *ifp = adapter->ifp; - - IGB_CORE_LOCK(adapter); - igb_init_locked(adapter); - igb_init_manageability(adapter); - - if ((ifp->if_flags & IFF_UP) && - (ifp->if_drv_flags & IFF_DRV_RUNNING) && adapter->link_active) { - for (int i = 0; i < adapter->num_queues; i++, txr++) { - IGB_TX_LOCK(txr); -#ifndef IGB_LEGACY_TX - /* Process the stack queue only if not depleted */ - if (((txr->queue_status & IGB_QUEUE_DEPLETED) == 0) && - !drbr_empty(ifp, txr->br)) - igb_mq_start_locked(ifp, txr); -#else - if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) - igb_start_locked(txr, ifp); -#endif - IGB_TX_UNLOCK(txr); - } - } - IGB_CORE_UNLOCK(adapter); - - return bus_generic_resume(dev); -} - - -#ifdef IGB_LEGACY_TX - -/********************************************************************* - * Transmit entry point - * - * igb_start is called by the stack to initiate a transmit. - * The driver will remain in this routine as long as there are - * packets to transmit and transmit resources are available. - * In case resources are not available stack is notified and - * the packet is requeued. - **********************************************************************/ - -static void -igb_start_locked(struct tx_ring *txr, struct ifnet *ifp) -{ - struct adapter *adapter = ifp->if_softc; - struct mbuf *m_head; - - IGB_TX_LOCK_ASSERT(txr); - - if ((ifp->if_drv_flags & (IFF_DRV_RUNNING|IFF_DRV_OACTIVE)) != - IFF_DRV_RUNNING) - return; - if (!adapter->link_active) - return; - - /* Call cleanup if number of TX descriptors low */ - if (txr->tx_avail <= IGB_TX_CLEANUP_THRESHOLD) - igb_txeof(txr); - - while (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) { - if (txr->tx_avail <= IGB_MAX_SCATTER) { - txr->queue_status |= IGB_QUEUE_DEPLETED; - break; - } - IFQ_DRV_DEQUEUE(&ifp->if_snd, m_head); - if (m_head == NULL) - break; - /* - * Encapsulation can modify our pointer, and or make it - * NULL on failure. In that event, we can't requeue. - */ - if (igb_xmit(txr, &m_head)) { - if (m_head != NULL) - IFQ_DRV_PREPEND(&ifp->if_snd, m_head); - if (txr->tx_avail <= IGB_MAX_SCATTER) - txr->queue_status |= IGB_QUEUE_DEPLETED; - break; - } - - /* Send a copy of the frame to the BPF listener */ - ETHER_BPF_MTAP(ifp, m_head); - - /* Set watchdog on */ - txr->watchdog_time = ticks; - txr->queue_status |= IGB_QUEUE_WORKING; - } -} - -/* - * Legacy TX driver routine, called from the - * stack, always uses tx[0], and spins for it. - * Should not be used with multiqueue tx - */ -static void -igb_start(struct ifnet *ifp) -{ - struct adapter *adapter = ifp->if_softc; - struct tx_ring *txr = adapter->tx_rings; - - if (ifp->if_drv_flags & IFF_DRV_RUNNING) { - IGB_TX_LOCK(txr); - igb_start_locked(txr, ifp); - IGB_TX_UNLOCK(txr); - } - return; -} - -#else /* ~IGB_LEGACY_TX */ - -/* -** Multiqueue Transmit Entry: -** quick turnaround to the stack -** -*/ -static int -igb_mq_start(struct ifnet *ifp, struct mbuf *m) -{ - struct adapter *adapter = ifp->if_softc; - struct igb_queue *que; - struct tx_ring *txr; - int i, err = 0; -#ifdef RSS - uint32_t bucket_id; -#endif - - /* Which queue to use */ - /* - * When doing RSS, map it to the same outbound queue - * as the incoming flow would be mapped to. - * - * If everything is setup correctly, it should be the - * same bucket that the current CPU we're on is. - */ - if (M_HASHTYPE_GET(m) != M_HASHTYPE_NONE) { -#ifdef RSS - if (rss_hash2bucket(m->m_pkthdr.flowid, - M_HASHTYPE_GET(m), &bucket_id) == 0) { - /* XXX TODO: spit out something if bucket_id > num_queues? */ - i = bucket_id % adapter->num_queues; - } else { -#endif - i = m->m_pkthdr.flowid % adapter->num_queues; -#ifdef RSS - } -#endif - } else { - i = curcpu % adapter->num_queues; - } - txr = &adapter->tx_rings[i]; - que = &adapter->queues[i]; - - err = drbr_enqueue(ifp, txr->br, m); - if (err) - return (err); - if (IGB_TX_TRYLOCK(txr)) { - igb_mq_start_locked(ifp, txr); - IGB_TX_UNLOCK(txr); - } else - taskqueue_enqueue(que->tq, &txr->txq_task); - - return (0); -} - -static int -igb_mq_start_locked(struct ifnet *ifp, struct tx_ring *txr) -{ - struct adapter *adapter = txr->adapter; - struct mbuf *next; - int err = 0, enq = 0; - - IGB_TX_LOCK_ASSERT(txr); - - if (((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) || - adapter->link_active == 0) - return (ENETDOWN); - - /* Process the queue */ - while ((next = drbr_peek(ifp, txr->br)) != NULL) { - if ((err = igb_xmit(txr, &next)) != 0) { - if (next == NULL) { - /* It was freed, move forward */ - drbr_advance(ifp, txr->br); - } else { - /* - * Still have one left, it may not be - * the same since the transmit function - * may have changed it. - */ - drbr_putback(ifp, txr->br, next); - } - break; - } - drbr_advance(ifp, txr->br); - enq++; - if (next->m_flags & M_MCAST && adapter->vf_ifp) - if_inc_counter(ifp, IFCOUNTER_OMCASTS, 1); - ETHER_BPF_MTAP(ifp, next); - if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) - break; - } - if (enq > 0) { - /* Set the watchdog */ - txr->queue_status |= IGB_QUEUE_WORKING; - txr->watchdog_time = ticks; - } - if (txr->tx_avail <= IGB_TX_CLEANUP_THRESHOLD) - igb_txeof(txr); - if (txr->tx_avail <= IGB_MAX_SCATTER) - txr->queue_status |= IGB_QUEUE_DEPLETED; - return (err); -} - -/* - * Called from a taskqueue to drain queued transmit packets. - */ -static void -igb_deferred_mq_start(void *arg, int pending) -{ - struct tx_ring *txr = arg; - struct adapter *adapter = txr->adapter; - struct ifnet *ifp = adapter->ifp; - - IGB_TX_LOCK(txr); - if (!drbr_empty(ifp, txr->br)) - igb_mq_start_locked(ifp, txr); - IGB_TX_UNLOCK(txr); -} - -/* -** Flush all ring buffers -*/ -static void -igb_qflush(struct ifnet *ifp) -{ - struct adapter *adapter = ifp->if_softc; - struct tx_ring *txr = adapter->tx_rings; - struct mbuf *m; - - for (int i = 0; i < adapter->num_queues; i++, txr++) { - IGB_TX_LOCK(txr); - while ((m = buf_ring_dequeue_sc(txr->br)) != NULL) - m_freem(m); - IGB_TX_UNLOCK(txr); - } - if_qflush(ifp); -} -#endif /* ~IGB_LEGACY_TX */ - -/********************************************************************* - * Ioctl entry point - * - * igb_ioctl is called when the user wants to configure the - * interface. - * - * return 0 on success, positive on failure - **********************************************************************/ - -static int -igb_ioctl(struct ifnet *ifp, u_long command, caddr_t data) -{ - struct adapter *adapter = ifp->if_softc; - struct ifreq *ifr = (struct ifreq *)data; -#if defined(INET) || defined(INET6) - struct ifaddr *ifa = (struct ifaddr *)data; -#endif - bool avoid_reset = FALSE; - int error = 0; - - if (adapter->in_detach) - return (error); - - switch (command) { - case SIOCSIFADDR: -#ifdef INET - if (ifa->ifa_addr->sa_family == AF_INET) - avoid_reset = TRUE; -#endif -#ifdef INET6 - if (ifa->ifa_addr->sa_family == AF_INET6) - avoid_reset = TRUE; -#endif - /* - ** Calling init results in link renegotiation, - ** so we avoid doing it when possible. - */ - if (avoid_reset) { - ifp->if_flags |= IFF_UP; - if (!(ifp->if_drv_flags & IFF_DRV_RUNNING)) - igb_init(adapter); -#ifdef INET - if (!(ifp->if_flags & IFF_NOARP)) - arp_ifinit(ifp, ifa); -#endif - } else - error = ether_ioctl(ifp, command, data); - break; - case SIOCSIFMTU: - { - int max_frame_size; - - IOCTL_DEBUGOUT("ioctl rcv'd: SIOCSIFMTU (Set Interface MTU)"); - - IGB_CORE_LOCK(adapter); - max_frame_size = 9234; - if (ifr->ifr_mtu > max_frame_size - ETHER_HDR_LEN - - ETHER_CRC_LEN) { - IGB_CORE_UNLOCK(adapter); - error = EINVAL; - break; - } - - ifp->if_mtu = ifr->ifr_mtu; - adapter->max_frame_size = - ifp->if_mtu + ETHER_HDR_LEN + ETHER_CRC_LEN; - if (ifp->if_drv_flags & IFF_DRV_RUNNING) - igb_init_locked(adapter); - IGB_CORE_UNLOCK(adapter); - break; - } - case SIOCSIFFLAGS: - IOCTL_DEBUGOUT("ioctl rcv'd:\ - SIOCSIFFLAGS (Set Interface Flags)"); - IGB_CORE_LOCK(adapter); - if (ifp->if_flags & IFF_UP) { - if ((ifp->if_drv_flags & IFF_DRV_RUNNING)) { - if ((ifp->if_flags ^ adapter->if_flags) & - (IFF_PROMISC | IFF_ALLMULTI)) { - igb_disable_promisc(adapter); - igb_set_promisc(adapter); - } - } else - igb_init_locked(adapter); - } else - if (ifp->if_drv_flags & IFF_DRV_RUNNING) - igb_stop(adapter); - adapter->if_flags = ifp->if_flags; - IGB_CORE_UNLOCK(adapter); - break; - case SIOCADDMULTI: - case SIOCDELMULTI: - IOCTL_DEBUGOUT("ioctl rcv'd: SIOC(ADD|DEL)MULTI"); - if (ifp->if_drv_flags & IFF_DRV_RUNNING) { - IGB_CORE_LOCK(adapter); - igb_disable_intr(adapter); - igb_set_multi(adapter); -#ifdef DEVICE_POLLING - if (!(ifp->if_capenable & IFCAP_POLLING)) -#endif - igb_enable_intr(adapter); - IGB_CORE_UNLOCK(adapter); - } - break; - case SIOCSIFMEDIA: - /* Check SOL/IDER usage */ - IGB_CORE_LOCK(adapter); - if (e1000_check_reset_block(&adapter->hw)) { - IGB_CORE_UNLOCK(adapter); - device_printf(adapter->dev, "Media change is" - " blocked due to SOL/IDER session.\n"); - break; - } - IGB_CORE_UNLOCK(adapter); - case SIOCGIFMEDIA: - IOCTL_DEBUGOUT("ioctl rcv'd: \ - SIOCxIFMEDIA (Get/Set Interface Media)"); - error = ifmedia_ioctl(ifp, ifr, &adapter->media, command); - break; - case SIOCSIFCAP: - { - int mask, reinit; - - IOCTL_DEBUGOUT("ioctl rcv'd: SIOCSIFCAP (Set Capabilities)"); - reinit = 0; - mask = ifr->ifr_reqcap ^ ifp->if_capenable; -#ifdef DEVICE_POLLING - if (mask & IFCAP_POLLING) { - if (ifr->ifr_reqcap & IFCAP_POLLING) { - error = ether_poll_register(igb_poll, ifp); - if (error) - return (error); - IGB_CORE_LOCK(adapter); - igb_disable_intr(adapter); - ifp->if_capenable |= IFCAP_POLLING; - IGB_CORE_UNLOCK(adapter); - } else { - error = ether_poll_deregister(ifp); - /* Enable interrupt even in error case */ - IGB_CORE_LOCK(adapter); - igb_enable_intr(adapter); - ifp->if_capenable &= ~IFCAP_POLLING; - IGB_CORE_UNLOCK(adapter); - } - } -#endif -#if __FreeBSD_version >= 1000000 - /* HW cannot turn these on/off separately */ - if (mask & (IFCAP_RXCSUM | IFCAP_RXCSUM_IPV6)) { - ifp->if_capenable ^= IFCAP_RXCSUM; - ifp->if_capenable ^= IFCAP_RXCSUM_IPV6; - reinit = 1; - } - if (mask & IFCAP_TXCSUM) { - ifp->if_capenable ^= IFCAP_TXCSUM; - reinit = 1; - } - if (mask & IFCAP_TXCSUM_IPV6) { - ifp->if_capenable ^= IFCAP_TXCSUM_IPV6; - reinit = 1; - } -#else - if (mask & IFCAP_HWCSUM) { - ifp->if_capenable ^= IFCAP_HWCSUM; - reinit = 1; - } -#endif - if (mask & IFCAP_TSO4) { - ifp->if_capenable ^= IFCAP_TSO4; - reinit = 1; - } - if (mask & IFCAP_TSO6) { - ifp->if_capenable ^= IFCAP_TSO6; - reinit = 1; - } - if (mask & IFCAP_VLAN_HWTAGGING) { - ifp->if_capenable ^= IFCAP_VLAN_HWTAGGING; - reinit = 1; - } - if (mask & IFCAP_VLAN_HWFILTER) { - ifp->if_capenable ^= IFCAP_VLAN_HWFILTER; - reinit = 1; - } - if (mask & IFCAP_VLAN_HWTSO) { - ifp->if_capenable ^= IFCAP_VLAN_HWTSO; - reinit = 1; - } - if (mask & IFCAP_LRO) { - ifp->if_capenable ^= IFCAP_LRO; - reinit = 1; - } - if (reinit && (ifp->if_drv_flags & IFF_DRV_RUNNING)) - igb_init(adapter); - VLAN_CAPABILITIES(ifp); - break; - } - - default: - error = ether_ioctl(ifp, command, data); - break; - } - - return (error); -} - - -/********************************************************************* - * Init entry point - * - * This routine is used in two ways. It is used by the stack as - * init entry point in network interface structure. It is also used - * by the driver as a hw/sw initialization routine to get to a - * consistent state. - * - * return 0 on success, positive on failure - **********************************************************************/ - -static void -igb_init_locked(struct adapter *adapter) -{ - struct ifnet *ifp = adapter->ifp; - device_t dev = adapter->dev; - - INIT_DEBUGOUT("igb_init: begin"); - - IGB_CORE_LOCK_ASSERT(adapter); - - igb_disable_intr(adapter); - callout_stop(&adapter->timer); - - /* Get the latest mac address, User can use a LAA */ - bcopy(IF_LLADDR(adapter->ifp), adapter->hw.mac.addr, - ETHER_ADDR_LEN); - - /* Put the address into the Receive Address Array */ - e1000_rar_set(&adapter->hw, adapter->hw.mac.addr, 0); - - igb_reset(adapter); - igb_update_link_status(adapter); - - E1000_WRITE_REG(&adapter->hw, E1000_VET, ETHERTYPE_VLAN); - - /* Set hardware offload abilities */ - ifp->if_hwassist = 0; - if (ifp->if_capenable & IFCAP_TXCSUM) { -#if __FreeBSD_version >= 1000000 - ifp->if_hwassist |= (CSUM_IP_TCP | CSUM_IP_UDP); - if (adapter->hw.mac.type != e1000_82575) - ifp->if_hwassist |= CSUM_IP_SCTP; -#else - ifp->if_hwassist |= (CSUM_TCP | CSUM_UDP); -#if __FreeBSD_version >= 800000 - if (adapter->hw.mac.type != e1000_82575) - ifp->if_hwassist |= CSUM_SCTP; -#endif -#endif - } - -#if __FreeBSD_version >= 1000000 - if (ifp->if_capenable & IFCAP_TXCSUM_IPV6) { - ifp->if_hwassist |= (CSUM_IP6_TCP | CSUM_IP6_UDP); - if (adapter->hw.mac.type != e1000_82575) - ifp->if_hwassist |= CSUM_IP6_SCTP; - } -#endif - if (ifp->if_capenable & IFCAP_TSO) - ifp->if_hwassist |= CSUM_TSO; - - /* Clear bad data from Rx FIFOs */ - e1000_rx_fifo_flush_82575(&adapter->hw); - - /* Configure for OS presence */ - igb_init_manageability(adapter); - - /* Prepare transmit descriptors and buffers */ - igb_setup_transmit_structures(adapter); - igb_initialize_transmit_units(adapter); - - /* Setup Multicast table */ - igb_set_multi(adapter); - - /* - ** Figure out the desired mbuf pool - ** for doing jumbo/packetsplit - */ - if (adapter->max_frame_size <= 2048) - adapter->rx_mbuf_sz = MCLBYTES; - else if (adapter->max_frame_size <= 4096) - adapter->rx_mbuf_sz = MJUMPAGESIZE; - else - adapter->rx_mbuf_sz = MJUM9BYTES; - - /* Prepare receive descriptors and buffers */ - if (igb_setup_receive_structures(adapter)) { - device_printf(dev, "Could not setup receive structures\n"); - return; - } - igb_initialize_receive_units(adapter); - - /* Enable VLAN support */ - if (ifp->if_capenable & IFCAP_VLAN_HWTAGGING) - igb_setup_vlan_hw_support(adapter); - - /* Don't lose promiscuous settings */ - igb_set_promisc(adapter); - - ifp->if_drv_flags |= IFF_DRV_RUNNING; - ifp->if_drv_flags &= ~IFF_DRV_OACTIVE; - - callout_reset(&adapter->timer, hz, igb_local_timer, adapter); - e1000_clear_hw_cntrs_base_generic(&adapter->hw); - - if (adapter->msix > 1) /* Set up queue routing */ - igb_configure_queues(adapter); - - /* this clears any pending interrupts */ - E1000_READ_REG(&adapter->hw, E1000_ICR); -#ifdef DEVICE_POLLING - /* - * Only enable interrupts if we are not polling, make sure - * they are off otherwise. - */ - if (ifp->if_capenable & IFCAP_POLLING) - igb_disable_intr(adapter); - else -#endif /* DEVICE_POLLING */ - { - igb_enable_intr(adapter); - E1000_WRITE_REG(&adapter->hw, E1000_ICS, E1000_ICS_LSC); - } - - /* Set Energy Efficient Ethernet */ - if (adapter->hw.phy.media_type == e1000_media_type_copper) { - if (adapter->hw.mac.type == e1000_i354) - e1000_set_eee_i354(&adapter->hw, TRUE, TRUE); - else - e1000_set_eee_i350(&adapter->hw, TRUE, TRUE); - } -} - -static void -igb_init(void *arg) -{ - struct adapter *adapter = arg; - - IGB_CORE_LOCK(adapter); - igb_init_locked(adapter); - IGB_CORE_UNLOCK(adapter); -} - - -static void -igb_handle_que(void *context, int pending) -{ - struct igb_queue *que = context; - struct adapter *adapter = que->adapter; - struct tx_ring *txr = que->txr; - struct ifnet *ifp = adapter->ifp; - - if (ifp->if_drv_flags & IFF_DRV_RUNNING) { - bool more; - - more = igb_rxeof(que, adapter->rx_process_limit, NULL); - - IGB_TX_LOCK(txr); - igb_txeof(txr); -#ifndef IGB_LEGACY_TX - /* Process the stack queue only if not depleted */ - if (((txr->queue_status & IGB_QUEUE_DEPLETED) == 0) && - !drbr_empty(ifp, txr->br)) - igb_mq_start_locked(ifp, txr); -#else - if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) - igb_start_locked(txr, ifp); -#endif - IGB_TX_UNLOCK(txr); - /* Do we need another? */ - if (more) { - taskqueue_enqueue(que->tq, &que->que_task); - return; - } - } - -#ifdef DEVICE_POLLING - if (ifp->if_capenable & IFCAP_POLLING) - return; -#endif - /* Reenable this interrupt */ - if (que->eims) - E1000_WRITE_REG(&adapter->hw, E1000_EIMS, que->eims); - else - igb_enable_intr(adapter); -} - -/* Deal with link in a sleepable context */ -static void -igb_handle_link(void *context, int pending) -{ - struct adapter *adapter = context; - - IGB_CORE_LOCK(adapter); - igb_handle_link_locked(adapter); - IGB_CORE_UNLOCK(adapter); -} - -static void -igb_handle_link_locked(struct adapter *adapter) -{ - struct tx_ring *txr = adapter->tx_rings; - struct ifnet *ifp = adapter->ifp; - - IGB_CORE_LOCK_ASSERT(adapter); - adapter->hw.mac.get_link_status = 1; - igb_update_link_status(adapter); - if ((ifp->if_drv_flags & IFF_DRV_RUNNING) && adapter->link_active) { - for (int i = 0; i < adapter->num_queues; i++, txr++) { - IGB_TX_LOCK(txr); -#ifndef IGB_LEGACY_TX - /* Process the stack queue only if not depleted */ - if (((txr->queue_status & IGB_QUEUE_DEPLETED) == 0) && - !drbr_empty(ifp, txr->br)) - igb_mq_start_locked(ifp, txr); -#else - if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) - igb_start_locked(txr, ifp); -#endif - IGB_TX_UNLOCK(txr); - } - } -} - -/********************************************************************* - * - * MSI/Legacy Deferred - * Interrupt Service routine - * - *********************************************************************/ -static int -igb_irq_fast(void *arg) -{ - struct adapter *adapter = arg; - struct igb_queue *que = adapter->queues; - u32 reg_icr; - - - reg_icr = E1000_READ_REG(&adapter->hw, E1000_ICR); - - /* Hot eject? */ - if (reg_icr == 0xffffffff) - return FILTER_STRAY; - - /* Definitely not our interrupt. */ - if (reg_icr == 0x0) - return FILTER_STRAY; - - if ((reg_icr & E1000_ICR_INT_ASSERTED) == 0) - return FILTER_STRAY; - - /* - * Mask interrupts until the taskqueue is finished running. This is - * cheap, just assume that it is needed. This also works around the - * MSI message reordering errata on certain systems. - */ - igb_disable_intr(adapter); - taskqueue_enqueue(que->tq, &que->que_task); - - /* Link status change */ - if (reg_icr & (E1000_ICR_RXSEQ | E1000_ICR_LSC)) - taskqueue_enqueue(que->tq, &adapter->link_task); - - if (reg_icr & E1000_ICR_RXO) - adapter->rx_overruns++; - return FILTER_HANDLED; -} - -#ifdef DEVICE_POLLING -#if __FreeBSD_version >= 800000 -#define POLL_RETURN_COUNT(a) (a) -static int -#else -#define POLL_RETURN_COUNT(a) -static void -#endif -igb_poll(struct ifnet *ifp, enum poll_cmd cmd, int count) -{ - struct adapter *adapter = ifp->if_softc; - struct igb_queue *que; - struct tx_ring *txr; - u32 reg_icr, rx_done = 0; - u32 loop = IGB_MAX_LOOP; - bool more; - - IGB_CORE_LOCK(adapter); - if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { - IGB_CORE_UNLOCK(adapter); - return POLL_RETURN_COUNT(rx_done); - } - - if (cmd == POLL_AND_CHECK_STATUS) { - reg_icr = E1000_READ_REG(&adapter->hw, E1000_ICR); - /* Link status change */ - if (reg_icr & (E1000_ICR_RXSEQ | E1000_ICR_LSC)) - igb_handle_link_locked(adapter); - - if (reg_icr & E1000_ICR_RXO) - adapter->rx_overruns++; - } - IGB_CORE_UNLOCK(adapter); - - for (int i = 0; i < adapter->num_queues; i++) { - que = &adapter->queues[i]; - txr = que->txr; - - igb_rxeof(que, count, &rx_done); - - IGB_TX_LOCK(txr); - do { - more = igb_txeof(txr); - } while (loop-- && more); -#ifndef IGB_LEGACY_TX - if (!drbr_empty(ifp, txr->br)) - igb_mq_start_locked(ifp, txr); -#else - if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) - igb_start_locked(txr, ifp); -#endif - IGB_TX_UNLOCK(txr); - } - - return POLL_RETURN_COUNT(rx_done); -} -#endif /* DEVICE_POLLING */ - -/********************************************************************* - * - * MSIX Que Interrupt Service routine - * - **********************************************************************/ -static void -igb_msix_que(void *arg) -{ - struct igb_queue *que = arg; - struct adapter *adapter = que->adapter; - struct ifnet *ifp = adapter->ifp; - struct tx_ring *txr = que->txr; - struct rx_ring *rxr = que->rxr; - u32 newitr = 0; - bool more_rx; - - /* Ignore spurious interrupts */ - if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) - return; - - E1000_WRITE_REG(&adapter->hw, E1000_EIMC, que->eims); - ++que->irqs; - - IGB_TX_LOCK(txr); - igb_txeof(txr); -#ifndef IGB_LEGACY_TX - /* Process the stack queue only if not depleted */ - if (((txr->queue_status & IGB_QUEUE_DEPLETED) == 0) && - !drbr_empty(ifp, txr->br)) - igb_mq_start_locked(ifp, txr); -#else - if (!IFQ_DRV_IS_EMPTY(&ifp->if_snd)) - igb_start_locked(txr, ifp); -#endif - IGB_TX_UNLOCK(txr); - - more_rx = igb_rxeof(que, adapter->rx_process_limit, NULL); - - if (adapter->enable_aim == FALSE) - goto no_calc; - /* - ** Do Adaptive Interrupt Moderation: - ** - Write out last calculated setting - ** - Calculate based on average size over - ** the last interval. - */ - if (que->eitr_setting) - E1000_WRITE_REG(&adapter->hw, - E1000_EITR(que->msix), que->eitr_setting); - - que->eitr_setting = 0; - - /* Idle, do nothing */ - if ((txr->bytes == 0) && (rxr->bytes == 0)) - goto no_calc; - - /* Used half Default if sub-gig */ - if (adapter->link_speed != 1000) - newitr = IGB_DEFAULT_ITR / 2; - else { - if ((txr->bytes) && (txr->packets)) - newitr = txr->bytes/txr->packets; - if ((rxr->bytes) && (rxr->packets)) - newitr = max(newitr, - (rxr->bytes / rxr->packets)); - newitr += 24; /* account for hardware frame, crc */ - /* set an upper boundary */ - newitr = min(newitr, 3000); - /* Be nice to the mid range */ - if ((newitr > 300) && (newitr < 1200)) - newitr = (newitr / 3); - else - newitr = (newitr / 2); - } - newitr &= 0x7FFC; /* Mask invalid bits */ - if (adapter->hw.mac.type == e1000_82575) - newitr |= newitr << 16; - else - newitr |= E1000_EITR_CNT_IGNR; - - /* save for next interrupt */ - que->eitr_setting = newitr; - - /* Reset state */ - txr->bytes = 0; - txr->packets = 0; - rxr->bytes = 0; - rxr->packets = 0; - -no_calc: - /* Schedule a clean task if needed*/ - if (more_rx) - taskqueue_enqueue(que->tq, &que->que_task); - else - /* Reenable this interrupt */ - E1000_WRITE_REG(&adapter->hw, E1000_EIMS, que->eims); - return; -} - - -/********************************************************************* - * - * MSIX Link Interrupt Service routine - * - **********************************************************************/ - -static void -igb_msix_link(void *arg) -{ - struct adapter *adapter = arg; - u32 icr; - - ++adapter->link_irq; - icr = E1000_READ_REG(&adapter->hw, E1000_ICR); - if (!(icr & E1000_ICR_LSC)) - goto spurious; - igb_handle_link(adapter, 0); - -spurious: - /* Rearm */ - E1000_WRITE_REG(&adapter->hw, E1000_IMS, E1000_IMS_LSC); - E1000_WRITE_REG(&adapter->hw, E1000_EIMS, adapter->link_mask); - return; -} - - -/********************************************************************* - * - * Media Ioctl callback - * - * This routine is called whenever the user queries the status of - * the interface using ifconfig. - * - **********************************************************************/ -static void -igb_media_status(struct ifnet *ifp, struct ifmediareq *ifmr) -{ - struct adapter *adapter = ifp->if_softc; - - INIT_DEBUGOUT("igb_media_status: begin"); - - IGB_CORE_LOCK(adapter); - igb_update_link_status(adapter); - - ifmr->ifm_status = IFM_AVALID; - ifmr->ifm_active = IFM_ETHER; - - if (!adapter->link_active) { - IGB_CORE_UNLOCK(adapter); - return; - } - - ifmr->ifm_status |= IFM_ACTIVE; - - switch (adapter->link_speed) { - case 10: - ifmr->ifm_active |= IFM_10_T; - break; - case 100: - /* - ** Support for 100Mb SFP - these are Fiber - ** but the media type appears as serdes - */ - if (adapter->hw.phy.media_type == - e1000_media_type_internal_serdes) - ifmr->ifm_active |= IFM_100_FX; - else - ifmr->ifm_active |= IFM_100_TX; - break; - case 1000: - ifmr->ifm_active |= IFM_1000_T; - break; - case 2500: - ifmr->ifm_active |= IFM_2500_SX; - break; - } - - if (adapter->link_duplex == FULL_DUPLEX) - ifmr->ifm_active |= IFM_FDX; - else - ifmr->ifm_active |= IFM_HDX; - - IGB_CORE_UNLOCK(adapter); -} - -/********************************************************************* - * - * Media Ioctl callback - * - * This routine is called when the user changes speed/duplex using - * media/mediopt option with ifconfig. - * - **********************************************************************/ -static int -igb_media_change(struct ifnet *ifp) -{ - struct adapter *adapter = ifp->if_softc; - struct ifmedia *ifm = &adapter->media; - - INIT_DEBUGOUT("igb_media_change: begin"); - - if (IFM_TYPE(ifm->ifm_media) != IFM_ETHER) - return (EINVAL); - - IGB_CORE_LOCK(adapter); - switch (IFM_SUBTYPE(ifm->ifm_media)) { - case IFM_AUTO: - adapter->hw.mac.autoneg = DO_AUTO_NEG; - adapter->hw.phy.autoneg_advertised = AUTONEG_ADV_DEFAULT; - break; - case IFM_1000_LX: - case IFM_1000_SX: - case IFM_1000_T: - adapter->hw.mac.autoneg = DO_AUTO_NEG; - adapter->hw.phy.autoneg_advertised = ADVERTISE_1000_FULL; - break; - case IFM_100_TX: - adapter->hw.mac.autoneg = FALSE; - adapter->hw.phy.autoneg_advertised = 0; - if ((ifm->ifm_media & IFM_GMASK) == IFM_FDX) - adapter->hw.mac.forced_speed_duplex = ADVERTISE_100_FULL; - else - adapter->hw.mac.forced_speed_duplex = ADVERTISE_100_HALF; - break; - case IFM_10_T: - adapter->hw.mac.autoneg = FALSE; - adapter->hw.phy.autoneg_advertised = 0; - if ((ifm->ifm_media & IFM_GMASK) == IFM_FDX) - adapter->hw.mac.forced_speed_duplex = ADVERTISE_10_FULL; - else - adapter->hw.mac.forced_speed_duplex = ADVERTISE_10_HALF; - break; - default: - device_printf(adapter->dev, "Unsupported media type\n"); - } - - igb_init_locked(adapter); - IGB_CORE_UNLOCK(adapter); - - return (0); -} - - -/********************************************************************* - * - * This routine maps the mbufs to Advanced TX descriptors. - * - **********************************************************************/ -static int -igb_xmit(struct tx_ring *txr, struct mbuf **m_headp) -{ - struct adapter *adapter = txr->adapter; - u32 olinfo_status = 0, cmd_type_len; - int i, j, error, nsegs; - int first; - bool remap = TRUE; - struct mbuf *m_head; - bus_dma_segment_t segs[IGB_MAX_SCATTER]; - bus_dmamap_t map; - struct igb_tx_buf *txbuf; - union e1000_adv_tx_desc *txd = NULL; - - m_head = *m_headp; - - /* Basic descriptor defines */ - cmd_type_len = (E1000_ADVTXD_DTYP_DATA | - E1000_ADVTXD_DCMD_IFCS | E1000_ADVTXD_DCMD_DEXT); - - if (m_head->m_flags & M_VLANTAG) - cmd_type_len |= E1000_ADVTXD_DCMD_VLE; - - /* - * Important to capture the first descriptor - * used because it will contain the index of - * the one we tell the hardware to report back - */ - first = txr->next_avail_desc; - txbuf = &txr->tx_buffers[first]; - map = txbuf->map; - - /* - * Map the packet for DMA. - */ -retry: - error = bus_dmamap_load_mbuf_sg(txr->txtag, map, - *m_headp, segs, &nsegs, BUS_DMA_NOWAIT); - - if (__predict_false(error)) { - struct mbuf *m; - - switch (error) { - case EFBIG: - /* Try it again? - one try */ - if (remap == TRUE) { - remap = FALSE; - m = m_collapse(*m_headp, M_NOWAIT, - IGB_MAX_SCATTER); - if (m == NULL) { - adapter->mbuf_defrag_failed++; - m_freem(*m_headp); - *m_headp = NULL; - return (ENOBUFS); - } - *m_headp = m; - goto retry; - } else - return (error); - default: - txr->no_tx_dma_setup++; - m_freem(*m_headp); - *m_headp = NULL; - return (error); - } - } - - /* Make certain there are enough descriptors */ - if (txr->tx_avail < (nsegs + 2)) { - txr->no_desc_avail++; - bus_dmamap_unload(txr->txtag, map); - return (ENOBUFS); - } - m_head = *m_headp; - - /* - ** Set up the appropriate offload context - ** this will consume the first descriptor - */ - error = igb_tx_ctx_setup(txr, m_head, &cmd_type_len, &olinfo_status); - if (__predict_false(error)) { - m_freem(*m_headp); - *m_headp = NULL; - return (error); - } - - /* 82575 needs the queue index added */ - if (adapter->hw.mac.type == e1000_82575) - olinfo_status |= txr->me << 4; - - i = txr->next_avail_desc; - for (j = 0; j < nsegs; j++) { - bus_size_t seglen; - bus_addr_t segaddr; - - txbuf = &txr->tx_buffers[i]; - txd = &txr->tx_base[i]; - seglen = segs[j].ds_len; - segaddr = htole64(segs[j].ds_addr); - - txd->read.buffer_addr = segaddr; - txd->read.cmd_type_len = htole32(E1000_TXD_CMD_IFCS | - cmd_type_len | seglen); - txd->read.olinfo_status = htole32(olinfo_status); - - if (++i == txr->num_desc) - i = 0; - } - - txd->read.cmd_type_len |= - htole32(E1000_TXD_CMD_EOP | E1000_TXD_CMD_RS); - txr->tx_avail -= nsegs; - txr->next_avail_desc = i; - - txbuf->m_head = m_head; - /* - ** Here we swap the map so the last descriptor, - ** which gets the completion interrupt has the - ** real map, and the first descriptor gets the - ** unused map from this descriptor. - */ - txr->tx_buffers[first].map = txbuf->map; - txbuf->map = map; - bus_dmamap_sync(txr->txtag, map, BUS_DMASYNC_PREWRITE); - - /* Set the EOP descriptor that will be marked done */ - txbuf = &txr->tx_buffers[first]; - txbuf->eop = txd; - - bus_dmamap_sync(txr->txdma.dma_tag, txr->txdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - /* - * Advance the Transmit Descriptor Tail (Tdt), this tells the - * hardware that this frame is available to transmit. - */ - ++txr->total_packets; - E1000_WRITE_REG(&adapter->hw, E1000_TDT(txr->me), i); - - return (0); -} -static void -igb_set_promisc(struct adapter *adapter) -{ - struct ifnet *ifp = adapter->ifp; - struct e1000_hw *hw = &adapter->hw; - u32 reg; - - if (adapter->vf_ifp) { - e1000_promisc_set_vf(hw, e1000_promisc_enabled); - return; - } - - reg = E1000_READ_REG(hw, E1000_RCTL); - if (ifp->if_flags & IFF_PROMISC) { - reg |= (E1000_RCTL_UPE | E1000_RCTL_MPE); - E1000_WRITE_REG(hw, E1000_RCTL, reg); - } else if (ifp->if_flags & IFF_ALLMULTI) { - reg |= E1000_RCTL_MPE; - reg &= ~E1000_RCTL_UPE; - E1000_WRITE_REG(hw, E1000_RCTL, reg); - } -} - -static void -igb_disable_promisc(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - struct ifnet *ifp = adapter->ifp; - u32 reg; - int mcnt = 0; - - if (adapter->vf_ifp) { - e1000_promisc_set_vf(hw, e1000_promisc_disabled); - return; - } - reg = E1000_READ_REG(hw, E1000_RCTL); - reg &= (~E1000_RCTL_UPE); - if (ifp->if_flags & IFF_ALLMULTI) - mcnt = MAX_NUM_MULTICAST_ADDRESSES; - else { - struct ifmultiaddr *ifma; -#if __FreeBSD_version < 800000 - IF_ADDR_LOCK(ifp); -#else - if_maddr_rlock(ifp); -#endif - TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) { - if (ifma->ifma_addr->sa_family != AF_LINK) - continue; - if (mcnt == MAX_NUM_MULTICAST_ADDRESSES) - break; - mcnt++; - } -#if __FreeBSD_version < 800000 - IF_ADDR_UNLOCK(ifp); -#else - if_maddr_runlock(ifp); -#endif - } - /* Don't disable if in MAX groups */ - if (mcnt < MAX_NUM_MULTICAST_ADDRESSES) - reg &= (~E1000_RCTL_MPE); - E1000_WRITE_REG(hw, E1000_RCTL, reg); -} - - -/********************************************************************* - * Multicast Update - * - * This routine is called whenever multicast address list is updated. - * - **********************************************************************/ - -static void -igb_set_multi(struct adapter *adapter) -{ - struct ifnet *ifp = adapter->ifp; - struct ifmultiaddr *ifma; - u32 reg_rctl = 0; - u8 *mta; - - int mcnt = 0; - - IOCTL_DEBUGOUT("igb_set_multi: begin"); - - mta = adapter->mta; - bzero(mta, sizeof(uint8_t) * ETH_ADDR_LEN * - MAX_NUM_MULTICAST_ADDRESSES); - -#if __FreeBSD_version < 800000 - IF_ADDR_LOCK(ifp); -#else - if_maddr_rlock(ifp); -#endif - TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) { - if (ifma->ifma_addr->sa_family != AF_LINK) - continue; - - if (mcnt == MAX_NUM_MULTICAST_ADDRESSES) - break; - - bcopy(LLADDR((struct sockaddr_dl *)ifma->ifma_addr), - &mta[mcnt * ETH_ADDR_LEN], ETH_ADDR_LEN); - mcnt++; - } -#if __FreeBSD_version < 800000 - IF_ADDR_UNLOCK(ifp); -#else - if_maddr_runlock(ifp); -#endif - - if (mcnt >= MAX_NUM_MULTICAST_ADDRESSES) { - reg_rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); - reg_rctl |= E1000_RCTL_MPE; - E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); - } else - e1000_update_mc_addr_list(&adapter->hw, mta, mcnt); -} - - -/********************************************************************* - * Timer routine: - * This routine checks for link status, - * updates statistics, and does the watchdog. - * - **********************************************************************/ - -static void -igb_local_timer(void *arg) -{ - struct adapter *adapter = arg; - device_t dev = adapter->dev; - struct ifnet *ifp = adapter->ifp; - struct tx_ring *txr = adapter->tx_rings; - struct igb_queue *que = adapter->queues; - int hung = 0, busy = 0; - - - IGB_CORE_LOCK_ASSERT(adapter); - - igb_update_link_status(adapter); - igb_update_stats_counters(adapter); - - /* - ** Check the TX queues status - ** - central locked handling of OACTIVE - ** - watchdog only if all queues show hung - */ - for (int i = 0; i < adapter->num_queues; i++, que++, txr++) { - if ((txr->queue_status & IGB_QUEUE_HUNG) && - (adapter->pause_frames == 0)) - ++hung; - if (txr->queue_status & IGB_QUEUE_DEPLETED) - ++busy; - if ((txr->queue_status & IGB_QUEUE_IDLE) == 0) - taskqueue_enqueue(que->tq, &que->que_task); - } - if (hung == adapter->num_queues) - goto timeout; - if (busy == adapter->num_queues) - ifp->if_drv_flags |= IFF_DRV_OACTIVE; - else if ((ifp->if_drv_flags & IFF_DRV_OACTIVE) && - (busy < adapter->num_queues)) - ifp->if_drv_flags &= ~IFF_DRV_OACTIVE; - - adapter->pause_frames = 0; - callout_reset(&adapter->timer, hz, igb_local_timer, adapter); -#ifndef DEVICE_POLLING - /* Schedule all queue interrupts - deadlock protection */ - E1000_WRITE_REG(&adapter->hw, E1000_EICS, adapter->que_mask); -#endif - return; - -timeout: - device_printf(adapter->dev, "Watchdog timeout -- resetting\n"); - device_printf(dev,"Queue(%d) tdh = %d, hw tdt = %d\n", txr->me, - E1000_READ_REG(&adapter->hw, E1000_TDH(txr->me)), - E1000_READ_REG(&adapter->hw, E1000_TDT(txr->me))); - device_printf(dev,"TX(%d) desc avail = %d," - "Next TX to Clean = %d\n", - txr->me, txr->tx_avail, txr->next_to_clean); - adapter->ifp->if_drv_flags &= ~IFF_DRV_RUNNING; - adapter->watchdog_events++; - igb_init_locked(adapter); -} - -static void -igb_update_link_status(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - struct e1000_fc_info *fc = &hw->fc; - struct ifnet *ifp = adapter->ifp; - device_t dev = adapter->dev; - struct tx_ring *txr = adapter->tx_rings; - u32 link_check, thstat, ctrl; - char *flowctl = NULL; - - link_check = thstat = ctrl = 0; - - /* Get the cached link value or read for real */ - switch (hw->phy.media_type) { - case e1000_media_type_copper: - if (hw->mac.get_link_status) { - /* Do the work to read phy */ - e1000_check_for_link(hw); - link_check = !hw->mac.get_link_status; - } else - link_check = TRUE; - break; - case e1000_media_type_fiber: - e1000_check_for_link(hw); - link_check = (E1000_READ_REG(hw, E1000_STATUS) & - E1000_STATUS_LU); - break; - case e1000_media_type_internal_serdes: - e1000_check_for_link(hw); - link_check = adapter->hw.mac.serdes_has_link; - break; - /* VF device is type_unknown */ - case e1000_media_type_unknown: - e1000_check_for_link(hw); - link_check = !hw->mac.get_link_status; - /* Fall thru */ - default: - break; - } - - /* Check for thermal downshift or shutdown */ - if (hw->mac.type == e1000_i350) { - thstat = E1000_READ_REG(hw, E1000_THSTAT); - ctrl = E1000_READ_REG(hw, E1000_CTRL_EXT); - } - - /* Get the flow control for display */ - switch (fc->current_mode) { - case e1000_fc_rx_pause: - flowctl = "RX"; - break; - case e1000_fc_tx_pause: - flowctl = "TX"; - break; - case e1000_fc_full: - flowctl = "Full"; - break; - case e1000_fc_none: - default: - flowctl = "None"; - break; - } - - /* Now we check if a transition has happened */ - if (link_check && (adapter->link_active == 0)) { - e1000_get_speed_and_duplex(&adapter->hw, - &adapter->link_speed, &adapter->link_duplex); - if (bootverbose) - device_printf(dev, "Link is up %d Mbps %s," - " Flow Control: %s\n", - adapter->link_speed, - ((adapter->link_duplex == FULL_DUPLEX) ? - "Full Duplex" : "Half Duplex"), flowctl); - adapter->link_active = 1; - ifp->if_baudrate = adapter->link_speed * 1000000; - if ((ctrl & E1000_CTRL_EXT_LINK_MODE_GMII) && - (thstat & E1000_THSTAT_LINK_THROTTLE)) - device_printf(dev, "Link: thermal downshift\n"); - /* Delay Link Up for Phy update */ - if (((hw->mac.type == e1000_i210) || - (hw->mac.type == e1000_i211)) && - (hw->phy.id == I210_I_PHY_ID)) - msec_delay(I210_LINK_DELAY); - /* Reset if the media type changed. */ - if (hw->dev_spec._82575.media_changed) { - hw->dev_spec._82575.media_changed = false; - adapter->flags |= IGB_MEDIA_RESET; - igb_reset(adapter); - } - /* This can sleep */ - if_link_state_change(ifp, LINK_STATE_UP); - } else if (!link_check && (adapter->link_active == 1)) { - ifp->if_baudrate = adapter->link_speed = 0; - adapter->link_duplex = 0; - if (bootverbose) - device_printf(dev, "Link is Down\n"); - if ((ctrl & E1000_CTRL_EXT_LINK_MODE_GMII) && - (thstat & E1000_THSTAT_PWR_DOWN)) - device_printf(dev, "Link: thermal shutdown\n"); - adapter->link_active = 0; - /* This can sleep */ - if_link_state_change(ifp, LINK_STATE_DOWN); - /* Reset queue state */ - for (int i = 0; i < adapter->num_queues; i++, txr++) - txr->queue_status = IGB_QUEUE_IDLE; - } -} - -/********************************************************************* - * - * This routine disables all traffic on the adapter by issuing a - * global reset on the MAC and deallocates TX/RX buffers. - * - **********************************************************************/ - -static void -igb_stop(void *arg) -{ - struct adapter *adapter = arg; - struct ifnet *ifp = adapter->ifp; - struct tx_ring *txr = adapter->tx_rings; - - IGB_CORE_LOCK_ASSERT(adapter); - - INIT_DEBUGOUT("igb_stop: begin"); - - igb_disable_intr(adapter); - - callout_stop(&adapter->timer); - - /* Tell the stack that the interface is no longer active */ - ifp->if_drv_flags &= ~IFF_DRV_RUNNING; - ifp->if_drv_flags |= IFF_DRV_OACTIVE; - - /* Disarm watchdog timer. */ - for (int i = 0; i < adapter->num_queues; i++, txr++) { - IGB_TX_LOCK(txr); - txr->queue_status = IGB_QUEUE_IDLE; - IGB_TX_UNLOCK(txr); - } - - e1000_reset_hw(&adapter->hw); - E1000_WRITE_REG(&adapter->hw, E1000_WUC, 0); - - e1000_led_off(&adapter->hw); - e1000_cleanup_led(&adapter->hw); -} - - -/********************************************************************* - * - * Determine hardware revision. - * - **********************************************************************/ -static void -igb_identify_hardware(struct adapter *adapter) -{ - device_t dev = adapter->dev; - - /* Make sure our PCI config space has the necessary stuff set */ - pci_enable_busmaster(dev); - adapter->hw.bus.pci_cmd_word = pci_read_config(dev, PCIR_COMMAND, 2); - - /* Save off the information about this board */ - adapter->hw.vendor_id = pci_get_vendor(dev); - adapter->hw.device_id = pci_get_device(dev); - adapter->hw.revision_id = pci_read_config(dev, PCIR_REVID, 1); - adapter->hw.subsystem_vendor_id = - pci_read_config(dev, PCIR_SUBVEND_0, 2); - adapter->hw.subsystem_device_id = - pci_read_config(dev, PCIR_SUBDEV_0, 2); - - /* Set MAC type early for PCI setup */ - e1000_set_mac_type(&adapter->hw); - - /* Are we a VF device? */ - if ((adapter->hw.mac.type == e1000_vfadapt) || - (adapter->hw.mac.type == e1000_vfadapt_i350)) - adapter->vf_ifp = 1; - else - adapter->vf_ifp = 0; -} - -static int -igb_allocate_pci_resources(struct adapter *adapter) -{ - device_t dev = adapter->dev; - int rid; - - rid = PCIR_BAR(0); - adapter->pci_mem = bus_alloc_resource_any(dev, SYS_RES_MEMORY, - &rid, RF_ACTIVE); - if (adapter->pci_mem == NULL) { - device_printf(dev, "Unable to allocate bus resource: memory\n"); - return (ENXIO); - } - adapter->osdep.mem_bus_space_tag = - rman_get_bustag(adapter->pci_mem); - adapter->osdep.mem_bus_space_handle = - rman_get_bushandle(adapter->pci_mem); - adapter->hw.hw_addr = (u8 *)&adapter->osdep.mem_bus_space_handle; - - adapter->num_queues = 1; /* Defaults for Legacy or MSI */ - - /* This will setup either MSI/X or MSI */ - adapter->msix = igb_setup_msix(adapter); - adapter->hw.back = &adapter->osdep; - - return (0); -} - -/********************************************************************* - * - * Setup the Legacy or MSI Interrupt handler - * - **********************************************************************/ -static int -igb_allocate_legacy(struct adapter *adapter) -{ - device_t dev = adapter->dev; - struct igb_queue *que = adapter->queues; -#ifndef IGB_LEGACY_TX - struct tx_ring *txr = adapter->tx_rings; -#endif - int error, rid = 0; - - /* Turn off all interrupts */ - E1000_WRITE_REG(&adapter->hw, E1000_IMC, 0xffffffff); - - /* MSI RID is 1 */ - if (adapter->msix == 1) - rid = 1; - - /* We allocate a single interrupt resource */ - adapter->res = bus_alloc_resource_any(dev, - SYS_RES_IRQ, &rid, RF_SHAREABLE | RF_ACTIVE); - if (adapter->res == NULL) { - device_printf(dev, "Unable to allocate bus resource: " - "interrupt\n"); - return (ENXIO); - } - -#ifndef IGB_LEGACY_TX - TASK_INIT(&txr->txq_task, 0, igb_deferred_mq_start, txr); -#endif - - /* - * Try allocating a fast interrupt and the associated deferred - * processing contexts. - */ - TASK_INIT(&que->que_task, 0, igb_handle_que, que); - /* Make tasklet for deferred link handling */ - TASK_INIT(&adapter->link_task, 0, igb_handle_link, adapter); - que->tq = taskqueue_create_fast("igb_taskq", M_NOWAIT, - taskqueue_thread_enqueue, &que->tq); - taskqueue_start_threads(&que->tq, 1, PI_NET, "%s taskq", - device_get_nameunit(adapter->dev)); - if ((error = bus_setup_intr(dev, adapter->res, - INTR_TYPE_NET | INTR_MPSAFE, igb_irq_fast, NULL, - adapter, &adapter->tag)) != 0) { - device_printf(dev, "Failed to register fast interrupt " - "handler: %d\n", error); - taskqueue_free(que->tq); - que->tq = NULL; - return (error); - } - - return (0); -} - - -/********************************************************************* - * - * Setup the MSIX Queue Interrupt handlers: - * - **********************************************************************/ -static int -igb_allocate_msix(struct adapter *adapter) -{ - device_t dev = adapter->dev; - struct igb_queue *que = adapter->queues; - int error, rid, vector = 0; - int cpu_id = 0; -#ifdef RSS - cpuset_t cpu_mask; -#endif - - /* Be sure to start with all interrupts disabled */ - E1000_WRITE_REG(&adapter->hw, E1000_IMC, ~0); - E1000_WRITE_FLUSH(&adapter->hw); - -#ifdef RSS - /* - * If we're doing RSS, the number of queues needs to - * match the number of RSS buckets that are configured. - * - * + If there's more queues than RSS buckets, we'll end - * up with queues that get no traffic. - * - * + If there's more RSS buckets than queues, we'll end - * up having multiple RSS buckets map to the same queue, - * so there'll be some contention. - */ - if (adapter->num_queues != rss_getnumbuckets()) { - device_printf(dev, - "%s: number of queues (%d) != number of RSS buckets (%d)" - "; performance will be impacted.\n", - __func__, - adapter->num_queues, - rss_getnumbuckets()); - } -#endif - - for (int i = 0; i < adapter->num_queues; i++, vector++, que++) { - rid = vector +1; - que->res = bus_alloc_resource_any(dev, - SYS_RES_IRQ, &rid, RF_SHAREABLE | RF_ACTIVE); - if (que->res == NULL) { - device_printf(dev, - "Unable to allocate bus resource: " - "MSIX Queue Interrupt\n"); - return (ENXIO); - } - error = bus_setup_intr(dev, que->res, - INTR_TYPE_NET | INTR_MPSAFE, NULL, - igb_msix_que, que, &que->tag); - if (error) { - que->res = NULL; - device_printf(dev, "Failed to register Queue handler"); - return (error); - } -#if __FreeBSD_version >= 800504 - bus_describe_intr(dev, que->res, que->tag, "que %d", i); -#endif - que->msix = vector; - if (adapter->hw.mac.type == e1000_82575) - que->eims = E1000_EICR_TX_QUEUE0 << i; - else - que->eims = 1 << vector; - -#ifdef RSS - /* - * The queue ID is used as the RSS layer bucket ID. - * We look up the queue ID -> RSS CPU ID and select - * that. - */ - cpu_id = rss_getcpu(i % rss_getnumbuckets()); -#else - /* - * Bind the msix vector, and thus the - * rings to the corresponding cpu. - * - * This just happens to match the default RSS round-robin - * bucket -> queue -> CPU allocation. - */ - if (adapter->num_queues > 1) { - if (igb_last_bind_cpu < 0) - igb_last_bind_cpu = CPU_FIRST(); - cpu_id = igb_last_bind_cpu; - } -#endif - - if (adapter->num_queues > 1) { - bus_bind_intr(dev, que->res, cpu_id); -#ifdef RSS - device_printf(dev, - "Bound queue %d to RSS bucket %d\n", - i, cpu_id); -#else - device_printf(dev, - "Bound queue %d to cpu %d\n", - i, cpu_id); -#endif - } - -#ifndef IGB_LEGACY_TX - TASK_INIT(&que->txr->txq_task, 0, igb_deferred_mq_start, - que->txr); -#endif - /* Make tasklet for deferred handling */ - TASK_INIT(&que->que_task, 0, igb_handle_que, que); - que->tq = taskqueue_create("igb_que", M_NOWAIT, - taskqueue_thread_enqueue, &que->tq); - if (adapter->num_queues > 1) { - /* - * Only pin the taskqueue thread to a CPU if - * RSS is in use. - * - * This again just happens to match the default RSS - * round-robin bucket -> queue -> CPU allocation. - */ -#ifdef RSS - CPU_SETOF(cpu_id, &cpu_mask); - taskqueue_start_threads_cpuset(&que->tq, 1, PI_NET, - &cpu_mask, - "%s que (bucket %d)", - device_get_nameunit(adapter->dev), - cpu_id); -#else - taskqueue_start_threads(&que->tq, 1, PI_NET, - "%s que (qid %d)", - device_get_nameunit(adapter->dev), - cpu_id); -#endif - } else { - taskqueue_start_threads(&que->tq, 1, PI_NET, "%s que", - device_get_nameunit(adapter->dev)); - } - - /* Finally update the last bound CPU id */ - if (adapter->num_queues > 1) - igb_last_bind_cpu = CPU_NEXT(igb_last_bind_cpu); - } - - /* And Link */ - rid = vector + 1; - adapter->res = bus_alloc_resource_any(dev, - SYS_RES_IRQ, &rid, RF_SHAREABLE | RF_ACTIVE); - if (adapter->res == NULL) { - device_printf(dev, - "Unable to allocate bus resource: " - "MSIX Link Interrupt\n"); - return (ENXIO); - } - if ((error = bus_setup_intr(dev, adapter->res, - INTR_TYPE_NET | INTR_MPSAFE, NULL, - igb_msix_link, adapter, &adapter->tag)) != 0) { - device_printf(dev, "Failed to register Link handler"); - return (error); - } -#if __FreeBSD_version >= 800504 - bus_describe_intr(dev, adapter->res, adapter->tag, "link"); -#endif - adapter->linkvec = vector; - - return (0); -} - - -static void -igb_configure_queues(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - struct igb_queue *que; - u32 tmp, ivar = 0, newitr = 0; - - /* First turn on RSS capability */ - if (adapter->hw.mac.type != e1000_82575) - E1000_WRITE_REG(hw, E1000_GPIE, - E1000_GPIE_MSIX_MODE | E1000_GPIE_EIAME | - E1000_GPIE_PBA | E1000_GPIE_NSICR); - - /* Turn on MSIX */ - switch (adapter->hw.mac.type) { - case e1000_82580: - case e1000_i350: - case e1000_i354: - case e1000_i210: - case e1000_i211: - case e1000_vfadapt: - case e1000_vfadapt_i350: - /* RX entries */ - for (int i = 0; i < adapter->num_queues; i++) { - u32 index = i >> 1; - ivar = E1000_READ_REG_ARRAY(hw, E1000_IVAR0, index); - que = &adapter->queues[i]; - if (i & 1) { - ivar &= 0xFF00FFFF; - ivar |= (que->msix | E1000_IVAR_VALID) << 16; - } else { - ivar &= 0xFFFFFF00; - ivar |= que->msix | E1000_IVAR_VALID; - } - E1000_WRITE_REG_ARRAY(hw, E1000_IVAR0, index, ivar); - } - /* TX entries */ - for (int i = 0; i < adapter->num_queues; i++) { - u32 index = i >> 1; - ivar = E1000_READ_REG_ARRAY(hw, E1000_IVAR0, index); - que = &adapter->queues[i]; - if (i & 1) { - ivar &= 0x00FFFFFF; - ivar |= (que->msix | E1000_IVAR_VALID) << 24; - } else { - ivar &= 0xFFFF00FF; - ivar |= (que->msix | E1000_IVAR_VALID) << 8; - } - E1000_WRITE_REG_ARRAY(hw, E1000_IVAR0, index, ivar); - adapter->que_mask |= que->eims; - } - - /* And for the link interrupt */ - ivar = (adapter->linkvec | E1000_IVAR_VALID) << 8; - adapter->link_mask = 1 << adapter->linkvec; - E1000_WRITE_REG(hw, E1000_IVAR_MISC, ivar); - break; - case e1000_82576: - /* RX entries */ - for (int i = 0; i < adapter->num_queues; i++) { - u32 index = i & 0x7; /* Each IVAR has two entries */ - ivar = E1000_READ_REG_ARRAY(hw, E1000_IVAR0, index); - que = &adapter->queues[i]; - if (i < 8) { - ivar &= 0xFFFFFF00; - ivar |= que->msix | E1000_IVAR_VALID; - } else { - ivar &= 0xFF00FFFF; - ivar |= (que->msix | E1000_IVAR_VALID) << 16; - } - E1000_WRITE_REG_ARRAY(hw, E1000_IVAR0, index, ivar); - adapter->que_mask |= que->eims; - } - /* TX entries */ - for (int i = 0; i < adapter->num_queues; i++) { - u32 index = i & 0x7; /* Each IVAR has two entries */ - ivar = E1000_READ_REG_ARRAY(hw, E1000_IVAR0, index); - que = &adapter->queues[i]; - if (i < 8) { - ivar &= 0xFFFF00FF; - ivar |= (que->msix | E1000_IVAR_VALID) << 8; - } else { - ivar &= 0x00FFFFFF; - ivar |= (que->msix | E1000_IVAR_VALID) << 24; - } - E1000_WRITE_REG_ARRAY(hw, E1000_IVAR0, index, ivar); - adapter->que_mask |= que->eims; - } - - /* And for the link interrupt */ - ivar = (adapter->linkvec | E1000_IVAR_VALID) << 8; - adapter->link_mask = 1 << adapter->linkvec; - E1000_WRITE_REG(hw, E1000_IVAR_MISC, ivar); - break; - - case e1000_82575: - /* enable MSI-X support*/ - tmp = E1000_READ_REG(hw, E1000_CTRL_EXT); - tmp |= E1000_CTRL_EXT_PBA_CLR; - /* Auto-Mask interrupts upon ICR read. */ - tmp |= E1000_CTRL_EXT_EIAME; - tmp |= E1000_CTRL_EXT_IRCA; - E1000_WRITE_REG(hw, E1000_CTRL_EXT, tmp); - - /* Queues */ - for (int i = 0; i < adapter->num_queues; i++) { - que = &adapter->queues[i]; - tmp = E1000_EICR_RX_QUEUE0 << i; - tmp |= E1000_EICR_TX_QUEUE0 << i; - que->eims = tmp; - E1000_WRITE_REG_ARRAY(hw, E1000_MSIXBM(0), - i, que->eims); - adapter->que_mask |= que->eims; - } - - /* Link */ - E1000_WRITE_REG(hw, E1000_MSIXBM(adapter->linkvec), - E1000_EIMS_OTHER); - adapter->link_mask |= E1000_EIMS_OTHER; - default: - break; - } - - /* Set the starting interrupt rate */ - if (igb_max_interrupt_rate > 0) - newitr = (4000000 / igb_max_interrupt_rate) & 0x7FFC; - - if (hw->mac.type == e1000_82575) - newitr |= newitr << 16; - else - newitr |= E1000_EITR_CNT_IGNR; - - for (int i = 0; i < adapter->num_queues; i++) { - que = &adapter->queues[i]; - E1000_WRITE_REG(hw, E1000_EITR(que->msix), newitr); - } - - return; -} - - -static void -igb_free_pci_resources(struct adapter *adapter) -{ - struct igb_queue *que = adapter->queues; - device_t dev = adapter->dev; - int rid; - - /* - ** There is a slight possibility of a failure mode - ** in attach that will result in entering this function - ** before interrupt resources have been initialized, and - ** in that case we do not want to execute the loops below - ** We can detect this reliably by the state of the adapter - ** res pointer. - */ - if (adapter->res == NULL) - goto mem; - - /* - * First release all the interrupt resources: - */ - for (int i = 0; i < adapter->num_queues; i++, que++) { - rid = que->msix + 1; - if (que->tag != NULL) { - bus_teardown_intr(dev, que->res, que->tag); - que->tag = NULL; - } - if (que->res != NULL) - bus_release_resource(dev, - SYS_RES_IRQ, rid, que->res); - } - - /* Clean the Legacy or Link interrupt last */ - if (adapter->linkvec) /* we are doing MSIX */ - rid = adapter->linkvec + 1; - else - (adapter->msix != 0) ? (rid = 1):(rid = 0); - - que = adapter->queues; - if (adapter->tag != NULL) { - taskqueue_drain(que->tq, &adapter->link_task); - bus_teardown_intr(dev, adapter->res, adapter->tag); - adapter->tag = NULL; - } - if (adapter->res != NULL) - bus_release_resource(dev, SYS_RES_IRQ, rid, adapter->res); - - for (int i = 0; i < adapter->num_queues; i++, que++) { - if (que->tq != NULL) { -#ifndef IGB_LEGACY_TX - taskqueue_drain(que->tq, &que->txr->txq_task); -#endif - taskqueue_drain(que->tq, &que->que_task); - taskqueue_free(que->tq); - } - } -mem: - if (adapter->msix) - pci_release_msi(dev); - - if (adapter->msix_mem != NULL) - bus_release_resource(dev, SYS_RES_MEMORY, - adapter->memrid, adapter->msix_mem); - - if (adapter->pci_mem != NULL) - bus_release_resource(dev, SYS_RES_MEMORY, - PCIR_BAR(0), adapter->pci_mem); - -} - -/* - * Setup Either MSI/X or MSI - */ -static int -igb_setup_msix(struct adapter *adapter) -{ - device_t dev = adapter->dev; - int bar, want, queues, msgs, maxqueues; - - /* tuneable override */ - if (igb_enable_msix == 0) - goto msi; - - /* First try MSI/X */ - msgs = pci_msix_count(dev); - if (msgs == 0) - goto msi; - /* - ** Some new devices, as with ixgbe, now may - ** use a different BAR, so we need to keep - ** track of which is used. - */ - adapter->memrid = PCIR_BAR(IGB_MSIX_BAR); - bar = pci_read_config(dev, adapter->memrid, 4); - if (bar == 0) /* use next bar */ - adapter->memrid += 4; - adapter->msix_mem = bus_alloc_resource_any(dev, - SYS_RES_MEMORY, &adapter->memrid, RF_ACTIVE); - if (adapter->msix_mem == NULL) { - /* May not be enabled */ - device_printf(adapter->dev, - "Unable to map MSIX table \n"); - goto msi; - } - - queues = (mp_ncpus > (msgs-1)) ? (msgs-1) : mp_ncpus; - - /* Override via tuneable */ - if (igb_num_queues != 0) - queues = igb_num_queues; - -#ifdef RSS - /* If we're doing RSS, clamp at the number of RSS buckets */ - if (queues > rss_getnumbuckets()) - queues = rss_getnumbuckets(); -#endif - - - /* Sanity check based on HW */ - switch (adapter->hw.mac.type) { - case e1000_82575: - maxqueues = 4; - break; - case e1000_82576: - case e1000_82580: - case e1000_i350: - case e1000_i354: - maxqueues = 8; - break; - case e1000_i210: - maxqueues = 4; - break; - case e1000_i211: - maxqueues = 2; - break; - default: /* VF interfaces */ - maxqueues = 1; - break; - } - - /* Final clamp on the actual hardware capability */ - if (queues > maxqueues) - queues = maxqueues; - - /* - ** One vector (RX/TX pair) per queue - ** plus an additional for Link interrupt - */ - want = queues + 1; - if (msgs >= want) - msgs = want; - else { - device_printf(adapter->dev, - "MSIX Configuration Problem, " - "%d vectors configured, but %d queues wanted!\n", - msgs, want); - goto msi; - } - if ((pci_alloc_msix(dev, &msgs) == 0) && (msgs == want)) { - device_printf(adapter->dev, - "Using MSIX interrupts with %d vectors\n", msgs); - adapter->num_queues = queues; - return (msgs); - } - /* - ** If MSIX alloc failed or provided us with - ** less than needed, free and fall through to MSI - */ - pci_release_msi(dev); - -msi: - if (adapter->msix_mem != NULL) { - bus_release_resource(dev, SYS_RES_MEMORY, - PCIR_BAR(IGB_MSIX_BAR), adapter->msix_mem); - adapter->msix_mem = NULL; - } - msgs = 1; - if (pci_alloc_msi(dev, &msgs) == 0) { - device_printf(adapter->dev," Using an MSI interrupt\n"); - return (msgs); - } - device_printf(adapter->dev," Using a Legacy interrupt\n"); - return (0); -} - -/********************************************************************* - * - * Initialize the DMA Coalescing feature - * - **********************************************************************/ -static void -igb_init_dmac(struct adapter *adapter, u32 pba) -{ - device_t dev = adapter->dev; - struct e1000_hw *hw = &adapter->hw; - u32 dmac, reg = ~E1000_DMACR_DMAC_EN; - u16 hwm; - - if (hw->mac.type == e1000_i211) - return; - - if (hw->mac.type > e1000_82580) { - - if (adapter->dmac == 0) { /* Disabling it */ - E1000_WRITE_REG(hw, E1000_DMACR, reg); - return; - } else - device_printf(dev, "DMA Coalescing enabled\n"); - - /* Set starting threshold */ - E1000_WRITE_REG(hw, E1000_DMCTXTH, 0); - - hwm = 64 * pba - adapter->max_frame_size / 16; - if (hwm < 64 * (pba - 6)) - hwm = 64 * (pba - 6); - reg = E1000_READ_REG(hw, E1000_FCRTC); - reg &= ~E1000_FCRTC_RTH_COAL_MASK; - reg |= ((hwm << E1000_FCRTC_RTH_COAL_SHIFT) - & E1000_FCRTC_RTH_COAL_MASK); - E1000_WRITE_REG(hw, E1000_FCRTC, reg); - - - dmac = pba - adapter->max_frame_size / 512; - if (dmac < pba - 10) - dmac = pba - 10; - reg = E1000_READ_REG(hw, E1000_DMACR); - reg &= ~E1000_DMACR_DMACTHR_MASK; - reg = ((dmac << E1000_DMACR_DMACTHR_SHIFT) - & E1000_DMACR_DMACTHR_MASK); - - /* transition to L0x or L1 if available..*/ - reg |= (E1000_DMACR_DMAC_EN | E1000_DMACR_DMAC_LX_MASK); - - /* Check if status is 2.5Gb backplane connection - * before configuration of watchdog timer, which is - * in msec values in 12.8usec intervals - * watchdog timer= msec values in 32usec intervals - * for non 2.5Gb connection - */ - if (hw->mac.type == e1000_i354) { - int status = E1000_READ_REG(hw, E1000_STATUS); - if ((status & E1000_STATUS_2P5_SKU) && - (!(status & E1000_STATUS_2P5_SKU_OVER))) - reg |= ((adapter->dmac * 5) >> 6); - else - reg |= (adapter->dmac >> 5); - } else { - reg |= (adapter->dmac >> 5); - } - - E1000_WRITE_REG(hw, E1000_DMACR, reg); - - E1000_WRITE_REG(hw, E1000_DMCRTRH, 0); - - /* Set the interval before transition */ - reg = E1000_READ_REG(hw, E1000_DMCTLX); - if (hw->mac.type == e1000_i350) - reg |= IGB_DMCTLX_DCFLUSH_DIS; - /* - ** in 2.5Gb connection, TTLX unit is 0.4 usec - ** which is 0x4*2 = 0xA. But delay is still 4 usec - */ - if (hw->mac.type == e1000_i354) { - int status = E1000_READ_REG(hw, E1000_STATUS); - if ((status & E1000_STATUS_2P5_SKU) && - (!(status & E1000_STATUS_2P5_SKU_OVER))) - reg |= 0xA; - else - reg |= 0x4; - } else { - reg |= 0x4; - } - - E1000_WRITE_REG(hw, E1000_DMCTLX, reg); - - /* free space in tx packet buffer to wake from DMA coal */ - E1000_WRITE_REG(hw, E1000_DMCTXTH, (IGB_TXPBSIZE - - (2 * adapter->max_frame_size)) >> 6); - - /* make low power state decision controlled by DMA coal */ - reg = E1000_READ_REG(hw, E1000_PCIEMISC); - reg &= ~E1000_PCIEMISC_LX_DECISION; - E1000_WRITE_REG(hw, E1000_PCIEMISC, reg); - - } else if (hw->mac.type == e1000_82580) { - u32 reg = E1000_READ_REG(hw, E1000_PCIEMISC); - E1000_WRITE_REG(hw, E1000_PCIEMISC, - reg & ~E1000_PCIEMISC_LX_DECISION); - E1000_WRITE_REG(hw, E1000_DMACR, 0); - } -} - - -/********************************************************************* - * - * Set up an fresh starting state - * - **********************************************************************/ -static void -igb_reset(struct adapter *adapter) -{ - device_t dev = adapter->dev; - struct e1000_hw *hw = &adapter->hw; - struct e1000_fc_info *fc = &hw->fc; - struct ifnet *ifp = adapter->ifp; - u32 pba = 0; - u16 hwm; - - INIT_DEBUGOUT("igb_reset: begin"); - - /* Let the firmware know the OS is in control */ - igb_get_hw_control(adapter); - - /* - * Packet Buffer Allocation (PBA) - * Writing PBA sets the receive portion of the buffer - * the remainder is used for the transmit buffer. - */ - switch (hw->mac.type) { - case e1000_82575: - pba = E1000_PBA_32K; - break; - case e1000_82576: - case e1000_vfadapt: - pba = E1000_READ_REG(hw, E1000_RXPBS); - pba &= E1000_RXPBS_SIZE_MASK_82576; - break; - case e1000_82580: - case e1000_i350: - case e1000_i354: - case e1000_vfadapt_i350: - pba = E1000_READ_REG(hw, E1000_RXPBS); - pba = e1000_rxpbs_adjust_82580(pba); - break; - case e1000_i210: - case e1000_i211: - pba = E1000_PBA_34K; - default: - break; - } - - /* Special needs in case of Jumbo frames */ - if ((hw->mac.type == e1000_82575) && (ifp->if_mtu > ETHERMTU)) { - u32 tx_space, min_tx, min_rx; - pba = E1000_READ_REG(hw, E1000_PBA); - tx_space = pba >> 16; - pba &= 0xffff; - min_tx = (adapter->max_frame_size + - sizeof(struct e1000_tx_desc) - ETHERNET_FCS_SIZE) * 2; - min_tx = roundup2(min_tx, 1024); - min_tx >>= 10; - min_rx = adapter->max_frame_size; - min_rx = roundup2(min_rx, 1024); - min_rx >>= 10; - if (tx_space < min_tx && - ((min_tx - tx_space) < pba)) { - pba = pba - (min_tx - tx_space); - /* - * if short on rx space, rx wins - * and must trump tx adjustment - */ - if (pba < min_rx) - pba = min_rx; - } - E1000_WRITE_REG(hw, E1000_PBA, pba); - } - - INIT_DEBUGOUT1("igb_init: pba=%dK",pba); - - /* - * These parameters control the automatic generation (Tx) and - * response (Rx) to Ethernet PAUSE frames. - * - High water mark should allow for at least two frames to be - * received after sending an XOFF. - * - Low water mark works best when it is very near the high water mark. - * This allows the receiver to restart by sending XON when it has - * drained a bit. - */ - hwm = min(((pba << 10) * 9 / 10), - ((pba << 10) - 2 * adapter->max_frame_size)); - - if (hw->mac.type < e1000_82576) { - fc->high_water = hwm & 0xFFF8; /* 8-byte granularity */ - fc->low_water = fc->high_water - 8; - } else { - fc->high_water = hwm & 0xFFF0; /* 16-byte granularity */ - fc->low_water = fc->high_water - 16; - } - - fc->pause_time = IGB_FC_PAUSE_TIME; - fc->send_xon = TRUE; - if (adapter->fc) - fc->requested_mode = adapter->fc; - else - fc->requested_mode = e1000_fc_default; - - /* Issue a global reset */ - e1000_reset_hw(hw); - E1000_WRITE_REG(hw, E1000_WUC, 0); - - /* Reset for AutoMediaDetect */ - if (adapter->flags & IGB_MEDIA_RESET) { - e1000_setup_init_funcs(hw, TRUE); - e1000_get_bus_info(hw); - adapter->flags &= ~IGB_MEDIA_RESET; - } - - if (e1000_init_hw(hw) < 0) - device_printf(dev, "Hardware Initialization Failed\n"); - - /* Setup DMA Coalescing */ - igb_init_dmac(adapter, pba); - - E1000_WRITE_REG(&adapter->hw, E1000_VET, ETHERTYPE_VLAN); - e1000_get_phy_info(hw); - e1000_check_for_link(hw); - return; -} - -/********************************************************************* - * - * Setup networking device structure and register an interface. - * - **********************************************************************/ -static int -igb_setup_interface(device_t dev, struct adapter *adapter) -{ - struct ifnet *ifp; - - INIT_DEBUGOUT("igb_setup_interface: begin"); - - ifp = adapter->ifp = if_alloc(IFT_ETHER); - if (ifp == NULL) { - device_printf(dev, "can not allocate ifnet structure\n"); - return (-1); - } - if_initname(ifp, device_get_name(dev), device_get_unit(dev)); - ifp->if_init = igb_init; - ifp->if_softc = adapter; - ifp->if_flags = IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST; - ifp->if_ioctl = igb_ioctl; - ifp->if_get_counter = igb_get_counter; - - /* TSO parameters */ - ifp->if_hw_tsomax = IP_MAXPACKET; - ifp->if_hw_tsomaxsegcount = IGB_MAX_SCATTER; - ifp->if_hw_tsomaxsegsize = IGB_TSO_SEG_SIZE; - -#ifndef IGB_LEGACY_TX - ifp->if_transmit = igb_mq_start; - ifp->if_qflush = igb_qflush; -#else - ifp->if_start = igb_start; - IFQ_SET_MAXLEN(&ifp->if_snd, adapter->num_tx_desc - 1); - ifp->if_snd.ifq_drv_maxlen = adapter->num_tx_desc - 1; - IFQ_SET_READY(&ifp->if_snd); -#endif - - ether_ifattach(ifp, adapter->hw.mac.addr); - - ifp->if_capabilities = ifp->if_capenable = 0; - - ifp->if_capabilities = IFCAP_HWCSUM | IFCAP_VLAN_HWCSUM; -#if __FreeBSD_version >= 1000000 - ifp->if_capabilities |= IFCAP_HWCSUM_IPV6; -#endif - ifp->if_capabilities |= IFCAP_TSO; - ifp->if_capabilities |= IFCAP_JUMBO_MTU; - ifp->if_capenable = ifp->if_capabilities; - - /* Don't enable LRO by default */ - ifp->if_capabilities |= IFCAP_LRO; - -#ifdef DEVICE_POLLING - ifp->if_capabilities |= IFCAP_POLLING; -#endif - - /* - * Tell the upper layer(s) we - * support full VLAN capability. - */ - ifp->if_hdrlen = sizeof(struct ether_vlan_header); - ifp->if_capabilities |= IFCAP_VLAN_HWTAGGING - | IFCAP_VLAN_HWTSO - | IFCAP_VLAN_MTU; - ifp->if_capenable |= IFCAP_VLAN_HWTAGGING - | IFCAP_VLAN_HWTSO - | IFCAP_VLAN_MTU; - - /* - ** Don't turn this on by default, if vlans are - ** created on another pseudo device (eg. lagg) - ** then vlan events are not passed thru, breaking - ** operation, but with HW FILTER off it works. If - ** using vlans directly on the igb driver you can - ** enable this and get full hardware tag filtering. - */ - ifp->if_capabilities |= IFCAP_VLAN_HWFILTER; - - /* - * Specify the media types supported by this adapter and register - * callbacks to update media and link information - */ - ifmedia_init(&adapter->media, IFM_IMASK, - igb_media_change, igb_media_status); - if ((adapter->hw.phy.media_type == e1000_media_type_fiber) || - (adapter->hw.phy.media_type == e1000_media_type_internal_serdes)) { - ifmedia_add(&adapter->media, IFM_ETHER | IFM_1000_SX | IFM_FDX, - 0, NULL); - ifmedia_add(&adapter->media, IFM_ETHER | IFM_1000_SX, 0, NULL); - } else { - ifmedia_add(&adapter->media, IFM_ETHER | IFM_10_T, 0, NULL); - ifmedia_add(&adapter->media, IFM_ETHER | IFM_10_T | IFM_FDX, - 0, NULL); - ifmedia_add(&adapter->media, IFM_ETHER | IFM_100_TX, - 0, NULL); - ifmedia_add(&adapter->media, IFM_ETHER | IFM_100_TX | IFM_FDX, - 0, NULL); - if (adapter->hw.phy.type != e1000_phy_ife) { - ifmedia_add(&adapter->media, - IFM_ETHER | IFM_1000_T | IFM_FDX, 0, NULL); - ifmedia_add(&adapter->media, - IFM_ETHER | IFM_1000_T, 0, NULL); - } - } - ifmedia_add(&adapter->media, IFM_ETHER | IFM_AUTO, 0, NULL); - ifmedia_set(&adapter->media, IFM_ETHER | IFM_AUTO); - return (0); -} - - -/* - * Manage DMA'able memory. - */ -static void -igb_dmamap_cb(void *arg, bus_dma_segment_t *segs, int nseg, int error) -{ - if (error) - return; - *(bus_addr_t *) arg = segs[0].ds_addr; -} - -static int -igb_dma_malloc(struct adapter *adapter, bus_size_t size, - struct igb_dma_alloc *dma, int mapflags) -{ - int error; - - error = bus_dma_tag_create(bus_get_dma_tag(adapter->dev), /* parent */ - IGB_DBA_ALIGN, 0, /* alignment, bounds */ - BUS_SPACE_MAXADDR, /* lowaddr */ - BUS_SPACE_MAXADDR, /* highaddr */ - NULL, NULL, /* filter, filterarg */ - size, /* maxsize */ - 1, /* nsegments */ - size, /* maxsegsize */ - 0, /* flags */ - NULL, /* lockfunc */ - NULL, /* lockarg */ - &dma->dma_tag); - if (error) { - device_printf(adapter->dev, - "%s: bus_dma_tag_create failed: %d\n", - __func__, error); - goto fail_0; - } - - error = bus_dmamem_alloc(dma->dma_tag, (void**) &dma->dma_vaddr, - BUS_DMA_NOWAIT | BUS_DMA_COHERENT, &dma->dma_map); - if (error) { - device_printf(adapter->dev, - "%s: bus_dmamem_alloc(%ju) failed: %d\n", - __func__, (uintmax_t)size, error); - goto fail_2; - } - - dma->dma_paddr = 0; - error = bus_dmamap_load(dma->dma_tag, dma->dma_map, dma->dma_vaddr, - size, igb_dmamap_cb, &dma->dma_paddr, mapflags | BUS_DMA_NOWAIT); - if (error || dma->dma_paddr == 0) { - device_printf(adapter->dev, - "%s: bus_dmamap_load failed: %d\n", - __func__, error); - goto fail_3; - } - - return (0); - -fail_3: - bus_dmamap_unload(dma->dma_tag, dma->dma_map); -fail_2: - bus_dmamem_free(dma->dma_tag, dma->dma_vaddr, dma->dma_map); - bus_dma_tag_destroy(dma->dma_tag); -fail_0: - dma->dma_tag = NULL; - - return (error); -} - -static void -igb_dma_free(struct adapter *adapter, struct igb_dma_alloc *dma) -{ - if (dma->dma_tag == NULL) - return; - if (dma->dma_paddr != 0) { - bus_dmamap_sync(dma->dma_tag, dma->dma_map, - BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); - bus_dmamap_unload(dma->dma_tag, dma->dma_map); - dma->dma_paddr = 0; - } - if (dma->dma_vaddr != NULL) { - bus_dmamem_free(dma->dma_tag, dma->dma_vaddr, dma->dma_map); - dma->dma_vaddr = NULL; - } - bus_dma_tag_destroy(dma->dma_tag); - dma->dma_tag = NULL; -} - - -/********************************************************************* - * - * Allocate memory for the transmit and receive rings, and then - * the descriptors associated with each, called only once at attach. - * - **********************************************************************/ -static int -igb_allocate_queues(struct adapter *adapter) -{ - device_t dev = adapter->dev; - struct igb_queue *que = NULL; - struct tx_ring *txr = NULL; - struct rx_ring *rxr = NULL; - int rsize, tsize, error = E1000_SUCCESS; - int txconf = 0, rxconf = 0; - - /* First allocate the top level queue structs */ - if (!(adapter->queues = - (struct igb_queue *) malloc(sizeof(struct igb_queue) * - adapter->num_queues, M_DEVBUF, M_NOWAIT | M_ZERO))) { - device_printf(dev, "Unable to allocate queue memory\n"); - error = ENOMEM; - goto fail; - } - - /* Next allocate the TX ring struct memory */ - if (!(adapter->tx_rings = - (struct tx_ring *) malloc(sizeof(struct tx_ring) * - adapter->num_queues, M_DEVBUF, M_NOWAIT | M_ZERO))) { - device_printf(dev, "Unable to allocate TX ring memory\n"); - error = ENOMEM; - goto tx_fail; - } - - /* Now allocate the RX */ - if (!(adapter->rx_rings = - (struct rx_ring *) malloc(sizeof(struct rx_ring) * - adapter->num_queues, M_DEVBUF, M_NOWAIT | M_ZERO))) { - device_printf(dev, "Unable to allocate RX ring memory\n"); - error = ENOMEM; - goto rx_fail; - } - - tsize = roundup2(adapter->num_tx_desc * - sizeof(union e1000_adv_tx_desc), IGB_DBA_ALIGN); - /* - * Now set up the TX queues, txconf is needed to handle the - * possibility that things fail midcourse and we need to - * undo memory gracefully - */ - for (int i = 0; i < adapter->num_queues; i++, txconf++) { - /* Set up some basics */ - txr = &adapter->tx_rings[i]; - txr->adapter = adapter; - txr->me = i; - txr->num_desc = adapter->num_tx_desc; - - /* Initialize the TX lock */ - snprintf(txr->mtx_name, sizeof(txr->mtx_name), "%s:tx(%d)", - device_get_nameunit(dev), txr->me); - mtx_init(&txr->tx_mtx, txr->mtx_name, NULL, MTX_DEF); - - if (igb_dma_malloc(adapter, tsize, - &txr->txdma, BUS_DMA_NOWAIT)) { - device_printf(dev, - "Unable to allocate TX Descriptor memory\n"); - error = ENOMEM; - goto err_tx_desc; - } - txr->tx_base = (union e1000_adv_tx_desc *)txr->txdma.dma_vaddr; - bzero((void *)txr->tx_base, tsize); - - /* Now allocate transmit buffers for the ring */ - if (igb_allocate_transmit_buffers(txr)) { - device_printf(dev, - "Critical Failure setting up transmit buffers\n"); - error = ENOMEM; - goto err_tx_desc; - } -#ifndef IGB_LEGACY_TX - /* Allocate a buf ring */ - txr->br = buf_ring_alloc(igb_buf_ring_size, M_DEVBUF, - M_WAITOK, &txr->tx_mtx); -#endif - } - - /* - * Next the RX queues... - */ - rsize = roundup2(adapter->num_rx_desc * - sizeof(union e1000_adv_rx_desc), IGB_DBA_ALIGN); - for (int i = 0; i < adapter->num_queues; i++, rxconf++) { - rxr = &adapter->rx_rings[i]; - rxr->adapter = adapter; - rxr->me = i; - - /* Initialize the RX lock */ - snprintf(rxr->mtx_name, sizeof(rxr->mtx_name), "%s:rx(%d)", - device_get_nameunit(dev), txr->me); - mtx_init(&rxr->rx_mtx, rxr->mtx_name, NULL, MTX_DEF); - - if (igb_dma_malloc(adapter, rsize, - &rxr->rxdma, BUS_DMA_NOWAIT)) { - device_printf(dev, - "Unable to allocate RxDescriptor memory\n"); - error = ENOMEM; - goto err_rx_desc; - } - rxr->rx_base = (union e1000_adv_rx_desc *)rxr->rxdma.dma_vaddr; - bzero((void *)rxr->rx_base, rsize); - - /* Allocate receive buffers for the ring*/ - if (igb_allocate_receive_buffers(rxr)) { - device_printf(dev, - "Critical Failure setting up receive buffers\n"); - error = ENOMEM; - goto err_rx_desc; - } - } - - /* - ** Finally set up the queue holding structs - */ - for (int i = 0; i < adapter->num_queues; i++) { - que = &adapter->queues[i]; - que->adapter = adapter; - que->txr = &adapter->tx_rings[i]; - que->rxr = &adapter->rx_rings[i]; - } - - return (0); - -err_rx_desc: - for (rxr = adapter->rx_rings; rxconf > 0; rxr++, rxconf--) - igb_dma_free(adapter, &rxr->rxdma); -err_tx_desc: - for (txr = adapter->tx_rings; txconf > 0; txr++, txconf--) - igb_dma_free(adapter, &txr->txdma); - free(adapter->rx_rings, M_DEVBUF); -rx_fail: -#ifndef IGB_LEGACY_TX - buf_ring_free(txr->br, M_DEVBUF); -#endif - free(adapter->tx_rings, M_DEVBUF); -tx_fail: - free(adapter->queues, M_DEVBUF); -fail: - return (error); -} - -/********************************************************************* - * - * Allocate memory for tx_buffer structures. The tx_buffer stores all - * the information needed to transmit a packet on the wire. This is - * called only once at attach, setup is done every reset. - * - **********************************************************************/ -static int -igb_allocate_transmit_buffers(struct tx_ring *txr) -{ - struct adapter *adapter = txr->adapter; - device_t dev = adapter->dev; - struct igb_tx_buf *txbuf; - int error, i; - - /* - * Setup DMA descriptor areas. - */ - if ((error = bus_dma_tag_create(bus_get_dma_tag(dev), - 1, 0, /* alignment, bounds */ - BUS_SPACE_MAXADDR, /* lowaddr */ - BUS_SPACE_MAXADDR, /* highaddr */ - NULL, NULL, /* filter, filterarg */ - IGB_TSO_SIZE, /* maxsize */ - IGB_MAX_SCATTER, /* nsegments */ - PAGE_SIZE, /* maxsegsize */ - 0, /* flags */ - NULL, /* lockfunc */ - NULL, /* lockfuncarg */ - &txr->txtag))) { - device_printf(dev,"Unable to allocate TX DMA tag\n"); - goto fail; - } - - if (!(txr->tx_buffers = - (struct igb_tx_buf *) malloc(sizeof(struct igb_tx_buf) * - adapter->num_tx_desc, M_DEVBUF, M_NOWAIT | M_ZERO))) { - device_printf(dev, "Unable to allocate tx_buffer memory\n"); - error = ENOMEM; - goto fail; - } - - /* Create the descriptor buffer dma maps */ - txbuf = txr->tx_buffers; - for (i = 0; i < adapter->num_tx_desc; i++, txbuf++) { - error = bus_dmamap_create(txr->txtag, 0, &txbuf->map); - if (error != 0) { - device_printf(dev, "Unable to create TX DMA map\n"); - goto fail; - } - } - - return 0; -fail: - /* We free all, it handles case where we are in the middle */ - igb_free_transmit_structures(adapter); - return (error); -} - -/********************************************************************* - * - * Initialize a transmit ring. - * - **********************************************************************/ -static void -igb_setup_transmit_ring(struct tx_ring *txr) -{ - struct adapter *adapter = txr->adapter; - struct igb_tx_buf *txbuf; - int i; -#ifdef DEV_NETMAP - struct netmap_adapter *na = NA(adapter->ifp); - struct netmap_slot *slot; -#endif /* DEV_NETMAP */ - - /* Clear the old descriptor contents */ - IGB_TX_LOCK(txr); -#ifdef DEV_NETMAP - slot = netmap_reset(na, NR_TX, txr->me, 0); -#endif /* DEV_NETMAP */ - bzero((void *)txr->tx_base, - (sizeof(union e1000_adv_tx_desc)) * adapter->num_tx_desc); - /* Reset indices */ - txr->next_avail_desc = 0; - txr->next_to_clean = 0; - - /* Free any existing tx buffers. */ - txbuf = txr->tx_buffers; - for (i = 0; i < adapter->num_tx_desc; i++, txbuf++) { - if (txbuf->m_head != NULL) { - bus_dmamap_sync(txr->txtag, txbuf->map, - BUS_DMASYNC_POSTWRITE); - bus_dmamap_unload(txr->txtag, txbuf->map); - m_freem(txbuf->m_head); - txbuf->m_head = NULL; - } -#ifdef DEV_NETMAP - if (slot) { - int si = netmap_idx_n2k(&na->tx_rings[txr->me], i); - /* no need to set the address */ - netmap_load_map(na, txr->txtag, txbuf->map, NMB(na, slot + si)); - } -#endif /* DEV_NETMAP */ - /* clear the watch index */ - txbuf->eop = NULL; - } - - /* Set number of descriptors available */ - txr->tx_avail = adapter->num_tx_desc; - - bus_dmamap_sync(txr->txdma.dma_tag, txr->txdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - IGB_TX_UNLOCK(txr); -} - -/********************************************************************* - * - * Initialize all transmit rings. - * - **********************************************************************/ -static void -igb_setup_transmit_structures(struct adapter *adapter) -{ - struct tx_ring *txr = adapter->tx_rings; - - for (int i = 0; i < adapter->num_queues; i++, txr++) - igb_setup_transmit_ring(txr); - - return; -} - -/********************************************************************* - * - * Enable transmit unit. - * - **********************************************************************/ -static void -igb_initialize_transmit_units(struct adapter *adapter) -{ - struct tx_ring *txr = adapter->tx_rings; - struct e1000_hw *hw = &adapter->hw; - u32 tctl, txdctl; - - INIT_DEBUGOUT("igb_initialize_transmit_units: begin"); - tctl = txdctl = 0; - - /* Setup the Tx Descriptor Rings */ - for (int i = 0; i < adapter->num_queues; i++, txr++) { - u64 bus_addr = txr->txdma.dma_paddr; - - E1000_WRITE_REG(hw, E1000_TDLEN(i), - adapter->num_tx_desc * sizeof(struct e1000_tx_desc)); - E1000_WRITE_REG(hw, E1000_TDBAH(i), - (uint32_t)(bus_addr >> 32)); - E1000_WRITE_REG(hw, E1000_TDBAL(i), - (uint32_t)bus_addr); - - /* Setup the HW Tx Head and Tail descriptor pointers */ - E1000_WRITE_REG(hw, E1000_TDT(i), 0); - E1000_WRITE_REG(hw, E1000_TDH(i), 0); - - HW_DEBUGOUT2("Base = %x, Length = %x\n", - E1000_READ_REG(hw, E1000_TDBAL(i)), - E1000_READ_REG(hw, E1000_TDLEN(i))); - - txr->queue_status = IGB_QUEUE_IDLE; - - txdctl |= IGB_TX_PTHRESH; - txdctl |= IGB_TX_HTHRESH << 8; - txdctl |= IGB_TX_WTHRESH << 16; - txdctl |= E1000_TXDCTL_QUEUE_ENABLE; - E1000_WRITE_REG(hw, E1000_TXDCTL(i), txdctl); - } - - if (adapter->vf_ifp) - return; - - e1000_config_collision_dist(hw); - - /* Program the Transmit Control Register */ - tctl = E1000_READ_REG(hw, E1000_TCTL); - tctl &= ~E1000_TCTL_CT; - tctl |= (E1000_TCTL_PSP | E1000_TCTL_RTLC | E1000_TCTL_EN | - (E1000_COLLISION_THRESHOLD << E1000_CT_SHIFT)); - - /* This write will effectively turn on the transmit unit. */ - E1000_WRITE_REG(hw, E1000_TCTL, tctl); -} - -/********************************************************************* - * - * Free all transmit rings. - * - **********************************************************************/ -static void -igb_free_transmit_structures(struct adapter *adapter) -{ - struct tx_ring *txr = adapter->tx_rings; - - for (int i = 0; i < adapter->num_queues; i++, txr++) { - IGB_TX_LOCK(txr); - igb_free_transmit_buffers(txr); - igb_dma_free(adapter, &txr->txdma); - IGB_TX_UNLOCK(txr); - IGB_TX_LOCK_DESTROY(txr); - } - free(adapter->tx_rings, M_DEVBUF); -} - -/********************************************************************* - * - * Free transmit ring related data structures. - * - **********************************************************************/ -static void -igb_free_transmit_buffers(struct tx_ring *txr) -{ - struct adapter *adapter = txr->adapter; - struct igb_tx_buf *tx_buffer; - int i; - - INIT_DEBUGOUT("free_transmit_ring: begin"); - - if (txr->tx_buffers == NULL) - return; - - tx_buffer = txr->tx_buffers; - for (i = 0; i < adapter->num_tx_desc; i++, tx_buffer++) { - if (tx_buffer->m_head != NULL) { - bus_dmamap_sync(txr->txtag, tx_buffer->map, - BUS_DMASYNC_POSTWRITE); - bus_dmamap_unload(txr->txtag, - tx_buffer->map); - m_freem(tx_buffer->m_head); - tx_buffer->m_head = NULL; - if (tx_buffer->map != NULL) { - bus_dmamap_destroy(txr->txtag, - tx_buffer->map); - tx_buffer->map = NULL; - } - } else if (tx_buffer->map != NULL) { - bus_dmamap_unload(txr->txtag, - tx_buffer->map); - bus_dmamap_destroy(txr->txtag, - tx_buffer->map); - tx_buffer->map = NULL; - } - } -#ifndef IGB_LEGACY_TX - if (txr->br != NULL) - buf_ring_free(txr->br, M_DEVBUF); -#endif - if (txr->tx_buffers != NULL) { - free(txr->tx_buffers, M_DEVBUF); - txr->tx_buffers = NULL; - } - if (txr->txtag != NULL) { - bus_dma_tag_destroy(txr->txtag); - txr->txtag = NULL; - } - return; -} - -/********************************************************************** - * - * Setup work for hardware segmentation offload (TSO) on - * adapters using advanced tx descriptors - * - **********************************************************************/ -static int -igb_tso_setup(struct tx_ring *txr, struct mbuf *mp, - u32 *cmd_type_len, u32 *olinfo_status) -{ - struct adapter *adapter = txr->adapter; - struct e1000_adv_tx_context_desc *TXD; - u32 vlan_macip_lens = 0, type_tucmd_mlhl = 0; - u32 mss_l4len_idx = 0, paylen; - u16 vtag = 0, eh_type; - int ctxd, ehdrlen, ip_hlen, tcp_hlen; - struct ether_vlan_header *eh; -#ifdef INET6 - struct ip6_hdr *ip6; -#endif -#ifdef INET - struct ip *ip; -#endif - struct tcphdr *th; - - - /* - * Determine where frame payload starts. - * Jump over vlan headers if already present - */ - eh = mtod(mp, struct ether_vlan_header *); - if (eh->evl_encap_proto == htons(ETHERTYPE_VLAN)) { - ehdrlen = ETHER_HDR_LEN + ETHER_VLAN_ENCAP_LEN; - eh_type = eh->evl_proto; - } else { - ehdrlen = ETHER_HDR_LEN; - eh_type = eh->evl_encap_proto; - } - - switch (ntohs(eh_type)) { -#ifdef INET6 - case ETHERTYPE_IPV6: - ip6 = (struct ip6_hdr *)(mp->m_data + ehdrlen); - /* XXX-BZ For now we do not pretend to support ext. hdrs. */ - if (ip6->ip6_nxt != IPPROTO_TCP) - return (ENXIO); - ip_hlen = sizeof(struct ip6_hdr); - ip6 = (struct ip6_hdr *)(mp->m_data + ehdrlen); - th = (struct tcphdr *)((caddr_t)ip6 + ip_hlen); - th->th_sum = in6_cksum_pseudo(ip6, 0, IPPROTO_TCP, 0); - type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_IPV6; - break; -#endif -#ifdef INET - case ETHERTYPE_IP: - ip = (struct ip *)(mp->m_data + ehdrlen); - if (ip->ip_p != IPPROTO_TCP) - return (ENXIO); - ip->ip_sum = 0; - ip_hlen = ip->ip_hl << 2; - th = (struct tcphdr *)((caddr_t)ip + ip_hlen); - th->th_sum = in_pseudo(ip->ip_src.s_addr, - ip->ip_dst.s_addr, htons(IPPROTO_TCP)); - type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_IPV4; - /* Tell transmit desc to also do IPv4 checksum. */ - *olinfo_status |= E1000_TXD_POPTS_IXSM << 8; - break; -#endif - default: - panic("%s: CSUM_TSO but no supported IP version (0x%04x)", - __func__, ntohs(eh_type)); - break; - } - - ctxd = txr->next_avail_desc; - TXD = (struct e1000_adv_tx_context_desc *) &txr->tx_base[ctxd]; - - tcp_hlen = th->th_off << 2; - - /* This is used in the transmit desc in encap */ - paylen = mp->m_pkthdr.len - ehdrlen - ip_hlen - tcp_hlen; - - /* VLAN MACLEN IPLEN */ - if (mp->m_flags & M_VLANTAG) { - vtag = htole16(mp->m_pkthdr.ether_vtag); - vlan_macip_lens |= (vtag << E1000_ADVTXD_VLAN_SHIFT); - } - - vlan_macip_lens |= ehdrlen << E1000_ADVTXD_MACLEN_SHIFT; - vlan_macip_lens |= ip_hlen; - TXD->vlan_macip_lens = htole32(vlan_macip_lens); - - /* ADV DTYPE TUCMD */ - type_tucmd_mlhl |= E1000_ADVTXD_DCMD_DEXT | E1000_ADVTXD_DTYP_CTXT; - type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_L4T_TCP; - TXD->type_tucmd_mlhl = htole32(type_tucmd_mlhl); - - /* MSS L4LEN IDX */ - mss_l4len_idx |= (mp->m_pkthdr.tso_segsz << E1000_ADVTXD_MSS_SHIFT); - mss_l4len_idx |= (tcp_hlen << E1000_ADVTXD_L4LEN_SHIFT); - /* 82575 needs the queue index added */ - if (adapter->hw.mac.type == e1000_82575) - mss_l4len_idx |= txr->me << 4; - TXD->mss_l4len_idx = htole32(mss_l4len_idx); - - TXD->seqnum_seed = htole32(0); - - if (++ctxd == txr->num_desc) - ctxd = 0; - - txr->tx_avail--; - txr->next_avail_desc = ctxd; - *cmd_type_len |= E1000_ADVTXD_DCMD_TSE; - *olinfo_status |= E1000_TXD_POPTS_TXSM << 8; - *olinfo_status |= paylen << E1000_ADVTXD_PAYLEN_SHIFT; - ++txr->tso_tx; - return (0); -} - -/********************************************************************* - * - * Advanced Context Descriptor setup for VLAN, CSUM or TSO - * - **********************************************************************/ - -static int -igb_tx_ctx_setup(struct tx_ring *txr, struct mbuf *mp, - u32 *cmd_type_len, u32 *olinfo_status) -{ - struct e1000_adv_tx_context_desc *TXD; - struct adapter *adapter = txr->adapter; - struct ether_vlan_header *eh; - struct ip *ip; - struct ip6_hdr *ip6; - u32 vlan_macip_lens = 0, type_tucmd_mlhl = 0, mss_l4len_idx = 0; - int ehdrlen, ip_hlen = 0; - u16 etype; - u8 ipproto = 0; - int offload = TRUE; - int ctxd = txr->next_avail_desc; - u16 vtag = 0; - - /* First check if TSO is to be used */ - if (mp->m_pkthdr.csum_flags & CSUM_TSO) - return (igb_tso_setup(txr, mp, cmd_type_len, olinfo_status)); - - if ((mp->m_pkthdr.csum_flags & CSUM_OFFLOAD) == 0) - offload = FALSE; - - /* Indicate the whole packet as payload when not doing TSO */ - *olinfo_status |= mp->m_pkthdr.len << E1000_ADVTXD_PAYLEN_SHIFT; - - /* Now ready a context descriptor */ - TXD = (struct e1000_adv_tx_context_desc *) &txr->tx_base[ctxd]; - - /* - ** In advanced descriptors the vlan tag must - ** be placed into the context descriptor. Hence - ** we need to make one even if not doing offloads. - */ - if (mp->m_flags & M_VLANTAG) { - vtag = htole16(mp->m_pkthdr.ether_vtag); - vlan_macip_lens |= (vtag << E1000_ADVTXD_VLAN_SHIFT); - } else if (offload == FALSE) /* ... no offload to do */ - return (0); - - /* - * Determine where frame payload starts. - * Jump over vlan headers if already present, - * helpful for QinQ too. - */ - eh = mtod(mp, struct ether_vlan_header *); - if (eh->evl_encap_proto == htons(ETHERTYPE_VLAN)) { - etype = ntohs(eh->evl_proto); - ehdrlen = ETHER_HDR_LEN + ETHER_VLAN_ENCAP_LEN; - } else { - etype = ntohs(eh->evl_encap_proto); - ehdrlen = ETHER_HDR_LEN; - } - - /* Set the ether header length */ - vlan_macip_lens |= ehdrlen << E1000_ADVTXD_MACLEN_SHIFT; - - switch (etype) { - case ETHERTYPE_IP: - ip = (struct ip *)(mp->m_data + ehdrlen); - ip_hlen = ip->ip_hl << 2; - ipproto = ip->ip_p; - type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_IPV4; - break; - case ETHERTYPE_IPV6: - ip6 = (struct ip6_hdr *)(mp->m_data + ehdrlen); - ip_hlen = sizeof(struct ip6_hdr); - /* XXX-BZ this will go badly in case of ext hdrs. */ - ipproto = ip6->ip6_nxt; - type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_IPV6; - break; - default: - offload = FALSE; - break; - } - - vlan_macip_lens |= ip_hlen; - type_tucmd_mlhl |= E1000_ADVTXD_DCMD_DEXT | E1000_ADVTXD_DTYP_CTXT; - - switch (ipproto) { - case IPPROTO_TCP: -#if __FreeBSD_version >= 1000000 - if (mp->m_pkthdr.csum_flags & (CSUM_IP_TCP | CSUM_IP6_TCP)) -#else - if (mp->m_pkthdr.csum_flags & CSUM_TCP) -#endif - type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_L4T_TCP; - break; - case IPPROTO_UDP: -#if __FreeBSD_version >= 1000000 - if (mp->m_pkthdr.csum_flags & (CSUM_IP_UDP | CSUM_IP6_UDP)) -#else - if (mp->m_pkthdr.csum_flags & CSUM_UDP) -#endif - type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_L4T_UDP; - break; - -#if __FreeBSD_version >= 800000 - case IPPROTO_SCTP: -#if __FreeBSD_version >= 1000000 - if (mp->m_pkthdr.csum_flags & (CSUM_IP_SCTP | CSUM_IP6_SCTP)) -#else - if (mp->m_pkthdr.csum_flags & CSUM_SCTP) -#endif - type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_L4T_SCTP; - break; -#endif - default: - offload = FALSE; - break; - } - - if (offload) /* For the TX descriptor setup */ - *olinfo_status |= E1000_TXD_POPTS_TXSM << 8; - - /* 82575 needs the queue index added */ - if (adapter->hw.mac.type == e1000_82575) - mss_l4len_idx = txr->me << 4; - - /* Now copy bits into descriptor */ - TXD->vlan_macip_lens = htole32(vlan_macip_lens); - TXD->type_tucmd_mlhl = htole32(type_tucmd_mlhl); - TXD->seqnum_seed = htole32(0); - TXD->mss_l4len_idx = htole32(mss_l4len_idx); - - /* We've consumed the first desc, adjust counters */ - if (++ctxd == txr->num_desc) - ctxd = 0; - txr->next_avail_desc = ctxd; - --txr->tx_avail; - - return (0); -} - -/********************************************************************** - * - * Examine each tx_buffer in the used queue. If the hardware is done - * processing the packet then free associated resources. The - * tx_buffer is put back on the free queue. - * - * TRUE return means there's work in the ring to clean, FALSE its empty. - **********************************************************************/ -static bool -igb_txeof(struct tx_ring *txr) -{ - struct adapter *adapter = txr->adapter; -#ifdef DEV_NETMAP - struct ifnet *ifp = adapter->ifp; -#endif /* DEV_NETMAP */ - u32 work, processed = 0; - int limit = adapter->tx_process_limit; - struct igb_tx_buf *buf; - union e1000_adv_tx_desc *txd; - - mtx_assert(&txr->tx_mtx, MA_OWNED); - -#ifdef DEV_NETMAP - if (netmap_tx_irq(ifp, txr->me)) - return (FALSE); -#endif /* DEV_NETMAP */ - - if (txr->tx_avail == txr->num_desc) { - txr->queue_status = IGB_QUEUE_IDLE; - return FALSE; - } - - /* Get work starting point */ - work = txr->next_to_clean; - buf = &txr->tx_buffers[work]; - txd = &txr->tx_base[work]; - work -= txr->num_desc; /* The distance to ring end */ - bus_dmamap_sync(txr->txdma.dma_tag, txr->txdma.dma_map, - BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); - do { - union e1000_adv_tx_desc *eop = buf->eop; - if (eop == NULL) /* No work */ - break; - - if ((eop->wb.status & E1000_TXD_STAT_DD) == 0) - break; /* I/O not complete */ - - if (buf->m_head) { - txr->bytes += - buf->m_head->m_pkthdr.len; - bus_dmamap_sync(txr->txtag, - buf->map, - BUS_DMASYNC_POSTWRITE); - bus_dmamap_unload(txr->txtag, - buf->map); - m_freem(buf->m_head); - buf->m_head = NULL; - } - buf->eop = NULL; - ++txr->tx_avail; - - /* We clean the range if multi segment */ - while (txd != eop) { - ++txd; - ++buf; - ++work; - /* wrap the ring? */ - if (__predict_false(!work)) { - work -= txr->num_desc; - buf = txr->tx_buffers; - txd = txr->tx_base; - } - if (buf->m_head) { - txr->bytes += - buf->m_head->m_pkthdr.len; - bus_dmamap_sync(txr->txtag, - buf->map, - BUS_DMASYNC_POSTWRITE); - bus_dmamap_unload(txr->txtag, - buf->map); - m_freem(buf->m_head); - buf->m_head = NULL; - } - ++txr->tx_avail; - buf->eop = NULL; - - } - ++txr->packets; - ++processed; - txr->watchdog_time = ticks; - - /* Try the next packet */ - ++txd; - ++buf; - ++work; - /* reset with a wrap */ - if (__predict_false(!work)) { - work -= txr->num_desc; - buf = txr->tx_buffers; - txd = txr->tx_base; - } - prefetch(txd); - } while (__predict_true(--limit)); - - bus_dmamap_sync(txr->txdma.dma_tag, txr->txdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - - work += txr->num_desc; - txr->next_to_clean = work; - - /* - ** Watchdog calculation, we know there's - ** work outstanding or the first return - ** would have been taken, so none processed - ** for too long indicates a hang. - */ - if ((!processed) && ((ticks - txr->watchdog_time) > IGB_WATCHDOG)) - txr->queue_status |= IGB_QUEUE_HUNG; - - if (txr->tx_avail >= IGB_QUEUE_THRESHOLD) - txr->queue_status &= ~IGB_QUEUE_DEPLETED; - - if (txr->tx_avail == txr->num_desc) { - txr->queue_status = IGB_QUEUE_IDLE; - return (FALSE); - } - - return (TRUE); -} - -/********************************************************************* - * - * Refresh mbuf buffers for RX descriptor rings - * - now keeps its own state so discards due to resource - * exhaustion are unnecessary, if an mbuf cannot be obtained - * it just returns, keeping its placeholder, thus it can simply - * be recalled to try again. - * - **********************************************************************/ -static void -igb_refresh_mbufs(struct rx_ring *rxr, int limit) -{ - struct adapter *adapter = rxr->adapter; - bus_dma_segment_t hseg[1]; - bus_dma_segment_t pseg[1]; - struct igb_rx_buf *rxbuf; - struct mbuf *mh, *mp; - int i, j, nsegs, error; - bool refreshed = FALSE; - - i = j = rxr->next_to_refresh; - /* - ** Get one descriptor beyond - ** our work mark to control - ** the loop. - */ - if (++j == adapter->num_rx_desc) - j = 0; - - while (j != limit) { - rxbuf = &rxr->rx_buffers[i]; - /* No hdr mbuf used with header split off */ - if (rxr->hdr_split == FALSE) - goto no_split; - if (rxbuf->m_head == NULL) { - mh = m_gethdr(M_NOWAIT, MT_DATA); - if (mh == NULL) - goto update; - } else - mh = rxbuf->m_head; - - mh->m_pkthdr.len = mh->m_len = MHLEN; - mh->m_len = MHLEN; - mh->m_flags |= M_PKTHDR; - /* Get the memory mapping */ - error = bus_dmamap_load_mbuf_sg(rxr->htag, - rxbuf->hmap, mh, hseg, &nsegs, BUS_DMA_NOWAIT); - if (error != 0) { - printf("Refresh mbufs: hdr dmamap load" - " failure - %d\n", error); - m_free(mh); - rxbuf->m_head = NULL; - goto update; - } - rxbuf->m_head = mh; - bus_dmamap_sync(rxr->htag, rxbuf->hmap, - BUS_DMASYNC_PREREAD); - rxr->rx_base[i].read.hdr_addr = - htole64(hseg[0].ds_addr); -no_split: - if (rxbuf->m_pack == NULL) { - mp = m_getjcl(M_NOWAIT, MT_DATA, - M_PKTHDR, adapter->rx_mbuf_sz); - if (mp == NULL) - goto update; - } else - mp = rxbuf->m_pack; - - mp->m_pkthdr.len = mp->m_len = adapter->rx_mbuf_sz; - /* Get the memory mapping */ - error = bus_dmamap_load_mbuf_sg(rxr->ptag, - rxbuf->pmap, mp, pseg, &nsegs, BUS_DMA_NOWAIT); - if (error != 0) { - printf("Refresh mbufs: payload dmamap load" - " failure - %d\n", error); - m_free(mp); - rxbuf->m_pack = NULL; - goto update; - } - rxbuf->m_pack = mp; - bus_dmamap_sync(rxr->ptag, rxbuf->pmap, - BUS_DMASYNC_PREREAD); - rxr->rx_base[i].read.pkt_addr = - htole64(pseg[0].ds_addr); - refreshed = TRUE; /* I feel wefreshed :) */ - - i = j; /* our next is precalculated */ - rxr->next_to_refresh = i; - if (++j == adapter->num_rx_desc) - j = 0; - } -update: - if (refreshed) /* update tail */ - E1000_WRITE_REG(&adapter->hw, - E1000_RDT(rxr->me), rxr->next_to_refresh); - return; -} - - -/********************************************************************* - * - * Allocate memory for rx_buffer structures. Since we use one - * rx_buffer per received packet, the maximum number of rx_buffer's - * that we'll need is equal to the number of receive descriptors - * that we've allocated. - * - **********************************************************************/ -static int -igb_allocate_receive_buffers(struct rx_ring *rxr) -{ - struct adapter *adapter = rxr->adapter; - device_t dev = adapter->dev; - struct igb_rx_buf *rxbuf; - int i, bsize, error; - - bsize = sizeof(struct igb_rx_buf) * adapter->num_rx_desc; - if (!(rxr->rx_buffers = - (struct igb_rx_buf *) malloc(bsize, - M_DEVBUF, M_NOWAIT | M_ZERO))) { - device_printf(dev, "Unable to allocate rx_buffer memory\n"); - error = ENOMEM; - goto fail; - } - - if ((error = bus_dma_tag_create(bus_get_dma_tag(dev), - 1, 0, /* alignment, bounds */ - BUS_SPACE_MAXADDR, /* lowaddr */ - BUS_SPACE_MAXADDR, /* highaddr */ - NULL, NULL, /* filter, filterarg */ - MSIZE, /* maxsize */ - 1, /* nsegments */ - MSIZE, /* maxsegsize */ - 0, /* flags */ - NULL, /* lockfunc */ - NULL, /* lockfuncarg */ - &rxr->htag))) { - device_printf(dev, "Unable to create RX DMA tag\n"); - goto fail; - } - - if ((error = bus_dma_tag_create(bus_get_dma_tag(dev), - 1, 0, /* alignment, bounds */ - BUS_SPACE_MAXADDR, /* lowaddr */ - BUS_SPACE_MAXADDR, /* highaddr */ - NULL, NULL, /* filter, filterarg */ - MJUM9BYTES, /* maxsize */ - 1, /* nsegments */ - MJUM9BYTES, /* maxsegsize */ - 0, /* flags */ - NULL, /* lockfunc */ - NULL, /* lockfuncarg */ - &rxr->ptag))) { - device_printf(dev, "Unable to create RX payload DMA tag\n"); - goto fail; - } - - for (i = 0; i < adapter->num_rx_desc; i++) { - rxbuf = &rxr->rx_buffers[i]; - error = bus_dmamap_create(rxr->htag, 0, &rxbuf->hmap); - if (error) { - device_printf(dev, - "Unable to create RX head DMA maps\n"); - goto fail; - } - error = bus_dmamap_create(rxr->ptag, 0, &rxbuf->pmap); - if (error) { - device_printf(dev, - "Unable to create RX packet DMA maps\n"); - goto fail; - } - } - - return (0); - -fail: - /* Frees all, but can handle partial completion */ - igb_free_receive_structures(adapter); - return (error); -} - - -static void -igb_free_receive_ring(struct rx_ring *rxr) -{ - struct adapter *adapter = rxr->adapter; - struct igb_rx_buf *rxbuf; - - - for (int i = 0; i < adapter->num_rx_desc; i++) { - rxbuf = &rxr->rx_buffers[i]; - if (rxbuf->m_head != NULL) { - bus_dmamap_sync(rxr->htag, rxbuf->hmap, - BUS_DMASYNC_POSTREAD); - bus_dmamap_unload(rxr->htag, rxbuf->hmap); - rxbuf->m_head->m_flags |= M_PKTHDR; - m_freem(rxbuf->m_head); - } - if (rxbuf->m_pack != NULL) { - bus_dmamap_sync(rxr->ptag, rxbuf->pmap, - BUS_DMASYNC_POSTREAD); - bus_dmamap_unload(rxr->ptag, rxbuf->pmap); - rxbuf->m_pack->m_flags |= M_PKTHDR; - m_freem(rxbuf->m_pack); - } - rxbuf->m_head = NULL; - rxbuf->m_pack = NULL; - } -} - - -/********************************************************************* - * - * Initialize a receive ring and its buffers. - * - **********************************************************************/ -static int -igb_setup_receive_ring(struct rx_ring *rxr) -{ - struct adapter *adapter; - struct ifnet *ifp; - device_t dev; - struct igb_rx_buf *rxbuf; - bus_dma_segment_t pseg[1], hseg[1]; - struct lro_ctrl *lro = &rxr->lro; - int rsize, nsegs, error = 0; -#ifdef DEV_NETMAP - struct netmap_adapter *na = NA(rxr->adapter->ifp); - struct netmap_slot *slot; -#endif /* DEV_NETMAP */ - - adapter = rxr->adapter; - dev = adapter->dev; - ifp = adapter->ifp; - - /* Clear the ring contents */ - IGB_RX_LOCK(rxr); -#ifdef DEV_NETMAP - slot = netmap_reset(na, NR_RX, rxr->me, 0); -#endif /* DEV_NETMAP */ - rsize = roundup2(adapter->num_rx_desc * - sizeof(union e1000_adv_rx_desc), IGB_DBA_ALIGN); - bzero((void *)rxr->rx_base, rsize); - - /* - ** Free current RX buffer structures and their mbufs - */ - igb_free_receive_ring(rxr); - - /* Configure for header split? */ - if (igb_header_split) - rxr->hdr_split = TRUE; - - /* Now replenish the ring mbufs */ - for (int j = 0; j < adapter->num_rx_desc; ++j) { - struct mbuf *mh, *mp; - - rxbuf = &rxr->rx_buffers[j]; -#ifdef DEV_NETMAP - if (slot) { - /* slot sj is mapped to the j-th NIC-ring entry */ - int sj = netmap_idx_n2k(&na->rx_rings[rxr->me], j); - uint64_t paddr; - void *addr; - - addr = PNMB(na, slot + sj, &paddr); - netmap_load_map(na, rxr->ptag, rxbuf->pmap, addr); - /* Update descriptor */ - rxr->rx_base[j].read.pkt_addr = htole64(paddr); - continue; - } -#endif /* DEV_NETMAP */ - if (rxr->hdr_split == FALSE) - goto skip_head; - - /* First the header */ - rxbuf->m_head = m_gethdr(M_NOWAIT, MT_DATA); - if (rxbuf->m_head == NULL) { - error = ENOBUFS; - goto fail; - } - m_adj(rxbuf->m_head, ETHER_ALIGN); - mh = rxbuf->m_head; - mh->m_len = mh->m_pkthdr.len = MHLEN; - mh->m_flags |= M_PKTHDR; - /* Get the memory mapping */ - error = bus_dmamap_load_mbuf_sg(rxr->htag, - rxbuf->hmap, rxbuf->m_head, hseg, - &nsegs, BUS_DMA_NOWAIT); - if (error != 0) /* Nothing elegant to do here */ - goto fail; - bus_dmamap_sync(rxr->htag, - rxbuf->hmap, BUS_DMASYNC_PREREAD); - /* Update descriptor */ - rxr->rx_base[j].read.hdr_addr = htole64(hseg[0].ds_addr); - -skip_head: - /* Now the payload cluster */ - rxbuf->m_pack = m_getjcl(M_NOWAIT, MT_DATA, - M_PKTHDR, adapter->rx_mbuf_sz); - if (rxbuf->m_pack == NULL) { - error = ENOBUFS; - goto fail; - } - mp = rxbuf->m_pack; - mp->m_pkthdr.len = mp->m_len = adapter->rx_mbuf_sz; - /* Get the memory mapping */ - error = bus_dmamap_load_mbuf_sg(rxr->ptag, - rxbuf->pmap, mp, pseg, - &nsegs, BUS_DMA_NOWAIT); - if (error != 0) - goto fail; - bus_dmamap_sync(rxr->ptag, - rxbuf->pmap, BUS_DMASYNC_PREREAD); - /* Update descriptor */ - rxr->rx_base[j].read.pkt_addr = htole64(pseg[0].ds_addr); - } - - /* Setup our descriptor indices */ - rxr->next_to_check = 0; - rxr->next_to_refresh = adapter->num_rx_desc - 1; - rxr->lro_enabled = FALSE; - rxr->rx_split_packets = 0; - rxr->rx_bytes = 0; - - rxr->fmp = NULL; - rxr->lmp = NULL; - - bus_dmamap_sync(rxr->rxdma.dma_tag, rxr->rxdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - - /* - ** Now set up the LRO interface, we - ** also only do head split when LRO - ** is enabled, since so often they - ** are undesirable in similar setups. - */ - if (ifp->if_capenable & IFCAP_LRO) { - error = tcp_lro_init(lro); - if (error) { - device_printf(dev, "LRO Initialization failed!\n"); - goto fail; - } - INIT_DEBUGOUT("RX LRO Initialized\n"); - rxr->lro_enabled = TRUE; - lro->ifp = adapter->ifp; - } - - IGB_RX_UNLOCK(rxr); - return (0); - -fail: - igb_free_receive_ring(rxr); - IGB_RX_UNLOCK(rxr); - return (error); -} - - -/********************************************************************* - * - * Initialize all receive rings. - * - **********************************************************************/ -static int -igb_setup_receive_structures(struct adapter *adapter) -{ - struct rx_ring *rxr = adapter->rx_rings; - int i; - - for (i = 0; i < adapter->num_queues; i++, rxr++) - if (igb_setup_receive_ring(rxr)) - goto fail; - - return (0); -fail: - /* - * Free RX buffers allocated so far, we will only handle - * the rings that completed, the failing case will have - * cleaned up for itself. 'i' is the endpoint. - */ - for (int j = 0; j < i; ++j) { - rxr = &adapter->rx_rings[j]; - IGB_RX_LOCK(rxr); - igb_free_receive_ring(rxr); - IGB_RX_UNLOCK(rxr); - } - - return (ENOBUFS); -} - -/* - * Initialise the RSS mapping for NICs that support multiple transmit/ - * receive rings. - */ -static void -igb_initialise_rss_mapping(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - int i; - int queue_id; - u32 reta; - u32 rss_key[10], mrqc, shift = 0; - - /* XXX? */ - if (adapter->hw.mac.type == e1000_82575) - shift = 6; - - /* - * The redirection table controls which destination - * queue each bucket redirects traffic to. - * Each DWORD represents four queues, with the LSB - * being the first queue in the DWORD. - * - * This just allocates buckets to queues using round-robin - * allocation. - * - * NOTE: It Just Happens to line up with the default - * RSS allocation method. - */ - - /* Warning FM follows */ - reta = 0; - for (i = 0; i < 128; i++) { -#ifdef RSS - queue_id = rss_get_indirection_to_bucket(i); - /* - * If we have more queues than buckets, we'll - * end up mapping buckets to a subset of the - * queues. - * - * If we have more buckets than queues, we'll - * end up instead assigning multiple buckets - * to queues. - * - * Both are suboptimal, but we need to handle - * the case so we don't go out of bounds - * indexing arrays and such. - */ - queue_id = queue_id % adapter->num_queues; -#else - queue_id = (i % adapter->num_queues); -#endif - /* Adjust if required */ - queue_id = queue_id << shift; - - /* - * The low 8 bits are for hash value (n+0); - * The next 8 bits are for hash value (n+1), etc. - */ - reta = reta >> 8; - reta = reta | ( ((uint32_t) queue_id) << 24); - if ((i & 3) == 3) { - E1000_WRITE_REG(hw, E1000_RETA(i >> 2), reta); - reta = 0; - } - } - - /* Now fill in hash table */ - - /* - * MRQC: Multiple Receive Queues Command - * Set queuing to RSS control, number depends on the device. - */ - mrqc = E1000_MRQC_ENABLE_RSS_8Q; - -#ifdef RSS - /* XXX ew typecasting */ - rss_getkey((uint8_t *) &rss_key); -#else - arc4rand(&rss_key, sizeof(rss_key), 0); -#endif - for (i = 0; i < 10; i++) - E1000_WRITE_REG_ARRAY(hw, - E1000_RSSRK(0), i, rss_key[i]); - - /* - * Configure the RSS fields to hash upon. - */ - mrqc |= (E1000_MRQC_RSS_FIELD_IPV4 | - E1000_MRQC_RSS_FIELD_IPV4_TCP); - mrqc |= (E1000_MRQC_RSS_FIELD_IPV6 | - E1000_MRQC_RSS_FIELD_IPV6_TCP); - mrqc |=( E1000_MRQC_RSS_FIELD_IPV4_UDP | - E1000_MRQC_RSS_FIELD_IPV6_UDP); - mrqc |=( E1000_MRQC_RSS_FIELD_IPV6_UDP_EX | - E1000_MRQC_RSS_FIELD_IPV6_TCP_EX); - - E1000_WRITE_REG(hw, E1000_MRQC, mrqc); -} - -/********************************************************************* - * - * Enable receive unit. - * - **********************************************************************/ -static void -igb_initialize_receive_units(struct adapter *adapter) -{ - struct rx_ring *rxr = adapter->rx_rings; - struct ifnet *ifp = adapter->ifp; - struct e1000_hw *hw = &adapter->hw; - u32 rctl, rxcsum, psize, srrctl = 0; - - INIT_DEBUGOUT("igb_initialize_receive_unit: begin"); - - /* - * Make sure receives are disabled while setting - * up the descriptor ring - */ - rctl = E1000_READ_REG(hw, E1000_RCTL); - E1000_WRITE_REG(hw, E1000_RCTL, rctl & ~E1000_RCTL_EN); - - /* - ** Set up for header split - */ - if (igb_header_split) { - /* Use a standard mbuf for the header */ - srrctl |= IGB_HDR_BUF << E1000_SRRCTL_BSIZEHDRSIZE_SHIFT; - srrctl |= E1000_SRRCTL_DESCTYPE_HDR_SPLIT_ALWAYS; - } else - srrctl |= E1000_SRRCTL_DESCTYPE_ADV_ONEBUF; - - /* - ** Set up for jumbo frames - */ - if (ifp->if_mtu > ETHERMTU) { - rctl |= E1000_RCTL_LPE; - if (adapter->rx_mbuf_sz == MJUMPAGESIZE) { - srrctl |= 4096 >> E1000_SRRCTL_BSIZEPKT_SHIFT; - rctl |= E1000_RCTL_SZ_4096 | E1000_RCTL_BSEX; - } else if (adapter->rx_mbuf_sz > MJUMPAGESIZE) { - srrctl |= 8192 >> E1000_SRRCTL_BSIZEPKT_SHIFT; - rctl |= E1000_RCTL_SZ_8192 | E1000_RCTL_BSEX; - } - /* Set maximum packet len */ - psize = adapter->max_frame_size; - /* are we on a vlan? */ - if (adapter->ifp->if_vlantrunk != NULL) - psize += VLAN_TAG_SIZE; - E1000_WRITE_REG(&adapter->hw, E1000_RLPML, psize); - } else { - rctl &= ~E1000_RCTL_LPE; - srrctl |= 2048 >> E1000_SRRCTL_BSIZEPKT_SHIFT; - rctl |= E1000_RCTL_SZ_2048; - } - - /* - * If TX flow control is disabled and there's >1 queue defined, - * enable DROP. - * - * This drops frames rather than hanging the RX MAC for all queues. - */ - if ((adapter->num_queues > 1) && - (adapter->fc == e1000_fc_none || - adapter->fc == e1000_fc_rx_pause)) { - srrctl |= E1000_SRRCTL_DROP_EN; - } - - /* Setup the Base and Length of the Rx Descriptor Rings */ - for (int i = 0; i < adapter->num_queues; i++, rxr++) { - u64 bus_addr = rxr->rxdma.dma_paddr; - u32 rxdctl; - - E1000_WRITE_REG(hw, E1000_RDLEN(i), - adapter->num_rx_desc * sizeof(struct e1000_rx_desc)); - E1000_WRITE_REG(hw, E1000_RDBAH(i), - (uint32_t)(bus_addr >> 32)); - E1000_WRITE_REG(hw, E1000_RDBAL(i), - (uint32_t)bus_addr); - E1000_WRITE_REG(hw, E1000_SRRCTL(i), srrctl); - /* Enable this Queue */ - rxdctl = E1000_READ_REG(hw, E1000_RXDCTL(i)); - rxdctl |= E1000_RXDCTL_QUEUE_ENABLE; - rxdctl &= 0xFFF00000; - rxdctl |= IGB_RX_PTHRESH; - rxdctl |= IGB_RX_HTHRESH << 8; - rxdctl |= IGB_RX_WTHRESH << 16; - E1000_WRITE_REG(hw, E1000_RXDCTL(i), rxdctl); - } - - /* - ** Setup for RX MultiQueue - */ - rxcsum = E1000_READ_REG(hw, E1000_RXCSUM); - if (adapter->num_queues >1) { - - /* rss setup */ - igb_initialise_rss_mapping(adapter); - - /* - ** NOTE: Receive Full-Packet Checksum Offload - ** is mutually exclusive with Multiqueue. However - ** this is not the same as TCP/IP checksums which - ** still work. - */ - rxcsum |= E1000_RXCSUM_PCSD; -#if __FreeBSD_version >= 800000 - /* For SCTP Offload */ - if ((hw->mac.type != e1000_82575) && - (ifp->if_capenable & IFCAP_RXCSUM)) - rxcsum |= E1000_RXCSUM_CRCOFL; -#endif - } else { - /* Non RSS setup */ - if (ifp->if_capenable & IFCAP_RXCSUM) { - rxcsum |= E1000_RXCSUM_IPPCSE; -#if __FreeBSD_version >= 800000 - if (adapter->hw.mac.type != e1000_82575) - rxcsum |= E1000_RXCSUM_CRCOFL; -#endif - } else - rxcsum &= ~E1000_RXCSUM_TUOFL; - } - E1000_WRITE_REG(hw, E1000_RXCSUM, rxcsum); - - /* Setup the Receive Control Register */ - rctl &= ~(3 << E1000_RCTL_MO_SHIFT); - rctl |= E1000_RCTL_EN | E1000_RCTL_BAM | E1000_RCTL_LBM_NO | - E1000_RCTL_RDMTS_HALF | - (hw->mac.mc_filter_type << E1000_RCTL_MO_SHIFT); - /* Strip CRC bytes. */ - rctl |= E1000_RCTL_SECRC; - /* Make sure VLAN Filters are off */ - rctl &= ~E1000_RCTL_VFE; - /* Don't store bad packets */ - rctl &= ~E1000_RCTL_SBP; - - /* Enable Receives */ - E1000_WRITE_REG(hw, E1000_RCTL, rctl); - - /* - * Setup the HW Rx Head and Tail Descriptor Pointers - * - needs to be after enable - */ - for (int i = 0; i < adapter->num_queues; i++) { - rxr = &adapter->rx_rings[i]; - E1000_WRITE_REG(hw, E1000_RDH(i), rxr->next_to_check); -#ifdef DEV_NETMAP - /* - * an init() while a netmap client is active must - * preserve the rx buffers passed to userspace. - * In this driver it means we adjust RDT to - * something different from next_to_refresh - * (which is not used in netmap mode). - */ - if (ifp->if_capenable & IFCAP_NETMAP) { - struct netmap_adapter *na = NA(adapter->ifp); - struct netmap_kring *kring = &na->rx_rings[i]; - int t = rxr->next_to_refresh - nm_kr_rxspace(kring); - - if (t >= adapter->num_rx_desc) - t -= adapter->num_rx_desc; - else if (t < 0) - t += adapter->num_rx_desc; - E1000_WRITE_REG(hw, E1000_RDT(i), t); - } else -#endif /* DEV_NETMAP */ - E1000_WRITE_REG(hw, E1000_RDT(i), rxr->next_to_refresh); - } - return; -} - -/********************************************************************* - * - * Free receive rings. - * - **********************************************************************/ -static void -igb_free_receive_structures(struct adapter *adapter) -{ - struct rx_ring *rxr = adapter->rx_rings; - - for (int i = 0; i < adapter->num_queues; i++, rxr++) { - struct lro_ctrl *lro = &rxr->lro; - igb_free_receive_buffers(rxr); - tcp_lro_free(lro); - igb_dma_free(adapter, &rxr->rxdma); - } - - free(adapter->rx_rings, M_DEVBUF); -} - -/********************************************************************* - * - * Free receive ring data structures. - * - **********************************************************************/ -static void -igb_free_receive_buffers(struct rx_ring *rxr) -{ - struct adapter *adapter = rxr->adapter; - struct igb_rx_buf *rxbuf; - int i; - - INIT_DEBUGOUT("free_receive_structures: begin"); - - /* Cleanup any existing buffers */ - if (rxr->rx_buffers != NULL) { - for (i = 0; i < adapter->num_rx_desc; i++) { - rxbuf = &rxr->rx_buffers[i]; - if (rxbuf->m_head != NULL) { - bus_dmamap_sync(rxr->htag, rxbuf->hmap, - BUS_DMASYNC_POSTREAD); - bus_dmamap_unload(rxr->htag, rxbuf->hmap); - rxbuf->m_head->m_flags |= M_PKTHDR; - m_freem(rxbuf->m_head); - } - if (rxbuf->m_pack != NULL) { - bus_dmamap_sync(rxr->ptag, rxbuf->pmap, - BUS_DMASYNC_POSTREAD); - bus_dmamap_unload(rxr->ptag, rxbuf->pmap); - rxbuf->m_pack->m_flags |= M_PKTHDR; - m_freem(rxbuf->m_pack); - } - rxbuf->m_head = NULL; - rxbuf->m_pack = NULL; - if (rxbuf->hmap != NULL) { - bus_dmamap_destroy(rxr->htag, rxbuf->hmap); - rxbuf->hmap = NULL; - } - if (rxbuf->pmap != NULL) { - bus_dmamap_destroy(rxr->ptag, rxbuf->pmap); - rxbuf->pmap = NULL; - } - } - if (rxr->rx_buffers != NULL) { - free(rxr->rx_buffers, M_DEVBUF); - rxr->rx_buffers = NULL; - } - } - - if (rxr->htag != NULL) { - bus_dma_tag_destroy(rxr->htag); - rxr->htag = NULL; - } - if (rxr->ptag != NULL) { - bus_dma_tag_destroy(rxr->ptag); - rxr->ptag = NULL; - } -} - -static __inline void -igb_rx_discard(struct rx_ring *rxr, int i) -{ - struct igb_rx_buf *rbuf; - - rbuf = &rxr->rx_buffers[i]; - - /* Partially received? Free the chain */ - if (rxr->fmp != NULL) { - rxr->fmp->m_flags |= M_PKTHDR; - m_freem(rxr->fmp); - rxr->fmp = NULL; - rxr->lmp = NULL; - } - - /* - ** With advanced descriptors the writeback - ** clobbers the buffer addrs, so its easier - ** to just free the existing mbufs and take - ** the normal refresh path to get new buffers - ** and mapping. - */ - if (rbuf->m_head) { - m_free(rbuf->m_head); - rbuf->m_head = NULL; - bus_dmamap_unload(rxr->htag, rbuf->hmap); - } - - if (rbuf->m_pack) { - m_free(rbuf->m_pack); - rbuf->m_pack = NULL; - bus_dmamap_unload(rxr->ptag, rbuf->pmap); - } - - return; -} - -static __inline void -igb_rx_input(struct rx_ring *rxr, struct ifnet *ifp, struct mbuf *m, u32 ptype) -{ - - /* - * ATM LRO is only for IPv4/TCP packets and TCP checksum of the packet - * should be computed by hardware. Also it should not have VLAN tag in - * ethernet header. - */ - if (rxr->lro_enabled && - (ifp->if_capenable & IFCAP_VLAN_HWTAGGING) != 0 && - (ptype & E1000_RXDADV_PKTTYPE_ETQF) == 0 && - (ptype & (E1000_RXDADV_PKTTYPE_IPV4 | E1000_RXDADV_PKTTYPE_TCP)) == - (E1000_RXDADV_PKTTYPE_IPV4 | E1000_RXDADV_PKTTYPE_TCP) && - (m->m_pkthdr.csum_flags & (CSUM_DATA_VALID | CSUM_PSEUDO_HDR)) == - (CSUM_DATA_VALID | CSUM_PSEUDO_HDR)) { - /* - * Send to the stack if: - ** - LRO not enabled, or - ** - no LRO resources, or - ** - lro enqueue fails - */ - if (rxr->lro.lro_cnt != 0) - if (tcp_lro_rx(&rxr->lro, m, 0) == 0) - return; - } - IGB_RX_UNLOCK(rxr); - (*ifp->if_input)(ifp, m); - IGB_RX_LOCK(rxr); -} - -/********************************************************************* - * - * This routine executes in interrupt context. It replenishes - * the mbufs in the descriptor and sends data which has been - * dma'ed into host memory to upper layer. - * - * We loop at most count times if count is > 0, or until done if - * count < 0. - * - * Return TRUE if more to clean, FALSE otherwise - *********************************************************************/ -static bool -igb_rxeof(struct igb_queue *que, int count, int *done) -{ - struct adapter *adapter = que->adapter; - struct rx_ring *rxr = que->rxr; - struct ifnet *ifp = adapter->ifp; - struct lro_ctrl *lro = &rxr->lro; - int i, processed = 0, rxdone = 0; - u32 ptype, staterr = 0; - union e1000_adv_rx_desc *cur; - - IGB_RX_LOCK(rxr); - /* Sync the ring. */ - bus_dmamap_sync(rxr->rxdma.dma_tag, rxr->rxdma.dma_map, - BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); - -#ifdef DEV_NETMAP - if (netmap_rx_irq(ifp, rxr->me, &processed)) { - IGB_RX_UNLOCK(rxr); - return (FALSE); - } -#endif /* DEV_NETMAP */ - - /* Main clean loop */ - for (i = rxr->next_to_check; count != 0;) { - struct mbuf *sendmp, *mh, *mp; - struct igb_rx_buf *rxbuf; - u16 hlen, plen, hdr, vtag, pkt_info; - bool eop = FALSE; - - cur = &rxr->rx_base[i]; - staterr = le32toh(cur->wb.upper.status_error); - if ((staterr & E1000_RXD_STAT_DD) == 0) - break; - if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) - break; - count--; - sendmp = mh = mp = NULL; - cur->wb.upper.status_error = 0; - rxbuf = &rxr->rx_buffers[i]; - plen = le16toh(cur->wb.upper.length); - ptype = le32toh(cur->wb.lower.lo_dword.data) & IGB_PKTTYPE_MASK; - if (((adapter->hw.mac.type == e1000_i350) || - (adapter->hw.mac.type == e1000_i354)) && - (staterr & E1000_RXDEXT_STATERR_LB)) - vtag = be16toh(cur->wb.upper.vlan); - else - vtag = le16toh(cur->wb.upper.vlan); - hdr = le16toh(cur->wb.lower.lo_dword.hs_rss.hdr_info); - pkt_info = le16toh(cur->wb.lower.lo_dword.hs_rss.pkt_info); - eop = ((staterr & E1000_RXD_STAT_EOP) == E1000_RXD_STAT_EOP); - - /* - * Free the frame (all segments) if we're at EOP and - * it's an error. - * - * The datasheet states that EOP + status is only valid for - * the final segment in a multi-segment frame. - */ - if (eop && ((staterr & E1000_RXDEXT_ERR_FRAME_ERR_MASK) != 0)) { - adapter->dropped_pkts++; - ++rxr->rx_discarded; - igb_rx_discard(rxr, i); - goto next_desc; - } - - /* - ** The way the hardware is configured to - ** split, it will ONLY use the header buffer - ** when header split is enabled, otherwise we - ** get normal behavior, ie, both header and - ** payload are DMA'd into the payload buffer. - ** - ** The fmp test is to catch the case where a - ** packet spans multiple descriptors, in that - ** case only the first header is valid. - */ - if (rxr->hdr_split && rxr->fmp == NULL) { - bus_dmamap_unload(rxr->htag, rxbuf->hmap); - hlen = (hdr & E1000_RXDADV_HDRBUFLEN_MASK) >> - E1000_RXDADV_HDRBUFLEN_SHIFT; - if (hlen > IGB_HDR_BUF) - hlen = IGB_HDR_BUF; - mh = rxr->rx_buffers[i].m_head; - mh->m_len = hlen; - /* clear buf pointer for refresh */ - rxbuf->m_head = NULL; - /* - ** Get the payload length, this - ** could be zero if its a small - ** packet. - */ - if (plen > 0) { - mp = rxr->rx_buffers[i].m_pack; - mp->m_len = plen; - mh->m_next = mp; - /* clear buf pointer */ - rxbuf->m_pack = NULL; - rxr->rx_split_packets++; - } - } else { - /* - ** Either no header split, or a - ** secondary piece of a fragmented - ** split packet. - */ - mh = rxr->rx_buffers[i].m_pack; - mh->m_len = plen; - /* clear buf info for refresh */ - rxbuf->m_pack = NULL; - } - bus_dmamap_unload(rxr->ptag, rxbuf->pmap); - - ++processed; /* So we know when to refresh */ - - /* Initial frame - setup */ - if (rxr->fmp == NULL) { - mh->m_pkthdr.len = mh->m_len; - /* Save the head of the chain */ - rxr->fmp = mh; - rxr->lmp = mh; - if (mp != NULL) { - /* Add payload if split */ - mh->m_pkthdr.len += mp->m_len; - rxr->lmp = mh->m_next; - } - } else { - /* Chain mbuf's together */ - rxr->lmp->m_next = mh; - rxr->lmp = rxr->lmp->m_next; - rxr->fmp->m_pkthdr.len += mh->m_len; - } - - if (eop) { - rxr->fmp->m_pkthdr.rcvif = ifp; - rxr->rx_packets++; - /* capture data for AIM */ - rxr->packets++; - rxr->bytes += rxr->fmp->m_pkthdr.len; - rxr->rx_bytes += rxr->fmp->m_pkthdr.len; - - if ((ifp->if_capenable & IFCAP_RXCSUM) != 0) - igb_rx_checksum(staterr, rxr->fmp, ptype); - - if ((ifp->if_capenable & IFCAP_VLAN_HWTAGGING) != 0 && - (staterr & E1000_RXD_STAT_VP) != 0) { - rxr->fmp->m_pkthdr.ether_vtag = vtag; - rxr->fmp->m_flags |= M_VLANTAG; - } - - /* - * In case of multiqueue, we have RXCSUM.PCSD bit set - * and never cleared. This means we have RSS hash - * available to be used. - */ - if (adapter->num_queues > 1) { - rxr->fmp->m_pkthdr.flowid = - le32toh(cur->wb.lower.hi_dword.rss); - switch (pkt_info & E1000_RXDADV_RSSTYPE_MASK) { - case E1000_RXDADV_RSSTYPE_IPV4_TCP: - M_HASHTYPE_SET(rxr->fmp, - M_HASHTYPE_RSS_TCP_IPV4); - break; - case E1000_RXDADV_RSSTYPE_IPV4: - M_HASHTYPE_SET(rxr->fmp, - M_HASHTYPE_RSS_IPV4); - break; - case E1000_RXDADV_RSSTYPE_IPV6_TCP: - M_HASHTYPE_SET(rxr->fmp, - M_HASHTYPE_RSS_TCP_IPV6); - break; - case E1000_RXDADV_RSSTYPE_IPV6_EX: - M_HASHTYPE_SET(rxr->fmp, - M_HASHTYPE_RSS_IPV6_EX); - break; - case E1000_RXDADV_RSSTYPE_IPV6: - M_HASHTYPE_SET(rxr->fmp, - M_HASHTYPE_RSS_IPV6); - break; - case E1000_RXDADV_RSSTYPE_IPV6_TCP_EX: - M_HASHTYPE_SET(rxr->fmp, - M_HASHTYPE_RSS_TCP_IPV6_EX); - break; - default: - /* XXX fallthrough */ - M_HASHTYPE_SET(rxr->fmp, - M_HASHTYPE_OPAQUE_HASH); - } - } else { -#ifndef IGB_LEGACY_TX - rxr->fmp->m_pkthdr.flowid = que->msix; - M_HASHTYPE_SET(rxr->fmp, M_HASHTYPE_OPAQUE); -#endif - } - sendmp = rxr->fmp; - /* Make sure to set M_PKTHDR. */ - sendmp->m_flags |= M_PKTHDR; - rxr->fmp = NULL; - rxr->lmp = NULL; - } - -next_desc: - bus_dmamap_sync(rxr->rxdma.dma_tag, rxr->rxdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - - /* Advance our pointers to the next descriptor. */ - if (++i == adapter->num_rx_desc) - i = 0; - /* - ** Send to the stack or LRO - */ - if (sendmp != NULL) { - rxr->next_to_check = i; - igb_rx_input(rxr, ifp, sendmp, ptype); - i = rxr->next_to_check; - rxdone++; - } - - /* Every 8 descriptors we go to refresh mbufs */ - if (processed == 8) { - igb_refresh_mbufs(rxr, i); - processed = 0; - } - } - - /* Catch any remainders */ - if (igb_rx_unrefreshed(rxr)) - igb_refresh_mbufs(rxr, i); - - rxr->next_to_check = i; - - /* - * Flush any outstanding LRO work - */ - tcp_lro_flush_all(lro); - - if (done != NULL) - *done += rxdone; - - IGB_RX_UNLOCK(rxr); - return ((staterr & E1000_RXD_STAT_DD) ? TRUE : FALSE); -} - -/********************************************************************* - * - * Verify that the hardware indicated that the checksum is valid. - * Inform the stack about the status of checksum so that stack - * doesn't spend time verifying the checksum. - * - *********************************************************************/ -static void -igb_rx_checksum(u32 staterr, struct mbuf *mp, u32 ptype) -{ - u16 status = (u16)staterr; - u8 errors = (u8) (staterr >> 24); - int sctp; - - /* Ignore Checksum bit is set */ - if (status & E1000_RXD_STAT_IXSM) { - mp->m_pkthdr.csum_flags = 0; - return; - } - - if ((ptype & E1000_RXDADV_PKTTYPE_ETQF) == 0 && - (ptype & E1000_RXDADV_PKTTYPE_SCTP) != 0) - sctp = 1; - else - sctp = 0; - if (status & E1000_RXD_STAT_IPCS) { - /* Did it pass? */ - if (!(errors & E1000_RXD_ERR_IPE)) { - /* IP Checksum Good */ - mp->m_pkthdr.csum_flags = CSUM_IP_CHECKED; - mp->m_pkthdr.csum_flags |= CSUM_IP_VALID; - } else - mp->m_pkthdr.csum_flags = 0; - } - - if (status & (E1000_RXD_STAT_TCPCS | E1000_RXD_STAT_UDPCS)) { - u64 type = (CSUM_DATA_VALID | CSUM_PSEUDO_HDR); -#if __FreeBSD_version >= 800000 - if (sctp) /* reassign */ - type = CSUM_SCTP_VALID; -#endif - /* Did it pass? */ - if (!(errors & E1000_RXD_ERR_TCPE)) { - mp->m_pkthdr.csum_flags |= type; - if (sctp == 0) - mp->m_pkthdr.csum_data = htons(0xffff); - } - } - return; -} - -/* - * This routine is run via an vlan - * config EVENT - */ -static void -igb_register_vlan(void *arg, struct ifnet *ifp, u16 vtag) -{ - struct adapter *adapter = ifp->if_softc; - u32 index, bit; - - if (ifp->if_softc != arg) /* Not our event */ - return; - - if ((vtag == 0) || (vtag > 4095)) /* Invalid */ - return; - - IGB_CORE_LOCK(adapter); - index = (vtag >> 5) & 0x7F; - bit = vtag & 0x1F; - adapter->shadow_vfta[index] |= (1 << bit); - ++adapter->num_vlans; - /* Change hw filter setting */ - if (ifp->if_capenable & IFCAP_VLAN_HWFILTER) - igb_setup_vlan_hw_support(adapter); - IGB_CORE_UNLOCK(adapter); -} - -/* - * This routine is run via an vlan - * unconfig EVENT - */ -static void -igb_unregister_vlan(void *arg, struct ifnet *ifp, u16 vtag) -{ - struct adapter *adapter = ifp->if_softc; - u32 index, bit; - - if (ifp->if_softc != arg) - return; - - if ((vtag == 0) || (vtag > 4095)) /* Invalid */ - return; - - IGB_CORE_LOCK(adapter); - index = (vtag >> 5) & 0x7F; - bit = vtag & 0x1F; - adapter->shadow_vfta[index] &= ~(1 << bit); - --adapter->num_vlans; - /* Change hw filter setting */ - if (ifp->if_capenable & IFCAP_VLAN_HWFILTER) - igb_setup_vlan_hw_support(adapter); - IGB_CORE_UNLOCK(adapter); -} - -static void -igb_setup_vlan_hw_support(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - struct ifnet *ifp = adapter->ifp; - u32 reg; - - if (adapter->vf_ifp) { - e1000_rlpml_set_vf(hw, - adapter->max_frame_size + VLAN_TAG_SIZE); - return; - } - - reg = E1000_READ_REG(hw, E1000_CTRL); - reg |= E1000_CTRL_VME; - E1000_WRITE_REG(hw, E1000_CTRL, reg); - - /* Enable the Filter Table */ - if (ifp->if_capenable & IFCAP_VLAN_HWFILTER) { - reg = E1000_READ_REG(hw, E1000_RCTL); - reg &= ~E1000_RCTL_CFIEN; - reg |= E1000_RCTL_VFE; - E1000_WRITE_REG(hw, E1000_RCTL, reg); - } - - /* Update the frame size */ - E1000_WRITE_REG(&adapter->hw, E1000_RLPML, - adapter->max_frame_size + VLAN_TAG_SIZE); - - /* Don't bother with table if no vlans */ - if ((adapter->num_vlans == 0) || - ((ifp->if_capenable & IFCAP_VLAN_HWFILTER) == 0)) - return; - /* - ** A soft reset zero's out the VFTA, so - ** we need to repopulate it now. - */ - for (int i = 0; i < IGB_VFTA_SIZE; i++) - if (adapter->shadow_vfta[i] != 0) { - if (adapter->vf_ifp) - e1000_vfta_set_vf(hw, - adapter->shadow_vfta[i], TRUE); - else - e1000_write_vfta(hw, - i, adapter->shadow_vfta[i]); - } -} - -static void -igb_enable_intr(struct adapter *adapter) -{ - /* With RSS set up what to auto clear */ - if (adapter->msix_mem) { - u32 mask = (adapter->que_mask | adapter->link_mask); - E1000_WRITE_REG(&adapter->hw, E1000_EIAC, mask); - E1000_WRITE_REG(&adapter->hw, E1000_EIAM, mask); - E1000_WRITE_REG(&adapter->hw, E1000_EIMS, mask); - E1000_WRITE_REG(&adapter->hw, E1000_IMS, - E1000_IMS_LSC); - } else { - E1000_WRITE_REG(&adapter->hw, E1000_IMS, - IMS_ENABLE_MASK); - } - E1000_WRITE_FLUSH(&adapter->hw); - - return; -} - -static void -igb_disable_intr(struct adapter *adapter) -{ - if (adapter->msix_mem) { - E1000_WRITE_REG(&adapter->hw, E1000_EIMC, ~0); - E1000_WRITE_REG(&adapter->hw, E1000_EIAC, 0); - } - E1000_WRITE_REG(&adapter->hw, E1000_IMC, ~0); - E1000_WRITE_FLUSH(&adapter->hw); - return; -} - -/* - * Bit of a misnomer, what this really means is - * to enable OS management of the system... aka - * to disable special hardware management features - */ -static void -igb_init_manageability(struct adapter *adapter) -{ - if (adapter->has_manage) { - int manc2h = E1000_READ_REG(&adapter->hw, E1000_MANC2H); - int manc = E1000_READ_REG(&adapter->hw, E1000_MANC); - - /* disable hardware interception of ARP */ - manc &= ~(E1000_MANC_ARP_EN); - - /* enable receiving management packets to the host */ - manc |= E1000_MANC_EN_MNG2HOST; - manc2h |= 1 << 5; /* Mng Port 623 */ - manc2h |= 1 << 6; /* Mng Port 664 */ - E1000_WRITE_REG(&adapter->hw, E1000_MANC2H, manc2h); - E1000_WRITE_REG(&adapter->hw, E1000_MANC, manc); - } -} - -/* - * Give control back to hardware management - * controller if there is one. - */ -static void -igb_release_manageability(struct adapter *adapter) -{ - if (adapter->has_manage) { - int manc = E1000_READ_REG(&adapter->hw, E1000_MANC); - - /* re-enable hardware interception of ARP */ - manc |= E1000_MANC_ARP_EN; - manc &= ~E1000_MANC_EN_MNG2HOST; - - E1000_WRITE_REG(&adapter->hw, E1000_MANC, manc); - } -} - -/* - * igb_get_hw_control sets CTRL_EXT:DRV_LOAD bit. - * For ASF and Pass Through versions of f/w this means that - * the driver is loaded. - * - */ -static void -igb_get_hw_control(struct adapter *adapter) -{ - u32 ctrl_ext; - - if (adapter->vf_ifp) - return; - - /* Let firmware know the driver has taken over */ - ctrl_ext = E1000_READ_REG(&adapter->hw, E1000_CTRL_EXT); - E1000_WRITE_REG(&adapter->hw, E1000_CTRL_EXT, - ctrl_ext | E1000_CTRL_EXT_DRV_LOAD); -} - -/* - * igb_release_hw_control resets CTRL_EXT:DRV_LOAD bit. - * For ASF and Pass Through versions of f/w this means that the - * driver is no longer loaded. - * - */ -static void -igb_release_hw_control(struct adapter *adapter) -{ - u32 ctrl_ext; - - if (adapter->vf_ifp) - return; - - /* Let firmware taken over control of h/w */ - ctrl_ext = E1000_READ_REG(&adapter->hw, E1000_CTRL_EXT); - E1000_WRITE_REG(&adapter->hw, E1000_CTRL_EXT, - ctrl_ext & ~E1000_CTRL_EXT_DRV_LOAD); -} - -static int -igb_is_valid_ether_addr(uint8_t *addr) -{ - char zero_addr[6] = { 0, 0, 0, 0, 0, 0 }; - - if ((addr[0] & 1) || (!bcmp(addr, zero_addr, ETHER_ADDR_LEN))) { - return (FALSE); - } - - return (TRUE); -} - - -/* - * Enable PCI Wake On Lan capability - */ -static void -igb_enable_wakeup(device_t dev) -{ - u16 cap, status; - u8 id; - - /* First find the capabilities pointer*/ - cap = pci_read_config(dev, PCIR_CAP_PTR, 2); - /* Read the PM Capabilities */ - id = pci_read_config(dev, cap, 1); - if (id != PCIY_PMG) /* Something wrong */ - return; - /* OK, we have the power capabilities, so - now get the status register */ - cap += PCIR_POWER_STATUS; - status = pci_read_config(dev, cap, 2); - status |= PCIM_PSTAT_PME | PCIM_PSTAT_PMEENABLE; - pci_write_config(dev, cap, status, 2); - return; -} - -static void -igb_led_func(void *arg, int onoff) -{ - struct adapter *adapter = arg; - - IGB_CORE_LOCK(adapter); - if (onoff) { - e1000_setup_led(&adapter->hw); - e1000_led_on(&adapter->hw); - } else { - e1000_led_off(&adapter->hw); - e1000_cleanup_led(&adapter->hw); - } - IGB_CORE_UNLOCK(adapter); -} - -static uint64_t -igb_get_vf_counter(if_t ifp, ift_counter cnt) -{ - struct adapter *adapter; - struct e1000_vf_stats *stats; -#ifndef IGB_LEGACY_TX - struct tx_ring *txr; - uint64_t rv; -#endif - - adapter = if_getsoftc(ifp); - stats = (struct e1000_vf_stats *)adapter->stats; - - switch (cnt) { - case IFCOUNTER_IPACKETS: - return (stats->gprc); - case IFCOUNTER_OPACKETS: - return (stats->gptc); - case IFCOUNTER_IBYTES: - return (stats->gorc); - case IFCOUNTER_OBYTES: - return (stats->gotc); - case IFCOUNTER_IMCASTS: - return (stats->mprc); - case IFCOUNTER_IERRORS: - return (adapter->dropped_pkts); - case IFCOUNTER_OERRORS: - return (adapter->watchdog_events); -#ifndef IGB_LEGACY_TX - case IFCOUNTER_OQDROPS: - rv = 0; - txr = adapter->tx_rings; - for (int i = 0; i < adapter->num_queues; i++, txr++) - rv += txr->br->br_drops; - return (rv); -#endif - default: - return (if_get_counter_default(ifp, cnt)); - } -} - -static uint64_t -igb_get_counter(if_t ifp, ift_counter cnt) -{ - struct adapter *adapter; - struct e1000_hw_stats *stats; -#ifndef IGB_LEGACY_TX - struct tx_ring *txr; - uint64_t rv; -#endif - - adapter = if_getsoftc(ifp); - if (adapter->vf_ifp) - return (igb_get_vf_counter(ifp, cnt)); - - stats = (struct e1000_hw_stats *)adapter->stats; - - switch (cnt) { - case IFCOUNTER_IPACKETS: - return (stats->gprc); - case IFCOUNTER_OPACKETS: - return (stats->gptc); - case IFCOUNTER_IBYTES: - return (stats->gorc); - case IFCOUNTER_OBYTES: - return (stats->gotc); - case IFCOUNTER_IMCASTS: - return (stats->mprc); - case IFCOUNTER_OMCASTS: - return (stats->mptc); - case IFCOUNTER_IERRORS: - return (adapter->dropped_pkts + stats->rxerrc + - stats->crcerrs + stats->algnerrc + - stats->ruc + stats->roc + stats->cexterr); - case IFCOUNTER_OERRORS: - return (stats->ecol + stats->latecol + - adapter->watchdog_events); - case IFCOUNTER_COLLISIONS: - return (stats->colc); - case IFCOUNTER_IQDROPS: - return (stats->mpc); -#ifndef IGB_LEGACY_TX - case IFCOUNTER_OQDROPS: - rv = 0; - txr = adapter->tx_rings; - for (int i = 0; i < adapter->num_queues; i++, txr++) - rv += txr->br->br_drops; - return (rv); -#endif - default: - return (if_get_counter_default(ifp, cnt)); - } -} - -/********************************************************************** - * - * Update the board statistics counters. - * - **********************************************************************/ -static void -igb_update_stats_counters(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - struct e1000_hw_stats *stats; - - /* - ** The virtual function adapter has only a - ** small controlled set of stats, do only - ** those and return. - */ - if (adapter->vf_ifp) { - igb_update_vf_stats_counters(adapter); - return; - } - - stats = (struct e1000_hw_stats *)adapter->stats; - - if (adapter->hw.phy.media_type == e1000_media_type_copper || - (E1000_READ_REG(hw, E1000_STATUS) & E1000_STATUS_LU)) { - stats->symerrs += - E1000_READ_REG(hw,E1000_SYMERRS); - stats->sec += E1000_READ_REG(hw, E1000_SEC); - } - - stats->crcerrs += E1000_READ_REG(hw, E1000_CRCERRS); - stats->mpc += E1000_READ_REG(hw, E1000_MPC); - stats->scc += E1000_READ_REG(hw, E1000_SCC); - stats->ecol += E1000_READ_REG(hw, E1000_ECOL); - - stats->mcc += E1000_READ_REG(hw, E1000_MCC); - stats->latecol += E1000_READ_REG(hw, E1000_LATECOL); - stats->colc += E1000_READ_REG(hw, E1000_COLC); - stats->dc += E1000_READ_REG(hw, E1000_DC); - stats->rlec += E1000_READ_REG(hw, E1000_RLEC); - stats->xonrxc += E1000_READ_REG(hw, E1000_XONRXC); - stats->xontxc += E1000_READ_REG(hw, E1000_XONTXC); - /* - ** For watchdog management we need to know if we have been - ** paused during the last interval, so capture that here. - */ - adapter->pause_frames = E1000_READ_REG(&adapter->hw, E1000_XOFFRXC); - stats->xoffrxc += adapter->pause_frames; - stats->xofftxc += E1000_READ_REG(hw, E1000_XOFFTXC); - stats->fcruc += E1000_READ_REG(hw, E1000_FCRUC); - stats->prc64 += E1000_READ_REG(hw, E1000_PRC64); - stats->prc127 += E1000_READ_REG(hw, E1000_PRC127); - stats->prc255 += E1000_READ_REG(hw, E1000_PRC255); - stats->prc511 += E1000_READ_REG(hw, E1000_PRC511); - stats->prc1023 += E1000_READ_REG(hw, E1000_PRC1023); - stats->prc1522 += E1000_READ_REG(hw, E1000_PRC1522); - stats->gprc += E1000_READ_REG(hw, E1000_GPRC); - stats->bprc += E1000_READ_REG(hw, E1000_BPRC); - stats->mprc += E1000_READ_REG(hw, E1000_MPRC); - stats->gptc += E1000_READ_REG(hw, E1000_GPTC); - - /* For the 64-bit byte counters the low dword must be read first. */ - /* Both registers clear on the read of the high dword */ - - stats->gorc += E1000_READ_REG(hw, E1000_GORCL) + - ((u64)E1000_READ_REG(hw, E1000_GORCH) << 32); - stats->gotc += E1000_READ_REG(hw, E1000_GOTCL) + - ((u64)E1000_READ_REG(hw, E1000_GOTCH) << 32); - - stats->rnbc += E1000_READ_REG(hw, E1000_RNBC); - stats->ruc += E1000_READ_REG(hw, E1000_RUC); - stats->rfc += E1000_READ_REG(hw, E1000_RFC); - stats->roc += E1000_READ_REG(hw, E1000_ROC); - stats->rjc += E1000_READ_REG(hw, E1000_RJC); - - stats->mgprc += E1000_READ_REG(hw, E1000_MGTPRC); - stats->mgpdc += E1000_READ_REG(hw, E1000_MGTPDC); - stats->mgptc += E1000_READ_REG(hw, E1000_MGTPTC); - - stats->tor += E1000_READ_REG(hw, E1000_TORL) + - ((u64)E1000_READ_REG(hw, E1000_TORH) << 32); - stats->tot += E1000_READ_REG(hw, E1000_TOTL) + - ((u64)E1000_READ_REG(hw, E1000_TOTH) << 32); - - stats->tpr += E1000_READ_REG(hw, E1000_TPR); - stats->tpt += E1000_READ_REG(hw, E1000_TPT); - stats->ptc64 += E1000_READ_REG(hw, E1000_PTC64); - stats->ptc127 += E1000_READ_REG(hw, E1000_PTC127); - stats->ptc255 += E1000_READ_REG(hw, E1000_PTC255); - stats->ptc511 += E1000_READ_REG(hw, E1000_PTC511); - stats->ptc1023 += E1000_READ_REG(hw, E1000_PTC1023); - stats->ptc1522 += E1000_READ_REG(hw, E1000_PTC1522); - stats->mptc += E1000_READ_REG(hw, E1000_MPTC); - stats->bptc += E1000_READ_REG(hw, E1000_BPTC); - - /* Interrupt Counts */ - - stats->iac += E1000_READ_REG(hw, E1000_IAC); - stats->icrxptc += E1000_READ_REG(hw, E1000_ICRXPTC); - stats->icrxatc += E1000_READ_REG(hw, E1000_ICRXATC); - stats->ictxptc += E1000_READ_REG(hw, E1000_ICTXPTC); - stats->ictxatc += E1000_READ_REG(hw, E1000_ICTXATC); - stats->ictxqec += E1000_READ_REG(hw, E1000_ICTXQEC); - stats->ictxqmtc += E1000_READ_REG(hw, E1000_ICTXQMTC); - stats->icrxdmtc += E1000_READ_REG(hw, E1000_ICRXDMTC); - stats->icrxoc += E1000_READ_REG(hw, E1000_ICRXOC); - - /* Host to Card Statistics */ - - stats->cbtmpc += E1000_READ_REG(hw, E1000_CBTMPC); - stats->htdpmc += E1000_READ_REG(hw, E1000_HTDPMC); - stats->cbrdpc += E1000_READ_REG(hw, E1000_CBRDPC); - stats->cbrmpc += E1000_READ_REG(hw, E1000_CBRMPC); - stats->rpthc += E1000_READ_REG(hw, E1000_RPTHC); - stats->hgptc += E1000_READ_REG(hw, E1000_HGPTC); - stats->htcbdpc += E1000_READ_REG(hw, E1000_HTCBDPC); - stats->hgorc += (E1000_READ_REG(hw, E1000_HGORCL) + - ((u64)E1000_READ_REG(hw, E1000_HGORCH) << 32)); - stats->hgotc += (E1000_READ_REG(hw, E1000_HGOTCL) + - ((u64)E1000_READ_REG(hw, E1000_HGOTCH) << 32)); - stats->lenerrs += E1000_READ_REG(hw, E1000_LENERRS); - stats->scvpc += E1000_READ_REG(hw, E1000_SCVPC); - stats->hrmpc += E1000_READ_REG(hw, E1000_HRMPC); - - stats->algnerrc += E1000_READ_REG(hw, E1000_ALGNERRC); - stats->rxerrc += E1000_READ_REG(hw, E1000_RXERRC); - stats->tncrs += E1000_READ_REG(hw, E1000_TNCRS); - stats->cexterr += E1000_READ_REG(hw, E1000_CEXTERR); - stats->tsctc += E1000_READ_REG(hw, E1000_TSCTC); - stats->tsctfc += E1000_READ_REG(hw, E1000_TSCTFC); - - /* Driver specific counters */ - adapter->device_control = E1000_READ_REG(hw, E1000_CTRL); - adapter->rx_control = E1000_READ_REG(hw, E1000_RCTL); - adapter->int_mask = E1000_READ_REG(hw, E1000_IMS); - adapter->eint_mask = E1000_READ_REG(hw, E1000_EIMS); - adapter->packet_buf_alloc_tx = - ((E1000_READ_REG(hw, E1000_PBA) & 0xffff0000) >> 16); - adapter->packet_buf_alloc_rx = - (E1000_READ_REG(hw, E1000_PBA) & 0xffff); -} - - -/********************************************************************** - * - * Initialize the VF board statistics counters. - * - **********************************************************************/ -static void -igb_vf_init_stats(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - struct e1000_vf_stats *stats; - - stats = (struct e1000_vf_stats *)adapter->stats; - if (stats == NULL) - return; - stats->last_gprc = E1000_READ_REG(hw, E1000_VFGPRC); - stats->last_gorc = E1000_READ_REG(hw, E1000_VFGORC); - stats->last_gptc = E1000_READ_REG(hw, E1000_VFGPTC); - stats->last_gotc = E1000_READ_REG(hw, E1000_VFGOTC); - stats->last_mprc = E1000_READ_REG(hw, E1000_VFMPRC); -} - -/********************************************************************** - * - * Update the VF board statistics counters. - * - **********************************************************************/ -static void -igb_update_vf_stats_counters(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - struct e1000_vf_stats *stats; - - if (adapter->link_speed == 0) - return; - - stats = (struct e1000_vf_stats *)adapter->stats; - - UPDATE_VF_REG(E1000_VFGPRC, - stats->last_gprc, stats->gprc); - UPDATE_VF_REG(E1000_VFGORC, - stats->last_gorc, stats->gorc); - UPDATE_VF_REG(E1000_VFGPTC, - stats->last_gptc, stats->gptc); - UPDATE_VF_REG(E1000_VFGOTC, - stats->last_gotc, stats->gotc); - UPDATE_VF_REG(E1000_VFMPRC, - stats->last_mprc, stats->mprc); -} - -/* Export a single 32-bit register via a read-only sysctl. */ -static int -igb_sysctl_reg_handler(SYSCTL_HANDLER_ARGS) -{ - struct adapter *adapter; - u_int val; - - adapter = oidp->oid_arg1; - val = E1000_READ_REG(&adapter->hw, oidp->oid_arg2); - return (sysctl_handle_int(oidp, &val, 0, req)); -} - -/* -** Tuneable interrupt rate handler -*/ -static int -igb_sysctl_interrupt_rate_handler(SYSCTL_HANDLER_ARGS) -{ - struct igb_queue *que = ((struct igb_queue *)oidp->oid_arg1); - int error; - u32 reg, usec, rate; - - reg = E1000_READ_REG(&que->adapter->hw, E1000_EITR(que->msix)); - usec = ((reg & 0x7FFC) >> 2); - if (usec > 0) - rate = 1000000 / usec; - else - rate = 0; - error = sysctl_handle_int(oidp, &rate, 0, req); - if (error || !req->newptr) - return error; - return 0; -} - -/* - * Add sysctl variables, one per statistic, to the system. - */ -static void -igb_add_hw_stats(struct adapter *adapter) -{ - device_t dev = adapter->dev; - - struct tx_ring *txr = adapter->tx_rings; - struct rx_ring *rxr = adapter->rx_rings; - - struct sysctl_ctx_list *ctx = device_get_sysctl_ctx(dev); - struct sysctl_oid *tree = device_get_sysctl_tree(dev); - struct sysctl_oid_list *child = SYSCTL_CHILDREN(tree); - struct e1000_hw_stats *stats = adapter->stats; - - struct sysctl_oid *stat_node, *queue_node, *int_node, *host_node; - struct sysctl_oid_list *stat_list, *queue_list, *int_list, *host_list; - -#define QUEUE_NAME_LEN 32 - char namebuf[QUEUE_NAME_LEN]; - - /* Driver Statistics */ - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "dropped", - CTLFLAG_RD, &adapter->dropped_pkts, - "Driver dropped packets"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "link_irq", - CTLFLAG_RD, &adapter->link_irq, - "Link MSIX IRQ Handled"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "mbuf_defrag_fail", - CTLFLAG_RD, &adapter->mbuf_defrag_failed, - "Defragmenting mbuf chain failed"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "tx_dma_fail", - CTLFLAG_RD, &adapter->no_tx_dma_setup, - "Driver tx dma failure in xmit"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "rx_overruns", - CTLFLAG_RD, &adapter->rx_overruns, - "RX overruns"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "watchdog_timeouts", - CTLFLAG_RD, &adapter->watchdog_events, - "Watchdog timeouts"); - - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "device_control", - CTLFLAG_RD, &adapter->device_control, - "Device Control Register"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "rx_control", - CTLFLAG_RD, &adapter->rx_control, - "Receiver Control Register"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "interrupt_mask", - CTLFLAG_RD, &adapter->int_mask, - "Interrupt Mask"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "extended_int_mask", - CTLFLAG_RD, &adapter->eint_mask, - "Extended Interrupt Mask"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "tx_buf_alloc", - CTLFLAG_RD, &adapter->packet_buf_alloc_tx, - "Transmit Buffer Packet Allocation"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "rx_buf_alloc", - CTLFLAG_RD, &adapter->packet_buf_alloc_rx, - "Receive Buffer Packet Allocation"); - SYSCTL_ADD_UINT(ctx, child, OID_AUTO, "fc_high_water", - CTLFLAG_RD, &adapter->hw.fc.high_water, 0, - "Flow Control High Watermark"); - SYSCTL_ADD_UINT(ctx, child, OID_AUTO, "fc_low_water", - CTLFLAG_RD, &adapter->hw.fc.low_water, 0, - "Flow Control Low Watermark"); - - for (int i = 0; i < adapter->num_queues; i++, rxr++, txr++) { - struct lro_ctrl *lro = &rxr->lro; - - snprintf(namebuf, QUEUE_NAME_LEN, "queue%d", i); - queue_node = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, namebuf, - CTLFLAG_RD, NULL, "Queue Name"); - queue_list = SYSCTL_CHILDREN(queue_node); - - SYSCTL_ADD_PROC(ctx, queue_list, OID_AUTO, "interrupt_rate", - CTLTYPE_UINT | CTLFLAG_RD, &adapter->queues[i], - sizeof(&adapter->queues[i]), - igb_sysctl_interrupt_rate_handler, - "IU", "Interrupt Rate"); - - SYSCTL_ADD_PROC(ctx, queue_list, OID_AUTO, "txd_head", - CTLTYPE_UINT | CTLFLAG_RD, adapter, E1000_TDH(txr->me), - igb_sysctl_reg_handler, "IU", - "Transmit Descriptor Head"); - SYSCTL_ADD_PROC(ctx, queue_list, OID_AUTO, "txd_tail", - CTLTYPE_UINT | CTLFLAG_RD, adapter, E1000_TDT(txr->me), - igb_sysctl_reg_handler, "IU", - "Transmit Descriptor Tail"); - SYSCTL_ADD_QUAD(ctx, queue_list, OID_AUTO, "no_desc_avail", - CTLFLAG_RD, &txr->no_desc_avail, - "Queue Descriptors Unavailable"); - SYSCTL_ADD_UQUAD(ctx, queue_list, OID_AUTO, "tx_packets", - CTLFLAG_RD, &txr->total_packets, - "Queue Packets Transmitted"); - - SYSCTL_ADD_PROC(ctx, queue_list, OID_AUTO, "rxd_head", - CTLTYPE_UINT | CTLFLAG_RD, adapter, E1000_RDH(rxr->me), - igb_sysctl_reg_handler, "IU", - "Receive Descriptor Head"); - SYSCTL_ADD_PROC(ctx, queue_list, OID_AUTO, "rxd_tail", - CTLTYPE_UINT | CTLFLAG_RD, adapter, E1000_RDT(rxr->me), - igb_sysctl_reg_handler, "IU", - "Receive Descriptor Tail"); - SYSCTL_ADD_QUAD(ctx, queue_list, OID_AUTO, "rx_packets", - CTLFLAG_RD, &rxr->rx_packets, - "Queue Packets Received"); - SYSCTL_ADD_QUAD(ctx, queue_list, OID_AUTO, "rx_bytes", - CTLFLAG_RD, &rxr->rx_bytes, - "Queue Bytes Received"); - SYSCTL_ADD_U64(ctx, queue_list, OID_AUTO, "lro_queued", - CTLFLAG_RD, &lro->lro_queued, 0, - "LRO Queued"); - SYSCTL_ADD_U64(ctx, queue_list, OID_AUTO, "lro_flushed", - CTLFLAG_RD, &lro->lro_flushed, 0, - "LRO Flushed"); - } - - /* MAC stats get their own sub node */ - - stat_node = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, "mac_stats", - CTLFLAG_RD, NULL, "MAC Statistics"); - stat_list = SYSCTL_CHILDREN(stat_node); - - /* - ** VF adapter has a very limited set of stats - ** since its not managing the metal, so to speak. - */ - if (adapter->vf_ifp) { - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "good_pkts_recvd", - CTLFLAG_RD, &stats->gprc, - "Good Packets Received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "good_pkts_txd", - CTLFLAG_RD, &stats->gptc, - "Good Packets Transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "good_octets_recvd", - CTLFLAG_RD, &stats->gorc, - "Good Octets Received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "good_octets_txd", - CTLFLAG_RD, &stats->gotc, - "Good Octets Transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "mcast_pkts_recvd", - CTLFLAG_RD, &stats->mprc, - "Multicast Packets Received"); - return; - } - - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "excess_coll", - CTLFLAG_RD, &stats->ecol, - "Excessive collisions"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "single_coll", - CTLFLAG_RD, &stats->scc, - "Single collisions"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "multiple_coll", - CTLFLAG_RD, &stats->mcc, - "Multiple collisions"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "late_coll", - CTLFLAG_RD, &stats->latecol, - "Late collisions"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "collision_count", - CTLFLAG_RD, &stats->colc, - "Collision Count"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "symbol_errors", - CTLFLAG_RD, &stats->symerrs, - "Symbol Errors"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "sequence_errors", - CTLFLAG_RD, &stats->sec, - "Sequence Errors"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "defer_count", - CTLFLAG_RD, &stats->dc, - "Defer Count"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "missed_packets", - CTLFLAG_RD, &stats->mpc, - "Missed Packets"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "recv_length_errors", - CTLFLAG_RD, &stats->rlec, - "Receive Length Errors"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "recv_no_buff", - CTLFLAG_RD, &stats->rnbc, - "Receive No Buffers"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "recv_undersize", - CTLFLAG_RD, &stats->ruc, - "Receive Undersize"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "recv_fragmented", - CTLFLAG_RD, &stats->rfc, - "Fragmented Packets Received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "recv_oversize", - CTLFLAG_RD, &stats->roc, - "Oversized Packets Received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "recv_jabber", - CTLFLAG_RD, &stats->rjc, - "Recevied Jabber"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "recv_errs", - CTLFLAG_RD, &stats->rxerrc, - "Receive Errors"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "crc_errs", - CTLFLAG_RD, &stats->crcerrs, - "CRC errors"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "alignment_errs", - CTLFLAG_RD, &stats->algnerrc, - "Alignment Errors"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "tx_no_crs", - CTLFLAG_RD, &stats->tncrs, - "Transmit with No CRS"); - /* On 82575 these are collision counts */ - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "coll_ext_errs", - CTLFLAG_RD, &stats->cexterr, - "Collision/Carrier extension errors"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "xon_recvd", - CTLFLAG_RD, &stats->xonrxc, - "XON Received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "xon_txd", - CTLFLAG_RD, &stats->xontxc, - "XON Transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "xoff_recvd", - CTLFLAG_RD, &stats->xoffrxc, - "XOFF Received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "xoff_txd", - CTLFLAG_RD, &stats->xofftxc, - "XOFF Transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "unsupported_fc_recvd", - CTLFLAG_RD, &stats->fcruc, - "Unsupported Flow Control Received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "mgmt_pkts_recvd", - CTLFLAG_RD, &stats->mgprc, - "Management Packets Received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "mgmt_pkts_drop", - CTLFLAG_RD, &stats->mgpdc, - "Management Packets Dropped"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "mgmt_pkts_txd", - CTLFLAG_RD, &stats->mgptc, - "Management Packets Transmitted"); - /* Packet Reception Stats */ - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "total_pkts_recvd", - CTLFLAG_RD, &stats->tpr, - "Total Packets Received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "good_pkts_recvd", - CTLFLAG_RD, &stats->gprc, - "Good Packets Received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "bcast_pkts_recvd", - CTLFLAG_RD, &stats->bprc, - "Broadcast Packets Received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "mcast_pkts_recvd", - CTLFLAG_RD, &stats->mprc, - "Multicast Packets Received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "rx_frames_64", - CTLFLAG_RD, &stats->prc64, - "64 byte frames received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "rx_frames_65_127", - CTLFLAG_RD, &stats->prc127, - "65-127 byte frames received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "rx_frames_128_255", - CTLFLAG_RD, &stats->prc255, - "128-255 byte frames received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "rx_frames_256_511", - CTLFLAG_RD, &stats->prc511, - "256-511 byte frames received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "rx_frames_512_1023", - CTLFLAG_RD, &stats->prc1023, - "512-1023 byte frames received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "rx_frames_1024_1522", - CTLFLAG_RD, &stats->prc1522, - "1023-1522 byte frames received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "good_octets_recvd", - CTLFLAG_RD, &stats->gorc, - "Good Octets Received"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "total_octets_recvd", - CTLFLAG_RD, &stats->tor, - "Total Octets Received"); - - /* Packet Transmission Stats */ - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "good_octets_txd", - CTLFLAG_RD, &stats->gotc, - "Good Octets Transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "total_octets_txd", - CTLFLAG_RD, &stats->tot, - "Total Octets Transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "total_pkts_txd", - CTLFLAG_RD, &stats->tpt, - "Total Packets Transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "good_pkts_txd", - CTLFLAG_RD, &stats->gptc, - "Good Packets Transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "bcast_pkts_txd", - CTLFLAG_RD, &stats->bptc, - "Broadcast Packets Transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "mcast_pkts_txd", - CTLFLAG_RD, &stats->mptc, - "Multicast Packets Transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "tx_frames_64", - CTLFLAG_RD, &stats->ptc64, - "64 byte frames transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "tx_frames_65_127", - CTLFLAG_RD, &stats->ptc127, - "65-127 byte frames transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "tx_frames_128_255", - CTLFLAG_RD, &stats->ptc255, - "128-255 byte frames transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "tx_frames_256_511", - CTLFLAG_RD, &stats->ptc511, - "256-511 byte frames transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "tx_frames_512_1023", - CTLFLAG_RD, &stats->ptc1023, - "512-1023 byte frames transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "tx_frames_1024_1522", - CTLFLAG_RD, &stats->ptc1522, - "1024-1522 byte frames transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "tso_txd", - CTLFLAG_RD, &stats->tsctc, - "TSO Contexts Transmitted"); - SYSCTL_ADD_QUAD(ctx, stat_list, OID_AUTO, "tso_ctx_fail", - CTLFLAG_RD, &stats->tsctfc, - "TSO Contexts Failed"); - - - /* Interrupt Stats */ - - int_node = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, "interrupts", - CTLFLAG_RD, NULL, "Interrupt Statistics"); - int_list = SYSCTL_CHILDREN(int_node); - - SYSCTL_ADD_QUAD(ctx, int_list, OID_AUTO, "asserts", - CTLFLAG_RD, &stats->iac, - "Interrupt Assertion Count"); - - SYSCTL_ADD_QUAD(ctx, int_list, OID_AUTO, "rx_pkt_timer", - CTLFLAG_RD, &stats->icrxptc, - "Interrupt Cause Rx Pkt Timer Expire Count"); - - SYSCTL_ADD_QUAD(ctx, int_list, OID_AUTO, "rx_abs_timer", - CTLFLAG_RD, &stats->icrxatc, - "Interrupt Cause Rx Abs Timer Expire Count"); - - SYSCTL_ADD_QUAD(ctx, int_list, OID_AUTO, "tx_pkt_timer", - CTLFLAG_RD, &stats->ictxptc, - "Interrupt Cause Tx Pkt Timer Expire Count"); - - SYSCTL_ADD_QUAD(ctx, int_list, OID_AUTO, "tx_abs_timer", - CTLFLAG_RD, &stats->ictxatc, - "Interrupt Cause Tx Abs Timer Expire Count"); - - SYSCTL_ADD_QUAD(ctx, int_list, OID_AUTO, "tx_queue_empty", - CTLFLAG_RD, &stats->ictxqec, - "Interrupt Cause Tx Queue Empty Count"); - - SYSCTL_ADD_QUAD(ctx, int_list, OID_AUTO, "tx_queue_min_thresh", - CTLFLAG_RD, &stats->ictxqmtc, - "Interrupt Cause Tx Queue Min Thresh Count"); - - SYSCTL_ADD_QUAD(ctx, int_list, OID_AUTO, "rx_desc_min_thresh", - CTLFLAG_RD, &stats->icrxdmtc, - "Interrupt Cause Rx Desc Min Thresh Count"); - - SYSCTL_ADD_QUAD(ctx, int_list, OID_AUTO, "rx_overrun", - CTLFLAG_RD, &stats->icrxoc, - "Interrupt Cause Receiver Overrun Count"); - - /* Host to Card Stats */ - - host_node = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, "host", - CTLFLAG_RD, NULL, - "Host to Card Statistics"); - - host_list = SYSCTL_CHILDREN(host_node); - - SYSCTL_ADD_QUAD(ctx, host_list, OID_AUTO, "breaker_tx_pkt", - CTLFLAG_RD, &stats->cbtmpc, - "Circuit Breaker Tx Packet Count"); - - SYSCTL_ADD_QUAD(ctx, host_list, OID_AUTO, "host_tx_pkt_discard", - CTLFLAG_RD, &stats->htdpmc, - "Host Transmit Discarded Packets"); - - SYSCTL_ADD_QUAD(ctx, host_list, OID_AUTO, "rx_pkt", - CTLFLAG_RD, &stats->rpthc, - "Rx Packets To Host"); - - SYSCTL_ADD_QUAD(ctx, host_list, OID_AUTO, "breaker_rx_pkts", - CTLFLAG_RD, &stats->cbrmpc, - "Circuit Breaker Rx Packet Count"); - - SYSCTL_ADD_QUAD(ctx, host_list, OID_AUTO, "breaker_rx_pkt_drop", - CTLFLAG_RD, &stats->cbrdpc, - "Circuit Breaker Rx Dropped Count"); - - SYSCTL_ADD_QUAD(ctx, host_list, OID_AUTO, "tx_good_pkt", - CTLFLAG_RD, &stats->hgptc, - "Host Good Packets Tx Count"); - - SYSCTL_ADD_QUAD(ctx, host_list, OID_AUTO, "breaker_tx_pkt_drop", - CTLFLAG_RD, &stats->htcbdpc, - "Host Tx Circuit Breaker Dropped Count"); - - SYSCTL_ADD_QUAD(ctx, host_list, OID_AUTO, "rx_good_bytes", - CTLFLAG_RD, &stats->hgorc, - "Host Good Octets Received Count"); - - SYSCTL_ADD_QUAD(ctx, host_list, OID_AUTO, "tx_good_bytes", - CTLFLAG_RD, &stats->hgotc, - "Host Good Octets Transmit Count"); - - SYSCTL_ADD_QUAD(ctx, host_list, OID_AUTO, "length_errors", - CTLFLAG_RD, &stats->lenerrs, - "Length Errors"); - - SYSCTL_ADD_QUAD(ctx, host_list, OID_AUTO, "serdes_violation_pkt", - CTLFLAG_RD, &stats->scvpc, - "SerDes/SGMII Code Violation Pkt Count"); - - SYSCTL_ADD_QUAD(ctx, host_list, OID_AUTO, "header_redir_missed", - CTLFLAG_RD, &stats->hrmpc, - "Header Redirection Missed Packet Count"); -} - - -/********************************************************************** - * - * This routine provides a way to dump out the adapter eeprom, - * often a useful debug/service tool. This only dumps the first - * 32 words, stuff that matters is in that extent. - * - **********************************************************************/ -static int -igb_sysctl_nvm_info(SYSCTL_HANDLER_ARGS) -{ - struct adapter *adapter; - int error; - int result; - - result = -1; - error = sysctl_handle_int(oidp, &result, 0, req); - - if (error || !req->newptr) - return (error); - - /* - * This value will cause a hex dump of the - * first 32 16-bit words of the EEPROM to - * the screen. - */ - if (result == 1) { - adapter = (struct adapter *)arg1; - igb_print_nvm_info(adapter); - } - - return (error); -} - -static void -igb_print_nvm_info(struct adapter *adapter) -{ - u16 eeprom_data; - int i, j, row = 0; - - /* Its a bit crude, but it gets the job done */ - printf("\nInterface EEPROM Dump:\n"); - printf("Offset\n0x0000 "); - for (i = 0, j = 0; i < 32; i++, j++) { - if (j == 8) { /* Make the offset block */ - j = 0; ++row; - printf("\n0x00%x0 ",row); - } - e1000_read_nvm(&adapter->hw, i, 1, &eeprom_data); - printf("%04x ", eeprom_data); - } - printf("\n"); -} - -static void -igb_set_sysctl_value(struct adapter *adapter, const char *name, - const char *description, int *limit, int value) -{ - *limit = value; - SYSCTL_ADD_INT(device_get_sysctl_ctx(adapter->dev), - SYSCTL_CHILDREN(device_get_sysctl_tree(adapter->dev)), - OID_AUTO, name, CTLFLAG_RW, limit, value, description); -} - -/* -** Set flow control using sysctl: -** Flow control values: -** 0 - off -** 1 - rx pause -** 2 - tx pause -** 3 - full -*/ -static int -igb_set_flowcntl(SYSCTL_HANDLER_ARGS) -{ - int error; - static int input = 3; /* default is full */ - struct adapter *adapter = (struct adapter *) arg1; - - error = sysctl_handle_int(oidp, &input, 0, req); - - if ((error) || (req->newptr == NULL)) - return (error); - - switch (input) { - case e1000_fc_rx_pause: - case e1000_fc_tx_pause: - case e1000_fc_full: - case e1000_fc_none: - adapter->hw.fc.requested_mode = input; - adapter->fc = input; - break; - default: - /* Do nothing */ - return (error); - } - - adapter->hw.fc.current_mode = adapter->hw.fc.requested_mode; - e1000_force_mac_fc(&adapter->hw); - /* XXX TODO: update DROP_EN on each RX queue if appropriate */ - return (error); -} - -/* -** Manage DMA Coalesce: -** Control values: -** 0/1 - off/on -** Legal timer values are: -** 250,500,1000-10000 in thousands -*/ -static int -igb_sysctl_dmac(SYSCTL_HANDLER_ARGS) -{ - struct adapter *adapter = (struct adapter *) arg1; - int error; - - error = sysctl_handle_int(oidp, &adapter->dmac, 0, req); - - if ((error) || (req->newptr == NULL)) - return (error); - - switch (adapter->dmac) { - case 0: - /* Disabling */ - break; - case 1: /* Just enable and use default */ - adapter->dmac = 1000; - break; - case 250: - case 500: - case 1000: - case 2000: - case 3000: - case 4000: - case 5000: - case 6000: - case 7000: - case 8000: - case 9000: - case 10000: - /* Legal values - allow */ - break; - default: - /* Do nothing, illegal value */ - adapter->dmac = 0; - return (EINVAL); - } - /* Reinit the interface */ - igb_init(adapter); - return (error); -} - -/* -** Manage Energy Efficient Ethernet: -** Control values: -** 0/1 - enabled/disabled -*/ -static int -igb_sysctl_eee(SYSCTL_HANDLER_ARGS) -{ - struct adapter *adapter = (struct adapter *) arg1; - int error, value; - - value = adapter->hw.dev_spec._82575.eee_disable; - error = sysctl_handle_int(oidp, &value, 0, req); - if (error || req->newptr == NULL) - return (error); - IGB_CORE_LOCK(adapter); - adapter->hw.dev_spec._82575.eee_disable = (value != 0); - igb_init_locked(adapter); - IGB_CORE_UNLOCK(adapter); - return (0); -} diff --git a/sys/dev/e1000/if_igb.h b/sys/dev/e1000/if_igb.h deleted file mode 100644 index e2f2219cc94c..000000000000 --- a/sys/dev/e1000/if_igb.h +++ /dev/null @@ -1,634 +0,0 @@ -/****************************************************************************** - - Copyright (c) 2001-2015, Intel Corporation - All rights reserved. - - Redistribution and use in source and binary forms, with or without - modification, are permitted provided that the following conditions are met: - - 1. Redistributions of source code must retain the above copyright notice, - this list of conditions and the following disclaimer. - - 2. Redistributions in binary form must reproduce the above copyright - notice, this list of conditions and the following disclaimer in the - documentation and/or other materials provided with the distribution. - - 3. Neither the name of the Intel Corporation nor the names of its - contributors may be used to endorse or promote products derived from - this software without specific prior written permission. - - THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" - AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE - LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR - CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF - SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS - INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN - CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) - ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE - POSSIBILITY OF SUCH DAMAGE. - -******************************************************************************/ -/*$FreeBSD$*/ - -#ifndef _IF_IGB_H_ -#define _IF_IGB_H_ - -#ifdef ALTQ -#define IGB_LEGACY_TX -#endif - -#include -#include -#ifndef IGB_LEGACY_TX -#include -#endif -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include -#include -#include -#include -#include -#ifdef RSS -#include -#include -#endif - -#include -#include - -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include -#include - -#include "e1000_api.h" -#include "e1000_82575.h" - -/* Tunables */ -/* - * IGB_TXD: Maximum number of Transmit Descriptors - * - * This value is the number of transmit descriptors allocated by the driver. - * Increasing this value allows the driver to queue more transmits. Each - * descriptor is 16 bytes. - * Since TDLEN should be multiple of 128bytes, the number of transmit - * desscriptors should meet the following condition. - * (num_tx_desc * sizeof(struct e1000_tx_desc)) % 128 == 0 - */ -#define IGB_MIN_TXD 256 -#define IGB_DEFAULT_TXD 1024 -#define IGB_MAX_TXD 4096 - -/* - * IGB_RXD: Maximum number of Receive Descriptors - * - * This value is the number of receive descriptors allocated by the driver. - * Increasing this value allows the driver to buffer more incoming packets. - * Each descriptor is 16 bytes. A receive buffer is also allocated for each - * descriptor. The maximum MTU size is 16110. - * Since TDLEN should be multiple of 128bytes, the number of transmit - * desscriptors should meet the following condition. - * (num_tx_desc * sizeof(struct e1000_tx_desc)) % 128 == 0 - */ -#define IGB_MIN_RXD 256 -#define IGB_DEFAULT_RXD 1024 -#define IGB_MAX_RXD 4096 - -/* - * IGB_TIDV - Transmit Interrupt Delay Value - * Valid Range: 0-65535 (0=off) - * Default Value: 64 - * This value delays the generation of transmit interrupts in units of - * 1.024 microseconds. Transmit interrupt reduction can improve CPU - * efficiency if properly tuned for specific network traffic. If the - * system is reporting dropped transmits, this value may be set too high - * causing the driver to run out of available transmit descriptors. - */ -#define IGB_TIDV 64 - -/* - * IGB_TADV - Transmit Absolute Interrupt Delay Value - * Valid Range: 0-65535 (0=off) - * Default Value: 64 - * This value, in units of 1.024 microseconds, limits the delay in which a - * transmit interrupt is generated. Useful only if IGB_TIDV is non-zero, - * this value ensures that an interrupt is generated after the initial - * packet is sent on the wire within the set amount of time. Proper tuning, - * along with IGB_TIDV, may improve traffic throughput in specific - * network conditions. - */ -#define IGB_TADV 64 - -/* - * IGB_RDTR - Receive Interrupt Delay Timer (Packet Timer) - * Valid Range: 0-65535 (0=off) - * Default Value: 0 - * This value delays the generation of receive interrupts in units of 1.024 - * microseconds. Receive interrupt reduction can improve CPU efficiency if - * properly tuned for specific network traffic. Increasing this value adds - * extra latency to frame reception and can end up decreasing the throughput - * of TCP traffic. If the system is reporting dropped receives, this value - * may be set too high, causing the driver to run out of available receive - * descriptors. - * - * CAUTION: When setting IGB_RDTR to a value other than 0, adapters - * may hang (stop transmitting) under certain network conditions. - * If this occurs a WATCHDOG message is logged in the system - * event log. In addition, the controller is automatically reset, - * restoring the network connection. To eliminate the potential - * for the hang ensure that IGB_RDTR is set to 0. - */ -#define IGB_RDTR 0 - -/* - * Receive Interrupt Absolute Delay Timer (Not valid for 82542/82543/82544) - * Valid Range: 0-65535 (0=off) - * Default Value: 64 - * This value, in units of 1.024 microseconds, limits the delay in which a - * receive interrupt is generated. Useful only if IGB_RDTR is non-zero, - * this value ensures that an interrupt is generated after the initial - * packet is received within the set amount of time. Proper tuning, - * along with IGB_RDTR, may improve traffic throughput in specific network - * conditions. - */ -#define IGB_RADV 64 - -/* - * This parameter controls the duration of transmit watchdog timer. - */ -#define IGB_WATCHDOG (10 * hz) - -/* - * This parameter controls when the driver calls the routine to reclaim - * transmit descriptors. Cleaning earlier seems a win. - */ -#define IGB_TX_CLEANUP_THRESHOLD (adapter->num_tx_desc / 2) - -/* - * This parameter controls whether or not autonegotation is enabled. - * 0 - Disable autonegotiation - * 1 - Enable autonegotiation - */ -#define DO_AUTO_NEG 1 - -/* - * This parameter control whether or not the driver will wait for - * autonegotiation to complete. - * 1 - Wait for autonegotiation to complete - * 0 - Don't wait for autonegotiation to complete - */ -#define WAIT_FOR_AUTO_NEG_DEFAULT 0 - -/* Tunables -- End */ - -#define AUTONEG_ADV_DEFAULT (ADVERTISE_10_HALF | ADVERTISE_10_FULL | \ - ADVERTISE_100_HALF | ADVERTISE_100_FULL | \ - ADVERTISE_1000_FULL) - -#define AUTO_ALL_MODES 0 - -/* PHY master/slave setting */ -#define IGB_MASTER_SLAVE e1000_ms_hw_default - -/* Support AutoMediaDetect for Marvell M88 PHY in i354 */ -#define IGB_MEDIA_RESET (1 << 0) - -/* - * Micellaneous constants - */ -#define IGB_INTEL_VENDOR_ID 0x8086 - -#define IGB_JUMBO_PBA 0x00000028 -#define IGB_DEFAULT_PBA 0x00000030 -#define IGB_SMARTSPEED_DOWNSHIFT 3 -#define IGB_SMARTSPEED_MAX 15 -#define IGB_MAX_LOOP 10 - -#define IGB_RX_PTHRESH ((hw->mac.type == e1000_i354) ? 12 : \ - ((hw->mac.type <= e1000_82576) ? 16 : 8)) -#define IGB_RX_HTHRESH 8 -#define IGB_RX_WTHRESH ((hw->mac.type == e1000_82576 && \ - adapter->msix_mem) ? 1 : 4) - -#define IGB_TX_PTHRESH ((hw->mac.type == e1000_i354) ? 20 : 8) -#define IGB_TX_HTHRESH 1 -#define IGB_TX_WTHRESH ((hw->mac.type != e1000_82575 && \ - adapter->msix_mem) ? 1 : 16) - -#define MAX_NUM_MULTICAST_ADDRESSES 128 -#define PCI_ANY_ID (~0U) -#define ETHER_ALIGN 2 -#define IGB_TX_BUFFER_SIZE ((uint32_t) 1514) -#define IGB_FC_PAUSE_TIME 0x0680 -#define IGB_EEPROM_APME 0x400; -/* Queue minimum free for use */ -#define IGB_QUEUE_THRESHOLD (adapter->num_tx_desc / 8) - -/* - * TDBA/RDBA should be aligned on 16 byte boundary. But TDLEN/RDLEN should be - * multiple of 128 bytes. So we align TDBA/RDBA on 128 byte boundary. This will - * also optimize cache line size effect. H/W supports up to cache line size 128. - */ -#define IGB_DBA_ALIGN 128 - -#define SPEED_MODE_BIT (1<<21) /* On PCI-E MACs only */ - -/* PCI Config defines */ -#define IGB_MSIX_BAR 3 - -/* Defines for printing debug information */ -#define DEBUG_INIT 0 -#define DEBUG_IOCTL 0 -#define DEBUG_HW 0 - -#define INIT_DEBUGOUT(S) if (DEBUG_INIT) printf(S "\n") -#define INIT_DEBUGOUT1(S, A) if (DEBUG_INIT) printf(S "\n", A) -#define INIT_DEBUGOUT2(S, A, B) if (DEBUG_INIT) printf(S "\n", A, B) -#define IOCTL_DEBUGOUT(S) if (DEBUG_IOCTL) printf(S "\n") -#define IOCTL_DEBUGOUT1(S, A) if (DEBUG_IOCTL) printf(S "\n", A) -#define IOCTL_DEBUGOUT2(S, A, B) if (DEBUG_IOCTL) printf(S "\n", A, B) -#define HW_DEBUGOUT(S) if (DEBUG_HW) printf(S "\n") -#define HW_DEBUGOUT1(S, A) if (DEBUG_HW) printf(S "\n", A) -#define HW_DEBUGOUT2(S, A, B) if (DEBUG_HW) printf(S "\n", A, B) - -#define IGB_MAX_SCATTER 40 -#define IGB_VFTA_SIZE 128 -#define IGB_BR_SIZE 4096 /* ring buf size */ -#define IGB_TSO_SIZE (65535 + sizeof(struct ether_vlan_header)) -#define IGB_TSO_SEG_SIZE 4096 /* Max dma segment size */ -#define IGB_TXPBSIZE 20408 -#define IGB_HDR_BUF 128 -#define IGB_PKTTYPE_MASK 0x0000FFF0 -#define IGB_DMCTLX_DCFLUSH_DIS 0x80000000 /* Disable DMA Coalesce Flush */ -#define ETH_ZLEN 60 -#define ETH_ADDR_LEN 6 - -/* Offload bits in mbuf flag */ -#if __FreeBSD_version >= 1000000 -#define CSUM_OFFLOAD_IPV4 (CSUM_IP|CSUM_IP_TCP|CSUM_IP_UDP|CSUM_IP_SCTP) -#define CSUM_OFFLOAD_IPV6 (CSUM_IP6_TCP|CSUM_IP6_UDP|CSUM_IP6_SCTP) -#define CSUM_OFFLOAD (CSUM_OFFLOAD_IPV4|CSUM_OFFLOAD_IPV6) -#elif __FreeBSD_version >= 800000 -#define CSUM_OFFLOAD (CSUM_IP|CSUM_TCP|CSUM_UDP|CSUM_SCTP) -#else -#define CSUM_OFFLOAD (CSUM_IP|CSUM_TCP|CSUM_UDP) -#endif - -/* Define the starting Interrupt rate per Queue */ -#define IGB_INTS_PER_SEC 8000 -#define IGB_DEFAULT_ITR ((1000000/IGB_INTS_PER_SEC) << 2) - -#define IGB_LINK_ITR 2000 -#define I210_LINK_DELAY 1000 - -/* Precision Time Sync (IEEE 1588) defines */ -#define ETHERTYPE_IEEE1588 0x88F7 -#define PICOSECS_PER_TICK 20833 -#define TSYNC_PORT 319 /* UDP port for the protocol */ - -/* - * Bus dma allocation structure used by - * e1000_dma_malloc and e1000_dma_free. - */ -struct igb_dma_alloc { - bus_addr_t dma_paddr; - caddr_t dma_vaddr; - bus_dma_tag_t dma_tag; - bus_dmamap_t dma_map; - bus_dma_segment_t dma_seg; - int dma_nseg; -}; - - -/* -** Driver queue struct: this is the interrupt container -** for the associated tx and rx ring. -*/ -struct igb_queue { - struct adapter *adapter; - u32 msix; /* This queue's MSIX vector */ - u32 eims; /* This queue's EIMS bit */ - u32 eitr_setting; - struct resource *res; - void *tag; - struct tx_ring *txr; - struct rx_ring *rxr; - struct task que_task; - struct taskqueue *tq; - u64 irqs; -}; - -/* - * The transmit ring, one per queue - */ -struct tx_ring { - struct adapter *adapter; - struct mtx tx_mtx; - u32 me; - int watchdog_time; - union e1000_adv_tx_desc *tx_base; - struct igb_tx_buf *tx_buffers; - struct igb_dma_alloc txdma; - volatile u16 tx_avail; - u16 next_avail_desc; - u16 next_to_clean; - u16 num_desc; - enum { - IGB_QUEUE_IDLE = 1, - IGB_QUEUE_WORKING = 2, - IGB_QUEUE_HUNG = 4, - IGB_QUEUE_DEPLETED = 8, - } queue_status; - u32 txd_cmd; - bus_dma_tag_t txtag; - char mtx_name[16]; -#ifndef IGB_LEGACY_TX - struct buf_ring *br; - struct task txq_task; -#endif - u32 bytes; /* used for AIM */ - u32 packets; - /* Soft Stats */ - unsigned long tso_tx; - unsigned long no_tx_map_avail; - unsigned long no_tx_dma_setup; - u64 no_desc_avail; - u64 total_packets; -}; - -/* - * Receive ring: one per queue - */ -struct rx_ring { - struct adapter *adapter; - u32 me; - struct igb_dma_alloc rxdma; - union e1000_adv_rx_desc *rx_base; - struct lro_ctrl lro; - bool lro_enabled; - bool hdr_split; - struct mtx rx_mtx; - char mtx_name[16]; - u32 next_to_refresh; - u32 next_to_check; - struct igb_rx_buf *rx_buffers; - bus_dma_tag_t htag; /* dma tag for rx head */ - bus_dma_tag_t ptag; /* dma tag for rx packet */ - /* - * First/last mbuf pointers, for - * collecting multisegment RX packets. - */ - struct mbuf *fmp; - struct mbuf *lmp; - - u32 bytes; - u32 packets; - int rdt; - int rdh; - - /* Soft stats */ - u64 rx_split_packets; - u64 rx_discarded; - u64 rx_packets; - u64 rx_bytes; -}; - -struct adapter { - struct ifnet *ifp; - struct e1000_hw hw; - - struct e1000_osdep osdep; - device_t dev; - struct cdev *led_dev; - - struct resource *pci_mem; - struct resource *msix_mem; - int memrid; - - /* - * Interrupt resources: this set is - * either used for legacy, or for Link - * when doing MSIX - */ - void *tag; - struct resource *res; - - struct ifmedia media; - struct callout timer; - int msix; - int if_flags; - int pause_frames; - - struct mtx core_mtx; - - eventhandler_tag vlan_attach; - eventhandler_tag vlan_detach; - - u16 num_vlans; - u16 num_queues; - - /* - ** Shadow VFTA table, this is needed because - ** the real vlan filter table gets cleared during - ** a soft reset and the driver needs to be able - ** to repopulate it. - */ - u32 shadow_vfta[IGB_VFTA_SIZE]; - - /* Info about the interface */ - u32 optics; - u32 fc; /* local flow ctrl setting */ - int advertise; /* link speeds */ - bool link_active; - u16 max_frame_size; - u16 num_segs; - u16 link_speed; - bool link_up; - u32 linkvec; - u16 link_duplex; - u32 dmac; - int link_mask; - - /* Flags */ - u32 flags; - - /* Mbuf cluster size */ - u32 rx_mbuf_sz; - - /* Support for pluggable optics */ - bool sfp_probe; - struct task link_task; /* Link tasklet */ - struct task mod_task; /* SFP tasklet */ - struct task msf_task; /* Multispeed Fiber */ - struct taskqueue *tq; - - /* - ** Queues: - ** This is the irq holder, it has - ** and RX/TX pair or rings associated - ** with it. - */ - struct igb_queue *queues; - - /* - * Transmit rings: - * Allocated at run time, an array of rings. - */ - struct tx_ring *tx_rings; - u32 num_tx_desc; - - /* - * Receive rings: - * Allocated at run time, an array of rings. - */ - struct rx_ring *rx_rings; - u64 que_mask; - u32 num_rx_desc; - - /* Multicast array memory */ - u8 *mta; - - /* Misc stats maintained by the driver */ - unsigned long device_control; - unsigned long dropped_pkts; - unsigned long eint_mask; - unsigned long int_mask; - unsigned long link_irq; - unsigned long mbuf_defrag_failed; - unsigned long no_tx_dma_setup; - unsigned long packet_buf_alloc_rx; - unsigned long packet_buf_alloc_tx; - unsigned long rx_control; - unsigned long rx_overruns; - unsigned long watchdog_events; - - /* Used in pf and vf */ - void *stats; - - int enable_aim; - int has_manage; - int wol; - int rx_process_limit; - int tx_process_limit; - u16 vf_ifp; /* a VF interface */ - bool in_detach; /* Used only in igb_ioctl */ - -}; - -/* ****************************************************************************** - * vendor_info_array - * - * This array contains the list of Subvendor/Subdevice IDs on which the driver - * should load. - * - * ******************************************************************************/ -typedef struct _igb_vendor_info_t { - unsigned int vendor_id; - unsigned int device_id; - unsigned int subvendor_id; - unsigned int subdevice_id; - unsigned int index; -} igb_vendor_info_t; - -struct igb_tx_buf { - union e1000_adv_tx_desc *eop; - struct mbuf *m_head; - bus_dmamap_t map; -}; - -struct igb_rx_buf { - struct mbuf *m_head; - struct mbuf *m_pack; - bus_dmamap_t hmap; /* bus_dma map for header */ - bus_dmamap_t pmap; /* bus_dma map for packet */ -}; - -/* -** Find the number of unrefreshed RX descriptors -*/ -static inline u16 -igb_rx_unrefreshed(struct rx_ring *rxr) -{ - struct adapter *adapter = rxr->adapter; - - if (rxr->next_to_check > rxr->next_to_refresh) - return (rxr->next_to_check - rxr->next_to_refresh - 1); - else - return ((adapter->num_rx_desc + rxr->next_to_check) - - rxr->next_to_refresh - 1); -} - -#define IGB_CORE_LOCK_INIT(_sc, _name) \ - mtx_init(&(_sc)->core_mtx, _name, "IGB Core Lock", MTX_DEF) -#define IGB_CORE_LOCK_DESTROY(_sc) mtx_destroy(&(_sc)->core_mtx) -#define IGB_CORE_LOCK(_sc) mtx_lock(&(_sc)->core_mtx) -#define IGB_CORE_UNLOCK(_sc) mtx_unlock(&(_sc)->core_mtx) -#define IGB_CORE_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->core_mtx, MA_OWNED) - -#define IGB_TX_LOCK_DESTROY(_sc) mtx_destroy(&(_sc)->tx_mtx) -#define IGB_TX_LOCK(_sc) mtx_lock(&(_sc)->tx_mtx) -#define IGB_TX_UNLOCK(_sc) mtx_unlock(&(_sc)->tx_mtx) -#define IGB_TX_TRYLOCK(_sc) mtx_trylock(&(_sc)->tx_mtx) -#define IGB_TX_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->tx_mtx, MA_OWNED) - -#define IGB_RX_LOCK_DESTROY(_sc) mtx_destroy(&(_sc)->rx_mtx) -#define IGB_RX_LOCK(_sc) mtx_lock(&(_sc)->rx_mtx) -#define IGB_RX_UNLOCK(_sc) mtx_unlock(&(_sc)->rx_mtx) -#define IGB_RX_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->rx_mtx, MA_OWNED) - -#define UPDATE_VF_REG(reg, last, cur) \ -{ \ - u32 new = E1000_READ_REG(hw, reg); \ - if (new < last) \ - cur += 0x100000000LL; \ - last = new; \ - cur &= 0xFFFFFFFF00000000LL; \ - cur |= new; \ -} - -#if __FreeBSD_version >= 800000 && __FreeBSD_version < 800504 -static __inline int -drbr_needs_enqueue(struct ifnet *ifp, struct buf_ring *br) -{ -#ifdef ALTQ - if (ALTQ_IS_ENABLED(&ifp->if_snd)) - return (1); -#endif - return (!buf_ring_empty(br)); -} -#endif - -#endif /* _IF_IGB_H_ */ - - diff --git a/sys/dev/e1000/if_lem.c b/sys/dev/e1000/if_lem.c deleted file mode 100644 index 537ccbe5ff11..000000000000 --- a/sys/dev/e1000/if_lem.c +++ /dev/null @@ -1,4730 +0,0 @@ -/****************************************************************************** - - Copyright (c) 2001-2015, Intel Corporation - All rights reserved. - - Redistribution and use in source and binary forms, with or without - modification, are permitted provided that the following conditions are met: - - 1. Redistributions of source code must retain the above copyright notice, - this list of conditions and the following disclaimer. - - 2. Redistributions in binary form must reproduce the above copyright - notice, this list of conditions and the following disclaimer in the - documentation and/or other materials provided with the distribution. - - 3. Neither the name of the Intel Corporation nor the names of its - contributors may be used to endorse or promote products derived from - this software without specific prior written permission. - - THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" - AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE - LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR - CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF - SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS - INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN - CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) - ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE - POSSIBILITY OF SUCH DAMAGE. - -******************************************************************************/ -/*$FreeBSD$*/ - -/* - * Uncomment the following extensions for better performance in a VM, - * especially if you have support in the hypervisor. - * See http://info.iet.unipi.it/~luigi/netmap/ - */ -// #define BATCH_DISPATCH -// #define NIC_SEND_COMBINING - -#include "opt_inet.h" -#include "opt_inet6.h" - -#ifdef HAVE_KERNEL_OPTION_HEADERS -#include "opt_device_polling.h" -#endif - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include -#include -#include -#include -#include - -#include -#include - -#include -#include -#include -#include -#include -#include -#include - -#include -#include -#include -#include - -#include "e1000_api.h" -#include "if_lem.h" - -/********************************************************************* - * Legacy Em Driver version: - *********************************************************************/ -char lem_driver_version[] = "1.1.0"; - -/********************************************************************* - * PCI Device ID Table - * - * Used by probe to select devices to load on - * Last field stores an index into e1000_strings - * Last entry must be all 0s - * - * { Vendor ID, Device ID, SubVendor ID, SubDevice ID, String Index } - *********************************************************************/ - -static em_vendor_info_t lem_vendor_info_array[] = -{ - /* Intel(R) PRO/1000 Network Connection */ - { 0x8086, E1000_DEV_ID_82540EM, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82540EM_LOM, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82540EP, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82540EP_LOM, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82540EP_LP, PCI_ANY_ID, PCI_ANY_ID, 0}, - - { 0x8086, E1000_DEV_ID_82541EI, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82541ER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82541ER_LOM, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82541EI_MOBILE, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82541GI, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82541GI_LF, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82541GI_MOBILE, PCI_ANY_ID, PCI_ANY_ID, 0}, - - { 0x8086, E1000_DEV_ID_82542, PCI_ANY_ID, PCI_ANY_ID, 0}, - - { 0x8086, E1000_DEV_ID_82543GC_FIBER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82543GC_COPPER, PCI_ANY_ID, PCI_ANY_ID, 0}, - - { 0x8086, E1000_DEV_ID_82544EI_COPPER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82544EI_FIBER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82544GC_COPPER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82544GC_LOM, PCI_ANY_ID, PCI_ANY_ID, 0}, - - { 0x8086, E1000_DEV_ID_82545EM_COPPER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82545EM_FIBER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82545GM_COPPER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82545GM_FIBER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82545GM_SERDES, PCI_ANY_ID, PCI_ANY_ID, 0}, - - { 0x8086, E1000_DEV_ID_82546EB_COPPER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82546EB_FIBER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82546EB_QUAD_COPPER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82546GB_COPPER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82546GB_FIBER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82546GB_SERDES, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82546GB_PCIE, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82546GB_QUAD_COPPER, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82546GB_QUAD_COPPER_KSP3, - PCI_ANY_ID, PCI_ANY_ID, 0}, - - { 0x8086, E1000_DEV_ID_82547EI, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82547EI_MOBILE, PCI_ANY_ID, PCI_ANY_ID, 0}, - { 0x8086, E1000_DEV_ID_82547GI, PCI_ANY_ID, PCI_ANY_ID, 0}, - /* required last entry */ - { 0, 0, 0, 0, 0} -}; - -/********************************************************************* - * Table of branding strings for all supported NICs. - *********************************************************************/ - -static char *lem_strings[] = { - "Intel(R) PRO/1000 Legacy Network Connection" -}; - -/********************************************************************* - * Function prototypes - *********************************************************************/ -static int lem_probe(device_t); -static int lem_attach(device_t); -static int lem_detach(device_t); -static int lem_shutdown(device_t); -static int lem_suspend(device_t); -static int lem_resume(device_t); -static void lem_start(if_t); -static void lem_start_locked(if_t ifp); -static int lem_ioctl(if_t, u_long, caddr_t); -static uint64_t lem_get_counter(if_t, ift_counter); -static void lem_init(void *); -static void lem_init_locked(struct adapter *); -static void lem_stop(void *); -static void lem_media_status(if_t, struct ifmediareq *); -static int lem_media_change(if_t); -static void lem_identify_hardware(struct adapter *); -static int lem_allocate_pci_resources(struct adapter *); -static int lem_allocate_irq(struct adapter *adapter); -static void lem_free_pci_resources(struct adapter *); -static void lem_local_timer(void *); -static int lem_hardware_init(struct adapter *); -static int lem_setup_interface(device_t, struct adapter *); -static void lem_setup_transmit_structures(struct adapter *); -static void lem_initialize_transmit_unit(struct adapter *); -static int lem_setup_receive_structures(struct adapter *); -static void lem_initialize_receive_unit(struct adapter *); -static void lem_enable_intr(struct adapter *); -static void lem_disable_intr(struct adapter *); -static void lem_free_transmit_structures(struct adapter *); -static void lem_free_receive_structures(struct adapter *); -static void lem_update_stats_counters(struct adapter *); -static void lem_add_hw_stats(struct adapter *adapter); -static void lem_txeof(struct adapter *); -static void lem_tx_purge(struct adapter *); -static int lem_allocate_receive_structures(struct adapter *); -static int lem_allocate_transmit_structures(struct adapter *); -static bool lem_rxeof(struct adapter *, int, int *); -#ifndef __NO_STRICT_ALIGNMENT -static int lem_fixup_rx(struct adapter *); -#endif -static void lem_receive_checksum(struct adapter *, struct e1000_rx_desc *, - struct mbuf *); -static void lem_transmit_checksum_setup(struct adapter *, struct mbuf *, - u32 *, u32 *); -static void lem_set_promisc(struct adapter *); -static void lem_disable_promisc(struct adapter *); -static void lem_set_multi(struct adapter *); -static void lem_update_link_status(struct adapter *); -static int lem_get_buf(struct adapter *, int); -static void lem_register_vlan(void *, if_t, u16); -static void lem_unregister_vlan(void *, if_t, u16); -static void lem_setup_vlan_hw_support(struct adapter *); -static int lem_xmit(struct adapter *, struct mbuf **); -static void lem_smartspeed(struct adapter *); -static int lem_82547_fifo_workaround(struct adapter *, int); -static void lem_82547_update_fifo_head(struct adapter *, int); -static int lem_82547_tx_fifo_reset(struct adapter *); -static void lem_82547_move_tail(void *); -static int lem_dma_malloc(struct adapter *, bus_size_t, - struct em_dma_alloc *, int); -static void lem_dma_free(struct adapter *, struct em_dma_alloc *); -static int lem_sysctl_nvm_info(SYSCTL_HANDLER_ARGS); -static void lem_print_nvm_info(struct adapter *); -static int lem_is_valid_ether_addr(u8 *); -static u32 lem_fill_descriptors (bus_addr_t address, u32 length, - PDESC_ARRAY desc_array); -static int lem_sysctl_int_delay(SYSCTL_HANDLER_ARGS); -static void lem_add_int_delay_sysctl(struct adapter *, const char *, - const char *, struct em_int_delay_info *, int, int); -static void lem_set_flow_cntrl(struct adapter *, const char *, - const char *, int *, int); -/* Management and WOL Support */ -static void lem_init_manageability(struct adapter *); -static void lem_release_manageability(struct adapter *); -static void lem_get_hw_control(struct adapter *); -static void lem_release_hw_control(struct adapter *); -static void lem_get_wakeup(device_t); -static void lem_enable_wakeup(device_t); -static int lem_enable_phy_wakeup(struct adapter *); -static void lem_led_func(void *, int); - -static void lem_intr(void *); -static int lem_irq_fast(void *); -static void lem_handle_rxtx(void *context, int pending); -static void lem_handle_link(void *context, int pending); -static void lem_add_rx_process_limit(struct adapter *, const char *, - const char *, int *, int); - -#ifdef DEVICE_POLLING -static poll_handler_t lem_poll; -#endif /* POLLING */ - -/********************************************************************* - * FreeBSD Device Interface Entry Points - *********************************************************************/ - -static device_method_t lem_methods[] = { - /* Device interface */ - DEVMETHOD(device_probe, lem_probe), - DEVMETHOD(device_attach, lem_attach), - DEVMETHOD(device_detach, lem_detach), - DEVMETHOD(device_shutdown, lem_shutdown), - DEVMETHOD(device_suspend, lem_suspend), - DEVMETHOD(device_resume, lem_resume), - DEVMETHOD_END -}; - -static driver_t lem_driver = { - "em", lem_methods, sizeof(struct adapter), -}; - -extern devclass_t em_devclass; -DRIVER_MODULE(lem, pci, lem_driver, em_devclass, 0, 0); -MODULE_DEPEND(lem, pci, 1, 1, 1); -MODULE_DEPEND(lem, ether, 1, 1, 1); -#ifdef DEV_NETMAP -MODULE_DEPEND(lem, netmap, 1, 1, 1); -#endif /* DEV_NETMAP */ - -/********************************************************************* - * Tunable default values. - *********************************************************************/ - -#define EM_TICKS_TO_USECS(ticks) ((1024 * (ticks) + 500) / 1000) -#define EM_USECS_TO_TICKS(usecs) ((1000 * (usecs) + 512) / 1024) - -#define MAX_INTS_PER_SEC 8000 -#define DEFAULT_ITR (1000000000/(MAX_INTS_PER_SEC * 256)) - -static int lem_tx_int_delay_dflt = EM_TICKS_TO_USECS(EM_TIDV); -static int lem_rx_int_delay_dflt = EM_TICKS_TO_USECS(EM_RDTR); -static int lem_tx_abs_int_delay_dflt = EM_TICKS_TO_USECS(EM_TADV); -static int lem_rx_abs_int_delay_dflt = EM_TICKS_TO_USECS(EM_RADV); -/* - * increase lem_rxd and lem_txd to at least 2048 in netmap mode - * for better performance. - */ -static int lem_rxd = EM_DEFAULT_RXD; -static int lem_txd = EM_DEFAULT_TXD; -static int lem_smart_pwr_down = FALSE; - -/* Controls whether promiscuous also shows bad packets */ -static int lem_debug_sbp = FALSE; - -TUNABLE_INT("hw.em.tx_int_delay", &lem_tx_int_delay_dflt); -TUNABLE_INT("hw.em.rx_int_delay", &lem_rx_int_delay_dflt); -TUNABLE_INT("hw.em.tx_abs_int_delay", &lem_tx_abs_int_delay_dflt); -TUNABLE_INT("hw.em.rx_abs_int_delay", &lem_rx_abs_int_delay_dflt); -TUNABLE_INT("hw.em.rxd", &lem_rxd); -TUNABLE_INT("hw.em.txd", &lem_txd); -TUNABLE_INT("hw.em.smart_pwr_down", &lem_smart_pwr_down); -TUNABLE_INT("hw.em.sbp", &lem_debug_sbp); - -/* Interrupt style - default to fast */ -static int lem_use_legacy_irq = 0; -TUNABLE_INT("hw.em.use_legacy_irq", &lem_use_legacy_irq); - -/* How many packets rxeof tries to clean at a time */ -static int lem_rx_process_limit = 100; -TUNABLE_INT("hw.em.rx_process_limit", &lem_rx_process_limit); - -/* Flow control setting - default to FULL */ -static int lem_fc_setting = e1000_fc_full; -TUNABLE_INT("hw.em.fc_setting", &lem_fc_setting); - -/* Global used in WOL setup with multiport cards */ -static int global_quad_port_a = 0; - -#ifdef DEV_NETMAP /* see ixgbe.c for details */ -#include -#endif /* DEV_NETMAP */ - -/********************************************************************* - * Device identification routine - * - * em_probe determines if the driver should be loaded on - * adapter based on PCI vendor/device id of the adapter. - * - * return BUS_PROBE_DEFAULT on success, positive on failure - *********************************************************************/ - -static int -lem_probe(device_t dev) -{ - char adapter_name[60]; - u16 pci_vendor_id = 0; - u16 pci_device_id = 0; - u16 pci_subvendor_id = 0; - u16 pci_subdevice_id = 0; - em_vendor_info_t *ent; - - INIT_DEBUGOUT("em_probe: begin"); - - pci_vendor_id = pci_get_vendor(dev); - if (pci_vendor_id != EM_VENDOR_ID) - return (ENXIO); - - pci_device_id = pci_get_device(dev); - pci_subvendor_id = pci_get_subvendor(dev); - pci_subdevice_id = pci_get_subdevice(dev); - - ent = lem_vendor_info_array; - while (ent->vendor_id != 0) { - if ((pci_vendor_id == ent->vendor_id) && - (pci_device_id == ent->device_id) && - - ((pci_subvendor_id == ent->subvendor_id) || - (ent->subvendor_id == PCI_ANY_ID)) && - - ((pci_subdevice_id == ent->subdevice_id) || - (ent->subdevice_id == PCI_ANY_ID))) { - sprintf(adapter_name, "%s %s", - lem_strings[ent->index], - lem_driver_version); - device_set_desc_copy(dev, adapter_name); - return (BUS_PROBE_DEFAULT); - } - ent++; - } - - return (ENXIO); -} - -/********************************************************************* - * Device initialization routine - * - * The attach entry point is called when the driver is being loaded. - * This routine identifies the type of hardware, allocates all resources - * and initializes the hardware. - * - * return 0 on success, positive on failure - *********************************************************************/ - -static int -lem_attach(device_t dev) -{ - struct adapter *adapter; - int tsize, rsize; - int error = 0; - - INIT_DEBUGOUT("lem_attach: begin"); - - adapter = device_get_softc(dev); - adapter->dev = adapter->osdep.dev = dev; - EM_CORE_LOCK_INIT(adapter, device_get_nameunit(dev)); - EM_TX_LOCK_INIT(adapter, device_get_nameunit(dev)); - EM_RX_LOCK_INIT(adapter, device_get_nameunit(dev)); - - /* SYSCTL stuff */ - SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), - SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), - OID_AUTO, "nvm", CTLTYPE_INT|CTLFLAG_RW, adapter, 0, - lem_sysctl_nvm_info, "I", "NVM Information"); - - callout_init_mtx(&adapter->timer, &adapter->core_mtx, 0); - callout_init_mtx(&adapter->tx_fifo_timer, &adapter->tx_mtx, 0); - - /* Determine hardware and mac info */ - lem_identify_hardware(adapter); - - /* Setup PCI resources */ - if (lem_allocate_pci_resources(adapter)) { - device_printf(dev, "Allocation of PCI resources failed\n"); - error = ENXIO; - goto err_pci; - } - - /* Do Shared Code initialization */ - if (e1000_setup_init_funcs(&adapter->hw, TRUE)) { - device_printf(dev, "Setup of Shared code failed\n"); - error = ENXIO; - goto err_pci; - } - - e1000_get_bus_info(&adapter->hw); - - /* Set up some sysctls for the tunable interrupt delays */ - lem_add_int_delay_sysctl(adapter, "rx_int_delay", - "receive interrupt delay in usecs", &adapter->rx_int_delay, - E1000_REGISTER(&adapter->hw, E1000_RDTR), lem_rx_int_delay_dflt); - lem_add_int_delay_sysctl(adapter, "tx_int_delay", - "transmit interrupt delay in usecs", &adapter->tx_int_delay, - E1000_REGISTER(&adapter->hw, E1000_TIDV), lem_tx_int_delay_dflt); - if (adapter->hw.mac.type >= e1000_82540) { - lem_add_int_delay_sysctl(adapter, "rx_abs_int_delay", - "receive interrupt delay limit in usecs", - &adapter->rx_abs_int_delay, - E1000_REGISTER(&adapter->hw, E1000_RADV), - lem_rx_abs_int_delay_dflt); - lem_add_int_delay_sysctl(adapter, "tx_abs_int_delay", - "transmit interrupt delay limit in usecs", - &adapter->tx_abs_int_delay, - E1000_REGISTER(&adapter->hw, E1000_TADV), - lem_tx_abs_int_delay_dflt); - lem_add_int_delay_sysctl(adapter, "itr", - "interrupt delay limit in usecs/4", - &adapter->tx_itr, - E1000_REGISTER(&adapter->hw, E1000_ITR), - DEFAULT_ITR); - } - - /* Sysctls for limiting the amount of work done in the taskqueue */ - lem_add_rx_process_limit(adapter, "rx_processing_limit", - "max number of rx packets to process", &adapter->rx_process_limit, - lem_rx_process_limit); - -#ifdef NIC_SEND_COMBINING - /* Sysctls to control mitigation */ - lem_add_rx_process_limit(adapter, "sc_enable", - "driver TDT mitigation", &adapter->sc_enable, 0); -#endif /* NIC_SEND_COMBINING */ -#ifdef BATCH_DISPATCH - lem_add_rx_process_limit(adapter, "batch_enable", - "driver rx batch", &adapter->batch_enable, 0); -#endif /* BATCH_DISPATCH */ - - /* Sysctl for setting the interface flow control */ - lem_set_flow_cntrl(adapter, "flow_control", - "flow control setting", - &adapter->fc_setting, lem_fc_setting); - - /* - * Validate number of transmit and receive descriptors. It - * must not exceed hardware maximum, and must be multiple - * of E1000_DBA_ALIGN. - */ - if (((lem_txd * sizeof(struct e1000_tx_desc)) % EM_DBA_ALIGN) != 0 || - (adapter->hw.mac.type >= e1000_82544 && lem_txd > EM_MAX_TXD) || - (adapter->hw.mac.type < e1000_82544 && lem_txd > EM_MAX_TXD_82543) || - (lem_txd < EM_MIN_TXD)) { - device_printf(dev, "Using %d TX descriptors instead of %d!\n", - EM_DEFAULT_TXD, lem_txd); - adapter->num_tx_desc = EM_DEFAULT_TXD; - } else - adapter->num_tx_desc = lem_txd; - if (((lem_rxd * sizeof(struct e1000_rx_desc)) % EM_DBA_ALIGN) != 0 || - (adapter->hw.mac.type >= e1000_82544 && lem_rxd > EM_MAX_RXD) || - (adapter->hw.mac.type < e1000_82544 && lem_rxd > EM_MAX_RXD_82543) || - (lem_rxd < EM_MIN_RXD)) { - device_printf(dev, "Using %d RX descriptors instead of %d!\n", - EM_DEFAULT_RXD, lem_rxd); - adapter->num_rx_desc = EM_DEFAULT_RXD; - } else - adapter->num_rx_desc = lem_rxd; - - adapter->hw.mac.autoneg = DO_AUTO_NEG; - adapter->hw.phy.autoneg_wait_to_complete = FALSE; - adapter->hw.phy.autoneg_advertised = AUTONEG_ADV_DEFAULT; - adapter->rx_buffer_len = 2048; - - e1000_init_script_state_82541(&adapter->hw, TRUE); - e1000_set_tbi_compatibility_82543(&adapter->hw, TRUE); - - /* Copper options */ - if (adapter->hw.phy.media_type == e1000_media_type_copper) { - adapter->hw.phy.mdix = AUTO_ALL_MODES; - adapter->hw.phy.disable_polarity_correction = FALSE; - adapter->hw.phy.ms_type = EM_MASTER_SLAVE; - } - - /* - * Set the frame limits assuming - * standard ethernet sized frames. - */ - adapter->max_frame_size = ETHERMTU + ETHER_HDR_LEN + ETHERNET_FCS_SIZE; - adapter->min_frame_size = ETH_ZLEN + ETHERNET_FCS_SIZE; - - /* - * This controls when hardware reports transmit completion - * status. - */ - adapter->hw.mac.report_tx_early = 1; - - /* - * It seems that the descriptor DMA engine on some PCI cards - * fetches memory past the end of the last descriptor in the - * ring. These reads are problematic when VT-d (DMAR) busdma - * is used. Allocate the scratch space to avoid getting - * faults from DMAR, by requesting scratch memory for one more - * descriptor. - */ - tsize = roundup2((adapter->num_tx_desc + 1) * - sizeof(struct e1000_tx_desc), EM_DBA_ALIGN); - - /* Allocate Transmit Descriptor ring */ - if (lem_dma_malloc(adapter, tsize, &adapter->txdma, BUS_DMA_NOWAIT)) { - device_printf(dev, "Unable to allocate tx_desc memory\n"); - error = ENOMEM; - goto err_tx_desc; - } - adapter->tx_desc_base = - (struct e1000_tx_desc *)adapter->txdma.dma_vaddr; - - /* - * See comment above txdma allocation for rationale behind +1. - */ - rsize = roundup2((adapter->num_rx_desc + 1) * - sizeof(struct e1000_rx_desc), EM_DBA_ALIGN); - - /* Allocate Receive Descriptor ring */ - if (lem_dma_malloc(adapter, rsize, &adapter->rxdma, BUS_DMA_NOWAIT)) { - device_printf(dev, "Unable to allocate rx_desc memory\n"); - error = ENOMEM; - goto err_rx_desc; - } - adapter->rx_desc_base = - (struct e1000_rx_desc *)adapter->rxdma.dma_vaddr; - - /* Allocate multicast array memory. */ - adapter->mta = malloc(sizeof(u8) * ETH_ADDR_LEN * - MAX_NUM_MULTICAST_ADDRESSES, M_DEVBUF, M_NOWAIT); - if (adapter->mta == NULL) { - device_printf(dev, "Can not allocate multicast setup array\n"); - error = ENOMEM; - goto err_hw_init; - } - - /* - ** Start from a known state, this is - ** important in reading the nvm and - ** mac from that. - */ - e1000_reset_hw(&adapter->hw); - - /* Make sure we have a good EEPROM before we read from it */ - if (e1000_validate_nvm_checksum(&adapter->hw) < 0) { - /* - ** Some PCI-E parts fail the first check due to - ** the link being in sleep state, call it again, - ** if it fails a second time its a real issue. - */ - if (e1000_validate_nvm_checksum(&adapter->hw) < 0) { - device_printf(dev, - "The EEPROM Checksum Is Not Valid\n"); - error = EIO; - goto err_hw_init; - } - } - - /* Copy the permanent MAC address out of the EEPROM */ - if (e1000_read_mac_addr(&adapter->hw) < 0) { - device_printf(dev, "EEPROM read error while reading MAC" - " address\n"); - error = EIO; - goto err_hw_init; - } - - if (!lem_is_valid_ether_addr(adapter->hw.mac.addr)) { - device_printf(dev, "Invalid MAC address\n"); - error = EIO; - goto err_hw_init; - } - - /* Initialize the hardware */ - if (lem_hardware_init(adapter)) { - device_printf(dev, "Unable to initialize the hardware\n"); - error = EIO; - goto err_hw_init; - } - - /* Allocate transmit descriptors and buffers */ - if (lem_allocate_transmit_structures(adapter)) { - device_printf(dev, "Could not setup transmit structures\n"); - error = ENOMEM; - goto err_tx_struct; - } - - /* Allocate receive descriptors and buffers */ - if (lem_allocate_receive_structures(adapter)) { - device_printf(dev, "Could not setup receive structures\n"); - error = ENOMEM; - goto err_rx_struct; - } - - /* - ** Do interrupt configuration - */ - error = lem_allocate_irq(adapter); - if (error) - goto err_rx_struct; - - /* - * Get Wake-on-Lan and Management info for later use - */ - lem_get_wakeup(dev); - - /* Setup OS specific network interface */ - if (lem_setup_interface(dev, adapter) != 0) - goto err_rx_struct; - - /* Initialize statistics */ - lem_update_stats_counters(adapter); - - adapter->hw.mac.get_link_status = 1; - lem_update_link_status(adapter); - - /* Indicate SOL/IDER usage */ - if (e1000_check_reset_block(&adapter->hw)) - device_printf(dev, - "PHY reset is blocked due to SOL/IDER session.\n"); - - /* Do we need workaround for 82544 PCI-X adapter? */ - if (adapter->hw.bus.type == e1000_bus_type_pcix && - adapter->hw.mac.type == e1000_82544) - adapter->pcix_82544 = TRUE; - else - adapter->pcix_82544 = FALSE; - - /* Register for VLAN events */ - adapter->vlan_attach = EVENTHANDLER_REGISTER(vlan_config, - lem_register_vlan, adapter, EVENTHANDLER_PRI_FIRST); - adapter->vlan_detach = EVENTHANDLER_REGISTER(vlan_unconfig, - lem_unregister_vlan, adapter, EVENTHANDLER_PRI_FIRST); - - lem_add_hw_stats(adapter); - - /* Non-AMT based hardware can now take control from firmware */ - if (adapter->has_manage && !adapter->has_amt) - lem_get_hw_control(adapter); - - /* Tell the stack that the interface is not active */ - if_setdrvflagbits(adapter->ifp, 0, IFF_DRV_OACTIVE | IFF_DRV_RUNNING); - - adapter->led_dev = led_create(lem_led_func, adapter, - device_get_nameunit(dev)); - -#ifdef DEV_NETMAP - lem_netmap_attach(adapter); -#endif /* DEV_NETMAP */ - INIT_DEBUGOUT("lem_attach: end"); - - return (0); - -err_rx_struct: - lem_free_transmit_structures(adapter); -err_tx_struct: -err_hw_init: - lem_release_hw_control(adapter); - lem_dma_free(adapter, &adapter->rxdma); -err_rx_desc: - lem_dma_free(adapter, &adapter->txdma); -err_tx_desc: -err_pci: - if (adapter->ifp != (void *)NULL) - if_free(adapter->ifp); - lem_free_pci_resources(adapter); - free(adapter->mta, M_DEVBUF); - EM_TX_LOCK_DESTROY(adapter); - EM_RX_LOCK_DESTROY(adapter); - EM_CORE_LOCK_DESTROY(adapter); - - return (error); -} - -/********************************************************************* - * Device removal routine - * - * The detach entry point is called when the driver is being removed. - * This routine stops the adapter and deallocates all the resources - * that were allocated for driver operation. - * - * return 0 on success, positive on failure - *********************************************************************/ - -static int -lem_detach(device_t dev) -{ - struct adapter *adapter = device_get_softc(dev); - if_t ifp = adapter->ifp; - - INIT_DEBUGOUT("em_detach: begin"); - - /* Make sure VLANS are not using driver */ - if (if_vlantrunkinuse(ifp)) { - device_printf(dev,"Vlan in use, detach first\n"); - return (EBUSY); - } - -#ifdef DEVICE_POLLING - if (if_getcapenable(ifp) & IFCAP_POLLING) - ether_poll_deregister(ifp); -#endif - - if (adapter->led_dev != NULL) - led_destroy(adapter->led_dev); - - EM_CORE_LOCK(adapter); - EM_TX_LOCK(adapter); - adapter->in_detach = 1; - lem_stop(adapter); - e1000_phy_hw_reset(&adapter->hw); - - lem_release_manageability(adapter); - - EM_TX_UNLOCK(adapter); - EM_CORE_UNLOCK(adapter); - - /* Unregister VLAN events */ - if (adapter->vlan_attach != NULL) - EVENTHANDLER_DEREGISTER(vlan_config, adapter->vlan_attach); - if (adapter->vlan_detach != NULL) - EVENTHANDLER_DEREGISTER(vlan_unconfig, adapter->vlan_detach); - - ether_ifdetach(adapter->ifp); - callout_drain(&adapter->timer); - callout_drain(&adapter->tx_fifo_timer); - -#ifdef DEV_NETMAP - netmap_detach(ifp); -#endif /* DEV_NETMAP */ - lem_free_pci_resources(adapter); - bus_generic_detach(dev); - if_free(ifp); - - lem_free_transmit_structures(adapter); - lem_free_receive_structures(adapter); - - /* Free Transmit Descriptor ring */ - if (adapter->tx_desc_base) { - lem_dma_free(adapter, &adapter->txdma); - adapter->tx_desc_base = NULL; - } - - /* Free Receive Descriptor ring */ - if (adapter->rx_desc_base) { - lem_dma_free(adapter, &adapter->rxdma); - adapter->rx_desc_base = NULL; - } - - lem_release_hw_control(adapter); - free(adapter->mta, M_DEVBUF); - EM_TX_LOCK_DESTROY(adapter); - EM_RX_LOCK_DESTROY(adapter); - EM_CORE_LOCK_DESTROY(adapter); - - return (0); -} - -/********************************************************************* - * - * Shutdown entry point - * - **********************************************************************/ - -static int -lem_shutdown(device_t dev) -{ - return lem_suspend(dev); -} - -/* - * Suspend/resume device methods. - */ -static int -lem_suspend(device_t dev) -{ - struct adapter *adapter = device_get_softc(dev); - - EM_CORE_LOCK(adapter); - - lem_release_manageability(adapter); - lem_release_hw_control(adapter); - lem_enable_wakeup(dev); - - EM_CORE_UNLOCK(adapter); - - return bus_generic_suspend(dev); -} - -static int -lem_resume(device_t dev) -{ - struct adapter *adapter = device_get_softc(dev); - if_t ifp = adapter->ifp; - - EM_CORE_LOCK(adapter); - lem_init_locked(adapter); - lem_init_manageability(adapter); - EM_CORE_UNLOCK(adapter); - lem_start(ifp); - - return bus_generic_resume(dev); -} - - -static void -lem_start_locked(if_t ifp) -{ - struct adapter *adapter = if_getsoftc(ifp); - struct mbuf *m_head; - - EM_TX_LOCK_ASSERT(adapter); - - if ((if_getdrvflags(ifp) & (IFF_DRV_RUNNING|IFF_DRV_OACTIVE)) != - IFF_DRV_RUNNING) - return; - if (!adapter->link_active) - return; - - /* - * Force a cleanup if number of TX descriptors - * available hits the threshold - */ - if (adapter->num_tx_desc_avail <= EM_TX_CLEANUP_THRESHOLD) { - lem_txeof(adapter); - /* Now do we at least have a minimal? */ - if (adapter->num_tx_desc_avail <= EM_TX_OP_THRESHOLD) { - adapter->no_tx_desc_avail1++; - return; - } - } - - while (!if_sendq_empty(ifp)) { - m_head = if_dequeue(ifp); - - if (m_head == NULL) - break; - /* - * Encapsulation can modify our pointer, and or make it - * NULL on failure. In that event, we can't requeue. - */ - if (lem_xmit(adapter, &m_head)) { - if (m_head == NULL) - break; - if_setdrvflagbits(ifp, IFF_DRV_OACTIVE, 0); - if_sendq_prepend(ifp, m_head); - break; - } - - /* Send a copy of the frame to the BPF listener */ - if_etherbpfmtap(ifp, m_head); - - /* Set timeout in case hardware has problems transmitting. */ - adapter->watchdog_check = TRUE; - adapter->watchdog_time = ticks; - } - if (adapter->num_tx_desc_avail <= EM_TX_OP_THRESHOLD) - if_setdrvflagbits(ifp, IFF_DRV_OACTIVE, 0); - - return; -} - -static void -lem_start(if_t ifp) -{ - struct adapter *adapter = if_getsoftc(ifp); - - EM_TX_LOCK(adapter); - if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) - lem_start_locked(ifp); - EM_TX_UNLOCK(adapter); -} - -/********************************************************************* - * Ioctl entry point - * - * em_ioctl is called when the user wants to configure the - * interface. - * - * return 0 on success, positive on failure - **********************************************************************/ - -static int -lem_ioctl(if_t ifp, u_long command, caddr_t data) -{ - struct adapter *adapter = if_getsoftc(ifp); - struct ifreq *ifr = (struct ifreq *)data; -#if defined(INET) || defined(INET6) - struct ifaddr *ifa = (struct ifaddr *)data; -#endif - bool avoid_reset = FALSE; - int error = 0; - - if (adapter->in_detach) - return (error); - - switch (command) { - case SIOCSIFADDR: -#ifdef INET - if (ifa->ifa_addr->sa_family == AF_INET) - avoid_reset = TRUE; -#endif -#ifdef INET6 - if (ifa->ifa_addr->sa_family == AF_INET6) - avoid_reset = TRUE; -#endif - /* - ** Calling init results in link renegotiation, - ** so we avoid doing it when possible. - */ - if (avoid_reset) { - if_setflagbits(ifp, IFF_UP, 0); - if (!(if_getdrvflags(ifp) & IFF_DRV_RUNNING)) - lem_init(adapter); -#ifdef INET - if (!(if_getflags(ifp) & IFF_NOARP)) - arp_ifinit(ifp, ifa); -#endif - } else - error = ether_ioctl(ifp, command, data); - break; - case SIOCSIFMTU: - { - int max_frame_size; - - IOCTL_DEBUGOUT("ioctl rcv'd: SIOCSIFMTU (Set Interface MTU)"); - - EM_CORE_LOCK(adapter); - switch (adapter->hw.mac.type) { - case e1000_82542: - max_frame_size = ETHER_MAX_LEN; - break; - default: - max_frame_size = MAX_JUMBO_FRAME_SIZE; - } - if (ifr->ifr_mtu > max_frame_size - ETHER_HDR_LEN - - ETHER_CRC_LEN) { - EM_CORE_UNLOCK(adapter); - error = EINVAL; - break; - } - - if_setmtu(ifp, ifr->ifr_mtu); - adapter->max_frame_size = - if_getmtu(ifp) + ETHER_HDR_LEN + ETHER_CRC_LEN; - if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) - lem_init_locked(adapter); - EM_CORE_UNLOCK(adapter); - break; - } - case SIOCSIFFLAGS: - IOCTL_DEBUGOUT("ioctl rcv'd:\ - SIOCSIFFLAGS (Set Interface Flags)"); - EM_CORE_LOCK(adapter); - if (if_getflags(ifp) & IFF_UP) { - if ((if_getdrvflags(ifp) & IFF_DRV_RUNNING)) { - if ((if_getflags(ifp) ^ adapter->if_flags) & - (IFF_PROMISC | IFF_ALLMULTI)) { - lem_disable_promisc(adapter); - lem_set_promisc(adapter); - } - } else - lem_init_locked(adapter); - } else - if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) { - EM_TX_LOCK(adapter); - lem_stop(adapter); - EM_TX_UNLOCK(adapter); - } - adapter->if_flags = if_getflags(ifp); - EM_CORE_UNLOCK(adapter); - break; - case SIOCADDMULTI: - case SIOCDELMULTI: - IOCTL_DEBUGOUT("ioctl rcv'd: SIOC(ADD|DEL)MULTI"); - if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) { - EM_CORE_LOCK(adapter); - lem_disable_intr(adapter); - lem_set_multi(adapter); - if (adapter->hw.mac.type == e1000_82542 && - adapter->hw.revision_id == E1000_REVISION_2) { - lem_initialize_receive_unit(adapter); - } -#ifdef DEVICE_POLLING - if (!(if_getcapenable(ifp) & IFCAP_POLLING)) -#endif - lem_enable_intr(adapter); - EM_CORE_UNLOCK(adapter); - } - break; - case SIOCSIFMEDIA: - /* Check SOL/IDER usage */ - EM_CORE_LOCK(adapter); - if (e1000_check_reset_block(&adapter->hw)) { - EM_CORE_UNLOCK(adapter); - device_printf(adapter->dev, "Media change is" - " blocked due to SOL/IDER session.\n"); - break; - } - EM_CORE_UNLOCK(adapter); - case SIOCGIFMEDIA: - IOCTL_DEBUGOUT("ioctl rcv'd: \ - SIOCxIFMEDIA (Get/Set Interface Media)"); - error = ifmedia_ioctl(ifp, ifr, &adapter->media, command); - break; - case SIOCSIFCAP: - { - int mask, reinit; - - IOCTL_DEBUGOUT("ioctl rcv'd: SIOCSIFCAP (Set Capabilities)"); - reinit = 0; - mask = ifr->ifr_reqcap ^ if_getcapenable(ifp); -#ifdef DEVICE_POLLING - if (mask & IFCAP_POLLING) { - if (ifr->ifr_reqcap & IFCAP_POLLING) { - error = ether_poll_register(lem_poll, ifp); - if (error) - return (error); - EM_CORE_LOCK(adapter); - lem_disable_intr(adapter); - if_setcapenablebit(ifp, IFCAP_POLLING, 0); - EM_CORE_UNLOCK(adapter); - } else { - error = ether_poll_deregister(ifp); - /* Enable interrupt even in error case */ - EM_CORE_LOCK(adapter); - lem_enable_intr(adapter); - if_setcapenablebit(ifp, 0, IFCAP_POLLING); - EM_CORE_UNLOCK(adapter); - } - } -#endif - if (mask & IFCAP_HWCSUM) { - if_togglecapenable(ifp, IFCAP_HWCSUM); - reinit = 1; - } - if (mask & IFCAP_VLAN_HWTAGGING) { - if_togglecapenable(ifp, IFCAP_VLAN_HWTAGGING); - reinit = 1; - } - if ((mask & IFCAP_WOL) && - (if_getcapabilities(ifp) & IFCAP_WOL) != 0) { - if (mask & IFCAP_WOL_MCAST) - if_togglecapenable(ifp, IFCAP_WOL_MCAST); - if (mask & IFCAP_WOL_MAGIC) - if_togglecapenable(ifp, IFCAP_WOL_MAGIC); - } - if (reinit && (if_getdrvflags(ifp) & IFF_DRV_RUNNING)) - lem_init(adapter); - if_vlancap(ifp); - break; - } - - default: - error = ether_ioctl(ifp, command, data); - break; - } - - return (error); -} - - -/********************************************************************* - * Init entry point - * - * This routine is used in two ways. It is used by the stack as - * init entry point in network interface structure. It is also used - * by the driver as a hw/sw initialization routine to get to a - * consistent state. - * - * return 0 on success, positive on failure - **********************************************************************/ - -static void -lem_init_locked(struct adapter *adapter) -{ - if_t ifp = adapter->ifp; - device_t dev = adapter->dev; - u32 pba; - - INIT_DEBUGOUT("lem_init: begin"); - - EM_CORE_LOCK_ASSERT(adapter); - - EM_TX_LOCK(adapter); - lem_stop(adapter); - EM_TX_UNLOCK(adapter); - - /* - * Packet Buffer Allocation (PBA) - * Writing PBA sets the receive portion of the buffer - * the remainder is used for the transmit buffer. - * - * Devices before the 82547 had a Packet Buffer of 64K. - * Default allocation: PBA=48K for Rx, leaving 16K for Tx. - * After the 82547 the buffer was reduced to 40K. - * Default allocation: PBA=30K for Rx, leaving 10K for Tx. - * Note: default does not leave enough room for Jumbo Frame >10k. - */ - switch (adapter->hw.mac.type) { - case e1000_82547: - case e1000_82547_rev_2: /* 82547: Total Packet Buffer is 40K */ - if (adapter->max_frame_size > 8192) - pba = E1000_PBA_22K; /* 22K for Rx, 18K for Tx */ - else - pba = E1000_PBA_30K; /* 30K for Rx, 10K for Tx */ - adapter->tx_fifo_head = 0; - adapter->tx_head_addr = pba << EM_TX_HEAD_ADDR_SHIFT; - adapter->tx_fifo_size = - (E1000_PBA_40K - pba) << EM_PBA_BYTES_SHIFT; - break; - default: - /* Devices before 82547 had a Packet Buffer of 64K. */ - if (adapter->max_frame_size > 8192) - pba = E1000_PBA_40K; /* 40K for Rx, 24K for Tx */ - else - pba = E1000_PBA_48K; /* 48K for Rx, 16K for Tx */ - } - - INIT_DEBUGOUT1("lem_init: pba=%dK",pba); - E1000_WRITE_REG(&adapter->hw, E1000_PBA, pba); - - /* Get the latest mac address, User can use a LAA */ - bcopy(if_getlladdr(adapter->ifp), adapter->hw.mac.addr, - ETHER_ADDR_LEN); - - /* Put the address into the Receive Address Array */ - e1000_rar_set(&adapter->hw, adapter->hw.mac.addr, 0); - - /* Initialize the hardware */ - if (lem_hardware_init(adapter)) { - device_printf(dev, "Unable to initialize the hardware\n"); - return; - } - lem_update_link_status(adapter); - - /* Setup VLAN support, basic and offload if available */ - E1000_WRITE_REG(&adapter->hw, E1000_VET, ETHERTYPE_VLAN); - - /* Set hardware offload abilities */ - if_clearhwassist(ifp); - if (adapter->hw.mac.type >= e1000_82543) { - if (if_getcapenable(ifp) & IFCAP_TXCSUM) - if_sethwassistbits(ifp, CSUM_TCP | CSUM_UDP, 0); - } - - /* Configure for OS presence */ - lem_init_manageability(adapter); - - /* Prepare transmit descriptors and buffers */ - lem_setup_transmit_structures(adapter); - lem_initialize_transmit_unit(adapter); - - /* Setup Multicast table */ - lem_set_multi(adapter); - - /* Prepare receive descriptors and buffers */ - if (lem_setup_receive_structures(adapter)) { - device_printf(dev, "Could not setup receive structures\n"); - EM_TX_LOCK(adapter); - lem_stop(adapter); - EM_TX_UNLOCK(adapter); - return; - } - lem_initialize_receive_unit(adapter); - - /* Use real VLAN Filter support? */ - if (if_getcapenable(ifp) & IFCAP_VLAN_HWTAGGING) { - if (if_getcapenable(ifp) & IFCAP_VLAN_HWFILTER) - /* Use real VLAN Filter support */ - lem_setup_vlan_hw_support(adapter); - else { - u32 ctrl; - ctrl = E1000_READ_REG(&adapter->hw, E1000_CTRL); - ctrl |= E1000_CTRL_VME; - E1000_WRITE_REG(&adapter->hw, E1000_CTRL, ctrl); - } - } - - /* Don't lose promiscuous settings */ - lem_set_promisc(adapter); - - if_setdrvflagbits(ifp, IFF_DRV_RUNNING, IFF_DRV_OACTIVE); - - callout_reset(&adapter->timer, hz, lem_local_timer, adapter); - e1000_clear_hw_cntrs_base_generic(&adapter->hw); - -#ifdef DEVICE_POLLING - /* - * Only enable interrupts if we are not polling, make sure - * they are off otherwise. - */ - if (if_getcapenable(ifp) & IFCAP_POLLING) - lem_disable_intr(adapter); - else -#endif /* DEVICE_POLLING */ - lem_enable_intr(adapter); - - /* AMT based hardware can now take control from firmware */ - if (adapter->has_manage && adapter->has_amt) - lem_get_hw_control(adapter); -} - -static void -lem_init(void *arg) -{ - struct adapter *adapter = arg; - - EM_CORE_LOCK(adapter); - lem_init_locked(adapter); - EM_CORE_UNLOCK(adapter); -} - - -#ifdef DEVICE_POLLING -/********************************************************************* - * - * Legacy polling routine - * - *********************************************************************/ -static int -lem_poll(if_t ifp, enum poll_cmd cmd, int count) -{ - struct adapter *adapter = if_getsoftc(ifp); - u32 reg_icr, rx_done = 0; - - EM_CORE_LOCK(adapter); - if ((if_getdrvflags(ifp) & IFF_DRV_RUNNING) == 0) { - EM_CORE_UNLOCK(adapter); - return (rx_done); - } - - if (cmd == POLL_AND_CHECK_STATUS) { - reg_icr = E1000_READ_REG(&adapter->hw, E1000_ICR); - if (reg_icr & (E1000_ICR_RXSEQ | E1000_ICR_LSC)) { - callout_stop(&adapter->timer); - adapter->hw.mac.get_link_status = 1; - lem_update_link_status(adapter); - callout_reset(&adapter->timer, hz, - lem_local_timer, adapter); - } - } - EM_CORE_UNLOCK(adapter); - - lem_rxeof(adapter, count, &rx_done); - - EM_TX_LOCK(adapter); - lem_txeof(adapter); - if(!if_sendq_empty(ifp)) - lem_start_locked(ifp); - EM_TX_UNLOCK(adapter); - return (rx_done); -} -#endif /* DEVICE_POLLING */ - -/********************************************************************* - * - * Legacy Interrupt Service routine - * - *********************************************************************/ -static void -lem_intr(void *arg) -{ - struct adapter *adapter = arg; - if_t ifp = adapter->ifp; - u32 reg_icr; - - - if ((if_getcapenable(ifp) & IFCAP_POLLING) || - ((if_getdrvflags(ifp) & IFF_DRV_RUNNING) == 0)) - return; - - EM_CORE_LOCK(adapter); - reg_icr = E1000_READ_REG(&adapter->hw, E1000_ICR); - if (reg_icr & E1000_ICR_RXO) - adapter->rx_overruns++; - - if ((reg_icr == 0xffffffff) || (reg_icr == 0)) { - EM_CORE_UNLOCK(adapter); - return; - } - - if (reg_icr & (E1000_ICR_RXSEQ | E1000_ICR_LSC)) { - callout_stop(&adapter->timer); - adapter->hw.mac.get_link_status = 1; - lem_update_link_status(adapter); - /* Deal with TX cruft when link lost */ - lem_tx_purge(adapter); - callout_reset(&adapter->timer, hz, - lem_local_timer, adapter); - EM_CORE_UNLOCK(adapter); - return; - } - - EM_CORE_UNLOCK(adapter); - lem_rxeof(adapter, -1, NULL); - - EM_TX_LOCK(adapter); - lem_txeof(adapter); - if ((if_getdrvflags(ifp) & IFF_DRV_RUNNING) && - (!if_sendq_empty(ifp))) - lem_start_locked(ifp); - EM_TX_UNLOCK(adapter); - return; -} - - -static void -lem_handle_link(void *context, int pending) -{ - struct adapter *adapter = context; - if_t ifp = adapter->ifp; - - if (!(if_getdrvflags(ifp) & IFF_DRV_RUNNING)) - return; - - EM_CORE_LOCK(adapter); - callout_stop(&adapter->timer); - lem_update_link_status(adapter); - /* Deal with TX cruft when link lost */ - lem_tx_purge(adapter); - callout_reset(&adapter->timer, hz, lem_local_timer, adapter); - EM_CORE_UNLOCK(adapter); -} - - -/* Combined RX/TX handler, used by Legacy and MSI */ -static void -lem_handle_rxtx(void *context, int pending) -{ - struct adapter *adapter = context; - if_t ifp = adapter->ifp; - - - if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) { - bool more = lem_rxeof(adapter, adapter->rx_process_limit, NULL); - EM_TX_LOCK(adapter); - lem_txeof(adapter); - if(!if_sendq_empty(ifp)) - lem_start_locked(ifp); - EM_TX_UNLOCK(adapter); - if (more) { - taskqueue_enqueue(adapter->tq, &adapter->rxtx_task); - return; - } - } - - if (if_getdrvflags(ifp) & IFF_DRV_RUNNING) - lem_enable_intr(adapter); -} - -/********************************************************************* - * - * Fast Legacy/MSI Combined Interrupt Service routine - * - *********************************************************************/ -static int -lem_irq_fast(void *arg) -{ - struct adapter *adapter = arg; - if_t ifp; - u32 reg_icr; - - ifp = adapter->ifp; - - reg_icr = E1000_READ_REG(&adapter->hw, E1000_ICR); - - /* Hot eject? */ - if (reg_icr == 0xffffffff) - return FILTER_STRAY; - - /* Definitely not our interrupt. */ - if (reg_icr == 0x0) - return FILTER_STRAY; - - /* - * Mask interrupts until the taskqueue is finished running. This is - * cheap, just assume that it is needed. This also works around the - * MSI message reordering errata on certain systems. - */ - lem_disable_intr(adapter); - taskqueue_enqueue(adapter->tq, &adapter->rxtx_task); - - /* Link status change */ - if (reg_icr & (E1000_ICR_RXSEQ | E1000_ICR_LSC)) { - adapter->hw.mac.get_link_status = 1; - taskqueue_enqueue(taskqueue_fast, &adapter->link_task); - } - - if (reg_icr & E1000_ICR_RXO) - adapter->rx_overruns++; - return FILTER_HANDLED; -} - - -/********************************************************************* - * - * Media Ioctl callback - * - * This routine is called whenever the user queries the status of - * the interface using ifconfig. - * - **********************************************************************/ -static void -lem_media_status(if_t ifp, struct ifmediareq *ifmr) -{ - struct adapter *adapter = if_getsoftc(ifp); - u_char fiber_type = IFM_1000_SX; - - INIT_DEBUGOUT("lem_media_status: begin"); - - EM_CORE_LOCK(adapter); - lem_update_link_status(adapter); - - ifmr->ifm_status = IFM_AVALID; - ifmr->ifm_active = IFM_ETHER; - - if (!adapter->link_active) { - EM_CORE_UNLOCK(adapter); - return; - } - - ifmr->ifm_status |= IFM_ACTIVE; - - if ((adapter->hw.phy.media_type == e1000_media_type_fiber) || - (adapter->hw.phy.media_type == e1000_media_type_internal_serdes)) { - if (adapter->hw.mac.type == e1000_82545) - fiber_type = IFM_1000_LX; - ifmr->ifm_active |= fiber_type | IFM_FDX; - } else { - switch (adapter->link_speed) { - case 10: - ifmr->ifm_active |= IFM_10_T; - break; - case 100: - ifmr->ifm_active |= IFM_100_TX; - break; - case 1000: - ifmr->ifm_active |= IFM_1000_T; - break; - } - if (adapter->link_duplex == FULL_DUPLEX) - ifmr->ifm_active |= IFM_FDX; - else - ifmr->ifm_active |= IFM_HDX; - } - EM_CORE_UNLOCK(adapter); -} - -/********************************************************************* - * - * Media Ioctl callback - * - * This routine is called when the user changes speed/duplex using - * media/mediopt option with ifconfig. - * - **********************************************************************/ -static int -lem_media_change(if_t ifp) -{ - struct adapter *adapter = if_getsoftc(ifp); - struct ifmedia *ifm = &adapter->media; - - INIT_DEBUGOUT("lem_media_change: begin"); - - if (IFM_TYPE(ifm->ifm_media) != IFM_ETHER) - return (EINVAL); - - EM_CORE_LOCK(adapter); - switch (IFM_SUBTYPE(ifm->ifm_media)) { - case IFM_AUTO: - adapter->hw.mac.autoneg = DO_AUTO_NEG; - adapter->hw.phy.autoneg_advertised = AUTONEG_ADV_DEFAULT; - break; - case IFM_1000_LX: - case IFM_1000_SX: - case IFM_1000_T: - adapter->hw.mac.autoneg = DO_AUTO_NEG; - adapter->hw.phy.autoneg_advertised = ADVERTISE_1000_FULL; - break; - case IFM_100_TX: - adapter->hw.mac.autoneg = FALSE; - adapter->hw.phy.autoneg_advertised = 0; - if ((ifm->ifm_media & IFM_GMASK) == IFM_FDX) - adapter->hw.mac.forced_speed_duplex = ADVERTISE_100_FULL; - else - adapter->hw.mac.forced_speed_duplex = ADVERTISE_100_HALF; - break; - case IFM_10_T: - adapter->hw.mac.autoneg = FALSE; - adapter->hw.phy.autoneg_advertised = 0; - if ((ifm->ifm_media & IFM_GMASK) == IFM_FDX) - adapter->hw.mac.forced_speed_duplex = ADVERTISE_10_FULL; - else - adapter->hw.mac.forced_speed_duplex = ADVERTISE_10_HALF; - break; - default: - device_printf(adapter->dev, "Unsupported media type\n"); - } - - lem_init_locked(adapter); - EM_CORE_UNLOCK(adapter); - - return (0); -} - -/********************************************************************* - * - * This routine maps the mbufs to tx descriptors. - * - * return 0 on success, positive on failure - **********************************************************************/ - -static int -lem_xmit(struct adapter *adapter, struct mbuf **m_headp) -{ - bus_dma_segment_t segs[EM_MAX_SCATTER]; - bus_dmamap_t map; - struct em_buffer *tx_buffer, *tx_buffer_mapped; - struct e1000_tx_desc *ctxd = NULL; - struct mbuf *m_head; - u32 txd_upper, txd_lower, txd_used, txd_saved; - int error, nsegs, i, j, first, last = 0; - - m_head = *m_headp; - txd_upper = txd_lower = txd_used = txd_saved = 0; - - /* - ** When doing checksum offload, it is critical to - ** make sure the first mbuf has more than header, - ** because that routine expects data to be present. - */ - if ((m_head->m_pkthdr.csum_flags & CSUM_OFFLOAD) && - (m_head->m_len < ETHER_HDR_LEN + sizeof(struct ip))) { - m_head = m_pullup(m_head, ETHER_HDR_LEN + sizeof(struct ip)); - *m_headp = m_head; - if (m_head == NULL) - return (ENOBUFS); - } - - /* - * Map the packet for DMA - * - * Capture the first descriptor index, - * this descriptor will have the index - * of the EOP which is the only one that - * now gets a DONE bit writeback. - */ - first = adapter->next_avail_tx_desc; - tx_buffer = &adapter->tx_buffer_area[first]; - tx_buffer_mapped = tx_buffer; - map = tx_buffer->map; - - error = bus_dmamap_load_mbuf_sg(adapter->txtag, map, - *m_headp, segs, &nsegs, BUS_DMA_NOWAIT); - - /* - * There are two types of errors we can (try) to handle: - * - EFBIG means the mbuf chain was too long and bus_dma ran - * out of segments. Defragment the mbuf chain and try again. - * - ENOMEM means bus_dma could not obtain enough bounce buffers - * at this point in time. Defer sending and try again later. - * All other errors, in particular EINVAL, are fatal and prevent the - * mbuf chain from ever going through. Drop it and report error. - */ - if (error == EFBIG) { - struct mbuf *m; - - m = m_collapse(*m_headp, M_NOWAIT, EM_MAX_SCATTER); - if (m == NULL) { - adapter->mbuf_defrag_failed++; - m_freem(*m_headp); - *m_headp = NULL; - return (ENOBUFS); - } - *m_headp = m; - - /* Try it again */ - error = bus_dmamap_load_mbuf_sg(adapter->txtag, map, - *m_headp, segs, &nsegs, BUS_DMA_NOWAIT); - - if (error) { - adapter->no_tx_dma_setup++; - m_freem(*m_headp); - *m_headp = NULL; - return (error); - } - } else if (error != 0) { - adapter->no_tx_dma_setup++; - return (error); - } - - if (adapter->num_tx_desc_avail < (nsegs + 2)) { - adapter->no_tx_desc_avail2++; - bus_dmamap_unload(adapter->txtag, map); - return (ENOBUFS); - } - m_head = *m_headp; - - /* Do hardware assists */ - if (m_head->m_pkthdr.csum_flags & CSUM_OFFLOAD) - lem_transmit_checksum_setup(adapter, m_head, - &txd_upper, &txd_lower); - - i = adapter->next_avail_tx_desc; - if (adapter->pcix_82544) - txd_saved = i; - - /* Set up our transmit descriptors */ - for (j = 0; j < nsegs; j++) { - bus_size_t seg_len; - bus_addr_t seg_addr; - /* If adapter is 82544 and on PCIX bus */ - if(adapter->pcix_82544) { - DESC_ARRAY desc_array; - u32 array_elements, counter; - /* - * Check the Address and Length combination and - * split the data accordingly - */ - array_elements = lem_fill_descriptors(segs[j].ds_addr, - segs[j].ds_len, &desc_array); - for (counter = 0; counter < array_elements; counter++) { - if (txd_used == adapter->num_tx_desc_avail) { - adapter->next_avail_tx_desc = txd_saved; - adapter->no_tx_desc_avail2++; - bus_dmamap_unload(adapter->txtag, map); - return (ENOBUFS); - } - tx_buffer = &adapter->tx_buffer_area[i]; - ctxd = &adapter->tx_desc_base[i]; - ctxd->buffer_addr = htole64( - desc_array.descriptor[counter].address); - ctxd->lower.data = htole32( - (adapter->txd_cmd | txd_lower | (u16) - desc_array.descriptor[counter].length)); - ctxd->upper.data = - htole32((txd_upper)); - last = i; - if (++i == adapter->num_tx_desc) - i = 0; - tx_buffer->m_head = NULL; - tx_buffer->next_eop = -1; - txd_used++; - } - } else { - tx_buffer = &adapter->tx_buffer_area[i]; - ctxd = &adapter->tx_desc_base[i]; - seg_addr = segs[j].ds_addr; - seg_len = segs[j].ds_len; - ctxd->buffer_addr = htole64(seg_addr); - ctxd->lower.data = htole32( - adapter->txd_cmd | txd_lower | seg_len); - ctxd->upper.data = - htole32(txd_upper); - last = i; - if (++i == adapter->num_tx_desc) - i = 0; - tx_buffer->m_head = NULL; - tx_buffer->next_eop = -1; - } - } - - adapter->next_avail_tx_desc = i; - - if (adapter->pcix_82544) - adapter->num_tx_desc_avail -= txd_used; - else - adapter->num_tx_desc_avail -= nsegs; - - if (m_head->m_flags & M_VLANTAG) { - /* Set the vlan id. */ - ctxd->upper.fields.special = - htole16(m_head->m_pkthdr.ether_vtag); - /* Tell hardware to add tag */ - ctxd->lower.data |= htole32(E1000_TXD_CMD_VLE); - } - - tx_buffer->m_head = m_head; - tx_buffer_mapped->map = tx_buffer->map; - tx_buffer->map = map; - bus_dmamap_sync(adapter->txtag, map, BUS_DMASYNC_PREWRITE); - - /* - * Last Descriptor of Packet - * needs End Of Packet (EOP) - * and Report Status (RS) - */ - ctxd->lower.data |= - htole32(E1000_TXD_CMD_EOP | E1000_TXD_CMD_RS); - /* - * Keep track in the first buffer which - * descriptor will be written back - */ - tx_buffer = &adapter->tx_buffer_area[first]; - tx_buffer->next_eop = last; - adapter->watchdog_time = ticks; - - /* - * Advance the Transmit Descriptor Tail (TDT), this tells the E1000 - * that this frame is available to transmit. - */ - bus_dmamap_sync(adapter->txdma.dma_tag, adapter->txdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - -#ifdef NIC_SEND_COMBINING - if (adapter->sc_enable) { - if (adapter->shadow_tdt & MIT_PENDING_INT) { - /* signal intr and data pending */ - adapter->shadow_tdt = MIT_PENDING_TDT | (i & 0xffff); - return (0); - } else { - adapter->shadow_tdt = MIT_PENDING_INT; - } - } -#endif /* NIC_SEND_COMBINING */ - - if (adapter->hw.mac.type == e1000_82547 && - adapter->link_duplex == HALF_DUPLEX) - lem_82547_move_tail(adapter); - else { - E1000_WRITE_REG(&adapter->hw, E1000_TDT(0), i); - if (adapter->hw.mac.type == e1000_82547) - lem_82547_update_fifo_head(adapter, - m_head->m_pkthdr.len); - } - - return (0); -} - -/********************************************************************* - * - * 82547 workaround to avoid controller hang in half-duplex environment. - * The workaround is to avoid queuing a large packet that would span - * the internal Tx FIFO ring boundary. We need to reset the FIFO pointers - * in this case. We do that only when FIFO is quiescent. - * - **********************************************************************/ -static void -lem_82547_move_tail(void *arg) -{ - struct adapter *adapter = arg; - struct e1000_tx_desc *tx_desc; - u16 hw_tdt, sw_tdt, length = 0; - bool eop = 0; - - EM_TX_LOCK_ASSERT(adapter); - - hw_tdt = E1000_READ_REG(&adapter->hw, E1000_TDT(0)); - sw_tdt = adapter->next_avail_tx_desc; - - while (hw_tdt != sw_tdt) { - tx_desc = &adapter->tx_desc_base[hw_tdt]; - length += tx_desc->lower.flags.length; - eop = tx_desc->lower.data & E1000_TXD_CMD_EOP; - if (++hw_tdt == adapter->num_tx_desc) - hw_tdt = 0; - - if (eop) { - if (lem_82547_fifo_workaround(adapter, length)) { - adapter->tx_fifo_wrk_cnt++; - callout_reset(&adapter->tx_fifo_timer, 1, - lem_82547_move_tail, adapter); - break; - } - E1000_WRITE_REG(&adapter->hw, E1000_TDT(0), hw_tdt); - lem_82547_update_fifo_head(adapter, length); - length = 0; - } - } -} - -static int -lem_82547_fifo_workaround(struct adapter *adapter, int len) -{ - int fifo_space, fifo_pkt_len; - - fifo_pkt_len = roundup2(len + EM_FIFO_HDR, EM_FIFO_HDR); - - if (adapter->link_duplex == HALF_DUPLEX) { - fifo_space = adapter->tx_fifo_size - adapter->tx_fifo_head; - - if (fifo_pkt_len >= (EM_82547_PKT_THRESH + fifo_space)) { - if (lem_82547_tx_fifo_reset(adapter)) - return (0); - else - return (1); - } - } - - return (0); -} - -static void -lem_82547_update_fifo_head(struct adapter *adapter, int len) -{ - int fifo_pkt_len = roundup2(len + EM_FIFO_HDR, EM_FIFO_HDR); - - /* tx_fifo_head is always 16 byte aligned */ - adapter->tx_fifo_head += fifo_pkt_len; - if (adapter->tx_fifo_head >= adapter->tx_fifo_size) { - adapter->tx_fifo_head -= adapter->tx_fifo_size; - } -} - - -static int -lem_82547_tx_fifo_reset(struct adapter *adapter) -{ - u32 tctl; - - if ((E1000_READ_REG(&adapter->hw, E1000_TDT(0)) == - E1000_READ_REG(&adapter->hw, E1000_TDH(0))) && - (E1000_READ_REG(&adapter->hw, E1000_TDFT) == - E1000_READ_REG(&adapter->hw, E1000_TDFH)) && - (E1000_READ_REG(&adapter->hw, E1000_TDFTS) == - E1000_READ_REG(&adapter->hw, E1000_TDFHS)) && - (E1000_READ_REG(&adapter->hw, E1000_TDFPC) == 0)) { - /* Disable TX unit */ - tctl = E1000_READ_REG(&adapter->hw, E1000_TCTL); - E1000_WRITE_REG(&adapter->hw, E1000_TCTL, - tctl & ~E1000_TCTL_EN); - - /* Reset FIFO pointers */ - E1000_WRITE_REG(&adapter->hw, E1000_TDFT, - adapter->tx_head_addr); - E1000_WRITE_REG(&adapter->hw, E1000_TDFH, - adapter->tx_head_addr); - E1000_WRITE_REG(&adapter->hw, E1000_TDFTS, - adapter->tx_head_addr); - E1000_WRITE_REG(&adapter->hw, E1000_TDFHS, - adapter->tx_head_addr); - - /* Re-enable TX unit */ - E1000_WRITE_REG(&adapter->hw, E1000_TCTL, tctl); - E1000_WRITE_FLUSH(&adapter->hw); - - adapter->tx_fifo_head = 0; - adapter->tx_fifo_reset_cnt++; - - return (TRUE); - } - else { - return (FALSE); - } -} - -static void -lem_set_promisc(struct adapter *adapter) -{ - if_t ifp = adapter->ifp; - u32 reg_rctl; - - reg_rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); - - if (if_getflags(ifp) & IFF_PROMISC) { - reg_rctl |= (E1000_RCTL_UPE | E1000_RCTL_MPE); - /* Turn this on if you want to see bad packets */ - if (lem_debug_sbp) - reg_rctl |= E1000_RCTL_SBP; - E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); - } else if (if_getflags(ifp) & IFF_ALLMULTI) { - reg_rctl |= E1000_RCTL_MPE; - reg_rctl &= ~E1000_RCTL_UPE; - E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); - } -} - -static void -lem_disable_promisc(struct adapter *adapter) -{ - if_t ifp = adapter->ifp; - u32 reg_rctl; - int mcnt = 0; - - reg_rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); - reg_rctl &= (~E1000_RCTL_UPE); - if (if_getflags(ifp) & IFF_ALLMULTI) - mcnt = MAX_NUM_MULTICAST_ADDRESSES; - else - mcnt = if_multiaddr_count(ifp, MAX_NUM_MULTICAST_ADDRESSES); - - /* Don't disable if in MAX groups */ - if (mcnt < MAX_NUM_MULTICAST_ADDRESSES) - reg_rctl &= (~E1000_RCTL_MPE); - reg_rctl &= (~E1000_RCTL_SBP); - E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); -} - - -/********************************************************************* - * Multicast Update - * - * This routine is called whenever multicast address list is updated. - * - **********************************************************************/ - -static void -lem_set_multi(struct adapter *adapter) -{ - if_t ifp = adapter->ifp; - u32 reg_rctl = 0; - u8 *mta; /* Multicast array memory */ - int mcnt = 0; - - IOCTL_DEBUGOUT("lem_set_multi: begin"); - - mta = adapter->mta; - bzero(mta, sizeof(u8) * ETH_ADDR_LEN * MAX_NUM_MULTICAST_ADDRESSES); - - if (adapter->hw.mac.type == e1000_82542 && - adapter->hw.revision_id == E1000_REVISION_2) { - reg_rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); - if (adapter->hw.bus.pci_cmd_word & CMD_MEM_WRT_INVALIDATE) - e1000_pci_clear_mwi(&adapter->hw); - reg_rctl |= E1000_RCTL_RST; - E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); - msec_delay(5); - } - - if_multiaddr_array(ifp, mta, &mcnt, MAX_NUM_MULTICAST_ADDRESSES); - - if (mcnt >= MAX_NUM_MULTICAST_ADDRESSES) { - reg_rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); - reg_rctl |= E1000_RCTL_MPE; - E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); - } else - e1000_update_mc_addr_list(&adapter->hw, mta, mcnt); - - if (adapter->hw.mac.type == e1000_82542 && - adapter->hw.revision_id == E1000_REVISION_2) { - reg_rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); - reg_rctl &= ~E1000_RCTL_RST; - E1000_WRITE_REG(&adapter->hw, E1000_RCTL, reg_rctl); - msec_delay(5); - if (adapter->hw.bus.pci_cmd_word & CMD_MEM_WRT_INVALIDATE) - e1000_pci_set_mwi(&adapter->hw); - } -} - - -/********************************************************************* - * Timer routine - * - * This routine checks for link status and updates statistics. - * - **********************************************************************/ - -static void -lem_local_timer(void *arg) -{ - struct adapter *adapter = arg; - - EM_CORE_LOCK_ASSERT(adapter); - - lem_update_link_status(adapter); - lem_update_stats_counters(adapter); - - lem_smartspeed(adapter); - - /* - * We check the watchdog: the time since - * the last TX descriptor was cleaned. - * This implies a functional TX engine. - */ - if ((adapter->watchdog_check == TRUE) && - (ticks - adapter->watchdog_time > EM_WATCHDOG)) - goto hung; - - callout_reset(&adapter->timer, hz, lem_local_timer, adapter); - return; -hung: - device_printf(adapter->dev, "Watchdog timeout -- resetting\n"); - if_setdrvflagbits(adapter->ifp, 0, IFF_DRV_RUNNING); - adapter->watchdog_events++; - lem_init_locked(adapter); -} - -static void -lem_update_link_status(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - if_t ifp = adapter->ifp; - device_t dev = adapter->dev; - u32 link_check = 0; - - /* Get the cached link value or read phy for real */ - switch (hw->phy.media_type) { - case e1000_media_type_copper: - if (hw->mac.get_link_status) { - /* Do the work to read phy */ - e1000_check_for_link(hw); - link_check = !hw->mac.get_link_status; - if (link_check) /* ESB2 fix */ - e1000_cfg_on_link_up(hw); - } else - link_check = TRUE; - break; - case e1000_media_type_fiber: - e1000_check_for_link(hw); - link_check = (E1000_READ_REG(hw, E1000_STATUS) & - E1000_STATUS_LU); - break; - case e1000_media_type_internal_serdes: - e1000_check_for_link(hw); - link_check = adapter->hw.mac.serdes_has_link; - break; - default: - case e1000_media_type_unknown: - break; - } - - /* Now check for a transition */ - if (link_check && (adapter->link_active == 0)) { - e1000_get_speed_and_duplex(hw, &adapter->link_speed, - &adapter->link_duplex); - if (bootverbose) - device_printf(dev, "Link is up %d Mbps %s\n", - adapter->link_speed, - ((adapter->link_duplex == FULL_DUPLEX) ? - "Full Duplex" : "Half Duplex")); - adapter->link_active = 1; - adapter->smartspeed = 0; - if_setbaudrate(ifp, adapter->link_speed * 1000000); - if_link_state_change(ifp, LINK_STATE_UP); - } else if (!link_check && (adapter->link_active == 1)) { - if_setbaudrate(ifp, 0); - adapter->link_speed = 0; - adapter->link_duplex = 0; - if (bootverbose) - device_printf(dev, "Link is Down\n"); - adapter->link_active = 0; - /* Link down, disable watchdog */ - adapter->watchdog_check = FALSE; - if_link_state_change(ifp, LINK_STATE_DOWN); - } -} - -/********************************************************************* - * - * This routine disables all traffic on the adapter by issuing a - * global reset on the MAC and deallocates TX/RX buffers. - * - * This routine should always be called with BOTH the CORE - * and TX locks. - **********************************************************************/ - -static void -lem_stop(void *arg) -{ - struct adapter *adapter = arg; - if_t ifp = adapter->ifp; - - EM_CORE_LOCK_ASSERT(adapter); - EM_TX_LOCK_ASSERT(adapter); - - INIT_DEBUGOUT("lem_stop: begin"); - - lem_disable_intr(adapter); - callout_stop(&adapter->timer); - callout_stop(&adapter->tx_fifo_timer); - - /* Tell the stack that the interface is no longer active */ - if_setdrvflagbits(ifp, 0, (IFF_DRV_RUNNING | IFF_DRV_OACTIVE)); - - e1000_reset_hw(&adapter->hw); - if (adapter->hw.mac.type >= e1000_82544) - E1000_WRITE_REG(&adapter->hw, E1000_WUC, 0); - - e1000_led_off(&adapter->hw); - e1000_cleanup_led(&adapter->hw); -} - - -/********************************************************************* - * - * Determine hardware revision. - * - **********************************************************************/ -static void -lem_identify_hardware(struct adapter *adapter) -{ - device_t dev = adapter->dev; - - /* Make sure our PCI config space has the necessary stuff set */ - pci_enable_busmaster(dev); - adapter->hw.bus.pci_cmd_word = pci_read_config(dev, PCIR_COMMAND, 2); - - /* Save off the information about this board */ - adapter->hw.vendor_id = pci_get_vendor(dev); - adapter->hw.device_id = pci_get_device(dev); - adapter->hw.revision_id = pci_read_config(dev, PCIR_REVID, 1); - adapter->hw.subsystem_vendor_id = - pci_read_config(dev, PCIR_SUBVEND_0, 2); - adapter->hw.subsystem_device_id = - pci_read_config(dev, PCIR_SUBDEV_0, 2); - - /* Do Shared Code Init and Setup */ - if (e1000_set_mac_type(&adapter->hw)) { - device_printf(dev, "Setup init failure\n"); - return; - } -} - -static int -lem_allocate_pci_resources(struct adapter *adapter) -{ - device_t dev = adapter->dev; - int val, rid, error = E1000_SUCCESS; - - rid = PCIR_BAR(0); - adapter->memory = bus_alloc_resource_any(dev, SYS_RES_MEMORY, - &rid, RF_ACTIVE); - if (adapter->memory == NULL) { - device_printf(dev, "Unable to allocate bus resource: memory\n"); - return (ENXIO); - } - adapter->osdep.mem_bus_space_tag = - rman_get_bustag(adapter->memory); - adapter->osdep.mem_bus_space_handle = - rman_get_bushandle(adapter->memory); - adapter->hw.hw_addr = (u8 *)&adapter->osdep.mem_bus_space_handle; - - /* Only older adapters use IO mapping */ - if (adapter->hw.mac.type > e1000_82543) { - /* Figure our where our IO BAR is ? */ - for (rid = PCIR_BAR(0); rid < PCIR_CIS;) { - val = pci_read_config(dev, rid, 4); - if (EM_BAR_TYPE(val) == EM_BAR_TYPE_IO) { - adapter->io_rid = rid; - break; - } - rid += 4; - /* check for 64bit BAR */ - if (EM_BAR_MEM_TYPE(val) == EM_BAR_MEM_TYPE_64BIT) - rid += 4; - } - if (rid >= PCIR_CIS) { - device_printf(dev, "Unable to locate IO BAR\n"); - return (ENXIO); - } - adapter->ioport = bus_alloc_resource_any(dev, - SYS_RES_IOPORT, &adapter->io_rid, RF_ACTIVE); - if (adapter->ioport == NULL) { - device_printf(dev, "Unable to allocate bus resource: " - "ioport\n"); - return (ENXIO); - } - adapter->hw.io_base = 0; - adapter->osdep.io_bus_space_tag = - rman_get_bustag(adapter->ioport); - adapter->osdep.io_bus_space_handle = - rman_get_bushandle(adapter->ioport); - } - - adapter->hw.back = &adapter->osdep; - - return (error); -} - -/********************************************************************* - * - * Setup the Legacy or MSI Interrupt handler - * - **********************************************************************/ -int -lem_allocate_irq(struct adapter *adapter) -{ - device_t dev = adapter->dev; - int error, rid = 0; - - /* Manually turn off all interrupts */ - E1000_WRITE_REG(&adapter->hw, E1000_IMC, 0xffffffff); - - /* We allocate a single interrupt resource */ - adapter->res[0] = bus_alloc_resource_any(dev, - SYS_RES_IRQ, &rid, RF_SHAREABLE | RF_ACTIVE); - if (adapter->res[0] == NULL) { - device_printf(dev, "Unable to allocate bus resource: " - "interrupt\n"); - return (ENXIO); - } - - /* Do Legacy setup? */ - if (lem_use_legacy_irq) { - if ((error = bus_setup_intr(dev, adapter->res[0], - INTR_TYPE_NET | INTR_MPSAFE, NULL, lem_intr, adapter, - &adapter->tag[0])) != 0) { - device_printf(dev, - "Failed to register interrupt handler"); - return (error); - } - return (0); - } - - /* - * Use a Fast interrupt and the associated - * deferred processing contexts. - */ - TASK_INIT(&adapter->rxtx_task, 0, lem_handle_rxtx, adapter); - TASK_INIT(&adapter->link_task, 0, lem_handle_link, adapter); - adapter->tq = taskqueue_create_fast("lem_taskq", M_NOWAIT, - taskqueue_thread_enqueue, &adapter->tq); - taskqueue_start_threads(&adapter->tq, 1, PI_NET, "%s taskq", - device_get_nameunit(adapter->dev)); - if ((error = bus_setup_intr(dev, adapter->res[0], - INTR_TYPE_NET, lem_irq_fast, NULL, adapter, - &adapter->tag[0])) != 0) { - device_printf(dev, "Failed to register fast interrupt " - "handler: %d\n", error); - taskqueue_free(adapter->tq); - adapter->tq = NULL; - return (error); - } - - return (0); -} - - -static void -lem_free_pci_resources(struct adapter *adapter) -{ - device_t dev = adapter->dev; - - - if (adapter->tag[0] != NULL) { - bus_teardown_intr(dev, adapter->res[0], - adapter->tag[0]); - adapter->tag[0] = NULL; - } - - if (adapter->res[0] != NULL) { - bus_release_resource(dev, SYS_RES_IRQ, - 0, adapter->res[0]); - } - - if (adapter->memory != NULL) - bus_release_resource(dev, SYS_RES_MEMORY, - PCIR_BAR(0), adapter->memory); - - if (adapter->ioport != NULL) - bus_release_resource(dev, SYS_RES_IOPORT, - adapter->io_rid, adapter->ioport); -} - - -/********************************************************************* - * - * Initialize the hardware to a configuration - * as specified by the adapter structure. - * - **********************************************************************/ -static int -lem_hardware_init(struct adapter *adapter) -{ - device_t dev = adapter->dev; - u16 rx_buffer_size; - - INIT_DEBUGOUT("lem_hardware_init: begin"); - - /* Issue a global reset */ - e1000_reset_hw(&adapter->hw); - - /* When hardware is reset, fifo_head is also reset */ - adapter->tx_fifo_head = 0; - - /* - * These parameters control the automatic generation (Tx) and - * response (Rx) to Ethernet PAUSE frames. - * - High water mark should allow for at least two frames to be - * received after sending an XOFF. - * - Low water mark works best when it is very near the high water mark. - * This allows the receiver to restart by sending XON when it has - * drained a bit. Here we use an arbitrary value of 1500 which will - * restart after one full frame is pulled from the buffer. There - * could be several smaller frames in the buffer and if so they will - * not trigger the XON until their total number reduces the buffer - * by 1500. - * - The pause time is fairly large at 1000 x 512ns = 512 usec. - */ - rx_buffer_size = ((E1000_READ_REG(&adapter->hw, E1000_PBA) & - 0xffff) << 10 ); - - adapter->hw.fc.high_water = rx_buffer_size - - roundup2(adapter->max_frame_size, 1024); - adapter->hw.fc.low_water = adapter->hw.fc.high_water - 1500; - - adapter->hw.fc.pause_time = EM_FC_PAUSE_TIME; - adapter->hw.fc.send_xon = TRUE; - - /* Set Flow control, use the tunable location if sane */ - if ((lem_fc_setting >= 0) && (lem_fc_setting < 4)) - adapter->hw.fc.requested_mode = lem_fc_setting; - else - adapter->hw.fc.requested_mode = e1000_fc_none; - - if (e1000_init_hw(&adapter->hw) < 0) { - device_printf(dev, "Hardware Initialization Failed\n"); - return (EIO); - } - - e1000_check_for_link(&adapter->hw); - - return (0); -} - -/********************************************************************* - * - * Setup networking device structure and register an interface. - * - **********************************************************************/ -static int -lem_setup_interface(device_t dev, struct adapter *adapter) -{ - if_t ifp; - - INIT_DEBUGOUT("lem_setup_interface: begin"); - - ifp = adapter->ifp = if_gethandle(IFT_ETHER); - if (ifp == (void *)NULL) { - device_printf(dev, "can not allocate ifnet structure\n"); - return (-1); - } - if_initname(ifp, device_get_name(dev), device_get_unit(dev)); - if_setinitfn(ifp, lem_init); - if_setsoftc(ifp, adapter); - if_setflags(ifp, IFF_BROADCAST | IFF_SIMPLEX | IFF_MULTICAST); - if_setioctlfn(ifp, lem_ioctl); - if_setstartfn(ifp, lem_start); - if_setgetcounterfn(ifp, lem_get_counter); - if_setsendqlen(ifp, adapter->num_tx_desc - 1); - if_setsendqready(ifp); - - ether_ifattach(ifp, adapter->hw.mac.addr); - - if_setcapabilities(ifp, 0); - - if (adapter->hw.mac.type >= e1000_82543) { - if_setcapabilitiesbit(ifp, IFCAP_HWCSUM | IFCAP_VLAN_HWCSUM, 0); - if_setcapenablebit(ifp, IFCAP_HWCSUM | IFCAP_VLAN_HWCSUM, 0); - } - - /* - * Tell the upper layer(s) we support long frames. - */ - if_setifheaderlen(ifp, sizeof(struct ether_vlan_header)); - if_setcapabilitiesbit(ifp, IFCAP_VLAN_HWTAGGING | IFCAP_VLAN_MTU, 0); - if_setcapenablebit(ifp, IFCAP_VLAN_HWTAGGING | IFCAP_VLAN_MTU, 0); - - /* - ** Dont turn this on by default, if vlans are - ** created on another pseudo device (eg. lagg) - ** then vlan events are not passed thru, breaking - ** operation, but with HW FILTER off it works. If - ** using vlans directly on the em driver you can - ** enable this and get full hardware tag filtering. - */ - if_setcapabilitiesbit(ifp, IFCAP_VLAN_HWFILTER, 0); - -#ifdef DEVICE_POLLING - if_setcapabilitiesbit(ifp, IFCAP_POLLING, 0); -#endif - - /* Enable only WOL MAGIC by default */ - if (adapter->wol) { - if_setcapabilitiesbit(ifp, IFCAP_WOL, 0); - if_setcapenablebit(ifp, IFCAP_WOL_MAGIC, 0); - } - - /* - * Specify the media types supported by this adapter and register - * callbacks to update media and link information - */ - ifmedia_init(&adapter->media, IFM_IMASK, - lem_media_change, lem_media_status); - if ((adapter->hw.phy.media_type == e1000_media_type_fiber) || - (adapter->hw.phy.media_type == e1000_media_type_internal_serdes)) { - u_char fiber_type = IFM_1000_SX; /* default type */ - - if (adapter->hw.mac.type == e1000_82545) - fiber_type = IFM_1000_LX; - ifmedia_add(&adapter->media, IFM_ETHER | fiber_type | IFM_FDX, - 0, NULL); - ifmedia_add(&adapter->media, IFM_ETHER | fiber_type, 0, NULL); - } else { - ifmedia_add(&adapter->media, IFM_ETHER | IFM_10_T, 0, NULL); - ifmedia_add(&adapter->media, IFM_ETHER | IFM_10_T | IFM_FDX, - 0, NULL); - ifmedia_add(&adapter->media, IFM_ETHER | IFM_100_TX, - 0, NULL); - ifmedia_add(&adapter->media, IFM_ETHER | IFM_100_TX | IFM_FDX, - 0, NULL); - if (adapter->hw.phy.type != e1000_phy_ife) { - ifmedia_add(&adapter->media, - IFM_ETHER | IFM_1000_T | IFM_FDX, 0, NULL); - ifmedia_add(&adapter->media, - IFM_ETHER | IFM_1000_T, 0, NULL); - } - } - ifmedia_add(&adapter->media, IFM_ETHER | IFM_AUTO, 0, NULL); - ifmedia_set(&adapter->media, IFM_ETHER | IFM_AUTO); - return (0); -} - - -/********************************************************************* - * - * Workaround for SmartSpeed on 82541 and 82547 controllers - * - **********************************************************************/ -static void -lem_smartspeed(struct adapter *adapter) -{ - u16 phy_tmp; - - if (adapter->link_active || (adapter->hw.phy.type != e1000_phy_igp) || - adapter->hw.mac.autoneg == 0 || - (adapter->hw.phy.autoneg_advertised & ADVERTISE_1000_FULL) == 0) - return; - - if (adapter->smartspeed == 0) { - /* If Master/Slave config fault is asserted twice, - * we assume back-to-back */ - e1000_read_phy_reg(&adapter->hw, PHY_1000T_STATUS, &phy_tmp); - if (!(phy_tmp & SR_1000T_MS_CONFIG_FAULT)) - return; - e1000_read_phy_reg(&adapter->hw, PHY_1000T_STATUS, &phy_tmp); - if (phy_tmp & SR_1000T_MS_CONFIG_FAULT) { - e1000_read_phy_reg(&adapter->hw, - PHY_1000T_CTRL, &phy_tmp); - if(phy_tmp & CR_1000T_MS_ENABLE) { - phy_tmp &= ~CR_1000T_MS_ENABLE; - e1000_write_phy_reg(&adapter->hw, - PHY_1000T_CTRL, phy_tmp); - adapter->smartspeed++; - if(adapter->hw.mac.autoneg && - !e1000_copper_link_autoneg(&adapter->hw) && - !e1000_read_phy_reg(&adapter->hw, - PHY_CONTROL, &phy_tmp)) { - phy_tmp |= (MII_CR_AUTO_NEG_EN | - MII_CR_RESTART_AUTO_NEG); - e1000_write_phy_reg(&adapter->hw, - PHY_CONTROL, phy_tmp); - } - } - } - return; - } else if(adapter->smartspeed == EM_SMARTSPEED_DOWNSHIFT) { - /* If still no link, perhaps using 2/3 pair cable */ - e1000_read_phy_reg(&adapter->hw, PHY_1000T_CTRL, &phy_tmp); - phy_tmp |= CR_1000T_MS_ENABLE; - e1000_write_phy_reg(&adapter->hw, PHY_1000T_CTRL, phy_tmp); - if(adapter->hw.mac.autoneg && - !e1000_copper_link_autoneg(&adapter->hw) && - !e1000_read_phy_reg(&adapter->hw, PHY_CONTROL, &phy_tmp)) { - phy_tmp |= (MII_CR_AUTO_NEG_EN | - MII_CR_RESTART_AUTO_NEG); - e1000_write_phy_reg(&adapter->hw, PHY_CONTROL, phy_tmp); - } - } - /* Restart process after EM_SMARTSPEED_MAX iterations */ - if(adapter->smartspeed++ == EM_SMARTSPEED_MAX) - adapter->smartspeed = 0; -} - - -/* - * Manage DMA'able memory. - */ -static void -lem_dmamap_cb(void *arg, bus_dma_segment_t *segs, int nseg, int error) -{ - if (error) - return; - *(bus_addr_t *) arg = segs[0].ds_addr; -} - -static int -lem_dma_malloc(struct adapter *adapter, bus_size_t size, - struct em_dma_alloc *dma, int mapflags) -{ - int error; - - error = bus_dma_tag_create(bus_get_dma_tag(adapter->dev), /* parent */ - EM_DBA_ALIGN, 0, /* alignment, bounds */ - BUS_SPACE_MAXADDR, /* lowaddr */ - BUS_SPACE_MAXADDR, /* highaddr */ - NULL, NULL, /* filter, filterarg */ - size, /* maxsize */ - 1, /* nsegments */ - size, /* maxsegsize */ - 0, /* flags */ - NULL, /* lockfunc */ - NULL, /* lockarg */ - &dma->dma_tag); - if (error) { - device_printf(adapter->dev, - "%s: bus_dma_tag_create failed: %d\n", - __func__, error); - goto fail_0; - } - - error = bus_dmamem_alloc(dma->dma_tag, (void**) &dma->dma_vaddr, - BUS_DMA_NOWAIT | BUS_DMA_COHERENT, &dma->dma_map); - if (error) { - device_printf(adapter->dev, - "%s: bus_dmamem_alloc(%ju) failed: %d\n", - __func__, (uintmax_t)size, error); - goto fail_2; - } - - dma->dma_paddr = 0; - error = bus_dmamap_load(dma->dma_tag, dma->dma_map, dma->dma_vaddr, - size, lem_dmamap_cb, &dma->dma_paddr, mapflags | BUS_DMA_NOWAIT); - if (error || dma->dma_paddr == 0) { - device_printf(adapter->dev, - "%s: bus_dmamap_load failed: %d\n", - __func__, error); - goto fail_3; - } - - return (0); - -fail_3: - bus_dmamap_unload(dma->dma_tag, dma->dma_map); -fail_2: - bus_dmamem_free(dma->dma_tag, dma->dma_vaddr, dma->dma_map); - bus_dma_tag_destroy(dma->dma_tag); -fail_0: - dma->dma_tag = NULL; - - return (error); -} - -static void -lem_dma_free(struct adapter *adapter, struct em_dma_alloc *dma) -{ - if (dma->dma_tag == NULL) - return; - if (dma->dma_paddr != 0) { - bus_dmamap_sync(dma->dma_tag, dma->dma_map, - BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE); - bus_dmamap_unload(dma->dma_tag, dma->dma_map); - dma->dma_paddr = 0; - } - if (dma->dma_vaddr != NULL) { - bus_dmamem_free(dma->dma_tag, dma->dma_vaddr, dma->dma_map); - dma->dma_vaddr = NULL; - } - bus_dma_tag_destroy(dma->dma_tag); - dma->dma_tag = NULL; -} - - -/********************************************************************* - * - * Allocate memory for tx_buffer structures. The tx_buffer stores all - * the information needed to transmit a packet on the wire. - * - **********************************************************************/ -static int -lem_allocate_transmit_structures(struct adapter *adapter) -{ - device_t dev = adapter->dev; - struct em_buffer *tx_buffer; - int error; - - /* - * Create DMA tags for tx descriptors - */ - if ((error = bus_dma_tag_create(bus_get_dma_tag(dev), /* parent */ - 1, 0, /* alignment, bounds */ - BUS_SPACE_MAXADDR, /* lowaddr */ - BUS_SPACE_MAXADDR, /* highaddr */ - NULL, NULL, /* filter, filterarg */ - MCLBYTES * EM_MAX_SCATTER, /* maxsize */ - EM_MAX_SCATTER, /* nsegments */ - MCLBYTES, /* maxsegsize */ - 0, /* flags */ - NULL, /* lockfunc */ - NULL, /* lockarg */ - &adapter->txtag)) != 0) { - device_printf(dev, "Unable to allocate TX DMA tag\n"); - goto fail; - } - - adapter->tx_buffer_area = malloc(sizeof(struct em_buffer) * - adapter->num_tx_desc, M_DEVBUF, M_NOWAIT | M_ZERO); - if (adapter->tx_buffer_area == NULL) { - device_printf(dev, "Unable to allocate tx_buffer memory\n"); - error = ENOMEM; - goto fail; - } - - /* Create the descriptor buffer dma maps */ - for (int i = 0; i < adapter->num_tx_desc; i++) { - tx_buffer = &adapter->tx_buffer_area[i]; - error = bus_dmamap_create(adapter->txtag, 0, &tx_buffer->map); - if (error != 0) { - device_printf(dev, "Unable to create TX DMA map\n"); - goto fail; - } - tx_buffer->next_eop = -1; - } - - return (0); -fail: - lem_free_transmit_structures(adapter); - return (error); -} - -/********************************************************************* - * - * (Re)Initialize transmit structures. - * - **********************************************************************/ -static void -lem_setup_transmit_structures(struct adapter *adapter) -{ - struct em_buffer *tx_buffer; -#ifdef DEV_NETMAP - /* we are already locked */ - struct netmap_adapter *na = netmap_getna(adapter->ifp); - struct netmap_slot *slot = netmap_reset(na, NR_TX, 0, 0); -#endif /* DEV_NETMAP */ - - /* Clear the old ring contents */ - bzero(adapter->tx_desc_base, - (sizeof(struct e1000_tx_desc)) * adapter->num_tx_desc); - - /* Free any existing TX buffers */ - for (int i = 0; i < adapter->num_tx_desc; i++, tx_buffer++) { - tx_buffer = &adapter->tx_buffer_area[i]; - bus_dmamap_sync(adapter->txtag, tx_buffer->map, - BUS_DMASYNC_POSTWRITE); - bus_dmamap_unload(adapter->txtag, tx_buffer->map); - m_freem(tx_buffer->m_head); - tx_buffer->m_head = NULL; -#ifdef DEV_NETMAP - if (slot) { - /* the i-th NIC entry goes to slot si */ - int si = netmap_idx_n2k(&na->tx_rings[0], i); - uint64_t paddr; - void *addr; - - addr = PNMB(na, slot + si, &paddr); - adapter->tx_desc_base[i].buffer_addr = htole64(paddr); - /* reload the map for netmap mode */ - netmap_load_map(na, adapter->txtag, tx_buffer->map, addr); - } -#endif /* DEV_NETMAP */ - tx_buffer->next_eop = -1; - } - - /* Reset state */ - adapter->last_hw_offload = 0; - adapter->next_avail_tx_desc = 0; - adapter->next_tx_to_clean = 0; - adapter->num_tx_desc_avail = adapter->num_tx_desc; - - bus_dmamap_sync(adapter->txdma.dma_tag, adapter->txdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - - return; -} - -/********************************************************************* - * - * Enable transmit unit. - * - **********************************************************************/ -static void -lem_initialize_transmit_unit(struct adapter *adapter) -{ - u32 tctl, tipg = 0; - u64 bus_addr; - - INIT_DEBUGOUT("lem_initialize_transmit_unit: begin"); - /* Setup the Base and Length of the Tx Descriptor Ring */ - bus_addr = adapter->txdma.dma_paddr; - E1000_WRITE_REG(&adapter->hw, E1000_TDLEN(0), - adapter->num_tx_desc * sizeof(struct e1000_tx_desc)); - E1000_WRITE_REG(&adapter->hw, E1000_TDBAH(0), - (u32)(bus_addr >> 32)); - E1000_WRITE_REG(&adapter->hw, E1000_TDBAL(0), - (u32)bus_addr); - /* Setup the HW Tx Head and Tail descriptor pointers */ - E1000_WRITE_REG(&adapter->hw, E1000_TDT(0), 0); - E1000_WRITE_REG(&adapter->hw, E1000_TDH(0), 0); - - HW_DEBUGOUT2("Base = %x, Length = %x\n", - E1000_READ_REG(&adapter->hw, E1000_TDBAL(0)), - E1000_READ_REG(&adapter->hw, E1000_TDLEN(0))); - - /* Set the default values for the Tx Inter Packet Gap timer */ - switch (adapter->hw.mac.type) { - case e1000_82542: - tipg = DEFAULT_82542_TIPG_IPGT; - tipg |= DEFAULT_82542_TIPG_IPGR1 << E1000_TIPG_IPGR1_SHIFT; - tipg |= DEFAULT_82542_TIPG_IPGR2 << E1000_TIPG_IPGR2_SHIFT; - break; - default: - if ((adapter->hw.phy.media_type == e1000_media_type_fiber) || - (adapter->hw.phy.media_type == - e1000_media_type_internal_serdes)) - tipg = DEFAULT_82543_TIPG_IPGT_FIBER; - else - tipg = DEFAULT_82543_TIPG_IPGT_COPPER; - tipg |= DEFAULT_82543_TIPG_IPGR1 << E1000_TIPG_IPGR1_SHIFT; - tipg |= DEFAULT_82543_TIPG_IPGR2 << E1000_TIPG_IPGR2_SHIFT; - } - - E1000_WRITE_REG(&adapter->hw, E1000_TIPG, tipg); - E1000_WRITE_REG(&adapter->hw, E1000_TIDV, adapter->tx_int_delay.value); - if(adapter->hw.mac.type >= e1000_82540) - E1000_WRITE_REG(&adapter->hw, E1000_TADV, - adapter->tx_abs_int_delay.value); - - /* Program the Transmit Control Register */ - tctl = E1000_READ_REG(&adapter->hw, E1000_TCTL); - tctl &= ~E1000_TCTL_CT; - tctl |= (E1000_TCTL_PSP | E1000_TCTL_RTLC | E1000_TCTL_EN | - (E1000_COLLISION_THRESHOLD << E1000_CT_SHIFT)); - - /* This write will effectively turn on the transmit unit. */ - E1000_WRITE_REG(&adapter->hw, E1000_TCTL, tctl); - - /* Setup Transmit Descriptor Base Settings */ - adapter->txd_cmd = E1000_TXD_CMD_IFCS; - - if (adapter->tx_int_delay.value > 0) - adapter->txd_cmd |= E1000_TXD_CMD_IDE; -} - -/********************************************************************* - * - * Free all transmit related data structures. - * - **********************************************************************/ -static void -lem_free_transmit_structures(struct adapter *adapter) -{ - struct em_buffer *tx_buffer; - - INIT_DEBUGOUT("free_transmit_structures: begin"); - - if (adapter->tx_buffer_area != NULL) { - for (int i = 0; i < adapter->num_tx_desc; i++) { - tx_buffer = &adapter->tx_buffer_area[i]; - if (tx_buffer->m_head != NULL) { - bus_dmamap_sync(adapter->txtag, tx_buffer->map, - BUS_DMASYNC_POSTWRITE); - bus_dmamap_unload(adapter->txtag, - tx_buffer->map); - m_freem(tx_buffer->m_head); - tx_buffer->m_head = NULL; - } else if (tx_buffer->map != NULL) - bus_dmamap_unload(adapter->txtag, - tx_buffer->map); - if (tx_buffer->map != NULL) { - bus_dmamap_destroy(adapter->txtag, - tx_buffer->map); - tx_buffer->map = NULL; - } - } - } - if (adapter->tx_buffer_area != NULL) { - free(adapter->tx_buffer_area, M_DEVBUF); - adapter->tx_buffer_area = NULL; - } - if (adapter->txtag != NULL) { - bus_dma_tag_destroy(adapter->txtag); - adapter->txtag = NULL; - } -} - -/********************************************************************* - * - * The offload context needs to be set when we transfer the first - * packet of a particular protocol (TCP/UDP). This routine has been - * enhanced to deal with inserted VLAN headers, and IPV6 (not complete) - * - * Added back the old method of keeping the current context type - * and not setting if unnecessary, as this is reported to be a - * big performance win. -jfv - **********************************************************************/ -static void -lem_transmit_checksum_setup(struct adapter *adapter, struct mbuf *mp, - u32 *txd_upper, u32 *txd_lower) -{ - struct e1000_context_desc *TXD = NULL; - struct em_buffer *tx_buffer; - struct ether_vlan_header *eh; - struct ip *ip = NULL; - struct ip6_hdr *ip6; - int curr_txd, ehdrlen; - u32 cmd, hdr_len, ip_hlen; - u16 etype; - u8 ipproto; - - - cmd = hdr_len = ipproto = 0; - *txd_upper = *txd_lower = 0; - curr_txd = adapter->next_avail_tx_desc; - - /* - * Determine where frame payload starts. - * Jump over vlan headers if already present, - * helpful for QinQ too. - */ - eh = mtod(mp, struct ether_vlan_header *); - if (eh->evl_encap_proto == htons(ETHERTYPE_VLAN)) { - etype = ntohs(eh->evl_proto); - ehdrlen = ETHER_HDR_LEN + ETHER_VLAN_ENCAP_LEN; - } else { - etype = ntohs(eh->evl_encap_proto); - ehdrlen = ETHER_HDR_LEN; - } - - /* - * We only support TCP/UDP for IPv4 and IPv6 for the moment. - * TODO: Support SCTP too when it hits the tree. - */ - switch (etype) { - case ETHERTYPE_IP: - ip = (struct ip *)(mp->m_data + ehdrlen); - ip_hlen = ip->ip_hl << 2; - - /* Setup of IP header checksum. */ - if (mp->m_pkthdr.csum_flags & CSUM_IP) { - /* - * Start offset for header checksum calculation. - * End offset for header checksum calculation. - * Offset of place to put the checksum. - */ - TXD = (struct e1000_context_desc *) - &adapter->tx_desc_base[curr_txd]; - TXD->lower_setup.ip_fields.ipcss = ehdrlen; - TXD->lower_setup.ip_fields.ipcse = - htole16(ehdrlen + ip_hlen); - TXD->lower_setup.ip_fields.ipcso = - ehdrlen + offsetof(struct ip, ip_sum); - cmd |= E1000_TXD_CMD_IP; - *txd_upper |= E1000_TXD_POPTS_IXSM << 8; - } - - hdr_len = ehdrlen + ip_hlen; - ipproto = ip->ip_p; - - break; - case ETHERTYPE_IPV6: - ip6 = (struct ip6_hdr *)(mp->m_data + ehdrlen); - ip_hlen = sizeof(struct ip6_hdr); /* XXX: No header stacking. */ - - /* IPv6 doesn't have a header checksum. */ - - hdr_len = ehdrlen + ip_hlen; - ipproto = ip6->ip6_nxt; - break; - - default: - return; - } - - switch (ipproto) { - case IPPROTO_TCP: - if (mp->m_pkthdr.csum_flags & CSUM_TCP) { - *txd_lower = E1000_TXD_CMD_DEXT | E1000_TXD_DTYP_D; - *txd_upper |= E1000_TXD_POPTS_TXSM << 8; - /* no need for context if already set */ - if (adapter->last_hw_offload == CSUM_TCP) - return; - adapter->last_hw_offload = CSUM_TCP; - /* - * Start offset for payload checksum calculation. - * End offset for payload checksum calculation. - * Offset of place to put the checksum. - */ - TXD = (struct e1000_context_desc *) - &adapter->tx_desc_base[curr_txd]; - TXD->upper_setup.tcp_fields.tucss = hdr_len; - TXD->upper_setup.tcp_fields.tucse = htole16(0); - TXD->upper_setup.tcp_fields.tucso = - hdr_len + offsetof(struct tcphdr, th_sum); - cmd |= E1000_TXD_CMD_TCP; - } - break; - case IPPROTO_UDP: - { - if (mp->m_pkthdr.csum_flags & CSUM_UDP) { - *txd_lower = E1000_TXD_CMD_DEXT | E1000_TXD_DTYP_D; - *txd_upper |= E1000_TXD_POPTS_TXSM << 8; - /* no need for context if already set */ - if (adapter->last_hw_offload == CSUM_UDP) - return; - adapter->last_hw_offload = CSUM_UDP; - /* - * Start offset for header checksum calculation. - * End offset for header checksum calculation. - * Offset of place to put the checksum. - */ - TXD = (struct e1000_context_desc *) - &adapter->tx_desc_base[curr_txd]; - TXD->upper_setup.tcp_fields.tucss = hdr_len; - TXD->upper_setup.tcp_fields.tucse = htole16(0); - TXD->upper_setup.tcp_fields.tucso = - hdr_len + offsetof(struct udphdr, uh_sum); - } - /* Fall Thru */ - } - default: - break; - } - - if (TXD == NULL) - return; - TXD->tcp_seg_setup.data = htole32(0); - TXD->cmd_and_length = - htole32(adapter->txd_cmd | E1000_TXD_CMD_DEXT | cmd); - tx_buffer = &adapter->tx_buffer_area[curr_txd]; - tx_buffer->m_head = NULL; - tx_buffer->next_eop = -1; - - if (++curr_txd == adapter->num_tx_desc) - curr_txd = 0; - - adapter->num_tx_desc_avail--; - adapter->next_avail_tx_desc = curr_txd; -} - - -/********************************************************************** - * - * Examine each tx_buffer in the used queue. If the hardware is done - * processing the packet then free associated resources. The - * tx_buffer is put back on the free queue. - * - **********************************************************************/ -static void -lem_txeof(struct adapter *adapter) -{ - int first, last, done, num_avail; - struct em_buffer *tx_buffer; - struct e1000_tx_desc *tx_desc, *eop_desc; - if_t ifp = adapter->ifp; - - EM_TX_LOCK_ASSERT(adapter); - -#ifdef DEV_NETMAP - if (netmap_tx_irq(ifp, 0)) - return; -#endif /* DEV_NETMAP */ - if (adapter->num_tx_desc_avail == adapter->num_tx_desc) - return; - - num_avail = adapter->num_tx_desc_avail; - first = adapter->next_tx_to_clean; - tx_desc = &adapter->tx_desc_base[first]; - tx_buffer = &adapter->tx_buffer_area[first]; - last = tx_buffer->next_eop; - eop_desc = &adapter->tx_desc_base[last]; - - /* - * What this does is get the index of the - * first descriptor AFTER the EOP of the - * first packet, that way we can do the - * simple comparison on the inner while loop. - */ - if (++last == adapter->num_tx_desc) - last = 0; - done = last; - - bus_dmamap_sync(adapter->txdma.dma_tag, adapter->txdma.dma_map, - BUS_DMASYNC_POSTREAD); - - while (eop_desc->upper.fields.status & E1000_TXD_STAT_DD) { - /* We clean the range of the packet */ - while (first != done) { - tx_desc->upper.data = 0; - tx_desc->lower.data = 0; - tx_desc->buffer_addr = 0; - ++num_avail; - - if (tx_buffer->m_head) { - if_inc_counter(ifp, IFCOUNTER_OPACKETS, 1); - bus_dmamap_sync(adapter->txtag, - tx_buffer->map, - BUS_DMASYNC_POSTWRITE); - bus_dmamap_unload(adapter->txtag, - tx_buffer->map); - - m_freem(tx_buffer->m_head); - tx_buffer->m_head = NULL; - } - tx_buffer->next_eop = -1; - adapter->watchdog_time = ticks; - - if (++first == adapter->num_tx_desc) - first = 0; - - tx_buffer = &adapter->tx_buffer_area[first]; - tx_desc = &adapter->tx_desc_base[first]; - } - /* See if we can continue to the next packet */ - last = tx_buffer->next_eop; - if (last != -1) { - eop_desc = &adapter->tx_desc_base[last]; - /* Get new done point */ - if (++last == adapter->num_tx_desc) last = 0; - done = last; - } else - break; - } - bus_dmamap_sync(adapter->txdma.dma_tag, adapter->txdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - - adapter->next_tx_to_clean = first; - adapter->num_tx_desc_avail = num_avail; - -#ifdef NIC_SEND_COMBINING - if ((adapter->shadow_tdt & MIT_PENDING_TDT) == MIT_PENDING_TDT) { - /* a tdt write is pending, do it */ - E1000_WRITE_REG(&adapter->hw, E1000_TDT(0), - 0xffff & adapter->shadow_tdt); - adapter->shadow_tdt = MIT_PENDING_INT; - } else { - adapter->shadow_tdt = 0; // disable - } -#endif /* NIC_SEND_COMBINING */ - /* - * If we have enough room, clear IFF_DRV_OACTIVE to - * tell the stack that it is OK to send packets. - * If there are no pending descriptors, clear the watchdog. - */ - if (adapter->num_tx_desc_avail > EM_TX_CLEANUP_THRESHOLD) { - if_setdrvflagbits(ifp, 0, IFF_DRV_OACTIVE); - if (adapter->num_tx_desc_avail == adapter->num_tx_desc) { - adapter->watchdog_check = FALSE; - return; - } - } -} - -/********************************************************************* - * - * When Link is lost sometimes there is work still in the TX ring - * which may result in a watchdog, rather than allow that we do an - * attempted cleanup and then reinit here. Note that this has been - * seens mostly with fiber adapters. - * - **********************************************************************/ -static void -lem_tx_purge(struct adapter *adapter) -{ - if ((!adapter->link_active) && (adapter->watchdog_check)) { - EM_TX_LOCK(adapter); - lem_txeof(adapter); - EM_TX_UNLOCK(adapter); - if (adapter->watchdog_check) /* Still outstanding? */ - lem_init_locked(adapter); - } -} - -/********************************************************************* - * - * Get a buffer from system mbuf buffer pool. - * - **********************************************************************/ -static int -lem_get_buf(struct adapter *adapter, int i) -{ - struct mbuf *m; - bus_dma_segment_t segs[1]; - bus_dmamap_t map; - struct em_buffer *rx_buffer; - int error, nsegs; - - m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR); - if (m == NULL) { - adapter->mbuf_cluster_failed++; - return (ENOBUFS); - } - m->m_len = m->m_pkthdr.len = MCLBYTES; - - if (adapter->max_frame_size <= (MCLBYTES - ETHER_ALIGN)) - m_adj(m, ETHER_ALIGN); - - /* - * Using memory from the mbuf cluster pool, invoke the - * bus_dma machinery to arrange the memory mapping. - */ - error = bus_dmamap_load_mbuf_sg(adapter->rxtag, - adapter->rx_sparemap, m, segs, &nsegs, BUS_DMA_NOWAIT); - if (error != 0) { - m_free(m); - return (error); - } - - /* If nsegs is wrong then the stack is corrupt. */ - KASSERT(nsegs == 1, ("Too many segments returned!")); - - rx_buffer = &adapter->rx_buffer_area[i]; - if (rx_buffer->m_head != NULL) - bus_dmamap_unload(adapter->rxtag, rx_buffer->map); - - map = rx_buffer->map; - rx_buffer->map = adapter->rx_sparemap; - adapter->rx_sparemap = map; - bus_dmamap_sync(adapter->rxtag, rx_buffer->map, BUS_DMASYNC_PREREAD); - rx_buffer->m_head = m; - - adapter->rx_desc_base[i].buffer_addr = htole64(segs[0].ds_addr); - return (0); -} - -/********************************************************************* - * - * Allocate memory for rx_buffer structures. Since we use one - * rx_buffer per received packet, the maximum number of rx_buffer's - * that we'll need is equal to the number of receive descriptors - * that we've allocated. - * - **********************************************************************/ -static int -lem_allocate_receive_structures(struct adapter *adapter) -{ - device_t dev = adapter->dev; - struct em_buffer *rx_buffer; - int i, error; - - adapter->rx_buffer_area = malloc(sizeof(struct em_buffer) * - adapter->num_rx_desc, M_DEVBUF, M_NOWAIT | M_ZERO); - if (adapter->rx_buffer_area == NULL) { - device_printf(dev, "Unable to allocate rx_buffer memory\n"); - return (ENOMEM); - } - - error = bus_dma_tag_create(bus_get_dma_tag(dev), /* parent */ - 1, 0, /* alignment, bounds */ - BUS_SPACE_MAXADDR, /* lowaddr */ - BUS_SPACE_MAXADDR, /* highaddr */ - NULL, NULL, /* filter, filterarg */ - MCLBYTES, /* maxsize */ - 1, /* nsegments */ - MCLBYTES, /* maxsegsize */ - 0, /* flags */ - NULL, /* lockfunc */ - NULL, /* lockarg */ - &adapter->rxtag); - if (error) { - device_printf(dev, "%s: bus_dma_tag_create failed %d\n", - __func__, error); - goto fail; - } - - /* Create the spare map (used by getbuf) */ - error = bus_dmamap_create(adapter->rxtag, 0, &adapter->rx_sparemap); - if (error) { - device_printf(dev, "%s: bus_dmamap_create failed: %d\n", - __func__, error); - goto fail; - } - - rx_buffer = adapter->rx_buffer_area; - for (i = 0; i < adapter->num_rx_desc; i++, rx_buffer++) { - error = bus_dmamap_create(adapter->rxtag, 0, &rx_buffer->map); - if (error) { - device_printf(dev, "%s: bus_dmamap_create failed: %d\n", - __func__, error); - goto fail; - } - } - - return (0); - -fail: - lem_free_receive_structures(adapter); - return (error); -} - -/********************************************************************* - * - * (Re)initialize receive structures. - * - **********************************************************************/ -static int -lem_setup_receive_structures(struct adapter *adapter) -{ - struct em_buffer *rx_buffer; - int i, error; -#ifdef DEV_NETMAP - /* we are already under lock */ - struct netmap_adapter *na = netmap_getna(adapter->ifp); - struct netmap_slot *slot = netmap_reset(na, NR_RX, 0, 0); -#endif - - /* Reset descriptor ring */ - bzero(adapter->rx_desc_base, - (sizeof(struct e1000_rx_desc)) * adapter->num_rx_desc); - - /* Free current RX buffers. */ - rx_buffer = adapter->rx_buffer_area; - for (i = 0; i < adapter->num_rx_desc; i++, rx_buffer++) { - if (rx_buffer->m_head != NULL) { - bus_dmamap_sync(adapter->rxtag, rx_buffer->map, - BUS_DMASYNC_POSTREAD); - bus_dmamap_unload(adapter->rxtag, rx_buffer->map); - m_freem(rx_buffer->m_head); - rx_buffer->m_head = NULL; - } - } - - /* Allocate new ones. */ - for (i = 0; i < adapter->num_rx_desc; i++) { -#ifdef DEV_NETMAP - if (slot) { - /* the i-th NIC entry goes to slot si */ - int si = netmap_idx_n2k(&na->rx_rings[0], i); - uint64_t paddr; - void *addr; - - addr = PNMB(na, slot + si, &paddr); - netmap_load_map(na, adapter->rxtag, rx_buffer->map, addr); - /* Update descriptor */ - adapter->rx_desc_base[i].buffer_addr = htole64(paddr); - continue; - } -#endif /* DEV_NETMAP */ - error = lem_get_buf(adapter, i); - if (error) - return (error); - } - - /* Setup our descriptor pointers */ - adapter->next_rx_desc_to_check = 0; - bus_dmamap_sync(adapter->rxdma.dma_tag, adapter->rxdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - - return (0); -} - -/********************************************************************* - * - * Enable receive unit. - * - **********************************************************************/ - -static void -lem_initialize_receive_unit(struct adapter *adapter) -{ - if_t ifp = adapter->ifp; - u64 bus_addr; - u32 rctl, rxcsum; - - INIT_DEBUGOUT("lem_initialize_receive_unit: begin"); - - /* - * Make sure receives are disabled while setting - * up the descriptor ring - */ - rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); - E1000_WRITE_REG(&adapter->hw, E1000_RCTL, rctl & ~E1000_RCTL_EN); - - if (adapter->hw.mac.type >= e1000_82540) { - E1000_WRITE_REG(&adapter->hw, E1000_RADV, - adapter->rx_abs_int_delay.value); - /* - * Set the interrupt throttling rate. Value is calculated - * as DEFAULT_ITR = 1/(MAX_INTS_PER_SEC * 256ns) - */ - E1000_WRITE_REG(&adapter->hw, E1000_ITR, DEFAULT_ITR); - } - - /* Setup the Base and Length of the Rx Descriptor Ring */ - bus_addr = adapter->rxdma.dma_paddr; - E1000_WRITE_REG(&adapter->hw, E1000_RDLEN(0), - adapter->num_rx_desc * sizeof(struct e1000_rx_desc)); - E1000_WRITE_REG(&adapter->hw, E1000_RDBAH(0), - (u32)(bus_addr >> 32)); - E1000_WRITE_REG(&adapter->hw, E1000_RDBAL(0), - (u32)bus_addr); - - /* Setup the Receive Control Register */ - rctl &= ~(3 << E1000_RCTL_MO_SHIFT); - rctl |= E1000_RCTL_EN | E1000_RCTL_BAM | E1000_RCTL_LBM_NO | - E1000_RCTL_RDMTS_HALF | - (adapter->hw.mac.mc_filter_type << E1000_RCTL_MO_SHIFT); - - /* Make sure VLAN Filters are off */ - rctl &= ~E1000_RCTL_VFE; - - if (e1000_tbi_sbp_enabled_82543(&adapter->hw)) - rctl |= E1000_RCTL_SBP; - else - rctl &= ~E1000_RCTL_SBP; - - switch (adapter->rx_buffer_len) { - default: - case 2048: - rctl |= E1000_RCTL_SZ_2048; - break; - case 4096: - rctl |= E1000_RCTL_SZ_4096 | - E1000_RCTL_BSEX | E1000_RCTL_LPE; - break; - case 8192: - rctl |= E1000_RCTL_SZ_8192 | - E1000_RCTL_BSEX | E1000_RCTL_LPE; - break; - case 16384: - rctl |= E1000_RCTL_SZ_16384 | - E1000_RCTL_BSEX | E1000_RCTL_LPE; - break; - } - - if (if_getmtu(ifp) > ETHERMTU) - rctl |= E1000_RCTL_LPE; - else - rctl &= ~E1000_RCTL_LPE; - - /* Enable 82543 Receive Checksum Offload for TCP and UDP */ - if ((adapter->hw.mac.type >= e1000_82543) && - (if_getcapenable(ifp) & IFCAP_RXCSUM)) { - rxcsum = E1000_READ_REG(&adapter->hw, E1000_RXCSUM); - rxcsum |= (E1000_RXCSUM_IPOFL | E1000_RXCSUM_TUOFL); - E1000_WRITE_REG(&adapter->hw, E1000_RXCSUM, rxcsum); - } - - /* Enable Receives */ - E1000_WRITE_REG(&adapter->hw, E1000_RCTL, rctl); - - /* - * Setup the HW Rx Head and - * Tail Descriptor Pointers - */ - E1000_WRITE_REG(&adapter->hw, E1000_RDH(0), 0); - rctl = adapter->num_rx_desc - 1; /* default RDT value */ -#ifdef DEV_NETMAP - /* preserve buffers already made available to clients */ - if (if_getcapenable(ifp) & IFCAP_NETMAP) { - struct netmap_adapter *na = netmap_getna(adapter->ifp); - rctl -= nm_kr_rxspace(&na->rx_rings[0]); - } -#endif /* DEV_NETMAP */ - E1000_WRITE_REG(&adapter->hw, E1000_RDT(0), rctl); - - return; -} - -/********************************************************************* - * - * Free receive related data structures. - * - **********************************************************************/ -static void -lem_free_receive_structures(struct adapter *adapter) -{ - struct em_buffer *rx_buffer; - int i; - - INIT_DEBUGOUT("free_receive_structures: begin"); - - if (adapter->rx_sparemap) { - bus_dmamap_destroy(adapter->rxtag, adapter->rx_sparemap); - adapter->rx_sparemap = NULL; - } - - /* Cleanup any existing buffers */ - if (adapter->rx_buffer_area != NULL) { - rx_buffer = adapter->rx_buffer_area; - for (i = 0; i < adapter->num_rx_desc; i++, rx_buffer++) { - if (rx_buffer->m_head != NULL) { - bus_dmamap_sync(adapter->rxtag, rx_buffer->map, - BUS_DMASYNC_POSTREAD); - bus_dmamap_unload(adapter->rxtag, - rx_buffer->map); - m_freem(rx_buffer->m_head); - rx_buffer->m_head = NULL; - } else if (rx_buffer->map != NULL) - bus_dmamap_unload(adapter->rxtag, - rx_buffer->map); - if (rx_buffer->map != NULL) { - bus_dmamap_destroy(adapter->rxtag, - rx_buffer->map); - rx_buffer->map = NULL; - } - } - } - - if (adapter->rx_buffer_area != NULL) { - free(adapter->rx_buffer_area, M_DEVBUF); - adapter->rx_buffer_area = NULL; - } - - if (adapter->rxtag != NULL) { - bus_dma_tag_destroy(adapter->rxtag); - adapter->rxtag = NULL; - } -} - -/********************************************************************* - * - * This routine executes in interrupt context. It replenishes - * the mbufs in the descriptor and sends data which has been - * dma'ed into host memory to upper layer. - * - * We loop at most count times if count is > 0, or until done if - * count < 0. - * - * For polling we also now return the number of cleaned packets - *********************************************************************/ -static bool -lem_rxeof(struct adapter *adapter, int count, int *done) -{ - if_t ifp = adapter->ifp; - struct mbuf *mp; - u8 status = 0, accept_frame = 0, eop = 0; - u16 len, desc_len, prev_len_adj; - int i, rx_sent = 0; - struct e1000_rx_desc *current_desc; - -#ifdef BATCH_DISPATCH - struct mbuf *mh = NULL, *mt = NULL; -#endif /* BATCH_DISPATCH */ - EM_RX_LOCK(adapter); - -#ifdef BATCH_DISPATCH - batch_again: -#endif /* BATCH_DISPATCH */ - i = adapter->next_rx_desc_to_check; - current_desc = &adapter->rx_desc_base[i]; - bus_dmamap_sync(adapter->rxdma.dma_tag, adapter->rxdma.dma_map, - BUS_DMASYNC_POSTREAD); - -#ifdef DEV_NETMAP - if (netmap_rx_irq(ifp, 0, &rx_sent)) { - EM_RX_UNLOCK(adapter); - return (FALSE); - } -#endif /* DEV_NETMAP */ - - if (!((current_desc->status) & E1000_RXD_STAT_DD)) { - if (done != NULL) - *done = rx_sent; - EM_RX_UNLOCK(adapter); - return (FALSE); - } - - while (count != 0 && if_getdrvflags(ifp) & IFF_DRV_RUNNING) { - struct mbuf *m = NULL; - - status = current_desc->status; - if ((status & E1000_RXD_STAT_DD) == 0) { - break; - } - - mp = adapter->rx_buffer_area[i].m_head; - /* - * Can't defer bus_dmamap_sync(9) because TBI_ACCEPT - * needs to access the last received byte in the mbuf. - */ - bus_dmamap_sync(adapter->rxtag, adapter->rx_buffer_area[i].map, - BUS_DMASYNC_POSTREAD); - - accept_frame = 1; - prev_len_adj = 0; - desc_len = le16toh(current_desc->length); - if (status & E1000_RXD_STAT_EOP) { - count--; - eop = 1; - if (desc_len < ETHER_CRC_LEN) { - len = 0; - prev_len_adj = ETHER_CRC_LEN - desc_len; - } else - len = desc_len - ETHER_CRC_LEN; - } else { - eop = 0; - len = desc_len; - } - - if (current_desc->errors & E1000_RXD_ERR_FRAME_ERR_MASK) { - u8 last_byte; - u32 pkt_len = desc_len; - - if (adapter->fmp != NULL) - pkt_len += adapter->fmp->m_pkthdr.len; - - last_byte = *(mtod(mp, caddr_t) + desc_len - 1); - if (TBI_ACCEPT(&adapter->hw, status, - current_desc->errors, pkt_len, last_byte, - adapter->min_frame_size, adapter->max_frame_size)) { - e1000_tbi_adjust_stats_82543(&adapter->hw, - &adapter->stats, pkt_len, - adapter->hw.mac.addr, - adapter->max_frame_size); - if (len > 0) - len--; - } else - accept_frame = 0; - } - - if (accept_frame) { - if (lem_get_buf(adapter, i) != 0) { - if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1); - goto discard; - } - - /* Assign correct length to the current fragment */ - mp->m_len = len; - - if (adapter->fmp == NULL) { - mp->m_pkthdr.len = len; - adapter->fmp = mp; /* Store the first mbuf */ - adapter->lmp = mp; - } else { - /* Chain mbuf's together */ - mp->m_flags &= ~M_PKTHDR; - /* - * Adjust length of previous mbuf in chain if - * we received less than 4 bytes in the last - * descriptor. - */ - if (prev_len_adj > 0) { - adapter->lmp->m_len -= prev_len_adj; - adapter->fmp->m_pkthdr.len -= - prev_len_adj; - } - adapter->lmp->m_next = mp; - adapter->lmp = adapter->lmp->m_next; - adapter->fmp->m_pkthdr.len += len; - } - - if (eop) { - if_setrcvif(adapter->fmp, ifp); - if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1); - lem_receive_checksum(adapter, current_desc, - adapter->fmp); -#ifndef __NO_STRICT_ALIGNMENT - if (adapter->max_frame_size > - (MCLBYTES - ETHER_ALIGN) && - lem_fixup_rx(adapter) != 0) - goto skip; -#endif - if (status & E1000_RXD_STAT_VP) { - adapter->fmp->m_pkthdr.ether_vtag = - le16toh(current_desc->special); - adapter->fmp->m_flags |= M_VLANTAG; - } -#ifndef __NO_STRICT_ALIGNMENT -skip: -#endif - m = adapter->fmp; - adapter->fmp = NULL; - adapter->lmp = NULL; - } - } else { - adapter->dropped_pkts++; -discard: - /* Reuse loaded DMA map and just update mbuf chain */ - mp = adapter->rx_buffer_area[i].m_head; - mp->m_len = mp->m_pkthdr.len = MCLBYTES; - mp->m_data = mp->m_ext.ext_buf; - mp->m_next = NULL; - if (adapter->max_frame_size <= - (MCLBYTES - ETHER_ALIGN)) - m_adj(mp, ETHER_ALIGN); - if (adapter->fmp != NULL) { - m_freem(adapter->fmp); - adapter->fmp = NULL; - adapter->lmp = NULL; - } - m = NULL; - } - - /* Zero out the receive descriptors status. */ - current_desc->status = 0; - bus_dmamap_sync(adapter->rxdma.dma_tag, adapter->rxdma.dma_map, - BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE); - - /* Advance our pointers to the next descriptor. */ - if (++i == adapter->num_rx_desc) - i = 0; - /* Call into the stack */ - if (m != NULL) { -#ifdef BATCH_DISPATCH - if (adapter->batch_enable) { - if (mh == NULL) - mh = mt = m; - else - mt->m_nextpkt = m; - mt = m; - m->m_nextpkt = NULL; - rx_sent++; - current_desc = &adapter->rx_desc_base[i]; - continue; - } -#endif /* BATCH_DISPATCH */ - adapter->next_rx_desc_to_check = i; - EM_RX_UNLOCK(adapter); - if_input(ifp, m); - EM_RX_LOCK(adapter); - rx_sent++; - i = adapter->next_rx_desc_to_check; - } - current_desc = &adapter->rx_desc_base[i]; - } - adapter->next_rx_desc_to_check = i; -#ifdef BATCH_DISPATCH - if (mh) { - EM_RX_UNLOCK(adapter); - while ( (mt = mh) != NULL) { - mh = mh->m_nextpkt; - mt->m_nextpkt = NULL; - if_input(ifp, mt); - } - EM_RX_LOCK(adapter); - i = adapter->next_rx_desc_to_check; /* in case of interrupts */ - if (count > 0) - goto batch_again; - } -#endif /* BATCH_DISPATCH */ - - /* Advance the E1000's Receive Queue #0 "Tail Pointer". */ - if (--i < 0) - i = adapter->num_rx_desc - 1; - E1000_WRITE_REG(&adapter->hw, E1000_RDT(0), i); - if (done != NULL) - *done = rx_sent; - EM_RX_UNLOCK(adapter); - return ((status & E1000_RXD_STAT_DD) ? TRUE : FALSE); -} - -#ifndef __NO_STRICT_ALIGNMENT -/* - * When jumbo frames are enabled we should realign entire payload on - * architecures with strict alignment. This is serious design mistake of 8254x - * as it nullifies DMA operations. 8254x just allows RX buffer size to be - * 2048/4096/8192/16384. What we really want is 2048 - ETHER_ALIGN to align its - * payload. On architecures without strict alignment restrictions 8254x still - * performs unaligned memory access which would reduce the performance too. - * To avoid copying over an entire frame to align, we allocate a new mbuf and - * copy ethernet header to the new mbuf. The new mbuf is prepended into the - * existing mbuf chain. - * - * Be aware, best performance of the 8254x is achieved only when jumbo frame is - * not used at all on architectures with strict alignment. - */ -static int -lem_fixup_rx(struct adapter *adapter) -{ - struct mbuf *m, *n; - int error; - - error = 0; - m = adapter->fmp; - if (m->m_len <= (MCLBYTES - ETHER_HDR_LEN)) { - bcopy(m->m_data, m->m_data + ETHER_HDR_LEN, m->m_len); - m->m_data += ETHER_HDR_LEN; - } else { - MGETHDR(n, M_NOWAIT, MT_DATA); - if (n != NULL) { - bcopy(m->m_data, n->m_data, ETHER_HDR_LEN); - m->m_data += ETHER_HDR_LEN; - m->m_len -= ETHER_HDR_LEN; - n->m_len = ETHER_HDR_LEN; - M_MOVE_PKTHDR(n, m); - n->m_next = m; - adapter->fmp = n; - } else { - adapter->dropped_pkts++; - m_freem(adapter->fmp); - adapter->fmp = NULL; - error = ENOMEM; - } - } - - return (error); -} -#endif - -/********************************************************************* - * - * Verify that the hardware indicated that the checksum is valid. - * Inform the stack about the status of checksum so that stack - * doesn't spend time verifying the checksum. - * - *********************************************************************/ -static void -lem_receive_checksum(struct adapter *adapter, - struct e1000_rx_desc *rx_desc, struct mbuf *mp) -{ - /* 82543 or newer only */ - if ((adapter->hw.mac.type < e1000_82543) || - /* Ignore Checksum bit is set */ - (rx_desc->status & E1000_RXD_STAT_IXSM)) { - mp->m_pkthdr.csum_flags = 0; - return; - } - - if (rx_desc->status & E1000_RXD_STAT_IPCS) { - /* Did it pass? */ - if (!(rx_desc->errors & E1000_RXD_ERR_IPE)) { - /* IP Checksum Good */ - mp->m_pkthdr.csum_flags = CSUM_IP_CHECKED; - mp->m_pkthdr.csum_flags |= CSUM_IP_VALID; - - } else { - mp->m_pkthdr.csum_flags = 0; - } - } - - if (rx_desc->status & E1000_RXD_STAT_TCPCS) { - /* Did it pass? */ - if (!(rx_desc->errors & E1000_RXD_ERR_TCPE)) { - mp->m_pkthdr.csum_flags |= - (CSUM_DATA_VALID | CSUM_PSEUDO_HDR); - mp->m_pkthdr.csum_data = htons(0xffff); - } - } -} - -/* - * This routine is run via an vlan - * config EVENT - */ -static void -lem_register_vlan(void *arg, if_t ifp, u16 vtag) -{ - struct adapter *adapter = if_getsoftc(ifp); - u32 index, bit; - - if (if_getsoftc(ifp) != arg) /* Not our event */ - return; - - if ((vtag == 0) || (vtag > 4095)) /* Invalid ID */ - return; - - EM_CORE_LOCK(adapter); - index = (vtag >> 5) & 0x7F; - bit = vtag & 0x1F; - adapter->shadow_vfta[index] |= (1 << bit); - ++adapter->num_vlans; - /* Re-init to load the changes */ - if (if_getcapenable(ifp) & IFCAP_VLAN_HWFILTER) - lem_init_locked(adapter); - EM_CORE_UNLOCK(adapter); -} - -/* - * This routine is run via an vlan - * unconfig EVENT - */ -static void -lem_unregister_vlan(void *arg, if_t ifp, u16 vtag) -{ - struct adapter *adapter = if_getsoftc(ifp); - u32 index, bit; - - if (if_getsoftc(ifp) != arg) - return; - - if ((vtag == 0) || (vtag > 4095)) /* Invalid */ - return; - - EM_CORE_LOCK(adapter); - index = (vtag >> 5) & 0x7F; - bit = vtag & 0x1F; - adapter->shadow_vfta[index] &= ~(1 << bit); - --adapter->num_vlans; - /* Re-init to load the changes */ - if (if_getcapenable(ifp) & IFCAP_VLAN_HWFILTER) - lem_init_locked(adapter); - EM_CORE_UNLOCK(adapter); -} - -static void -lem_setup_vlan_hw_support(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - u32 reg; - - /* - ** We get here thru init_locked, meaning - ** a soft reset, this has already cleared - ** the VFTA and other state, so if there - ** have been no vlan's registered do nothing. - */ - if (adapter->num_vlans == 0) - return; - - /* - ** A soft reset zero's out the VFTA, so - ** we need to repopulate it now. - */ - for (int i = 0; i < EM_VFTA_SIZE; i++) - if (adapter->shadow_vfta[i] != 0) - E1000_WRITE_REG_ARRAY(hw, E1000_VFTA, - i, adapter->shadow_vfta[i]); - - reg = E1000_READ_REG(hw, E1000_CTRL); - reg |= E1000_CTRL_VME; - E1000_WRITE_REG(hw, E1000_CTRL, reg); - - /* Enable the Filter Table */ - reg = E1000_READ_REG(hw, E1000_RCTL); - reg &= ~E1000_RCTL_CFIEN; - reg |= E1000_RCTL_VFE; - E1000_WRITE_REG(hw, E1000_RCTL, reg); -} - -static void -lem_enable_intr(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - u32 ims_mask = IMS_ENABLE_MASK; - - E1000_WRITE_REG(hw, E1000_IMS, ims_mask); -} - -static void -lem_disable_intr(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - - E1000_WRITE_REG(hw, E1000_IMC, 0xffffffff); -} - -/* - * Bit of a misnomer, what this really means is - * to enable OS management of the system... aka - * to disable special hardware management features - */ -static void -lem_init_manageability(struct adapter *adapter) -{ - /* A shared code workaround */ - if (adapter->has_manage) { - int manc = E1000_READ_REG(&adapter->hw, E1000_MANC); - /* disable hardware interception of ARP */ - manc &= ~(E1000_MANC_ARP_EN); - E1000_WRITE_REG(&adapter->hw, E1000_MANC, manc); - } -} - -/* - * Give control back to hardware management - * controller if there is one. - */ -static void -lem_release_manageability(struct adapter *adapter) -{ - if (adapter->has_manage) { - int manc = E1000_READ_REG(&adapter->hw, E1000_MANC); - - /* re-enable hardware interception of ARP */ - manc |= E1000_MANC_ARP_EN; - E1000_WRITE_REG(&adapter->hw, E1000_MANC, manc); - } -} - -/* - * lem_get_hw_control sets the {CTRL_EXT|FWSM}:DRV_LOAD bit. - * For ASF and Pass Through versions of f/w this means - * that the driver is loaded. For AMT version type f/w - * this means that the network i/f is open. - */ -static void -lem_get_hw_control(struct adapter *adapter) -{ - u32 ctrl_ext; - - ctrl_ext = E1000_READ_REG(&adapter->hw, E1000_CTRL_EXT); - E1000_WRITE_REG(&adapter->hw, E1000_CTRL_EXT, - ctrl_ext | E1000_CTRL_EXT_DRV_LOAD); - return; -} - -/* - * lem_release_hw_control resets {CTRL_EXT|FWSM}:DRV_LOAD bit. - * For ASF and Pass Through versions of f/w this means that - * the driver is no longer loaded. For AMT versions of the - * f/w this means that the network i/f is closed. - */ -static void -lem_release_hw_control(struct adapter *adapter) -{ - u32 ctrl_ext; - - if (!adapter->has_manage) - return; - - ctrl_ext = E1000_READ_REG(&adapter->hw, E1000_CTRL_EXT); - E1000_WRITE_REG(&adapter->hw, E1000_CTRL_EXT, - ctrl_ext & ~E1000_CTRL_EXT_DRV_LOAD); - return; -} - -static int -lem_is_valid_ether_addr(u8 *addr) -{ - char zero_addr[6] = { 0, 0, 0, 0, 0, 0 }; - - if ((addr[0] & 1) || (!bcmp(addr, zero_addr, ETHER_ADDR_LEN))) { - return (FALSE); - } - - return (TRUE); -} - -/* -** Parse the interface capabilities with regard -** to both system management and wake-on-lan for -** later use. -*/ -static void -lem_get_wakeup(device_t dev) -{ - struct adapter *adapter = device_get_softc(dev); - u16 eeprom_data = 0, device_id, apme_mask; - - adapter->has_manage = e1000_enable_mng_pass_thru(&adapter->hw); - apme_mask = EM_EEPROM_APME; - - switch (adapter->hw.mac.type) { - case e1000_82542: - case e1000_82543: - break; - case e1000_82544: - e1000_read_nvm(&adapter->hw, - NVM_INIT_CONTROL2_REG, 1, &eeprom_data); - apme_mask = EM_82544_APME; - break; - case e1000_82546: - case e1000_82546_rev_3: - if (adapter->hw.bus.func == 1) { - e1000_read_nvm(&adapter->hw, - NVM_INIT_CONTROL3_PORT_B, 1, &eeprom_data); - break; - } else - e1000_read_nvm(&adapter->hw, - NVM_INIT_CONTROL3_PORT_A, 1, &eeprom_data); - break; - default: - e1000_read_nvm(&adapter->hw, - NVM_INIT_CONTROL3_PORT_A, 1, &eeprom_data); - break; - } - if (eeprom_data & apme_mask) - adapter->wol = (E1000_WUFC_MAG | E1000_WUFC_MC); - /* - * We have the eeprom settings, now apply the special cases - * where the eeprom may be wrong or the board won't support - * wake on lan on a particular port - */ - device_id = pci_get_device(dev); - switch (device_id) { - case E1000_DEV_ID_82546GB_PCIE: - adapter->wol = 0; - break; - case E1000_DEV_ID_82546EB_FIBER: - case E1000_DEV_ID_82546GB_FIBER: - /* Wake events only supported on port A for dual fiber - * regardless of eeprom setting */ - if (E1000_READ_REG(&adapter->hw, E1000_STATUS) & - E1000_STATUS_FUNC_1) - adapter->wol = 0; - break; - case E1000_DEV_ID_82546GB_QUAD_COPPER_KSP3: - /* if quad port adapter, disable WoL on all but port A */ - if (global_quad_port_a != 0) - adapter->wol = 0; - /* Reset for multiple quad port adapters */ - if (++global_quad_port_a == 4) - global_quad_port_a = 0; - break; - } - return; -} - - -/* - * Enable PCI Wake On Lan capability - */ -static void -lem_enable_wakeup(device_t dev) -{ - struct adapter *adapter = device_get_softc(dev); - if_t ifp = adapter->ifp; - u32 pmc, ctrl, ctrl_ext, rctl; - u16 status; - - if ((pci_find_cap(dev, PCIY_PMG, &pmc) != 0)) - return; - - /* Advertise the wakeup capability */ - ctrl = E1000_READ_REG(&adapter->hw, E1000_CTRL); - ctrl |= (E1000_CTRL_SWDPIN2 | E1000_CTRL_SWDPIN3); - E1000_WRITE_REG(&adapter->hw, E1000_CTRL, ctrl); - E1000_WRITE_REG(&adapter->hw, E1000_WUC, E1000_WUC_PME_EN); - - /* Keep the laser running on Fiber adapters */ - if (adapter->hw.phy.media_type == e1000_media_type_fiber || - adapter->hw.phy.media_type == e1000_media_type_internal_serdes) { - ctrl_ext = E1000_READ_REG(&adapter->hw, E1000_CTRL_EXT); - ctrl_ext |= E1000_CTRL_EXT_SDP3_DATA; - E1000_WRITE_REG(&adapter->hw, E1000_CTRL_EXT, ctrl_ext); - } - - /* - ** Determine type of Wakeup: note that wol - ** is set with all bits on by default. - */ - if ((if_getcapenable(ifp) & IFCAP_WOL_MAGIC) == 0) - adapter->wol &= ~E1000_WUFC_MAG; - - if ((if_getcapenable(ifp) & IFCAP_WOL_MCAST) == 0) - adapter->wol &= ~E1000_WUFC_MC; - else { - rctl = E1000_READ_REG(&adapter->hw, E1000_RCTL); - rctl |= E1000_RCTL_MPE; - E1000_WRITE_REG(&adapter->hw, E1000_RCTL, rctl); - } - - if (adapter->hw.mac.type == e1000_pchlan) { - if (lem_enable_phy_wakeup(adapter)) - return; - } else { - E1000_WRITE_REG(&adapter->hw, E1000_WUC, E1000_WUC_PME_EN); - E1000_WRITE_REG(&adapter->hw, E1000_WUFC, adapter->wol); - } - - - /* Request PME */ - status = pci_read_config(dev, pmc + PCIR_POWER_STATUS, 2); - status &= ~(PCIM_PSTAT_PME | PCIM_PSTAT_PMEENABLE); - if (if_getcapenable(ifp) & IFCAP_WOL) - status |= PCIM_PSTAT_PME | PCIM_PSTAT_PMEENABLE; - pci_write_config(dev, pmc + PCIR_POWER_STATUS, status, 2); - - return; -} - -/* -** WOL in the newer chipset interfaces (pchlan) -** require thing to be copied into the phy -*/ -static int -lem_enable_phy_wakeup(struct adapter *adapter) -{ - struct e1000_hw *hw = &adapter->hw; - u32 mreg, ret = 0; - u16 preg; - - /* copy MAC RARs to PHY RARs */ - for (int i = 0; i < adapter->hw.mac.rar_entry_count; i++) { - mreg = E1000_READ_REG(hw, E1000_RAL(i)); - e1000_write_phy_reg(hw, BM_RAR_L(i), (u16)(mreg & 0xFFFF)); - e1000_write_phy_reg(hw, BM_RAR_M(i), - (u16)((mreg >> 16) & 0xFFFF)); - mreg = E1000_READ_REG(hw, E1000_RAH(i)); - e1000_write_phy_reg(hw, BM_RAR_H(i), (u16)(mreg & 0xFFFF)); - e1000_write_phy_reg(hw, BM_RAR_CTRL(i), - (u16)((mreg >> 16) & 0xFFFF)); - } - - /* copy MAC MTA to PHY MTA */ - for (int i = 0; i < adapter->hw.mac.mta_reg_count; i++) { - mreg = E1000_READ_REG_ARRAY(hw, E1000_MTA, i); - e1000_write_phy_reg(hw, BM_MTA(i), (u16)(mreg & 0xFFFF)); - e1000_write_phy_reg(hw, BM_MTA(i) + 1, - (u16)((mreg >> 16) & 0xFFFF)); - } - - /* configure PHY Rx Control register */ - e1000_read_phy_reg(&adapter->hw, BM_RCTL, &preg); - mreg = E1000_READ_REG(hw, E1000_RCTL); - if (mreg & E1000_RCTL_UPE) - preg |= BM_RCTL_UPE; - if (mreg & E1000_RCTL_MPE) - preg |= BM_RCTL_MPE; - preg &= ~(BM_RCTL_MO_MASK); - if (mreg & E1000_RCTL_MO_3) - preg |= (((mreg & E1000_RCTL_MO_3) >> E1000_RCTL_MO_SHIFT) - << BM_RCTL_MO_SHIFT); - if (mreg & E1000_RCTL_BAM) - preg |= BM_RCTL_BAM; - if (mreg & E1000_RCTL_PMCF) - preg |= BM_RCTL_PMCF; - mreg = E1000_READ_REG(hw, E1000_CTRL); - if (mreg & E1000_CTRL_RFCE) - preg |= BM_RCTL_RFCE; - e1000_write_phy_reg(&adapter->hw, BM_RCTL, preg); - - /* enable PHY wakeup in MAC register */ - E1000_WRITE_REG(hw, E1000_WUC, - E1000_WUC_PHY_WAKE | E1000_WUC_PME_EN); - E1000_WRITE_REG(hw, E1000_WUFC, adapter->wol); - - /* configure and enable PHY wakeup in PHY registers */ - e1000_write_phy_reg(&adapter->hw, BM_WUFC, adapter->wol); - e1000_write_phy_reg(&adapter->hw, BM_WUC, E1000_WUC_PME_EN); - - /* activate PHY wakeup */ - ret = hw->phy.ops.acquire(hw); - if (ret) { - printf("Could not acquire PHY\n"); - return ret; - } - e1000_write_phy_reg_mdic(hw, IGP01E1000_PHY_PAGE_SELECT, - (BM_WUC_ENABLE_PAGE << IGP_PAGE_SHIFT)); - ret = e1000_read_phy_reg_mdic(hw, BM_WUC_ENABLE_REG, &preg); - if (ret) { - printf("Could not read PHY page 769\n"); - goto out; - } - preg |= BM_WUC_ENABLE_BIT | BM_WUC_HOST_WU_BIT; - ret = e1000_write_phy_reg_mdic(hw, BM_WUC_ENABLE_REG, preg); - if (ret) - printf("Could not set PHY Host Wakeup bit\n"); -out: - hw->phy.ops.release(hw); - - return ret; -} - -static void -lem_led_func(void *arg, int onoff) -{ - struct adapter *adapter = arg; - - EM_CORE_LOCK(adapter); - if (onoff) { - e1000_setup_led(&adapter->hw); - e1000_led_on(&adapter->hw); - } else { - e1000_led_off(&adapter->hw); - e1000_cleanup_led(&adapter->hw); - } - EM_CORE_UNLOCK(adapter); -} - -/********************************************************************* -* 82544 Coexistence issue workaround. -* There are 2 issues. -* 1. Transmit Hang issue. -* To detect this issue, following equation can be used... -* SIZE[3:0] + ADDR[2:0] = SUM[3:0]. -* If SUM[3:0] is in between 1 to 4, we will have this issue. -* -* 2. DAC issue. -* To detect this issue, following equation can be used... -* SIZE[3:0] + ADDR[2:0] = SUM[3:0]. -* If SUM[3:0] is in between 9 to c, we will have this issue. -* -* -* WORKAROUND: -* Make sure we do not have ending address -* as 1,2,3,4(Hang) or 9,a,b,c (DAC) -* -*************************************************************************/ -static u32 -lem_fill_descriptors (bus_addr_t address, u32 length, - PDESC_ARRAY desc_array) -{ - u32 safe_terminator; - - /* Since issue is sensitive to length and address.*/ - /* Let us first check the address...*/ - if (length <= 4) { - desc_array->descriptor[0].address = address; - desc_array->descriptor[0].length = length; - desc_array->elements = 1; - return (desc_array->elements); - } - safe_terminator = (u32)((((u32)address & 0x7) + - (length & 0xF)) & 0xF); - /* if it does not fall between 0x1 to 0x4 and 0x9 to 0xC then return */ - if (safe_terminator == 0 || - (safe_terminator > 4 && - safe_terminator < 9) || - (safe_terminator > 0xC && - safe_terminator <= 0xF)) { - desc_array->descriptor[0].address = address; - desc_array->descriptor[0].length = length; - desc_array->elements = 1; - return (desc_array->elements); - } - - desc_array->descriptor[0].address = address; - desc_array->descriptor[0].length = length - 4; - desc_array->descriptor[1].address = address + (length - 4); - desc_array->descriptor[1].length = 4; - desc_array->elements = 2; - return (desc_array->elements); -} - -/********************************************************************** - * - * Update the board statistics counters. - * - **********************************************************************/ -static void -lem_update_stats_counters(struct adapter *adapter) -{ - - if(adapter->hw.phy.media_type == e1000_media_type_copper || - (E1000_READ_REG(&adapter->hw, E1000_STATUS) & E1000_STATUS_LU)) { - adapter->stats.symerrs += E1000_READ_REG(&adapter->hw, E1000_SYMERRS); - adapter->stats.sec += E1000_READ_REG(&adapter->hw, E1000_SEC); - } - adapter->stats.crcerrs += E1000_READ_REG(&adapter->hw, E1000_CRCERRS); - adapter->stats.mpc += E1000_READ_REG(&adapter->hw, E1000_MPC); - adapter->stats.scc += E1000_READ_REG(&adapter->hw, E1000_SCC); - adapter->stats.ecol += E1000_READ_REG(&adapter->hw, E1000_ECOL); - - adapter->stats.mcc += E1000_READ_REG(&adapter->hw, E1000_MCC); - adapter->stats.latecol += E1000_READ_REG(&adapter->hw, E1000_LATECOL); - adapter->stats.colc += E1000_READ_REG(&adapter->hw, E1000_COLC); - adapter->stats.dc += E1000_READ_REG(&adapter->hw, E1000_DC); - adapter->stats.rlec += E1000_READ_REG(&adapter->hw, E1000_RLEC); - adapter->stats.xonrxc += E1000_READ_REG(&adapter->hw, E1000_XONRXC); - adapter->stats.xontxc += E1000_READ_REG(&adapter->hw, E1000_XONTXC); - adapter->stats.xoffrxc += E1000_READ_REG(&adapter->hw, E1000_XOFFRXC); - adapter->stats.xofftxc += E1000_READ_REG(&adapter->hw, E1000_XOFFTXC); - adapter->stats.fcruc += E1000_READ_REG(&adapter->hw, E1000_FCRUC); - adapter->stats.prc64 += E1000_READ_REG(&adapter->hw, E1000_PRC64); - adapter->stats.prc127 += E1000_READ_REG(&adapter->hw, E1000_PRC127); - adapter->stats.prc255 += E1000_READ_REG(&adapter->hw, E1000_PRC255); - adapter->stats.prc511 += E1000_READ_REG(&adapter->hw, E1000_PRC511); - adapter->stats.prc1023 += E1000_READ_REG(&adapter->hw, E1000_PRC1023); - adapter->stats.prc1522 += E1000_READ_REG(&adapter->hw, E1000_PRC1522); - adapter->stats.gprc += E1000_READ_REG(&adapter->hw, E1000_GPRC); - adapter->stats.bprc += E1000_READ_REG(&adapter->hw, E1000_BPRC); - adapter->stats.mprc += E1000_READ_REG(&adapter->hw, E1000_MPRC); - adapter->stats.gptc += E1000_READ_REG(&adapter->hw, E1000_GPTC); - - /* For the 64-bit byte counters the low dword must be read first. */ - /* Both registers clear on the read of the high dword */ - - adapter->stats.gorc += E1000_READ_REG(&adapter->hw, E1000_GORCL) + - ((u64)E1000_READ_REG(&adapter->hw, E1000_GORCH) << 32); - adapter->stats.gotc += E1000_READ_REG(&adapter->hw, E1000_GOTCL) + - ((u64)E1000_READ_REG(&adapter->hw, E1000_GOTCH) << 32); - - adapter->stats.rnbc += E1000_READ_REG(&adapter->hw, E1000_RNBC); - adapter->stats.ruc += E1000_READ_REG(&adapter->hw, E1000_RUC); - adapter->stats.rfc += E1000_READ_REG(&adapter->hw, E1000_RFC); - adapter->stats.roc += E1000_READ_REG(&adapter->hw, E1000_ROC); - adapter->stats.rjc += E1000_READ_REG(&adapter->hw, E1000_RJC); - - adapter->stats.tor += E1000_READ_REG(&adapter->hw, E1000_TORH); - adapter->stats.tot += E1000_READ_REG(&adapter->hw, E1000_TOTH); - - adapter->stats.tpr += E1000_READ_REG(&adapter->hw, E1000_TPR); - adapter->stats.tpt += E1000_READ_REG(&adapter->hw, E1000_TPT); - adapter->stats.ptc64 += E1000_READ_REG(&adapter->hw, E1000_PTC64); - adapter->stats.ptc127 += E1000_READ_REG(&adapter->hw, E1000_PTC127); - adapter->stats.ptc255 += E1000_READ_REG(&adapter->hw, E1000_PTC255); - adapter->stats.ptc511 += E1000_READ_REG(&adapter->hw, E1000_PTC511); - adapter->stats.ptc1023 += E1000_READ_REG(&adapter->hw, E1000_PTC1023); - adapter->stats.ptc1522 += E1000_READ_REG(&adapter->hw, E1000_PTC1522); - adapter->stats.mptc += E1000_READ_REG(&adapter->hw, E1000_MPTC); - adapter->stats.bptc += E1000_READ_REG(&adapter->hw, E1000_BPTC); - - if (adapter->hw.mac.type >= e1000_82543) { - adapter->stats.algnerrc += - E1000_READ_REG(&adapter->hw, E1000_ALGNERRC); - adapter->stats.rxerrc += - E1000_READ_REG(&adapter->hw, E1000_RXERRC); - adapter->stats.tncrs += - E1000_READ_REG(&adapter->hw, E1000_TNCRS); - adapter->stats.cexterr += - E1000_READ_REG(&adapter->hw, E1000_CEXTERR); - adapter->stats.tsctc += - E1000_READ_REG(&adapter->hw, E1000_TSCTC); - adapter->stats.tsctfc += - E1000_READ_REG(&adapter->hw, E1000_TSCTFC); - } -} - -static uint64_t -lem_get_counter(if_t ifp, ift_counter cnt) -{ - struct adapter *adapter; - - adapter = if_getsoftc(ifp); - - switch (cnt) { - case IFCOUNTER_COLLISIONS: - return (adapter->stats.colc); - case IFCOUNTER_IERRORS: - return (adapter->dropped_pkts + adapter->stats.rxerrc + - adapter->stats.crcerrs + adapter->stats.algnerrc + - adapter->stats.ruc + adapter->stats.roc + - adapter->stats.mpc + adapter->stats.cexterr); - case IFCOUNTER_OERRORS: - return (adapter->stats.ecol + adapter->stats.latecol + - adapter->watchdog_events); - default: - return (if_get_counter_default(ifp, cnt)); - } -} - -/* Export a single 32-bit register via a read-only sysctl. */ -static int -lem_sysctl_reg_handler(SYSCTL_HANDLER_ARGS) -{ - struct adapter *adapter; - u_int val; - - adapter = oidp->oid_arg1; - val = E1000_READ_REG(&adapter->hw, oidp->oid_arg2); - return (sysctl_handle_int(oidp, &val, 0, req)); -} - -/* - * Add sysctl variables, one per statistic, to the system. - */ -static void -lem_add_hw_stats(struct adapter *adapter) -{ - device_t dev = adapter->dev; - - struct sysctl_ctx_list *ctx = device_get_sysctl_ctx(dev); - struct sysctl_oid *tree = device_get_sysctl_tree(dev); - struct sysctl_oid_list *child = SYSCTL_CHILDREN(tree); - struct e1000_hw_stats *stats = &adapter->stats; - - struct sysctl_oid *stat_node; - struct sysctl_oid_list *stat_list; - - /* Driver Statistics */ - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "cluster_alloc_fail", - CTLFLAG_RD, &adapter->mbuf_cluster_failed, - "Std mbuf cluster failed"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "mbuf_defrag_fail", - CTLFLAG_RD, &adapter->mbuf_defrag_failed, - "Defragmenting mbuf chain failed"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "dropped", - CTLFLAG_RD, &adapter->dropped_pkts, - "Driver dropped packets"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "tx_dma_fail", - CTLFLAG_RD, &adapter->no_tx_dma_setup, - "Driver tx dma failure in xmit"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "tx_desc_fail1", - CTLFLAG_RD, &adapter->no_tx_desc_avail1, - "Not enough tx descriptors failure in xmit"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "tx_desc_fail2", - CTLFLAG_RD, &adapter->no_tx_desc_avail2, - "Not enough tx descriptors failure in xmit"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "rx_overruns", - CTLFLAG_RD, &adapter->rx_overruns, - "RX overruns"); - SYSCTL_ADD_ULONG(ctx, child, OID_AUTO, "watchdog_timeouts", - CTLFLAG_RD, &adapter->watchdog_events, - "Watchdog timeouts"); - - SYSCTL_ADD_PROC(ctx, child, OID_AUTO, "device_control", - CTLTYPE_UINT | CTLFLAG_RD, adapter, E1000_CTRL, - lem_sysctl_reg_handler, "IU", - "Device Control Register"); - SYSCTL_ADD_PROC(ctx, child, OID_AUTO, "rx_control", - CTLTYPE_UINT | CTLFLAG_RD, adapter, E1000_RCTL, - lem_sysctl_reg_handler, "IU", - "Receiver Control Register"); - SYSCTL_ADD_UINT(ctx, child, OID_AUTO, "fc_high_water", - CTLFLAG_RD, &adapter->hw.fc.high_water, 0, - "Flow Control High Watermark"); - SYSCTL_ADD_UINT(ctx, child, OID_AUTO, "fc_low_water", - CTLFLAG_RD, &adapter->hw.fc.low_water, 0, - "Flow Control Low Watermark"); - SYSCTL_ADD_UQUAD(ctx, child, OID_AUTO, "fifo_workaround", - CTLFLAG_RD, &adapter->tx_fifo_wrk_cnt, - "TX FIFO workaround events"); - SYSCTL_ADD_UQUAD(ctx, child, OID_AUTO, "fifo_reset", - CTLFLAG_RD, &adapter->tx_fifo_reset_cnt, - "TX FIFO resets"); - - SYSCTL_ADD_PROC(ctx, child, OID_AUTO, "txd_head", - CTLTYPE_UINT | CTLFLAG_RD, adapter, E1000_TDH(0), - lem_sysctl_reg_handler, "IU", - "Transmit Descriptor Head"); - SYSCTL_ADD_PROC(ctx, child, OID_AUTO, "txd_tail", - CTLTYPE_UINT | CTLFLAG_RD, adapter, E1000_TDT(0), - lem_sysctl_reg_handler, "IU", - "Transmit Descriptor Tail"); - SYSCTL_ADD_PROC(ctx, child, OID_AUTO, "rxd_head", - CTLTYPE_UINT | CTLFLAG_RD, adapter, E1000_RDH(0), - lem_sysctl_reg_handler, "IU", - "Receive Descriptor Head"); - SYSCTL_ADD_PROC(ctx, child, OID_AUTO, "rxd_tail", - CTLTYPE_UINT | CTLFLAG_RD, adapter, E1000_RDT(0), - lem_sysctl_reg_handler, "IU", - "Receive Descriptor Tail"); - - - /* MAC stats get their own sub node */ - - stat_node = SYSCTL_ADD_NODE(ctx, child, OID_AUTO, "mac_stats", - CTLFLAG_RD, NULL, "Statistics"); - stat_list = SYSCTL_CHILDREN(stat_node); - - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "excess_coll", - CTLFLAG_RD, &stats->ecol, - "Excessive collisions"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "single_coll", - CTLFLAG_RD, &stats->scc, - "Single collisions"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "multiple_coll", - CTLFLAG_RD, &stats->mcc, - "Multiple collisions"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "late_coll", - CTLFLAG_RD, &stats->latecol, - "Late collisions"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "collision_count", - CTLFLAG_RD, &stats->colc, - "Collision Count"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "symbol_errors", - CTLFLAG_RD, &adapter->stats.symerrs, - "Symbol Errors"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "sequence_errors", - CTLFLAG_RD, &adapter->stats.sec, - "Sequence Errors"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "defer_count", - CTLFLAG_RD, &adapter->stats.dc, - "Defer Count"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "missed_packets", - CTLFLAG_RD, &adapter->stats.mpc, - "Missed Packets"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_no_buff", - CTLFLAG_RD, &adapter->stats.rnbc, - "Receive No Buffers"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_undersize", - CTLFLAG_RD, &adapter->stats.ruc, - "Receive Undersize"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_fragmented", - CTLFLAG_RD, &adapter->stats.rfc, - "Fragmented Packets Received "); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_oversize", - CTLFLAG_RD, &adapter->stats.roc, - "Oversized Packets Received"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_jabber", - CTLFLAG_RD, &adapter->stats.rjc, - "Recevied Jabber"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "recv_errs", - CTLFLAG_RD, &adapter->stats.rxerrc, - "Receive Errors"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "crc_errs", - CTLFLAG_RD, &adapter->stats.crcerrs, - "CRC errors"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "alignment_errs", - CTLFLAG_RD, &adapter->stats.algnerrc, - "Alignment Errors"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "coll_ext_errs", - CTLFLAG_RD, &adapter->stats.cexterr, - "Collision/Carrier extension errors"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "xon_recvd", - CTLFLAG_RD, &adapter->stats.xonrxc, - "XON Received"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "xon_txd", - CTLFLAG_RD, &adapter->stats.xontxc, - "XON Transmitted"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "xoff_recvd", - CTLFLAG_RD, &adapter->stats.xoffrxc, - "XOFF Received"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "xoff_txd", - CTLFLAG_RD, &adapter->stats.xofftxc, - "XOFF Transmitted"); - - /* Packet Reception Stats */ - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "total_pkts_recvd", - CTLFLAG_RD, &adapter->stats.tpr, - "Total Packets Received "); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "good_pkts_recvd", - CTLFLAG_RD, &adapter->stats.gprc, - "Good Packets Received"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "bcast_pkts_recvd", - CTLFLAG_RD, &adapter->stats.bprc, - "Broadcast Packets Received"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "mcast_pkts_recvd", - CTLFLAG_RD, &adapter->stats.mprc, - "Multicast Packets Received"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_64", - CTLFLAG_RD, &adapter->stats.prc64, - "64 byte frames received "); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_65_127", - CTLFLAG_RD, &adapter->stats.prc127, - "65-127 byte frames received"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_128_255", - CTLFLAG_RD, &adapter->stats.prc255, - "128-255 byte frames received"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_256_511", - CTLFLAG_RD, &adapter->stats.prc511, - "256-511 byte frames received"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_512_1023", - CTLFLAG_RD, &adapter->stats.prc1023, - "512-1023 byte frames received"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "rx_frames_1024_1522", - CTLFLAG_RD, &adapter->stats.prc1522, - "1023-1522 byte frames received"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "good_octets_recvd", - CTLFLAG_RD, &adapter->stats.gorc, - "Good Octets Received"); - - /* Packet Transmission Stats */ - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "good_octets_txd", - CTLFLAG_RD, &adapter->stats.gotc, - "Good Octets Transmitted"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "total_pkts_txd", - CTLFLAG_RD, &adapter->stats.tpt, - "Total Packets Transmitted"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "good_pkts_txd", - CTLFLAG_RD, &adapter->stats.gptc, - "Good Packets Transmitted"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "bcast_pkts_txd", - CTLFLAG_RD, &adapter->stats.bptc, - "Broadcast Packets Transmitted"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "mcast_pkts_txd", - CTLFLAG_RD, &adapter->stats.mptc, - "Multicast Packets Transmitted"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_64", - CTLFLAG_RD, &adapter->stats.ptc64, - "64 byte frames transmitted "); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_65_127", - CTLFLAG_RD, &adapter->stats.ptc127, - "65-127 byte frames transmitted"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_128_255", - CTLFLAG_RD, &adapter->stats.ptc255, - "128-255 byte frames transmitted"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_256_511", - CTLFLAG_RD, &adapter->stats.ptc511, - "256-511 byte frames transmitted"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_512_1023", - CTLFLAG_RD, &adapter->stats.ptc1023, - "512-1023 byte frames transmitted"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tx_frames_1024_1522", - CTLFLAG_RD, &adapter->stats.ptc1522, - "1024-1522 byte frames transmitted"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tso_txd", - CTLFLAG_RD, &adapter->stats.tsctc, - "TSO Contexts Transmitted"); - SYSCTL_ADD_UQUAD(ctx, stat_list, OID_AUTO, "tso_ctx_fail", - CTLFLAG_RD, &adapter->stats.tsctfc, - "TSO Contexts Failed"); -} - -/********************************************************************** - * - * This routine provides a way to dump out the adapter eeprom, - * often a useful debug/service tool. This only dumps the first - * 32 words, stuff that matters is in that extent. - * - **********************************************************************/ - -static int -lem_sysctl_nvm_info(SYSCTL_HANDLER_ARGS) -{ - struct adapter *adapter; - int error; - int result; - - result = -1; - error = sysctl_handle_int(oidp, &result, 0, req); - - if (error || !req->newptr) - return (error); - - /* - * This value will cause a hex dump of the - * first 32 16-bit words of the EEPROM to - * the screen. - */ - if (result == 1) { - adapter = (struct adapter *)arg1; - lem_print_nvm_info(adapter); - } - - return (error); -} - -static void -lem_print_nvm_info(struct adapter *adapter) -{ - u16 eeprom_data; - int i, j, row = 0; - - /* Its a bit crude, but it gets the job done */ - printf("\nInterface EEPROM Dump:\n"); - printf("Offset\n0x0000 "); - for (i = 0, j = 0; i < 32; i++, j++) { - if (j == 8) { /* Make the offset block */ - j = 0; ++row; - printf("\n0x00%x0 ",row); - } - e1000_read_nvm(&adapter->hw, i, 1, &eeprom_data); - printf("%04x ", eeprom_data); - } - printf("\n"); -} - -static int -lem_sysctl_int_delay(SYSCTL_HANDLER_ARGS) -{ - struct em_int_delay_info *info; - struct adapter *adapter; - u32 regval; - int error; - int usecs; - int ticks; - - info = (struct em_int_delay_info *)arg1; - usecs = info->value; - error = sysctl_handle_int(oidp, &usecs, 0, req); - if (error != 0 || req->newptr == NULL) - return (error); - if (usecs < 0 || usecs > EM_TICKS_TO_USECS(65535)) - return (EINVAL); - info->value = usecs; - ticks = EM_USECS_TO_TICKS(usecs); - if (info->offset == E1000_ITR) /* units are 256ns here */ - ticks *= 4; - - adapter = info->adapter; - - EM_CORE_LOCK(adapter); - regval = E1000_READ_OFFSET(&adapter->hw, info->offset); - regval = (regval & ~0xffff) | (ticks & 0xffff); - /* Handle a few special cases. */ - switch (info->offset) { - case E1000_RDTR: - break; - case E1000_TIDV: - if (ticks == 0) { - adapter->txd_cmd &= ~E1000_TXD_CMD_IDE; - /* Don't write 0 into the TIDV register. */ - regval++; - } else - adapter->txd_cmd |= E1000_TXD_CMD_IDE; - break; - } - E1000_WRITE_OFFSET(&adapter->hw, info->offset, regval); - EM_CORE_UNLOCK(adapter); - return (0); -} - -static void -lem_add_int_delay_sysctl(struct adapter *adapter, const char *name, - const char *description, struct em_int_delay_info *info, - int offset, int value) -{ - info->adapter = adapter; - info->offset = offset; - info->value = value; - SYSCTL_ADD_PROC(device_get_sysctl_ctx(adapter->dev), - SYSCTL_CHILDREN(device_get_sysctl_tree(adapter->dev)), - OID_AUTO, name, CTLTYPE_INT|CTLFLAG_RW, - info, 0, lem_sysctl_int_delay, "I", description); -} - -static void -lem_set_flow_cntrl(struct adapter *adapter, const char *name, - const char *description, int *limit, int value) -{ - *limit = value; - SYSCTL_ADD_INT(device_get_sysctl_ctx(adapter->dev), - SYSCTL_CHILDREN(device_get_sysctl_tree(adapter->dev)), - OID_AUTO, name, CTLFLAG_RW, limit, value, description); -} - -static void -lem_add_rx_process_limit(struct adapter *adapter, const char *name, - const char *description, int *limit, int value) -{ - *limit = value; - SYSCTL_ADD_INT(device_get_sysctl_ctx(adapter->dev), - SYSCTL_CHILDREN(device_get_sysctl_tree(adapter->dev)), - OID_AUTO, name, CTLFLAG_RW, limit, value, description); -} diff --git a/sys/dev/e1000/if_lem.h b/sys/dev/e1000/if_lem.h deleted file mode 100644 index 4a27c34bd772..000000000000 --- a/sys/dev/e1000/if_lem.h +++ /dev/null @@ -1,519 +0,0 @@ -/****************************************************************************** - - Copyright (c) 2001-2015, Intel Corporation - All rights reserved. - - Redistribution and use in source and binary forms, with or without - modification, are permitted provided that the following conditions are met: - - 1. Redistributions of source code must retain the above copyright notice, - this list of conditions and the following disclaimer. - - 2. Redistributions in binary form must reproduce the above copyright - notice, this list of conditions and the following disclaimer in the - documentation and/or other materials provided with the distribution. - - 3. Neither the name of the Intel Corporation nor the names of its - contributors may be used to endorse or promote products derived from - this software without specific prior written permission. - - THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" - AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE - LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR - CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF - SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS - INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN - CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) - ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE - POSSIBILITY OF SUCH DAMAGE. - -******************************************************************************/ -/*$FreeBSD$*/ - - -#ifndef _LEM_H_DEFINED_ -#define _LEM_H_DEFINED_ - - -/* Tunables */ - -/* - * EM_TXD: Maximum number of Transmit Descriptors - * Valid Range: 80-256 for 82542 and 82543-based adapters - * 80-4096 for others - * Default Value: 256 - * This value is the number of transmit descriptors allocated by the driver. - * Increasing this value allows the driver to queue more transmits. Each - * descriptor is 16 bytes. - * Since TDLEN should be multiple of 128bytes, the number of transmit - * desscriptors should meet the following condition. - * (num_tx_desc * sizeof(struct e1000_tx_desc)) % 128 == 0 - */ -#define EM_MIN_TXD 80 -#define EM_MAX_TXD_82543 256 -#define EM_MAX_TXD 4096 -#define EM_DEFAULT_TXD EM_MAX_TXD_82543 - -/* - * EM_RXD - Maximum number of receive Descriptors - * Valid Range: 80-256 for 82542 and 82543-based adapters - * 80-4096 for others - * Default Value: 256 - * This value is the number of receive descriptors allocated by the driver. - * Increasing this value allows the driver to buffer more incoming packets. - * Each descriptor is 16 bytes. A receive buffer is also allocated for each - * descriptor. The maximum MTU size is 16110. - * Since TDLEN should be multiple of 128bytes, the number of transmit - * desscriptors should meet the following condition. - * (num_tx_desc * sizeof(struct e1000_tx_desc)) % 128 == 0 - */ -#define EM_MIN_RXD 80 -#define EM_MAX_RXD_82543 256 -#define EM_MAX_RXD 4096 -#define EM_DEFAULT_RXD EM_MAX_RXD_82543 - -/* - * EM_TIDV - Transmit Interrupt Delay Value - * Valid Range: 0-65535 (0=off) - * Default Value: 64 - * This value delays the generation of transmit interrupts in units of - * 1.024 microseconds. Transmit interrupt reduction can improve CPU - * efficiency if properly tuned for specific network traffic. If the - * system is reporting dropped transmits, this value may be set too high - * causing the driver to run out of available transmit descriptors. - */ -#define EM_TIDV 64 - -/* - * EM_TADV - Transmit Absolute Interrupt Delay Value - * (Not valid for 82542/82543/82544) - * Valid Range: 0-65535 (0=off) - * Default Value: 64 - * This value, in units of 1.024 microseconds, limits the delay in which a - * transmit interrupt is generated. Useful only if EM_TIDV is non-zero, - * this value ensures that an interrupt is generated after the initial - * packet is sent on the wire within the set amount of time. Proper tuning, - * along with EM_TIDV, may improve traffic throughput in specific - * network conditions. - */ -#define EM_TADV 64 - -/* - * EM_RDTR - Receive Interrupt Delay Timer (Packet Timer) - * Valid Range: 0-65535 (0=off) - * Default Value: 0 - * This value delays the generation of receive interrupts in units of 1.024 - * microseconds. Receive interrupt reduction can improve CPU efficiency if - * properly tuned for specific network traffic. Increasing this value adds - * extra latency to frame reception and can end up decreasing the throughput - * of TCP traffic. If the system is reporting dropped receives, this value - * may be set too high, causing the driver to run out of available receive - * descriptors. - * - * CAUTION: When setting EM_RDTR to a value other than 0, adapters - * may hang (stop transmitting) under certain network conditions. - * If this occurs a WATCHDOG message is logged in the system - * event log. In addition, the controller is automatically reset, - * restoring the network connection. To eliminate the potential - * for the hang ensure that EM_RDTR is set to 0. - */ -#define EM_RDTR 0 - -/* - * Receive Interrupt Absolute Delay Timer (Not valid for 82542/82543/82544) - * Valid Range: 0-65535 (0=off) - * Default Value: 64 - * This value, in units of 1.024 microseconds, limits the delay in which a - * receive interrupt is generated. Useful only if EM_RDTR is non-zero, - * this value ensures that an interrupt is generated after the initial - * packet is received within the set amount of time. Proper tuning, - * along with EM_RDTR, may improve traffic throughput in specific network - * conditions. - */ -#define EM_RADV 64 - -/* - * This parameter controls the max duration of transmit watchdog. - */ -#define EM_WATCHDOG (10 * hz) - -/* - * This parameter controls when the driver calls the routine to reclaim - * transmit descriptors. - */ -#define EM_TX_CLEANUP_THRESHOLD (adapter->num_tx_desc / 8) -#define EM_TX_OP_THRESHOLD (adapter->num_tx_desc / 32) - -/* - * This parameter controls whether or not autonegotation is enabled. - * 0 - Disable autonegotiation - * 1 - Enable autonegotiation - */ -#define DO_AUTO_NEG 1 - -/* - * This parameter control whether or not the driver will wait for - * autonegotiation to complete. - * 1 - Wait for autonegotiation to complete - * 0 - Don't wait for autonegotiation to complete - */ -#define WAIT_FOR_AUTO_NEG_DEFAULT 0 - -/* Tunables -- End */ - -#define AUTONEG_ADV_DEFAULT (ADVERTISE_10_HALF | ADVERTISE_10_FULL | \ - ADVERTISE_100_HALF | ADVERTISE_100_FULL | \ - ADVERTISE_1000_FULL) - -#define AUTO_ALL_MODES 0 - -/* PHY master/slave setting */ -#define EM_MASTER_SLAVE e1000_ms_hw_default - -/* - * Micellaneous constants - */ -#define EM_VENDOR_ID 0x8086 -#define EM_FLASH 0x0014 - -#define EM_JUMBO_PBA 0x00000028 -#define EM_DEFAULT_PBA 0x00000030 -#define EM_SMARTSPEED_DOWNSHIFT 3 -#define EM_SMARTSPEED_MAX 15 -#define EM_MAX_LOOP 10 - -#define MAX_NUM_MULTICAST_ADDRESSES 128 -#define PCI_ANY_ID (~0U) -#define ETHER_ALIGN 2 -#define EM_FC_PAUSE_TIME 0x0680 -#define EM_EEPROM_APME 0x400; -#define EM_82544_APME 0x0004; - -/* Code compatilbility between 6 and 7 */ -#ifndef ETHER_BPF_MTAP -#define ETHER_BPF_MTAP BPF_MTAP -#endif - -/* - * TDBA/RDBA should be aligned on 16 byte boundary. But TDLEN/RDLEN should be - * multiple of 128 bytes. So we align TDBA/RDBA on 128 byte boundary. This will - * also optimize cache line size effect. H/W supports up to cache line size 128. - */ -#define EM_DBA_ALIGN 128 - -#define SPEED_MODE_BIT (1<<21) /* On PCI-E MACs only */ - -/* PCI Config defines */ -#define EM_BAR_TYPE(v) ((v) & EM_BAR_TYPE_MASK) -#define EM_BAR_TYPE_MASK 0x00000001 -#define EM_BAR_TYPE_MMEM 0x00000000 -#define EM_BAR_TYPE_IO 0x00000001 -#define EM_BAR_TYPE_FLASH 0x0014 -#define EM_BAR_MEM_TYPE(v) ((v) & EM_BAR_MEM_TYPE_MASK) -#define EM_BAR_MEM_TYPE_MASK 0x00000006 -#define EM_BAR_MEM_TYPE_32BIT 0x00000000 -#define EM_BAR_MEM_TYPE_64BIT 0x00000004 -#define EM_MSIX_BAR 3 /* On 82575 */ - -#if __FreeBSD_version < 900000 -#define SYSCTL_ADD_UQUAD SYSCTL_ADD_QUAD -#endif - -/* Defines for printing debug information */ -#define DEBUG_INIT 0 -#define DEBUG_IOCTL 0 -#define DEBUG_HW 0 - -#define INIT_DEBUGOUT(S) if (DEBUG_INIT) printf(S "\n") -#define INIT_DEBUGOUT1(S, A) if (DEBUG_INIT) printf(S "\n", A) -#define INIT_DEBUGOUT2(S, A, B) if (DEBUG_INIT) printf(S "\n", A, B) -#define IOCTL_DEBUGOUT(S) if (DEBUG_IOCTL) printf(S "\n") -#define IOCTL_DEBUGOUT1(S, A) if (DEBUG_IOCTL) printf(S "\n", A) -#define IOCTL_DEBUGOUT2(S, A, B) if (DEBUG_IOCTL) printf(S "\n", A, B) -#define HW_DEBUGOUT(S) if (DEBUG_HW) printf(S "\n") -#define HW_DEBUGOUT1(S, A) if (DEBUG_HW) printf(S "\n", A) -#define HW_DEBUGOUT2(S, A, B) if (DEBUG_HW) printf(S "\n", A, B) - -#define EM_MAX_SCATTER 40 -#define EM_VFTA_SIZE 128 -#define EM_MSIX_MASK 0x01F00000 /* For 82574 use */ -#define ETH_ZLEN 60 -#define ETH_ADDR_LEN 6 -#define CSUM_OFFLOAD 7 /* Offload bits in mbuf flag */ - -/* - * 82574 has a nonstandard address for EIAC - * and since its only used in MSIX, and in - * the em driver only 82574 uses MSIX we can - * solve it just using this define. - */ -#define EM_EIAC 0x000DC - -/* Used in for 82547 10Mb Half workaround */ -#define EM_PBA_BYTES_SHIFT 0xA -#define EM_TX_HEAD_ADDR_SHIFT 7 -#define EM_PBA_TX_MASK 0xFFFF0000 -#define EM_FIFO_HDR 0x10 -#define EM_82547_PKT_THRESH 0x3e0 - -/* Precision Time Sync (IEEE 1588) defines */ -#define ETHERTYPE_IEEE1588 0x88F7 -#define PICOSECS_PER_TICK 20833 -#define TSYNC_PORT 319 /* UDP port for the protocol */ - -#ifdef NIC_PARAVIRT -#define E1000_PARA_SUBDEV 0x1101 /* special id */ -#define E1000_CSBAL 0x02830 /* csb phys. addr. low */ -#define E1000_CSBAH 0x02834 /* csb phys. addr. hi */ -#include -#endif /* NIC_PARAVIRT */ - -/* - * Bus dma allocation structure used by - * e1000_dma_malloc and e1000_dma_free. - */ -struct em_dma_alloc { - bus_addr_t dma_paddr; - caddr_t dma_vaddr; - bus_dma_tag_t dma_tag; - bus_dmamap_t dma_map; - bus_dma_segment_t dma_seg; - int dma_nseg; -}; - -struct adapter; - -struct em_int_delay_info { - struct adapter *adapter; /* Back-pointer to the adapter struct */ - int offset; /* Register offset to read/write */ - int value; /* Current value in usecs */ -}; - -/* Our adapter structure */ -struct adapter { - if_t ifp; - struct e1000_hw hw; - - /* FreeBSD operating-system-specific structures. */ - struct e1000_osdep osdep; - device_t dev; - struct cdev *led_dev; - - struct resource *memory; - struct resource *flash; - struct resource *msix; - - struct resource *ioport; - int io_rid; - - /* 82574 may use 3 int vectors */ - struct resource *res[3]; - void *tag[3]; - int rid[3]; - - struct ifmedia media; - struct callout timer; - struct callout tx_fifo_timer; - bool watchdog_check; - int watchdog_time; - int msi; - int if_flags; - int max_frame_size; - int min_frame_size; - struct mtx core_mtx; - struct mtx tx_mtx; - struct mtx rx_mtx; - int em_insert_vlan_header; - - /* Task for FAST handling */ - struct task link_task; - struct task rxtx_task; - struct task rx_task; - struct task tx_task; - struct taskqueue *tq; /* private task queue */ - - eventhandler_tag vlan_attach; - eventhandler_tag vlan_detach; - u32 num_vlans; - - /* Management and WOL features */ - u32 wol; - bool has_manage; - bool has_amt; - - /* Multicast array memory */ - u8 *mta; - - /* - ** Shadow VFTA table, this is needed because - ** the real vlan filter table gets cleared during - ** a soft reset and the driver needs to be able - ** to repopulate it. - */ - u32 shadow_vfta[EM_VFTA_SIZE]; - - /* Info about the interface */ - uint8_t link_active; - uint16_t link_speed; - uint16_t link_duplex; - uint32_t smartspeed; - uint32_t fc_setting; - - struct em_int_delay_info tx_int_delay; - struct em_int_delay_info tx_abs_int_delay; - struct em_int_delay_info rx_int_delay; - struct em_int_delay_info rx_abs_int_delay; - struct em_int_delay_info tx_itr; - - /* - * Transmit definitions - * - * We have an array of num_tx_desc descriptors (handled - * by the controller) paired with an array of tx_buffers - * (at tx_buffer_area). - * The index of the next available descriptor is next_avail_tx_desc. - * The number of remaining tx_desc is num_tx_desc_avail. - */ - struct em_dma_alloc txdma; /* bus_dma glue for tx desc */ - struct e1000_tx_desc *tx_desc_base; - uint32_t next_avail_tx_desc; - uint32_t next_tx_to_clean; - volatile uint16_t num_tx_desc_avail; - uint16_t num_tx_desc; - uint16_t last_hw_offload; - uint32_t txd_cmd; - struct em_buffer *tx_buffer_area; - bus_dma_tag_t txtag; /* dma tag for tx */ - uint32_t tx_tso; /* last tx was tso */ - - /* - * Receive definitions - * - * we have an array of num_rx_desc rx_desc (handled by the - * controller), and paired with an array of rx_buffers - * (at rx_buffer_area). - * The next pair to check on receive is at offset next_rx_desc_to_check - */ - struct em_dma_alloc rxdma; /* bus_dma glue for rx desc */ - struct e1000_rx_desc *rx_desc_base; - uint32_t next_rx_desc_to_check; - uint32_t rx_buffer_len; - uint16_t num_rx_desc; - int rx_process_limit; - struct em_buffer *rx_buffer_area; - bus_dma_tag_t rxtag; - bus_dmamap_t rx_sparemap; - - /* - * First/last mbuf pointers, for - * collecting multisegment RX packets. - */ - struct mbuf *fmp; - struct mbuf *lmp; - - /* Misc stats maintained by the driver */ - unsigned long dropped_pkts; - unsigned long link_irq; - unsigned long mbuf_cluster_failed; - unsigned long mbuf_defrag_failed; - unsigned long no_tx_desc_avail1; - unsigned long no_tx_desc_avail2; - unsigned long no_tx_dma_setup; - unsigned long no_tx_map_avail; - unsigned long watchdog_events; - unsigned long rx_irq; - unsigned long rx_overruns; - unsigned long tx_irq; - - /* 82547 workaround */ - uint32_t tx_fifo_size; - uint32_t tx_fifo_head; - uint32_t tx_fifo_head_addr; - uint64_t tx_fifo_reset_cnt; - uint64_t tx_fifo_wrk_cnt; - uint32_t tx_head_addr; - - /* For 82544 PCIX Workaround */ - boolean_t pcix_82544; - boolean_t in_detach; - -#ifdef NIC_SEND_COMBINING - /* 0 = idle; 1xxxx int-pending; 3xxxx int + d pending + tdt */ -#define MIT_PENDING_INT 0x10000 /* pending interrupt */ -#define MIT_PENDING_TDT 0x30000 /* both intr and tdt write are pending */ - uint32_t shadow_tdt; - uint32_t sc_enable; -#endif /* NIC_SEND_COMBINING */ -#ifdef BATCH_DISPATCH - uint32_t batch_enable; -#endif /* BATCH_DISPATCH */ - -#ifdef NIC_PARAVIRT - struct em_dma_alloc csb_mem; /* phys address */ - struct paravirt_csb *csb; /* virtual addr */ - uint32_t rx_retries; /* optimize rx loop */ - uint32_t tdt_csb_count;// XXX stat - uint32_t tdt_reg_count;// XXX stat - uint32_t tdt_int_count;// XXX stat - uint32_t guest_need_kick_count;// XXX stat -#endif /* NIC_PARAVIRT */ - - struct e1000_hw_stats stats; -}; - -/* ****************************************************************************** - * vendor_info_array - * - * This array contains the list of Subvendor/Subdevice IDs on which the driver - * should load. - * - * ******************************************************************************/ -typedef struct _em_vendor_info_t { - unsigned int vendor_id; - unsigned int device_id; - unsigned int subvendor_id; - unsigned int subdevice_id; - unsigned int index; -} em_vendor_info_t; - -struct em_buffer { - int next_eop; /* Index of the desc to watch */ - struct mbuf *m_head; - bus_dmamap_t map; /* bus_dma map for packet */ -}; - -/* For 82544 PCIX Workaround */ -typedef struct _ADDRESS_LENGTH_PAIR -{ - uint64_t address; - uint32_t length; -} ADDRESS_LENGTH_PAIR, *PADDRESS_LENGTH_PAIR; - -typedef struct _DESCRIPTOR_PAIR -{ - ADDRESS_LENGTH_PAIR descriptor[4]; - uint32_t elements; -} DESC_ARRAY, *PDESC_ARRAY; - -#define EM_CORE_LOCK_INIT(_sc, _name) \ - mtx_init(&(_sc)->core_mtx, _name, "EM Core Lock", MTX_DEF) -#define EM_TX_LOCK_INIT(_sc, _name) \ - mtx_init(&(_sc)->tx_mtx, _name, "EM TX Lock", MTX_DEF) -#define EM_RX_LOCK_INIT(_sc, _name) \ - mtx_init(&(_sc)->rx_mtx, _name, "EM RX Lock", MTX_DEF) -#define EM_CORE_LOCK_DESTROY(_sc) mtx_destroy(&(_sc)->core_mtx) -#define EM_TX_LOCK_DESTROY(_sc) mtx_destroy(&(_sc)->tx_mtx) -#define EM_RX_LOCK_DESTROY(_sc) mtx_destroy(&(_sc)->rx_mtx) -#define EM_CORE_LOCK(_sc) mtx_lock(&(_sc)->core_mtx) -#define EM_TX_LOCK(_sc) mtx_lock(&(_sc)->tx_mtx) -#define EM_TX_TRYLOCK(_sc) mtx_trylock(&(_sc)->tx_mtx) -#define EM_RX_LOCK(_sc) mtx_lock(&(_sc)->rx_mtx) -#define EM_CORE_UNLOCK(_sc) mtx_unlock(&(_sc)->core_mtx) -#define EM_TX_UNLOCK(_sc) mtx_unlock(&(_sc)->tx_mtx) -#define EM_RX_UNLOCK(_sc) mtx_unlock(&(_sc)->rx_mtx) -#define EM_CORE_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->core_mtx, MA_OWNED) -#define EM_TX_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->tx_mtx, MA_OWNED) - -#endif /* _LEM_H_DEFINED_ */ diff --git a/sys/dev/e1000/igb_txrx.c b/sys/dev/e1000/igb_txrx.c new file mode 100644 index 000000000000..f69e4fe339e2 --- /dev/null +++ b/sys/dev/e1000/igb_txrx.c @@ -0,0 +1,594 @@ +/* $FreeBSD$ */ +#include "if_em.h" + +#ifdef RSS +#include +#include +#endif + +#ifdef VERBOSE_DEBUG +#define DPRINTF device_printf +#else +#define DPRINTF(...) +#endif + +/********************************************************************* + * Local Function prototypes + *********************************************************************/ +static int igb_isc_txd_encap(void *arg, if_pkt_info_t pi); +static void igb_isc_txd_flush(void *arg, uint16_t txqid, uint32_t pidx); +static int igb_isc_txd_credits_update(void *arg, uint16_t txqid, uint32_t cidx, bool clear); + +static void igb_isc_rxd_refill(void *arg, uint16_t rxqid, uint8_t flid __unused, + uint32_t pidx, uint64_t *paddrs, caddr_t *vaddrs __unused, uint16_t count, uint16_t buf_len __unused); +static void igb_isc_rxd_flush(void *arg, uint16_t rxqid, uint8_t flid __unused, uint32_t pidx); +static int igb_isc_rxd_available(void *arg, uint16_t rxqid, uint32_t idx, + int budget); +static int igb_isc_rxd_pkt_get(void *arg, if_rxd_info_t ri); + +static int igb_tx_ctx_setup(struct tx_ring *txr, if_pkt_info_t pi, u32 *cmd_type_len, u32 *olinfo_status); +static int igb_tso_setup(struct tx_ring *txr, if_pkt_info_t pi, u32 *cmd_type_len, u32 *olinfo_status); + +static void igb_rx_checksum(u32 staterr, if_rxd_info_t ri, u32 ptype); +static int igb_determine_rsstype(u16 pkt_info); + +extern void igb_if_enable_intr(if_ctx_t ctx); +extern int em_intr(void *arg); + +struct if_txrx igb_txrx = { + igb_isc_txd_encap, + igb_isc_txd_flush, + igb_isc_txd_credits_update, + igb_isc_rxd_available, + igb_isc_rxd_pkt_get, + igb_isc_rxd_refill, + igb_isc_rxd_flush, + em_intr +}; + +extern if_shared_ctx_t em_sctx; + +/********************************************************************** + * + * Setup work for hardware segmentation offload (TSO) on + * adapters using advanced tx descriptors + * + **********************************************************************/ +static int +igb_tso_setup(struct tx_ring *txr, if_pkt_info_t pi, u32 *cmd_type_len, u32 *olinfo_status) +{ + struct e1000_adv_tx_context_desc *TXD; + struct adapter *adapter = txr->adapter; + u32 type_tucmd_mlhl = 0, vlan_macip_lens = 0; + u32 mss_l4len_idx = 0; + u32 paylen; + + switch(pi->ipi_etype) { + case ETHERTYPE_IPV6: + type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_IPV6; + break; + case ETHERTYPE_IP: + type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_IPV4; + /* Tell transmit desc to also do IPv4 checksum. */ + *olinfo_status |= E1000_TXD_POPTS_IXSM << 8; + break; + default: + panic("%s: CSUM_TSO but no supported IP version (0x%04x)", + __func__, ntohs(pi->ipi_etype)); + break; + } + + TXD = (struct e1000_adv_tx_context_desc *) &txr->tx_base[pi->ipi_pidx]; + + /* This is used in the transmit desc in encap */ + paylen = pi->ipi_len - pi->ipi_ehdrlen - pi->ipi_ip_hlen - pi->ipi_tcp_hlen; + + /* VLAN MACLEN IPLEN */ + if (pi->ipi_mflags & M_VLANTAG) { + vlan_macip_lens |= (pi->ipi_vtag << E1000_ADVTXD_VLAN_SHIFT); + } + + vlan_macip_lens |= pi->ipi_ehdrlen << E1000_ADVTXD_MACLEN_SHIFT; + vlan_macip_lens |= pi->ipi_ip_hlen; + TXD->vlan_macip_lens = htole32(vlan_macip_lens); + + /* ADV DTYPE TUCMD */ + type_tucmd_mlhl |= E1000_ADVTXD_DCMD_DEXT | E1000_ADVTXD_DTYP_CTXT; + type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_L4T_TCP; + TXD->type_tucmd_mlhl = htole32(type_tucmd_mlhl); + + /* MSS L4LEN IDX */ + mss_l4len_idx |= (pi->ipi_tso_segsz << E1000_ADVTXD_MSS_SHIFT); + mss_l4len_idx |= (pi->ipi_tcp_hlen << E1000_ADVTXD_L4LEN_SHIFT); + /* 82575 needs the queue index added */ + if (adapter->hw.mac.type == e1000_82575) + mss_l4len_idx |= txr->me << 4; + TXD->mss_l4len_idx = htole32(mss_l4len_idx); + + TXD->seqnum_seed = htole32(0); + *cmd_type_len |= E1000_ADVTXD_DCMD_TSE; + *olinfo_status |= E1000_TXD_POPTS_TXSM << 8; + *olinfo_status |= paylen << E1000_ADVTXD_PAYLEN_SHIFT; + + return (1); +} + +/********************************************************************* + * + * Advanced Context Descriptor setup for VLAN, CSUM or TSO + * + **********************************************************************/ +static int +igb_tx_ctx_setup(struct tx_ring *txr, if_pkt_info_t pi, u32 *cmd_type_len, u32 *olinfo_status) +{ + struct e1000_adv_tx_context_desc *TXD; + struct adapter *adapter = txr->adapter; + u32 vlan_macip_lens, type_tucmd_mlhl; + u32 mss_l4len_idx; + mss_l4len_idx = vlan_macip_lens = type_tucmd_mlhl = 0; + int offload = TRUE; + + /* First check if TSO is to be used */ + if (pi->ipi_csum_flags & CSUM_TSO) + return (igb_tso_setup(txr, pi, cmd_type_len, olinfo_status)); + + /* Indicate the whole packet as payload when not doing TSO */ + *olinfo_status |= pi->ipi_len << E1000_ADVTXD_PAYLEN_SHIFT; + + /* Now ready a context descriptor */ + TXD = (struct e1000_adv_tx_context_desc *) &txr->tx_base[pi->ipi_pidx]; + + /* + ** In advanced descriptors the vlan tag must + ** be placed into the context descriptor. Hence + ** we need to make one even if not doing offloads. + */ + if (pi->ipi_mflags & M_VLANTAG) { + vlan_macip_lens |= (pi->ipi_vtag << E1000_ADVTXD_VLAN_SHIFT); + } else if ((pi->ipi_csum_flags & CSUM_OFFLOAD) == 0) { + return (0); + } + + /* Set the ether header length */ + vlan_macip_lens |= pi->ipi_ehdrlen << E1000_ADVTXD_MACLEN_SHIFT; + + switch(pi->ipi_etype) { + case ETHERTYPE_IP: + type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_IPV4; + break; + case ETHERTYPE_IPV6: + type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_IPV6; + break; + default: + offload = FALSE; + break; + } + + vlan_macip_lens |= pi->ipi_ip_hlen; + type_tucmd_mlhl |= E1000_ADVTXD_DCMD_DEXT | E1000_ADVTXD_DTYP_CTXT; + + switch (pi->ipi_ipproto) { + case IPPROTO_TCP: + #if __FreeBSD_version >= 1000000 + if (pi->ipi_csum_flags & (CSUM_IP_TCP | CSUM_IP6_TCP)) +#else + if (pi->ipi_csum_flags & CSUM_TCP) +#endif + type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_L4T_TCP; + break; + case IPPROTO_UDP: +#if __FreeBSD_version >= 1000000 + if (pi->ipi_csum_flags & (CSUM_IP_UDP | CSUM_IP6_UDP)) +#else + if (pi->ipi_csum_flags & CSUM_UDP) +#endif + type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_L4T_UDP; + break; + +#if __FreeBSD_version >= 800000 + case IPPROTO_SCTP: +#if __FreeBSD_version >= 1000000 + if (pi->ipi_csum_flags & (CSUM_IP_SCTP | CSUM_IP6_SCTP)) +#else + if (pi->ipi_csum_flags & CSUM_SCTP) +#endif + type_tucmd_mlhl |= E1000_ADVTXD_TUCMD_L4T_SCTP; + break; +#endif + default: + offload = FALSE; + break; + } + + if (offload) /* For the TX descriptor setup */ + *olinfo_status |= E1000_TXD_POPTS_TXSM << 8; + + /* 82575 needs the queue index added */ + if (adapter->hw.mac.type == e1000_82575) + mss_l4len_idx = txr->me << 4; + + /* Now copy bits into descriptor */ + TXD->vlan_macip_lens = htole32(vlan_macip_lens); + TXD->type_tucmd_mlhl = htole32(type_tucmd_mlhl); + TXD->seqnum_seed = htole32(0); + TXD->mss_l4len_idx = htole32(mss_l4len_idx); + + return (1); +} + +static int +igb_isc_txd_encap(void *arg, if_pkt_info_t pi) +{ + struct adapter *sc = arg; + if_softc_ctx_t scctx = sc->shared; + struct em_tx_queue *que = &sc->tx_queues[pi->ipi_qsidx]; + struct tx_ring *txr = &que->txr; + int nsegs = pi->ipi_nsegs; + bus_dma_segment_t *segs = pi->ipi_segs; + struct em_txbuffer *txbuf; + union e1000_adv_tx_desc *txd = NULL; + + int i, j, first, pidx_last; + u32 olinfo_status, cmd_type_len; + + pidx_last = olinfo_status = 0; + /* Basic descriptor defines */ + cmd_type_len = (E1000_ADVTXD_DTYP_DATA | + E1000_ADVTXD_DCMD_IFCS | E1000_ADVTXD_DCMD_DEXT); + + if (pi->ipi_mflags & M_VLANTAG) + cmd_type_len |= E1000_ADVTXD_DCMD_VLE; + + first = i = pi->ipi_pidx; + + /* Consume the first descriptor */ + i += igb_tx_ctx_setup(txr, pi, &cmd_type_len, &olinfo_status); + if (i == scctx->isc_ntxd[0]) + i = 0; + + /* 82575 needs the queue index added */ + if (sc->hw.mac.type == e1000_82575) + olinfo_status |= txr->me << 4; + + for (j = 0; j < nsegs; j++) { + bus_size_t seglen; + bus_addr_t segaddr; + + txbuf = &txr->tx_buffers[i]; + txd = (union e1000_adv_tx_desc *)&txr->tx_base[i]; + seglen = segs[j].ds_len; + segaddr = htole64(segs[j].ds_addr); + + txd->read.buffer_addr = segaddr; + txd->read.cmd_type_len = htole32(E1000_TXD_CMD_IFCS | + cmd_type_len | seglen); + txd->read.olinfo_status = htole32(olinfo_status); + pidx_last = i; + if (++i == scctx->isc_ntxd[0]) { + i = 0; + } + } + + txd->read.cmd_type_len |= + htole32(E1000_TXD_CMD_EOP | E1000_TXD_CMD_RS); + + /* Set the EOP descriptor that will be marked done */ + txbuf = &txr->tx_buffers[first]; + txbuf->eop = pidx_last; + + pi->ipi_new_pidx = i; + + return (0); +} + +static void +igb_isc_txd_flush(void *arg, uint16_t txqid, uint32_t pidx) +{ + struct adapter *adapter = arg; + struct em_tx_queue *que = &adapter->tx_queues[txqid]; + struct tx_ring *txr = &que->txr; + + E1000_WRITE_REG(&adapter->hw, E1000_TDT(txr->me), pidx); +} + +static int +igb_isc_txd_credits_update(void *arg, uint16_t txqid, uint32_t cidx_init, bool clear) +{ + struct adapter *adapter = arg; + if_softc_ctx_t scctx = adapter->shared; + struct em_tx_queue *que = &adapter->tx_queues[txqid]; + struct tx_ring *txr = &que->txr; + + u32 cidx, ntxd, processed = 0; + + struct em_txbuffer *buf; + union e1000_adv_tx_desc *txd, *eop; + int limit; + + cidx = cidx_init; + + buf = &txr->tx_buffers[cidx]; + txd = (union e1000_adv_tx_desc *)&txr->tx_base[cidx]; + ntxd = scctx->isc_ntxd[0]; + limit = adapter->tx_process_limit; + + do { + if (buf->eop == -1) /* No work */ + break; + + eop = (union e1000_adv_tx_desc *)&txr->tx_base[buf->eop]; + if ((eop->wb.status & E1000_TXD_STAT_DD) == 0) + break; /* I/O not complete */ + + if (clear) + buf->eop = -1; /* clear indicate processed */ + + /* We clean the range if multi segment */ + while (txd != eop) { + ++txd; + ++buf; + /* wrap the ring? */ + if (++cidx == scctx->isc_ntxd[0]) { + cidx = 0; + buf = txr->tx_buffers; + txd = (union e1000_adv_tx_desc *)txr->tx_base; + } + + buf = &txr->tx_buffers[cidx]; + if (clear) + buf->eop = -1; + processed++; + } + processed++; + + /* Try the next packet */ + txd++; + buf++; + + /* reset with a wrap */ + if (++cidx == scctx->isc_ntxd[0]) { + cidx = 0; + buf = txr->tx_buffers; + txd = (union e1000_adv_tx_desc *)txr->tx_base; + } + prefetch(txd); + prefetch(txd+1); + } while (__predict_true(--limit) && cidx != cidx_init); + + return (processed); +} + +static void +igb_isc_rxd_refill(void *arg, uint16_t rxqid, uint8_t flid __unused, + uint32_t pidx, uint64_t *paddrs, caddr_t *vaddrs __unused, + uint16_t count, uint16_t buf_len __unused) +{ + struct adapter *sc = arg; + if_softc_ctx_t scctx = sc->shared; + struct em_rx_queue *que = &sc->rx_queues[rxqid]; + union e1000_adv_rx_desc *rxd; + struct rx_ring *rxr = &que->rxr; + int i; + uint32_t next_pidx; + + for (i = 0, next_pidx = pidx; i < count; i++) { + rxd = (union e1000_adv_rx_desc *)&rxr->rx_base[next_pidx]; + + rxd->read.pkt_addr = htole64(paddrs[i]); + if (++next_pidx == scctx->isc_nrxd[0]) + next_pidx = 0; + } +} + +static void +igb_isc_rxd_flush(void *arg, uint16_t rxqid, uint8_t flid __unused, uint32_t pidx) +{ + struct adapter *sc = arg; + struct em_rx_queue *que = &sc->rx_queues[rxqid]; + struct rx_ring *rxr = &que->rxr; + + E1000_WRITE_REG(&sc->hw, E1000_RDT(rxr->me), pidx); +} + +static int +igb_isc_rxd_available(void *arg, uint16_t rxqid, uint32_t idx, int budget) +{ + struct adapter *sc = arg; + if_softc_ctx_t scctx = sc->shared; + struct em_rx_queue *que = &sc->rx_queues[rxqid]; + struct rx_ring *rxr = &que->rxr; + union e1000_adv_rx_desc *rxd; + u32 staterr = 0; + int cnt, i, iter; + + for (iter = cnt = 0, i = idx; iter < scctx->isc_nrxd[0] && iter <= budget;) { + rxd = (union e1000_adv_rx_desc *)&rxr->rx_base[i]; + staterr = le32toh(rxd->wb.upper.status_error); + + if ((staterr & E1000_RXD_STAT_DD) == 0) + break; + + if (++i == scctx->isc_nrxd[0]) { + i = 0; + } + + if (staterr & E1000_RXD_STAT_EOP) + cnt++; + iter++; + } + { + struct e1000_hw *hw = &sc->hw; + int rdt, rdh; + rdt = E1000_READ_REG(hw, E1000_RDT(rxr->me)); + rdh = E1000_READ_REG(hw, E1000_RDH(rxr->me)); + DPRINTF(iflib_get_dev(sc->ctx), "sidx:%d eidx:%d iter=%d pktcnt=%d RDT=%d RDH=%d\n", idx, i, iter, cnt, rdt, rdh); + } + return (cnt); +} + +/**************************************************************** + * Routine sends data which has been dma'ed into host memory + * to upper layer. Initialize ri structure. + * + * Returns 0 upon success, errno on failure + ***************************************************************/ + +static int +igb_isc_rxd_pkt_get(void *arg, if_rxd_info_t ri) +{ + struct adapter *adapter = arg; + if_softc_ctx_t scctx = adapter->shared; + struct em_rx_queue *que = &adapter->rx_queues[ri->iri_qsidx]; + struct rx_ring *rxr = &que->rxr; + struct ifnet *ifp = iflib_get_ifp(adapter->ctx); + union e1000_adv_rx_desc *rxd; + + u16 pkt_info, len; + u16 vtag = 0; + u32 ptype; + u32 staterr = 0; + bool eop; + int i = 0; + int cidx = ri->iri_cidx; + + do { + rxd = (union e1000_adv_rx_desc *)&rxr->rx_base[cidx]; + staterr = le32toh(rxd->wb.upper.status_error); + pkt_info = le16toh(rxd->wb.lower.lo_dword.hs_rss.pkt_info); + + MPASS ((staterr & E1000_RXD_STAT_DD) != 0); + + len = le16toh(rxd->wb.upper.length); + ptype = le32toh(rxd->wb.lower.lo_dword.data) & IGB_PKTTYPE_MASK; + + ri->iri_len += len; + rxr->rx_bytes += ri->iri_len; + + rxd->wb.upper.status_error = 0; + eop = ((staterr & E1000_RXD_STAT_EOP) == E1000_RXD_STAT_EOP); + + if (((adapter->hw.mac.type == e1000_i350) || + (adapter->hw.mac.type == e1000_i354)) && + (staterr & E1000_RXDEXT_STATERR_LB)) + vtag = be16toh(rxd->wb.upper.vlan); + else + vtag = le16toh(rxd->wb.upper.vlan); + + /* Make sure bad packets are discarded */ + if (eop && ((staterr & E1000_RXDEXT_ERR_FRAME_ERR_MASK) != 0)) { + adapter->dropped_pkts++; + ++rxr->rx_discarded; + return (EBADMSG); + } + ri->iri_frags[i].irf_flid = 0; + ri->iri_frags[i].irf_idx = cidx; + ri->iri_frags[i].irf_len = len; + + if (++cidx == scctx->isc_nrxd[0]) + cidx = 0; +#ifdef notyet + if (rxr->hdr_split == TRUE) { + ri->iri_frags[i].irf_flid = 1; + ri->iri_frags[i].irf_idx = cidx; + if (++cidx == scctx->isc_nrxd[0]) + cidx = 0; + } +#endif + i++; + } while (!eop); + + rxr->rx_packets++; + + if ((ifp->if_capenable & IFCAP_RXCSUM) != 0) + igb_rx_checksum(staterr, ri, ptype); + + if ((ifp->if_capenable & IFCAP_VLAN_HWTAGGING) != 0 && + (staterr & E1000_RXD_STAT_VP) != 0) { + ri->iri_vtag = vtag; + ri->iri_flags |= M_VLANTAG; + } + ri->iri_flowid = + le32toh(rxd->wb.lower.hi_dword.rss); + ri->iri_rsstype = igb_determine_rsstype(pkt_info); + ri->iri_nfrags = i; + + return (0); +} + +/********************************************************************* + * + * Verify that the hardware indicated that the checksum is valid. + * Inform the stack about the status of checksum so that stack + * doesn't spend time verifying the checksum. + * + *********************************************************************/ +static void +igb_rx_checksum(u32 staterr, if_rxd_info_t ri, u32 ptype) +{ + u16 status = (u16)staterr; + u8 errors = (u8) (staterr >> 24); + bool sctp = FALSE; + + /* Ignore Checksum bit is set */ + if (status & E1000_RXD_STAT_IXSM) { + ri->iri_csum_flags = 0; + return; + } + + if ((ptype & E1000_RXDADV_PKTTYPE_ETQF) == 0 && + (ptype & E1000_RXDADV_PKTTYPE_SCTP) != 0) + sctp = 1; + else + sctp = 0; + + if (status & E1000_RXD_STAT_IPCS) { + /* Did it pass? */ + if (!(errors & E1000_RXD_ERR_IPE)) { + /* IP Checksum Good */ + ri->iri_csum_flags = CSUM_IP_CHECKED; + ri->iri_csum_flags |= CSUM_IP_VALID; + } else + ri->iri_csum_flags = 0; + } + + if (status & (E1000_RXD_STAT_TCPCS | E1000_RXD_STAT_UDPCS)) { + u64 type = (CSUM_DATA_VALID | CSUM_PSEUDO_HDR); +#if __FreeBSD_version >= 800000 + if (sctp) /* reassign */ + type = CSUM_SCTP_VALID; +#endif + /* Did it pass? */ + if (!(errors & E1000_RXD_ERR_TCPE)) { + ri->iri_csum_flags |= type; + if (sctp == 0) + ri->iri_csum_data = htons(0xffff); + } + } + return; +} + +/******************************************************************** + * + * Parse the packet type to determine the appropriate hash + * + ******************************************************************/ +static int +igb_determine_rsstype(u16 pkt_info) +{ + switch (pkt_info & E1000_RXDADV_RSSTYPE_MASK) { + case E1000_RXDADV_RSSTYPE_IPV4_TCP: + return M_HASHTYPE_RSS_TCP_IPV4; + case E1000_RXDADV_RSSTYPE_IPV4: + return M_HASHTYPE_RSS_IPV4; + case E1000_RXDADV_RSSTYPE_IPV6_TCP: + return M_HASHTYPE_RSS_TCP_IPV6; + case E1000_RXDADV_RSSTYPE_IPV6_EX: + return M_HASHTYPE_RSS_IPV6_EX; + case E1000_RXDADV_RSSTYPE_IPV6: + return M_HASHTYPE_RSS_IPV6; + case E1000_RXDADV_RSSTYPE_IPV6_TCP_EX: + return M_HASHTYPE_RSS_TCP_IPV6_EX; + default: + return M_HASHTYPE_OPAQUE; + } +} diff --git a/sys/i386/conf/GENERIC b/sys/i386/conf/GENERIC index cbb15a91933e..f536927fc2c5 100644 --- a/sys/i386/conf/GENERIC +++ b/sys/i386/conf/GENERIC @@ -235,7 +235,6 @@ device puc # Multi I/O cards and multi-channel UARTs device bxe # Broadcom NetXtreme II BCM5771X/BCM578XX 10GbE device de # DEC/Intel DC21x4x (``Tulip'') device em # Intel PRO/1000 Gigabit Ethernet Family -device igb # Intel PRO/1000 PCIE Server Gigabit Family device ixgb # Intel PRO/10GbE Ethernet Card device le # AMD Am7900 LANCE and Am79C9xx PCnet device ti # Alteon Networks Tigon I/II gigabit Ethernet diff --git a/sys/mips/conf/OCTEON1 b/sys/mips/conf/OCTEON1 index d3ca58ca7faf..34f8ef1d6408 100644 --- a/sys/mips/conf/OCTEON1 +++ b/sys/mips/conf/OCTEON1 @@ -195,7 +195,6 @@ device octm # PCI Ethernet NICs. device de # DEC/Intel DC21x4x (``Tulip'') device em # Intel PRO/1000 Gigabit Ethernet Family -device igb # Intel PRO/1000 PCIE Server Gigabit Family device ix # Intel PRO/10GbE PF PCIE Ethernet Family device ixv # Intel PRO/10GbE VF PCIE Ethernet Family device le # AMD Am7900 LANCE and Am79C9xx PCnet diff --git a/sys/modules/Makefile b/sys/modules/Makefile index 69e5fc4d64e4..2bccd1883e55 100644 --- a/sys/modules/Makefile +++ b/sys/modules/Makefile @@ -163,7 +163,6 @@ SUBDIR= \ if_tun \ if_vlan \ if_vxlan \ - ${_igb} \ ${_iir} \ imgact_binmisc \ ${_intelspi} \ @@ -544,7 +543,6 @@ _cxgb= cxgb .if ${MACHINE_CPUARCH} == "aarch64" _armv8crypto= armv8crypto _em= em -_igb= igb .endif .if ${MACHINE_CPUARCH} == "i386" || ${MACHINE_CPUARCH} == "amd64" @@ -572,7 +570,6 @@ _fe= fe _ibcore= ibcore .endif _if_ndis= if_ndis -_igb= igb _io= io .if ${MK_OFED} != "no" || defined(ALL_MODULES) _ipoib= ipoib @@ -779,7 +776,6 @@ _nvram= powermac_nvram _auxio= auxio _em= em _epic= epic -_igb= igb .endif .if (${MACHINE_CPUARCH} == "amd64" || ${MACHINE_ARCH} == "armv6" || \ diff --git a/sys/modules/em/Makefile b/sys/modules/em/Makefile index 1de7c62298e9..7747d09f233b 100644 --- a/sys/modules/em/Makefile +++ b/sys/modules/em/Makefile @@ -1,23 +1,28 @@ + # $FreeBSD$ .PATH: ${.CURDIR}/../../dev/e1000 KMOD = if_em SRCS = device_if.h bus_if.h pci_if.h opt_ddb.h opt_em.h opt_inet.h \ - opt_inet6.h + opt_inet6.h ifdi_if.h SRCS += $(CORE_SRC) $(LEGACY_SRC) SRCS += $(COMMON_SHARED) $(LEGACY_SHARED) $(PCIE_SHARED) -CORE_SRC = if_em.c e1000_osdep.c +CORE_SRC = if_em.c em_txrx.c e1000_osdep.c +CORE_SRC += igb_txrx.c # This is the Legacy, pre-PCIE source, it can be # undefined when using modular driver if not needed -LEGACY_SRC += if_lem.c COMMON_SHARED = e1000_api.c e1000_phy.c e1000_nvm.c e1000_mac.c \ e1000_manage.c e1000_vf.c e1000_mbx.c e1000_i210.c PCIE_SHARED = e1000_80003es2lan.c e1000_ich8lan.c e1000_82571.c e1000_82575.c LEGACY_SHARED = e1000_82540.c e1000_82542.c e1000_82541.c e1000_82543.c -CFLAGS += -I${.CURDIR}/../../dev/e1000 + +CFLAGS += -I${.CURDIR}/../../../dev/e1000 # DEVICE_POLLING for a non-interrupt-driven method #CFLAGS += -DDEVICE_POLLING +afterinstall: + ln -sf ${DESTDIR}${KMODDIR}/${KMOD}.ko ${DESTDIR}${KMODDIR}/if_igb.ko + .include diff --git a/sys/modules/igb/Makefile b/sys/modules/igb/Makefile deleted file mode 100644 index 4db960826593..000000000000 --- a/sys/modules/igb/Makefile +++ /dev/null @@ -1,24 +0,0 @@ -#$FreeBSD$ - -.PATH: ${.CURDIR}/../../dev/e1000 -KMOD = if_igb -SRCS = device_if.h bus_if.h pci_if.h opt_inet.h opt_inet6.h opt_rss.h -SRCS += if_igb.c $(SHARED_SRCS) -SHARED_SRCS = e1000_api.c e1000_phy.c e1000_nvm.c e1000_mac.c e1000_manage.c -SHARED_SRCS += e1000_80003es2lan.c e1000_82542.c e1000_82541.c e1000_82543.c -SHARED_SRCS += e1000_82540.c e1000_ich8lan.c e1000_82571.c e1000_osdep.c -SHARED_SRCS += e1000_82575.c e1000_vf.c e1000_mbx.c e1000_i210.c - -CFLAGS += -I${.CURDIR}/../../dev/e1000 -DSMP - -# DEVICE_POLLING gives you non-interrupt handling -# not advisable since MSIX gives better results -#CFLAGS += -DDEVICE_POLLING - -# IGB_LEGACY_TX will override the stack if_transmit path and -# instead use the older if_start non-multiqueue capable interface. -# This might be desirable for testing, or to enable the use of -# ALTQ. -#CFLAGS += -DIGB_LEGACY_TX - -.include diff --git a/sys/powerpc/conf/GENERIC64 b/sys/powerpc/conf/GENERIC64 index 7ebcb7a8f21b..2325bb059d78 100644 --- a/sys/powerpc/conf/GENERIC64 +++ b/sys/powerpc/conf/GENERIC64 @@ -139,7 +139,6 @@ device uart_z8530 # Ethernet hardware device em # Intel PRO/1000 Gigabit Ethernet Family -device igb # Intel PRO/1000 PCIE Server Gigabit Family device ix # Intel PRO/10GbE PCIE PF Ethernet Family device ixv # Intel PRO/10GbE PCIE VF Ethernet Family device glc # Sony Playstation 3 Ethernet From cfed2e639ec4efd032b1d6d4fd8e9441f8241d58 Mon Sep 17 00:00:00 2001 From: Ian Lepore Date: Tue, 10 Jan 2017 03:53:38 +0000 Subject: [PATCH 18/79] Use the post-reset hook to force the controller to host mode. This will make both usb ports work on imx6 systems (the OTG port of course will only work in host mode). --- sys/dev/usb/controller/ehci_imx.c | 21 +++++++++++++++++++-- 1 file changed, 19 insertions(+), 2 deletions(-) diff --git a/sys/dev/usb/controller/ehci_imx.c b/sys/dev/usb/controller/ehci_imx.c index 07f73105f9ab..df4d202c38ef 100644 --- a/sys/dev/usb/controller/ehci_imx.c +++ b/sys/dev/usb/controller/ehci_imx.c @@ -157,6 +157,18 @@ struct imx_ehci_softc { struct resource *ehci_irq_res; /* EHCI core IRQ. */ }; +static void +imx_ehci_post_reset(struct ehci_softc *ehci_softc) +{ + uint32_t usbmode; + + /* Force HOST mode */ + usbmode = EOREAD4(ehci_softc, EHCI_USBMODE_NOLPM); + usbmode &= ~EHCI_UM_CM; + usbmode |= EHCI_UM_CM_HOST; + EOWRITE4(ehci_softc, EHCI_USBMODE_NOLPM, usbmode); +} + static int imx_ehci_probe(device_t dev) { @@ -282,8 +294,13 @@ imx_ehci_attach(device_t dev) esc->sc_id_vendor = USB_VENDOR_FREESCALE; strlcpy(esc->sc_vendor, "Freescale", sizeof(esc->sc_vendor)); - /* Set flags that affect ehci_init() behavior. */ - esc->sc_flags |= EHCI_SCFLG_DONTRESET | EHCI_SCFLG_NORESTERM; + /* + * Set flags that affect ehci_init() behavior, and hook our post-reset + * code into the standard controller code. + */ + esc->sc_flags |= EHCI_SCFLG_NORESTERM; + esc->sc_vendor_post_reset = imx_ehci_post_reset; + err = ehci_init(esc); if (err != 0) { device_printf(dev, "USB init failed, usb_err_t=%d\n", From bedd46202f908ad4fff93203e5eb254f1f4127be Mon Sep 17 00:00:00 2001 From: Mateusz Guzik Date: Tue, 10 Jan 2017 04:17:53 +0000 Subject: [PATCH 19/79] cd9660: fix up compilation on sparc after r311665 Reported by: linimon --- sys/fs/cd9660/cd9660_vfsops.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/sys/fs/cd9660/cd9660_vfsops.c b/sys/fs/cd9660/cd9660_vfsops.c index 2d97029a6642..96c7dcfbaee4 100644 --- a/sys/fs/cd9660/cd9660_vfsops.c +++ b/sys/fs/cd9660/cd9660_vfsops.c @@ -88,7 +88,7 @@ static struct vfsops cd9660_vfsops = { VFS_SET(cd9660_vfsops, cd9660, VFCF_READONLY); MODULE_VERSION(cd9660, 1); -static int cd9660_vfs_hash_cmp(struct vnode *vp, cd_ino_t *pino); +static int cd9660_vfs_hash_cmp(struct vnode *vp, void *pino); static int iso_mountfs(struct vnode *devvp, struct mount *mp); /* @@ -650,12 +650,14 @@ cd9660_vget(mp, ino, flags, vpp) static int cd9660_vfs_hash_cmp(vp, pino) struct vnode *vp; - cd_ino_t *pino; + void *pino; { struct iso_node *ip; + cd_ino_t ino; ip = VTOI(vp); - return (ip->i_number != *pino); + ino = *(cd_ino_t *)pino; + return (ip->i_number != ino); } int From b088abafd5f5fb5f5770cbd7e55b1d51c1a94a2b Mon Sep 17 00:00:00 2001 From: Larry Rosenman Date: Tue, 10 Jan 2017 04:31:56 +0000 Subject: [PATCH 20/79] Add myself to committers-ports.dot Approved by: adamw (mentor) Differential Revision: https://reviews.freebsd.org/D9117 --- share/misc/committers-ports.dot | 3 +++ 1 file changed, 3 insertions(+) diff --git a/share/misc/committers-ports.dot b/share/misc/committers-ports.dot index 3da0e274a68a..85dfac4a78cd 100644 --- a/share/misc/committers-ports.dot +++ b/share/misc/committers-ports.dot @@ -144,6 +144,7 @@ laszlof [label="Frank Laszlo\nlaszlof@FreeBSD.org\n2006/11/07"] lawrance [label="Sam Lawrance\nlawrance@FreeBSD.org\n2005/04/11\n2007/02/21"] lbr [label="Lars Balker Rasmussen\nlbr@FreeBSD.org\n2006/04/30"] leeym [label="Yen-Ming Lee\nleeym@FreeBSD.org\n2002/08/14"] +ler [label="Larry Rosenman\nler@FreeBSD.org\n2017/01/09"] lev [label="Lev Serebryakov\nlev@FreeBSD.org\n2003/06/17"] lifanov [label="Nikolai Lifanov\nlifanov@FreeBSD.org\n2016/12/11"] linimon [label="Mark Linimon\nlinimon@FreeBSD.org\n2003/10/23"] @@ -252,6 +253,7 @@ znerd [label="Ernst de Haan\nznerd@FreeBSD.org\n2001/11/15"] adamw -> ahze adamw -> jylefort +adamw -> ler adamw -> mezz adamw -> pav adamw -> woodsb02 @@ -554,6 +556,7 @@ rene -> bar rene -> cmt rene -> crees rene -> jgh +rene -> ler rene -> olivierd rm -> koobs From 994df66b4c30062aa064040abb64084ff62cd6e0 Mon Sep 17 00:00:00 2001 From: Adam Weinberger Date: Tue, 10 Jan 2017 04:49:59 +0000 Subject: [PATCH 21/79] As much as I've enjoyed being listed as emeritus for the last 10+ years, it's probably time to admit that I am an active committer. --- share/misc/committers-ports.dot | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/share/misc/committers-ports.dot b/share/misc/committers-ports.dot index 85dfac4a78cd..842620d7bf62 100644 --- a/share/misc/committers-ports.dot +++ b/share/misc/committers-ports.dot @@ -29,7 +29,6 @@ node [color=grey62, style=filled, bgcolor=black]; # Alumni go here.. Try to keep things sorted. -adamw [label="Adam Weinberger\nadamw@FreeBSD.org\n2002/10/16\n2006/09/25"] asami [label="Satoshi Asami\nasami@FreeBSD.org\n1994/11/18\n2001/09/11"] billf [label="Bill Fumerola\nbillf@FreeBSD.org\n1998/11/11\n2006/12/14"] jmallett [label="Juli Mallett\njmallett@FreeBSD.org\n2003/01/16\n2006/08/10"] @@ -43,6 +42,7 @@ node [color=lightblue2, style=filled, bgcolor=black]; ache [label="Andrey Chernov\nache@FreeBSD.org\n1994/11/15"] acm [label="Jose Alonso Cardenas Marquez\nacm@FreeBSD.org\n2006/07/18"] +adamw [label="Adam Weinberger\nadamw@FreeBSD.org\n2002/10/16"] ahze [label="Michael Johnson\nahze@FreeBSD.org\n2004/10/29"] ak [label="Alex Kozlov\nak@FreeBSD.org\n2012/02/29"] ale [label="Alex Dupre\nale@FreeBSD.org\n2004/01/12"] From d37cece2b019bd03119c514eedbd7a349cf6fba0 Mon Sep 17 00:00:00 2001 From: Sean Bruno Date: Tue, 10 Jan 2017 04:50:26 +0000 Subject: [PATCH 22/79] Add copywrite notices, 2-clause BSD. Reported by: jmallett --- sys/dev/e1000/em_txrx.c | 26 ++++++++++++++++++++++++++ sys/dev/e1000/if_em.c | 26 ++++++++++++++++++++++++++ sys/dev/e1000/if_em.h | 26 ++++++++++++++++++++++++++ sys/dev/e1000/igb_txrx.c | 26 ++++++++++++++++++++++++++ 4 files changed, 104 insertions(+) diff --git a/sys/dev/e1000/em_txrx.c b/sys/dev/e1000/em_txrx.c index cf52656c41c6..54c8880bb758 100644 --- a/sys/dev/e1000/em_txrx.c +++ b/sys/dev/e1000/em_txrx.c @@ -1,3 +1,29 @@ +/*- + * Copyright (c) 2016 Matt Macy + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + /* $FreeBSD$ */ #include "if_em.h" diff --git a/sys/dev/e1000/if_em.c b/sys/dev/e1000/if_em.c index a49f578682c4..629ae91025d3 100644 --- a/sys/dev/e1000/if_em.c +++ b/sys/dev/e1000/if_em.c @@ -1,3 +1,29 @@ +/*- + * Copyright (c) 2016 Matt Macy + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + /* $FreeBSD$ */ #include "if_em.h" #include diff --git a/sys/dev/e1000/if_em.h b/sys/dev/e1000/if_em.h index f4cc2338c133..35755d751b3b 100644 --- a/sys/dev/e1000/if_em.h +++ b/sys/dev/e1000/if_em.h @@ -1,3 +1,29 @@ +/*- + * Copyright (c) 2016 Matt Macy + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + /*$FreeBSD$*/ #include "opt_em.h" #include "opt_ddb.h" diff --git a/sys/dev/e1000/igb_txrx.c b/sys/dev/e1000/igb_txrx.c index f69e4fe339e2..28178a59074b 100644 --- a/sys/dev/e1000/igb_txrx.c +++ b/sys/dev/e1000/igb_txrx.c @@ -1,3 +1,29 @@ +/*- + * Copyright (c) 2016 Matt Macy + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + /* $FreeBSD$ */ #include "if_em.h" From 791be271f18af0b66d0a4e3a5c48915a4d5b8b0c Mon Sep 17 00:00:00 2001 From: Adrian Chadd Date: Tue, 10 Jan 2017 05:30:15 +0000 Subject: [PATCH 23/79] [net80211] create a helper function to calculate the station facing VHT capabilities. This is needed for two reasons: * Drivers will need to know what the negotiated set of VHT capabilities and rates are in order to configure (and reconfigure for opmode/chanwidth changes) how to speak to a given peer; and * Because some vendors are "special", we should be careful in what we announce to them during peer association. This isn't the complete solution, as I still need to make sure that when sending out probe requests before we know what we want, we don't limit the capabilities being announced. This is important for IBSS/mesh work later on as probe request/response exchanges are the first hint at what a peer supports. I'll look at adding that to the API soon. --- sys/net80211/ieee80211_vht.c | 397 ++++++++++++++++++++++++++++++++--- sys/net80211/ieee80211_vht.h | 5 + 2 files changed, 370 insertions(+), 32 deletions(-) diff --git a/sys/net80211/ieee80211_vht.c b/sys/net80211/ieee80211_vht.c index 274d94b25208..15012f254e70 100644 --- a/sys/net80211/ieee80211_vht.c +++ b/sys/net80211/ieee80211_vht.c @@ -70,6 +70,15 @@ __FBSDID("$FreeBSD$"); frm += 4; \ } while (0) +/* + * Immediate TODO: + * + * + handle WLAN_ACTION_VHT_OPMODE_NOTIF and other VHT action frames + * + ensure vhtinfo/vhtcap parameters correctly use the negotiated + * capabilities and ratesets + * + group ID management operation + */ + /* * XXX TODO: handle WLAN_ACTION_VHT_OPMODE_NOTIF * @@ -153,9 +162,9 @@ ieee80211_vht_announce(struct ieee80211com *ic) /* Channel width */ ic_printf(ic, "[VHT] Channel Widths: 20MHz, 40MHz, 80MHz"); - if (ic->ic_vhtcaps & IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_160_80PLUS80MHZ) + if (MS(ic->ic_vhtcaps, IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_MASK) == 2) printf(" 80+80MHz"); - if (ic->ic_vhtcaps & IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_160MHZ) + if (MS(ic->ic_vhtcaps, IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_MASK) >= 1) printf(" 160MHz"); printf("\n"); @@ -280,10 +289,342 @@ ieee80211_vht_node_leave(struct ieee80211_node *ni) "%s: called", __func__); } +/* + * Calculate the VHTCAP IE for a given node. + * + * This includes calculating the capability intersection based on the + * current operating mode and intersection of the TX/RX MCS maps. + * + * The standard only makes it clear about MCS rate negotiation + * and MCS basic rates (which must be a subset of the general + * negotiated rates). It doesn't make it clear that the AP should + * figure out the minimum functional overlap with the STA and + * support that. + * + * Note: this is in host order, not in 802.11 endian order. + * + * TODO: ensure I re-read 9.7.11 Rate Selection for VHT STAs. + * + * TODO: investigate what we should negotiate for MU-MIMO beamforming + * options. + * + * opmode is '1' for "vhtcap as if I'm a STA", 0 otherwise. + */ +void +ieee80211_vht_get_vhtcap_ie(struct ieee80211_node *ni, + struct ieee80211_ie_vhtcap *vhtcap, int opmode) +{ + struct ieee80211vap *vap = ni->ni_vap; +// struct ieee80211com *ic = vap->iv_ic; + uint32_t val, val1, val2; + uint32_t new_vhtcap; + int i; + + vhtcap->ie = IEEE80211_ELEMID_VHT_CAP; + vhtcap->len = sizeof(struct ieee80211_ie_vhtcap) - 2; + + /* + * Capabilities - it depends on whether we are a station + * or not. + */ + new_vhtcap = 0; + + /* + * Station - use our desired configuration based on + * local config, local device bits and the already-learnt + * vhtcap/vhtinfo IE in the node. + */ + + /* Limit MPDU size to the smaller of the two */ + val2 = val1 = MS(vap->iv_vhtcaps, IEEE80211_VHTCAP_MAX_MPDU_MASK); + if (opmode == 1) { + val2 = MS(ni->ni_vhtcap, IEEE80211_VHTCAP_MAX_MPDU_MASK); + } + val = MIN(val1, val2); + new_vhtcap |= SM(val, IEEE80211_VHTCAP_MAX_MPDU_MASK); + + /* Limit supp channel config */ + val2 = val1 = MS(vap->iv_vhtcaps, + IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_MASK); + if (opmode == 1) { + val2 = MS(ni->ni_vhtcap, + IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_MASK); + } + if ((val2 == 2) && + ((vap->iv_flags_vht & IEEE80211_FVHT_USEVHT80P80) == 0)) + val2 = 1; + if ((val2 == 1) && + ((vap->iv_flags_vht & IEEE80211_FVHT_USEVHT160) == 0)) + val2 = 0; + val = MIN(val1, val2); + new_vhtcap |= SM(val, IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_MASK); + + /* RX LDPC */ + val2 = val1 = MS(vap->iv_vhtcaps, IEEE80211_VHTCAP_RXLDPC); + if (opmode == 1) { + val2 = MS(ni->ni_vhtcap, IEEE80211_VHTCAP_RXLDPC); + } + val = MIN(val1, val2); + new_vhtcap |= SM(val, IEEE80211_VHTCAP_RXLDPC); + + /* Short-GI 80 */ + val2 = val1 = MS(vap->iv_vhtcaps, IEEE80211_VHTCAP_SHORT_GI_80); + if (opmode == 1) { + val2 = MS(ni->ni_vhtcap, IEEE80211_VHTCAP_SHORT_GI_80); + } + val = MIN(val1, val2); + new_vhtcap |= SM(val, IEEE80211_VHTCAP_SHORT_GI_80); + + /* Short-GI 160 */ + val2 = val1 = MS(vap->iv_vhtcaps, IEEE80211_VHTCAP_SHORT_GI_160); + if (opmode == 1) { + val2 = MS(ni->ni_vhtcap, IEEE80211_VHTCAP_SHORT_GI_160); + } + val = MIN(val1, val2); + new_vhtcap |= SM(val, IEEE80211_VHTCAP_SHORT_GI_160); + + /* + * STBC is slightly more complicated. + * + * In non-STA mode, we just announce our capabilities and that + * is that. + * + * In STA mode, we should calculate our capabilities based on + * local capabilities /and/ what the remote says. So: + * + * + Only TX STBC if we support it and the remote supports RX STBC; + * + Only announce RX STBC if we support it and the remote supports + * TX STBC; + * + RX STBC should be the minimum of local and remote RX STBC; + */ + + /* TX STBC */ + val2 = val1 = MS(vap->iv_vhtcaps, IEEE80211_VHTCAP_TXSTBC); + if (opmode == 1) { + /* STA mode - enable it only if node RXSTBC is non-zero */ + val2 = !! MS(ni->ni_vhtcap, IEEE80211_VHTCAP_RXSTBC_MASK); + } + val = MIN(val1, val2); + /* XXX For now, use the 11n config flag */ + if ((vap->iv_flags_ht & IEEE80211_FHT_STBC_TX) == 0) + val = 0; + new_vhtcap |= SM(val, IEEE80211_VHTCAP_TXSTBC); + + /* RX STBC1..4 */ + val2 = val1 = MS(vap->iv_vhtcaps, IEEE80211_VHTCAP_RXSTBC_MASK); + if (opmode == 1) { + /* STA mode - enable it only if node TXSTBC is non-zero */ + val2 = MS(ni->ni_vhtcap, IEEE80211_VHTCAP_TXSTBC); + } + val = MIN(val1, val2); + /* XXX For now, use the 11n config flag */ + if ((vap->iv_flags_ht & IEEE80211_FHT_STBC_RX) == 0) + val = 0; + new_vhtcap |= SM(val, IEEE80211_VHTCAP_RXSTBC_MASK); + + /* + * Finally - if RXSTBC is 0, then don't enable TXSTBC. + * Strictly speaking a device can TXSTBC and not RXSTBC, but + * it would be silly. + */ + if (val == 0) + new_vhtcap &= ~IEEE80211_VHTCAP_TXSTBC; + + /* + * Some of these fields require other fields to exist. + * So before using it, the parent field needs to be checked + * otherwise the overridden value may be wrong. + * + * For example, if SU beamformee is set to 0, then BF STS + * needs to be 0. + */ + + /* SU Beamformer capable */ + val2 = val1 = MS(vap->iv_vhtcaps, + IEEE80211_VHTCAP_SU_BEAMFORMER_CAPABLE); + if (opmode == 1) { + val2 = MS(ni->ni_vhtcap, + IEEE80211_VHTCAP_SU_BEAMFORMER_CAPABLE); + } + val = MIN(val1, val2); + new_vhtcap |= SM(val, IEEE80211_VHTCAP_SU_BEAMFORMER_CAPABLE); + + /* SU Beamformee capable */ + val2 = val1 = MS(vap->iv_vhtcaps, + IEEE80211_VHTCAP_SU_BEAMFORMEE_CAPABLE); + if (opmode == 1) { + val2 = MS(ni->ni_vhtcap, + IEEE80211_VHTCAP_SU_BEAMFORMEE_CAPABLE); + } + val = MIN(val1, val2); + new_vhtcap |= SM(val, IEEE80211_VHTCAP_SU_BEAMFORMEE_CAPABLE); + + /* Beamformee STS capability - only if SU beamformee capable */ + val2 = val1 = MS(vap->iv_vhtcaps, IEEE80211_VHTCAP_BEAMFORMEE_STS_MASK); + if (opmode == 1) { + val2 = MS(ni->ni_vhtcap, IEEE80211_VHTCAP_BEAMFORMEE_STS_MASK); + } + val = MIN(val1, val2); + if ((new_vhtcap & IEEE80211_VHTCAP_SU_BEAMFORMEE_CAPABLE) == 0) + val = 0; + new_vhtcap |= SM(val, IEEE80211_VHTCAP_BEAMFORMEE_STS_MASK); + + /* Sounding dimensions - only if SU beamformer capable */ + val2 = val1 = MS(vap->iv_vhtcaps, + IEEE80211_VHTCAP_SOUNDING_DIMENSIONS_MASK); + if (opmode == 1) + val2 = MS(ni->ni_vhtcap, + IEEE80211_VHTCAP_SOUNDING_DIMENSIONS_MASK); + val = MIN(val1, val2); + if ((new_vhtcap & IEEE80211_VHTCAP_SU_BEAMFORMER_CAPABLE) == 0) + val = 0; + new_vhtcap |= SM(val, IEEE80211_VHTCAP_SOUNDING_DIMENSIONS_MASK); + + /* + * MU Beamformer capable - only if SU BFF capable, MU BFF capable + * and STA (not AP) + */ + val2 = val1 = MS(vap->iv_vhtcaps, + IEEE80211_VHTCAP_MU_BEAMFORMER_CAPABLE); + if (opmode == 1) + val2 = MS(ni->ni_vhtcap, + IEEE80211_VHTCAP_MU_BEAMFORMER_CAPABLE); + val = MIN(val1, val2); + if ((new_vhtcap & IEEE80211_VHTCAP_SU_BEAMFORMER_CAPABLE) == 0) + val = 0; + if (opmode != 1) /* Only enable for STA mode */ + val = 0; + new_vhtcap |= SM(val, IEEE80211_VHTCAP_SU_BEAMFORMER_CAPABLE); + + /* + * MU Beamformee capable - only if SU BFE capable, MU BFE capable + * and AP (not STA) + */ + val2 = val1 = MS(vap->iv_vhtcaps, + IEEE80211_VHTCAP_MU_BEAMFORMEE_CAPABLE); + if (opmode == 1) + val2 = MS(ni->ni_vhtcap, + IEEE80211_VHTCAP_MU_BEAMFORMEE_CAPABLE); + val = MIN(val1, val2); + if ((new_vhtcap & IEEE80211_VHTCAP_SU_BEAMFORMEE_CAPABLE) == 0) + val = 0; + if (opmode != 0) /* Only enable for AP mode */ + val = 0; + new_vhtcap |= SM(val, IEEE80211_VHTCAP_SU_BEAMFORMEE_CAPABLE); + + /* VHT TXOP PS */ + val2 = val1 = MS(vap->iv_vhtcaps, IEEE80211_VHTCAP_VHT_TXOP_PS); + if (opmode == 1) + val2 = MS(ni->ni_vhtcap, IEEE80211_VHTCAP_VHT_TXOP_PS); + val = MIN(val1, val2); + new_vhtcap |= SM(val, IEEE80211_VHTCAP_VHT_TXOP_PS); + + /* HTC_VHT */ + val2 = val1 = MS(vap->iv_vhtcaps, IEEE80211_VHTCAP_HTC_VHT); + if (opmode == 1) + val2 = MS(ni->ni_vhtcap, IEEE80211_VHTCAP_HTC_VHT); + val = MIN(val1, val2); + new_vhtcap |= SM(val, IEEE80211_VHTCAP_HTC_VHT); + + /* A-MPDU length max */ + /* XXX TODO: we need a userland config knob for this */ + val2 = val1 = MS(vap->iv_vhtcaps, + IEEE80211_VHTCAP_MAX_A_MPDU_LENGTH_EXPONENT_MASK); + if (opmode == 1) + val2 = MS(ni->ni_vhtcap, + IEEE80211_VHTCAP_MAX_A_MPDU_LENGTH_EXPONENT_MASK); + val = MIN(val1, val2); + new_vhtcap |= SM(val, IEEE80211_VHTCAP_MAX_A_MPDU_LENGTH_EXPONENT_MASK); + + /* + * Link adaptation is only valid if HTC-VHT capable is 1. + * Otherwise, always set it to 0. + */ + val2 = val1 = MS(vap->iv_vhtcaps, + IEEE80211_VHTCAP_VHT_LINK_ADAPTATION_VHT_MASK); + if (opmode == 1) + val2 = MS(ni->ni_vhtcap, + IEEE80211_VHTCAP_VHT_LINK_ADAPTATION_VHT_MASK); + val = MIN(val1, val2); + if ((new_vhtcap & IEEE80211_VHTCAP_HTC_VHT) == 0) + val = 0; + new_vhtcap |= SM(val, IEEE80211_VHTCAP_VHT_LINK_ADAPTATION_VHT_MASK); + + /* + * The following two options are 0 if the pattern may change, 1 if it + * does not change. So, downgrade to the higher value. + */ + + /* RX antenna pattern */ + val2 = val1 = MS(vap->iv_vhtcaps, IEEE80211_VHTCAP_RX_ANTENNA_PATTERN); + if (opmode == 1) + val2 = MS(ni->ni_vhtcap, IEEE80211_VHTCAP_RX_ANTENNA_PATTERN); + val = MAX(val1, val2); + new_vhtcap |= SM(val, IEEE80211_VHTCAP_RX_ANTENNA_PATTERN); + + /* TX antenna pattern */ + val2 = val1 = MS(vap->iv_vhtcaps, IEEE80211_VHTCAP_TX_ANTENNA_PATTERN); + if (opmode == 1) + val2 = MS(ni->ni_vhtcap, IEEE80211_VHTCAP_TX_ANTENNA_PATTERN); + val = MAX(val1, val2); + new_vhtcap |= SM(val, IEEE80211_VHTCAP_TX_ANTENNA_PATTERN); + + /* + * MCS set - again, we announce what we want to use + * based on configuration, device capabilities and + * already-learnt vhtcap/vhtinfo IE information. + */ + + /* MCS set - start with whatever the device supports */ + vhtcap->supp_mcs.rx_mcs_map = vap->iv_vht_mcsinfo.rx_mcs_map; + vhtcap->supp_mcs.rx_highest = 0; + vhtcap->supp_mcs.tx_mcs_map = vap->iv_vht_mcsinfo.tx_mcs_map; + vhtcap->supp_mcs.tx_highest = 0; + + vhtcap->vht_cap_info = new_vhtcap; + + /* + * Now, if we're a STA, mask off whatever the AP doesn't support. + * Ie, we continue to state we can receive whatever we can do, + * but we only announce that we will transmit rates that meet + * the AP requirement. + * + * Note: 0 - MCS0..7; 1 - MCS0..8; 2 - MCS0..9; 3 = not supported. + * We can't just use MIN() because '3' means "no", so special case it. + */ + if (opmode) { + for (i = 0; i < 8; i++) { + val1 = (vhtcap->supp_mcs.tx_mcs_map >> (i*2)) & 0x3; + val2 = (ni->ni_vht_mcsinfo.tx_mcs_map >> (i*2)) & 0x3; + val = MIN(val1, val2); + if (val1 == 3 || val2 == 3) + val = 3; + vhtcap->supp_mcs.tx_mcs_map &= ~(0x3 << (i*2)); + vhtcap->supp_mcs.tx_mcs_map |= (val << (i*2)); + } + } +} + +/* + * Add a VHTCAP field. + * + * If in station mode, we announce what we would like our + * desired configuration to be. + * + * Else, we announce our capabilities based on our current + * configuration. + */ uint8_t * ieee80211_add_vhtcap(uint8_t *frm, struct ieee80211_node *ni) { - uint32_t cap; + struct ieee80211_ie_vhtcap vhtcap; + int opmode; + + opmode = 0; + if (ni->ni_vap->iv_opmode == IEEE80211_M_STA) + opmode = 1; + + ieee80211_vht_get_vhtcap_ie(ni, &vhtcap, opmode); memset(frm, '\0', sizeof(struct ieee80211_ie_vhtcap)); @@ -291,27 +632,14 @@ ieee80211_add_vhtcap(uint8_t *frm, struct ieee80211_node *ni) frm[1] = sizeof(struct ieee80211_ie_vhtcap) - 2; frm += 2; - /* - * For now, don't do any configuration. - * Just populate the node configuration. - * We can worry about making it configurable later. - */ - - cap = ni->ni_vhtcap; - - /* - * XXX TODO: any capability changes required by - * configuration. - */ - /* 32-bit VHT capability */ - ADDWORD(frm, cap); + ADDWORD(frm, vhtcap.vht_cap_info); /* suppmcs */ - ADDSHORT(frm, ni->ni_vht_mcsinfo.rx_mcs_map); - ADDSHORT(frm, ni->ni_vht_mcsinfo.rx_highest); - ADDSHORT(frm, ni->ni_vht_mcsinfo.tx_mcs_map); - ADDSHORT(frm, ni->ni_vht_mcsinfo.tx_highest); + ADDSHORT(frm, vhtcap.supp_mcs.rx_mcs_map); + ADDSHORT(frm, vhtcap.supp_mcs.rx_highest); + ADDSHORT(frm, vhtcap.supp_mcs.tx_mcs_map); + ADDSHORT(frm, vhtcap.supp_mcs.tx_highest); return (frm); } @@ -370,17 +698,6 @@ ieee80211_add_vhtinfo(uint8_t *frm, struct ieee80211_node *ni) frm[1] = sizeof(struct ieee80211_ie_vht_operation) - 2; frm += 2; - /* - * XXX if it's a station, then see if we have a node - * channel or ANYC. If it's ANYC then assume we're - * scanning, and announce our capabilities. - * - * This should set the "20/40/80/160MHz wide config"; - * the 80/80 or 160MHz wide config is done in VHTCAP. - * - * Other modes - just limit it to the channel. - */ - /* 8-bit chanwidth */ *frm++ = ieee80211_vht_get_chwidth_ie(ni->ni_chan); @@ -475,3 +792,19 @@ ieee80211_vht_adjust_channel(struct ieee80211com *ic, #endif return (chan); } + +/* + * Calculate the VHT operation IE for a given node. + * + * This includes calculating the suitable channel width/parameters + * and basic MCS set. + * + * TODO: ensure I read 9.7.11 Rate Selection for VHT STAs. + * TODO: ensure I read 10.39.7 - BSS Basic VHT-MCS and NSS set operation. + */ +void +ieee80211_vht_get_vhtinfo_ie(struct ieee80211_node *ni, + struct ieee80211_ie_vht_operation *vhtop, int opmode) +{ + printf("%s: called; TODO!\n", __func__); +} diff --git a/sys/net80211/ieee80211_vht.h b/sys/net80211/ieee80211_vht.h index 9e68e7d258d8..791762b14a31 100644 --- a/sys/net80211/ieee80211_vht.h +++ b/sys/net80211/ieee80211_vht.h @@ -60,4 +60,9 @@ struct ieee80211_channel * ieee80211_vht_adjust_channel(struct ieee80211com *, struct ieee80211_channel *, int); +void ieee80211_vht_get_vhtcap_ie(struct ieee80211_node *ni, + struct ieee80211_ie_vhtcap *, int); +void ieee80211_vht_get_vhtinfo_ie(struct ieee80211_node *ni, + struct ieee80211_ie_vht_operation *, int); + #endif /* _NET80211_IEEE80211_VHT_H_ */ From 8fde59a7da76417a9779ba29632393d36c5fdf65 Mon Sep 17 00:00:00 2001 From: Adrian Chadd Date: Tue, 10 Jan 2017 05:32:02 +0000 Subject: [PATCH 24/79] [net80211] add VHT EDCA parameters for WME/QoS mode. --- sys/net80211/ieee80211_proto.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+) diff --git a/sys/net80211/ieee80211_proto.c b/sys/net80211/ieee80211_proto.c index 36070b7f1947..275401fa5747 100644 --- a/sys/net80211/ieee80211_proto.c +++ b/sys/net80211/ieee80211_proto.c @@ -842,6 +842,9 @@ setbasicrates(struct ieee80211_rateset *rs, [IEEE80211_MODE_11NA] = { 3, { 12, 24, 48 } }, /* NB: mixed b/g */ [IEEE80211_MODE_11NG] = { 4, { 2, 4, 11, 22 } }, + /* NB: mixed b/g */ + [IEEE80211_MODE_VHT_2GHZ] = { 4, { 2, 4, 11, 22 } }, + [IEEE80211_MODE_VHT_5GHZ] = { 3, { 12, 24, 48 } }, }; int i, j; @@ -906,6 +909,8 @@ static const struct phyParamType phyParamForAC_BE[IEEE80211_MODE_MAX] = { [IEEE80211_MODE_QUARTER]= { 3, 4, 6, 0, 0 }, [IEEE80211_MODE_11NA] = { 3, 4, 6, 0, 0 }, [IEEE80211_MODE_11NG] = { 3, 4, 6, 0, 0 }, + [IEEE80211_MODE_VHT_2GHZ] = { 3, 4, 6, 0, 0 }, + [IEEE80211_MODE_VHT_5GHZ] = { 3, 4, 6, 0, 0 }, }; static const struct phyParamType phyParamForAC_BK[IEEE80211_MODE_MAX] = { [IEEE80211_MODE_AUTO] = { 7, 4, 10, 0, 0 }, @@ -920,6 +925,8 @@ static const struct phyParamType phyParamForAC_BK[IEEE80211_MODE_MAX] = { [IEEE80211_MODE_QUARTER]= { 7, 4, 10, 0, 0 }, [IEEE80211_MODE_11NA] = { 7, 4, 10, 0, 0 }, [IEEE80211_MODE_11NG] = { 7, 4, 10, 0, 0 }, + [IEEE80211_MODE_VHT_2GHZ] = { 7, 4, 10, 0, 0 }, + [IEEE80211_MODE_VHT_5GHZ] = { 7, 4, 10, 0, 0 }, }; static const struct phyParamType phyParamForAC_VI[IEEE80211_MODE_MAX] = { [IEEE80211_MODE_AUTO] = { 1, 3, 4, 94, 0 }, @@ -934,6 +941,8 @@ static const struct phyParamType phyParamForAC_VI[IEEE80211_MODE_MAX] = { [IEEE80211_MODE_QUARTER]= { 1, 3, 4, 94, 0 }, [IEEE80211_MODE_11NA] = { 1, 3, 4, 94, 0 }, [IEEE80211_MODE_11NG] = { 1, 3, 4, 94, 0 }, + [IEEE80211_MODE_VHT_2GHZ] = { 1, 3, 4, 94, 0 }, + [IEEE80211_MODE_VHT_5GHZ] = { 1, 3, 4, 94, 0 }, }; static const struct phyParamType phyParamForAC_VO[IEEE80211_MODE_MAX] = { [IEEE80211_MODE_AUTO] = { 1, 2, 3, 47, 0 }, @@ -948,6 +957,8 @@ static const struct phyParamType phyParamForAC_VO[IEEE80211_MODE_MAX] = { [IEEE80211_MODE_QUARTER]= { 1, 2, 3, 47, 0 }, [IEEE80211_MODE_11NA] = { 1, 2, 3, 47, 0 }, [IEEE80211_MODE_11NG] = { 1, 2, 3, 47, 0 }, + [IEEE80211_MODE_VHT_2GHZ] = { 1, 2, 3, 47, 0 }, + [IEEE80211_MODE_VHT_5GHZ] = { 1, 2, 3, 47, 0 }, }; static const struct phyParamType bssPhyParamForAC_BE[IEEE80211_MODE_MAX] = { @@ -1123,6 +1134,8 @@ ieee80211_wme_updateparams_locked(struct ieee80211vap *vap) [IEEE80211_MODE_QUARTER] = { 2, 4, 10, 64, 0 }, [IEEE80211_MODE_11NA] = { 2, 4, 10, 64, 0 }, /* XXXcheck*/ [IEEE80211_MODE_11NG] = { 2, 4, 10, 64, 0 }, /* XXXcheck*/ + [IEEE80211_MODE_VHT_2GHZ] = { 2, 4, 10, 64, 0 }, /* XXXcheck*/ + [IEEE80211_MODE_VHT_5GHZ] = { 2, 4, 10, 64, 0 }, /* XXXcheck*/ }; struct ieee80211com *ic = vap->iv_ic; struct ieee80211_wme_state *wme = &ic->ic_wme; @@ -1243,6 +1256,8 @@ ieee80211_wme_updateparams_locked(struct ieee80211vap *vap) [IEEE80211_MODE_QUARTER] = 3, [IEEE80211_MODE_11NA] = 3, [IEEE80211_MODE_11NG] = 3, + [IEEE80211_MODE_VHT_2GHZ] = 3, + [IEEE80211_MODE_VHT_5GHZ] = 3, }; chanp = &wme->wme_chanParams.cap_wmeParams[WME_AC_BE]; bssp = &wme->wme_bssChanParams.cap_wmeParams[WME_AC_BE]; From 94338935eedf58299427d38299758f1ebe724f37 Mon Sep 17 00:00:00 2001 From: Adrian Chadd Date: Tue, 10 Jan 2017 05:32:30 +0000 Subject: [PATCH 25/79] [net80211] add CHAN_VHT2G/CHAN_VHT5G macros. --- sys/net80211/_ieee80211.h | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/sys/net80211/_ieee80211.h b/sys/net80211/_ieee80211.h index 13155ea399a3..12f3ad8c84d2 100644 --- a/sys/net80211/_ieee80211.h +++ b/sys/net80211/_ieee80211.h @@ -313,6 +313,12 @@ struct ieee80211_channel { #define IEEE80211_IS_CHAN_VHT(_c) \ (((_c)->ic_flags & IEEE80211_CHAN_VHT) != 0) +#define IEEE80211_IS_CHAN_VHT_2GHZ(_c) \ + (IEEE80211_IS_CHAN_2GHZ(_c) && \ + ((_c)->ic_flags & IEEE80211_CHAN_VHT) != 0) +#define IEEE80211_IS_CHAN_VHT_5GHZ(_c) \ + (IEEE80211_IS_CHAN_5GHZ(_c) && \ + ((_c)->ic_flags & IEEE80211_CHAN_VHT) != 0) #define IEEE80211_IS_CHAN_VHT20(_c) \ (((_c)->ic_flags & IEEE80211_CHAN_VHT20) != 0) #define IEEE80211_IS_CHAN_VHT40(_c) \ From 5fd74bfae8b88679aab8ae7b8aeee4af4e9346c2 Mon Sep 17 00:00:00 2001 From: Adrian Chadd Date: Tue, 10 Jan 2017 05:33:34 +0000 Subject: [PATCH 26/79] [net80211] add missing VHTCAP declaration changes. These are required for the recent ieee80211_vht.[ch] changes - they make things start to work with MS() / SM() macros. --- sys/net80211/ieee80211.h | 40 ++++++++++++++++++++++++++++++++++++++-- 1 file changed, 38 insertions(+), 2 deletions(-) diff --git a/sys/net80211/ieee80211.h b/sys/net80211/ieee80211.h index aa2ddb0968ec..93740eec36b3 100644 --- a/sys/net80211/ieee80211.h +++ b/sys/net80211/ieee80211.h @@ -798,37 +798,73 @@ struct ieee80211_ie_vht_operation { #define IEEE80211_VHTCAP_MAX_MPDU_LENGTH_7991 0x00000001 #define IEEE80211_VHTCAP_MAX_MPDU_LENGTH_11454 0x00000002 #define IEEE80211_VHTCAP_MAX_MPDU_MASK 0x00000003 -#define IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_160MHZ 0x00000004 -#define IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_160_80PLUS80MHZ 0x00000008 +#define IEEE80211_VHTCAP_MAX_MPDU_MASK_S 0 + #define IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_MASK 0x0000000C +#define IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_MASK_S 2 +#define IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_NONE 0 +#define IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_160MHZ 1 +#define IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_160_80P80MHZ 2 +#define IEEE80211_VHTCAP_SUPP_CHAN_WIDTH_RESERVED 3 + #define IEEE80211_VHTCAP_RXLDPC 0x00000010 +#define IEEE80211_VHTCAP_RXLDPC_S 4 + #define IEEE80211_VHTCAP_SHORT_GI_80 0x00000020 +#define IEEE80211_VHTCAP_SHORT_GI_80_S 5 + #define IEEE80211_VHTCAP_SHORT_GI_160 0x00000040 +#define IEEE80211_VHTCAP_SHORT_GI_160_S 6 + #define IEEE80211_VHTCAP_TXSTBC 0x00000080 +#define IEEE80211_VHTCAP_TXSTBC_S 7 + #define IEEE80211_VHTCAP_RXSTBC_1 0x00000100 #define IEEE80211_VHTCAP_RXSTBC_2 0x00000200 #define IEEE80211_VHTCAP_RXSTBC_3 0x00000300 #define IEEE80211_VHTCAP_RXSTBC_4 0x00000400 #define IEEE80211_VHTCAP_RXSTBC_MASK 0x00000700 +#define IEEE80211_VHTCAP_RXSTBC_MASK_S 8 + #define IEEE80211_VHTCAP_SU_BEAMFORMER_CAPABLE 0x00000800 +#define IEEE80211_VHTCAP_SU_BEAMFORMER_CAPABLE_S 11 + #define IEEE80211_VHTCAP_SU_BEAMFORMEE_CAPABLE 0x00001000 +#define IEEE80211_VHTCAP_SU_BEAMFORMEE_CAPABLE_S 12 + #define IEEE80211_VHTCAP_BEAMFORMEE_STS_SHIFT 13 #define IEEE80211_VHTCAP_BEAMFORMEE_STS_MASK \ (7 << IEEE80211_VHTCAP_BEAMFORMEE_STS_SHIFT) +#define IEEE80211_VHTCAP_BEAMFORMEE_STS_MASK_S 13 + #define IEEE80211_VHTCAP_SOUNDING_DIMENSIONS_SHIFT 16 #define IEEE80211_VHTCAP_SOUNDING_DIMENSIONS_MASK \ (7 << IEEE80211_VHTCAP_SOUNDING_DIMENSIONS_SHIFT) +#define IEEE80211_VHTCAP_SOUNDING_DIMENSIONS_MASK_S 16 + #define IEEE80211_VHTCAP_MU_BEAMFORMER_CAPABLE 0x00080000 +#define IEEE80211_VHTCAP_MU_BEAMFORMER_CAPABLE_S 19 #define IEEE80211_VHTCAP_MU_BEAMFORMEE_CAPABLE 0x00100000 +#define IEEE80211_VHTCAP_MU_BEAMFORMEE_CAPABLE_S 20 #define IEEE80211_VHTCAP_VHT_TXOP_PS 0x00200000 +#define IEEE80211_VHTCAP_VHT_TXOP_PS_S 21 #define IEEE80211_VHTCAP_HTC_VHT 0x00400000 +#define IEEE80211_VHTCAP_HTC_VHT_S 22 + #define IEEE80211_VHTCAP_MAX_A_MPDU_LENGTH_EXPONENT_SHIFT 23 #define IEEE80211_VHTCAP_MAX_A_MPDU_LENGTH_EXPONENT_MASK \ (7 << IEEE80211_VHTCAP_MAX_A_MPDU_LENGTH_EXPONENT_SHIFT) +#define IEEE80211_VHTCAP_MAX_A_MPDU_LENGTH_EXPONENT_MASK_S 23 + +#define IEEE80211_VHTCAP_VHT_LINK_ADAPTATION_VHT_MASK 0x0c000000 #define IEEE80211_VHTCAP_VHT_LINK_ADAPTATION_VHT_UNSOL_MFB 0x08000000 #define IEEE80211_VHTCAP_VHT_LINK_ADAPTATION_VHT_MRQ_MFB 0x0c000000 +#define IEEE80211_VHTCAP_VHT_LINK_ADAPTATION_VHT_MASK_S 26 + #define IEEE80211_VHTCAP_RX_ANTENNA_PATTERN 0x10000000 +#define IEEE80211_VHTCAP_RX_ANTENNA_PATTERN_S 28 #define IEEE80211_VHTCAP_TX_ANTENNA_PATTERN 0x20000000 +#define IEEE80211_VHTCAP_TX_ANTENNA_PATTERN_S 29 /* * XXX TODO: add the rest of the bits From 33740d0754d419c7d147b0b83d31979450076b50 Mon Sep 17 00:00:00 2001 From: Larry Rosenman Date: Tue, 10 Jan 2017 05:37:53 +0000 Subject: [PATCH 27/79] Add my birthday to calendar.freebsd Approved by: adamw (Mentor) Differential Revision: https://reviews.freebsd.org/D9119 --- usr.bin/calendar/calendars/calendar.freebsd | 1 + 1 file changed, 1 insertion(+) diff --git a/usr.bin/calendar/calendars/calendar.freebsd b/usr.bin/calendar/calendars/calendar.freebsd index 841896eb1d81..3a36f6eccee9 100644 --- a/usr.bin/calendar/calendars/calendar.freebsd +++ b/usr.bin/calendar/calendars/calendar.freebsd @@ -310,6 +310,7 @@ 09/20 Kevin Lo born in Taipei, Taiwan, Republic of China, 1972 09/21 Gleb Kurtsou born in Minsk, Belarus, 1984 09/22 Bryan Drewery born in San Diego, California, United States, 1984 +09/24 Larry Rosenman born in Queens, New York, United States, 1957 09/27 Neil Blakey-Milner born in Port Elizabeth, South Africa, 1978 09/27 Renato Botelho born in Araras, Sao Paulo, Brazil, 1979 09/28 Greg Lehey born in Melbourne, Victoria, Australia, 1948 From 86fee26330807e433ca2e8a94fa6657c5d44e09e Mon Sep 17 00:00:00 2001 From: Adrian Chadd Date: Tue, 10 Jan 2017 07:21:07 +0000 Subject: [PATCH 28/79] [net80211] add VHT action frame placeholders for when it's time to implement. --- sys/net80211/ieee80211_vht.c | 40 ++++++++++++++++++++++++++++++++++++ 1 file changed, 40 insertions(+) diff --git a/sys/net80211/ieee80211_vht.c b/sys/net80211/ieee80211_vht.c index 15012f254e70..deb3811026af 100644 --- a/sys/net80211/ieee80211_vht.c +++ b/sys/net80211/ieee80211_vht.c @@ -85,9 +85,49 @@ __FBSDID("$FreeBSD$"); * Look at mac80211/vht.c:ieee80211_vht_handle_opmode() for further details. */ +static int +vht_recv_action_placeholder(struct ieee80211_node *ni, + const struct ieee80211_frame *wh, + const uint8_t *frm, const uint8_t *efrm) +{ + + ieee80211_note(ni->ni_vap, "%s: called; fc=0x%.2x/0x%.2x", + __func__, + wh->i_fc[0], + wh->i_fc[1]); + + return (0); +} + +static int +vht_send_action_placeholder(struct ieee80211_node *ni, + int category, int action, void *arg0) +{ + + ieee80211_note(ni->ni_vap, "%s: called; category=%d, action=%d", + __func__, + category, + action); + return (EINVAL); +} + static void ieee80211_vht_init(void) { + + ieee80211_recv_action_register(IEEE80211_ACTION_CAT_VHT, + WLAN_ACTION_VHT_COMPRESSED_BF, vht_recv_action_placeholder); + ieee80211_recv_action_register(IEEE80211_ACTION_CAT_VHT, + WLAN_ACTION_VHT_GROUPID_MGMT, vht_recv_action_placeholder); + ieee80211_recv_action_register(IEEE80211_ACTION_CAT_VHT, + WLAN_ACTION_VHT_OPMODE_NOTIF, vht_recv_action_placeholder); + + ieee80211_send_action_register(IEEE80211_ACTION_CAT_VHT, + WLAN_ACTION_VHT_COMPRESSED_BF, vht_send_action_placeholder); + ieee80211_send_action_register(IEEE80211_ACTION_CAT_VHT, + WLAN_ACTION_VHT_GROUPID_MGMT, vht_send_action_placeholder); + ieee80211_send_action_register(IEEE80211_ACTION_CAT_VHT, + WLAN_ACTION_VHT_OPMODE_NOTIF, vht_send_action_placeholder); } SYSINIT(wlan_vht, SI_SUB_DRIVERS, SI_ORDER_FIRST, ieee80211_vht_init, NULL); From 930dc01620ed6c2e6b88d6503c888a13f217be2f Mon Sep 17 00:00:00 2001 From: Adrian Chadd Date: Tue, 10 Jan 2017 07:24:29 +0000 Subject: [PATCH 29/79] [net80211] Add default parameters for 11ac. I doubt TDMA code will ever work for 11ac, but you never know, someone may one day make it happen. --- sys/net80211/ieee80211_tdma.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/sys/net80211/ieee80211_tdma.c b/sys/net80211/ieee80211_tdma.c index aac8c7fd55c1..93eba3b2efe8 100644 --- a/sys/net80211/ieee80211_tdma.c +++ b/sys/net80211/ieee80211_tdma.c @@ -176,6 +176,8 @@ ieee80211_tdma_vattach(struct ieee80211vap *vap) settxparms(vap, IEEE80211_MODE_11NG, TDMA_TXRATE_11NG_DEFAULT); settxparms(vap, IEEE80211_MODE_HALF, TDMA_TXRATE_HALF_DEFAULT); settxparms(vap, IEEE80211_MODE_QUARTER, TDMA_TXRATE_QUARTER_DEFAULT); + settxparms(vap, IEEE80211_MODE_VHT_2GHZ, TDMA_TXRATE_11NG_DEFAULT); + settxparms(vap, IEEE80211_MODE_VHT_5GHZ, TDMA_TXRATE_11NA_DEFAULT); setackpolicy(vap->iv_ic, 1); /* disable ACK's */ From 0c67d389f412a1f982bf9015b8d36dd0d560e301 Mon Sep 17 00:00:00 2001 From: Adrian Chadd Date: Tue, 10 Jan 2017 07:50:21 +0000 Subject: [PATCH 30/79] [net80211] add VHT mediatype initialisation and update helper functions. --- sys/net80211/ieee80211.c | 29 ++++++++++++++++++++++++++++- 1 file changed, 28 insertions(+), 1 deletion(-) diff --git a/sys/net80211/ieee80211.c b/sys/net80211/ieee80211.c index e4990868f06c..487e1505ef08 100644 --- a/sys/net80211/ieee80211.c +++ b/sys/net80211/ieee80211.c @@ -70,6 +70,8 @@ const char *ieee80211_phymode_name[IEEE80211_MODE_MAX] = { [IEEE80211_MODE_QUARTER] = "quarter", [IEEE80211_MODE_11NA] = "11na", [IEEE80211_MODE_11NG] = "11ng", + [IEEE80211_MODE_VHT_2GHZ] = "11acg", + [IEEE80211_MODE_VHT_5GHZ] = "11ac", }; /* map ieee80211_opmode to the corresponding capability bit */ const int ieee80211_opcap[IEEE80211_OPMODE_MAX] = { @@ -181,6 +183,10 @@ ieee80211_chan_init(struct ieee80211com *ic) setbit(ic->ic_modecaps, IEEE80211_MODE_11NA); if (IEEE80211_IS_CHAN_HTG(c)) setbit(ic->ic_modecaps, IEEE80211_MODE_11NG); + if (IEEE80211_IS_CHAN_VHTA(c)) + setbit(ic->ic_modecaps, IEEE80211_MODE_VHT_5GHZ); + if (IEEE80211_IS_CHAN_VHTG(c)) + setbit(ic->ic_modecaps, IEEE80211_MODE_VHT_2GHZ); } /* initialize candidate channels to all available */ memcpy(ic->ic_chan_active, ic->ic_chan_avail, @@ -208,6 +214,8 @@ ieee80211_chan_init(struct ieee80211com *ic) DEFAULTRATES(IEEE80211_MODE_QUARTER, ieee80211_rateset_quarter); DEFAULTRATES(IEEE80211_MODE_11NA, ieee80211_rateset_11a); DEFAULTRATES(IEEE80211_MODE_11NG, ieee80211_rateset_11g); + DEFAULTRATES(IEEE80211_MODE_VHT_2GHZ, ieee80211_rateset_11g); + DEFAULTRATES(IEEE80211_MODE_VHT_5GHZ, ieee80211_rateset_11a); /* * Setup required information to fill the mcsset field, if driver did @@ -1492,6 +1500,8 @@ addmedia(struct ifmedia *media, int caps, int addsta, int mode, int mword) [IEEE80211_MODE_QUARTER] = IFM_IEEE80211_11A, /* XXX */ [IEEE80211_MODE_11NA] = IFM_IEEE80211_11NA, [IEEE80211_MODE_11NG] = IFM_IEEE80211_11NG, + [IEEE80211_MODE_VHT_2GHZ] = IFM_IEEE80211_VHT2G, + [IEEE80211_MODE_VHT_5GHZ] = IFM_IEEE80211_VHT5G, }; u_int mopt; @@ -1604,6 +1614,19 @@ ieee80211_media_setup(struct ieee80211com *ic, if (rate > maxrate) maxrate = rate; } + + /* + * Add VHT media. + */ + for (; mode <= IEEE80211_MODE_VHT_5GHZ; mode++) { + if (isclr(ic->ic_modecaps, mode)) + continue; + addmedia(media, caps, addsta, mode, IFM_AUTO); + addmedia(media, caps, addsta, mode, IFM_IEEE80211_VHT); + + /* XXX TODO: VHT maxrate */ + } + return maxrate; } @@ -1883,7 +1906,11 @@ enum ieee80211_phymode ieee80211_chan2mode(const struct ieee80211_channel *chan) { - if (IEEE80211_IS_CHAN_HTA(chan)) + if (IEEE80211_IS_CHAN_VHT_2GHZ(chan)) + return IEEE80211_MODE_VHT_2GHZ; + else if (IEEE80211_IS_CHAN_VHT_5GHZ(chan)) + return IEEE80211_MODE_VHT_5GHZ; + else if (IEEE80211_IS_CHAN_HTA(chan)) return IEEE80211_MODE_11NA; else if (IEEE80211_IS_CHAN_HTG(chan)) return IEEE80211_MODE_11NG; From 887eb0f8da295d19a096df458e0fabd0e1fc7fd9 Mon Sep 17 00:00:00 2001 From: Enji Cooper Date: Tue, 10 Jan 2017 10:02:44 +0000 Subject: [PATCH 31/79] t_raise: import a minor grammar error fix to diff reduce with NetBSD --- lib/libc/gen/t_raise.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/lib/libc/gen/t_raise.c b/lib/libc/gen/t_raise.c index adc032fdd98a..4258cc16444e 100644 --- a/lib/libc/gen/t_raise.c +++ b/lib/libc/gen/t_raise.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_raise.c,v 1.5 2011/05/10 12:43:42 jruoho Exp $ */ +/* $NetBSD: t_raise.c,v 1.6 2016/11/03 22:08:31 kamil Exp $ */ /*- * Copyright (c) 2011 The NetBSD Foundation, Inc. @@ -29,7 +29,7 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_raise.c,v 1.5 2011/05/10 12:43:42 jruoho Exp $"); +__RCSID("$NetBSD: t_raise.c,v 1.6 2016/11/03 22:08:31 kamil Exp $"); #include @@ -176,7 +176,7 @@ ATF_TC_BODY(raise_stress, tc) (void)raise(SIGUSR1); if (count != maxiter) - atf_tc_fail("not all signals were catched"); + atf_tc_fail("not all signals were caught"); } ATF_TP_ADD_TCS(tp) From 4023a42882d5f9419fef21524cb32cb410a18ad3 Mon Sep 17 00:00:00 2001 From: Enji Cooper Date: Tue, 10 Jan 2017 10:06:15 +0000 Subject: [PATCH 32/79] Import lib/libc/ttyio/t_ttyio.c,1.3 --- lib/libc/ttyio/t_ttyio.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/lib/libc/ttyio/t_ttyio.c b/lib/libc/ttyio/t_ttyio.c index 5a5ec0f988ba..7f60d786a75f 100644 --- a/lib/libc/ttyio/t_ttyio.c +++ b/lib/libc/ttyio/t_ttyio.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_ttyio.c,v 1.2 2011/04/19 20:07:53 martin Exp $ */ +/* $NetBSD: t_ttyio.c,v 1.3 2017/01/10 01:31:40 christos Exp $ */ /* * Copyright (c) 2001, 2008 The NetBSD Foundation, Inc. @@ -32,7 +32,7 @@ #include __COPYRIGHT("@(#) Copyright (c) 2008\ The NetBSD Foundation, inc. All rights reserved."); -__RCSID("$NetBSD: t_ttyio.c,v 1.2 2011/04/19 20:07:53 martin Exp $"); +__RCSID("$NetBSD: t_ttyio.c,v 1.3 2017/01/10 01:31:40 christos Exp $"); #include #include @@ -150,8 +150,9 @@ ATF_TC_BODY(ioctl, tc) /* wait for last child */ sa.sa_handler = SIG_DFL; REQUIRE_ERRNO(sigaction(SIGCHLD, &sa, NULL), -1); - (void) wait(NULL); + (void)wait(NULL); + (void)close(s); ATF_REQUIRE_EQ(rc, 0); } From 7931520d400a5b3b4df7584d4ebfd28a8767f9d8 Mon Sep 17 00:00:00 2001 From: Enji Cooper Date: Tue, 10 Jan 2017 10:27:09 +0000 Subject: [PATCH 33/79] Diff reduce with upstream by removing signal.h #include MFC after: 3 days --- contrib/netbsd-tests/lib/libc/stdio/t_printf.c | 4 ---- 1 file changed, 4 deletions(-) diff --git a/contrib/netbsd-tests/lib/libc/stdio/t_printf.c b/contrib/netbsd-tests/lib/libc/stdio/t_printf.c index d6ca662273fc..95b4b2c9dffe 100644 --- a/contrib/netbsd-tests/lib/libc/stdio/t_printf.c +++ b/contrib/netbsd-tests/lib/libc/stdio/t_printf.c @@ -36,10 +36,6 @@ #include #include -#ifndef __NetBSD__ -#include -#endif - ATF_TC(snprintf_c99); ATF_TC_HEAD(snprintf_c99, tc) { From 7c07ea9aa23363a9aa160505cc7fa33f3b2739d7 Mon Sep 17 00:00:00 2001 From: Alexander Motin Date: Tue, 10 Jan 2017 10:33:36 +0000 Subject: [PATCH 34/79] Fix malloc(M_WAITOK) under mutex, introduced at r311787. MFC after: 13 days --- sys/cam/ctl/ctl.c | 38 +++++++++++++++++--------------------- 1 file changed, 17 insertions(+), 21 deletions(-) diff --git a/sys/cam/ctl/ctl.c b/sys/cam/ctl/ctl.c index 0c31ef9e8fff..2efad7c7b7ad 100644 --- a/sys/cam/ctl/ctl.c +++ b/sys/cam/ctl/ctl.c @@ -4593,6 +4593,8 @@ ctl_alloc_lun(struct ctl_softc *ctl_softc, struct ctl_lun *ctl_lun, printf("ctl: requested LUN ID %d is already " "in use\n", be_lun->req_lun_id); } +fail: + free(lun->lun_devid, M_CTL); if (lun->flags & CTL_LUN_MALLOCED) free(lun, M_CTL); be_lun->lun_config_status(be_lun->be_lun, @@ -4605,14 +4607,11 @@ ctl_alloc_lun(struct ctl_softc *ctl_softc, struct ctl_lun *ctl_lun, if (lun_number == -1) { mtx_unlock(&ctl_softc->ctl_lock); printf("ctl: can't allocate LUN, out of LUNs\n"); - if (lun->flags & CTL_LUN_MALLOCED) - free(lun, M_CTL); - be_lun->lun_config_status(be_lun->be_lun, - CTL_LUN_CONFIG_FAILURE); - return (ENOSPC); + goto fail; } } ctl_set_mask(ctl_softc->ctl_lun_mask, lun_number); + mtx_unlock(&ctl_softc->ctl_lock); mtx_init(&lun->lun_lock, "CTL LUN", NULL, MTX_DEF); lun->lun = lun_number; @@ -4664,22 +4663,6 @@ ctl_alloc_lun(struct ctl_softc *ctl_softc, struct ctl_lun *ctl_lun, ctl_init_page_index(lun); ctl_init_log_page_index(lun); - /* - * Now, before we insert this lun on the lun list, set the lun - * inventory changed UA for all other luns. - */ - STAILQ_FOREACH(nlun, &ctl_softc->lun_list, links) { - mtx_lock(&nlun->lun_lock); - ctl_est_ua_all(nlun, -1, CTL_UA_LUN_CHANGE); - mtx_unlock(&nlun->lun_lock); - } - - STAILQ_INSERT_TAIL(&ctl_softc->lun_list, lun, links); - - ctl_softc->ctl_luns[lun_number] = lun; - - ctl_softc->num_luns++; - /* Setup statistics gathering */ #ifdef CTL_LEGACY_STATS lun->legacy_stats.device_type = be_lun->lun_type; @@ -4692,6 +4675,19 @@ ctl_alloc_lun(struct ctl_softc *ctl_softc, struct ctl_lun *ctl_lun, #endif /* CTL_LEGACY_STATS */ lun->stats.item = lun_number; + /* + * Now, before we insert this lun on the lun list, set the lun + * inventory changed UA for all other luns. + */ + mtx_lock(&ctl_softc->ctl_lock); + STAILQ_FOREACH(nlun, &ctl_softc->lun_list, links) { + mtx_lock(&nlun->lun_lock); + ctl_est_ua_all(nlun, -1, CTL_UA_LUN_CHANGE); + mtx_unlock(&nlun->lun_lock); + } + STAILQ_INSERT_TAIL(&ctl_softc->lun_list, lun, links); + ctl_softc->ctl_luns[lun_number] = lun; + ctl_softc->num_luns++; mtx_unlock(&ctl_softc->ctl_lock); lun->be_lun->lun_config_status(lun->be_lun->be_lun, CTL_LUN_CONFIG_OK); From 2647410d059e4b1a042625eda57dfcf0c00fb0f0 Mon Sep 17 00:00:00 2001 From: Andrew Turner Date: Tue, 10 Jan 2017 10:56:33 +0000 Subject: [PATCH 35/79] Add an ACPI attachment to the existing ahci_generic driver. This is used in some arm64 hardware, for example the AMD Opteron A1100. Reviewed by: mav Obtained from: ABT Systems Ltd Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D8852 --- sys/conf/files.arm64 | 2 +- sys/dev/ahci/ahci_generic.c | 88 +++++++++++++++++++++++++++++++------ 2 files changed, 75 insertions(+), 15 deletions(-) diff --git a/sys/conf/files.arm64 b/sys/conf/files.arm64 index 4351bdcf31ec..2ff24eb1ef94 100644 --- a/sys/conf/files.arm64 +++ b/sys/conf/files.arm64 @@ -145,7 +145,7 @@ armv8_crypto_wrap.o optional armv8crypto \ crypto/blowfish/bf_enc.c optional crypto | ipsec crypto/des/des_enc.c optional crypto | ipsec | netsmb dev/acpica/acpi_if.m optional acpi -dev/ahci/ahci_generic.c optional ahci fdt +dev/ahci/ahci_generic.c optional ahci dev/cpufreq/cpufreq_dt.c optional cpufreq fdt dev/hwpmc/hwpmc_arm64.c optional hwpmc dev/hwpmc/hwpmc_arm64_md.c optional hwpmc diff --git a/sys/dev/ahci/ahci_generic.c b/sys/dev/ahci/ahci_generic.c index 80f7fedee049..a427e7aeaeea 100644 --- a/sys/dev/ahci/ahci_generic.c +++ b/sys/dev/ahci/ahci_generic.c @@ -27,6 +27,9 @@ #include __FBSDID("$FreeBSD$"); +#include "opt_acpi.h" +#include "opt_platform.h" + #include #include #include @@ -44,6 +47,15 @@ __FBSDID("$FreeBSD$"); #include +#ifdef DEV_ACPI +#include +#include + +#include +#include +#endif + +#ifdef FDT #include #include @@ -54,14 +66,7 @@ static struct ofw_compat_data compat_data[] = { }; static int -ahci_gen_ctlr_reset(device_t dev) -{ - - return ahci_ctlr_reset(dev); -} - -static int -ahci_probe(device_t dev) +ahci_fdt_probe(device_t dev) { if (!ofw_bus_status_okay(dev)) @@ -73,6 +78,34 @@ ahci_probe(device_t dev) device_set_desc_copy(dev, "AHCI SATA controller"); return (BUS_PROBE_DEFAULT); } +#endif + +#ifdef DEV_ACPI +static int +ahci_acpi_probe(device_t dev) +{ + ACPI_HANDLE h; + + if ((h = acpi_get_handle(dev)) == NULL) + return (ENXIO); + + if (pci_get_class(dev) == PCIC_STORAGE && + pci_get_subclass(dev) == PCIS_STORAGE_SATA && + pci_get_progif(dev) == PCIP_STORAGE_SATA_AHCI_1_0) { + device_set_desc_copy(dev, "AHCI SATA controller"); + return (BUS_PROBE_DEFAULT); + } + + return (ENXIO); +} +#endif + +static int +ahci_gen_ctlr_reset(device_t dev) +{ + + return ahci_ctlr_reset(dev); +} static int ahci_gen_attach(device_t dev) @@ -109,9 +142,10 @@ ahci_gen_detach(device_t dev) return (0); } -static devclass_t ahci_gen_devclass; -static device_method_t ahci_methods[] = { - DEVMETHOD(device_probe, ahci_probe), +#ifdef FDT +static devclass_t ahci_gen_fdt_devclass; +static device_method_t ahci_fdt_methods[] = { + DEVMETHOD(device_probe, ahci_fdt_probe), DEVMETHOD(device_attach, ahci_gen_attach), DEVMETHOD(device_detach, ahci_gen_detach), DEVMETHOD(bus_print_child, ahci_print_child), @@ -123,9 +157,35 @@ static device_method_t ahci_methods[] = { DEVMETHOD(bus_get_dma_tag, ahci_get_dma_tag), DEVMETHOD_END }; -static driver_t ahci_driver = { +static driver_t ahci_fdt_driver = { "ahci", - ahci_methods, + ahci_fdt_methods, sizeof(struct ahci_controller) }; -DRIVER_MODULE(ahci, simplebus, ahci_driver, ahci_gen_devclass, NULL, NULL); +DRIVER_MODULE(ahci_fdt, simplebus, ahci_fdt_driver, ahci_gen_fdt_devclass, + NULL, NULL); +#endif + +#ifdef DEV_ACPI +static devclass_t ahci_gen_acpi_devclass; +static device_method_t ahci_acpi_methods[] = { + DEVMETHOD(device_probe, ahci_acpi_probe), + DEVMETHOD(device_attach, ahci_gen_attach), + DEVMETHOD(device_detach, ahci_gen_detach), + DEVMETHOD(bus_print_child, ahci_print_child), + DEVMETHOD(bus_alloc_resource, ahci_alloc_resource), + DEVMETHOD(bus_release_resource, ahci_release_resource), + DEVMETHOD(bus_setup_intr, ahci_setup_intr), + DEVMETHOD(bus_teardown_intr,ahci_teardown_intr), + DEVMETHOD(bus_child_location_str, ahci_child_location_str), + DEVMETHOD(bus_get_dma_tag, ahci_get_dma_tag), + DEVMETHOD_END +}; +static driver_t ahci_acpi_driver = { + "ahci", + ahci_acpi_methods, + sizeof(struct ahci_controller) +}; +DRIVER_MODULE(ahci_acpi, acpi, ahci_acpi_driver, ahci_gen_acpi_devclass, + NULL, NULL); +#endif From ce790c9008056b4198d987269bdb52170dc86231 Mon Sep 17 00:00:00 2001 From: Andrew Turner Date: Tue, 10 Jan 2017 13:36:33 +0000 Subject: [PATCH 36/79] Add acpi_if.h to SRCS so we have it when building ahci_generic.c with ACPI. Obtained from: ABT Systems Ltd Sponsored by: The FreeBSD Foundation --- sys/modules/ahci/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sys/modules/ahci/Makefile b/sys/modules/ahci/Makefile index 5666870532a1..cc947d3b910a 100644 --- a/sys/modules/ahci/Makefile +++ b/sys/modules/ahci/Makefile @@ -6,7 +6,7 @@ KMOD= ahci SRCS= ahci.c ahci_pci.c ahciem.c ahci.h device_if.h bus_if.h pci_if.h opt_cam.h .if ${MACHINE_CPUARCH} == "aarch64" -SRCS+= ahci_generic.c ofw_bus_if.h +SRCS+= ahci_generic.c acpi_if.h ofw_bus_if.h .endif .include From 4a6768e64cc5229568f481cd3d7d1d539d6b375f Mon Sep 17 00:00:00 2001 From: "Bjoern A. Zeeb" Date: Tue, 10 Jan 2017 14:40:30 +0000 Subject: [PATCH 37/79] Teach crunchide about EM_S390 to make bootstrapping from future releases easier unless someone will fix the PR properly. MFC after: 3 days PR: 215940 --- usr.sbin/crunch/crunchide/exec_elf32.c | 1 + 1 file changed, 1 insertion(+) diff --git a/usr.sbin/crunch/crunchide/exec_elf32.c b/usr.sbin/crunch/crunchide/exec_elf32.c index 9aa29fd8a97b..d81775bfba71 100644 --- a/usr.sbin/crunch/crunchide/exec_elf32.c +++ b/usr.sbin/crunch/crunchide/exec_elf32.c @@ -191,6 +191,7 @@ ELFNAMEEND(check)(int fd, const char *fn) #define EM_RISCV 243 #endif case EM_RISCV: break; + case EM_S390: break; case EM_SPARCV9: break; case EM_X86_64: break; /* ELFDEFNNAME(MACHDEP_ID_CASES) */ From e3ef7bb21608dd77fae2ac9790b12f3ea9ebe541 Mon Sep 17 00:00:00 2001 From: Andrew Rybchenko Date: Tue, 10 Jan 2017 16:25:39 +0000 Subject: [PATCH 38/79] sfxge(4): avoid unnecessary mbuf data prefetch Unnecessary prefetch just loads HW prefetcher and displaces other cache entries (which could be really useful). If we parse mbuf for TSO early and use firmware-assisted TSO, we do not expect mbuf data access when we compose firmware-assisted TSO (v1 or v2) option descriptors. If packet header needs to be linearized or finally FATSO cannot be used because of, for example, too big header, we do not care about a bit more performance degradation because of prefetch absence (it is better to optimize more common case). Reviewed by: gnn Sponsored by: Solarflare Communications, Inc. MFC after: 2 days Differential Revision: https://reviews.freebsd.org/D9120 --- sys/dev/sfxge/sfxge_tx.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/sys/dev/sfxge/sfxge_tx.c b/sys/dev/sfxge/sfxge_tx.c index 4e31aeccbb2f..dd78e6e8a24f 100644 --- a/sys/dev/sfxge/sfxge_tx.c +++ b/sys/dev/sfxge/sfxge_tx.c @@ -363,8 +363,22 @@ static int sfxge_tx_queue_mbuf(struct sfxge_txq *txq, struct mbuf *mbuf) KASSERT(!txq->blocked, ("txq->blocked")); +#if SFXGE_TX_PARSE_EARLY + /* + * If software TSO is used, we still need to copy packet header, + * even if we have already parsed it early before enqueue. + */ + if ((mbuf->m_pkthdr.csum_flags & CSUM_TSO) && + (txq->tso_fw_assisted == 0)) + prefetch_read_many(mbuf->m_data); +#else + /* + * Prefetch packet header since we need to parse it and extract + * IP ID, TCP sequence number and flags. + */ if (mbuf->m_pkthdr.csum_flags & CSUM_TSO) prefetch_read_many(mbuf->m_data); +#endif if (__predict_false(txq->init_state != SFXGE_TXQ_STARTED)) { rc = EINTR; From e7bfd34bdf8fd0f38f652234801f6639692a7c64 Mon Sep 17 00:00:00 2001 From: Konstantin Belousov Date: Tue, 10 Jan 2017 17:05:34 +0000 Subject: [PATCH 39/79] Use ANSI C definitions, update comment. Sponsored by: The FreeBSD Foundation MFC after: 1 week --- libexec/rtld-elf/rtld_lock.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/libexec/rtld-elf/rtld_lock.c b/libexec/rtld-elf/rtld_lock.c index f31546c4ac91..eeaea5e578b0 100644 --- a/libexec/rtld-elf/rtld_lock.c +++ b/libexec/rtld-elf/rtld_lock.c @@ -38,8 +38,8 @@ * In this algorithm the lock is a single word. Its low-order bit is * set when a writer holds the lock. The remaining high-order bits * contain a count of readers desiring the lock. The algorithm requires - * atomic "compare_and_store" and "add" operations, which we implement - * using assembly language sequences in "rtld_start.S". + * atomic "compare_and_store" and "add" operations, which we take + * from machine/atomic.h. */ #include @@ -67,7 +67,7 @@ static sigset_t fullsigmask, oldsigmask; static int thread_flag; static void * -def_lock_create() +def_lock_create(void) { void *base; char *p; @@ -269,7 +269,7 @@ lock_restart_for_upgrade(RtldLockState *lockstate) } void -lockdflt_init() +lockdflt_init(void) { int i; From c70c4d71dec943ff4af7bd9b0ae924985b8d3a9a Mon Sep 17 00:00:00 2001 From: Navdeep Parhar Date: Tue, 10 Jan 2017 18:39:53 +0000 Subject: [PATCH 40/79] The iw_cxgb and iw_cxgbe drivers should not use a FreeBSD device_t where a linuxkpi style device is expected. If OFED/linuxkpi actually starts using this field then we'll have to figure out whether to create fake devices for these drivers or have linuxkpi deal with NULL device. This mismatch was first reported as part of D6585. --- sys/dev/cxgb/ulp/iw_cxgb/iw_cxgb_provider.c | 2 +- sys/dev/cxgbe/iw_cxgbe/provider.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/sys/dev/cxgb/ulp/iw_cxgb/iw_cxgb_provider.c b/sys/dev/cxgb/ulp/iw_cxgb/iw_cxgb_provider.c index 448b99348fb2..c2b508039a47 100644 --- a/sys/dev/cxgb/ulp/iw_cxgb/iw_cxgb_provider.c +++ b/sys/dev/cxgb/ulp/iw_cxgb/iw_cxgb_provider.c @@ -1094,7 +1094,7 @@ int iwch_register_device(struct iwch_dev *dev) memcpy(dev->ibdev.node_desc, IWCH_NODE_DESC, sizeof(IWCH_NODE_DESC)); dev->ibdev.phys_port_cnt = sc->params.nports; dev->ibdev.num_comp_vectors = 1; - dev->ibdev.dma_device = dev->rdev.adap->dev; + dev->ibdev.dma_device = NULL; dev->ibdev.query_device = iwch_query_device; dev->ibdev.query_port = iwch_query_port; dev->ibdev.modify_port = iwch_modify_port; diff --git a/sys/dev/cxgbe/iw_cxgbe/provider.c b/sys/dev/cxgbe/iw_cxgbe/provider.c index af9da14ce5b7..423c33b82bce 100644 --- a/sys/dev/cxgbe/iw_cxgbe/provider.c +++ b/sys/dev/cxgbe/iw_cxgbe/provider.c @@ -429,7 +429,7 @@ c4iw_register_device(struct c4iw_dev *dev) strlcpy(ibdev->node_desc, C4IW_NODE_DESC, sizeof(ibdev->node_desc)); ibdev->phys_port_cnt = sc->params.nports; ibdev->num_comp_vectors = 1; - ibdev->dma_device = sc->dev; + ibdev->dma_device = NULL; ibdev->query_device = c4iw_query_device; ibdev->query_port = c4iw_query_port; ibdev->modify_port = c4iw_modify_port; From 0a5023893550da8537932e02225f239a3ba3652b Mon Sep 17 00:00:00 2001 From: Li-Wen Hsu Date: Tue, 10 Jan 2017 18:46:40 +0000 Subject: [PATCH 41/79] Replace using of objdump with elfdump In-tree objdump is too old to dump new ELF headers. But for example if we use: `make CROSS_TOOLCHAIN=riscv64-gcc TARGET_ARCH=riscv64` and do not specify CROSS_BINUTILS_PREFIX in env, embed_mfs.sh cannot find the correct objdump. This patch just replaces using of objdump with elfdump to collect needed information. Later we may also put an ELFDUMP in CROSSENV and use it in embed_mfs.sh . Reviewed by: emaste, br MFC after: 3 days Differential Revision: https://reviews.freebsd.org/D9062 --- sys/tools/embed_mfs.sh | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/sys/tools/embed_mfs.sh b/sys/tools/embed_mfs.sh index 66c50c2da111..a7ac80c8f8d3 100644 --- a/sys/tools/embed_mfs.sh +++ b/sys/tools/embed_mfs.sh @@ -36,12 +36,12 @@ mfs_size=`stat -f '%z' $2 2> /dev/null` # If we can't determine MFS image size - bail. [ -z ${mfs_size} ] && echo "Can't determine MFS image size" && exit 1 -sec_info=`${CROSS_BINUTILS_PREFIX}objdump -h $1 2> /dev/null | grep " oldmfs "` +sec_info=`elfdump -c $1 2> /dev/null | grep -A 5 -E "sh_name: oldmfs$"` # If we can't find the mfs section within the given kernel - bail. [ -z "${sec_info}" ] && echo "Can't locate mfs section within kernel" && exit 1 -sec_size=`echo ${sec_info} | awk '{printf("%d", "0x" $3)}' 2> /dev/null` -sec_start=`echo ${sec_info} | awk '{printf("%d", "0x" $6)}' 2> /dev/null` +sec_size=`echo "${sec_info}" | awk '/sh_size/ {print $2}' 2> /dev/null` +sec_start=`echo "${sec_info}" | awk '/sh_offset/ {print $2}' 2> /dev/null` # If the mfs section size is smaller than the mfs image - bail. [ ${sec_size} -lt ${mfs_size} ] && echo "MFS image too large" && exit 1 From b88a8d3d1d04a73764b09d94065ff4678cde6ce3 Mon Sep 17 00:00:00 2001 From: Konstantin Belousov Date: Tue, 10 Jan 2017 19:26:55 +0000 Subject: [PATCH 42/79] Fix acquisition of nested write compat rtld locks. Obtaining compat rtld lock in write mode sets process signal mask to block all signals. Previous mask is stored in the global variable oldsigmask. If a lock is write-locked while another lock is already write-locked, oldsigmask is overwritten by the total mask and on the last unlock, all signals except traps appear to be blocked. Fix this by counting the write-lock nested level, and only storing to oldsigmask/restoring from it at the outermost level. Masking signals disables involuntary preemption for libc_r, and there could be no voluntary context switches in the locked code (dl_iterate_phdr(3) keeps a lock around user callback, but it was added long after libc_r was renounced). Due to this, remembering the level in the global variable after the lock is obtained should be safe, because no two libc_r threads can acquire different write locks in parallel. PR: 215826 Reported by: kami Tested by: yamagi@yamagi.org (previous version) To be reviewed by: kan Sponsored by: The FreeBSD Foundation MFC after: 2 weeks --- libexec/rtld-elf/rtld_lock.c | 51 ++++++++++++++++++++---------------- 1 file changed, 28 insertions(+), 23 deletions(-) diff --git a/libexec/rtld-elf/rtld_lock.c b/libexec/rtld-elf/rtld_lock.c index eeaea5e578b0..c4d2a0b0a7b1 100644 --- a/libexec/rtld-elf/rtld_lock.c +++ b/libexec/rtld-elf/rtld_lock.c @@ -64,7 +64,7 @@ typedef struct Struct_Lock { } Lock; static sigset_t fullsigmask, oldsigmask; -static int thread_flag; +static int thread_flag, wnested; static void * def_lock_create(void) @@ -117,29 +117,34 @@ def_rlock_acquire(void *lock) static void def_wlock_acquire(void *lock) { - Lock *l = (Lock *)lock; - sigset_t tmp_oldsigmask; + Lock *l; + sigset_t tmp_oldsigmask; - for ( ; ; ) { - sigprocmask(SIG_BLOCK, &fullsigmask, &tmp_oldsigmask); - if (atomic_cmpset_acq_int(&l->lock, 0, WAFLAG)) - break; - sigprocmask(SIG_SETMASK, &tmp_oldsigmask, NULL); - } - oldsigmask = tmp_oldsigmask; + l = (Lock *)lock; + for (;;) { + sigprocmask(SIG_BLOCK, &fullsigmask, &tmp_oldsigmask); + if (atomic_cmpset_acq_int(&l->lock, 0, WAFLAG)) + break; + sigprocmask(SIG_SETMASK, &tmp_oldsigmask, NULL); + } + if (atomic_fetchadd_int(&wnested, 1) == 0) + oldsigmask = tmp_oldsigmask; } static void def_lock_release(void *lock) { - Lock *l = (Lock *)lock; + Lock *l; - if ((l->lock & WAFLAG) == 0) - atomic_add_rel_int(&l->lock, -RC_INCR); - else { - atomic_add_rel_int(&l->lock, -WAFLAG); - sigprocmask(SIG_SETMASK, &oldsigmask, NULL); - } + l = (Lock *)lock; + if ((l->lock & WAFLAG) == 0) + atomic_add_rel_int(&l->lock, -RC_INCR); + else { + assert(wnested > 0); + atomic_add_rel_int(&l->lock, -WAFLAG); + if (atomic_fetchadd_int(&wnested, -1) == 1) + sigprocmask(SIG_SETMASK, &oldsigmask, NULL); + } } static int @@ -373,12 +378,12 @@ _rtld_atfork_pre(int *locks) return; /* - * Warning: this does not work with the rtld compat locks - * above, since the thread signal mask is corrupted (set to - * all signals blocked) if two locks are taken in write mode. - * The caller of the _rtld_atfork_pre() must provide the - * working implementation of the locks, and libthr locks are - * fine. + * Warning: this did not worked well with the rtld compat + * locks above, when the thread signal mask was corrupted (set + * to all signals blocked) if two locks were taken + * simultaneously in the write mode. The caller of the + * _rtld_atfork_pre() must provide the working implementation + * of the locks anyway, and libthr locks are fine. */ wlock_acquire(rtld_phdr_lock, &ls[0]); wlock_acquire(rtld_bind_lock, &ls[1]); From 30007e3fdcce2adeeac5fb2c08db5da27fc631f9 Mon Sep 17 00:00:00 2001 From: Sergey Kandaurov Date: Tue, 10 Jan 2017 19:28:40 +0000 Subject: [PATCH 43/79] Fix build without IEEE80211_DEBUG. Reported by: many --- sys/net80211/ieee80211_vht.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/sys/net80211/ieee80211_vht.c b/sys/net80211/ieee80211_vht.c index deb3811026af..e50c11c5edcd 100644 --- a/sys/net80211/ieee80211_vht.c +++ b/sys/net80211/ieee80211_vht.c @@ -91,11 +91,12 @@ vht_recv_action_placeholder(struct ieee80211_node *ni, const uint8_t *frm, const uint8_t *efrm) { +#ifdef IEEE80211_DEBUG ieee80211_note(ni->ni_vap, "%s: called; fc=0x%.2x/0x%.2x", __func__, wh->i_fc[0], wh->i_fc[1]); - +#endif return (0); } @@ -104,10 +105,12 @@ vht_send_action_placeholder(struct ieee80211_node *ni, int category, int action, void *arg0) { +#ifdef IEEE80211_DEBUG ieee80211_note(ni->ni_vap, "%s: called; category=%d, action=%d", __func__, category, action); +#endif return (EINVAL); } From 06212e046742e30fff834abade9592afbcef0a98 Mon Sep 17 00:00:00 2001 From: Oleksandr Tymoshenko Date: Tue, 10 Jan 2017 20:08:21 +0000 Subject: [PATCH 44/79] [efi] Fix off-by-one error in ARM .bss zeroing code in loader's _start __bss_end should not be included in .bss zeroing code. Otherwise first 4 bytes of the section that follows .bss (in loader's case it's .sdata) are overwritten by zero. Reviewed by: andrew MFC after: 3 days Differential Revision: https://reviews.freebsd.org/D9108 --- sys/boot/efi/loader/arch/arm/start.S | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sys/boot/efi/loader/arch/arm/start.S b/sys/boot/efi/loader/arch/arm/start.S index 443de4af1f87..9bc0c9f0f08b 100644 --- a/sys/boot/efi/loader/arch/arm/start.S +++ b/sys/boot/efi/loader/arch/arm/start.S @@ -161,7 +161,7 @@ _start: mov r2, #0 1: cmp r0, r1 - bgt 2f + bge 2f str r2, [r0], #4 b 1b 2: From 546fa3ada3928679231dcf7566a880dd56c7243e Mon Sep 17 00:00:00 2001 From: Oleksandr Tymoshenko Date: Tue, 10 Jan 2017 20:12:07 +0000 Subject: [PATCH 45/79] [efi] Fix .rel.data.* being erroneously merged into .data on ARM Fix section pattern code to exclude .rel.data.* sections from being merged into .data. Otherwise relocations in those sections are lost in final binary Reviewed by: andrew MFC after: 3 days Differential Revision: https://reviews.freebsd.org/D9108 --- sys/boot/efi/loader/arch/arm/ldscript.arm | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sys/boot/efi/loader/arch/arm/ldscript.arm b/sys/boot/efi/loader/arch/arm/ldscript.arm index 7ff4cc6a4cb2..9f6e7c29117e 100644 --- a/sys/boot/efi/loader/arch/arm/ldscript.arm +++ b/sys/boot/efi/loader/arch/arm/ldscript.arm @@ -18,7 +18,7 @@ SECTIONS . = ALIGN(16); .data : { - *(.data *.data.*) + *(.data .data.*) *(.gnu.linkonce.d*) *(.rodata) *(.rodata.*) From 759e131362059b638c5640c6f17a572006287bf1 Mon Sep 17 00:00:00 2001 From: Oleksandr Tymoshenko Date: Tue, 10 Jan 2017 20:15:24 +0000 Subject: [PATCH 46/79] [efi] Build EFI bits with -fPIC on ARM clang 3.9.0 without -fPIC generates absolute jump table for switch/case statement which trips boot1.efi and loader.efi on ARM platform. Reviewed by: andrew MFC after: 3 days Differential Revision: https://reviews.freebsd.org/D9108 --- sys/boot/efi/Makefile.inc | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/sys/boot/efi/Makefile.inc b/sys/boot/efi/Makefile.inc index 671033c58e80..eac5992a4c40 100644 --- a/sys/boot/efi/Makefile.inc +++ b/sys/boot/efi/Makefile.inc @@ -23,4 +23,8 @@ CFLAGS+= -fshort-wchar CFLAGS+= -fPIC .endif +.if ${MACHINE_CPUARCH} == "arm" +CFLAGS+= -fPIC +.endif + .include "../Makefile.inc" From 4982dca18d2dde64fbc0483208b961af1d4ef030 Mon Sep 17 00:00:00 2001 From: Alexander Motin Date: Tue, 10 Jan 2017 20:16:16 +0000 Subject: [PATCH 47/79] Do not wait for HA thread shutdown if scheduler is stopped. This wait loop made system hang on panic instead of reboot. MFC after: 1 week --- sys/cam/ctl/ctl_ha.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sys/cam/ctl/ctl_ha.c b/sys/cam/ctl/ctl_ha.c index d7f21c7a74b6..3402f040ee02 100644 --- a/sys/cam/ctl/ctl_ha.c +++ b/sys/cam/ctl/ctl_ha.c @@ -1001,7 +1001,7 @@ ctl_ha_msg_shutdown(struct ctl_softc *ctl_softc) softc->ha_shutdown = 1; softc->ha_wakeup = 1; wakeup(&softc->ha_wakeup); - while (softc->ha_shutdown < 2) { + while (softc->ha_shutdown < 2 && !SCHEDULER_STOPPED()) { msleep(&softc->ha_wakeup, &softc->ha_lock, 0, "shutdown", hz); } From 9a1fe327ae3ce405d1686352a3797788c7165f5d Mon Sep 17 00:00:00 2001 From: Alan Somers Date: Tue, 10 Jan 2017 20:35:09 +0000 Subject: [PATCH 48/79] ATFify the geom gate tests. This ensures their cleanup routines will be run even if they should timeout. tests/sys/geom/class/gate/ggate_test.sh tests/sys/geom/class/gate/Makefile Add an ATF test with three testcases, one for each TAP test. Use ATF-style cleanup functions, and convert sleeps to polling loops. ObsoleteFiles.inc tests/sys/geom/class/gate/conf.sh tests/sys/geom/class/gate/1_test.sh tests/sys/geom/class/gate/2_test.sh tests/sys/geom/class/gate/3_test.sh Delete TAP test files Reviewed by: ngie MFC after: 4 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D8891 --- ObsoleteFiles.inc | 5 + tests/sys/geom/class/gate/1_test.sh | 63 -------- tests/sys/geom/class/gate/2_test.sh | 48 ------ tests/sys/geom/class/gate/3_test.sh | 49 ------ tests/sys/geom/class/gate/Makefile | 10 +- tests/sys/geom/class/gate/conf.sh | 8 - tests/sys/geom/class/gate/ggate_test.sh | 193 ++++++++++++++++++++++++ 7 files changed, 199 insertions(+), 177 deletions(-) delete mode 100644 tests/sys/geom/class/gate/1_test.sh delete mode 100644 tests/sys/geom/class/gate/2_test.sh delete mode 100644 tests/sys/geom/class/gate/3_test.sh delete mode 100755 tests/sys/geom/class/gate/conf.sh create mode 100755 tests/sys/geom/class/gate/ggate_test.sh diff --git a/ObsoleteFiles.inc b/ObsoleteFiles.inc index bbd23609823e..c266d2edcc98 100644 --- a/ObsoleteFiles.inc +++ b/ObsoleteFiles.inc @@ -38,6 +38,11 @@ # xargs -n1 | sort | uniq -d; # done +# 20170110: Four files from ggate tests consolidated into one +OLD_FILES+=usr/tests/sys/geom/class/gate/1_test +OLD_FILES+=usr/tests/sys/geom/class/gate/2_test +OLD_FILES+=usr/tests/sys/geom/class/gate/3_test +OLD_FILES+=usr/tests/sys/geom/class/gate/conf.sh # 20170103: libbsnmptools.so made into an INTERNALLIB OLD_FILES+=usr/lib/libbsnmptools.a OLD_FILES+=usr/lib/libbsnmptools_p.a diff --git a/tests/sys/geom/class/gate/1_test.sh b/tests/sys/geom/class/gate/1_test.sh deleted file mode 100644 index ba573bb6e122..000000000000 --- a/tests/sys/geom/class/gate/1_test.sh +++ /dev/null @@ -1,63 +0,0 @@ -#!/bin/sh -# $FreeBSD$ - -. `dirname $0`/conf.sh - -echo '1..2' - -base=`basename $0` -us=0 -while [ -c /dev/ggate${us} ]; do - : $(( us += 1 )) -done -pidfile=ggated.$$.pid -conf=`mktemp $base.XXXXXX` || exit 1 -port=33080 - -work=$(attach_md -t malloc -s 1M) -src=$(attach_md -t malloc -s 1M) - -test_cleanup() -{ - ggatec destroy -f -u $us - pkill -F $pidfile - geom_test_cleanup -} -trap test_cleanup ABRT EXIT INT TERM - -dd if=/dev/random of=/dev/$work bs=1m count=1 conv=sync -dd if=/dev/random of=/dev/$src bs=1m count=1 conv=sync -src_checksum=$(md5 -q /dev/$src) - -echo "127.0.0.1 RW /dev/$work" > $conf - -if ! ggated -p $port -F $pidfile $conf; then - echo 'ggated failed to start' - echo 'Bail out!' - exit 1 -fi -sleep 1 -if ! ggatec create -p $port -u $us 127.0.0.1 /dev/$work; then - echo 'ggatec create failed' - echo 'Bail out!' - exit 1 -fi -sleep 1 - -dd if=/dev/${src} of=/dev/ggate${us} bs=1m count=1 -sleep 1 - -work_checksum=$(md5 -q /dev/$work) -if [ "$work_checksum" != "$src_checksum" ]; then - echo "not ok 1 - md5 checksums didn't match ($work_checksum != $src_checksum)" - echo "not ok 2 # SKIP" -else - echo 'ok 1 - md5 checksum' - - ggate_checksum=$(md5 -q /dev/ggate${us}) - if [ "$ggate_checksum" != "$src_checksum" ]; then - echo "not ok 2 - md5 checksums didn't match ($ggate_checksum != $src_checksum)" - else - echo 'ok 2 - md5 checksum' - fi -fi diff --git a/tests/sys/geom/class/gate/2_test.sh b/tests/sys/geom/class/gate/2_test.sh deleted file mode 100644 index be89accfcb9c..000000000000 --- a/tests/sys/geom/class/gate/2_test.sh +++ /dev/null @@ -1,48 +0,0 @@ -#!/bin/sh -# $FreeBSD$ - -. `dirname $0`/conf.sh - -base=`basename $0` -us=46 -work=`mktemp -u $base.XXXXXX` || exit 1 -src=`mktemp -u $base.XXXXXX` || exit 1 - -test_cleanup() -{ - ggatel destroy -f -u $us - rm -f $work $src - - geom_test_cleanup -} -trap test_cleanup ABRT EXIT INT TERM - -dd if=/dev/random of=$work bs=1m count=1 conv=sync -dd if=/dev/random of=$src bs=1m count=1 conv=sync - -if ! ggatel create -u $us $work; then - echo 'ggatel create failed' - echo 'Bail out!' - exit 1 -fi - -dd if=${src} of=/dev/ggate${us} bs=1m count=1 -sleep 1 - -echo '1..2' - -src_checksum=$(md5 -q $src) -work_checksum=$(md5 -q $work) -if [ "$work_checksum" != "$src_checksum" ]; then - echo "not ok 1 - md5 checksums didn't match ($work_checksum != $src_checksum) # TODO: bug 204616" - echo 'not ok 2 # SKIP' -else - echo 'ok 1 - md5 checksum' - - ggate_checksum=$(md5 -q /dev/ggate${us}) - if [ "$ggate_checksum" != "$src_checksum" ]; then - echo "not ok 2 - md5 checksums didn't match ($ggate_checksum != $src_checksum)" - else - echo 'ok 2 - md5 checksum' - fi -fi diff --git a/tests/sys/geom/class/gate/3_test.sh b/tests/sys/geom/class/gate/3_test.sh deleted file mode 100644 index 3511df761ca6..000000000000 --- a/tests/sys/geom/class/gate/3_test.sh +++ /dev/null @@ -1,49 +0,0 @@ -#!/bin/sh -# $FreeBSD$ - -. `dirname $0`/conf.sh - -base=`basename $0` -us=47 - -test_cleanup() -{ - ggatel destroy -f -u $us - - geom_test_cleanup -} -trap test_cleanup ABRT EXIT INT TERM - -work=$(attach_md -t malloc -s 1M) -src=$(attach_md -t malloc -s 1M) - -dd if=/dev/random of=/dev/$work bs=1m count=1 conv=sync -dd if=/dev/random of=/dev/$src bs=1m count=1 conv=sync -src_checksum=$(md5 -q /dev/$src) - -if ! ggatel create -u $us /dev/$work; then - echo 'ggatel create failed' - echo 'Bail out!' - exit 1 -fi - -sleep 1 -dd if=/dev/${src} of=/dev/ggate${us} bs=1m count=1 conv=sync -sleep 1 - -echo '1..2' - -work_checksum=$(md5 -q /dev/$work) -if [ "$work_checksum" != "$src_checksum" ]; then - echo "not ok 1 - md5 checksums didn't match ($work_checksum != $src_checksum)" - echo 'not ok 2 # SKIP' -else - echo 'ok 1 - md5 checksum' - - ggate_checksum=$(md5 -q /dev/ggate${us}) - if [ "$ggate_checksum" != "$src_checksum" ]; then - echo "not ok 2 - md5 checksums didn't match ($ggate_checksum != $src_checksum)" - else - echo 'ok 2 - md5 checksum' - fi -fi diff --git a/tests/sys/geom/class/gate/Makefile b/tests/sys/geom/class/gate/Makefile index c034bf037efb..417fa7dfd197 100644 --- a/tests/sys/geom/class/gate/Makefile +++ b/tests/sys/geom/class/gate/Makefile @@ -4,14 +4,6 @@ PACKAGE= tests TESTSDIR= ${TESTSBASE}/sys/geom/class/${.CURDIR:T} -TAP_TESTS_SH+= 1_test -TAP_TESTS_SH+= 2_test -TAP_TESTS_SH+= 3_test - -${PACKAGE}FILES+= conf.sh - -.for t in ${TAP_TESTS_SH} -TEST_METADATA.$t+= required_user="root" -.endfor +ATF_TESTS_SH+= ggate_test .include diff --git a/tests/sys/geom/class/gate/conf.sh b/tests/sys/geom/class/gate/conf.sh deleted file mode 100755 index 7e22ce46af66..000000000000 --- a/tests/sys/geom/class/gate/conf.sh +++ /dev/null @@ -1,8 +0,0 @@ -#!/bin/sh -# $FreeBSD$ - -name="$(mktemp -u gate.XXXXXX)" -class="gate" -base=`basename $0` - -. `dirname $0`/../geom_subr.sh diff --git a/tests/sys/geom/class/gate/ggate_test.sh b/tests/sys/geom/class/gate/ggate_test.sh new file mode 100755 index 000000000000..ffdb341a4b28 --- /dev/null +++ b/tests/sys/geom/class/gate/ggate_test.sh @@ -0,0 +1,193 @@ +# $FreeBSD$ + +PIDFILE=ggated.pid +PLAINFILES=plainfiles +PORT=33080 +CONF=gg.exports +RETRIES=16 + +atf_test_case ggated cleanup +ggated_head() +{ + atf_set "descr" "ggated can proxy geoms" + atf_set "require.progs" "ggatec ggated" + atf_set "require.user" "root" + atf_set "timeout" 60 +} + +ggated_body() +{ + us=$(alloc_ggate_dev) + work=$(alloc_md) + src=$(alloc_md) + + dd if=/dev/random of=/dev/$work bs=1m count=1 conv=notrunc + dd if=/dev/random of=/dev/$src bs=1m count=1 conv=notrunc + + echo $CONF >> $PLAINFILES + echo "127.0.0.1 RW /dev/$work" > $CONF + + atf_check ggated -p $PORT -F $PIDFILE $CONF + for try in `jot $RETRIES`; do + ggatec create -p $PORT -u $us 127.0.0.1 /dev/$work && break + # wait for ggated to be ready + sleep 0.25 + done + if [ "$try" -eq "$RETRIES" ]; then + atf_fail "ggatec create failed" + fi + + for try in `jot $RETRIES`; do + dd if=/dev/${src} of=/dev/ggate${us} bs=1m count=1 conv=notrunc\ + && break + # Wait for /dev/ggate${us} to be ready + sleep 0.25 + done + if [ "$try" -eq "$RETRIES" ]; then + atf_fail "dd failed; /dev/ggate${us} isn't working" + fi + + checksum /dev/$src /dev/$work +} + +ggated_cleanup() +{ + common_cleanup +} + +atf_test_case ggatel_file cleanup +ggatel_file_head() +{ + atf_set "descr" "ggatel can proxy files" + atf_set "require.progs" "ggatel" + atf_set "require.user" "root" + atf_set "timeout" 15 +} + +ggatel_file_body() +{ + us=$(alloc_ggate_dev) + + echo src work >> ${PLAINFILES} + dd if=/dev/random of=work bs=1m count=1 + dd if=/dev/random of=src bs=1m count=1 + + atf_check ggatel create -u $us work + + dd if=src of=/dev/ggate${us} bs=1m count=1 conv=notrunc + + checksum src work +} + +ggatel_file_cleanup() +{ + common_cleanup +} + +atf_test_case ggatel_md cleanup +ggatel_md_head() +{ + atf_set "descr" "ggatel can proxy files" + atf_set "require.progs" "ggatel" + atf_set "require.user" "root" + atf_set "timeout" 15 +} + +ggatel_md_body() +{ + us=$(alloc_ggate_dev) + work=$(alloc_md) + src=$(alloc_md) + + dd if=/dev/random of=$work bs=1m count=1 conv=notrunc + dd if=/dev/random of=$src bs=1m count=1 conv=notrunc + + atf_check ggatel create -u $us /dev/$work + + dd if=/dev/$src of=/dev/ggate${us} bs=1m count=1 conv=notrunc + + checksum /dev/$src /dev/$work +} + +ggatel_md_cleanup() +{ + common_cleanup +} + +atf_init_test_cases() +{ + atf_add_test_case ggated + atf_add_test_case ggatel_file + atf_add_test_case ggatel_md +} + +alloc_ggate_dev() +{ + local us + + us=0 + while [ -c /dev/ggate${us} ]; do + : $(( us += 1 )) + done + echo ${us} > ggate.devs + echo ${us} +} + +alloc_md() +{ + local md + + md=$(mdconfig -a -t malloc -s 1M) || \ + atf_fail "failed to allocate md device" + echo ${md} >> md.devs + echo ${md} +} + +checksum() +{ + local src work + src=$1 + work=$2 + + src_checksum=$(md5 -q $src) + work_checksum=$(md5 -q $work) + + if [ "$work_checksum" != "$src_checksum" ]; then + atf_fail "work md5 checksum didn't match" + fi + + ggate_checksum=$(md5 -q /dev/ggate${us}) + if [ "$ggate_checksum" != "$src_checksum" ]; then + atf_fail "ggate md5 checksum didn't match" + fi +} + +common_cleanup() +{ + if [ -f "ggate.devs" ]; then + while read test_ggate; do + ggatec destroy -f -u $test_ggate >/dev/null + done < ggate.devs + rm ggate.devs + fi + + if [ -f "$PIDFILE" ]; then + pkill -F "$PIDFILE" + rm $PIDFILE + fi + + if [ -f "PLAINFILES" ]; then + while read f; do + rm -f ${f} + done < ${PLAINFILES} + rm ${PLAINFILES} + fi + + if [ -f "md.devs" ]; then + while read test_md; do + mdconfig -d -u $test_md 2>/dev/null + done < md.devs + rm md.devs + fi + true +} From 442f47d697a12c484b7fbd399afe457eefc4a031 Mon Sep 17 00:00:00 2001 From: Alan Somers Date: Tue, 10 Jan 2017 20:37:44 +0000 Subject: [PATCH 49/79] Fix typo from change 310985 in ObsoleteFiles.inc MFC after: 16 days X-MFC-With: 310803 Sponsored by: Spectra Logic Corp --- ObsoleteFiles.inc | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/ObsoleteFiles.inc b/ObsoleteFiles.inc index c266d2edcc98..e97bb0099470 100644 --- a/ObsoleteFiles.inc +++ b/ObsoleteFiles.inc @@ -53,8 +53,8 @@ OLD_FILES+=usr/share/man/man3/sysdecode_getfsstat_flags.3.gz # 20161230: libarchive ACL pax test renamed to test_acl_pax_posix1e.tar.uu OLD_FILES+=usr/tests/lib/libarchive/test_acl_pax.tar.uu # 20161229: Three files from gnop tests consolidated into one -OLD_FILES+=usr/tests/sys/geom/class/nop/1_test.sh -OLD_FILES+=usr/tests/sys/geom/class/nop/2_test.sh +OLD_FILES+=usr/tests/sys/geom/class/nop/1_test +OLD_FILES+=usr/tests/sys/geom/class/nop/2_test OLD_FILES+=usr/tests/sys/geom/class/nop/conf.sh # 20161217: new clang import which bumps version from 3.9.0 to 3.9.1. OLD_FILES+=usr/lib/clang/3.9.0/include/sanitizer/allocator_interface.h From cdb7a6fc4244d794a3e0c312c161adfd8442eef6 Mon Sep 17 00:00:00 2001 From: Alan Somers Date: Tue, 10 Jan 2017 20:43:32 +0000 Subject: [PATCH 50/79] Fix memory leaks during "tail -r" of an irregular file * Rewrite r_buf to use standard tail queues instead of a hand-rolled circular linked list. Free dynamic allocations when done. * Remove an optimization for the case where the file is a multiple of 128KB in size and there is a scarcity of memory. * Add ATF tests for "tail -r" and its variants. Reported by: Valgrind Reviewed by: ngie MFC after: 4 weeks Sponsored by: Spectra Logic Corp Differential Revision: https://reviews.freebsd.org/D9067 --- etc/mtree/BSD.tests.dist | 2 + usr.bin/tail/Makefile | 6 + usr.bin/tail/reverse.c | 140 +++++++++---------- usr.bin/tail/tests/Makefile | 7 + usr.bin/tail/tests/tail_test.sh | 233 ++++++++++++++++++++++++++++++++ 5 files changed, 311 insertions(+), 77 deletions(-) create mode 100644 usr.bin/tail/tests/Makefile create mode 100755 usr.bin/tail/tests/tail_test.sh diff --git a/etc/mtree/BSD.tests.dist b/etc/mtree/BSD.tests.dist index 620f95c04665..60e2dbea9ae6 100644 --- a/etc/mtree/BSD.tests.dist +++ b/etc/mtree/BSD.tests.dist @@ -644,6 +644,8 @@ .. soelim .. + tail + .. tar .. timeout diff --git a/usr.bin/tail/Makefile b/usr.bin/tail/Makefile index 672cbed008e6..6366f6f15f72 100644 --- a/usr.bin/tail/Makefile +++ b/usr.bin/tail/Makefile @@ -1,7 +1,13 @@ # $FreeBSD$ # @(#)Makefile 8.1 (Berkeley) 6/6/93 +.include + PROG= tail SRCS= forward.c misc.c read.c reverse.c tail.c +.if ${MK_TESTS} != "no" +SUBDIR+= tests +.endif + .include diff --git a/usr.bin/tail/reverse.c b/usr.bin/tail/reverse.c index 87269051ed68..3ad9e4bc9fdc 100644 --- a/usr.bin/tail/reverse.c +++ b/usr.bin/tail/reverse.c @@ -40,6 +40,7 @@ static char sccsid[] = "@(#)reverse.c 8.1 (Berkeley) 6/6/93"; __FBSDID("$FreeBSD$"); #include +#include #include #include @@ -169,12 +170,12 @@ r_reg(FILE *fp, const char *fn, enum STYLE style, off_t off, struct stat *sbp) ierr(fn); } -typedef struct bf { - struct bf *next; - struct bf *prev; - int len; - char *l; -} BF; +static const size_t bsz = 128 * 1024; +typedef struct bfelem { + TAILQ_ENTRY(bfelem) entries; + size_t len; + char l[bsz]; +} bfelem_t; /* * r_buf -- display a non-regular file in reverse order by line. @@ -189,64 +190,42 @@ typedef struct bf { static void r_buf(FILE *fp, const char *fn) { - BF *mark, *tl, *tr; - int ch, len, llen; + struct bfelem *tl, *temp, *first = NULL; + size_t len, llen; char *p; - off_t enomem; + off_t enomem = 0; + TAILQ_HEAD(bfhead, bfelem) head; - tl = NULL; -#define BSZ (128 * 1024) - for (mark = NULL, enomem = 0;;) { + TAILQ_INIT(&head); + + while (!feof(fp)) { /* * Allocate a new block and link it into place in a doubly * linked list. If out of memory, toss the LRU block and * keep going. */ - if (enomem || (tl = malloc(sizeof(BF))) == NULL || - (tl->l = malloc(BSZ)) == NULL) { - if (!mark) + while ((tl = malloc(sizeof(bfelem_t))) == NULL) { + first = TAILQ_FIRST(&head); + if (TAILQ_EMPTY(&head)) err(1, "malloc"); - if (enomem) - tl = tl->next; - else { - if (tl) - free(tl); - tl = mark; - } - enomem += tl->len; - } else if (mark) { - tl->next = mark; - tl->prev = mark->prev; - mark->prev->next = tl; - mark->prev = tl; - } else { - mark = tl; - mark->next = mark->prev = mark; + enomem += first->len; + TAILQ_REMOVE(&head, first, entries); + free(first); } + TAILQ_INSERT_TAIL(&head, tl, entries); /* Fill the block with input data. */ - for (p = tl->l, len = 0; - len < BSZ && (ch = getc(fp)) != EOF; ++len) - *p++ = ch; - - if (ferror(fp)) { - ierr(fn); - return; - } - - /* - * If no input data for this block and we tossed some data, - * recover it. - */ - if (!len && enomem) { - enomem -= tl->len; - tl = tl->prev; - break; + len = 0; + while ((!feof(fp)) && len < bsz) { + p = tl->l + len; + len += fread(p, 1, bsz - len, fp); + if (ferror(fp)) { + ierr(fn); + return; + } } tl->len = len; - if (ch == EOF) - break; } if (enomem) { @@ -255,37 +234,44 @@ r_buf(FILE *fp, const char *fn) } /* - * Step through the blocks in the reverse order read. The last char - * is special, ignore whether newline or not. + * Now print the lines in reverse order + * Outline: + * Scan backward for "\n", + * print forward to the end of the buffers + * free any buffers that start after the "\n" just found + * Loop */ - for (mark = tl;;) { - for (p = tl->l + (len = tl->len) - 1, llen = 0; len--; - --p, ++llen) - if (*p == '\n') { - if (llen) { + tl = TAILQ_LAST(&head, bfhead); + first = TAILQ_FIRST(&head); + while (tl != NULL) { + for (p = tl->l + tl->len - 1, llen = 0; p >= tl->l; + --p, ++llen) { + int start = (tl == first && p == tl->l); + + if ((*p == '\n') || start) { + struct bfelem *tr; + + if (start && len) + WR(p, llen + 1); + else if (llen) WR(p + 1, llen); - llen = 0; - } - if (tl == mark) - continue; - for (tr = tl->next; tr->len; tr = tr->next) { - WR(tr->l, tr->len); - tr->len = 0; - if (tr == mark) - break; + tr = TAILQ_NEXT(tl, entries); + llen = 0; + if (tr != NULL) { + TAILQ_FOREACH_FROM_SAFE(tr, &head, + entries, temp) { + if (tr->len) + WR(&tr->l, tr->len); + TAILQ_REMOVE(&head, tr, + entries); + free(tr); + } } } + } tl->len = llen; - if ((tl = tl->prev) == mark) - break; - } - tl = tl->next; - if (tl->len) { - WR(tl->l, tl->len); - tl->len = 0; - } - while ((tl = tl->next)->len) { - WR(tl->l, tl->len); - tl->len = 0; + tl = TAILQ_PREV(tl, bfhead, entries); } + TAILQ_REMOVE(&head, first, entries); + free(first); } diff --git a/usr.bin/tail/tests/Makefile b/usr.bin/tail/tests/Makefile new file mode 100644 index 000000000000..84338e292dc0 --- /dev/null +++ b/usr.bin/tail/tests/Makefile @@ -0,0 +1,7 @@ +# $FreeBSD$ + +PACKAGE= tests + +ATF_TESTS_SH= tail_test + +.include diff --git a/usr.bin/tail/tests/tail_test.sh b/usr.bin/tail/tests/tail_test.sh new file mode 100755 index 000000000000..3e407ea3ff00 --- /dev/null +++ b/usr.bin/tail/tests/tail_test.sh @@ -0,0 +1,233 @@ +# Copyright (c) 2016 Alan Somers +# All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# 1. Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# 2. Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in the +# documentation and/or other materials provided with the distribution. +# +# THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND +# ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +# ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE +# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS +# OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) +# HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT +# LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY +# OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF +# SUCH DAMAGE. +# +# $FreeBSD$ + +atf_test_case empty_r +empty_r_head() +{ + atf_set "descr" "Reverse an empty file" +} +empty_r_body() +{ + touch infile expectfile + tail -r infile > outfile + tail -r < infile > outpipe + atf_check cmp expectfile outfile + atf_check cmp expectfile outpipe +} + +atf_test_case file_r +file_r_head() +{ + atf_set "descr" "Reverse a file" +} +file_r_body() +{ + cat > infile < expectfile << HERE +This is the third line +This is the second line +This is the first line +HERE + tail -r infile > outfile + tail -r < infile > outpipe + atf_check cmp expectfile outfile + atf_check cmp expectfile outpipe +} + +atf_test_case file_rn2 +file_rn2_head() +{ + atf_set "descr" "Reverse the last two lines of a file" +} +file_rn2_body() +{ + cat > infile < expectfile << HERE +This is the third line +This is the second line +HERE + tail -rn2 infile > outfile + tail -rn2 < infile > outpipe + atf_check cmp expectfile outfile + atf_check cmp expectfile outpipe +} + +atf_test_case file_rc28 +file_rc28_head() +{ + atf_set "descr" "Reverse a file and display the last 28 characters" +} +file_rc28_body() +{ + cat > infile < expectfile << HERE +This is the third line +line +HERE + tail -rc28 infile > outfile + tail -rc28 < infile > outpipe + atf_check cmp expectfile outfile + atf_check cmp expectfile outpipe +} + +atf_test_case longfile_r +longfile_r_head() +{ + atf_set "descr" "Reverse a long file" +} +longfile_r_body() +{ + jot -w "%0511d" 1030 0 > infile + jot -w "%0511d" 1030 1029 0 -1 > expectfile + tail -r infile > outfile + tail -r < infile > outpipe + atf_check cmp expectfile outfile + atf_check cmp expectfile outpipe +} + +atf_test_case longfile_r_enomem +longfile_r_enomem_head() +{ + atf_set "descr" "Reverse a file that's too long to store in RAM" +} +longfile_r_enomem_body() +{ + # When we reverse a file that's too long for RAM, tail should drop the + # first part and just print what it can. We'll check that the last + # part is ok + { + ulimit -v 32768 || atf_skip "Can't adjust ulimit" + jot -w "%01023d" 32768 0 | tail -r > outfile ; + } + if [ "$?" -ne 1 ]; then + atf_skip "Didn't get ENOMEM. Adjust test parameters" + fi + # We don't know how much of the input we dropped. So just check that + # the first ten lines of tail's output are the same as the last ten of + # the input + jot -w "%01023d" 10 32767 0 -1 > expectfile + head -n 10 outfile > outtrunc + diff expectfile outtrunc + atf_check cmp expectfile outtrunc +} + +atf_test_case longfile_r_longlines +longfile_r_longlines_head() +{ + atf_set "descr" "Reverse a long file with extremely long lines" +} +longfile_r_longlines_body() +{ + jot -s " " -w "%07d" 18000 0 > infile + jot -s " " -w "%07d" 18000 18000 >> infile + jot -s " " -w "%07d" 18000 36000 >> infile + jot -s " " -w "%07d" 18000 36000 > expectfile + jot -s " " -w "%07d" 18000 18000 >> expectfile + jot -s " " -w "%07d" 18000 0 >> expectfile + tail -r infile > outfile + tail -r < infile > outpipe + atf_check cmp expectfile outfile + atf_check cmp expectfile outpipe +} + +atf_test_case longfile_rc135782 +longfile_rc135782_head() +{ + atf_set "descr" "Reverse a long file and print the last 135,782 bytes" +} +longfile_rc135782_body() +{ + jot -w "%063d" 9000 0 > infile + jot -w "%063d" 2121 8999 0 -1 > expectfile + echo "0000000000000000000000000000000006878" >> expectfile + tail -rc135782 infile > outfile + tail -rc135782 < infile > outpipe + atf_check cmp expectfile outfile + atf_check cmp expectfile outpipe +} + +atf_test_case longfile_rc145782_longlines +longfile_rc145782_longlines_head() +{ + atf_set "descr" "Reverse a long file with extremely long lines and print the last 145,782 bytes" +} +longfile_rc145782_longlines_body() +{ + jot -s " " -w "%07d" 18000 0 > infile + jot -s " " -w "%07d" 18000 18000 >> infile + jot -s " " -w "%07d" 18000 36000 >> infile + jot -s " " -w "%07d" 18000 36000 > expectfile + echo -n "35777 " >> expectfile + jot -s " " -w "%07d" 222 35778 >> expectfile + tail -rc145782 infile > outfile + tail -rc145782 < infile > outpipe + atf_check cmp expectfile outfile + atf_check cmp expectfile outpipe +} + +atf_test_case longfile_rn2500 +longfile_rn2500_head() +{ + atf_set "descr" "Reverse a long file and print the last 2,500 lines" +} +longfile_rn2500_body() +{ + jot -w "%063d" 9000 0 > infile + jot -w "%063d" 2500 8999 0 -1 > expectfile + tail -rn2500 infile > outfile + tail -rn2500 < infile > outpipe + atf_check cmp expectfile outfile + atf_check cmp expectfile outpipe +} + + +atf_init_test_cases() +{ + atf_add_test_case empty_r + atf_add_test_case file_r + atf_add_test_case file_rc28 + atf_add_test_case file_rn2 + # The longfile tests are designed to exercise behavior in r_buf(), + # which operates on 128KB blocks + atf_add_test_case longfile_r + atf_add_test_case longfile_r_enomem + atf_add_test_case longfile_r_longlines + atf_add_test_case longfile_rc135782 + atf_add_test_case longfile_rc145782_longlines + atf_add_test_case longfile_rn2500 +} From 990c731f0d7724fb1c53b594327314db9692da35 Mon Sep 17 00:00:00 2001 From: "Pedro F. Giffuni" Date: Tue, 10 Jan 2017 20:44:31 +0000 Subject: [PATCH 51/79] Remove unused __gnu_inline() attribute. This was meant to be used by a future FORTIFY_SOURCE implementation. Probably for good, FORTIFY_SOURCE and this particular GCCism were never well supported by clang or other compilers. Furthermore, the technology has long since been replaced by either static checkers, sanitizers, or even just the strong stack protector that was enabled by default. Drop __gnu_inline to avoid cluttering the headers. MFC after: 5 days --- sys/sys/cdefs.h | 16 ---------------- 1 file changed, 16 deletions(-) diff --git a/sys/sys/cdefs.h b/sys/sys/cdefs.h index 453b395a806d..7db4f25542b3 100644 --- a/sys/sys/cdefs.h +++ b/sys/sys/cdefs.h @@ -543,22 +543,6 @@ __attribute__((__format__ (__strftime__, fmtarg, firstvararg))) #endif -/* - * FORTIFY_SOURCE, and perhaps other compiler-specific features, require - * the use of non-standard inlining. In general we should try to avoid - * using these but GCC-compatible compilers tend to support the extensions - * well enough to use them in limited cases. - */ -#if defined(__GNUC_GNU_INLINE__) || defined(__GNUC_STDC_INLINE__) -#if __GNUC_PREREQ__(4, 3) || __has_attribute(__artificial__) -#define __gnu_inline __attribute__((__gnu_inline__, __artificial__)) -#else -#define __gnu_inline __attribute__((__gnu_inline__)) -#endif /* artificial */ -#else -#define __gnu_inline -#endif - /* Compiler-dependent macros that rely on FreeBSD-specific extensions. */ #if defined(__FreeBSD_cc_version) && __FreeBSD_cc_version >= 300001 && \ defined(__GNUC__) && !defined(__INTEL_COMPILER) From 36289c33368ffb8c930da60cbcacf65523e0c544 Mon Sep 17 00:00:00 2001 From: Alexander Motin Date: Tue, 10 Jan 2017 20:52:44 +0000 Subject: [PATCH 52/79] Add checks for received mode page length. If our buffer is too small, we may receive part of the page, and should not try read/write past the end of the buffer. Reported by: Coverity CID: 1368374, 1368375 MFC after: 1 week --- sbin/camcontrol/modeedit.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/sbin/camcontrol/modeedit.c b/sbin/camcontrol/modeedit.c index cbc138caf391..90862db193be 100644 --- a/sbin/camcontrol/modeedit.c +++ b/sbin/camcontrol/modeedit.c @@ -557,7 +557,7 @@ editlist_populate(struct cam_device *device, int dbd, int pc, int page, struct scsi_mode_header_6 *mh; /* Location of mode header. */ struct scsi_mode_page_header *mph; struct scsi_mode_page_header_sp *mphsp; - int len; + size_t len; STAILQ_INIT(&editlist); @@ -575,6 +575,7 @@ editlist_populate(struct cam_device *device, int dbd, int pc, int page, mode_pars = (uint8_t *)(mphsp + 1); len = scsi_2btoul(mphsp->page_length); } + len = MIN(len, sizeof(data) - (mode_pars - data)); /* Decode the value data, creating edit_entries for each value. */ buff_decode_visit(mode_pars, len, format, editentry_create, 0); @@ -594,7 +595,7 @@ editlist_save(struct cam_device *device, int dbd, int pc, int page, struct scsi_mode_header_6 *mh; /* Location of mode header. */ struct scsi_mode_page_header *mph; struct scsi_mode_page_header_sp *mphsp; - int len, hlen; + size_t len, hlen; /* Make sure that something changed before continuing. */ if (! editlist_changed) @@ -617,6 +618,7 @@ editlist_save(struct cam_device *device, int dbd, int pc, int page, mode_pars = (uint8_t *)(mphsp + 1); len = scsi_2btoul(mphsp->page_length); } + len = MIN(len, sizeof(data) - (mode_pars - data)); /* Encode the value data to be passed back to the device. */ buff_encode_visit(mode_pars, len, format, editentry_save, 0); @@ -814,7 +816,7 @@ modepage_dump(struct cam_device *device, int dbd, int pc, int page, int subpage, struct scsi_mode_header_6 *mh; /* Location of mode header. */ struct scsi_mode_page_header *mph; struct scsi_mode_page_header_sp *mphsp; - int indx, len; + size_t indx, len; mode_sense(device, dbd, pc, page, subpage, retries, timeout, data, sizeof(data)); @@ -829,6 +831,7 @@ modepage_dump(struct cam_device *device, int dbd, int pc, int page, int subpage, mode_pars = (uint8_t *)(mphsp + 1); len = scsi_2btoul(mphsp->page_length); } + len = MIN(len, sizeof(data) - (mode_pars - data)); /* Print the raw mode page data with newlines each 8 bytes. */ for (indx = 0; indx < len; indx++) { From 4cf80fbd79aa026cc2650cd33e73e33bd7cacebc Mon Sep 17 00:00:00 2001 From: Mateusz Guzik Date: Tue, 10 Jan 2017 21:10:20 +0000 Subject: [PATCH 53/79] sparc64: add atomic_fcmpset Tested on hardware provided by feld. Reviewed by: marius --- sys/sparc64/include/atomic.h | 34 ++++++++++++++++++++++++++++++++++ 1 file changed, 34 insertions(+) diff --git a/sys/sparc64/include/atomic.h b/sys/sparc64/include/atomic.h index c70e99aea70b..b2fd8cc5a101 100644 --- a/sys/sparc64/include/atomic.h +++ b/sys/sparc64/include/atomic.h @@ -219,6 +219,40 @@ atomic_cmpset_rel_ ## name(volatile ptype p, vtype e, vtype s) \ return (((vtype)atomic_cas_rel((p), (e), (s), sz)) == (e)); \ } \ \ +static __inline int \ +atomic_fcmpset_ ## name(volatile ptype p, vtype *ep, vtype s) \ +{ \ + vtype t; \ + \ + t = (vtype)atomic_cas((p), (*ep), (s), sz); \ + if (t == (*ep)) \ + return (1); \ + *ep = t; \ + return (0); \ +} \ +static __inline int \ +atomic_fcmpset_acq_ ## name(volatile ptype p, vtype *ep, vtype s) \ +{ \ + vtype t; \ + \ + t = (vtype)atomic_cas_acq((p), (*ep), (s), sz); \ + if (t == (*ep)) \ + return (1); \ + *ep = t; \ + return (0); \ +} \ +static __inline int \ +atomic_fcmpset_rel_ ## name(volatile ptype p, vtype *ep, vtype s) \ +{ \ + vtype t; \ + \ + t = (vtype)atomic_cas_rel((p), (*ep), (s), sz); \ + if (t == (*ep)) \ + return (1); \ + *ep = t; \ + return (0); \ +} \ + \ static __inline vtype \ atomic_load_ ## name(volatile ptype p) \ { \ From 2a59734ec264ae1b692d282b01434fa4d4a1a083 Mon Sep 17 00:00:00 2001 From: Martin Matuska Date: Tue, 10 Jan 2017 21:18:32 +0000 Subject: [PATCH 54/79] Update vendor/libarchive to git 22f2d190639e6bd496a3b82f70c01fba0d38b40a Vendor changes: #691: Support for SCHILY.xattr extended attributes #854: Spelling fixes Multiple fixes in ACL code: - prefer acl_set_fd_np() to acl_set_fd() - if acl_set_fd_np() fails, do no fallback to acl_set_file() - do not warn if trying to write ACLs to a filesystem without ACL support - fix id handling in archive_acl_(from_to)_text*() for NFSv4 ACLs --- Makefile.am | 1 + libarchive/archive_acl.c | 8 +- .../archive_read_disk_entry_from_file.c | 10 +- libarchive/archive_read_disk_posix.c | 6 +- libarchive/archive_read_disk_windows.c | 2 +- libarchive/archive_read_support_filter_lz4.c | 6 +- libarchive/archive_read_support_filter_lzop.c | 2 +- libarchive/archive_read_support_format_7zip.c | 10 +- .../archive_read_support_format_iso9660.c | 2 +- libarchive/archive_read_support_format_lha.c | 2 +- libarchive/archive_read_support_format_rar.c | 2 +- libarchive/archive_read_support_format_tar.c | 39 ++- libarchive/archive_read_support_format_warc.c | 4 +- libarchive/archive_read_support_format_zip.c | 14 +- libarchive/archive_string.c | 26 +- libarchive/archive_string.h | 4 + libarchive/archive_string_composition.h | 2 +- libarchive/archive_write.c | 2 +- libarchive/archive_write_add_filter_xz.c | 2 +- libarchive/archive_write_disk_acl.c | 49 ++-- libarchive/archive_write_set_format_7zip.c | 2 +- libarchive/archive_write_set_format_pax.c | 65 ++++- libarchive/archive_write_set_format_xar.c | 2 +- libarchive/archive_write_set_format_zip.c | 10 +- libarchive/test/CMakeLists.txt | 1 + .../test/test_archive_read_add_passphrase.c | 2 +- libarchive/test/test_compat_uudecode.c | 2 +- libarchive/test/test_read_format_cpio_afio.c | 2 +- ...d_format_zip_traditional_encryption_data.c | 4 +- .../test/test_read_format_zip_winzip_aes.c | 2 +- .../test_read_format_zip_winzip_aes_large.c | 2 +- libarchive/test/test_read_pax_schily_xattr.c | 70 ++++++ .../test/test_read_pax_schily_xattr.tar.uu | 231 ++++++++++++++++++ libarchive/test/test_sparse_basic.c | 2 +- libarchive/xxhash.c | 5 +- 35 files changed, 498 insertions(+), 97 deletions(-) create mode 100644 libarchive/test/test_read_pax_schily_xattr.c create mode 100644 libarchive/test/test_read_pax_schily_xattr.tar.uu diff --git a/Makefile.am b/Makefile.am index 580fa2a6ec3b..f92d723a4280 100644 --- a/Makefile.am +++ b/Makefile.am @@ -497,6 +497,7 @@ libarchive_test_SOURCES= \ libarchive/test/test_read_format_zip_winzip_aes_large.c \ libarchive/test/test_read_format_zip_zip64.c \ libarchive/test/test_read_large.c \ + libarchive/test/test_read_pax_schily_xattr.c \ libarchive/test/test_read_pax_truncated.c \ libarchive/test/test_read_position.c \ libarchive/test/test_read_set_format.c \ diff --git a/libarchive/archive_acl.c b/libarchive/archive_acl.c index b8e8ba40d2ec..7c682095962a 100644 --- a/libarchive/archive_acl.c +++ b/libarchive/archive_acl.c @@ -786,7 +786,8 @@ append_entry_w(wchar_t **wp, const wchar_t *prefix, int type, } else if (tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { append_id_w(wp, id); - id = -1; + if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) == 0) + id = -1; } /* Solaris style has no second colon after other and mask */ if (((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) == 0) @@ -1042,7 +1043,8 @@ append_entry(char **p, const char *prefix, int type, } else if (tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { append_id(p, id); - id = -1; + if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) == 0) + id = -1; } /* Solaris style has no second colon after other and mask */ if (((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) == 0) @@ -1328,6 +1330,7 @@ archive_acl_from_text_w(struct archive_acl *acl, const wchar_t *text, tag == ARCHIVE_ENTRY_ACL_GROUP) { n = 1; name = field[1]; + isint_w(name.start, name.end, &id); } else n = 0; @@ -1799,6 +1802,7 @@ archive_acl_from_text_l(struct archive_acl *acl, const char *text, tag == ARCHIVE_ENTRY_ACL_GROUP) { n = 1; name = field[1]; + isint(name.start, name.end, &id); } else n = 0; diff --git a/libarchive/archive_read_disk_entry_from_file.c b/libarchive/archive_read_disk_entry_from_file.c index 0a5453c640ba..ce576e0acda2 100644 --- a/libarchive/archive_read_disk_entry_from_file.c +++ b/libarchive/archive_read_disk_entry_from_file.c @@ -526,6 +526,11 @@ setup_acls(struct archive_read_disk *a, /* Only directories can have default ACLs. */ if (S_ISDIR(archive_entry_mode(entry))) { +#if HAVE_ACL_GET_FD_NP + if (*fd >= 0) + acl = acl_get_fd_np(*fd, ACL_TYPE_DEFAULT); + else +#endif acl = acl_get_file(accpath, ACL_TYPE_DEFAULT); if (acl != NULL) { r = translate_acl(a, entry, acl, @@ -581,7 +586,10 @@ static struct { {ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, ACL_ENTRY_FILE_INHERIT}, {ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT, ACL_ENTRY_DIRECTORY_INHERIT}, {ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT, ACL_ENTRY_NO_PROPAGATE_INHERIT}, - {ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, ACL_ENTRY_INHERIT_ONLY} + {ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, ACL_ENTRY_INHERIT_ONLY}, + {ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS, ACL_ENTRY_SUCCESSFUL_ACCESS}, + {ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS, ACL_ENTRY_FAILED_ACCESS}, + {ARCHIVE_ENTRY_ACL_ENTRY_INHERITED, ACL_ENTRY_INHERITED} }; #endif static int diff --git a/libarchive/archive_read_disk_posix.c b/libarchive/archive_read_disk_posix.c index bc7fad1d9d2e..b89370421c07 100644 --- a/libarchive/archive_read_disk_posix.c +++ b/libarchive/archive_read_disk_posix.c @@ -675,7 +675,7 @@ setup_suitable_read_buffer(struct archive_read_disk *a) asize = cf->min_xfer_size; /* Increase a buffer size up to 64K bytes in - * a proper incremant size. */ + * a proper increment size. */ while (asize < 1024*64) asize += incr; /* Take a margin to adjust to the filesystem @@ -1656,7 +1656,7 @@ setup_current_filesystem(struct archive_read_disk *a) archive_set_error(&a->archive, errno, "statvfs failed"); return (ARCHIVE_FAILED); } else if (xr == 1) { - /* Usuall come here unless NetBSD supports _PC_REC_XFER_ALIGN + /* Usually come here unless NetBSD supports _PC_REC_XFER_ALIGN * for pathconf() function. */ t->current_filesystem->xfer_align = sfs.f_frsize; t->current_filesystem->max_xfer_size = -1; @@ -1944,7 +1944,7 @@ setup_current_filesystem(struct archive_read_disk *a) if (nm == -1) # endif /* _PC_NAME_MAX */ /* - * Some sysmtes (HP-UX or others?) incorrectly defined + * Some systems (HP-UX or others?) incorrectly defined * NAME_MAX macro to be a smaller value. */ # if defined(NAME_MAX) && NAME_MAX >= 255 diff --git a/libarchive/archive_read_disk_windows.c b/libarchive/archive_read_disk_windows.c index 6791b2c13620..27b75e3b10ce 100644 --- a/libarchive/archive_read_disk_windows.c +++ b/libarchive/archive_read_disk_windows.c @@ -1401,7 +1401,7 @@ close_and_restore_time(HANDLE h, struct tree *t, struct restore_time *rt) if (h == INVALID_HANDLE_VALUE && AE_IFLNK == rt->filetype) return (0); - /* Close a file descritor. + /* Close a file descriptor. * It will not be used for SetFileTime() because it has been opened * by a read only mode. */ diff --git a/libarchive/archive_read_support_filter_lz4.c b/libarchive/archive_read_support_filter_lz4.c index 37b2f59004b1..4c66ed04dc95 100644 --- a/libarchive/archive_read_support_filter_lz4.c +++ b/libarchive/archive_read_support_filter_lz4.c @@ -180,7 +180,7 @@ lz4_reader_bid(struct archive_read_filter_bidder *self, return (0); bits_checked += 8; BD = buffer[5]; - /* A block maximum size shuld be more than 3. */ + /* A block maximum size should be more than 3. */ if (((BD & 0x70) >> 4) < 4) return (0); /* Reserved bits must be "0". */ @@ -417,7 +417,7 @@ lz4_filter_read_descriptor(struct archive_read_filter *self) /* Reserved bits must be zero. */ if (bd & 0x8f) goto malformed_error; - /* Get a maxinum block size. */ + /* Get a maximum block size. */ switch (read_buf[1] >> 4) { case 4: /* 64 KB */ state->flags.block_maximum_size = 64 * 1024; @@ -627,7 +627,7 @@ lz4_filter_read_default_stream(struct archive_read_filter *self, const void **p) if (state->stage == SELECT_STREAM) { state->stage = READ_DEFAULT_STREAM; - /* First, read a desciprtor. */ + /* First, read a descriptor. */ if((ret = lz4_filter_read_descriptor(self)) != ARCHIVE_OK) return (ret); state->stage = READ_DEFAULT_BLOCK; diff --git a/libarchive/archive_read_support_filter_lzop.c b/libarchive/archive_read_support_filter_lzop.c index 44ac9964ae11..a1c392f4f364 100644 --- a/libarchive/archive_read_support_filter_lzop.c +++ b/libarchive/archive_read_support_filter_lzop.c @@ -436,7 +436,7 @@ lzop_filter_read(struct archive_read_filter *self, const void **p) } /* - * Drive lzo uncompresison. + * Drive lzo uncompression. */ out_size = (lzo_uint)state->uncompressed_size; r = lzo1x_decompress_safe(b, (lzo_uint)state->compressed_size, diff --git a/libarchive/archive_read_support_format_7zip.c b/libarchive/archive_read_support_format_7zip.c index f8d52fb71797..3387eaf7e7f5 100644 --- a/libarchive/archive_read_support_format_7zip.c +++ b/libarchive/archive_read_support_format_7zip.c @@ -552,7 +552,7 @@ skip_sfx(struct archive_read *a, ssize_t bytes_avail) /* * If bytes_avail > SFX_MIN_ADDR we do not have to call * __archive_read_seek() at this time since we have - * alredy had enough data. + * already had enough data. */ if (bytes_avail > SFX_MIN_ADDR) __archive_read_consume(a, SFX_MIN_ADDR); @@ -760,7 +760,7 @@ archive_read_format_7zip_read_header(struct archive_read *a, symsize += size; } if (symsize == 0) { - /* If there is no synname, handle it as a regular + /* If there is no symname, handle it as a regular * file. */ zip_entry->mode &= ~AE_IFMT; zip_entry->mode |= AE_IFREG; @@ -3288,7 +3288,7 @@ read_stream(struct archive_read *a, const void **buff, size_t size, return (r); /* - * Skip the bytes we alrady has skipped in skip_stream(). + * Skip the bytes we already has skipped in skip_stream(). */ while (skip_bytes) { ssize_t skipped; @@ -3506,7 +3506,7 @@ setup_decode_folder(struct archive_read *a, struct _7z_folder *folder, return (ARCHIVE_FATAL); } - /* Allocate memory for the decorded data of a sub + /* Allocate memory for the decoded data of a sub * stream. */ b[i] = malloc((size_t)zip->folder_outbytes_remaining); if (b[i] == NULL) { @@ -3591,7 +3591,7 @@ skip_stream(struct archive_read *a, size_t skip_bytes) if (zip->folder_index == 0) { /* * Optimization for a list mode. - * Avoid unncecessary decoding operations. + * Avoid unnecessary decoding operations. */ zip->si.ci.folders[zip->entry->folderIndex].skipped_bytes += skip_bytes; diff --git a/libarchive/archive_read_support_format_iso9660.c b/libarchive/archive_read_support_format_iso9660.c index 476490dc00c9..549aa835bc00 100644 --- a/libarchive/archive_read_support_format_iso9660.c +++ b/libarchive/archive_read_support_format_iso9660.c @@ -322,7 +322,7 @@ struct iso9660 { struct archive_string pathname; char seenRockridge; /* Set true if RR extensions are used. */ - char seenSUSP; /* Set true if SUSP is beging used. */ + char seenSUSP; /* Set true if SUSP is being used. */ char seenJoliet; unsigned char suspOffset; diff --git a/libarchive/archive_read_support_format_lha.c b/libarchive/archive_read_support_format_lha.c index a7f1d8d949f5..52a5531b0ff9 100644 --- a/libarchive/archive_read_support_format_lha.c +++ b/libarchive/archive_read_support_format_lha.c @@ -1711,7 +1711,7 @@ lha_crc16(uint16_t crc, const void *pp, size_t len) */ for (;len >= 8; len -= 8) { /* This if statement expects compiler optimization will - * remove the stament which will not be executed. */ + * remove the statement which will not be executed. */ #undef bswap16 #if defined(_MSC_VER) && _MSC_VER >= 1400 /* Visual Studio */ # define bswap16(x) _byteswap_ushort(x) diff --git a/libarchive/archive_read_support_format_rar.c b/libarchive/archive_read_support_format_rar.c index 9c9f6f12ec01..1e9849fdd629 100644 --- a/libarchive/archive_read_support_format_rar.c +++ b/libarchive/archive_read_support_format_rar.c @@ -906,7 +906,7 @@ archive_read_format_rar_read_header(struct archive_read *a, sizeof(rar->reserved2)); } - /* Main header is password encrytped, so we cannot read any + /* Main header is password encrypted, so we cannot read any file names or any other info about files from the header. */ if (rar->main_flags & MHD_PASSWORD) { diff --git a/libarchive/archive_read_support_format_tar.c b/libarchive/archive_read_support_format_tar.c index b18edcc12029..33732530f8b9 100644 --- a/libarchive/archive_read_support_format_tar.c +++ b/libarchive/archive_read_support_format_tar.c @@ -204,13 +204,14 @@ static int archive_read_format_tar_read_header(struct archive_read *, struct archive_entry *); static int checksum(struct archive_read *, const void *); static int pax_attribute(struct archive_read *, struct tar *, - struct archive_entry *, const char *key, const char *value); + struct archive_entry *, const char *key, const char *value, + size_t value_length); static int pax_attribute_acl(struct archive_read *, struct tar *, struct archive_entry *, const char *, int); static int pax_attribute_xattr(struct archive_entry *, const char *, const char *); static int pax_header(struct archive_read *, struct tar *, - struct archive_entry *, char *attr); + struct archive_entry *, struct archive_string *); static void pax_time(const char *, int64_t *sec, long *nanos); static ssize_t readline(struct archive_read *, struct tar *, const char **, ssize_t limit, size_t *); @@ -1483,7 +1484,7 @@ header_pax_extensions(struct archive_read *a, struct tar *tar, * and then skip any fields in the standard header that were * defined in the pax header. */ - err2 = pax_header(a, tar, entry, tar->pax_header.s); + err2 = pax_header(a, tar, entry, &tar->pax_header); err = err_combine(err, err2); tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); return (err); @@ -1564,16 +1565,17 @@ header_ustar(struct archive_read *a, struct tar *tar, */ static int pax_header(struct archive_read *a, struct tar *tar, - struct archive_entry *entry, char *attr) + struct archive_entry *entry, struct archive_string *in_as) { - size_t attr_length, l, line_length; + size_t attr_length, l, line_length, value_length; char *p; char *key, *value; struct archive_string *as; struct archive_string_conv *sconv; int err, err2; + char *attr = in_as->s; - attr_length = strlen(attr); + attr_length = in_as->length; tar->pax_hdrcharset_binary = 0; archive_string_empty(&(tar->entry_gname)); archive_string_empty(&(tar->entry_linkpath)); @@ -1638,11 +1640,13 @@ pax_header(struct archive_read *a, struct tar *tar, } *p = '\0'; - /* Identify null-terminated 'value' portion. */ value = p + 1; + /* Some values may be binary data */ + value_length = attr + line_length - 1 - value; + /* Identify this attribute and set it in the entry. */ - err2 = pax_attribute(a, tar, entry, key, value); + err2 = pax_attribute(a, tar, entry, key, value, value_length); if (err2 == ARCHIVE_FATAL) return (err2); err = err_combine(err, err2); @@ -1763,6 +1767,20 @@ pax_attribute_xattr(struct archive_entry *entry, return 0; } +static int +pax_attribute_schily_xattr(struct archive_entry *entry, + const char *name, const char *value, size_t value_length) +{ + if (strlen(name) < 14 || (memcmp(name, "SCHILY.xattr.", 13)) != 0) + return 1; + + name += 13; + + archive_entry_xattr_add_entry(entry, name, value, value_length); + + return 0; +} + static int pax_attribute_acl(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const char *value, int type) @@ -1824,7 +1842,7 @@ pax_attribute_acl(struct archive_read *a, struct tar *tar, */ static int pax_attribute(struct archive_read *a, struct tar *tar, - struct archive_entry *entry, const char *key, const char *value) + struct archive_entry *entry, const char *key, const char *value, size_t value_length) { int64_t s; long n; @@ -1959,6 +1977,9 @@ pax_attribute(struct archive_read *a, struct tar *tar, } else if (strcmp(key, "SCHILY.realsize") == 0) { tar->realsize = tar_atol10(value, strlen(value)); archive_entry_set_size(entry, tar->realsize); + } else if (strncmp(key, "SCHILY.xattr.", 13) == 0) { + pax_attribute_schily_xattr(entry, key, value, + value_length); } else if (strcmp(key, "SUN.holesdata") == 0) { /* A Solaris extension for sparse. */ r = solaris_sparse_parse(a, tar, entry, value); diff --git a/libarchive/archive_read_support_format_warc.c b/libarchive/archive_read_support_format_warc.c index a287fc2a06b5..8eda519eb149 100644 --- a/libarchive/archive_read_support_format_warc.c +++ b/libarchive/archive_read_support_format_warc.c @@ -88,7 +88,7 @@ typedef enum { WT_RVIS, /* conversion, unsupported */ WT_CONV, - /* continutation, unsupported at the moment */ + /* continuation, unsupported at the moment */ WT_CONT, /* invalid type */ LAST_WT @@ -562,7 +562,7 @@ xstrpisotime(const char *s, char **endptr) goto out; } - /* massage TM to fulfill some of POSIX' contraints */ + /* massage TM to fulfill some of POSIX' constraints */ tm.tm_year -= 1900; tm.tm_mon--; diff --git a/libarchive/archive_read_support_format_zip.c b/libarchive/archive_read_support_format_zip.c index d19e7914e11c..cde62ee62b9d 100644 --- a/libarchive/archive_read_support_format_zip.c +++ b/libarchive/archive_read_support_format_zip.c @@ -199,7 +199,7 @@ struct zip { struct trad_enc_ctx tctx; char tctx_valid; - /* WinZip AES decyption. */ + /* WinZip AES decryption. */ /* Contexts used for AES decryption. */ archive_crypto_ctx cctx; char cctx_valid; @@ -242,7 +242,7 @@ trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) } static uint8_t -trad_enc_decypt_byte(struct trad_enc_ctx *ctx) +trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) { unsigned temp = ctx->keys[2] | 2; return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; @@ -257,7 +257,7 @@ trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, max = (unsigned)((in_len < out_len)? in_len: out_len); for (i = 0; i < max; i++) { - uint8_t t = in[i] ^ trad_enc_decypt_byte(ctx); + uint8_t t = in[i] ^ trad_enc_decrypt_byte(ctx); out[i] = t; trad_enc_update_keys(ctx, t); } @@ -710,7 +710,7 @@ process_extra(struct archive_read *a, const char *p, size_t extra_length, struct break; } case 0x9901: - /* WinZIp AES extra data field. */ + /* WinZip AES extra data field. */ if (p[offset + 2] == 'A' && p[offset + 3] == 'E') { /* Vendor version. */ zip_entry->aes_extra.vendor = @@ -1518,7 +1518,7 @@ read_decryption_header(struct archive_read *a) case 0x6720:/* Blowfish */ case 0x6721:/* Twofish */ case 0x6801:/* RC4 */ - /* Suuported encryption algorithm. */ + /* Supported encryption algorithm. */ break; default: archive_set_error(&a->archive, @@ -1627,7 +1627,7 @@ read_decryption_header(struct archive_read *a) __archive_read_consume(a, 4); /*return (ARCHIVE_OK); - * This is not fully implemnted yet.*/ + * This is not fully implemented yet.*/ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Encrypted file is unsupported"); return (ARCHIVE_FAILED); @@ -1709,7 +1709,7 @@ init_traditional_PKWARE_decryption(struct archive_read *a) } /* - * Initialize ctx for Traditional PKWARE Decyption. + * Initialize ctx for Traditional PKWARE Decryption. */ r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), p, ENC_HEADER_SIZE, &crcchk); diff --git a/libarchive/archive_string.c b/libarchive/archive_string.c index bbb1e4585180..592ead2bdd58 100644 --- a/libarchive/archive_string.c +++ b/libarchive/archive_string.c @@ -219,6 +219,12 @@ archive_wstring_append(struct archive_wstring *as, const wchar_t *p, size_t s) return (as); } +struct archive_string * +archive_array_append(struct archive_string *as, const char *p, size_t s) +{ + return archive_string_append(as, p, s); +} + void archive_string_concat(struct archive_string *dest, struct archive_string *src) { @@ -597,7 +603,7 @@ archive_wstring_append_from_mbs(struct archive_wstring *dest, wcs = dest->s + dest->length; /* * We cannot use mbsrtowcs/mbstowcs here because those may convert - * extra MBS when strlen(p) > len and one wide character consis of + * extra MBS when strlen(p) > len and one wide character consists of * multi bytes. */ while (*mbs && mbs_length > 0) { @@ -1248,7 +1254,7 @@ create_sconv_object(const char *fc, const char *tc, sc->cd = iconv_open(tc, fc); if (sc->cd == (iconv_t)-1 && (sc->flag & SCONV_BEST_EFFORT)) { /* - * Unfortunaly, all of iconv implements do support + * Unfortunately, all of iconv implements do support * "CP932" character-set, so we should use "SJIS" * instead if iconv_open failed. */ @@ -1261,7 +1267,7 @@ create_sconv_object(const char *fc, const char *tc, /* * archive_mstring on Windows directly convert multi-bytes * into archive_wstring in order not to depend on locale - * so that you can do a I18N programing. This will be + * so that you can do a I18N programming. This will be * used only in archive_mstring_copy_mbs_len_l so far. */ if (flag & SCONV_FROM_CHARSET) { @@ -1726,7 +1732,7 @@ archive_string_conversion_from_charset(struct archive *a, const char *charset, * in tar or zip files. But mbstowcs/wcstombs(CRT) usually use CP_ACP * unless you use setlocale(LC_ALL, ".OCP")(specify CP_OEMCP). * So we should make a string conversion between CP_ACP and CP_OEMCP - * for compatibillty. + * for compatibility. */ #if defined(_WIN32) && !defined(__CYGWIN__) struct archive_string_conv * @@ -2220,7 +2226,7 @@ best_effort_strncat_in_locale(struct archive_string *as, const void *_p, /* * If a character is ASCII, this just copies it. If not, this - * assigns '?' charater instead but in UTF-8 locale this assigns + * assigns '?' character instead but in UTF-8 locale this assigns * byte sequence 0xEF 0xBD 0xBD, which are code point U+FFFD, * a Replacement Character in Unicode. */ @@ -2554,7 +2560,7 @@ utf16_to_unicode(uint32_t *pwc, const char *s, size_t n, int be) /* * Surrogate pair values(0xd800 through 0xdfff) are only - * used by UTF-16, so, after above culculation, the code + * used by UTF-16, so, after above calculation, the code * must not be surrogate values, and Unicode has no codes * larger than 0x10ffff. Thus, those are not legal Unicode * values. @@ -2903,7 +2909,7 @@ get_nfc(uint32_t uc, uint32_t uc2) /* * Normalize UTF-8/UTF-16BE characters to Form C and copy the result. * - * TODO: Convert composition exclusions,which are never converted + * TODO: Convert composition exclusions, which are never converted * from NFC,NFD,NFKC and NFKD, to Form C. */ static int @@ -3437,7 +3443,7 @@ strncat_from_utf8_libarchive2(struct archive_string *as, } /* - * As libarchie 2.x, translates the UTF-8 characters into + * As libarchive 2.x, translates the UTF-8 characters into * wide-characters in the assumption that WCS is Unicode. */ if (n < 0) { @@ -3947,7 +3953,7 @@ archive_mstring_get_mbs_l(struct archive_mstring *aes, #if defined(_WIN32) && !defined(__CYGWIN__) /* - * Internationalization programing on Windows must use Wide + * Internationalization programming on Windows must use Wide * characters because Windows platform cannot make locale UTF-8. */ if (sc != NULL && (aes->aes_set & AES_SET_WCS) != 0) { @@ -4079,7 +4085,7 @@ archive_mstring_copy_mbs_len_l(struct archive_mstring *aes, archive_string_empty(&(aes->aes_utf8)); #if defined(_WIN32) && !defined(__CYGWIN__) /* - * Internationalization programing on Windows must use Wide + * Internationalization programming on Windows must use Wide * characters because Windows platform cannot make locale UTF-8. */ if (sc == NULL) { diff --git a/libarchive/archive_string.h b/libarchive/archive_string.h index bac22e6d2945..56dfbb28f287 100644 --- a/libarchive/archive_string.h +++ b/libarchive/archive_string.h @@ -81,6 +81,10 @@ archive_strappend_char(struct archive_string *, char); struct archive_wstring * archive_wstrappend_wchar(struct archive_wstring *, wchar_t); +/* Append a raw array to an archive_string, resizing as necessary */ +struct archive_string * +archive_array_append(struct archive_string *, const char *, size_t); + /* Convert a Unicode string to current locale and append the result. */ /* Returns -1 if conversion fails. */ int diff --git a/libarchive/archive_string_composition.h b/libarchive/archive_string_composition.h index be41e3365124..8902ac1f7f30 100644 --- a/libarchive/archive_string_composition.h +++ b/libarchive/archive_string_composition.h @@ -1009,7 +1009,7 @@ static const char u_decomposable_blocks[0x1D2+1] = { (((uc) > 0x1D244)?0:\ ccc_val[ccc_val_index[ccc_index[(uc)>>8]][((uc)>>4)&0x0F]][(uc)&0x0F]) -/* The table of the value of Canonical Cimbining Class */ +/* The table of the value of Canonical Combining Class */ static const unsigned char ccc_val[][16] = { /* idx=0: XXXX0 - XXXXF */ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, diff --git a/libarchive/archive_write.c b/libarchive/archive_write.c index 0697db1dd919..0634a22967f8 100644 --- a/libarchive/archive_write.c +++ b/libarchive/archive_write.c @@ -231,7 +231,7 @@ __archive_write_filter(struct archive_write_filter *f, if (length == 0) return(ARCHIVE_OK); if (f->write == NULL) - /* If unset, a fatal error has already ocuured, so this filter + /* If unset, a fatal error has already occurred, so this filter * didn't open. We cannot write anything. */ return(ARCHIVE_FATAL); r = (f->write)(f, buff, length); diff --git a/libarchive/archive_write_add_filter_xz.c b/libarchive/archive_write_add_filter_xz.c index 46a6c38aa6e1..b0f25a6ef0ed 100644 --- a/libarchive/archive_write_add_filter_xz.c +++ b/libarchive/archive_write_add_filter_xz.c @@ -233,7 +233,7 @@ archive_compressor_xz_init_stream(struct archive_write_filter *f, if (f->code == ARCHIVE_FILTER_XZ) { #ifdef HAVE_LZMA_STREAM_ENCODER_MT if (data->threads != 1) { - bzero(&mt_options, sizeof(mt_options)); + memset(&mt_options, 0, sizeof(mt_options)); mt_options.threads = data->threads; mt_options.timeout = 300; mt_options.filters = data->lzmafilters; diff --git a/libarchive/archive_write_disk_acl.c b/libarchive/archive_write_disk_acl.c index e47384a6efaf..706ab62a07c8 100644 --- a/libarchive/archive_write_disk_acl.c +++ b/libarchive/archive_write_disk_acl.c @@ -124,7 +124,10 @@ static struct { {ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, ACL_ENTRY_FILE_INHERIT}, {ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT, ACL_ENTRY_DIRECTORY_INHERIT}, {ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT, ACL_ENTRY_NO_PROPAGATE_INHERIT}, - {ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, ACL_ENTRY_INHERIT_ONLY} + {ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, ACL_ENTRY_INHERIT_ONLY}, + {ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS, ACL_ENTRY_SUCCESSFUL_ACCESS}, + {ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS, ACL_ENTRY_FAILED_ACCESS}, + {ARCHIVE_ENTRY_ACL_ENTRY_INHERITED, ACL_ENTRY_INHERITED} }; #endif @@ -292,29 +295,41 @@ set_acl(struct archive *a, int fd, const char *name, } /* Try restoring the ACL through 'fd' if we can. */ -#if HAVE_ACL_SET_FD - if (fd >= 0 && acl_type == ACL_TYPE_ACCESS && acl_set_fd(fd, acl) == 0) - ret = ARCHIVE_OK; - else -#else +#if HAVE_ACL_SET_FD_NP || HAVE_ACL_SET_FD #if HAVE_ACL_SET_FD_NP - if (fd >= 0 && acl_set_fd_np(fd, acl, acl_type) == 0) - ret = ARCHIVE_OK; - else -#endif + if (fd >= 0) { + if (acl_set_fd_np(fd, acl, acl_type) == 0) +#else /* HAVE_ACL_SET_FD */ + if (fd >= 0 && acl_type == ACL_TYPE_ACCESS) { + if (acl_set_fd(fd, acl) == 0) #endif + ret = ARCHIVE_OK; + else { + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, + "Failed to set %s acl on fd", tname); + } + } + } else +#endif /* HAVE_ACL_SET_FD_NP || HAVE_ACL_SET_FD */ #if HAVE_ACL_SET_LINK_NP - if (acl_set_link_np(name, acl_type, acl) != 0) { - archive_set_error(a, errno, "Failed to set %s acl", tname); - ret = ARCHIVE_WARN; - } + if (acl_set_link_np(name, acl_type, acl) != 0) { #else /* TODO: Skip this if 'name' is a symlink. */ if (acl_set_file(name, acl_type, acl) != 0) { - archive_set_error(a, errno, "Failed to set %s acl", tname); - ret = ARCHIVE_WARN; - } #endif + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, "Failed to set %s acl", + tname); + ret = ARCHIVE_WARN; + } + } exit_free: acl_free(acl); return (ret); diff --git a/libarchive/archive_write_set_format_7zip.c b/libarchive/archive_write_set_format_7zip.c index abd521a4641c..41ed74daf1b9 100644 --- a/libarchive/archive_write_set_format_7zip.c +++ b/libarchive/archive_write_set_format_7zip.c @@ -1358,7 +1358,7 @@ make_header(struct archive_write *a, uint64_t offset, uint64_t pack_size, if (r < 0) return (r); - /* Write Nume size. */ + /* Write Name size. */ r = enc_uint64(a, zip->total_bytes_entry_name+1); if (r < 0) return (r); diff --git a/libarchive/archive_write_set_format_pax.c b/libarchive/archive_write_set_format_pax.c index 0063d23e1d2a..cb0e5c31598c 100644 --- a/libarchive/archive_write_set_format_pax.c +++ b/libarchive/archive_write_set_format_pax.c @@ -62,10 +62,17 @@ struct pax { struct sparse_block *sparse_tail; struct archive_string_conv *sconv_utf8; int opt_binary; + + unsigned flags; +#define WRITE_SCHILY_XATTR (1 << 0) +#define WRITE_LIBARCHIVE_XATTR (1 << 1) }; static void add_pax_attr(struct archive_string *, const char *key, const char *value); +static void add_pax_attr_binary(struct archive_string *, + const char *key, + const char *value, size_t value_len); static void add_pax_attr_int(struct archive_string *, const char *key, int64_t value); static void add_pax_attr_time(struct archive_string *, @@ -136,6 +143,8 @@ archive_write_set_format_pax(struct archive *_a) "Can't allocate pax data"); return (ARCHIVE_FATAL); } + pax->flags = WRITE_LIBARCHIVE_XATTR | WRITE_SCHILY_XATTR; + a->format_data = pax; a->format_name = "pax"; a->format_options = archive_write_pax_options; @@ -274,6 +283,17 @@ add_pax_attr_int(struct archive_string *as, const char *key, int64_t value) */ static void add_pax_attr(struct archive_string *as, const char *key, const char *value) +{ + add_pax_attr_binary(as, key, value, strlen(value)); +} + +/* + * Add a key/value attribute to the pax header. This function handles + * binary values. + */ +static void +add_pax_attr_binary(struct archive_string *as, const char *key, + const char *value, size_t value_len) { int digits, i, len, next_ten; char tmp[1 + 3 * sizeof(int)]; /* < 3 base-10 digits per byte */ @@ -282,7 +302,7 @@ add_pax_attr(struct archive_string *as, const char *key, const char *value) * PAX attributes have the following layout: * <=> */ - len = 1 + (int)strlen(key) + 1 + (int)strlen(value) + 1; + len = 1 + (int)strlen(key) + 1 + (int)value_len + 1; /* * The field includes the length of the field, so @@ -313,21 +333,47 @@ add_pax_attr(struct archive_string *as, const char *key, const char *value) archive_strappend_char(as, ' '); archive_strcat(as, key); archive_strappend_char(as, '='); - archive_strcat(as, value); + archive_array_append(as, value, value_len); archive_strappend_char(as, '\n'); } +static void +archive_write_pax_header_xattr(struct pax *pax, const char *encoded_name, + const void *value, size_t value_len) +{ + struct archive_string s; + char *encoded_value; + + if (pax->flags & WRITE_LIBARCHIVE_XATTR) { + encoded_value = base64_encode((const char *)value, value_len); + + if (encoded_name != NULL && encoded_value != NULL) { + archive_string_init(&s); + archive_strcpy(&s, "LIBARCHIVE.xattr."); + archive_strcat(&s, encoded_name); + add_pax_attr(&(pax->pax_header), s.s, encoded_value); + archive_string_free(&s); + } + free(encoded_value); + } + if (pax->flags & WRITE_SCHILY_XATTR) { + archive_string_init(&s); + archive_strcpy(&s, "SCHILY.xattr."); + archive_strcat(&s, encoded_name); + add_pax_attr_binary(&(pax->pax_header), s.s, value, value_len); + archive_string_free(&s); + } +} + static int archive_write_pax_header_xattrs(struct archive_write *a, struct pax *pax, struct archive_entry *entry) { - struct archive_string s; int i = archive_entry_xattr_reset(entry); while (i--) { const char *name; const void *value; - char *encoded_value; char *url_encoded_name = NULL, *encoded_name = NULL; size_t size; int r; @@ -348,16 +394,9 @@ archive_write_pax_header_xattrs(struct archive_write *a, } } - encoded_value = base64_encode((const char *)value, size); + archive_write_pax_header_xattr(pax, encoded_name, + value, size); - if (encoded_name != NULL && encoded_value != NULL) { - archive_string_init(&s); - archive_strcpy(&s, "LIBARCHIVE.xattr."); - archive_strcat(&s, encoded_name); - add_pax_attr(&(pax->pax_header), s.s, encoded_value); - archive_string_free(&s); - } - free(encoded_value); } return (ARCHIVE_OK); } diff --git a/libarchive/archive_write_set_format_xar.c b/libarchive/archive_write_set_format_xar.c index ef9ad96ad51f..c9f5ac239582 100644 --- a/libarchive/archive_write_set_format_xar.c +++ b/libarchive/archive_write_set_format_xar.c @@ -2913,7 +2913,7 @@ compression_init_encoder_xz(struct archive *a, *strm = lzma_init_data; #ifdef HAVE_LZMA_STREAM_ENCODER_MT if (threads > 1) { - bzero(&mt_options, sizeof(mt_options)); + memset(&mt_options, 0, sizeof(mt_options)); mt_options.threads = threads; mt_options.timeout = 300; mt_options.filters = lzmafilters; diff --git a/libarchive/archive_write_set_format_zip.c b/libarchive/archive_write_set_format_zip.c index 810b8d7ffd70..f69b8467f440 100644 --- a/libarchive/archive_write_set_format_zip.c +++ b/libarchive/archive_write_set_format_zip.c @@ -878,7 +878,7 @@ archive_write_zip_header(struct archive_write *a, struct archive_entry *entry) || zip->entry_encryption == ENCRYPTION_WINZIP_AES256)) { memcpy(e, "\001\231\007\000\001\000AE", 8); - /* AES vendoer version AE-2 does not store a CRC. + /* AES vendor version AE-2 does not store a CRC. * WinZip 11 uses AE-1, which does store the CRC, * but it does not store the CRC when the file size * is less than 20 bytes. So we simulate what @@ -1013,7 +1013,7 @@ archive_write_zip_data(struct archive_write *a, const void *buff, size_t s) if (zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) { switch (zip->entry_encryption) { case ENCRYPTION_TRADITIONAL: - /* Initialize traditoinal PKWARE encryption context. */ + /* Initialize traditional PKWARE encryption context. */ if (!zip->tctx_valid) { ret = init_traditional_pkware_encryption(a); if (ret != ARCHIVE_OK) @@ -1499,7 +1499,7 @@ trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) } static uint8_t -trad_enc_decypt_byte(struct trad_enc_ctx *ctx) +trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) { unsigned temp = ctx->keys[2] | 2; return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; @@ -1515,7 +1515,7 @@ trad_enc_encrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, for (i = 0; i < max; i++) { uint8_t t = in[i]; - out[i] = t ^ trad_enc_decypt_byte(ctx); + out[i] = t ^ trad_enc_decrypt_byte(ctx); trad_enc_update_keys(ctx, t); } return i; @@ -1626,7 +1626,7 @@ init_winzip_aes_encryption(struct archive_write *a) return (ARCHIVE_FAILED); } - /* Set a passowrd verification value after the 'salt'. */ + /* Set a password verification value after the 'salt'. */ salt[salt_len] = derived_key[key_len * 2]; salt[salt_len + 1] = derived_key[key_len * 2 + 1]; diff --git a/libarchive/test/CMakeLists.txt b/libarchive/test/CMakeLists.txt index 2f0e9e7b3d23..06b27638e9de 100644 --- a/libarchive/test/CMakeLists.txt +++ b/libarchive/test/CMakeLists.txt @@ -183,6 +183,7 @@ IF(ENABLE_TEST) test_read_format_zip_winzip_aes_large.c test_read_format_zip_zip64.c test_read_large.c + test_read_pax_schily_xattr.c test_read_pax_truncated.c test_read_position.c test_read_set_format.c diff --git a/libarchive/test/test_archive_read_add_passphrase.c b/libarchive/test/test_archive_read_add_passphrase.c index 68dec10d9a12..0ce5a76aedbc 100644 --- a/libarchive/test/test_archive_read_add_passphrase.c +++ b/libarchive/test/test_archive_read_add_passphrase.c @@ -191,7 +191,7 @@ DEFINE_TEST(test_archive_read_add_passphrase_set_callback3) /* Fist call, we should get "passCallBack" as a passphrase. */ assertEqualString("passCallBack", __archive_read_next_passphrase(ar)); __archive_read_reset_passphrase(ar); - /* After reset passphrase, we should get "passCallBack"passphrase. */ + /* After reset passphrase, we should get "passCallBack" passphrase. */ assertEqualString("passCallBack", __archive_read_next_passphrase(ar)); /* Second call, we should get NULL which means all the passphrases * are passed already. */ diff --git a/libarchive/test/test_compat_uudecode.c b/libarchive/test/test_compat_uudecode.c index 95b1c9a8b836..cfb17c85788d 100644 --- a/libarchive/test/test_compat_uudecode.c +++ b/libarchive/test/test_compat_uudecode.c @@ -40,7 +40,7 @@ static char archive_data[] = { }; /* - * Compatibility: uudecode command ignores junk data placed ater the "end" + * Compatibility: uudecode command ignores junk data placed after the "end" * marker. */ DEFINE_TEST(test_compat_uudecode) diff --git a/libarchive/test/test_read_format_cpio_afio.c b/libarchive/test/test_read_format_cpio_afio.c index 16065eb0a512..95d3171e72cc 100644 --- a/libarchive/test/test_read_format_cpio_afio.c +++ b/libarchive/test/test_read_format_cpio_afio.c @@ -27,7 +27,7 @@ __FBSDID("$FreeBSD$"); /* -ecute the following to rebuild the data for this program: +execute the following to rebuild the data for this program: tail -n +33 test_read_format_cpio_afio.c | /bin/sh # How to make a sample data. diff --git a/libarchive/test/test_read_format_zip_traditional_encryption_data.c b/libarchive/test/test_read_format_zip_traditional_encryption_data.c index 2700be15ac7c..305261567ec8 100644 --- a/libarchive/test/test_read_format_zip_traditional_encryption_data.c +++ b/libarchive/test/test_read_format_zip_traditional_encryption_data.c @@ -28,7 +28,7 @@ __FBSDID("$FreeBSD$"); DEFINE_TEST(test_read_format_zip_traditional_encryption_data) { - /* This file is password protected (Traditional PKWARE Enctypted). + /* This file is password protected (Traditional PKWARE Encrypted). The headers are NOT encrypted. Password is "12345678". */ const char *refname = "test_read_format_zip_traditional_encryption_data.zip"; @@ -36,7 +36,7 @@ DEFINE_TEST(test_read_format_zip_traditional_encryption_data) struct archive *a; char buff[512]; - /* Check if running system has cryptographic functionarity. */ + /* Check if running system has cryptographic functionality. */ assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_write_add_filter_none(a)); diff --git a/libarchive/test/test_read_format_zip_winzip_aes.c b/libarchive/test/test_read_format_zip_winzip_aes.c index 082337d7eabf..cc1e3110d65b 100644 --- a/libarchive/test/test_read_format_zip_winzip_aes.c +++ b/libarchive/test/test_read_format_zip_winzip_aes.c @@ -33,7 +33,7 @@ test_winzip_aes(const char *refname, int need_libz) struct archive *a; char buff[512]; - /* Check if running system has cryptographic functionarity. */ + /* Check if running system has cryptographic functionality. */ assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_write_add_filter_none(a)); diff --git a/libarchive/test/test_read_format_zip_winzip_aes_large.c b/libarchive/test/test_read_format_zip_winzip_aes_large.c index a40d5cf1329a..6c40ae76695a 100644 --- a/libarchive/test/test_read_format_zip_winzip_aes_large.c +++ b/libarchive/test/test_read_format_zip_winzip_aes_large.c @@ -34,7 +34,7 @@ DEFINE_TEST(test_read_format_zip_winzip_aes256_large) char buff[512]; - /* Check if running system has cryptographic functionarity. */ + /* Check if running system has cryptographic functionality. */ assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_write_add_filter_none(a)); diff --git a/libarchive/test/test_read_pax_schily_xattr.c b/libarchive/test/test_read_pax_schily_xattr.c new file mode 100644 index 000000000000..7554f6d5d12a --- /dev/null +++ b/libarchive/test/test_read_pax_schily_xattr.c @@ -0,0 +1,70 @@ +/*- + * Copyright (c) 2016 IBM Corporation + * Copyright (c) 2003-2007 Tim Kientzle + * + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * This test case's code has been derived from test_entry.c + */ +#include "test.h" + +DEFINE_TEST(test_schily_xattr_pax) +{ + struct archive *a; + struct archive_entry *ae; + const char *refname = "test_read_pax_schily_xattr.tar"; + const char *xname; /* For xattr tests. */ + const void *xval; /* For xattr tests. */ + size_t xsize; /* For xattr tests. */ + const char *string, *array; + + assert((a = archive_read_new()) != NULL); + assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); + + extract_reference_file(refname); + assertEqualIntA(a, ARCHIVE_OK, + archive_read_open_filename(a, refname, 10240)); + + assertEqualInt(ARCHIVE_OK, archive_read_next_header(a, &ae)); + assertEqualInt(2, archive_entry_xattr_count(ae)); + assertEqualInt(2, archive_entry_xattr_reset(ae)); + + assertEqualInt(0, archive_entry_xattr_next(ae, &xname, &xval, &xsize)); + assertEqualString(xname, "security.selinux"); + string = "system_u:object_r:unlabeled_t:s0"; + assertEqualString(xval, string); + /* the xattr's value also contains the terminating \0 */ + assertEqualInt((int)xsize, strlen(string) + 1); + + assertEqualInt(0, archive_entry_xattr_next(ae, &xname, &xval, &xsize)); + assertEqualString(xname, "security.ima"); + assertEqualInt((int)xsize, 265); + /* we only compare the first 12 bytes */ + array = "\x03\x02\x04\xb0\xe9\xd6\x79\x01\x00\x2b\xad\x1e"; + assertEqualMem(xval, array, 12); + + /* Close the archive. */ + assertEqualInt(ARCHIVE_OK, archive_read_close(a)); + assertEqualInt(ARCHIVE_OK, archive_read_free(a)); +} diff --git a/libarchive/test/test_read_pax_schily_xattr.tar.uu b/libarchive/test/test_read_pax_schily_xattr.tar.uu new file mode 100644 index 000000000000..52f7a8f0dc08 --- /dev/null +++ b/libarchive/test/test_read_pax_schily_xattr.tar.uu @@ -0,0 +1,231 @@ +begin 644 test_schily_xattr_pax.tar +M+B]087A(96%D97)S+C$U,C4O8V]N9F9I;&5S```````````````````````` +M```````````````````````````````````````````````````````````` +M`````````````#`P,#`V-#0`,#`P,#`P,``P,#`P,#`P`#`P,#`P,#`P-C0W +M`#$R-S$R,C$P-3`V`#`Q,C4V-@`@>``````````````````````````````` +M```````````````````````````````````````````````````````````` +M``````````````````````````````````````````!UW6X5O?Y6: +M9^':P2MZR[4)$@W?)B6GX0U@<,0M%6YNMO%OG+IS%/.< +M,"A(N&S.F9]=!*5=\).X."2$GUGJ,0C:@+G#$M_E8UQP,LU-G(8IKW^K^<8* +M*3_.N0'%8.^$8S$`D9XOF+DK<<)U34U'_"O5/22YS96QI;G5X +M/7-Y libarchive_test'. */ diff --git a/libarchive/xxhash.c b/libarchive/xxhash.c index d7f8e96de6fb..6f5ba52fac32 100644 --- a/libarchive/xxhash.c +++ b/libarchive/xxhash.c @@ -29,10 +29,11 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. You can contact the author at : - xxHash source repository : http://code.google.com/p/xxhash/ */ +#include "archive_platform.h" + #include #include -#include "archive_platform.h" #include "archive_xxhash.h" #ifdef HAVE_LIBLZ4 @@ -60,7 +61,7 @@ You can contact the author at : ** By default, xxHash library provides endian-independent Hash values, based on little-endian convention. ** Results are therefore identical for little-endian and big-endian CPU. ** This comes at a performance cost for big-endian CPU, since some swapping is required to emulate little-endian format. -** Should endian-independance be of no importance for your application, you may set the #define below to 1. +** Should endian-independence be of no importance for your application, you may set the #define below to 1. ** It will improve speed for Big-endian CPU. ** This option has no impact on Little_Endian CPU. */ From 093cf246190c8348e8c80232c23f9659a10e941a Mon Sep 17 00:00:00 2001 From: Sean Bruno Date: Tue, 10 Jan 2017 21:21:00 +0000 Subject: [PATCH 55/79] Set CFLAGS correctly for sys/modules/em Unbreak gcc sparc64 builds (or any gcc build that uses em(4)). Reported by: lidl@freebsd.org --- sys/modules/em/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sys/modules/em/Makefile b/sys/modules/em/Makefile index 7747d09f233b..ead132978052 100644 --- a/sys/modules/em/Makefile +++ b/sys/modules/em/Makefile @@ -17,7 +17,7 @@ PCIE_SHARED = e1000_80003es2lan.c e1000_ich8lan.c e1000_82571.c e1000_82575.c LEGACY_SHARED = e1000_82540.c e1000_82542.c e1000_82541.c e1000_82543.c -CFLAGS += -I${.CURDIR}/../../../dev/e1000 +CFLAGS += -I${.CURDIR}/../../dev/e1000 # DEVICE_POLLING for a non-interrupt-driven method #CFLAGS += -DDEVICE_POLLING From 90e17792c89650ad586ba763d15acaec74cb97b1 Mon Sep 17 00:00:00 2001 From: Mark Johnston Date: Tue, 10 Jan 2017 21:41:28 +0000 Subject: [PATCH 56/79] Do not set BIO_DONE if the BIO specifies a completion handler. biowait() will otherwise race with completions of such BIOs. In-tree code only calls biowait() on BIOs that do not specify a handler, so this change should not have any functional impact. Reviewed by: mav MFC after: 1 month Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D9070 --- sys/kern/vfs_bio.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/sys/kern/vfs_bio.c b/sys/kern/vfs_bio.c index b746eb93eb4d..c4498a191f55 100644 --- a/sys/kern/vfs_bio.c +++ b/sys/kern/vfs_bio.c @@ -3905,10 +3905,8 @@ biodone(struct bio *bp) bp->bio_flags |= BIO_DONE; wakeup(bp); mtx_unlock(mtxp); - } else { - bp->bio_flags |= BIO_DONE; + } else done(bp); - } } /* From bd7abab0c93ddff7c0a303edb9724ea2b77b61aa Mon Sep 17 00:00:00 2001 From: Mark Johnston Date: Tue, 10 Jan 2017 21:52:48 +0000 Subject: [PATCH 57/79] Coalesce TLB shootdowns of global PTEs in pmap_advise() on x86. We would previously invalidate such entries individually, resulting in more IPIs than necessary. Reviewed by: alc, kib MFC after: 3 weeks Differential Revision: https://reviews.freebsd.org/D9094 --- sys/amd64/amd64/pmap.c | 26 ++++++++++++++++++-------- sys/i386/i386/pmap.c | 25 +++++++++++++++++-------- 2 files changed, 35 insertions(+), 16 deletions(-) diff --git a/sys/amd64/amd64/pmap.c b/sys/amd64/amd64/pmap.c index 85202302c9d7..817cdbe97c9e 100644 --- a/sys/amd64/amd64/pmap.c +++ b/sys/amd64/amd64/pmap.c @@ -6010,7 +6010,7 @@ pmap_advise(pmap_t pmap, vm_offset_t sva, vm_offset_t eva, int advice) pdp_entry_t *pdpe; pd_entry_t oldpde, *pde; pt_entry_t *pte, PG_A, PG_G, PG_M, PG_RW, PG_V; - vm_offset_t va_next; + vm_offset_t va, va_next; vm_page_t m; boolean_t anychanged; @@ -6090,11 +6090,11 @@ pmap_advise(pmap_t pmap, vm_offset_t sva, vm_offset_t eva, int advice) } if (va_next > eva) va_next = eva; + va = va_next; for (pte = pmap_pde_to_pte(pde, sva); sva != va_next; pte++, sva += PAGE_SIZE) { - if ((*pte & (PG_MANAGED | PG_V)) != (PG_MANAGED | - PG_V)) - continue; + if ((*pte & (PG_MANAGED | PG_V)) != (PG_MANAGED | PG_V)) + goto maybe_invlrng; else if ((*pte & (PG_M | PG_RW)) == (PG_M | PG_RW)) { if (advice == MADV_DONTNEED) { /* @@ -6109,12 +6109,22 @@ pmap_advise(pmap_t pmap, vm_offset_t sva, vm_offset_t eva, int advice) } else if ((*pte & PG_A) != 0) atomic_clear_long(pte, PG_A); else - continue; - if ((*pte & PG_G) != 0) - pmap_invalidate_page(pmap, sva); - else + goto maybe_invlrng; + + if ((*pte & PG_G) != 0) { + if (va == va_next) + va = sva; + } else anychanged = TRUE; + continue; +maybe_invlrng: + if (va != va_next) { + pmap_invalidate_range(pmap, va, sva); + va = va_next; + } } + if (va != va_next) + pmap_invalidate_range(pmap, va, sva); } if (anychanged) pmap_invalidate_all(pmap); diff --git a/sys/i386/i386/pmap.c b/sys/i386/i386/pmap.c index c5296f69654a..b4eef419a698 100644 --- a/sys/i386/i386/pmap.c +++ b/sys/i386/i386/pmap.c @@ -4905,7 +4905,7 @@ pmap_advise(pmap_t pmap, vm_offset_t sva, vm_offset_t eva, int advice) { pd_entry_t oldpde, *pde; pt_entry_t *pte; - vm_offset_t pdnxt; + vm_offset_t va, pdnxt; vm_page_t m; boolean_t anychanged, pv_lists_locked; @@ -4966,11 +4966,11 @@ pmap_advise(pmap_t pmap, vm_offset_t sva, vm_offset_t eva, int advice) } if (pdnxt > eva) pdnxt = eva; + va = pdnxt; for (pte = pmap_pte_quick(pmap, sva); sva != pdnxt; pte++, sva += PAGE_SIZE) { - if ((*pte & (PG_MANAGED | PG_V)) != (PG_MANAGED | - PG_V)) - continue; + if ((*pte & (PG_MANAGED | PG_V)) != (PG_MANAGED | PG_V)) + goto maybe_invlrng; else if ((*pte & (PG_M | PG_RW)) == (PG_M | PG_RW)) { if (advice == MADV_DONTNEED) { /* @@ -4985,12 +4985,21 @@ pmap_advise(pmap_t pmap, vm_offset_t sva, vm_offset_t eva, int advice) } else if ((*pte & PG_A) != 0) atomic_clear_int((u_int *)pte, PG_A); else - continue; - if ((*pte & PG_G) != 0) - pmap_invalidate_page(pmap, sva); - else + goto maybe_invlrng; + if ((*pte & PG_G) != 0) { + if (va == pdnxt) + va = sva; + } else anychanged = TRUE; + continue; +maybe_invlrng: + if (va != pdnxt) { + pmap_invalidate_range(pmap, va, sva); + va = pdnxt; + } } + if (va != pdnxt) + pmap_invalidate_range(pmap, va, sva); } if (anychanged) pmap_invalidate_all(pmap); From 9520841213e9762cf6e483d0463ef0b9e106fa0a Mon Sep 17 00:00:00 2001 From: Martin Matuska Date: Tue, 10 Jan 2017 22:01:37 +0000 Subject: [PATCH 58/79] Build libarchive tests missing in r311899 MFC after: 1 week X-MFC with: r311899 --- lib/libarchive/tests/Makefile | 2 ++ 1 file changed, 2 insertions(+) diff --git a/lib/libarchive/tests/Makefile b/lib/libarchive/tests/Makefile index dbf34dd93dbe..1249458b1a20 100644 --- a/lib/libarchive/tests/Makefile +++ b/lib/libarchive/tests/Makefile @@ -196,6 +196,7 @@ TESTS_SRCS= \ test_read_format_zip_winzip_aes_large.c \ test_read_format_zip_zip64.c \ test_read_large.c \ + test_read_pax_schily_xattr.c \ test_read_pax_truncated.c \ test_read_position.c \ test_read_set_format.c \ @@ -549,6 +550,7 @@ ${PACKAGE}FILES+= test_read_large_splitted_rar_ab.uu ${PACKAGE}FILES+= test_read_large_splitted_rar_ac.uu ${PACKAGE}FILES+= test_read_large_splitted_rar_ad.uu ${PACKAGE}FILES+= test_read_large_splitted_rar_ae.uu +${PACKAGE}FILES+= test_read_pax_schily_xattr.tar.uu ${PACKAGE}FILES+= test_read_splitted_rar_aa.uu ${PACKAGE}FILES+= test_read_splitted_rar_ab.uu ${PACKAGE}FILES+= test_read_splitted_rar_ac.uu From e116c444a02313d5d113ef6c60fafeffb4773253 Mon Sep 17 00:00:00 2001 From: Conrad Meyer Date: Wed, 11 Jan 2017 00:02:51 +0000 Subject: [PATCH 59/79] Revert r311843, r311667 As jkim@ points out, it isn't needed. --- .../acpica/components/namespace/nsxfeval.c | 40 ++++--------------- .../dev/acpica/components/tables/tbxface.c | 40 +++---------------- sys/contrib/dev/acpica/include/acpixf.h | 16 -------- 3 files changed, 12 insertions(+), 84 deletions(-) diff --git a/sys/contrib/dev/acpica/components/namespace/nsxfeval.c b/sys/contrib/dev/acpica/components/namespace/nsxfeval.c index 4c50040156e1..6467d7e2f54f 100644 --- a/sys/contrib/dev/acpica/components/namespace/nsxfeval.c +++ b/sys/contrib/dev/acpica/components/namespace/nsxfeval.c @@ -1022,25 +1022,23 @@ ACPI_EXPORT_SYMBOL (AcpiDetachData) /******************************************************************************* * - * FUNCTION: AcpiGetDataFull + * FUNCTION: AcpiGetData * * PARAMETERS: ObjHandle - Namespace node - * Handle - Handler used in call to attach_data + * Handler - Handler used in call to AttachData * Data - Where the data is returned - * Callback - function to execute before returning * * RETURN: Status * - * DESCRIPTION: Retrieve data that was previously attached to a namespace node - * and execute a callback before returning. + * DESCRIPTION: Retrieve data that was previously attached to a namespace node. * ******************************************************************************/ + ACPI_STATUS -AcpiGetDataFull ( +AcpiGetData ( ACPI_HANDLE ObjHandle, ACPI_OBJECT_HANDLER Handler, - void **Data, - void (*Callback)(void *)) + void **Data) { ACPI_NAMESPACE_NODE *Node; ACPI_STATUS Status; @@ -1071,34 +1069,10 @@ AcpiGetDataFull ( } Status = AcpiNsGetAttachedData (Node, Handler, Data); - if (ACPI_SUCCESS(Status) && Callback) { - Callback(*Data); - } + UnlockAndExit: (void) AcpiUtReleaseMutex (ACPI_MTX_NAMESPACE); return (Status); } -ACPI_EXPORT_SYMBOL (AcpiGetDataFull) -/******************************************************************************* - * - * FUNCTION: AcpiGetData - * - * PARAMETERS: ObjHandle - Namespace node - * Handler - Handler used in call to AttachData - * Data - Where the data is returned - * - * RETURN: Status - * - * DESCRIPTION: Retrieve data that was previously attached to a namespace node. - * - ******************************************************************************/ -ACPI_STATUS -AcpiGetData ( - ACPI_HANDLE ObjHandle, - ACPI_OBJECT_HANDLER Handler, - void **Data) -{ - return (AcpiGetDataFull(ObjHandle, Handler, Data, NULL)); -} ACPI_EXPORT_SYMBOL (AcpiGetData) diff --git a/sys/contrib/dev/acpica/components/tables/tbxface.c b/sys/contrib/dev/acpica/components/tables/tbxface.c index 07e2fe846e61..6e231cd8fceb 100644 --- a/sys/contrib/dev/acpica/components/tables/tbxface.c +++ b/sys/contrib/dev/acpica/components/tables/tbxface.c @@ -314,12 +314,11 @@ ACPI_EXPORT_SYMBOL (AcpiGetTableHeader) /******************************************************************************* * - * FUNCTION: AcpiGetTableWithSize + * FUNCTION: AcpiGetTable * * PARAMETERS: Signature - ACPI signature of needed table * Instance - Which instance (for SSDTs) * OutTable - Where the pointer to the table is returned - * TblSize - Size of the table * * RETURN: Status and pointer to the requested table * @@ -334,11 +333,10 @@ ACPI_EXPORT_SYMBOL (AcpiGetTableHeader) ******************************************************************************/ ACPI_STATUS -AcpiGetTableWithSize ( +AcpiGetTable ( char *Signature, UINT32 Instance, - ACPI_TABLE_HEADER **OutTable, - ACPI_SIZE *TblSize) + ACPI_TABLE_HEADER **OutTable) { UINT32 i; UINT32 j; @@ -386,7 +384,7 @@ AcpiGetTableWithSize ( return (Status); } -ACPI_EXPORT_SYMBOL (AcpiGetTableWithSize) +ACPI_EXPORT_SYMBOL (AcpiGetTable) /******************************************************************************* @@ -436,38 +434,10 @@ AcpiPutTable ( (void) AcpiUtReleaseMutex (ACPI_MTX_TABLES); return_VOID; } + ACPI_EXPORT_SYMBOL (AcpiPutTable) -/******************************************************************************* - * - * FUNCTION: AcpiGetTable - * - * PARAMETERS: Signature - ACPI signature of needed table - * Instance - Which instance (for SSDTs) - * OutTable - Where the pointer to the table is returned - * - * RETURN: Status and pointer to the requested table - * - * DESCRIPTION: Finds and verifies an ACPI table. Table must be in the - * RSDT/XSDT. - * - ******************************************************************************/ - -ACPI_STATUS -AcpiGetTable ( - char *Signature, - UINT32 Instance, - ACPI_TABLE_HEADER **OutTable) -{ - ACPI_SIZE Size; - - return (AcpiGetTableWithSize(Signature, Instance, OutTable, &Size)); -} - -ACPI_EXPORT_SYMBOL (AcpiGetTable) - - /******************************************************************************* * * FUNCTION: AcpiGetTableByIndex diff --git a/sys/contrib/dev/acpica/include/acpixf.h b/sys/contrib/dev/acpica/include/acpixf.h index 7a0edc091ddd..7234ae8e331d 100644 --- a/sys/contrib/dev/acpica/include/acpixf.h +++ b/sys/contrib/dev/acpica/include/acpixf.h @@ -584,14 +584,6 @@ AcpiGetTableHeader ( UINT32 Instance, ACPI_TABLE_HEADER *OutTableHeader)) -ACPI_EXTERNAL_RETURN_STATUS ( -ACPI_STATUS -AcpiGetTableWithSize ( - ACPI_STRING Signature, - UINT32 Instance, - ACPI_TABLE_HEADER **OutTable, - ACPI_SIZE *TblSize)) - ACPI_EXTERNAL_RETURN_STATUS ( ACPI_STATUS AcpiGetTable ( @@ -678,14 +670,6 @@ AcpiGetData ( ACPI_OBJECT_HANDLER Handler, void **Data)) -ACPI_EXTERNAL_RETURN_STATUS ( -ACPI_STATUS -AcpiGetDataFull ( - ACPI_HANDLE Object, - ACPI_OBJECT_HANDLER Handler, - void **Data, - void (*Callback)(void *))) - ACPI_EXTERNAL_RETURN_STATUS ( ACPI_STATUS AcpiDebugTrace ( From 6a4b451a11263d63626eeb53b09df55bc1705454 Mon Sep 17 00:00:00 2001 From: Ian Lepore Date: Wed, 11 Jan 2017 00:14:47 +0000 Subject: [PATCH 60/79] Follow r311103: add "pool" to the keywords that rc.d/ntpdate examines to find a server address in ntp.conf. Submitted by: Ronald Klop Pointy hat to: ian --- etc/rc.d/ntpdate | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/etc/rc.d/ntpdate b/etc/rc.d/ntpdate index 8f5b273d0fa2..90f9589ab00f 100755 --- a/etc/rc.d/ntpdate +++ b/etc/rc.d/ntpdate @@ -20,7 +20,7 @@ ntpdate_start() if [ -z "$ntpdate_hosts" -a -f "$ntpdate_config" ]; then ntpdate_hosts=`awk ' /^server[ \t]*127.127/ {next} - /^(server|peer)/ { + /^(server|peer|pool)/ { if ($2 ~/^-/) {print $3} else {print $2}} ' < "$ntpdate_config"` From 4153c9b93241e7887f2b240c144605a5e5ad1da8 Mon Sep 17 00:00:00 2001 From: Mark Johnston Date: Wed, 11 Jan 2017 01:15:55 +0000 Subject: [PATCH 61/79] Ignore LC_SLEEPABLE when testing whether a mutex is adaptive. MFC after: 1 week --- sys/cddl/contrib/opensolaris/uts/common/dtrace/dtrace.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/sys/cddl/contrib/opensolaris/uts/common/dtrace/dtrace.c b/sys/cddl/contrib/opensolaris/uts/common/dtrace/dtrace.c index c47fb9b65398..3bfa8ec86974 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/dtrace/dtrace.c +++ b/sys/cddl/contrib/opensolaris/uts/common/dtrace/dtrace.c @@ -4356,9 +4356,7 @@ dtrace_dif_subr(uint_t subr, uint_t rd, uint64_t *regs, break; } l.lx = dtrace_loadptr((uintptr_t)&tupregs[0].dttk_value); - /* XXX - should be only LC_SLEEPABLE? */ - regs[rd] = (LOCK_CLASS(l.li)->lc_flags & - (LC_SLEEPLOCK | LC_SLEEPABLE)) != 0; + regs[rd] = (LOCK_CLASS(l.li)->lc_flags & LC_SLEEPLOCK) != 0; break; case DIF_SUBR_MUTEX_TYPE_SPIN: From d01e6ad41b27fefb9adac14315b978b9bff915ca Mon Sep 17 00:00:00 2001 From: Mark Johnston Date: Wed, 11 Jan 2017 01:18:06 +0000 Subject: [PATCH 62/79] Have DTrace handle faults when dereferencing a lock object pointer. MFC after: 1 week --- .../contrib/opensolaris/uts/common/dtrace/dtrace.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/sys/cddl/contrib/opensolaris/uts/common/dtrace/dtrace.c b/sys/cddl/contrib/opensolaris/uts/common/dtrace/dtrace.c index 3bfa8ec86974..ed0db960d16a 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/dtrace/dtrace.c +++ b/sys/cddl/contrib/opensolaris/uts/common/dtrace/dtrace.c @@ -4335,7 +4335,9 @@ dtrace_dif_subr(uint_t subr, uint_t rd, uint64_t *regs, break; } l.lx = dtrace_loadptr((uintptr_t)&tupregs[0].dttk_value); + DTRACE_CPUFLAG_SET(CPU_DTRACE_NOFAULT); regs[rd] = LOCK_CLASS(l.li)->lc_owner(l.li, &lowner); + DTRACE_CPUFLAG_CLEAR(CPU_DTRACE_NOFAULT); break; case DIF_SUBR_MUTEX_OWNER: @@ -4345,7 +4347,9 @@ dtrace_dif_subr(uint_t subr, uint_t rd, uint64_t *regs, break; } l.lx = dtrace_loadptr((uintptr_t)&tupregs[0].dttk_value); + DTRACE_CPUFLAG_SET(CPU_DTRACE_NOFAULT); LOCK_CLASS(l.li)->lc_owner(l.li, &lowner); + DTRACE_CPUFLAG_CLEAR(CPU_DTRACE_NOFAULT); regs[rd] = (uintptr_t)lowner; break; @@ -4356,7 +4360,9 @@ dtrace_dif_subr(uint_t subr, uint_t rd, uint64_t *regs, break; } l.lx = dtrace_loadptr((uintptr_t)&tupregs[0].dttk_value); + DTRACE_CPUFLAG_SET(CPU_DTRACE_NOFAULT); regs[rd] = (LOCK_CLASS(l.li)->lc_flags & LC_SLEEPLOCK) != 0; + DTRACE_CPUFLAG_CLEAR(CPU_DTRACE_NOFAULT); break; case DIF_SUBR_MUTEX_TYPE_SPIN: @@ -4366,7 +4372,9 @@ dtrace_dif_subr(uint_t subr, uint_t rd, uint64_t *regs, break; } l.lx = dtrace_loadptr((uintptr_t)&tupregs[0].dttk_value); + DTRACE_CPUFLAG_SET(CPU_DTRACE_NOFAULT); regs[rd] = (LOCK_CLASS(l.li)->lc_flags & LC_SPINLOCK) != 0; + DTRACE_CPUFLAG_CLEAR(CPU_DTRACE_NOFAULT); break; case DIF_SUBR_RW_READ_HELD: @@ -4377,8 +4385,10 @@ dtrace_dif_subr(uint_t subr, uint_t rd, uint64_t *regs, break; } l.lx = dtrace_loadptr((uintptr_t)&tupregs[0].dttk_value); + DTRACE_CPUFLAG_SET(CPU_DTRACE_NOFAULT); regs[rd] = LOCK_CLASS(l.li)->lc_owner(l.li, &lowner) && lowner == NULL; + DTRACE_CPUFLAG_CLEAR(CPU_DTRACE_NOFAULT); break; case DIF_SUBR_RW_WRITE_HELD: @@ -4389,8 +4399,10 @@ dtrace_dif_subr(uint_t subr, uint_t rd, uint64_t *regs, break; } l.lx = dtrace_loadptr(tupregs[0].dttk_value); + DTRACE_CPUFLAG_SET(CPU_DTRACE_NOFAULT); regs[rd] = LOCK_CLASS(l.li)->lc_owner(l.li, &lowner) && lowner != NULL; + DTRACE_CPUFLAG_CLEAR(CPU_DTRACE_NOFAULT); break; case DIF_SUBR_RW_ISWRITER: @@ -4401,7 +4413,9 @@ dtrace_dif_subr(uint_t subr, uint_t rd, uint64_t *regs, break; } l.lx = dtrace_loadptr(tupregs[0].dttk_value); + DTRACE_CPUFLAG_SET(CPU_DTRACE_NOFAULT); LOCK_CLASS(l.li)->lc_owner(l.li, &lowner); + DTRACE_CPUFLAG_CLEAR(CPU_DTRACE_NOFAULT); regs[rd] = (lowner == curthread); break; #endif /* illumos */ From e5d519fdbc4e3223bd02d1589042e92c1c9d8b09 Mon Sep 17 00:00:00 2001 From: Oleksandr Tymoshenko Date: Wed, 11 Jan 2017 01:53:54 +0000 Subject: [PATCH 63/79] [sdhci] Add ACPI platform support for SDHCI driver - Create ACPI version of SDHCI attach/detach/accessors logic. Some platforms (e.g. BayTrail-based Minnowboard) expose SDHCI devices via ACPI, not PCI - Add sdchi_acpi kernel module Reviewed by: ian, imp MFC after: 1 week Differential Revision: https://reviews.freebsd.org/D9112 --- sys/conf/files | 1 + sys/dev/sdhci/sdhci_acpi.c | 370 ++++++++++++++++++++++++++++++++ sys/modules/Makefile | 2 + sys/modules/sdhci_acpi/Makefile | 8 + 4 files changed, 381 insertions(+) create mode 100644 sys/dev/sdhci/sdhci_acpi.c create mode 100644 sys/modules/sdhci_acpi/Makefile diff --git a/sys/conf/files b/sys/conf/files index d9b7972be69f..c221bbec85bf 100644 --- a/sys/conf/files +++ b/sys/conf/files @@ -2826,6 +2826,7 @@ dev/scc/scc_dev_z8530.c optional scc dev/sdhci/sdhci.c optional sdhci dev/sdhci/sdhci_fdt_gpio.c optional sdhci fdt gpio dev/sdhci/sdhci_if.m optional sdhci +dev/sdhci/sdhci_acpi.c optional sdhci acpi dev/sdhci/sdhci_pci.c optional sdhci pci dev/sf/if_sf.c optional sf pci dev/sge/if_sge.c optional sge pci diff --git a/sys/dev/sdhci/sdhci_acpi.c b/sys/dev/sdhci/sdhci_acpi.c new file mode 100644 index 000000000000..a2a2cab0ffd2 --- /dev/null +++ b/sys/dev/sdhci/sdhci_acpi.c @@ -0,0 +1,370 @@ +/*- + * Copyright (c) 2017 Oleksandr Tymoshenko + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#include +__FBSDID("$FreeBSD$"); + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include + +#include +#include + +#include +#include +#include + +#include "sdhci.h" +#include "mmcbr_if.h" +#include "sdhci_if.h" + +static const struct sdhci_acpi_device { + const char* hid; + int uid; + const char *desc; + u_int quirks; +} sdhci_acpi_devices[] = { + { "80860F14", 1, "Intel Bay Trail eMMC 4.5 Controller", + SDHCI_QUIRK_ALL_SLOTS_NON_REMOVABLE | + SDHCI_QUIRK_INTEL_POWER_UP_RESET }, + { "80860F16", 0, "Intel Bay Trail SD Host Controller", + 0 }, + { NULL, 0, NULL, 0} +}; + +static char *sdhci_ids[] = { + "80860F14", + "80860F16", + NULL +}; + +struct sdhci_acpi_softc { + u_int quirks; /* Chip specific quirks */ + struct resource *irq_res; /* IRQ resource */ + void *intrhand; /* Interrupt handle */ + + struct sdhci_slot slot; + struct resource *mem_res; /* Memory resource */ +}; + +static void sdhci_acpi_intr(void *arg); +static int sdhci_acpi_detach(device_t dev); + +static uint8_t +sdhci_acpi_read_1(device_t dev, struct sdhci_slot *slot, bus_size_t off) +{ + struct sdhci_acpi_softc *sc = device_get_softc(dev); + + bus_barrier(sc->mem_res, 0, 0xFF, + BUS_SPACE_BARRIER_READ | BUS_SPACE_BARRIER_WRITE); + return bus_read_1(sc->mem_res, off); +} + +static void +sdhci_acpi_write_1(device_t dev, struct sdhci_slot *slot, bus_size_t off, uint8_t val) +{ + struct sdhci_acpi_softc *sc = device_get_softc(dev); + + bus_barrier(sc->mem_res, 0, 0xFF, + BUS_SPACE_BARRIER_READ | BUS_SPACE_BARRIER_WRITE); + bus_write_1(sc->mem_res, off, val); +} + +static uint16_t +sdhci_acpi_read_2(device_t dev, struct sdhci_slot *slot, bus_size_t off) +{ + struct sdhci_acpi_softc *sc = device_get_softc(dev); + + bus_barrier(sc->mem_res, 0, 0xFF, + BUS_SPACE_BARRIER_READ | BUS_SPACE_BARRIER_WRITE); + return bus_read_2(sc->mem_res, off); +} + +static void +sdhci_acpi_write_2(device_t dev, struct sdhci_slot *slot, bus_size_t off, uint16_t val) +{ + struct sdhci_acpi_softc *sc = device_get_softc(dev); + + bus_barrier(sc->mem_res, 0, 0xFF, + BUS_SPACE_BARRIER_READ | BUS_SPACE_BARRIER_WRITE); + bus_write_2(sc->mem_res, off, val); +} + +static uint32_t +sdhci_acpi_read_4(device_t dev, struct sdhci_slot *slot, bus_size_t off) +{ + struct sdhci_acpi_softc *sc = device_get_softc(dev); + + bus_barrier(sc->mem_res, 0, 0xFF, + BUS_SPACE_BARRIER_READ | BUS_SPACE_BARRIER_WRITE); + return bus_read_4(sc->mem_res, off); +} + +static void +sdhci_acpi_write_4(device_t dev, struct sdhci_slot *slot, bus_size_t off, uint32_t val) +{ + struct sdhci_acpi_softc *sc = device_get_softc(dev); + + bus_barrier(sc->mem_res, 0, 0xFF, + BUS_SPACE_BARRIER_READ | BUS_SPACE_BARRIER_WRITE); + bus_write_4(sc->mem_res, off, val); +} + +static void +sdhci_acpi_read_multi_4(device_t dev, struct sdhci_slot *slot, + bus_size_t off, uint32_t *data, bus_size_t count) +{ + struct sdhci_acpi_softc *sc = device_get_softc(dev); + + bus_read_multi_stream_4(sc->mem_res, off, data, count); +} + +static void +sdhci_acpi_write_multi_4(device_t dev, struct sdhci_slot *slot, + bus_size_t off, uint32_t *data, bus_size_t count) +{ + struct sdhci_acpi_softc *sc = device_get_softc(dev); + + bus_write_multi_stream_4(sc->mem_res, off, data, count); +} + +static const struct sdhci_acpi_device * +sdhci_acpi_find_device(device_t dev) +{ + const char *hid; + int i, uid; + ACPI_HANDLE handle; + ACPI_STATUS status; + + hid = ACPI_ID_PROBE(device_get_parent(dev), dev, sdhci_ids); + if (hid == NULL) + return (NULL); + + handle = acpi_get_handle(dev); + status = acpi_GetInteger(handle, "_UID", &uid); + if (ACPI_FAILURE(status)) + uid = 0; + + for (i = 0; sdhci_acpi_devices[i].hid != NULL; i++) { + if (strcmp(sdhci_acpi_devices[i].hid, hid) != 0) + continue; + if ((sdhci_acpi_devices[i].uid != 0) && + (sdhci_acpi_devices[i].uid != uid)) + continue; + return &sdhci_acpi_devices[i]; + } + + return (NULL); +} + +static int +sdhci_acpi_probe(device_t dev) +{ + const struct sdhci_acpi_device *acpi_dev; + + acpi_dev = sdhci_acpi_find_device(dev); + if (acpi_dev == NULL) + return (ENXIO); + + device_set_desc(dev, acpi_dev->desc); + + return (BUS_PROBE_DEFAULT); +} + +static int +sdhci_acpi_attach(device_t dev) +{ + struct sdhci_acpi_softc *sc = device_get_softc(dev); + int rid, err; + const struct sdhci_acpi_device *acpi_dev; + + acpi_dev = sdhci_acpi_find_device(dev); + if (acpi_dev == NULL) + return (ENXIO); + + sc->quirks = acpi_dev->quirks; + + /* Allocate IRQ. */ + rid = 0; + sc->irq_res = bus_alloc_resource_any(dev, SYS_RES_IRQ, &rid, + RF_ACTIVE); + if (sc->irq_res == NULL) { + device_printf(dev, "can't allocate IRQ\n"); + return (ENOMEM); + } + + rid = 0; + sc->mem_res = bus_alloc_resource_any(dev, SYS_RES_MEMORY, + &rid, RF_ACTIVE); + if (sc->mem_res == NULL) { + device_printf(dev, "can't allocate memory resource for slot\n"); + sdhci_acpi_detach(dev); + return (ENOMEM); + } + + sc->slot.quirks = sc->quirks; + + err = sdhci_init_slot(dev, &sc->slot, 0); + if (err) { + device_printf(dev, "failed to init slot\n"); + sdhci_acpi_detach(dev); + return (err); + } + + /* Activate the interrupt */ + err = bus_setup_intr(dev, sc->irq_res, INTR_TYPE_MISC | INTR_MPSAFE, + NULL, sdhci_acpi_intr, sc, &sc->intrhand); + if (err) { + device_printf(dev, "can't setup IRQ\n"); + sdhci_acpi_detach(dev); + return (err); + } + + /* Process cards detection. */ + sdhci_start_slot(&sc->slot); + + return (0); +} + +static int +sdhci_acpi_detach(device_t dev) +{ + struct sdhci_acpi_softc *sc = device_get_softc(dev); + + if (sc->intrhand) + bus_teardown_intr(dev, sc->irq_res, sc->intrhand); + if (sc->irq_res) + bus_release_resource(dev, SYS_RES_IRQ, + rman_get_rid(sc->irq_res), sc->irq_res); + + if (sc->mem_res) { + sdhci_cleanup_slot(&sc->slot); + bus_release_resource(dev, SYS_RES_MEMORY, + rman_get_rid(sc->mem_res), sc->mem_res); + } + + return (0); +} + +static int +sdhci_acpi_shutdown(device_t dev) +{ + + return (0); +} + +static int +sdhci_acpi_suspend(device_t dev) +{ + struct sdhci_acpi_softc *sc = device_get_softc(dev); + int err; + + err = bus_generic_suspend(dev); + if (err) + return (err); + sdhci_generic_suspend(&sc->slot); + return (0); +} + +static int +sdhci_acpi_resume(device_t dev) +{ + struct sdhci_acpi_softc *sc = device_get_softc(dev); + int err; + + sdhci_generic_resume(&sc->slot); + err = bus_generic_resume(dev); + if (err) + return (err); + return (0); +} + +static void +sdhci_acpi_intr(void *arg) +{ + struct sdhci_acpi_softc *sc = (struct sdhci_acpi_softc *)arg; + + sdhci_generic_intr(&sc->slot); +} + +static device_method_t sdhci_methods[] = { + /* device_if */ + DEVMETHOD(device_probe, sdhci_acpi_probe), + DEVMETHOD(device_attach, sdhci_acpi_attach), + DEVMETHOD(device_detach, sdhci_acpi_detach), + DEVMETHOD(device_shutdown, sdhci_acpi_shutdown), + DEVMETHOD(device_suspend, sdhci_acpi_suspend), + DEVMETHOD(device_resume, sdhci_acpi_resume), + + /* Bus interface */ + DEVMETHOD(bus_read_ivar, sdhci_generic_read_ivar), + DEVMETHOD(bus_write_ivar, sdhci_generic_write_ivar), + + /* mmcbr_if */ + DEVMETHOD(mmcbr_update_ios, sdhci_generic_update_ios), + DEVMETHOD(mmcbr_request, sdhci_generic_request), + DEVMETHOD(mmcbr_get_ro, sdhci_generic_get_ro), + DEVMETHOD(mmcbr_acquire_host, sdhci_generic_acquire_host), + DEVMETHOD(mmcbr_release_host, sdhci_generic_release_host), + + /* SDHCI registers accessors */ + DEVMETHOD(sdhci_read_1, sdhci_acpi_read_1), + DEVMETHOD(sdhci_read_2, sdhci_acpi_read_2), + DEVMETHOD(sdhci_read_4, sdhci_acpi_read_4), + DEVMETHOD(sdhci_read_multi_4, sdhci_acpi_read_multi_4), + DEVMETHOD(sdhci_write_1, sdhci_acpi_write_1), + DEVMETHOD(sdhci_write_2, sdhci_acpi_write_2), + DEVMETHOD(sdhci_write_4, sdhci_acpi_write_4), + DEVMETHOD(sdhci_write_multi_4, sdhci_acpi_write_multi_4), + + DEVMETHOD_END +}; + +static driver_t sdhci_acpi_driver = { + "sdhci_acpi", + sdhci_methods, + sizeof(struct sdhci_acpi_softc), +}; +static devclass_t sdhci_acpi_devclass; + +DRIVER_MODULE(sdhci_acpi, acpi, sdhci_acpi_driver, sdhci_acpi_devclass, NULL, + NULL); +MODULE_DEPEND(sdhci_acpi, sdhci, 1, 1, 1); +DRIVER_MODULE(mmc, sdhci_acpi, mmc_driver, mmc_devclass, NULL, NULL); +MODULE_DEPEND(sdhci_acpi, mmc, 1, 1, 1); diff --git a/sys/modules/Makefile b/sys/modules/Makefile index 2bccd1883e55..0e016f7ce820 100644 --- a/sys/modules/Makefile +++ b/sys/modules/Makefile @@ -332,6 +332,7 @@ SUBDIR= \ scc \ ${_scsi_low} \ sdhci \ + ${_sdhci_acpi} \ sdhci_pci \ sem \ send \ @@ -665,6 +666,7 @@ _padlock_rng= padlock_rng _rdrand_rng= rdrand_rng .endif _s3= s3 +_sdhci_acpi= sdhci_acpi _tpm= tpm _twa= twa _vesa= vesa diff --git a/sys/modules/sdhci_acpi/Makefile b/sys/modules/sdhci_acpi/Makefile new file mode 100644 index 000000000000..621b4062ee60 --- /dev/null +++ b/sys/modules/sdhci_acpi/Makefile @@ -0,0 +1,8 @@ +# $FreeBSD$ + +.PATH: ${.CURDIR}/../../dev/sdhci + +KMOD= sdhci_acpi +SRCS= sdhci_acpi.c sdhci.h sdhci_if.h device_if.h bus_if.h pci_if.h mmcbr_if.h + +.include From 6b3e2169c5ab3eb877e21343fa7044f2f5bfc325 Mon Sep 17 00:00:00 2001 From: Justin Hibbits Date: Wed, 11 Jan 2017 02:21:34 +0000 Subject: [PATCH 64/79] Force all TOC references in asm to include '@toc' This reportedly fixes one problem with booting a clang kernel. PR: kern/215819 Submitted by: Mark Millard MFC after: 2 weeks --- sys/powerpc/include/asm.h | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/sys/powerpc/include/asm.h b/sys/powerpc/include/asm.h index 3aec5d3a4791..08aab9854c62 100644 --- a/sys/powerpc/include/asm.h +++ b/sys/powerpc/include/asm.h @@ -89,10 +89,11 @@ name: #ifdef __powerpc64__ -#define TOC_REF(name) __CONCAT(.L,name) +#define TOC_NAME_FOR_REF(name) __CONCAT(.L,name) +#define TOC_REF(name) TOC_NAME_FOR_REF(name)@toc #define TOC_ENTRY(name) \ .section ".toc","aw"; \ - TOC_REF(name): \ + TOC_NAME_FOR_REF(name): \ .tc name[TC],name #endif From 8f8c559269bccf737d318b89630a0ef812865b99 Mon Sep 17 00:00:00 2001 From: Xin LI Date: Wed, 11 Jan 2017 05:42:06 +0000 Subject: [PATCH 65/79] Apply upstream fix for CVE-2016-10009 and CVE-2016-10010: add a whitelist of paths from which ssh-agent will load (via ssh-pkcs11-helper) a PKCS#11 module; ok markus@ disable Unix-domain socket forwarding when privsep is disabled (Note that this is a backport of upstream fixes, and this commit is mainly to ease future imports). Obtained from: OpenBSD --- serverloop.c | 4 ++-- ssh-agent.1 | 15 ++++++++++++++- ssh-agent.c | 41 ++++++++++++++++++++++++++++++++++------- 3 files changed, 50 insertions(+), 10 deletions(-) diff --git a/serverloop.c b/serverloop.c index 80d1db5490bc..f5c362dfcc59 100644 --- a/serverloop.c +++ b/serverloop.c @@ -995,7 +995,7 @@ server_request_direct_streamlocal(void) /* XXX fine grained permissions */ if ((options.allow_streamlocal_forwarding & FORWARD_LOCAL) != 0 && - !no_port_forwarding_flag) { + !no_port_forwarding_flag && use_privsep) { c = channel_connect_to_path(target, "direct-streamlocal@openssh.com", "direct-streamlocal"); } else { @@ -1279,7 +1279,7 @@ server_input_global_request(int type, u_int32_t seq, void *ctxt) /* check permissions */ if ((options.allow_streamlocal_forwarding & FORWARD_REMOTE) == 0 - || no_port_forwarding_flag) { + || no_port_forwarding_flag || !use_privsep) { success = 0; packet_send_debug("Server has disabled port forwarding."); } else { diff --git a/ssh-agent.1 b/ssh-agent.1 index c4b50bbdfe74..c349426de6b5 100644 --- a/ssh-agent.1 +++ b/ssh-agent.1 @@ -1,4 +1,4 @@ -.\" $OpenBSD: ssh-agent.1,v 1.62 2015/11/15 23:54:15 jmc Exp $ +.\" $OpenBSD: ssh-agent.1,v 1.63 2016/11/30 03:07:37 djm Exp $ .\" .\" Author: Tatu Ylonen .\" Copyright (c) 1995 Tatu Ylonen , Espoo, Finland @@ -47,6 +47,7 @@ .Op Fl a Ar bind_address .Op Fl E Ar fingerprint_hash .Op Fl t Ar life +.Op Fl P Ar pkcs11_whitelist .Op Ar command Op Ar arg ... .Nm ssh-agent .Op Fl c | s @@ -121,6 +122,18 @@ The default is Kill the current agent (given by the .Ev SSH_AGENT_PID environment variable). +.It Fl P +Specify a pattern-list of acceptable paths for PKCS#11 shared libraries +that may be added using the +.Fl s +option to +.Xr ssh-add 1 . +The default is to allow loading PKCS#11 libraries from +.Dq /usr/lib/*,/usr/local/lib/* . +PKCS#11 libraries that do not match the whitelist will be refused. +See PATTERNS in +.Xr ssh_config 5 +for a description of pattern-list syntax. .It Fl s Generate Bourne shell commands on .Dv stdout . diff --git a/ssh-agent.c b/ssh-agent.c index c38906d94278..99525d2a48e3 100644 --- a/ssh-agent.c +++ b/ssh-agent.c @@ -83,11 +83,16 @@ #include "misc.h" #include "digest.h" #include "ssherr.h" +#include "match.h" #ifdef ENABLE_PKCS11 #include "ssh-pkcs11.h" #endif +#ifndef DEFAULT_PKCS11_WHITELIST +# define DEFAULT_PKCS11_WHITELIST "/usr/lib/*,/usr/local/lib/*" +#endif + #if defined(HAVE_SYS_PRCTL_H) #include /* For prctl() and PR_SET_DUMPABLE */ #endif @@ -139,6 +144,9 @@ pid_t cleanup_pid = 0; char socket_name[PATH_MAX]; char socket_dir[PATH_MAX]; +/* PKCS#11 path whitelist */ +static char *pkcs11_whitelist; + /* locking */ #define LOCK_SIZE 32 #define LOCK_SALT_SIZE 16 @@ -741,7 +749,7 @@ no_identities(SocketEntry *e, u_int type) static void process_add_smartcard_key(SocketEntry *e) { - char *provider = NULL, *pin; + char *provider = NULL, *pin, canonical_provider[PATH_MAX]; int r, i, version, count = 0, success = 0, confirm = 0; u_int seconds; time_t death = 0; @@ -773,10 +781,21 @@ process_add_smartcard_key(SocketEntry *e) goto send; } } + if (realpath(provider, canonical_provider) == NULL) { + verbose("failed PKCS#11 add of \"%.100s\": realpath: %s", + provider, strerror(errno)); + goto send; + } + if (match_pattern_list(canonical_provider, pkcs11_whitelist, 0) != 1) { + verbose("refusing PKCS#11 add of \"%.100s\": " + "provider not whitelisted", canonical_provider); + goto send; + } + debug("%s: add %.100s", __func__, canonical_provider); if (lifetime && !death) death = monotime() + lifetime; - count = pkcs11_add_provider(provider, pin, &keys); + count = pkcs11_add_provider(canonical_provider, pin, &keys); for (i = 0; i < count; i++) { k = keys[i]; version = k->type == KEY_RSA1 ? 1 : 2; @@ -784,8 +803,8 @@ process_add_smartcard_key(SocketEntry *e) if (lookup_identity(k, version) == NULL) { id = xcalloc(1, sizeof(Identity)); id->key = k; - id->provider = xstrdup(provider); - id->comment = xstrdup(provider); /* XXX */ + id->provider = xstrdup(canonical_provider); + id->comment = xstrdup(canonical_provider); /* XXX */ id->death = death; id->confirm = confirm; TAILQ_INSERT_TAIL(&tab->idlist, id, next); @@ -1176,7 +1195,7 @@ usage(void) { fprintf(stderr, "usage: ssh-agent [-c | -s] [-Dd] [-a bind_address] [-E fingerprint_hash]\n" - " [-t life] [command [arg ...]]\n" + " [-P pkcs11_whitelist] [-t life] [command [arg ...]]\n" " ssh-agent [-c | -s] -k\n"); exit(1); } @@ -1220,7 +1239,7 @@ main(int ac, char **av) __progname = ssh_get_progname(av[0]); seed_rng(); - while ((ch = getopt(ac, av, "cDdksE:a:t:")) != -1) { + while ((ch = getopt(ac, av, "cDdksE:a:P:t:")) != -1) { switch (ch) { case 'E': fingerprint_hash = ssh_digest_alg_by_name(optarg); @@ -1235,6 +1254,11 @@ main(int ac, char **av) case 'k': k_flag++; break; + case 'P': + if (pkcs11_whitelist != NULL) + fatal("-P option already specified"); + pkcs11_whitelist = xstrdup(optarg); + break; case 's': if (c_flag) usage(); @@ -1269,6 +1293,9 @@ main(int ac, char **av) if (ac > 0 && (c_flag || k_flag || s_flag || d_flag || D_flag)) usage(); + if (pkcs11_whitelist == NULL) + pkcs11_whitelist = xstrdup(DEFAULT_PKCS11_WHITELIST); + if (ac == 0 && !c_flag && !s_flag) { shell = getenv("SHELL"); if (shell != NULL && (len = strlen(shell)) > 2 && @@ -1416,7 +1443,7 @@ main(int ac, char **av) signal(SIGTERM, cleanup_handler); nalloc = 0; - if (pledge("stdio cpath unix id proc exec", NULL) == -1) + if (pledge("stdio rpath cpath unix id proc exec", NULL) == -1) fatal("%s: pledge: %s", __progname, strerror(errno)); platform_pledge_agent(); From 19f27f3c3447a903b2c426433de3e1c4b6b58079 Mon Sep 17 00:00:00 2001 From: Enji Cooper Date: Wed, 11 Jan 2017 07:17:03 +0000 Subject: [PATCH 66/79] Fix up r311227 Check for creat returning a value != -1, not a non-zero value MFC after: 3 days Pointyhat to: ngie Reported by: Coverity CID: 1368366 --- contrib/netbsd-tests/lib/libc/gen/t_dir.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/contrib/netbsd-tests/lib/libc/gen/t_dir.c b/contrib/netbsd-tests/lib/libc/gen/t_dir.c index cc38c3b0e772..3b4c6cb4bb42 100644 --- a/contrib/netbsd-tests/lib/libc/gen/t_dir.c +++ b/contrib/netbsd-tests/lib/libc/gen/t_dir.c @@ -59,12 +59,12 @@ ATF_TC_BODY(seekdir_basic, tc) long here; #ifdef __FreeBSD__ -#define CREAT(x, m) do { \ - int _creat_fd; \ - ATF_REQUIRE_MSG((_creat_fd = creat((x), (m))), \ - "creat(%s, %x) failed: %s", (x), (m), \ - strerror(errno)); \ - (void)close(_creat_fd); \ +#define CREAT(x, m) do { \ + int _creat_fd; \ + ATF_REQUIRE_MSG((_creat_fd = creat((x), (m))) != -1, \ + "creat(%s, %x) failed: %s", (x), (m), \ + strerror(errno)); \ + (void)close(_creat_fd); \ } while(0); ATF_REQUIRE_MSG(mkdir("t", 0755) == 0, From b4d60f9802690baf2e1b89212879ba495e53ee38 Mon Sep 17 00:00:00 2001 From: Hiroki Sato Date: Wed, 11 Jan 2017 07:21:59 +0000 Subject: [PATCH 67/79] Add more #ifdef INET and INET6. --- usr.sbin/syslogd/syslogd.c | 27 +++++++++++++++++++++++++-- 1 file changed, 25 insertions(+), 2 deletions(-) diff --git a/usr.sbin/syslogd/syslogd.c b/usr.sbin/syslogd/syslogd.c index 81254bd564bf..278e4be57535 100644 --- a/usr.sbin/syslogd/syslogd.c +++ b/usr.sbin/syslogd/syslogd.c @@ -89,9 +89,11 @@ __FBSDID("$FreeBSD$"); #include #include +#if defined(INET) || defined(INET6) #include -#include #include +#endif +#include #include #include @@ -127,8 +129,11 @@ static const char include_ext[] = ".conf"; #define MAXUNAMES 20 /* maximum number of user names */ #define sstosa(ss) ((struct sockaddr *)(ss)) +#ifdef INET #define sstosin(ss) ((struct sockaddr_in *)(void *)(ss)) #define satosin(sa) ((struct sockaddr_in *)(void *)(sa)) +#endif +#ifdef INET6 #define sstosin6(ss) ((struct sockaddr_in6 *)(void *)(ss)) #define satosin6(sa) ((struct sockaddr_in6 *)(void *)(sa)) #define s6_addr32 __u6_addr.__u6_addr32 @@ -137,6 +142,7 @@ static const char include_ext[] = ".conf"; (((d)->s6_addr32[1] ^ (a)->s6_addr32[1]) & (m)->s6_addr32[1]) == 0 && \ (((d)->s6_addr32[2] ^ (a)->s6_addr32[2]) & (m)->s6_addr32[2]) == 0 && \ (((d)->s6_addr32[3] ^ (a)->s6_addr32[3]) & (m)->s6_addr32[3]) == 0 ) +#endif /* * List of peers and sockets for binding. */ @@ -1305,10 +1311,12 @@ fprintlog(struct filed *f, int flags, const char *msg) case F_FORW: dprintf(" %s", f->fu_forw_hname); switch (f->fu_forw_addr->ai_addr->sa_family) { +#ifdef INET case AF_INET: dprintf(":%d\n", ntohs(satosin(f->fu_forw_addr->ai_addr)->sin_port)); break; +#endif #ifdef INET6 case AF_INET6: dprintf(":%d\n", @@ -1929,7 +1937,20 @@ init(int signo) break; case F_FORW: - port = ntohs(satosin(f->fu_forw_addr->ai_addr)->sin_port); + switch (f->fu_forw_addr->ai_addr->sa_family) { +#ifdef INET + case AF_INET: + port = ntohs(satosin(f->fu_forw_addr->ai_addr)->sin_port); + break; +#endif +#ifdef INET6 + case AF_INET6: + port = ntohs(satosin6(f->fu_forw_addr->ai_addr)->sin6_port); + break; +#endif + default: + port = 0; + } if (port != 514) { printf("%s:%d", f->fu_forw_hname, port); @@ -2410,6 +2431,7 @@ timedout(int sig __unused) static int allowaddr(char *s) { +#if defined(INET) || defined(INET6) char *cp1, *cp2; struct allowedpeer *ap; struct servent *se; @@ -2571,6 +2593,7 @@ allowaddr(char *s) } printf("port = %d\n", ap->port); } +#endif return (0); } From 0e0dd023066b7cd3b8355e153c9ef768e131dff7 Mon Sep 17 00:00:00 2001 From: Enji Cooper Date: Wed, 11 Jan 2017 07:22:21 +0000 Subject: [PATCH 68/79] Partially revert r311236 There's no sense in trying to close a file descriptor from the negative cases with unlink_test; it's best to ignore these cases. The mkfifo case does make sense to keep though. MFC after: 3 days --- contrib/netbsd-tests/lib/libc/sys/t_unlink.c | 10 ---------- 1 file changed, 10 deletions(-) diff --git a/contrib/netbsd-tests/lib/libc/sys/t_unlink.c b/contrib/netbsd-tests/lib/libc/sys/t_unlink.c index e72974c7d133..b504bb06debc 100644 --- a/contrib/netbsd-tests/lib/libc/sys/t_unlink.c +++ b/contrib/netbsd-tests/lib/libc/sys/t_unlink.c @@ -63,12 +63,7 @@ ATF_TC_BODY(unlink_basic, tc) ATF_REQUIRE(unlink(path) == 0); errno = 0; -#ifdef __FreeBSD__ - ATF_REQUIRE_ERRNO(ENOENT, (fd = open(path, O_RDONLY)) == -1); - (void)close(fd); -#else ATF_REQUIRE_ERRNO(ENOENT, open(path, O_RDONLY) == -1); -#endif } } @@ -128,12 +123,7 @@ ATF_TC_BODY(unlink_fifo, tc) ATF_REQUIRE(unlink(path) == 0); errno = 0; -#ifdef __FreeBSD__ - ATF_REQUIRE_ERRNO(ENOENT, (fd = open(path, O_RDONLY)) == -1); - (void)close(fd); -#else ATF_REQUIRE_ERRNO(ENOENT, open(path, O_RDONLY) == -1); -#endif } ATF_TC_CLEANUP(unlink_fifo, tc) From d87ba4cf3a533294eacc9336a78661813906a740 Mon Sep 17 00:00:00 2001 From: Enji Cooper Date: Wed, 11 Jan 2017 08:15:18 +0000 Subject: [PATCH 69/79] Pull in changes from upstream for lib/libc/{c063,gen,string,sys} to address issues resolved in FreeBSD or support added to testcases In collaboration with: --- lib/libc/c063/t_faccessat.c | 7 +- lib/libc/c063/t_fchmodat.c | 7 +- lib/libc/c063/t_fchownat.c | 7 +- lib/libc/c063/t_fexecve.c | 5 +- lib/libc/c063/t_fstatat.c | 7 +- lib/libc/c063/t_mkfifoat.c | 5 +- lib/libc/c063/t_mknodat.c | 5 +- lib/libc/c063/t_o_search.c | 9 +- lib/libc/c063/t_openat.c | 7 +- lib/libc/c063/t_readlinkat.c | 7 +- lib/libc/c063/t_unlinkat.c | 7 +- lib/libc/c063/t_utimensat.c | 9 +- lib/libc/gen/posix_spawn/t_fileactions.c | 7 +- lib/libc/gen/t_assert.c | 20 +++- lib/libc/gen/t_dir.c | 29 ++++-- lib/libc/gen/t_ftok.c | 5 +- lib/libc/gen/t_humanize_number.c | 3 +- lib/libc/gen/t_sleep.c | 12 ++- lib/libc/gen/t_time.c | 5 +- lib/libc/gen/t_ttyname.c | 5 +- lib/libc/gen/t_vis.c | 6 +- lib/libc/string/t_strchr.c | 10 +- lib/libc/string/t_strerror.c | 5 +- lib/libc/sys/t_access.c | 10 +- lib/libc/sys/t_chroot.c | 5 +- lib/libc/sys/t_mincore.c | 5 +- lib/libc/sys/t_mmap.c | 5 +- lib/libc/sys/t_wait.c | 114 +++++++++++++++++------ lib/libc/t_cdb.c | 7 +- 29 files changed, 229 insertions(+), 106 deletions(-) diff --git a/lib/libc/c063/t_faccessat.c b/lib/libc/c063/t_faccessat.c index c9e0cc803158..5e6829f7ea45 100644 --- a/lib/libc/c063/t_faccessat.c +++ b/lib/libc/c063/t_faccessat.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_faccessat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $ */ +/* $NetBSD: t_faccessat.c,v 1.3 2017/01/10 15:13:56 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. @@ -29,8 +29,10 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_faccessat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $"); +__RCSID("$NetBSD: t_faccessat.c,v 1.3 2017/01/10 15:13:56 christos Exp $"); +#include +#include #include #include #include @@ -39,7 +41,6 @@ __RCSID("$NetBSD: t_faccessat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $"); #include #include #include -#include #define DIR "dir" #define FILE "dir/faccessat" diff --git a/lib/libc/c063/t_fchmodat.c b/lib/libc/c063/t_fchmodat.c index 462d53d9e961..a7bb6831b86c 100644 --- a/lib/libc/c063/t_fchmodat.c +++ b/lib/libc/c063/t_fchmodat.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_fchmodat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $ */ +/* $NetBSD: t_fchmodat.c,v 1.3 2017/01/10 15:13:56 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. @@ -29,8 +29,10 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_fchmodat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $"); +__RCSID("$NetBSD: t_fchmodat.c,v 1.3 2017/01/10 15:13:56 christos Exp $"); +#include +#include #include #include #include @@ -39,7 +41,6 @@ __RCSID("$NetBSD: t_fchmodat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $"); #include #include #include -#include #define DIR "dir" #define FILE "dir/fchmodat" diff --git a/lib/libc/c063/t_fchownat.c b/lib/libc/c063/t_fchownat.c index 80c760683eb9..631d55ac521c 100644 --- a/lib/libc/c063/t_fchownat.c +++ b/lib/libc/c063/t_fchownat.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_fchownat.c,v 1.3 2013/03/17 04:46:06 jmmv Exp $ */ +/* $NetBSD: t_fchownat.c,v 1.4 2017/01/10 15:13:56 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. @@ -29,8 +29,10 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_fchownat.c,v 1.3 2013/03/17 04:46:06 jmmv Exp $"); +__RCSID("$NetBSD: t_fchownat.c,v 1.4 2017/01/10 15:13:56 christos Exp $"); +#include +#include #include #include #include @@ -40,7 +42,6 @@ __RCSID("$NetBSD: t_fchownat.c,v 1.3 2013/03/17 04:46:06 jmmv Exp $"); #include #include #include -#include #define DIR "dir" #define FILE "dir/fchownat" diff --git a/lib/libc/c063/t_fexecve.c b/lib/libc/c063/t_fexecve.c index d557b00d92c2..b9aa0170f885 100644 --- a/lib/libc/c063/t_fexecve.c +++ b/lib/libc/c063/t_fexecve.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_fexecve.c,v 1.2 2013/03/17 04:35:59 jmmv Exp $ */ +/* $NetBSD: t_fexecve.c,v 1.3 2017/01/10 15:15:09 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. @@ -29,7 +29,7 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_fexecve.c,v 1.2 2013/03/17 04:35:59 jmmv Exp $"); +__RCSID("$NetBSD: t_fexecve.c,v 1.3 2017/01/10 15:15:09 christos Exp $"); #include @@ -70,6 +70,7 @@ ATF_TC_BODY(fexecve, tc) error = 76; else error = EXIT_FAILURE; + (void)close(fd); err(error, "fexecve"); } } diff --git a/lib/libc/c063/t_fstatat.c b/lib/libc/c063/t_fstatat.c index a48cd5781384..c17961f26494 100644 --- a/lib/libc/c063/t_fstatat.c +++ b/lib/libc/c063/t_fstatat.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_fstatat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $ */ +/* $NetBSD: t_fstatat.c,v 1.3 2017/01/10 15:13:56 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. @@ -29,8 +29,10 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_fstatat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $"); +__RCSID("$NetBSD: t_fstatat.c,v 1.3 2017/01/10 15:13:56 christos Exp $"); +#include +#include #include #include #include @@ -39,7 +41,6 @@ __RCSID("$NetBSD: t_fstatat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $"); #include #include #include -#include #define DIR "dir" #define FILE "dir/fstatat" diff --git a/lib/libc/c063/t_mkfifoat.c b/lib/libc/c063/t_mkfifoat.c index 1ae023c3f4a6..5c496c8d0987 100644 --- a/lib/libc/c063/t_mkfifoat.c +++ b/lib/libc/c063/t_mkfifoat.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_mkfifoat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $ */ +/* $NetBSD: t_mkfifoat.c,v 1.3 2017/01/10 15:15:09 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. @@ -29,7 +29,7 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_mkfifoat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $"); +__RCSID("$NetBSD: t_mkfifoat.c,v 1.3 2017/01/10 15:15:09 christos Exp $"); #include #include @@ -63,6 +63,7 @@ ATF_TC_BODY(mkfifoat_fd, tc) ATF_REQUIRE((fd = mkfifoat(dfd, BASEFIFO, mode)) != -1); ATF_REQUIRE(close(fd) == 0); ATF_REQUIRE(access(FIFO, F_OK) == 0); + (void)close(dfd); } ATF_TC(mkfifoat_fdcwd); diff --git a/lib/libc/c063/t_mknodat.c b/lib/libc/c063/t_mknodat.c index b04a1595d6e5..7c3ab9b2edbe 100644 --- a/lib/libc/c063/t_mknodat.c +++ b/lib/libc/c063/t_mknodat.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_mknodat.c,v 1.3 2013/03/17 04:46:06 jmmv Exp $ */ +/* $NetBSD: t_mknodat.c,v 1.4 2017/01/10 15:15:09 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. @@ -29,7 +29,7 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_mknodat.c,v 1.3 2013/03/17 04:46:06 jmmv Exp $"); +__RCSID("$NetBSD: t_mknodat.c,v 1.4 2017/01/10 15:15:09 christos Exp $"); #include #include @@ -80,6 +80,7 @@ ATF_TC_BODY(mknodat_fd, tc) ATF_REQUIRE((fd = mknodat(dfd, BASEFILE, mode, dev)) != -1); ATF_REQUIRE(close(fd) == 0); ATF_REQUIRE(access(FILE, F_OK) == 0); + (void)close(dfd); } ATF_TC(mknodat_fdcwd); diff --git a/lib/libc/c063/t_o_search.c b/lib/libc/c063/t_o_search.c index d9dbe19e899f..7cefa8bf8822 100644 --- a/lib/libc/c063/t_o_search.c +++ b/lib/libc/c063/t_o_search.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_o_search.c,v 1.4 2013/03/17 04:46:06 jmmv Exp $ */ +/* $NetBSD: t_o_search.c,v 1.5 2017/01/10 22:25:01 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. @@ -29,9 +29,13 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_o_search.c,v 1.4 2013/03/17 04:46:06 jmmv Exp $"); +__RCSID("$NetBSD: t_o_search.c,v 1.5 2017/01/10 22:25:01 christos Exp $"); #include + +#include +#include + #include #include #include @@ -40,7 +44,6 @@ __RCSID("$NetBSD: t_o_search.c,v 1.4 2013/03/17 04:46:06 jmmv Exp $"); #include #include #include -#include /* * dholland 20130112: disable tests that require O_SEARCH semantics diff --git a/lib/libc/c063/t_openat.c b/lib/libc/c063/t_openat.c index 79b5f38d81fd..f7c8c74b5b15 100644 --- a/lib/libc/c063/t_openat.c +++ b/lib/libc/c063/t_openat.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_openat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $ */ +/* $NetBSD: t_openat.c,v 1.3 2017/01/10 15:13:56 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. @@ -29,8 +29,10 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_openat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $"); +__RCSID("$NetBSD: t_openat.c,v 1.3 2017/01/10 15:13:56 christos Exp $"); +#include +#include #include #include #include @@ -39,7 +41,6 @@ __RCSID("$NetBSD: t_openat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $"); #include #include #include -#include #define DIR "dir" #define FILE "dir/openat" diff --git a/lib/libc/c063/t_readlinkat.c b/lib/libc/c063/t_readlinkat.c index d354ff59b886..cf14652a9efe 100644 --- a/lib/libc/c063/t_readlinkat.c +++ b/lib/libc/c063/t_readlinkat.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_readlinkat.c,v 1.3 2013/03/17 04:46:06 jmmv Exp $ */ +/* $NetBSD: t_readlinkat.c,v 1.4 2017/01/10 15:13:56 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. @@ -29,8 +29,10 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_readlinkat.c,v 1.3 2013/03/17 04:46:06 jmmv Exp $"); +__RCSID("$NetBSD: t_readlinkat.c,v 1.4 2017/01/10 15:13:56 christos Exp $"); +#include +#include #include #include #include @@ -39,7 +41,6 @@ __RCSID("$NetBSD: t_readlinkat.c,v 1.3 2013/03/17 04:46:06 jmmv Exp $"); #include #include #include -#include #define DIR "dir" #define FILE "dir/readlinkat" diff --git a/lib/libc/c063/t_unlinkat.c b/lib/libc/c063/t_unlinkat.c index 79aa7aa2bf70..9897b928c516 100644 --- a/lib/libc/c063/t_unlinkat.c +++ b/lib/libc/c063/t_unlinkat.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_unlinkat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $ */ +/* $NetBSD: t_unlinkat.c,v 1.3 2017/01/10 15:13:56 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. @@ -29,8 +29,10 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_unlinkat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $"); +__RCSID("$NetBSD: t_unlinkat.c,v 1.3 2017/01/10 15:13:56 christos Exp $"); +#include +#include #include #include #include @@ -39,7 +41,6 @@ __RCSID("$NetBSD: t_unlinkat.c,v 1.2 2013/03/17 04:46:06 jmmv Exp $"); #include #include #include -#include #define DIR "dir" #define FILE "dir/unlinkat" diff --git a/lib/libc/c063/t_utimensat.c b/lib/libc/c063/t_utimensat.c index 9f21fd6f126a..682c2df06974 100644 --- a/lib/libc/c063/t_utimensat.c +++ b/lib/libc/c063/t_utimensat.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_utimensat.c,v 1.5 2013/03/17 04:46:06 jmmv Exp $ */ +/* $NetBSD: t_utimensat.c,v 1.6 2017/01/10 15:13:56 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. @@ -29,8 +29,11 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_utimensat.c,v 1.5 2013/03/17 04:46:06 jmmv Exp $"); +__RCSID("$NetBSD: t_utimensat.c,v 1.6 2017/01/10 15:13:56 christos Exp $"); +#include +#include +#include #include #include #include @@ -39,8 +42,6 @@ __RCSID("$NetBSD: t_utimensat.c,v 1.5 2013/03/17 04:46:06 jmmv Exp $"); #include #include #include -#include -#include #define DIR "dir" #define FILE "dir/utimensat" diff --git a/lib/libc/gen/posix_spawn/t_fileactions.c b/lib/libc/gen/posix_spawn/t_fileactions.c index 1cf643365c3f..a955d6a65190 100644 --- a/lib/libc/gen/posix_spawn/t_fileactions.c +++ b/lib/libc/gen/posix_spawn/t_fileactions.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_fileactions.c,v 1.5 2012/04/09 19:42:07 martin Exp $ */ +/* $NetBSD: t_fileactions.c,v 1.6 2017/01/10 22:36:29 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. @@ -32,6 +32,10 @@ #include + +#include +#include + #include #include #include @@ -39,7 +43,6 @@ #include #include #include -#include ATF_TC(t_spawn_openmode); diff --git a/lib/libc/gen/t_assert.c b/lib/libc/gen/t_assert.c index 140417a38aa8..ce73015142d6 100644 --- a/lib/libc/gen/t_assert.c +++ b/lib/libc/gen/t_assert.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_assert.c,v 1.2 2011/06/14 05:28:00 jruoho Exp $ */ +/* $NetBSD: t_assert.c,v 1.3 2017/01/10 15:17:57 christos Exp $ */ /*- * Copyright (c) 2011 The NetBSD Foundation, Inc. @@ -29,8 +29,11 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_assert.c,v 1.2 2011/06/14 05:28:00 jruoho Exp $"); +__RCSID("$NetBSD: t_assert.c,v 1.3 2017/01/10 15:17:57 christos Exp $"); +#include +#include +#include #include #include @@ -40,6 +43,17 @@ __RCSID("$NetBSD: t_assert.c,v 1.2 2011/06/14 05:28:00 jruoho Exp $"); #include #include +static void +disable_corefile(void) +{ + struct rlimit limits; + + limits.rlim_cur = 0; + limits.rlim_max = 0; + + ATF_REQUIRE(setrlimit(RLIMIT_CORE, &limits) == 0); +} + static void handler(int); static void @@ -65,6 +79,7 @@ ATF_TC_BODY(assert_false, tc) if (pid == 0) { + disable_corefile(); (void)closefrom(0); (void)memset(&sa, 0, sizeof(struct sigaction)); @@ -102,6 +117,7 @@ ATF_TC_BODY(assert_true, tc) if (pid == 0) { + disable_corefile(); (void)closefrom(0); (void)memset(&sa, 0, sizeof(struct sigaction)); diff --git a/lib/libc/gen/t_dir.c b/lib/libc/gen/t_dir.c index 81412c1ae898..6142f68caf3c 100644 --- a/lib/libc/gen/t_dir.c +++ b/lib/libc/gen/t_dir.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_dir.c,v 1.6 2013/10/19 17:45:00 christos Exp $ */ +/* $NetBSD: t_dir.c,v 1.8 2017/01/11 07:26:17 christos Exp $ */ /*- * Copyright (c) 2010 The NetBSD Foundation, Inc. @@ -26,18 +26,19 @@ * POSSIBILITY OF SUCH DAMAGE. */ -#include - +#include #include +#include #include #include +#include #include #include #include #include #include -#include + ATF_TC(seekdir_basic); ATF_TC_HEAD(seekdir_basic, tc) @@ -54,10 +55,19 @@ ATF_TC_BODY(seekdir_basic, tc) struct dirent *entry; long here; - mkdir("t", 0755); - creat("t/a", 0600); - creat("t/b", 0600); - creat("t/c", 0600); +#define CREAT(x, m) do { \ + int _creat_fd; \ + ATF_REQUIRE_MSG((_creat_fd = creat((x), (m)) != -1), \ + "creat(%s, %x) failed: %s", (x), (m), \ + strerror(errno)); \ + (void)close(_creat_fd); \ + } while(0); + + ATF_REQUIRE_MSG(mkdir("t", 0755) == 0, + "mkdir failed: %s", strerror(errno)); + CREAT("t/a", 0600); + CREAT("t/b", 0600); + CREAT("t/c", 0600); dp = opendir("t"); if ( dp == NULL) @@ -70,6 +80,8 @@ ATF_TC_BODY(seekdir_basic, tc) /* get first entry */ entry = readdir(dp); here = telldir(dp); + ATF_REQUIRE_MSG(here != -1, + "telldir failed: %s", strerror(errno)); /* get second entry */ entry = readdir(dp); @@ -109,6 +121,7 @@ ATF_TC_BODY(seekdir_basic, tc) atf_tc_fail("3rd seekdir found wrong name"); closedir(dp); + free(wasname); } ATF_TC(telldir_leak); diff --git a/lib/libc/gen/t_ftok.c b/lib/libc/gen/t_ftok.c index 100bd1bb982f..4c1ab18950eb 100644 --- a/lib/libc/gen/t_ftok.c +++ b/lib/libc/gen/t_ftok.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_ftok.c,v 1.1 2011/11/08 05:47:00 jruoho Exp $ */ +/* $NetBSD: t_ftok.c,v 1.2 2017/01/10 15:19:52 christos Exp $ */ /*- * Copyright (c) 2011 The NetBSD Foundation, Inc. @@ -29,7 +29,7 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_ftok.c,v 1.1 2011/11/08 05:47:00 jruoho Exp $"); +__RCSID("$NetBSD: t_ftok.c,v 1.2 2017/01/10 15:19:52 christos Exp $"); #include #include @@ -68,6 +68,7 @@ ATF_TC_BODY(ftok_link, tc) fd = open(path, O_RDONLY | O_CREAT); ATF_REQUIRE(fd >= 0); + (void)close(fd); ATF_REQUIRE(link(path, hlnk) == 0); ATF_REQUIRE(symlink(path, slnk) == 0); diff --git a/lib/libc/gen/t_humanize_number.c b/lib/libc/gen/t_humanize_number.c index 1af579e61c02..cf701d1112d9 100644 --- a/lib/libc/gen/t_humanize_number.c +++ b/lib/libc/gen/t_humanize_number.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_humanize_number.c,v 1.8 2012/03/18 07:14:08 jruoho Exp $ */ +/* $NetBSD: t_humanize_number.c,v 1.9 2017/01/10 15:20:44 christos Exp $ */ /*- * Copyright (c) 2010, 2011 The NetBSD Foundation, Inc. @@ -241,6 +241,7 @@ ATF_TC_BODY(humanize_number_basic, tc) newline(); atf_tc_fail_nonfatal("Failed for table entry %d", i); } + free(buf); } ATF_TC(humanize_number_big); diff --git a/lib/libc/gen/t_sleep.c b/lib/libc/gen/t_sleep.c index b24f8dd830d5..4545a6941a76 100644 --- a/lib/libc/gen/t_sleep.c +++ b/lib/libc/gen/t_sleep.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_sleep.c,v 1.9 2016/08/11 21:34:11 kre Exp $ */ +/* $NetBSD: t_sleep.c,v 1.11 2017/01/10 15:43:59 maya Exp $ */ /*- * Copyright (c) 2006 Frank Kardel @@ -26,8 +26,14 @@ * POSSIBILITY OF SUCH DAMAGE. */ +#include +#include +#include +#include /* for TIMESPEC_TO_TIMEVAL on FreeBSD */ + #include #include +#include #include #include #include @@ -35,10 +41,6 @@ #include #include -#include -#include -#include - #include "isqemu.h" #define BILLION 1000000000LL /* nano-seconds per second */ diff --git a/lib/libc/gen/t_time.c b/lib/libc/gen/t_time.c index 4496f59efaf0..15a8d585c847 100644 --- a/lib/libc/gen/t_time.c +++ b/lib/libc/gen/t_time.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_time.c,v 1.3 2014/10/31 12:22:38 justin Exp $ */ +/* $NetBSD: t_time.c,v 1.4 2017/01/10 15:32:46 christos Exp $ */ /*- * Copyright (c) 2011 The NetBSD Foundation, Inc. @@ -29,7 +29,7 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_time.c,v 1.3 2014/10/31 12:22:38 justin Exp $"); +__RCSID("$NetBSD: t_time.c,v 1.4 2017/01/10 15:32:46 christos Exp $"); #include #include @@ -38,6 +38,7 @@ __RCSID("$NetBSD: t_time.c,v 1.3 2014/10/31 12:22:38 justin Exp $"); #include #include #include +#include #include ATF_TC(time_copy); diff --git a/lib/libc/gen/t_ttyname.c b/lib/libc/gen/t_ttyname.c index 0c10c2496443..61121b8b9c0d 100644 --- a/lib/libc/gen/t_ttyname.c +++ b/lib/libc/gen/t_ttyname.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_ttyname.c,v 1.3 2011/05/01 18:14:01 jruoho Exp $ */ +/* $NetBSD: t_ttyname.c,v 1.4 2017/01/10 15:33:40 christos Exp $ */ /*- * Copyright (c) 2011 The NetBSD Foundation, Inc. @@ -29,7 +29,7 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_ttyname.c,v 1.3 2011/05/01 18:14:01 jruoho Exp $"); +__RCSID("$NetBSD: t_ttyname.c,v 1.4 2017/01/10 15:33:40 christos Exp $"); #include #include @@ -78,6 +78,7 @@ ATF_TC_BODY(ttyname_err, tc) ATF_REQUIRE(ttyname(fd) == NULL); ATF_REQUIRE(errno == ENOTTY); + (void)close(fd); } } diff --git a/lib/libc/gen/t_vis.c b/lib/libc/gen/t_vis.c index 38decb3069c9..adb0930a300a 100644 --- a/lib/libc/gen/t_vis.c +++ b/lib/libc/gen/t_vis.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_vis.c,v 1.8 2015/05/23 14:02:11 christos Exp $ */ +/* $NetBSD: t_vis.c,v 1.9 2017/01/10 15:16:57 christos Exp $ */ /*- * Copyright (c) 2002 The NetBSD Foundation, Inc. @@ -144,6 +144,7 @@ ATF_TC_BODY(strunvis_hex, tc) } } +#ifdef VIS_NOLOCALE ATF_TC(strvis_locale); ATF_TC_HEAD(strvis_locale, tc) { @@ -172,6 +173,7 @@ ATF_TC_BODY(strvis_locale, tc) setlocale(LC_CTYPE, ol); free(ol); } +#endif /* VIS_NOLOCALE */ ATF_TP_ADD_TCS(tp) { @@ -180,7 +182,9 @@ ATF_TP_ADD_TCS(tp) ATF_TP_ADD_TC(tp, strvis_null); ATF_TP_ADD_TC(tp, strvis_empty); ATF_TP_ADD_TC(tp, strunvis_hex); +#ifdef VIS_NOLOCALE ATF_TP_ADD_TC(tp, strvis_locale); +#endif /* VIS_NOLOCALE */ return atf_no_error(); } diff --git a/lib/libc/string/t_strchr.c b/lib/libc/string/t_strchr.c index 958b186e9ceb..5dd9a62213ab 100644 --- a/lib/libc/string/t_strchr.c +++ b/lib/libc/string/t_strchr.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_strchr.c,v 1.1 2011/07/07 08:59:33 jruoho Exp $ */ +/* $NetBSD: t_strchr.c,v 1.2 2017/01/10 15:34:49 christos Exp $ */ /* * Written by J.T. Conklin @@ -58,9 +58,10 @@ ATF_TC_HEAD(strchr_basic, tc) ATF_TC_BODY(strchr_basic, tc) { - unsigned int t, a; + void *dl_handle; char *off; char buf[32]; + unsigned int t, a; const char *tab[] = { "", @@ -245,8 +246,8 @@ ATF_TC_BODY(strchr_basic, tc) "abcdefgh/abcdefgh/", }; - - strchr_fn = dlsym(dlopen(0, RTLD_LAZY), "test_strchr"); + dl_handle = dlopen(NULL, RTLD_LAZY); + strchr_fn = dlsym(dl_handle, "test_strlen"); if (!strchr_fn) strchr_fn = strchr; @@ -281,6 +282,7 @@ ATF_TC_BODY(strchr_basic, tc) verify_strchr(buf + a, 0xff, t, a); } } + (void)dlclose(dl_handle); } ATF_TP_ADD_TCS(tp) diff --git a/lib/libc/string/t_strerror.c b/lib/libc/string/t_strerror.c index c0e9c06e9185..99b95b4266c6 100644 --- a/lib/libc/string/t_strerror.c +++ b/lib/libc/string/t_strerror.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_strerror.c,v 1.3 2011/05/10 06:55:27 jruoho Exp $ */ +/* $NetBSD: t_strerror.c,v 1.4 2017/01/10 20:35:49 christos Exp $ */ /*- * Copyright (c) 2011 The NetBSD Foundation, Inc. @@ -29,10 +29,11 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_strerror.c,v 1.3 2011/05/10 06:55:27 jruoho Exp $"); +__RCSID("$NetBSD: t_strerror.c,v 1.4 2017/01/10 20:35:49 christos Exp $"); #include #include +#include /* Needed for sys_nerr on FreeBSD */ #include #include #include diff --git a/lib/libc/sys/t_access.c b/lib/libc/sys/t_access.c index f15afc25a6f2..f75617235972 100644 --- a/lib/libc/sys/t_access.c +++ b/lib/libc/sys/t_access.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_access.c,v 1.1 2011/07/07 06:57:53 jruoho Exp $ */ +/* $NetBSD: t_access.c,v 1.2 2017/01/10 22:36:29 christos Exp $ */ /*- * Copyright (c) 2011 The NetBSD Foundation, Inc. @@ -29,7 +29,11 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_access.c,v 1.1 2011/07/07 06:57:53 jruoho Exp $"); +__RCSID("$NetBSD: t_access.c,v 1.2 2017/01/10 22:36:29 christos Exp $"); + +#include + +#include #include #include @@ -38,8 +42,6 @@ __RCSID("$NetBSD: t_access.c,v 1.1 2011/07/07 06:57:53 jruoho Exp $"); #include #include -#include - static const char path[] = "access"; static const int mode[4] = { R_OK, W_OK, X_OK, F_OK }; diff --git a/lib/libc/sys/t_chroot.c b/lib/libc/sys/t_chroot.c index ce7170875f24..aa761b80967f 100644 --- a/lib/libc/sys/t_chroot.c +++ b/lib/libc/sys/t_chroot.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_chroot.c,v 1.1 2011/07/07 06:57:53 jruoho Exp $ */ +/* $NetBSD: t_chroot.c,v 1.2 2017/01/10 22:36:29 christos Exp $ */ /*- * Copyright (c) 2011 The NetBSD Foundation, Inc. @@ -29,9 +29,10 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_chroot.c,v 1.1 2011/07/07 06:57:53 jruoho Exp $"); +__RCSID("$NetBSD: t_chroot.c,v 1.2 2017/01/10 22:36:29 christos Exp $"); #include +#include #include #include diff --git a/lib/libc/sys/t_mincore.c b/lib/libc/sys/t_mincore.c index 553207a8af7f..431970953e19 100644 --- a/lib/libc/sys/t_mincore.c +++ b/lib/libc/sys/t_mincore.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_mincore.c,v 1.8 2012/06/08 07:18:58 martin Exp $ */ +/* $NetBSD: t_mincore.c,v 1.9 2017/01/10 22:36:29 christos Exp $ */ /*- * Copyright (c) 2011 The NetBSD Foundation, Inc. @@ -59,9 +59,10 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_mincore.c,v 1.8 2012/06/08 07:18:58 martin Exp $"); +__RCSID("$NetBSD: t_mincore.c,v 1.9 2017/01/10 22:36:29 christos Exp $"); #include +#include #include #include diff --git a/lib/libc/sys/t_mmap.c b/lib/libc/sys/t_mmap.c index 8576403fdea3..963b3c5cebeb 100644 --- a/lib/libc/sys/t_mmap.c +++ b/lib/libc/sys/t_mmap.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_mmap.c,v 1.9 2015/02/28 13:57:08 martin Exp $ */ +/* $NetBSD: t_mmap.c,v 1.10 2017/01/10 22:36:29 christos Exp $ */ /*- * Copyright (c) 2011 The NetBSD Foundation, Inc. @@ -55,10 +55,11 @@ * SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_mmap.c,v 1.9 2015/02/28 13:57:08 martin Exp $"); +__RCSID("$NetBSD: t_mmap.c,v 1.10 2017/01/10 22:36:29 christos Exp $"); #include #include +#include #include #include #include diff --git a/lib/libc/sys/t_wait.c b/lib/libc/sys/t_wait.c index c06bb5519cae..163709bf3e78 100644 --- a/lib/libc/sys/t_wait.c +++ b/lib/libc/sys/t_wait.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_wait.c,v 1.4 2016/04/27 21:14:24 christos Exp $ */ +/* $NetBSD: t_wait.c,v 1.7 2016/11/06 15:04:14 kamil Exp $ */ /*- * Copyright (c) 2016 The NetBSD Foundation, Inc. @@ -29,7 +29,7 @@ * POSSIBILITY OF SUCH DAMAGE. */ #include -__RCSID("$NetBSD: t_wait.c,v 1.4 2016/04/27 21:14:24 christos Exp $"); +__RCSID("$NetBSD: t_wait.c,v 1.7 2016/11/06 15:04:14 kamil Exp $"); #include #include @@ -60,22 +60,6 @@ ATF_TC_BODY(wait6_invalid, tc) && errno == EINVAL); } -ATF_TC(wait6_noproc); -ATF_TC_HEAD(wait6_noproc, tc) -{ - atf_tc_set_md_var(tc, "descr", - "Test that wait6(2) returns ECHILD with for no processes"); -} - -ATF_TC_BODY(wait6_noproc, tc) -{ - siginfo_t si; - struct wrusage wru; - int st; - ATF_REQUIRE(wait6(P_ALL, 0, &st, WEXITED, &wru, &si) == -1 - && errno == ECHILD); -} - ATF_TC(wait6_exited); ATF_TC_HEAD(wait6_exited, tc) { @@ -92,12 +76,12 @@ ATF_TC_BODY(wait6_exited, tc) switch (pid = fork()) { case -1: - ATF_REQUIRE(pid > 0); + ATF_REQUIRE(pid > 0); case 0: exit(0x5a5a5a5a); /*NOTREACHED*/ default: - ATF_REQUIRE(wait6(P_PID, pid, &st, WEXITED, &wru, &si) == pid); + ATF_REQUIRE(wait6(P_PID, pid, &st, WEXITED, &wru, &si) == pid); ATF_REQUIRE(WIFEXITED(st) && WEXITSTATUS(st) == 0x5a); ATF_REQUIRE(si.si_status = 0x5a5a5a5a); ATF_REQUIRE(si.si_pid == pid); @@ -128,10 +112,10 @@ ATF_TC_BODY(wait6_terminated, tc) sleep(100); /*FALLTHROUGH*/ case -1: - ATF_REQUIRE(pid > 0); + ATF_REQUIRE(pid > 0); default: ATF_REQUIRE(kill(pid, SIGTERM) == 0); - ATF_REQUIRE(wait6(P_PID, pid, &st, WEXITED, &wru, &si) == pid); + ATF_REQUIRE(wait6(P_PID, pid, &st, WEXITED, &wru, &si) == pid); ATF_REQUIRE(WIFSIGNALED(st) && WTERMSIG(st) == SIGTERM); ATF_REQUIRE(si.si_status == SIGTERM); ATF_REQUIRE(si.si_pid == pid); @@ -164,9 +148,9 @@ ATF_TC_BODY(wait6_coredumped, tc) *(char *)8 = 0; /*FALLTHROUGH*/ case -1: - ATF_REQUIRE(pid > 0); + ATF_REQUIRE(pid > 0); default: - ATF_REQUIRE(wait6(P_PID, pid, &st, WEXITED, &wru, &si) == pid); + ATF_REQUIRE(wait6(P_PID, pid, &st, WEXITED, &wru, &si) == pid); ATF_REQUIRE(WIFSIGNALED(st) && WTERMSIG(st) == SIGSEGV && WCOREDUMP(st)); ATF_REQUIRE(si.si_status == SIGSEGV); @@ -200,11 +184,14 @@ ATF_TC_BODY(wait6_stop_and_go, tc) sleep(100); /*FALLTHROUGH*/ case -1: - ATF_REQUIRE(pid > 0); + ATF_REQUIRE(pid > 0); default: ATF_REQUIRE(kill(pid, SIGSTOP) == 0); - ATF_REQUIRE(wait6(P_PID, pid, &st, WSTOPPED, &wru, &si) == pid); + ATF_REQUIRE(wait6(P_PID, pid, &st, WSTOPPED, &wru, &si) == pid); + ATF_REQUIRE(!WIFEXITED(st)); + ATF_REQUIRE(!WIFSIGNALED(st)); ATF_REQUIRE(WIFSTOPPED(st) && WSTOPSIG(st) == SIGSTOP); + ATF_REQUIRE(!WIFCONTINUED(st)); ATF_REQUIRE(si.si_status == SIGSTOP); ATF_REQUIRE(si.si_pid == pid); ATF_REQUIRE(si.si_uid == getuid()); @@ -213,8 +200,11 @@ ATF_TC_BODY(wait6_stop_and_go, tc) (uintmax_t)si.si_utime); ATF_REQUIRE(kill(pid, SIGCONT) == 0); - ATF_REQUIRE(wait6(P_PID, pid, &st, WCONTINUED, &wru, &si) == pid); + ATF_REQUIRE(wait6(P_PID, pid, &st, WCONTINUED, &wru, &si) == pid); + ATF_REQUIRE(!WIFEXITED(st)); + ATF_REQUIRE(!WIFSIGNALED(st)); ATF_REQUIRE(WIFCONTINUED(st)); + ATF_REQUIRE(!WIFSTOPPED(st)); ATF_REQUIRE(si.si_status == SIGCONT); ATF_REQUIRE(si.si_pid == pid); ATF_REQUIRE(si.si_uid == getuid()); @@ -223,8 +213,11 @@ ATF_TC_BODY(wait6_stop_and_go, tc) (uintmax_t)si.si_utime); ATF_REQUIRE(kill(pid, SIGQUIT) == 0); - ATF_REQUIRE(wait6(P_PID, pid, &st, WEXITED, &wru, &si) == pid); + ATF_REQUIRE(wait6(P_PID, pid, &st, WEXITED, &wru, &si) == pid); + ATF_REQUIRE(!WIFEXITED(st)); ATF_REQUIRE(WIFSIGNALED(st) && WTERMSIG(st) == SIGQUIT); + ATF_REQUIRE(!WIFSTOPPED(st)); + ATF_REQUIRE(!WIFCONTINUED(st)); ATF_REQUIRE(si.si_status == SIGQUIT); ATF_REQUIRE(si.si_pid == pid); ATF_REQUIRE(si.si_uid == getuid()); @@ -235,15 +228,78 @@ ATF_TC_BODY(wait6_stop_and_go, tc) } } +ATF_TC(wait6_stopgo_loop); +ATF_TC_HEAD(wait6_stopgo_loop, tc) +{ + atf_tc_set_md_var(tc, "descr", + "Test that wait6(2) handled stopped/continued process loop"); +} + +ATF_TC_BODY(wait6_stopgo_loop, tc) +{ + siginfo_t si; + struct wrusage wru; + int st; + pid_t pid; + static const struct rlimit rl = { 0, 0 }; + size_t N = 100; + + ATF_REQUIRE(setrlimit(RLIMIT_CORE, &rl) == 0); + switch (pid = fork()) { + case 0: + sleep(100); + /*FALLTHROUGH*/ + case -1: + ATF_REQUIRE(pid > 0); + } + + printf("Before loop of SIGSTOP/SIGCONT sequence %zu times\n", N); + while (N --> 0) { + ATF_REQUIRE(kill(pid, SIGSTOP) == 0); + ATF_REQUIRE(wait6(P_PID, pid, &st, WSTOPPED, &wru, &si) == pid); + ATF_REQUIRE(!WIFEXITED(st)); + ATF_REQUIRE(!WIFSIGNALED(st)); + ATF_REQUIRE(WIFSTOPPED(st) && WSTOPSIG(st) == SIGSTOP); + ATF_REQUIRE(!WIFCONTINUED(st)); + ATF_REQUIRE(si.si_status == SIGSTOP); + ATF_REQUIRE(si.si_pid == pid); + ATF_REQUIRE(si.si_uid == getuid()); + ATF_REQUIRE(si.si_code == CLD_STOPPED); + + ATF_REQUIRE(kill(pid, SIGCONT) == 0); + ATF_REQUIRE(wait6(P_PID, pid, &st, WCONTINUED, &wru, &si) == pid); + ATF_REQUIRE(!WIFEXITED(st)); + ATF_REQUIRE(!WIFSIGNALED(st)); + ATF_REQUIRE(WIFCONTINUED(st)); + ATF_REQUIRE(!WIFSTOPPED(st)); + ATF_REQUIRE(si.si_status == SIGCONT); + ATF_REQUIRE(si.si_pid == pid); + ATF_REQUIRE(si.si_uid == getuid()); + ATF_REQUIRE(si.si_code == CLD_CONTINUED); + } + ATF_REQUIRE(kill(pid, SIGQUIT) == 0); + ATF_REQUIRE(wait6(P_PID, pid, &st, WEXITED, &wru, &si) == pid); + ATF_REQUIRE(!WIFEXITED(st)); + ATF_REQUIRE(WIFSIGNALED(st) && WTERMSIG(st) == SIGQUIT); + ATF_REQUIRE(!WIFSTOPPED(st)); + ATF_REQUIRE(!WIFCONTINUED(st)); + ATF_REQUIRE(si.si_status == SIGQUIT); + ATF_REQUIRE(si.si_pid == pid); + ATF_REQUIRE(si.si_uid == getuid()); + ATF_REQUIRE(si.si_code == CLD_KILLED); + printf("user: %ju system: %ju\n", (uintmax_t)si.si_utime, + (uintmax_t)si.si_utime); +} + ATF_TP_ADD_TCS(tp) { ATF_TP_ADD_TC(tp, wait6_invalid); - ATF_TP_ADD_TC(tp, wait6_noproc); ATF_TP_ADD_TC(tp, wait6_exited); ATF_TP_ADD_TC(tp, wait6_terminated); ATF_TP_ADD_TC(tp, wait6_coredumped); ATF_TP_ADD_TC(tp, wait6_stop_and_go); + ATF_TP_ADD_TC(tp, wait6_stopgo_loop); return atf_no_error(); } diff --git a/lib/libc/t_cdb.c b/lib/libc/t_cdb.c index 5e88e6523a26..97da4a34f412 100644 --- a/lib/libc/t_cdb.c +++ b/lib/libc/t_cdb.c @@ -1,4 +1,4 @@ -/* $NetBSD: t_cdb.c,v 1.1 2012/09/27 00:38:57 joerg Exp $ */ +/* $NetBSD: t_cdb.c,v 1.2 2017/01/10 22:24:29 christos Exp $ */ /*- * Copyright (c) 2012 The NetBSD Foundation, Inc. * All rights reserved. @@ -32,9 +32,12 @@ */ #include -__RCSID("$NetBSD: t_cdb.c,v 1.1 2012/09/27 00:38:57 joerg Exp $"); +__RCSID("$NetBSD: t_cdb.c,v 1.2 2017/01/10 22:24:29 christos Exp $"); #include + +#include + #include #include #include From 8a4c0f80640475c06f73a507b0b48e6a7da80661 Mon Sep 17 00:00:00 2001 From: Enji Cooper Date: Wed, 11 Jan 2017 08:43:58 +0000 Subject: [PATCH 70/79] Add acpi_if.h and opt_acpi.h to Makefile to unbreak "make depend" with sys/modules/sdhci_acpi MFC after: 6 days X-MFC with: r311911 Reported by: Jenkins --- sys/modules/sdhci_acpi/Makefile | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/sys/modules/sdhci_acpi/Makefile b/sys/modules/sdhci_acpi/Makefile index 621b4062ee60..0d348050d116 100644 --- a/sys/modules/sdhci_acpi/Makefile +++ b/sys/modules/sdhci_acpi/Makefile @@ -3,6 +3,7 @@ .PATH: ${.CURDIR}/../../dev/sdhci KMOD= sdhci_acpi -SRCS= sdhci_acpi.c sdhci.h sdhci_if.h device_if.h bus_if.h pci_if.h mmcbr_if.h +SRCS= sdhci_acpi.c sdhci.h sdhci_if.h +SRCS+= acpi_if.h device_if.h bus_if.h opt_acpi.h pci_if.h mmcbr_if.h .include From d1a6b5fb2f0bfe55713867a4e795518293ced8c9 Mon Sep 17 00:00:00 2001 From: Enji Cooper Date: Wed, 11 Jan 2017 09:34:42 +0000 Subject: [PATCH 71/79] Fix whitespace in comment MFC after: 3 days --- contrib/netbsd-tests/lib/libc/gen/t_setdomainname.c | 2 +- contrib/netbsd-tests/lib/libc/gen/t_sethostname.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/contrib/netbsd-tests/lib/libc/gen/t_setdomainname.c b/contrib/netbsd-tests/lib/libc/gen/t_setdomainname.c index f51eb2a9d893..02dd17694740 100644 --- a/contrib/netbsd-tests/lib/libc/gen/t_setdomainname.c +++ b/contrib/netbsd-tests/lib/libc/gen/t_setdomainname.c @@ -64,7 +64,7 @@ ATF_TC_BODY(setdomainname_basic, tc) (void)memset(name, 0, sizeof(name)); #ifdef __FreeBSD__ - /* + /* * Sanity checks to ensure that the wrong invariant isn't being * tested for per PR # 181127 */ diff --git a/contrib/netbsd-tests/lib/libc/gen/t_sethostname.c b/contrib/netbsd-tests/lib/libc/gen/t_sethostname.c index 1972f7d5747b..136fa7cdfca0 100644 --- a/contrib/netbsd-tests/lib/libc/gen/t_sethostname.c +++ b/contrib/netbsd-tests/lib/libc/gen/t_sethostname.c @@ -66,7 +66,7 @@ ATF_TC_BODY(sethostname_basic, tc) (void)memset(name, 0, sizeof(name)); #ifdef __FreeBSD__ - /* + /* * Sanity checks to ensure that the wrong invariant isn't being * tested for per PR # 181127 */ From bccef7f6d9f28d4d60dc3d0ba70aaa552f096cb9 Mon Sep 17 00:00:00 2001 From: Enji Cooper Date: Wed, 11 Jan 2017 10:20:35 +0000 Subject: [PATCH 72/79] Consolidate __NetBSD__ #ifdef MFC after: 3 days --- contrib/netbsd-tests/lib/libc/regex/debug.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/contrib/netbsd-tests/lib/libc/regex/debug.c b/contrib/netbsd-tests/lib/libc/regex/debug.c index 3fc6d5b5867a..22f0b90d2272 100644 --- a/contrib/netbsd-tests/lib/libc/regex/debug.c +++ b/contrib/netbsd-tests/lib/libc/regex/debug.c @@ -48,9 +48,7 @@ #ifdef __NetBSD__ static void s_print(struct re_guts *, FILE *); static char *regchar(int); -#endif -#ifdef __NetBSD__ /* * regprint - print a regexp for debugging */ From 43e9ad02f6e9d646ec70b34d59df2a2fa77eb218 Mon Sep 17 00:00:00 2001 From: Alan Somers Date: Wed, 11 Jan 2017 16:09:25 +0000 Subject: [PATCH 73/79] Fix build of usr.bin/tail with GCC Submitted by: pluknet Reported by: pluknet MFC after: 27 days X-MFC-with: 311895 Sponsored by: Spectra Logic Corp --- usr.bin/tail/reverse.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-) diff --git a/usr.bin/tail/reverse.c b/usr.bin/tail/reverse.c index 3ad9e4bc9fdc..88c328a3aafb 100644 --- a/usr.bin/tail/reverse.c +++ b/usr.bin/tail/reverse.c @@ -170,11 +170,11 @@ r_reg(FILE *fp, const char *fn, enum STYLE style, off_t off, struct stat *sbp) ierr(fn); } -static const size_t bsz = 128 * 1024; +#define BSZ (128 * 1024) typedef struct bfelem { TAILQ_ENTRY(bfelem) entries; size_t len; - char l[bsz]; + char l[BSZ]; } bfelem_t; /* @@ -190,8 +190,8 @@ typedef struct bfelem { static void r_buf(FILE *fp, const char *fn) { - struct bfelem *tl, *temp, *first = NULL; - size_t len, llen; + struct bfelem *tl, *first = NULL; + size_t llen; char *p; off_t enomem = 0; TAILQ_HEAD(bfhead, bfelem) head; @@ -199,6 +199,8 @@ r_buf(FILE *fp, const char *fn) TAILQ_INIT(&head); while (!feof(fp)) { + size_t len; + /* * Allocate a new block and link it into place in a doubly * linked list. If out of memory, toss the LRU block and @@ -216,9 +218,9 @@ r_buf(FILE *fp, const char *fn) /* Fill the block with input data. */ len = 0; - while ((!feof(fp)) && len < bsz) { + while ((!feof(fp)) && len < BSZ) { p = tl->l + len; - len += fread(p, 1, bsz - len, fp); + len += fread(p, 1, BSZ - len, fp); if (ferror(fp)) { ierr(fn); return; @@ -244,6 +246,8 @@ r_buf(FILE *fp, const char *fn) tl = TAILQ_LAST(&head, bfhead); first = TAILQ_FIRST(&head); while (tl != NULL) { + struct bfelem *temp; + for (p = tl->l + tl->len - 1, llen = 0; p >= tl->l; --p, ++llen) { int start = (tl == first && p == tl->l); @@ -251,7 +255,7 @@ r_buf(FILE *fp, const char *fn) if ((*p == '\n') || start) { struct bfelem *tr; - if (start && len) + if (start && llen) WR(p, llen + 1); else if (llen) WR(p + 1, llen); From 7ce48226374ac39637c87ce9c296011983f6e551 Mon Sep 17 00:00:00 2001 From: Dimitry Andric Date: Wed, 11 Jan 2017 18:47:00 +0000 Subject: [PATCH 74/79] Don't include in reloc_elf.c, as it includes just after it, which has a conflicting definition of errno. This leads to the following warning with clang 4.0.0: In file included from sys/boot/common/reloc_elf32.c:6: In file included from sys/boot/common/reloc_elf.c:37: /usr/obj/usr/src/tmp/usr/include/stand.h:155:12: error: this function declaration is not a prototype [-Werror,-Wstrict-prototypes] extern int errno; ^ sys/sys/errno.h:46:26: note: expanded from macro 'errno' #define errno (* __error()) ^ MFC after: 3 days --- sys/boot/common/reloc_elf.c | 1 - 1 file changed, 1 deletion(-) diff --git a/sys/boot/common/reloc_elf.c b/sys/boot/common/reloc_elf.c index 2b60d18b5aff..6d4a00ff8f68 100644 --- a/sys/boot/common/reloc_elf.c +++ b/sys/boot/common/reloc_elf.c @@ -33,7 +33,6 @@ __FBSDID("$FreeBSD$"); #include #include -#include #include #define FREEBSD_ELF From 6377daf2ec8595543980cab89fcf8a6100d229ef Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jean-S=C3=A9bastien=20P=C3=A9dron?= Date: Wed, 11 Jan 2017 19:29:28 +0000 Subject: [PATCH 75/79] committers-ports.dot: Add myself Approved by: antoine (mentor) Differential Revision: https://reviews.freebsd.org/D9143 --- share/misc/committers-ports.dot | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/share/misc/committers-ports.dot b/share/misc/committers-ports.dot index 842620d7bf62..79c5f2e18e04 100644 --- a/share/misc/committers-ports.dot +++ b/share/misc/committers-ports.dot @@ -90,6 +90,7 @@ delphij [label="Xin Li\ndelphij@FreeBSD.org\n2006/05/01"] demon [label="Dmitry Sivachenko\ndemon@FreeBSD.org\n2000/11/13"] dhn [label="Dennis Herrmann\ndhn@FreeBSD.org\n2009/03/03"] dryice [label="Dryice Dong Liu\ndryice@FreeBSD.org\n2006/12/25"] +dumbbell [label="Jean-Sebastien Pedron\ndumbbell@FreeBSD.org\n2017/01/10"] dvl [label="Dan Langille\ndvl@FreeBSD.org\n2014/08/10"] eadler [label="Eitan Adler\neadler@FreeBSD.org\n2011/08/17"] edwin [label="Edwin Groothuis\nedwin@FreeBSD.org\n2002/10/22"] @@ -265,6 +266,8 @@ ahze -> tmclaugh amdmi3 -> jrm +antoine -> dumbbell + araujo -> lippe araujo -> pclin araujo -> pgollucci @@ -283,6 +286,7 @@ bdrewery -> trociny bapt -> bdrewery bapt -> bofh +bapt -> dumbbell bapt -> eadler bapt -> grembo bapt -> jbeich From 8237905651c3b5077999f023efdd7150580813aa Mon Sep 17 00:00:00 2001 From: Sean Bruno Date: Wed, 11 Jan 2017 19:29:33 +0000 Subject: [PATCH 76/79] Restore v6 offload caps for igb(4) class devices. Reported by: tuxen --- sys/dev/e1000/em_txrx.c | 2 +- sys/dev/e1000/if_em.h | 3 ++- sys/dev/e1000/igb_txrx.c | 2 +- 3 files changed, 4 insertions(+), 3 deletions(-) diff --git a/sys/dev/e1000/em_txrx.c b/sys/dev/e1000/em_txrx.c index 54c8880bb758..33a9b9d115f4 100644 --- a/sys/dev/e1000/em_txrx.c +++ b/sys/dev/e1000/em_txrx.c @@ -304,7 +304,7 @@ em_isc_txd_encap(void *arg, if_pkt_info_t pi) if (do_tso) { i = em_tso_setup(sc, pi, &txd_upper, &txd_lower); tso_desc = TRUE; - } else if (csum_flags & CSUM_OFFLOAD) { + } else if (csum_flags & EM_CSUM_OFFLOAD) { i = em_transmit_checksum_setup(sc, pi, &txd_upper, &txd_lower); } diff --git a/sys/dev/e1000/if_em.h b/sys/dev/e1000/if_em.h index 35755d751b3b..635d172bf458 100644 --- a/sys/dev/e1000/if_em.h +++ b/sys/dev/e1000/if_em.h @@ -330,7 +330,8 @@ #define EM_MSIX_LINK 0x01000000 /* For 82574 use */ #define ETH_ZLEN 60 #define ETH_ADDR_LEN 6 -#define CSUM_OFFLOAD 7 /* Offload bits in mbuf flag */ +#define EM_CSUM_OFFLOAD 7 /* Offload bits in mbuf flag */ +#define IGB_CSUM_OFFLOAD 0x0E0F /* Offload bits in mbuf flag */ #define IGB_PKTTYPE_MASK 0x0000FFF0 #define IGB_DMCTLX_DCFLUSH_DIS 0x80000000 /* Disable DMA Coalesce Flush */ diff --git a/sys/dev/e1000/igb_txrx.c b/sys/dev/e1000/igb_txrx.c index 28178a59074b..039269caad95 100644 --- a/sys/dev/e1000/igb_txrx.c +++ b/sys/dev/e1000/igb_txrx.c @@ -171,7 +171,7 @@ igb_tx_ctx_setup(struct tx_ring *txr, if_pkt_info_t pi, u32 *cmd_type_len, u32 * */ if (pi->ipi_mflags & M_VLANTAG) { vlan_macip_lens |= (pi->ipi_vtag << E1000_ADVTXD_VLAN_SHIFT); - } else if ((pi->ipi_csum_flags & CSUM_OFFLOAD) == 0) { + } else if ((pi->ipi_csum_flags & IGB_CSUM_OFFLOAD) == 0) { return (0); } From 5b89bdba3db97af89a49bb0a2420941529d6910b Mon Sep 17 00:00:00 2001 From: Dimitry Andric Date: Wed, 11 Jan 2017 19:59:27 +0000 Subject: [PATCH 77/79] Make EFI_RESERVED_SERVICE a proper prototype With clang 4.0.0, the EFI API header causes the following warning: In file included from sys/boot/efi/loader/bootinfo.c:43: In file included from sys/boot/efi/loader/../include/efi.h:52: sys/boot/efi/include/efiapi.h:534:32: error: this function declaration is not a prototype [-Werror,-Wstrict-prototypes] (EFIAPI *EFI_RESERVED_SERVICE) ( ^ Add VOID to make it into a real prototype. Reviewed by: imp, emaste, tsoome MFC after: 3 days Differential Revision: https://reviews.freebsd.org/D9132 --- sys/boot/efi/include/efiapi.h | 1 + 1 file changed, 1 insertion(+) diff --git a/sys/boot/efi/include/efiapi.h b/sys/boot/efi/include/efiapi.h index b1a7b45eeb53..92eb5131f2d7 100644 --- a/sys/boot/efi/include/efiapi.h +++ b/sys/boot/efi/include/efiapi.h @@ -532,6 +532,7 @@ EFI_STATUS typedef EFI_STATUS (EFIAPI *EFI_RESERVED_SERVICE) ( + VOID ); typedef From 400e78e45af1edc7cfd2b0f46dc6bf0938dca693 Mon Sep 17 00:00:00 2001 From: Dimitry Andric Date: Wed, 11 Jan 2017 20:00:24 +0000 Subject: [PATCH 78/79] Use proper prototypes in struct boot_module_t With clang 4.0.0, we are getting the following warnings about struct boot_module_t in efi's boot_module.h: In file included from sys/boot/efi/boot1/ufs_module.c:41: sys/boot/efi/boot1/boot_module.h:67:14: error: this function declaration is not a prototype [-Werror,-Wstrict-prototypes] void (*init)(); ^ void sys/boot/efi/boot1/boot_module.h:92:16: error: this function declaration is not a prototype [-Werror,-Wstrict-prototypes] void (*status)(); ^ void sys/boot/efi/boot1/boot_module.h:95:24: error: this function declaration is not a prototype [-Werror,-Wstrict-prototypes] dev_info_t *(*devices)(); ^ void 3 errors generated. Fix this by adding 'void' to the parameter lists. No functional change. Reviewed by: emaste, imp, smh MFC after: 3 days Differential Revision: https://reviews.freebsd.org/D9144 --- sys/boot/efi/boot1/boot_module.h | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/sys/boot/efi/boot1/boot_module.h b/sys/boot/efi/boot1/boot_module.h index 296d5a67a10b..3a6b8270a5cc 100644 --- a/sys/boot/efi/boot1/boot_module.h +++ b/sys/boot/efi/boot1/boot_module.h @@ -64,7 +64,7 @@ typedef struct boot_module_t const char *name; /* init is the optional initialiser for the module. */ - void (*init)(); + void (*init)(void); /* * probe checks to see if the module can handle dev. @@ -89,10 +89,10 @@ typedef struct boot_module_t void **buf, size_t *bufsize); /* status outputs information about the probed devices. */ - void (*status)(); + void (*status)(void); /* valid devices as found by probe. */ - dev_info_t *(*devices)(); + dev_info_t *(*devices)(void); } boot_module_t; /* Standard boot modules. */ From 47967ab7416a40f6be84f4649bd2747c6ec5d009 Mon Sep 17 00:00:00 2001 From: Alexander Motin Date: Wed, 11 Jan 2017 20:23:45 +0000 Subject: [PATCH 79/79] Pretend we support some IOCTLs to not scary upper layers. MFC after: 2 weeks --- sys/dev/ntb/if_ntb/if_ntb.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/sys/dev/ntb/if_ntb/if_ntb.c b/sys/dev/ntb/if_ntb/if_ntb.c index c67ae0dc682f..ace4861e6a89 100644 --- a/sys/dev/ntb/if_ntb/if_ntb.c +++ b/sys/dev/ntb/if_ntb/if_ntb.c @@ -237,6 +237,11 @@ ntb_ioctl(if_t ifp, u_long command, caddr_t data) int error = 0; switch (command) { + case SIOCSIFFLAGS: + case SIOCADDMULTI: + case SIOCDELMULTI: + break; + case SIOCSIFMTU: { if (ifr->ifr_mtu > sc->mtu - ETHER_HDR_LEN) {