Commit Graph

89 Commits

Author SHA1 Message Date
Matt Macy
d7c5a620e2 ifnet: Replace if_addr_lock rwlock with epoch + mutex
Run on LLNW canaries and tested by pho@

gallatin:
Using a 14-core, 28-HTT single socket E5-2697 v3 with a 40GbE MLX5
based ConnectX 4-LX NIC, I see an almost 12% improvement in received
packet rate, and a larger improvement in bytes delivered all the way
to userspace.

When the host receiving 64 streams of netperf -H $DUT -t UDP_STREAM -- -m 1,
I see, using nstat -I mce0 1 before the patch:

InMpps OMpps  InGbs  OGbs err TCP Est %CPU syscalls csw     irq GBfree
4.98   0.00   4.42   0.00 4235592     33   83.80 4720653 2149771   1235 247.32
4.73   0.00   4.20   0.00 4025260     33   82.99 4724900 2139833   1204 247.32
4.72   0.00   4.20   0.00 4035252     33   82.14 4719162 2132023   1264 247.32
4.71   0.00   4.21   0.00 4073206     33   83.68 4744973 2123317   1347 247.32
4.72   0.00   4.21   0.00 4061118     33   80.82 4713615 2188091   1490 247.32
4.72   0.00   4.21   0.00 4051675     33   85.29 4727399 2109011   1205 247.32
4.73   0.00   4.21   0.00 4039056     33   84.65 4724735 2102603   1053 247.32

After the patch

InMpps OMpps  InGbs  OGbs err TCP Est %CPU syscalls csw     irq GBfree
5.43   0.00   4.20   0.00 3313143     33   84.96 5434214 1900162   2656 245.51
5.43   0.00   4.20   0.00 3308527     33   85.24 5439695 1809382   2521 245.51
5.42   0.00   4.19   0.00 3316778     33   87.54 5416028 1805835   2256 245.51
5.42   0.00   4.19   0.00 3317673     33   90.44 5426044 1763056   2332 245.51
5.42   0.00   4.19   0.00 3314839     33   88.11 5435732 1792218   2499 245.52
5.44   0.00   4.19   0.00 3293228     33   91.84 5426301 1668597   2121 245.52

Similarly, netperf reports 230Mb/s before the patch, and 270Mb/s after the patch

Reviewed by:	gallatin
Sponsored by:	Limelight Networks
Differential Revision:	https://reviews.freebsd.org/D15366
2018-05-18 20:13:34 +00:00
Pedro F. Giffuni
718cf2ccb9 sys/dev: further adoption of SPDX licensing ID tags.
Mainly focus on files that use BSD 2-Clause license, however the tool I
was using misidentified many licenses so this was mostly a manual - error
prone - task.

The Software Package Data Exchange (SPDX) group provides a specification
to make it easier for automated tools to detect and summarize well known
opensource licenses. We are gradually adopting the specification, noting
that the tags are considered only advisory and do not, in any way,
superceed or replace the license texts.
2017-11-27 14:52:40 +00:00
Gleb Smirnoff
f98131f881 Mechanically convert to if_inc_counter(). 2014-09-18 21:07:05 +00:00
Gleb Smirnoff
1bffa9511f Use define from if_var.h to access a field inside struct if_data,
that resides in struct ifnet.

Sponsored by:	Nginx, Inc.
2014-08-30 19:55:54 +00:00
Gleb Smirnoff
d6a773db08 Provide necessary includes. 2013-10-28 23:34:05 +00:00
Gleb Smirnoff
76039bc84f The r48589 promised to remove implicit inclusion of if_var.h soon. Prepare
to this event, adding if_var.h to files that do need it. Also, include
all includes that now are included due to implicit pollution via if_var.h

Sponsored by:	Netflix
Sponsored by:	Nginx, Inc.
2013-10-26 17:58:36 +00:00
Pyun YongHyeon
37d17b6b63 Avoid controller reinitialization which could be triggered by
dhclient(8) or alias addresses are added.

Tested by:	dcx dcy <dcbsdx@hotmail.com>
2013-07-10 06:46:46 +00:00
Gleb Smirnoff
c6499eccad Mechanically substitute flags from historic mbuf allocator with
malloc(9) flags in sys/dev.
2012-12-04 09:32:43 +00:00
Marius Strobl
4b7ec27007 - There's no need to overwrite the default device method with the default
one. Interestingly, these are actually the default for quite some time
  (bus_generic_driver_added(9) since r52045 and bus_generic_print_child(9)
  since r52045) but even recently added device drivers do this unnecessarily.
  Discussed with: jhb, marcel
- While at it, use DEVMETHOD_END.
  Discussed with: jhb
- Also while at it, use __FBSDID.
2011-11-22 21:28:20 +00:00
Marius Strobl
3fcb7a5365 - Remove attempts to implement setting of BMCR_LOOP/MIIF_NOLOOP
(reporting IFM_LOOP based on BMCR_LOOP is left in place though as
  it might provide useful for debugging). For most mii(4) drivers it
  was unclear whether the PHYs driven by them actually support
  loopback or not. Moreover, typically loopback mode also needs to
  be activated on the MAC, which none of the Ethernet drivers using
  mii(4) implements. Given that loopback media has no real use (and
  obviously hardly had a chance to actually work) besides for driver
  development (which just loopback mode should be sufficient for
  though, i.e one doesn't necessary need support for loopback media)
  support for it is just dropped as both NetBSD and OpenBSD already
  did quite some time ago.
- Let mii_phy_add_media() also announce the support of IFM_NONE.
- Restructure the PHY entry points to use a structure of entry points
  instead of discrete function pointers, and extend this to include
  a "reset" entry point. Make sure any PHY-specific reset routine is
  always used, and provide one for lxtphy(4) which disables MII
  interrupts (as is done for a few other PHYs we have drivers for).
  This includes changing NIC drivers which previously just called the
  generic mii_phy_reset() to now actually call the PHY-specific reset
  routine, which might be crucial in some cases. While at it, the
  redundant checks in these NIC drivers for mii->mii_instance not being
  zero before calling the reset routines were removed because as soon
  as one PHY driver attaches mii->mii_instance is incremented and we
  hardly can end up in their media change callbacks etc if no PHY driver
  has attached as mii_attach() would have failed in that case and not
  attach a miibus(4) instance.
  Consequently, NIC drivers now no longer should call mii_phy_reset()
  directly, so it was removed from EXPORT_SYMS.
- Add a mii_phy_dev_attach() as a companion helper to mii_phy_dev_probe().
  The purpose of that function is to perform the common steps to attach
  a PHY driver instance and to hook it up to the miibus(4) instance and to
  optionally also handle the probing, addition and initialization of the
  supported media. So all a PHY driver without any special requirements
  has to do in its bus attach method is to call mii_phy_dev_attach()
  along with PHY-specific MIIF_* flags, a pointer to its PHY functions
  and the add_media set to one. All PHY drivers were updated to take
  advantage of mii_phy_dev_attach() as appropriate. Along with these
  changes the capability mask was added to the mii_softc structure so
  PHY drivers taking advantage of mii_phy_dev_attach() but still
  handling media on their own do not need to fiddle with the MII attach
  arguments anyway.
- Keep track of the PHY offset in the mii_softc structure. This is done
  for compatibility with NetBSD/OpenBSD.
- Keep track of the PHY's OUI, model and revision in the mii_softc
  structure. Several PHY drivers require this information also after
  attaching and previously had to wrap their own softc around mii_softc.
  NetBSD/OpenBSD also keep track of the model and revision on their
  mii_softc structure. All PHY drivers were updated to take advantage
  as appropriate.
- Convert the mebers of the MII data structure to unsigned where
  appropriate. This is partly inspired by NetBSD/OpenBSD.
- According to IEEE 802.3-2002 the bits actually have to be reversed
  when mapping an OUI to the MII ID registers. All PHY drivers and
  miidevs where changed as necessary. Actually this now again allows to
  largely share miidevs with NetBSD, which fixed this problem already
  9 years ago. Consequently miidevs was synced as far as possible.
- Add MIIF_NOMANPAUSE and mii_phy_flowstatus() calls to drivers that
  weren't explicitly converted to support flow control before. It's
  unclear whether flow control actually works with these but typically
  it should and their net behavior should be more correct with these
  changes in place than without if the MAC driver sets MIIF_DOPAUSE.

Obtained from:	NetBSD (partially)
Reviewed by:	yongari (earlier version), silence on arch@ and net@
2011-05-03 19:51:29 +00:00
Marius Strobl
19db2a7ba0 Correct spelling in comments.
Submitted by:	brucec
2011-04-22 09:39:05 +00:00
Marius Strobl
8e5d93dbb4 Convert the PHY drivers to honor the mii_flags passed down and convert
the NIC drivers as well as the PHY drivers to take advantage of the
mii_attach() introduced in r213878 to get rid of certain hacks. For
the most part these were:
- Artificially limiting miibus_{read,write}reg methods to certain PHY
  addresses; we now let mii_attach() only probe the PHY at the desired
  address(es) instead.
- PHY drivers setting MIIF_* flags based on the NIC driver they hang
  off from, partly even based on grabbing and using the softc of the
  parent; we now pass these flags down from the NIC to the PHY drivers
  via mii_attach(). This got us rid of all such hacks except those of
  brgphy() in combination with bce(4) and bge(4), which is way beyond
  what can be expressed with simple flags.

While at it, I took the opportunity to change the NIC drivers to pass
up the error returned by mii_attach() (previously by mii_phy_probe())
and unify the error message used in this case where and as appropriate
as mii_attach() actually can fail for a number of reasons, not just
because of no PHY(s) being present at the expected address(es).

Reviewed by:	jhb, yongari
2010-10-15 14:52:11 +00:00
Attilio Rao
651aa2d896 KTR_CTx are long time aliased by existing classes so they can't serve
their purpose anymore. Axe them out.

Sponsored by:	Sandvine Incorporated
Discussed with:	jhb, emaste
Possible MFC:	TBD
2010-07-21 10:05:07 +00:00
Joel Dahl
1edcf74de7 The NetBSD Foundation has granted permission to remove clause 3 and 4 from
the software.

Obtained from:	NetBSD
2010-03-03 17:55:51 +00:00
Robert Watson
eb956cd041 Use if_maddr_rlock()/if_maddr_runlock() rather than IF_ADDR_LOCK()/
IF_ADDR_UNLOCK() across network device drivers when accessing the
per-interface multicast address list, if_multiaddrs.  This will
allow us to change the locking strategy without affecting our driver
programming interface or binary interface.

For two wireless drivers, remove unnecessary locking, since they
don't actually access the multicast address list.

Approved by:	re (kib)
MFC after:	6 weeks
2009-06-26 11:45:06 +00:00
Marius Strobl
c15fccb8f2 - Use the revamped code from the gem(4) PCI front-end, which
doesn't require parts of the Expansion ROM to be copied around,
  for obtaining the MAC address on !OFW platforms.
- Don't unnecessarily cache bus space tag and handle nor RIDs
  in the softcs of the front-ends.
- Don't use function calls in initializers.
- Let the SBus front-end depend on sbus(4).
2008-04-26 14:17:21 +00:00
Marius Strobl
da28723ecd o Disable HMEDEBUG by default.
o Add CTASSERTs ensuring that HME_NRXDESC and HME_NTXDESC are set to
  legal values.
o Use appropriate maxsize, nsegments and maxsegsize parameters when
  creating DMA tags and correct some comments related to them.
o The FreeBSD bus_dmamap_sync(9) supports ored together flags for quite
  some time now so collapse calls accordingly.
o Add missing BUS_DMASYNC_PREREAD when syncing the control DMA maps in
  hme_rint() and hme_start_locked().
o Keep state of the link state and use it to enable or disable the MAC
  in hme_mii_statchg() accordingly as well as to return early from
  hme_start_locked() in case the link is down.
o Introduce a sc_flags and use it to replace individual members like
  sc_pci.
o Add bus_barrier(9) calls to hme_mac_bitflip(), hme_mii_readreg(),
  hme_mii_writereg() and hme_stop() to ensure the respective bit
  has been written before we starting polling on it and for the right
  bits to change.
o Rather just returning in case hme_mac_bitflip() fails and leaving us
  in an undefined state report the problem and move on; chances are
  the requested configuration will become active shortly after.
o Don't call hme_start_locked() in hme_init_locked() unconditionally
  but only after calls to hme_init_locked() when it's appropriate, i.e.
  in hme_watchdog().
o Add a KASSERT which asserts nsegs is valid also to hme_load_txmbuf().
o In hme_load_txmbuf():
  - use a maximum of the newly introduced HME_NTXSEGS segments instead
    of the incorrect HME_NTXQ, which reflects the maximum TX queue
    length, for loading the mbufs and put the DMA segments back onto
    the stack instead of the softc as 16 should be ok there.
  - use the common errno(2) return values instead of homegrown ones,
  - given that hme_load_txmbuf() is allowed to fail resulting in a
    packet drop for quite some time now implement the functionality of
    hme_txcksum() by means of m_pullup(9), which de-obfuscates the code
    and allows to always retrieve the correct length of the IP header, [1]
  - also add a KASSERT which asserts nsegs is valid,
  - take advantage of m_collapse(9) instead of m_defrag(9) for
    performance reasons.
o Don't bother to check whether the interface is running or whether its
  queue is empty before calling hme_start_locked() in hme_tint(), the
  former will check these anyway.
o In hme_intr() call hme_rint() before hme_tint() as gem_tint() may
  take quite a while to return when it calls hme_start_locked().
o Get rid of sc_debug and just check if_flags for IFF_DEBUG directly.
o Add a shadow sc_ifflags so we don't reset the chip when unnecessary.
o Handle IFF_ALLMULTI correctly. [2]
o Use PCIR_BAR instead of a homegrown macro.
o Replace sc_enaddr[6] with sc_enaddr[ETHER_ADDR_LEN].
o Use the maximum of 256 TX descriptors for better performance as using
  all of them has no additional static cost rather than using just half
  of them.

Reported by:	rwatson [2]
Suggested by:	yongari [1]
Reviewed by:	yongari
MFC after:	1 month
2008-04-24 23:12:03 +00:00
Marius Strobl
e3b78ec974 Remove invalid BUS_DMA_ALLOCNOW when creating a tag which is used for
a "static" memory allocation only.
2007-05-01 11:50:11 +00:00
Paolo Pisati
ef544f6312 o break newbus api: add a new argument of type driver_filter_t to
bus_setup_intr()

o add an int return code to all fast handlers

o retire INTR_FAST/IH_FAST

For more info: http://docs.freebsd.org/cgi/getmsg.cgi?fetch=465712+0+current/freebsd-current

Reviewed by: many
Approved by: re@
2007-02-23 12:19:07 +00:00
Marius Strobl
05856e0b8f - Use the hme_tick() callout instead of if_slowtimo() for driving
hme_watchdog() in order to avoid races accessing if_timer.
- Use bus_get_dma_tag() so hme(4) works on platforms requiring it.
- Don't bother to set if_mtu to ETHERMTU, ether_ifattach() does that.
2006-12-06 02:07:20 +00:00
Marius Strobl
e46af1289e Remove the HME_LOCK_ASSERT() in hme_mifinit(), which was added in the
previous revision; it's actually ok when invoking hme_mifinit() from
hme_config() without the lock held.
2006-12-04 01:53:40 +00:00
Marius Strobl
f279f5ced4 - In hme_stop() mask all interrupts.
- In hme_eint() print MIF register contents on MIF interrupts.
- In hme_mifinit() don't bother to preserve the previous MIF config.
  This was mainly done in order to preserve the PHY select bit (external
  or internal PHY) but which only needs to be set as appropriate when
  reading from or writing to the desired PHY in hme_mii_{read,write}reg().
  Similarly don't bother to set the PHY select bit in hme_mii_statchg().
- In hme_mii_{read,write}reg() ignore requests to PHYs other than the
  external and internal PHY one.
- Move enabling/disabling the MII drivers of the external transceiver
  from hme_init_locked() and based on the sheer presence of an external
  to hme_mifinit() and based on the currently selected media, defaulting
  to the internal transceiver when the media hasn't been set, yet.
  Invoke hme_mifinit() from the newly added hme_mediachange_locked() so
  the setting of the MII drivers is updated when changing media.
  These changes keep the MII bus from wedging (which manifests in the HME
  and the PHYs no longer being able to communicate with each other) when
  the PHY device drivers isolate the unused PHY in two-PHY configurations
  as present in f.e. Netra t1 100 while changing media, either from
  hme_init_locked() (see also below) or via ifconfig(8). They also allow
  for using both transceivers/PHYs.
- In the newly added hme_mediachange_locked() also reset the PHYs in two-
  PHY configurations before invoking mii_mediachg(). This is required
  for successfully unisolating the previously unused PHY when switching
  between PHYs.
- Now that changing media should no longer cause problems back out rev.
  1.27 and re-enable setting the current media in hme_init_locked() (see
  the commit message of rev. 1.23 for more info).

These changes are roughly a merge of NetBSD gem.c rev. 1.32 - 1.35 (1.30
was already fixed differently in our 1.36; 1.31 and 1.32 were wrong) with
some parts reworked and things that don't make sense like setting the MII
drivers and restoring the previous MIF and XIF settings in hme_mii_{read,
write}reg() omitted.

MFC after:	2 weeks
2006-12-04 00:51:08 +00:00
Pyun YongHyeon
d13bfc93ea Fix invalid reference of mbuf chains.
Use proper pointer dereference to inform modified mbuf chains to
caller.

While I'm here perform checksum offload setup after loading DMA
maps as m_defrag(9) can return new mbuf chains.

In collaboration with:	glebius
2006-08-12 01:19:37 +00:00
Simon L. B. Nielsen
f6f4187202 Fix typo in printf string.
MFC after:	1 week
Approved by:	cperciva (mentor)
2006-05-27 09:28:59 +00:00
Pyun YongHyeon
5dbde3ff57 Backout rev. 1.46. It caused Rx checksum offload breakage on little
endian systems.

Reported by:	joerg
Tested by:	joerg
2006-03-21 12:21:51 +00:00
Pyun YongHyeon
1a98c9b4a2 fix Rx checksum computation on little endian systems.
Reported & Tested by:	brad@OpenBSD
2006-01-17 06:02:22 +00:00
Ruslan Ermilov
4a0d6638b3 - Store pointer to the link-level address right in "struct ifnet"
rather than in ifindex_table[]; all (except one) accesses are
  through ifp anyway.  IF_LLADDR() works faster, and all (except
  one) ifaddr_byindex() users were converted to use ifp->if_addr.

- Stop storing a (pointer to) Ethernet address in "struct arpcom",
  and drop the IFP2ENADDR() macro; all users have been converted
  to use IF_LLADDR() instead.
2005-11-11 16:04:59 +00:00
Pyun YongHyeon
38eae97c07 - Convert hme(4) to use TX side bus_dmamap_load_mbuf_sg(9).
- Move hardware counter reading/zeroing to hme_tick(). This saves
   8 register access per interrupt. [1]
- Use imax macro for getting max. argument between two integers.
- Invoke bus_dmamap_sync(9) first before freeing mbuf.
- Check driver queue first to reduce locking operation in hme_start_locked()
  and interrupt handler.
- Simplyfy watchdog timer setup in interrupt handler.
- Don't log normal errors such as RX overrun. If we have DMA stuck
  condition, reinitialize the driver and log it.

Reviewed by:	marius
Obtained from:	OpenBSD [1]
2005-10-25 03:56:21 +00:00
Ken Smith
55915fe666 Move hme_stop() after ether_ifdetach() and if_free() to prevent a
memory-referenced-after-free panic if the hme interface fails to
attach.

Patch obtained from:	marius
Suggested same thing:	brooks

MFC after:	3 days
2005-09-08 13:50:16 +00:00
John Baldwin
00d101cf1f - Remove redundant assertions that the driver lock is not held in attach()
and detach() since mtx_lock() will assert that already since the driver
  lock is not recursive.
- Move the call to callout_init_mtx() before hme_stop() so that the
  callout_stop() in hme_stop() doesn't operate on an uninitialized callout
  structure during attach.

Reported by:	yongari (2)
MFC after:	3 days
2005-08-24 20:28:56 +00:00
John Baldwin
9eda9d7ac5 Add callout_drain()'s to foo_detach() after calling foo_stop() to make sure
that if softclock is running on another CPU and is blocked on our driver
lock, we will wait until it has acquired the lock, seen that it was
cancelled, dropped the lock, and awakened us so that we can safely destroy
the mutex.

MFC after:	3 days
2005-08-17 17:44:32 +00:00
John Baldwin
3e17bf524e - Use callout_init_mtx() to close races between hme_stop() and hme_tick().
- Use the driver lock instead of Giant in a bus dma callback.
- Clear IFF_DRV_(RUNNING|OACTIVE) in hme_stop() instead of just clearing
  RUNNING in hme_ioctl() to be more like other ethernet drivers.
- Lock the driver lock around mii operations.
- Remove spls.
- Cleanup locking in hme_ioctl().

MFC after:	1 week
2005-08-10 20:12:25 +00:00
Robert Watson
13f4c340ae Propagate rename of IFF_OACTIVE and IFF_RUNNING to IFF_DRV_OACTIVE and
IFF_DRV_RUNNING, as well as the move from ifnet.if_flags to
ifnet.if_drv_flags.  Device drivers are now responsible for
synchronizing access to these flags, as they are in if_drv_flags.  This
helps prevent races between the network stack and device driver in
maintaining the interface flags field.

Many __FreeBSD__ and __FreeBSD_version checks maintained and continued;
some less so.

Reviewed by:	pjd, bz
MFC after:	7 days
2005-08-09 10:20:02 +00:00
Robert Watson
13b203d0d7 Modify device drivers supporting multicast addresses to lock if_addr_mtx
over iteration of their multicast address lists when synchronizing the
hardware address filter with the network stack-maintained list.

Problem reported by:	Ed Maste (emaste at phaedrus dot sandvine dot ca>
MFC after:		1 week
2005-08-03 00:18:35 +00:00
Marius Strobl
51a753f9d8 Add a missing mtx_destroy() in hme_pci_detach().
Reviewed by:	yongari
Approved by:	re (scottl)
2005-07-10 10:36:45 +00:00
Brooks Davis
fc74a9f93a Stop embedding struct ifnet at the top of driver softcs. Instead the
struct ifnet or the layer 2 common structure it was embedded in have
been replaced with a struct ifnet pointer to be filled by a call to the
new function, if_alloc(). The layer 2 common structure is also allocated
via if_alloc() based on the interface type. It is hung off the new
struct ifnet member, if_l2com.

This change removes the size of these structures from the kernel ABI and
will allow us to better manage them as interfaces come and go.

Other changes of note:
 - Struct arpcom is no longer referenced in normal interface code.
   Instead the Ethernet address is accessed via the IFP2ENADDR() macro.
   To enforce this ac_enaddr has been renamed to _ac_enaddr.
 - The second argument to ether_ifattach is now always the mac address
   from driver private storage rather than sometimes being ac_enaddr.

Reviewed by:	sobomax, sam
2005-06-10 16:49:24 +00:00
Pyun YongHyeon
264166b8bb According to STP2002QFP User's Guide, it seems that driver should
program RXMAC to discard frames with SA field matching the stations's
MAC address. Experimentation shows that HME receives its own frames
when it operates at 10Mbps half-duplex. With this change HME runs at
10Mbps half-duplx should work with IPv6.
(No more "DAD detected duplicate IPv6 address".)

Reported by:	jacques brierre <jbrierre AT bellsouth DOT net>
Reviewed by:	marius
2005-05-23 05:45:36 +00:00
Marius Strobl
8386475dc8 Release the IRQ resource on detach. This fixes repeatedly loading and
unloading the hme(4) module in case it attaches to sbus(4).

MFC after:	1 week
2005-04-05 17:51:56 +00:00
Warner Losh
538565c4a5 Use BUS_PROBE_DEFAULT for pci probe return value 2005-03-05 18:30:12 +00:00
Marius Strobl
e12405280b - Fix braino introduced in rev. 1.17, unlike the X1032A (HME-ISP1040-combo)
and the X1034A (quad HME; QFE) cards the X1033A (single HME) don't have a
  PCI-PCI-bridge so we can't rely on the PCI slot number being useable as
  index for the network address to read from the VPD on the latter. Use
  the end tag to determine whether it is a QFE VPD with 4 NAs and only use
  the slot number as index in this case.
- Remove a useless check.

Prodded by:		joerg
Additional testing by:	joerg
MFC after:		1 day
2005-03-02 00:49:37 +00:00
Marius Strobl
6ff81db1a7 Add a note that all four HME chips on a SUNW,qfe card share the same PROM,
making it a bit more clear why we can read four NAs from the VPD.

MFC after:	1 day
2005-02-20 01:52:43 +00:00
Pyun YongHyeon
fee4d88e0f Fix "Duplicate mbuf free panic".
The cause of "Duplicate mbuf free panic" is in the programming
error of hme_load_txmbuf(). The code path of the panic is the
following.

1. Due to unknown reason DMA engine was freezed. So TX descritors
   of HME become full and the last failed attempt to transmit a
   packet had set its associated mbuf address to hme_txdesc
   structure. Also the failed packet is requeued into interface
   queue structure in order to retrasmit it when there are more
   available TX descritors.

2. Since DMA engine was freezed, if_timer starts to decrement its
   counter. When if_timer expires it tries to reset HME. During
   the reset phase, hme_meminit() is called and it frees all
   associated mbuf with descriptors. The last failed mbuf is also
   freed here.

3. After HME reset completed, HME starts to retransmit packets
   by dequeing the first packet in interface queue.(Note! the
   packet was already freed in hme_meminit()!)

4. When a TX completion interrupt is posted by the HME, driver
   tries to free the successfylly transmitted mbuf. Since the
   mbuf was freed in step2, now we get "Duplicate mbuf free panic".

However, the real cause is in DMA engine freeze. Since no fatal
errors reported via interrupts, there might be other cause of
the freeze. I tried hard to understand the cause of DMA engine
freeze but couldn't find any clues. It seems that the freeze
happens under very high network loads(e.g. 7.5-8.0 MB/s TX speed).

Though this fix is not enough to eliminate DMA engine freeze it's
better than panic.

Reported by:	jhb via sparc64 ML
2005-02-02 08:35:11 +00:00
David E. O'Brien
7c804097a4 Fix kernel builds with INVARIANTS. 2005-01-16 02:39:18 +00:00
Scott Long
bd1954cc37 Use bus_dmamap_load_mbuf_sg() for loading rx buffers. 2005-01-15 22:05:59 +00:00
Marius Strobl
be063afc20 Some changes related to reading the MAC-address from the VPD on systems
without Open Firmware:
- The PCI data structure of some HME PROMs contains a non-zero interface
  revision in the class code. Thus remove the checks for matching class
  code and PCI data structure length and revsion. These were pretty much
  useless anyway as we only really need the pointer to the VPD which is
  located before the structure length and revision fields.
- On Sun QFE (Quad FastEthernet) cards read the Nth MAC-address for the
  Nth HME controller instead of always the first one for all four HMEs. [1]
- Improve the comment describing the used VPD format to better reflect
  reality.
- Minor clean-up.

Prodded by:	joerg [1]
2004-12-12 00:32:51 +00:00
Pyun YongHyeon
38fa13a6a2 Make hme(4) mpsafe
- Let hme_start()/hme_init() acquire lock and then call
   hme_start_locked()/hme_init_locked() respectivly.
 - Teardown interrupt handler before hme_detach().
 - Remove IFF_NEEDSGIANT flag and mark interrupt handler INTR_MPSAFE.
 - Set callout handler to CALLOUT_MPSAFE.
 - Add locks in hme MII interface.

Reviewed by:	jake
Tested by:	Julian C. Dunn  <jdunn at opentrend dot net>
MFC after:	2 weeks
2004-11-22 06:46:30 +00:00
Marius Strobl
f348e5ac4b Make hme(4), i.e. the PCI-variant, MI by reading the MAC address on sytems
without Open Firmware directly instead of using OF_getetheraddr(). This is
a bit painful though, as the MAC address is contained in the NA field of
the VPD of the EBus bridge, which is is another function of the same chip.
To make it worse, the VPD of the EBus bridge can't be accessed via the PCI
capability pointer but has to be digged out from the Boot PROM and has a
non-standard format.
The PCI VPD struct and macros used here should be part of the FreeBSD PCI
code nevertheless.

Approved by:	tmm
Based on:	NetBSD
Tested with:	Sun X1032A (hme(4)-isp(4)-combo card) on alpha and i386
2004-08-14 22:38:20 +00:00
Robert Watson
3acd3b73c5 Since if_hme doesn't contain locking or run with INTR_MPSAFE, mark
the interface as IFF_NEEDSGIANT so if_start is run holding Giant.
2004-08-13 23:14:50 +00:00
Marius Strobl
f7abebde9e - Use bus_space_subregion() rather than arithmetic on bus_space_handle_t. [1]
- Properly use the error variable and return it on failure in the attach-
  routines.

Reviewed by:	tmm
Inspired by:	NetBSD [1]
2004-08-12 20:37:02 +00:00
Marius Strobl
26280d88d7 - Introduce an ofw_bus kobj-interface for retrieving the OFW node and a
subset ("compatible", "device_type", "model" and "name") of the standard
  properties in drivers for devices on Open Firmware supported busses. The
  standard properties "reg", "interrupts" und "address" are not covered by
  this interface because they are only of interest in the respective bridge
  code. There's a remaining standard property "status" which is unclear how
  to support properly but which also isn't used in FreeBSD at present.
  This ofw_bus kobj-interface allows to replace the various (ebus_get_node(),
  ofw_pci_get_node(), etc.) and partially inconsistent (central_get_type()
  vs. sbus_get_device_type(), etc.) existing IVAR ones with a common one.
  This in turn allows to simplify and remove code-duplication in drivers for
  devices that can hang off of more than one OFW supported bus.
- Convert the sparc64 Central, EBus, FHC, PCI and SBus bus drivers and the
  drivers for their children to use the ofw_bus kobj-interface. The IVAR-
  interfaces of the Central, EBus and FHC are entirely replaced by this. The
  PCI bus driver used its own kobj-interface and now also uses the ofw_bus
  one. The IVARs special to the SBus, e.g. for retrieving the burst size,
  remain.
  Beware: this causes an ABI-breakage for modules of drivers which used the
  IVAR-interfaces, i.e. esp(4), hme(4), isp(4) and uart(4), which need to be
  recompiled.
  The style-inconsistencies introduced in some of the bus drivers will be
  fixed by tmm@ in a generic clean-up of the respective drivers later (he
  requested to add the changes in the "new" style).
- Convert the powerpc MacIO bus driver and the drivers for its children to
  use the ofw_bus kobj-interface. This invloves removing the IVARs related
  to the "reg" property which were unused and a leftover from the NetBSD
  origini of the code. There's no ABI-breakage caused by this because none
  of these driver are currently built as modules.
  There are other powerpc bus drivers which can be converted to the ofw_bus
  kobj-interface, e.g. the PCI bus driver, which should be done together
  with converting powerpc to use the OFW PCI code from sparc64.
- Make the SBus and FHC front-end of zs(4) and the sparc64 eeprom(4) take
  advantage of the ofw_bus kobj-interface and simplify them a bit.

Reviewed by:	grehan, tmm
Approved by:	re (scottl)
Discussed with:	tmm
Tested with:	Sun AX1105, AXe, Ultra 2, Ultra 60; PPC cross-build on i386
2004-08-12 17:41:33 +00:00