in parallel in several threads, one symbol lookup could
clear db_last_symtab when another one going to use it as
starting point for traversal.
Approved by: pjd (mentor)
MFC after: 1 month
Now GCC likes to stick a "mov %eax, %FOO" instruction before
"addl $BAR, %esp" if the function just called returns an int,
which is a very common case in the kernel.
Sponsored by: RiNet (Cronyx Plus LLC)
filesystem agnostic. We are not touching any file system specific functions
in this code path. Since we have a cache lock, there is really no need to
keep Giant around here.
This eliminates Giant acquisitions for any syscall which is auditing pathnames.
Discussed with: jeff
an explicit comment that it's needed for the linuxolator. This is not the
case anymore. For all other architectures there was only a "KEEP THIS".
I'm (and other people too) running a COMPAT_43-less kernel since it's not
necessary anymore for the linuxolator. Roman is running such a kernel for a
for longer time. No problems so far. And I doubt other (newer than ia32
or alpha) architectures really depend on it.
This may result in a small performance increase for some workloads.
If the removal of COMPAT_43 results in a not working program, please
recompile it and all dependencies and try again before reporting a
problem.
The only place where COMPAT_43 is needed (as in: does not compile without
it) is in the (outdated/not usable since too old) svr4 code.
Note: this does not remove the COMPAT_43TTY option.
Nagging by: rdivacky
- Add more device IDs, ASIC revisions and chip IDs.
- Rewrite a bit code that picks the description for device.
- Introduce several macros to shorten quirks for bugs and
features.[*]
- Use some magic values, that OpenBSD has successfully
possessed from Linux (Broadcom supplied) driver.
- Remove disabled code that tried to access VPD.
[*] The macro that matches Jumbo capable NICs is
rewritten to preserve our current behavior. I
need clarify whether our or theirs is correct.
PR: 68351 (and may be others)
Obtained from: OpenBSD, brad@ mostly
- 'tag' & 'untag' action parameters.
- 'tagged' & 'limit' rule options.
Rule examples:
pipe 1 tag tablearg ip from table(1) to any
allow ip from any to table(2) tagged tablearg
allow tcp from table(3) to any 25 setup limit src-addr tablearg
sbin/ipfw/ipfw2.c:
1) new macros
GET_UINT_ARG - support of 'tablearg' keyword, argument range checking.
PRINT_UINT_ARG - support of 'tablearg' keyword.
2) strtoport(): do not silently truncate/accept invalid port list expressions
like: '1,2-abc' or '1,2-3-4' or '1,2-3x4'. style(9) cleanup.
Approved by: glebius (mentor)
MFC after: 1 month
yield() and sched_yield() syscalls. Every scheduler has its own way
to relinquish cpu, the ULE and CORE schedulers have two internal run-
queues, a timesharing thread which calls yield() syscall should be
moved to inactive queue.
There is a race with the current locking scheme and removing
it should have no measurable performance impact.
This fixes page faults leading to panics in pmap_enter_quick_locked()
on amd64/i386.
Reviewed by: alc,jhb,peter,ps
!IFQ_DRV_IS_EMPTY(). Taking this into account, I re-structured the
transmit routine so as to avoid adding another if/then in the
critical path.
Thanks to brueffer for showing my how to test with altq/pf.
Found mapped cache page. Specifically, if cnt.v_free_count dips below
cnt.v_free_reserved after p_start has been set to a non-NULL value,
then vm_map_pmap_enter() would break out of the loop and incorrectly
call pmap_enter_object() for the remaining address range. To correct
this error, this revision truncates the address range so that
pmap_enter_object() will not map any cache pages.
In collaboration with: tegge@
Reported by: kris@
- Update the firmware to the latest released firmware
(1.4.3), which corresponds to the firmware in the
latest shipping drivers from Myricom. This firmware
fixes several bugs in the firmware's PCI-e implementation,
and it also changes the driver/firmware interface:
o TSO was added, and changed the format of the transmit
descriptors.
o The firmware no longer counts transmits descriptors,
but frames. So the driver needs to keep a count
of the number of frames sent.
o The weird interrupt strategy changed to a normal receive
return ring. This ring is much bigger, and we may be
able to support DEVICE_POLLING.
o Myricom's header files changed the name of firmware
related #define's and enums (s/_MCP_/FW_).
- Stopped spamming the console with lots of printfs unless
mxge_verbose (or bootverbose) is set.
- Made additional information available via sysctl, including
the results of a PCI-e DMA benchmark run at device reset.
- Decreased the excessively long timeouts when sending commands
from 2 seconds to 20ms.
Sponsored by: Myricom Inc.
KASSERT(ke->ke_runq == NULL) panic when the sched_add is recursively
called by maybe_preempt.
Reported by: Wojciech A. Koszek < dunstan at freebsd dot czest dot pl >
pointer to a zeroed, statically allocated bpf_if structure. This way the
LIST_EMPTY() macro will always return true. This allows us to remove the
additional unconditional memory reference for each packet in the fast path.
Discussed with: sam
we intend for the user to be able to unload them later via kldunload(2)
instead of calling linker_load_module() and then directly adjusting the
ref count on the linker file structure. This makes the resulting
consumer code simpler and cleaner and better hides the linker internals
making it possible to sanely lock the linker.
Giant down in it.
- Push Giant down in kern_kldunload() and reorganize it slightly to avoid
using gotos. Also, expose this function to the rest of the kernel.
- Use a 'struct kld_file_stat' on the stack to read data under the lock
and then do one copyout() w/o holding the lock at the end to push the
data out to userland.
linker_file_unload() instead of in the middle of a bunch of code for
the case of dropping the last reference to improve readability and sanity.
While I'm here, remove pointless goto's that were just jumping to a
return statement.
moving the typedef of bus_space_tag_t from sys/sparc64/include/bus.h
to sys/sparc64/include/_bus.h. This brings sparc64 in sync with the
other platforms and fixes the compilation of drivers which include
<sys/rman.h> before <machine/bus.h> after sys/sys/rman.h rev. 1.34.
- Remove the definition of bus_type_t from sys/sparc64/include/_bus.h
as it's unused since sys/sparc64/include/bus.h rev. 1.6 and
sys/sparc64/sparc64/bus_machdep.c rev. 1.3.
- Remove some pointless comments.
Update of syscall.master:
o Adding of several new dummy syscalls (268-310)
o Synchronization of amd64 syscall.master with i386 one
o Auditing added to amd64 syscall.master
o Change auditing type for lstat syscall (bugfix). [1]
P4-Changes: 98672, 98674
Noticed by: rwatson [1]
Sponsored by: Google SoC 2006
Submitted by: rdivacky
sockets:
1) A sender sends SCM_CREDS message to a reciever, struct cmsgcred;
2) A reciever sets LOCAL_CREDS socket option and gets sender
credentials in control message, struct sockcred.
Both methods use the same control message type SCM_CREDS with the
same control message level SOL_SOCKET, so they are indistinguishable
for the receiver. A difference in struct cmsgcred and struct sockcred
layouts may lead to unwanted effects.
Now for sockets with LOCAL_CREDS option remove all previous linked
SCM_CREDS control messages and then add a control message with
struct sockcred so the process specifically asked for the peer
credentials by LOCAL_CREDS option always gets struct sockcred.
PR: kern/90800
Submitted by: Andrey Simonenko
Regres. tests: tools/regression/sockets/unix_cmsg/
MFC after: 1 month
the description so we don't have to do any more queries. Disable the
event query code until it figured out since but it is similar to the
AEN detail so we should be able to get that working.
s/myri10ge/mxge/g replacement in the myri10ge files. A few contuation
lines were joined because of the regained columns.
- Hook the mxge driver back to the build.
I picked it up again. The scheduler is forked from ULE, but the
algorithm to detect an interactive process is almost completely
different with ULE, it comes from Linux paper "Understanding the
Linux 2.6.8.1 CPU Scheduler", although I still use same word
"score" as a priority boost in ULE scheduler.
Briefly, the scheduler has following characteristic:
1. Timesharing process's nice value is seriously respected,
timeslice and interaction detecting algorithm are based
on nice value.
2. per-cpu scheduling queue and load balancing.
3. O(1) scheduling.
4. Some cpu affinity code in wakeup path.
5. Support POSIX SCHED_FIFO and SCHED_RR.
Unlike scheduler 4BSD and ULE which using fuzzy RQ_PPQ, the scheduler
uses 256 priority queues. Unlike ULE which using pull and push, the
scheduelr uses pull method, the main reason is to let relative idle
cpu do the work, but current the whole scheduler is protected by the
big sched_lock, so the benefit is not visible, it really can be worse
than nothing because all other cpu are locked out when we are doing
balancing work, which the 4BSD scheduelr does not have this problem.
The scheduler does not support hyperthreading very well, in fact,
the scheduler does not make the difference between physical CPU and
logical CPU, this should be improved in feature. The scheduler has
priority inversion problem on MP machine, it is not good for
realtime scheduling, it can cause realtime process starving.
As a result, it seems the MySQL super-smack runs better on my
Pentium-D machine when using libthr, despite on UP or SMP kernel.
- Try hard to calculate a safe sp, so that the stack doesn't get smashed
while uncompressing or relocating the kernel.
- Bring in code needed to calculate the cacheline size etc, needed for
arm9_idcache_wbinv_all.
for CBUS-PNP cards there) by default, as there are no amd64 and sparc64
machines with ISA slots and which therefore could make use of this code
known to exist. For sparc64 this additionally allows to get rid of the
compat shims for in{b,w,l}()/out{b,w,l}() etc and the associated hacks.
OK'ed by: imp, peter
o Implement a bunch of sysctl's to report the information
that's now always reported. Mvoe reporting of that info
to bootverbose, but maybe it can go away entirely.
dev.ed.X.type: string name
dev.ed.X.TxMem: amount of memory used for tx side of the card
dev.ed.X.RxMem: amount of memory used for rx side of the card
dev.ed.X.Mem: Total amount of mem on card.
o Better comments about where NE-2000 (and clones) gets their MAC
address from.
the arm to compile without all the extras that don't appear, at least
not in the flavors of ARM I deal with. This helps us save about 100k.
If I've botched the available devices on a platform, please let me
know and I'll correct ASAP.
systems. Introduce a new sysctl "hw.acpi.disable_on_reboot" that allows
users to re-enable the old behavior in case it's needed for some systems.
We never disable in the power-off path.
Original approach submitted by Alexander Logvinov <abuse@akavia.ru> with
reworking by Jung-uk Kim and myself.
current interface with the machine-independent layer. Without this change,
the page daemon would only have been awakened the first time that the
number of pv entries went above the high water mark, not each time.
S_IFDIR when making a directory
S_IFLNK when making a symbolic link
S_IFIFO when making a pipe
xfs_ialloc() checks this field for these flags when figuring
out whether to make a directory, make a symbolic link or make a pipe.
firmware_get() will not work while resuming. Note that we can't
simply drop the FIRMWARE_UNLOAD flag, because that will result in
a firmware image that can never be unloaded by the user since the
firmware subsystem will hold a linker reference to it (it's not
clear that firmware_put() without FIRMWARE_UNLOAD ever does quite
what you'd want).
with firmware_unregister(). Previously when the last driver reference
had been dropped we would clear the list entry under the assumption
that the firmware module was about to be unloaded, but this was not
true if the firmware image had been loaded manually with kldload.
This makes it possible to manually kldload firmware images as a
workaround for drivers such as ipw that attempt to load firmware
while resuming after a suspend.
Reviewed by: mlaier (an earlier version of the patch)
- Move sonewconn(), which creates new sockets for incoming connections on
listen sockets, so that all socket allocate code is together in
uipc_socket.c.
- Move 'maxsockets' and associated sysctls to uipc_socket.c with the
socket allocation code.
- Move kern.ipc sysctl node to uipc_socket.c, add a SYSCTL_DECL() for it
to sysctl.h and remove lots of scattered implementations in various
IPC modules.
- Sort sodealloc() after soalloc() in uipc_socket.c for dependency order
reasons. Statisticize soalloc() and sodealloc() as they are now
required only in uipc_socket.c, and are internal to the socket
implementation.
After this change, socket allocation and deallocation is entirely
centralized in one file, and uipc_socket2.c consists entirely of socket
buffer manipulation and default protocol switch functions.
MFC after: 1 month
not be necessary but might be helpful and at least reduce fragmentation.
* Add an assert to detect if the wakecode ever grows too big. We include
1 KB for stack, which should be more than enough also.
* Remove unnecessary initialization of static variables.
* Add comments and a bootverbose print giving the page phys address.
to 4. There is no need to be more strict at assembly time since we copy
the code anyway to a private page.
* Clear the direction flag and eflags. Probably not necessary but it won't
hurt to be safe.
* Add prefixes to all instructions to prevent any assembler mistakes.
* Remove zeroing of eax - edi. We use those registers immediately after
to transfer values to protected mode so this was pointless.
* Update comments to reflect info found during code review.
get a consistent snapshot, as well as get consistent values (i.e.,
that p_comm is properly nul-terminated).
Perforce CID: 98824
Obtained from: TrustedBSD Project
Map the device memory belonging to resources of type SYS_RES_MEMORY into
KVA upon activation so that rman_get_virtual() works as expected.
- In sbus_alloc_resource() checking whether toffs is 0 as an indication
that no applicable child range was found isn't appropriate as it's
perfectly valid for the requested SYS_RES_MEMORY resource to start at
the beginning of a child range. So check for the RMAN still being NULL
instead.
- As a minor runtime speed optimization break out of the loop where we
search for the applicable child range in sbus_alloc_resource() as soon
as it's found.
- Let sbus_setup_intr() return ENOMEM rather than 0 if it can't allocate
memory for the interrupt clearing info.
- Actually do what the comment in sbus_setup_intr() says and disable the
respective interrupt while fiddling with it.
- Remove some superfluous INTVEC() around inr, which already only contains
the interrupt vector, in sbus_setup_intr().
- While here, fix a style(9) bug in sbus_setup_intr() (don't use function
calls in initializers).
The first two changes are required for a CG6 driver.
MFC after: 2 weeks
- Removed updates to if_ibytes, if_obytes, if_imcasts, and if_omcasts.
These should not be handled by the driver.
- Add code to handle excessively fragmented mbufs when mapping TX frames.
Reviewed by: ps
Approved by: ps (mentor)
MFC after: 1 week
* Add hw.acpi.resume_beep tunable and sysctl, default to 0. Beeps the PC
speaker soon after waking to diagnose whether the wakeup code is even
getting run before other drivers possibly hang the system. To stop the beep,
cause another beep (i.e. keyboard bell). Submitted by takawata@, I changed
the frequency to be lower.
* Use 4096 instead of 4 byte alignment. Might be useful although doesn't
seem to be necessary.
* Remove a useless assignment to acpi_reset_video. It was overwritten by
the default sysctl value anyway.
- Between 1996 and 1997, wollman eliminated pr_usrreq() and replaced it
with direct function pointers. Update comment to reflect these changes.
- In 2003, I added pru_sosetlabel(). Update comment to reflect this
change.
MFC after: 1 week
SK-NET GENESIS document says reading SK_ISSR should stop generating
further interrupts(Since we drop a driver lock before invoking
ifp->if_input handler we should disable interrupts in ISR in order
to protect integrity of softc from subsequent interrupts). But it
seems that there is possibility of loosing interrupts between
reading SK_ISSR and determining which interrupts are reported.
To cope with the situation we continuously read SK_ISSR register
until there are no interrupts. However, it seems that the above
work around doesn't fix all cases. To protect watchdog handler
from triggering false alarm add a work around code which try to
reclaim pending Tx descriptors before resetting hardware. This
should fix occasional watchdog timeout errors seen on this driver.
Reported by: Frank Behrens <frank AT pinky dot sax dot de >
Tested by: Frank Behrens <frank AT pinky dot sax dot de >
taskqueued interrupt mode is going to be quite complex. Since
the polling mode is considered legacy feature for em(4) driver,
the decision is made to make polling and new interrupt handler
mutually exclusive, selected at compile time.
If kernel is compiled with DEVICE_POLLING, the fast taskqueued
interrupt handler code is disabled and the em_poll() and legacy
em_intr() functions are enabled. Otherwise, legacy functions
are disabled and only em_intr_fast() code is compiled.
Discussed with: scottl
Clean out the abortive start to homegrown, per-mpt,
Domain Validation. This should really be done at a
higher level.
Use the PIM_SEQSCAN flag for U320- this seems to correct
cases of being unable to consistently negotiate U320 in
the cases where I'd seen this before.
Between this and other recent checkins, this driver is
pretty close to being ready for MFC.
Reviewed by: scottl, ken, scsi@
MFC after: 1 week
what to do with it.
This forces us to scan targets sequentially, not in parallel.
The reason we might want to do this is that SPI negotiation
might not work right at the SIM level if we try to do it
in parallel. We *could* fix this for each SIM where this is
broken, but it's a lot harder to do that when we can simply
ask CAM to probe sequentially.
If PIM_SEQSCAN is not set (default), the original behaviour for
probing is unchanged.
LUN probing is still done in parallel for each target in either
case.
While we're at it, clean up some resource leakage for error
cases.
Reviewed by: ken, scott, scsi@
MFC after: 1 week
255.255.255.0, and a default route with gateway x.x.x.1. Now if
the address mask is changed to something more specific, e.g.,
255.255.255.128, then after the mask change the default gateway
is no longer reachable.
Since the default route is still present in the routing table,
when the output code tries to resolve the address of the default
gateway in function rt_check(), again, the default route will be
returned by rtalloc1(). Because the lock is currently held on the
rtentry structure, one more attempt to hold the lock will trigger
a crash due to "lock recursed on non-recursive mutex ..."
This is a general problem. The fix checks for the above condition
so that an existing route entry is not mistaken for a new cloned
route. Approriately, an ENETUNREACH error is returned back to the
caller
Approved by: andre
o define HAL_SOFTC, HAL_BUS_TAG, and HAL_BUS_HANDLE to be machine
independent; this fixes portability issues with bsd systems
o add ah_disable api for turning off operation of both MAC and PHY
o add ah_getAntennaSwitch and ah_setAntennaSwitch api's for better control
of antenna usage and diversity
o add ah_setAckCTSRate and ah_setAckCTSRate for controlling tx rate of
h/w generated frames
o add ah_setBeaconTimers api for simpler setting of the beacon timer registers
o remove ah_waitForBeaconDone api
o add HAL_TXDESC_DURENA flag to enable h/w duration setting in tx descriptor
o correct documentation of min/max tx power units (.5 dBm)
o switch arm, mips, and powerpc builds to use functions for register
read/write operations
o fix sparc builds to not reference %g2 and %g3 registers
o add public builds for SoC's
MFC after: 1 month
with fast_ipsec(4) and geli(8) authentication (comming soon).
If consumer requests only for HMAC algorithm (without encryption), return
EINVAL.
- Add support for the CRD_F_KEY_EXPLICIT flag, for both encryption and
authentication.
process was sucessfully audited. Otherwise, generate the PID
token. This change covers the pid < 0 cases, and pid lookup
failure cases.
Submitted by: wsalamon
Obtained from: TrustedBSD Project
set to ILACC rather than PCnet-PCI as VMware doesn't implement ILACC
compatibility, resulting in the VMware virtual machine to crash if
enabled. Add a comment regarding usage of ILACC vs. PCnet-PCI mode.
Reported and tested by: gnn, wsalamon
global audit trail configuration. This allows applications consuming
audit trails to specify parameters for which audit records are of
interest, including selecting records not required by the global trail.
Allowing application interest specification without changing the global
configuration allows intrusion detection systems to run without
interfering with global auditing or each other (if multiple are
present). To implement this:
- Kernel audit records now carry a flag to indicate whether they have
been selected by the global trail or by the audit pipe subsystem,
set during record commit, so that this information is available
after BSM conversion when delivering the BSM to the trail and audit
pipes in the audit worker thread asynchronously. Preselection by
either record target will cause the record to be kept.
- Similar changes to preselection when the audit record is created
when the system call is entering: consult both the global trail and
pipes.
- au_preselect() now accepts the class in order to avoid repeatedly
looking up the mask for each preselection test.
- Define a series of ioctls that allow applications to specify whether
they want to track the global trail, or program their own
preselection parameters: they may specify their own flags and naflags
masks, similar to the global masks of the same name, as well as a set
of per-auid masks. They also set a per-pipe mode specifying whether
they track the global trail, or user their own -- the door is left
open for future additional modes. A new ioctl is defined to allow a
user process to flush the current audit pipe queue, which can be used
after reprogramming pre-selection to make sure that only records of
interest are received in future reads.
- Audit pipe data structures are extended to hold the additional fields
necessary to support preselection. By default, audit pipes track the
global trail, so "praudit /dev/auditpipe" will track the global audit
trail even though praudit doesn't program the audit pipe selection
model.
- Comment about the complexities of potentially adding partial read
support to audit pipes.
By using a set of ioctls, applications can select which records are of
interest, and toggle the preselection mode.
Obtained from: TrustedBSD Project
device went away while open or if you tried to change the config
number while devices were open. Based on the patch from the PR with
a number of changes as discussed with the submitter.
PR: usb/97271
Submitted by: Anish Mistry
knowledge of user vs. kernel audit records into
audit_worker_process_record(). This largely confines vnode
knowledge to audit_record_write(), but avoids that logic knowing
about BSM as opposed to byte streams. This will allow us to
improve our ability to support real-time audit stream processing
by audit pipe consumers while auditing is disabled, but this
support is not yet complete.
Obtained from: TrustedBSD Project
Break out logic to call audit_record_write() and handle error
conditions into audit_worker_process_record(). This will be the
future home of some logic now present in audit_record_write()
also.
Obtained from: TrustedBSD Project
worker.
Rename audit_commit_cv to audit_watermark_cv, since it is there to
wake up threads waiting on hitting the low watermark. Describe
properly in comment.
Obtained from: TrustedBSD Project
src/sys/security/audit:
- Clarify and clean up AUR_ types to match Solaris.
- Clean up use of host vs. network byte order for IP addresses.
- Remove combined user/kernel implementations of some token creation
calls, such as au_to_file(), header calls, etc.
Obtained from: TrustedBSD Project
- Cleanup of AUR_ data types.
- Comment fixes.
- au_close_token() definition.
- Break out of kernel vs. user space token interfaces for headers.
Note: this may briefly break the kernel build until other kernel files are
updated to match.
Obtained from: TrustedBSD Project
Eliminate unnecessary, recursive acquisitions and releases of the page
queues lock by free_pv_entry() and pmap_remove_pages().
Reduce the scope of the page queues lock in pmap_remove_pages().
Before the change if a hardware crypto driver was loaded after
the software crypto driver, calling crypto_newsession() with
hard=0, will always choose software crypto.
By using a pointer to struct dos_partition, we implicitly tell the
compiler that the pointer is 4-bytes aligned, even though we know
that's not the case. The fact that we only dereference the pointer
to access a byte-wide field (field dp_ptyp) is not a guarantee that
the compiler will in fact use a byte-wide load. On some platforms
it's more efficient to use long word or quad word loads and use
bit-shifting and bit-masking to get the intended byte. On those
platforms an misaligned load will be the result.
The fix is to use byte-wide pointer arithmetic based on sizeof() and
offsetof() to avoid invalid casts which avoids that the compiler
makes invalid assumptions.
Backtrace provided by: wilko@
MFC after: 1 week
axe_cmd() calls. Without this the device can get confused if multiple
threads attempt these operations concurrently. The problem was
easily reproducible by running "ifconfig axe0" in a loop because
eventually it would conflict with axe_tick_task().
A similar approach is probably required in all USB ethernet drivers.
- Add defines with block length for each HMAC algorithm.
- Add AES_BLOCK_LEN define which is an alias for RIJNDAEL128_BLOCK_LEN.
- Add NULL_BLOCK_LEN define.
Move the code for printing timer statistics into a test function instead of
an ifdef (accessible via the debug.acpi.hpet_test tunable). Also use defines
for register offsets instead of magic values.
Courtesy of: slow flight to HK
bread() the UFS superblock. Should eliminate crashes when trying
to do: mount -t ufs on an audio CD.
PR: kern/85893
Reported by: Russell Francis <rfrancis at ev dot net>
MFC after: 1 week
non-intuitive for the ~ to be built into the mask. All the users now
explicitly ~ the mask. In addition, add MTX_UNOWNED to the mask even
though it technically isn't a flag. This should unbreak mtx_owner().
Quickly spotted by: kris
it. We just moved it to be pci specific, so this was causing compile
problems (linking problems, so I didn't notice since I unwisely just
built the module).
vendor-specific device ids across vendors.
- Include the revision in the dc_devs[] array instead of special casing
the revid handling in dc_devtype().
- Use PCI bus accessors to read registers instead of pci_read_config()
where possible.
- Use an 8-bit write to update the latency timer.
- Use PCIR_xxx constants and remove unused DC_xxx related to standard
PCI config registers.
MFC after: 1 week
dropped. This prevents a bug introduced during the socket/pcb refcounting
work from occuring, in which occasionally the retransmit timer may fire
after a connection has been reset, resulting in the resulting R|A TCP
packet having a source port of 0, as the port reservation has been
released.
While here, fixing up some RUNLOCK->WUNLOCK bugs.
MFC after: 1 month
resulting in some build failures. Instead, to fix the problem of bpf not
being present, check the pointer before dereferencing it.
This is a temporary bandaid until we can decide on how we want to handle
the bpf code not being present. This will be fixed shortly.
forget to unbusy file system before its destruction.
This fixes the following warning on mount failure:
Mount point <X> had 1 dangling refs
Tested by: wkoszek
(1) bpf peer attaches to interface netif0
(2) Packet is received by netif0
(3) ifp->if_bpf pointer is checked and handed off to bpf
(4) bpf peer detaches from netif0 resulting in ifp->if_bpf being
initialized to NULL.
(5) ifp->if_bpf is dereferenced by bpf machinery
(6) Kaboom
This race condition likely explains the various different kernel panics
reported around sending SIGINT to tcpdump or dhclient processes. But really
this race can result in kernel panics anywhere you have frequent bpf attach
and detach operations with high packet per second load.
Summary of changes:
- Remove the bpf interface's "driverp" member
- When we attach bpf interfaces, we now set the ifp->if_bpf member to the
bpf interface structure. Once this is done, ifp->if_bpf should never be
NULL. [1]
- Introduce bpf_peers_present function, an inline operation which will do
a lockless read bpf peer list associated with the interface. It should
be noted that the bpf code will pickup the bpf_interface lock before adding
or removing bpf peers. This should serialize the access to the bpf descriptor
list, removing the race.
- Expose the bpf_if structure in bpf.h so that the bpf_peers_present function
can use it. This also removes the struct bpf_if; hack that was there.
- Adjust all consumers of the raw if_bpf structure to use bpf_peers_present
Now what happens is:
(1) Packet is received by netif0
(2) Check to see if bpf descriptor list is empty
(3) Pickup the bpf interface lock
(4) Hand packet off to process
From the attach/detach side:
(1) Pickup the bpf interface lock
(2) Add/remove from bpf descriptor list
Now that we are storing the bpf interface structure with the ifnet, there is
is no need to walk the bpf interface list to locate the correct bpf interface.
We now simply look up the interface, and initialize the pointer. This has a
nice side effect of changing a bpf interface attach operation from O(N) (where
N is the number of bpf interfaces), to O(1).
[1] From now on, we can no longer check ifp->if_bpf to tell us whether or
not we have any bpf peers that might be interested in receiving packets.
In collaboration with: sam@
MFC after: 1 month
fixing speed negotiation.
Also fix the mpt_execute_req function to actually
match mpt_execute_req_a64. This may explain why
i386 users were having more grief.
