]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
15 years agoImplement simple machine check support for amd64 and i386.
jhb [Wed, 13 May 2009 17:53:04 +0000 (17:53 +0000)]
Implement simple machine check support for amd64 and i386.
- For CPUs that only support MCE (the machine check exception) but not MCA
  (i.e. Pentium), all this does is print out the value of the machine check
  registers and then panic when a machine check exception occurs.
- For CPUs that support MCA (the machine check architecture), the support is
  a bit more involved.
  - First, there is limited support for decoding the CPU-independent MCA
    error codes in the kernel, and the kernel uses this to output a short
    description of any machine check events that occur.
  - When a machine check exception occurs, all of the MCx banks on the
    current CPU are scanned and any events are reported to the console
    before panic'ing.
  - To catch events for correctable errors, a periodic timer kicks off a
    task which scans the MCx banks on all CPUs.  The frequency of these
    checks is controlled via the "hw.mca.interval" sysctl.
  - Userland can request an immediate scan of the MCx banks by writing
    a non-zero value to "hw.mca.force_scan".
  - If any correctable events are encountered, the appropriate details
    are stored in a 'struct mca_record' (defined in <machine/mca.h>).
    The "hw.mca.count" is a count of such records and each record may
    be queried via the "hw.mca.records" tree by specifying the record
    index (0 .. count - 1) as the next name in the MIB similar to using
    PIDs with the kern.proc.* sysctls.  The idea is to export machine
    check events to userland for more detailed processing.
  - The periodic timer and hw.mca sysctls are only present if the CPU
    supports MCA.

Discussed with: emaste (briefly)
MFC after: 1 month

15 years agoGarbage collect now-unused NETISR_FORCEQUEUE, which overrode the global
rwatson [Wed, 13 May 2009 17:22:33 +0000 (17:22 +0000)]
Garbage collect now-unused NETISR_FORCEQUEUE, which overrode the global
direct dispatch policy for specific protocols (NETISR_USB).  We leave
the additional 'flags' argument to netisr_register() for the time being,
even though it is no longer required.

15 years agoRemove now-unused NETISR_USB.
rwatson [Wed, 13 May 2009 17:17:05 +0000 (17:17 +0000)]
Remove now-unused NETISR_USB.

15 years agoGarbage collect legacy upgt driver now that it is available in the new
rwatson [Wed, 13 May 2009 17:11:25 +0000 (17:11 +0000)]
Garbage collect legacy upgt driver now that it is available in the new
USB implementation.

Garbage collect legacy USB ethernet framework now that it is unused.

15 years agoAdd a comment to motivate my last change.
nwhitehorn [Wed, 13 May 2009 16:19:05 +0000 (16:19 +0000)]
Add a comment to motivate my last change.

Suggested by: sam, imp

15 years agoifp->if_softc is managed entirely by the driver. We never set it to
imp [Wed, 13 May 2009 14:43:26 +0000 (14:43 +0000)]
ifp->if_softc is managed entirely by the driver.  We never set it to
NULL or change it.  We initialize it before we set if_ioctl.  It can
therefore never be NULL, and most other drivers don't bother with this
sanity check.

15 years agoAdd a short delay after programming PHY registers to give some time for
nwhitehorn [Wed, 13 May 2009 14:25:55 +0000 (14:25 +0000)]
Add a short delay after programming PHY registers to give some time for
the engine to catch up. This prevents a machine check exception from
illegal memory requests with a BCM4318.

15 years agoFix print_syscall_ret parameter order.
dds [Wed, 13 May 2009 13:00:52 +0000 (13:00 +0000)]
Fix print_syscall_ret parameter order.

15 years agoFix compilation error introduced in r192025.
dds [Wed, 13 May 2009 12:43:37 +0000 (12:43 +0000)]
Fix compilation error introduced in r192025.

15 years agoFix memory leak in an error case.
brueffer [Wed, 13 May 2009 08:50:13 +0000 (08:50 +0000)]
Fix memory leak in an error case.

Found with: Coverity Prevent(tm)
CID: 371
MFC after: 2 weeks

15 years agoCorrect a rare use-after-free error in pmap_copy(). This error was
alc [Wed, 13 May 2009 07:42:53 +0000 (07:42 +0000)]
Correct a rare use-after-free error in pmap_copy().  This error was
introduced in amd64 revision 1.540 and i386 revision 1.547.  However, it
had no harmful effects until after a recent change, r189698, on amd64.
(In other words, the error is harmless in RELENG_7.)

The error is triggered by the failure to allocate a pv entry for the one
and only mapping in a page table page.  I am addressing the error by
changing pmap_copy() to abort if either pv entry allocation or page
table page allocation fails.  This is appropriate because the creation of
mappings by pmap_copy() is optional.  They are a (possible) optimization,
and not a requirement.

Correct a nearby whitespace error in the i386 pmap_copy().

Crash reported by: jeff@
MFC after: 6 weeks

15 years agoEliminate page queues locking from bufdone_finish() through the
alc [Wed, 13 May 2009 05:39:39 +0000 (05:39 +0000)]
Eliminate page queues locking from bufdone_finish() through the
following changes:

Rename vfs_page_set_valid() to vfs_page_set_validclean() to reflect
what this function actually does.  Suggested by: tegge

Introduce a new version of vfs_page_set_valid() that does no more than
what the function's name implies.  Specifically, it does not update
the page's dirty mask, and thus it does not require the page queues
lock to be held.

Update two of the three callers to the old vfs_page_set_valid() to
call vfs_page_set_validclean() instead because they actually require
the page's dirty mask to be cleared.

Introduce vm_page_set_valid().

Reviewed by: tegge

15 years agoFix copy-paste bug in NGM_NETFLOW_SETCONFIG argument size verification.
mav [Wed, 13 May 2009 02:26:34 +0000 (02:26 +0000)]
Fix copy-paste bug in NGM_NETFLOW_SETCONFIG argument size verification.

PR: kern/134220
Submitted by: Eugene Mychlo
MFC after: 1 week

15 years agoEliminate duplicate error messages from "tar c".
kientzle [Wed, 13 May 2009 00:04:08 +0000 (00:04 +0000)]
Eliminate duplicate error messages from "tar c".

Reported by: pav@

15 years agoRemove unused variables.
brueffer [Tue, 12 May 2009 23:22:58 +0000 (23:22 +0000)]
Remove unused variables.

Found with: Coverity Prevent(tm)
CID: 544, 545

15 years agoRemove unused variables.
brueffer [Tue, 12 May 2009 22:11:02 +0000 (22:11 +0000)]
Remove unused variables.

Found with: Coverity Prevent(tm)
CID: 4285, 4286

15 years ago- Resurrect the debug printf message I accidentally dropped
stas [Tue, 12 May 2009 21:28:41 +0000 (21:28 +0000)]
- Resurrect the debug printf message I accidentally dropped
  in the previous commit.
- Use device_printf instead of printf.
- Put all printfs in the interrupt handler under bootverbose.

15 years ago- Eliminate extra register reads by using a variable to store
stas [Tue, 12 May 2009 21:14:36 +0000 (21:14 +0000)]
- Eliminate extra register reads by using a variable to store
  registers contents.
- Use memory barriers to preserve the order of buffer space operations.
  This might be needed if we'll ever use this driver on architectures
  where ordering is not guaranteed.

15 years agoCorrect r190283 (partially reverting it) as on sparc64 BUS_DMA_NOCACHE
marius [Tue, 12 May 2009 20:56:34 +0000 (20:56 +0000)]
Correct r190283 (partially reverting it) as on sparc64 BUS_DMA_NOCACHE
actually is only valid for bus_dmamap_load().

MFC after: 3 days

15 years agoAdd -c option to summarize number of calls, errors, and system time.
dds [Tue, 12 May 2009 20:42:12 +0000 (20:42 +0000)]
Add -c option to summarize number of calls, errors, and system time.

Reviewed by: alfred

15 years agoRemove unused variable.
brueffer [Tue, 12 May 2009 19:33:36 +0000 (19:33 +0000)]
Remove unused variable.

Found with: Coverity Prevent(tm)
CID: 549

15 years agoRemove unused variable.
brueffer [Tue, 12 May 2009 19:30:46 +0000 (19:30 +0000)]
Remove unused variable.

Found with: Coverity Prevent(tm)
CID: 550

15 years agoAdd missing 'break' statement.
trasz [Tue, 12 May 2009 17:05:40 +0000 (17:05 +0000)]
Add missing 'break' statement.

Found with: Coverity Prevent(tm)
CID: 3919

15 years agoCheck return value of gctl_get_asciiparam().
trasz [Tue, 12 May 2009 16:59:50 +0000 (16:59 +0000)]
Check return value of gctl_get_asciiparam().

Found with: Coverity Prevent(tm)
CID: 1118

15 years agoRemove dead code.
trasz [Tue, 12 May 2009 16:39:43 +0000 (16:39 +0000)]
Remove dead code.

Found with: Coverity Prevent(tm)
CID: 556

15 years agoRemove dead code.
trasz [Tue, 12 May 2009 16:38:32 +0000 (16:38 +0000)]
Remove dead code.

Found with: Coverity Prevent(tm)
CID: 3667

15 years ago- Implement detach path.
stas [Tue, 12 May 2009 16:07:08 +0000 (16:07 +0000)]
- Implement detach path.
- Release memory and DMA resources on stop.
- Unload the associated DMA maps after transmit is complete.

15 years agoModify the experimental nfs server to use the new nfsd_nfsd_args
rmacklem [Tue, 12 May 2009 16:04:51 +0000 (16:04 +0000)]
Modify the experimental nfs server to use the new nfsd_nfsd_args
structure for nfsd. Includes a change that clarifies the use of
an empty principal name string to indicate AUTH_SYS only.

Approved by: kib (mentor)

15 years agoAdd missing free(9) in error case.
trasz [Tue, 12 May 2009 15:14:37 +0000 (15:14 +0000)]
Add missing free(9) in error case.

Found with: Coverity Prevent(tm)
CID: 4224

15 years agoAdd missing 'break' statements.
trasz [Tue, 12 May 2009 15:03:47 +0000 (15:03 +0000)]
Add missing 'break' statements.

Found with: Coverity Prevent(tm)
CID: 3936, 3937
Reviewed by: scottl@

15 years agoReport all fdescfs vnodes as VCHR for stat(2). Fake the unique
kib [Tue, 12 May 2009 09:28:45 +0000 (09:28 +0000)]
Report all fdescfs vnodes as VCHR for stat(2). Fake the unique
major/minor numbers of the devices.

Pretending that the vnodes are character devices prevents file tree
walkers from descending into the directories opened by current process.
Also, not doing stat on the filedescriptors prevents the recursive entry
into the VFS.

Requested by: kientzle
Discussed with: Jilles Tjoelker <jilles stack nl>

15 years agoReturn controlled EINVAL when the fdescfs lookup routine is given string
kib [Tue, 12 May 2009 09:22:33 +0000 (09:22 +0000)]
Return controlled EINVAL when the fdescfs lookup routine is given string
representing too large integer, instead of overflowing and possibly
returning a random but valid vnode.

Noted by: Jilles Tjoelker <jilles stack nl>
MFC after: 3 days

15 years agoThis patch adds a host route to an interface address (that is assigned
qingli [Tue, 12 May 2009 07:41:20 +0000 (07:41 +0000)]
This patch adds a host route to an interface address (that is assigned
to a non loopback/ppp link types) through the loopback interface. Prior
to the new L2/L3 rewrite, this host route is implicitly added by the L2
code during RTM_RESOLVE of that interface address. This host route is
deleted when that interface is removed.

Reviewed by: kmacy

15 years agoEliminate gratuitous clearing of the page's dirty mask.
alc [Tue, 12 May 2009 05:49:02 +0000 (05:49 +0000)]
Eliminate gratuitous clearing of the page's dirty mask.

15 years agofix bug introduced by last change
kmacy [Tue, 12 May 2009 03:30:25 +0000 (03:30 +0000)]
fix bug introduced by last change

Submitted by: Navdeep Parhar

15 years agoconnect upgt(4) to the build. It should work on all architectures.
weongyo [Tue, 12 May 2009 02:08:56 +0000 (02:08 +0000)]
connect upgt(4) to the build.  It should work on all architectures.

15 years agoAdd WUSB54AG and XM142 entries for upgt(4)
weongyo [Tue, 12 May 2009 02:05:42 +0000 (02:05 +0000)]
Add WUSB54AG and XM142 entries for upgt(4)

15 years agodon't acquire tty lock with console lock held
kmacy [Mon, 11 May 2009 23:03:15 +0000 (23:03 +0000)]
don't acquire tty lock with console lock held

15 years agoxen console lock needs to be a spin lock in case it is acquired from an interrupt...
kmacy [Mon, 11 May 2009 22:55:49 +0000 (22:55 +0000)]
xen console lock needs to be a spin lock in case it is acquired from an interrupt context

15 years ago*sigh*, while the kernel built, userland C did not. Revert the previous
jhb [Mon, 11 May 2009 21:13:00 +0000 (21:13 +0000)]
*sigh*, while the kernel built, userland C did not.  Revert the previous
commit and fix it correctly by removing the _KERNEL check entirely.  Now
the kernel always sees the same value of NULL as userland meaning that it
sees __null, 0, or 0L when compiled as C++, and '(void *)0' when compiled
as C.

15 years agoPrevent overflow of uio_resid.
kib [Mon, 11 May 2009 19:58:03 +0000 (19:58 +0000)]
Prevent overflow of uio_resid.

Noted by: jhb
MFC after: 3 days

15 years ago Change the name of the nfs server addsock structure from nfsd_args
rmacklem [Mon, 11 May 2009 19:37:05 +0000 (19:37 +0000)]
Change the name of the nfs server addsock structure from nfsd_args
to nfsd_addsock_args, so that it is consistent with the one in
sys/nfsserver/nfs.h.

Approved by: kib (mentor)

15 years agoAdd macros around the sx operations in snp(4).
ed [Mon, 11 May 2009 18:52:46 +0000 (18:52 +0000)]
Add macros around the sx operations in snp(4).

As an experiment, I changed snp(4) to use a mutex instead of an sx lock.
We can't enable this right now, because Syscons still picks up Giant.
It's nice to already have the framework there.

15 years ago Modify nfsvno_fhtovp() to ensure that it always sets the credp
rmacklem [Mon, 11 May 2009 18:45:04 +0000 (18:45 +0000)]
Modify nfsvno_fhtovp() to ensure that it always sets the credp
argument. Returning without credp set could result in a caller
doing crfree() on garbage.

Reviewed by: kan
Approved by: kib (mentor)

15 years agoAlways use __null to define NULL for GCC 4+. Use '0' rather than
jhb [Mon, 11 May 2009 17:29:11 +0000 (17:29 +0000)]
Always use __null to define NULL for GCC 4+.  Use '0' rather than
'(void *)0' for NULL for C++ compilers compiling kernel code.  Together this
makes it easier to build kernel modules using C++.

Reviewed by: imp
MFC after: 3 days

15 years agoUpdate a few XXX comments
imp [Mon, 11 May 2009 17:13:52 +0000 (17:13 +0000)]
Update a few XXX comments

Submitted by: ddk ddk ddkprog at yahoo dot com

15 years agoLFS cannot be loaded, compiled, torn, spindled or mutilated in
imp [Mon, 11 May 2009 17:05:41 +0000 (17:05 +0000)]
LFS cannot be loaded, compiled, torn, spindled or mutilated in
FreeBSD: it was deleted around FreeBSD 3.x...

15 years agoThese are some cosmetic changes to improve the clarity of libthr's fork implementation.
green [Mon, 11 May 2009 16:45:53 +0000 (16:45 +0000)]
These are some cosmetic changes to improve the clarity of libthr's fork implementation.

15 years agoFix a kernel compilation error, introduced after r191990, by defining
attilio [Mon, 11 May 2009 16:32:58 +0000 (16:32 +0000)]
Fix a kernel compilation error, introduced after r191990, by defining
thread with curthread in the AUDIT case.

Reported by: dchagin

15 years agoRemove the thread argument from the FSD (File-System Dependent) parts of
attilio [Mon, 11 May 2009 15:33:26 +0000 (15:33 +0000)]
Remove the thread argument from the FSD (File-System Dependent) parts of
the VFS.  Now all the VFS_* functions and relating parts don't want the
context as long as it always refers to curthread.

In some points, in particular when dealing with VOPs and functions living
in the same namespace (eg. vflush) which still need to be converted,
pass curthread explicitly in order to retain the old behaviour.
Such loose ends will be fixed ASAP.

While here fix a bug: now, UFS_EXTATTR can be compiled alone without the
UFS_EXTATTR_AUTOSTART option.

VFS KPI is heavilly changed by this commit so thirdy parts modules needs
to be recompiled.  Bump __FreeBSD_version in order to signal such
situation.

15 years agoTranslate l_timeval arg to native struct timeval in
dchagin [Mon, 11 May 2009 13:50:42 +0000 (13:50 +0000)]
Translate l_timeval arg to native struct timeval in
linux_setsockopt()/linux_getsockopt() for SO_RCVTIMEO,
SO_SNDTIMEO opts as l_timeval has MD members.

Remove bogus __packed attribute from l_timeval struct on __amd64__.

PR: kern/134276
Submitted by: Thomas Mueller <tmueller sysgo com>
Approved by: kib (mentor)
MFC after: 2 weeks

15 years agoAdd forgotten linux to bsd flags argument mapping into the linux_recv().
dchagin [Mon, 11 May 2009 13:42:40 +0000 (13:42 +0000)]
Add forgotten linux to bsd flags argument mapping into the linux_recv().

PR: kern/134276
Submitted by: Thomas Mueller <tmueller sysgo com>
Approved by: kib (mentor)
MFC after: 2 weeks

15 years agoless has been updated to v429
delphij [Mon, 11 May 2009 07:57:29 +0000 (07:57 +0000)]
less has been updated to v429

15 years agoRevert CVS revision 1.94 (svn r16840). Current pmap implementations don't
alc [Mon, 11 May 2009 05:16:57 +0000 (05:16 +0000)]
Revert CVS revision 1.94 (svn r16840).  Current pmap implementations don't
suffer from the race condition that motivated revision 1.94.  Consequently,
the work-around that was implemented by revision 1.94 is no longer needed.
Moreover, reverting this work-around eliminates the need for
vfs_busy_pages() to acquire the page queues lock when preparing a buffer
for read.

Reviewed by: tegge

15 years agoFix typo in bwi driver.
imp [Mon, 11 May 2009 04:57:40 +0000 (04:57 +0000)]
Fix typo in bwi driver.

15 years agorename xdr support files to avoid conflicts when linking in to the kernel
kmacy [Mon, 11 May 2009 04:18:58 +0000 (04:18 +0000)]
rename xdr support files to avoid conflicts when linking in to the kernel

15 years agoports upgt(4) driver for USB2.
weongyo [Mon, 11 May 2009 02:39:49 +0000 (02:39 +0000)]
ports upgt(4) driver for USB2.

15 years agoJust like in cpu_halt(), use cpu_shutdown() rather than ofw_exit()
marius [Sun, 10 May 2009 20:41:52 +0000 (20:41 +0000)]
Just like in cpu_halt(), use cpu_shutdown() rather than ofw_exit()
directly in cpu_reset() in order to idle the APs before exiting
the kernel and letting the BSP enter the firmware so that processes
like init(8) which still might be running on an AP at that point
don't cause a panic there when it crashes due to the fact it no
longer can be supported by the kernel.

MFC after: 3 days

15 years ago- Fix style.
marius [Sun, 10 May 2009 20:22:41 +0000 (20:22 +0000)]
- Fix style.
- Use __FBSDID.

15 years agoChange uses of the struct ccb_hdr timeout_ch missed when isp(4) was
marius [Sun, 10 May 2009 20:14:19 +0000 (20:14 +0000)]
Change uses of the struct ccb_hdr timeout_ch missed when isp(4) was
adapted to MPSAFE cam(4) to a isp(4) specific callout structure.
Thanks to Florian Smeets for providing access to a machine exhibiting
this problem for debugging.

Approved by: mjacob
MFC after: 3 days

15 years agoDo not export AT_CLKTCK when emulating Linux kernel prior
dchagin [Sun, 10 May 2009 18:43:43 +0000 (18:43 +0000)]
Do not export AT_CLKTCK when emulating Linux kernel prior
to 2.4.0, as it has appeared in the 2.4.0-rc7 first time.
Being exported, AT_CLKTCK is returned by sysconf(_SC_CLK_TCK),
glibc falls back to the hard-coded CLK_TCK value when aux entry
is not present.

Glibc versions prior to 2.2.1 always use hard-coded CLK_TCK value.

For older applications/libc's which depends on hard-coded CLK_TCK
value user should set compat.linux.osrelease less than 2.4.0.

Approved by: kib (mentor)

15 years agoIntroduce linux_kernver() interface which is intended for an exact
dchagin [Sun, 10 May 2009 18:27:20 +0000 (18:27 +0000)]
Introduce linux_kernver() interface which is intended for an exact
designation of the emulated kernel version.

linux_kernver() returns integer value formatted as 'VVVMMMIII' where
VVV - version, MMM - major revision, III - minor revision.

Approved by: kib (mentor)

15 years agoRework r189362, r191883.
dchagin [Sun, 10 May 2009 18:16:07 +0000 (18:16 +0000)]
Rework r189362, r191883.
The frequency of the statistics clock is given by stathz.
Use stathz if it is available, otherwise use hz.

Pointed out by: bde

Approved by: kib (mentor)

15 years agoEliminate stale comments.
alc [Sun, 10 May 2009 17:05:43 +0000 (17:05 +0000)]
Eliminate stale comments.

Eliminate a case of unnecessary page queues locking.

15 years agoPowerPC does not, in general, have ISA sound devices or an ISA bus,
nwhitehorn [Sun, 10 May 2009 16:00:41 +0000 (16:00 +0000)]
PowerPC does not, in general, have ISA sound devices or an ISA bus,
so add PowerPC to list of platforms for which we don't want to depend
on ISA.

15 years ago- Fix build with INVARIANTS enabled.
stas [Sun, 10 May 2009 11:05:22 +0000 (11:05 +0000)]
- Fix build with INVARIANTS enabled.

15 years ago- Fix multicast operation that I broke in previous commit.
stas [Sun, 10 May 2009 10:32:29 +0000 (10:32 +0000)]
- Fix multicast operation that I broke in previous commit.
- Do not enable multicast hash lookup if no multicast addresses
  were configured or if promisc mode is enabled.

15 years ago- Fix promisc/multicast/broadcast parameters setting by introducing the
stas [Sun, 10 May 2009 08:54:10 +0000 (08:54 +0000)]
- Fix promisc/multicast/broadcast parameters setting by introducing the
  new ate_rxfilter function to set requested parameters. Use this function
  on parameters change rather than  reinitializing the chip.

15 years agoAbort any scan on a fatal firmware. ic_scan_curchan is overridden to perform
thompsa [Sun, 10 May 2009 02:44:19 +0000 (02:44 +0000)]
Abort any scan on a fatal firmware. ic_scan_curchan is overridden to perform
the scan in firmware and this relies on the firmware to wake up the scan task
on completion.

15 years agoCancel the scan when going to INIT state. Should do this for other states here
thompsa [Sun, 10 May 2009 02:21:19 +0000 (02:21 +0000)]
Cancel the scan when going to INIT state. Should do this for other states here
too as once the protocol newstate handler runs the scan has always ended.

15 years ago- Use "device\t" and "options \t" for consistency.
kuriyama [Sun, 10 May 2009 00:00:25 +0000 (00:00 +0000)]
- Use "device\t" and "options \t" for consistency.

15 years agoDescribe the topology_spec sysctl and do some minor adjustments to
ivoras [Sat, 9 May 2009 20:18:20 +0000 (20:18 +0000)]
Describe the topology_spec sysctl and do some minor adjustments to
relating documentation.

Reviewed by: brooks (older version)
Approved by: gnn (mentor)

15 years agohookup iwnfw.4
sam [Sat, 9 May 2009 19:44:23 +0000 (19:44 +0000)]
hookup iwnfw.4

15 years agopush wme parameter setting to the taskq thread; the update callback from
sam [Sat, 9 May 2009 19:22:37 +0000 (19:22 +0000)]
push wme parameter setting to the taskq thread; the update callback from
net80211 can happen from the ithread and submitting the fw cmd requires
a sleepable context

15 years agorestore ref to iwnfw
sam [Sat, 9 May 2009 19:19:30 +0000 (19:19 +0000)]
restore ref to iwnfw

15 years agoiwn firmware
sam [Sat, 9 May 2009 19:19:19 +0000 (19:19 +0000)]
iwn firmware

15 years agoSpell NULL properly, use (void) rather than () for functions with no
imp [Sat, 9 May 2009 19:08:22 +0000 (19:08 +0000)]
Spell NULL properly, use (void) rather than () for functions with no
parameters.  Mark two items as static that aren't used elsewhere...

15 years agoClean up <sys/ioctl.h>.
ed [Sat, 9 May 2009 19:01:24 +0000 (19:01 +0000)]
Clean up <sys/ioctl.h>.

- Just use #error when including <sys/ioctl.h> in the kernel. Code
  hasn't used this header for years now and probably doesn't compile
  anyway, because of -Werror.

- Get rid of struct ttysize, TIOCGSIZE and TIOCSSIZE. All code nowadays
  use both TIOC[GS]SIZE and TIOC[GS]WINSZ. Because we have other popular
  systems that don't implement the first, it's of little use to support
  interfaces nowadays.

15 years agoRetire kern.vm.kmem.size. It was marked as obsolete prior to 5.2, so
imp [Sat, 9 May 2009 19:00:47 +0000 (19:00 +0000)]
Retire kern.vm.kmem.size.  It was marked as obsolete prior to 5.2, so
it can go.

15 years agoMakefile to record the technique by which the .uu files are generated
sam [Sat, 9 May 2009 18:54:18 +0000 (18:54 +0000)]
Makefile to record the technique by which the .uu files are generated
from the Intel-distributed .fw files

15 years agoupdate to v3.1
sam [Sat, 9 May 2009 18:53:07 +0000 (18:53 +0000)]
update to v3.1

15 years agoRemove bogus comment.
imp [Sat, 9 May 2009 18:50:01 +0000 (18:50 +0000)]
Remove bogus comment.

15 years agoImplement RFC 5095 more fully. Rather than marking this no-op code as
imp [Sat, 9 May 2009 18:25:58 +0000 (18:25 +0000)]
Implement RFC 5095 more fully.  Rather than marking this no-op code as
BURN_BRIDGES, just remove it.  Adjust comments.

Reviewed by: dwhite, emaste, battlez

15 years agoRemove reference to nonexistant manpage.
roberto [Sat, 9 May 2009 18:09:18 +0000 (18:09 +0000)]
Remove reference to nonexistant manpage.

PR:     ports/131898
Submitted by:     Niclas Zeising <niclas.zeising@gmail.com>
Prodded by:         sbruno

15 years agoDo not embed struct ucred into larger netcred parent structures.
kan [Sat, 9 May 2009 18:09:17 +0000 (18:09 +0000)]
Do not embed struct ucred into larger netcred parent structures.

Credential might need to hang around longer than its parent and be used
outside of mnt_explock scope controlling netcred lifetime. Use separate
reference-counted ucred allocated separately instead.

While there, extend mnt_explock coverage in vfs_stdexpcheck and clean-up
some unused declarations in new NFS code.

Reported by: John Hickey
PR: kern/133439
Reviewed by: dfr, kib

15 years agoUnmark the ASR Compat IOCTLs as BURN_BRIDGES, per scottl@
imp [Sat, 9 May 2009 17:47:42 +0000 (17:47 +0000)]
Unmark the ASR Compat IOCTLs as BURN_BRIDGES, per scottl@

15 years agoTTYs don't necessarily use /dev/ttyxx.
ed [Sat, 9 May 2009 16:42:57 +0000 (16:42 +0000)]
TTYs don't necessarily use /dev/ttyxx.

Submitted by: csjp

15 years agoWork around non-standard behaviour of rpcbind in some versions of Linux (FC4?).
dfr [Sat, 9 May 2009 15:41:22 +0000 (15:41 +0000)]
Work around non-standard behaviour of rpcbind in some versions of Linux (FC4?).

Submitted by: zachary dot loafman at isilon dot com

15 years agoAdd support for old TTY ioctls to kdump.
ed [Sat, 9 May 2009 15:09:40 +0000 (15:09 +0000)]
Add support for old TTY ioctls to kdump.

At first I allowed ioctl_compat.h to be included, but it just returned
an empty file. I had to do this, to keep kdump happy. I really want to
raise a compiler error when including this header, so now it will just
throw an error if you don't set COMPAT_43TTY.

15 years agoFix a race involving vnode_pager_input_smlfs(). Specifically, in the case
alc [Sat, 9 May 2009 08:30:44 +0000 (08:30 +0000)]
Fix a race involving vnode_pager_input_smlfs().  Specifically, in the case
that vnode_pager_input_smlfs() zeroes the page, it should not mark the page
as valid until after the page is zeroed.  Otherwise, the page could be
mapped for read access (e.g., by vm_map_pmap_enter()) before the page is
zeroed.  Reviewed by: tegge

Eliminate gratuitous clearing of the page's dirty mask by
vnode_pager_input_smlfs().  Instead, assert that the page is clean.
Reviewed by: tegge

Eliminate some blank lines.

Eliminate pointless calls to pmap_clear_modify() and vm_page_undirty() from
vnode_pager_input_old().  The page is not mapped.  Therefore, it cannot have
any page table entries that are modified.

Eliminate an incorrect comment from vnode_pager_generic_getpages().

15 years agofix atomic.S rename and vimage breakage
kmacy [Sat, 9 May 2009 05:45:13 +0000 (05:45 +0000)]
fix atomic.S rename and vimage breakage

The latter was pointed out by Artem Belevich

15 years agoatomic.S has been renamed opensolaris_atomic.S to avoid collisions
kmacy [Sat, 9 May 2009 05:39:35 +0000 (05:39 +0000)]
atomic.S has been renamed opensolaris_atomic.S to avoid collisions

15 years agoConvert IPFW_DEFAULT_TO_ACCEPT into a loader tunable
jhb [Sat, 9 May 2009 05:07:36 +0000 (05:07 +0000)]
Convert IPFW_DEFAULT_TO_ACCEPT into a loader tunable
'net.inet.ip.fw.default_to_accept'.  The current value can also be queried
via a read-only sysctl of the same name.

Requested by: plosher
MFC after: 1 week

15 years ago- rename atomic.S and crc32.c to avoid collisions when linking zfs in to the kernel
kmacy [Sat, 9 May 2009 01:45:55 +0000 (01:45 +0000)]
- rename atomic.S and crc32.c to avoid collisions when linking zfs in to the kernel
- update Makefile
- ifdef out acl_{alloc, free}, they aren't used by zfs and conflict with existing in-kernel routines

15 years agoUpdate to less v429.
delphij [Sat, 9 May 2009 01:35:27 +0000 (01:35 +0000)]
Update to less v429.

15 years ago- Fixed incorrect packet length problem caused be earlier change to
davidch [Fri, 8 May 2009 22:20:45 +0000 (22:20 +0000)]
- Fixed incorrect packet length problem caused be earlier change to
  support ZERO_COPY_SOCKETS.
- Created #define for context initialization retry count.

MFC after: 1 week

15 years agoRegenerate system call tables to use SVN ids.
ed [Fri, 8 May 2009 20:16:04 +0000 (20:16 +0000)]
Regenerate system call tables to use SVN ids.

15 years agoRegenerate ibcs2 system call table.
ed [Fri, 8 May 2009 20:08:43 +0000 (20:08 +0000)]
Regenerate ibcs2 system call table.

15 years agoBurn TTY ioctl bridges in compat layers.
ed [Fri, 8 May 2009 20:06:37 +0000 (20:06 +0000)]
Burn TTY ioctl bridges in compat layers.

I really don't want any pieces of code to include ioctl_compat.h, so let
the ibcs2 and svr4 compat leave sgtty alone. If they want to support
sgtty, they should emulate it on top of termios, not sgtty.

The code has been marked with BURN_BRIDGES for a long time. ibcs2 and
svr4 are not really popular pieces of code anyway.

15 years agoUse log(9) for debug and status messages and hide some of the details with
dfr [Fri, 8 May 2009 15:14:52 +0000 (15:14 +0000)]
Use log(9) for debug and status messages and hide some of the details with
macros to allow for future flexibility in logging.

Submitted by: zachary dot loafman at isilon dot com