]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
14 years agoWhen the MNTK_EXTENDED_SHARED mount option was added, some filesystems were
John Baldwin [Fri, 16 Jul 2010 19:20:20 +0000 (19:20 +0000)]
When the MNTK_EXTENDED_SHARED mount option was added, some filesystems were
changed to defer the setting of VN_LOCK_ASHARE() (which clears LK_NOSHARE
in the vnode lock's flags) until after they had determined if the vnode was
a FIFO.  This occurs after the vnode has been inserted a VFS hash or some
similar table, so it is possible for another thread to find this vnode via
vget() on an i-node number and block on the vnode lock.  If the lockmgr
interlock (vnode interlock for vnode locks) is not held when clearing the
LK_NOSHARE flag, then the lk_flags field can be clobbered.  As a result
the thread blocked on the vnode lock may never get woken up.  Fix this by
holding the vnode interlock while modifying the lock flags in this case.

MFC after: 3 days

14 years agoNervously welcoming myself to the family.
Joseph S. Atkinson [Fri, 16 Jul 2010 18:06:19 +0000 (18:06 +0000)]
Nervously welcoming myself to the family.

Add myself to all the appropriate places, upload my pgp key, and announce
my arrival to the FreeBSD world.

Approved by: wxs@ (mentor)

14 years agoMake legacy ATA emulation detection more strict. This should fix false
Alexander Motin [Fri, 16 Jul 2010 17:27:43 +0000 (17:27 +0000)]
Make legacy ATA emulation detection more strict. This should fix false
positive legacy detection and attach failure/panic for Marvell 88SX6141
controller and potentially some others.

PR: kern/145064

14 years agoRename documentation source files to avoid filesystem case and restricted
Ed Maste [Fri, 16 Jul 2010 17:01:36 +0000 (17:01 +0000)]
Rename documentation source files to avoid filesystem case and restricted
filename issues when checking out the source on other operating systems.

I've verified the generated paper.ascii is identical before and after the
change.

14 years agoImprove interrupt setup errors handling.
Alexander Motin [Fri, 16 Jul 2010 10:05:00 +0000 (10:05 +0000)]
Improve interrupt setup errors handling.

14 years agoMove common macros into asm.h. Replace MIPS_CPU_NOP_DELAY with
Warner Losh [Fri, 16 Jul 2010 06:35:17 +0000 (06:35 +0000)]
Move common macros into asm.h.  Replace MIPS_CPU_NOP_DELAY with
HAZARD_DELAY.  Move HAZARD_DELAY and ITLBNOPFIX into asm.h, for
possible later optimization...

Reviewed by: jmallet, jchandra

14 years agomachine/cpu.h isn't appropriate for this file,so remove it
Warner Losh [Fri, 16 Jul 2010 06:32:38 +0000 (06:32 +0000)]
machine/cpu.h isn't appropriate for this file,so remove it

14 years agoThis file appears not to be used.
Warner Losh [Fri, 16 Jul 2010 06:31:37 +0000 (06:31 +0000)]
This file appears not to be used.

14 years agoUse #define for get_cyclecount rather than inline function.
Warner Losh [Fri, 16 Jul 2010 06:09:51 +0000 (06:09 +0000)]
Use #define for get_cyclecount rather than inline function.
mips_rd_count() isn't defined in userland, and cpu.h is included there
in alias_scpt.h (maybe they don't need it in the first place).

14 years agoSimplify AcpiOsReadPort() and AcpiOsWritePort() with iodev_read_*() and
Jung-uk Kim [Fri, 16 Jul 2010 04:27:38 +0000 (04:27 +0000)]
Simplify AcpiOsReadPort() and AcpiOsWritePort() with iodev_read_*() and
iodev_write_*().  This removes unnecessary uses of temporary macros as well.
There is no functional change after this (verified with md5(1) on amd64).

14 years agoFix KASSERT() messages to reflect reality.
Adrian Chadd [Fri, 16 Jul 2010 04:26:37 +0000 (04:26 +0000)]
Fix KASSERT() messages to reflect reality.

14 years agoUse pmap_mapdev()/pmap_unmapdev() to map device memory instead of using
Jung-uk Kim [Fri, 16 Jul 2010 03:59:50 +0000 (03:59 +0000)]
Use pmap_mapdev()/pmap_unmapdev() to map device memory instead of using
AcpiOsMapMemory()/AcpiOsUnmapMemory() (-> pmap_mapbios()/pmap_unmapbios())
for AcpiOsReadMemory() and AcpiOsWriteMemory().  Although they do not sound
too obvious, these functions are exclusively used to access memory mapped
IO in ACPICA.

14 years agoDelete comments related to soft clock interrupts that don't apply
Rick Macklem [Fri, 16 Jul 2010 01:44:49 +0000 (01:44 +0000)]
Delete comments related to soft clock interrupts that don't apply
to the FreeBSD port of the experimental NFSv4 server.

Submitted by: zack.kirsch at isilon.com
MFC after: 2 weeks

14 years agoUse Miscellaneous Configuration Register bit definition instead of
Pyun YongHyeon [Thu, 15 Jul 2010 23:34:58 +0000 (23:34 +0000)]
Use Miscellaneous Configuration Register bit definition instead of
magic number.

14 years agoSimple compatibility hacks for building on older systems where
Warner Losh [Thu, 15 Jul 2010 23:32:53 +0000 (23:32 +0000)]
Simple compatibility hacks for building on older systems where
MACHINE_CPUARCH isn't defined.  I believe that this will cover all
options.

I didn't define it in kern.mk because $M is set to MACHINE_CPUARCH and
then is expanded for the genassym.o rule in kern.post.mk and kern.mk
is included after this, so the expansion isn't quite right.  I think
this is a bug in make, but don't have the time to track it to ground
(and even if I did, fixing it would require a MFC of the change to the
very old systems we're targetting with this fix).

14 years agoIf there is any pending sleep request, disallow entering S5 state.
Jung-uk Kim [Thu, 15 Jul 2010 23:24:06 +0000 (23:24 +0000)]
If there is any pending sleep request, disallow entering S5 state.
Otherwise, bad things may happen. ;-)

14 years agoWhen we are not switching VTs, just mark all buffer to be updated.
Jung-uk Kim [Thu, 15 Jul 2010 23:11:51 +0000 (23:11 +0000)]
When we are not switching VTs, just mark all buffer to be updated.

14 years agoPut warnings out to stderr rather than stdout.
Warner Losh [Thu, 15 Jul 2010 22:28:19 +0000 (22:28 +0000)]
Put warnings out to stderr rather than stdout.

MFC after: 3 days

14 years agoBetter description of this file
Warner Losh [Thu, 15 Jul 2010 21:56:45 +0000 (21:56 +0000)]
Better description of this file

14 years agoRetire td_syscalls now that it is no longer needed.
John Baldwin [Thu, 15 Jul 2010 20:24:37 +0000 (20:24 +0000)]
Retire td_syscalls now that it is no longer needed.

14 years ago- AcpiOsReadPciConfiguration() needs similar fixes as r209965 and r210129.
Jung-uk Kim [Thu, 15 Jul 2010 19:52:54 +0000 (19:52 +0000)]
- AcpiOsReadPciConfiguration() needs similar fixes as r209965 and r210129.
According to ACPICA User Guide and Programmer Reference, the read data must
be zero extended to fill the 64-bit return value even if the bit width of
the location is less than 64.
- Return error when 64-bit access is requested as we do not support 64-bit
PCI register access (yet).  XXX We may have to split it up into two 32-bit
accesses if it is really required.

14 years agoRetire the NFS access cache timestamp structure. It was used in VOP_OPEN()
John Baldwin [Thu, 15 Jul 2010 19:40:48 +0000 (19:40 +0000)]
Retire the NFS access cache timestamp structure.  It was used in VOP_OPEN()
to avoid sending multiple ACCESS/GETATTR RPCs during a single open()
between VOP_LOOKUP() and VOP_OPEN().  Now we always send the RPC in
VOP_LOOKUP() and not VOP_OPEN() in the cases that multiple RPCs could be
sent.

MFC after: 2 weeks

14 years agoMerge 208603, 209946, and 209948 to the new NFS client:
John Baldwin [Thu, 15 Jul 2010 19:21:48 +0000 (19:21 +0000)]
Merge 208603, 209946, and 209948 to the new NFS client:
Move attribute cache flushes from VOP_OPEN() to VOP_LOOKUP() to provide
more graceful recovery for stale filehandles and eliminate the need for
conditionally clearing the attribute cache in the !NMODIFIED case in
VOP_OPEN().

Reviewed by: rmacklem
MFC after: 2 weeks

14 years agoDon't copy and return a potentially unset buffer when jail_get fails.
Jamie Gritton [Thu, 15 Jul 2010 19:21:33 +0000 (19:21 +0000)]
Don't copy and return a potentially unset buffer when jail_get fails.

14 years agoDon't import parameter values in jail_getv, except for the search key.
Jamie Gritton [Thu, 15 Jul 2010 19:21:07 +0000 (19:21 +0000)]
Don't import parameter values in jail_getv, except for the search key.
Remove the internal jailparam_vlist, in favor of using variants of its
 logic separately in jail_setv and jail_getv.
Free the temporary parameter list and exported values in jail_setv
 and jail_getv.

Noted by: Stanislav Uzunchev
MFC after: 3 days

14 years agoMake svr4(4) version of poll(2) use the same limit of file descriptors as the
Edward Tomasz Napierala [Thu, 15 Jul 2010 18:44:58 +0000 (18:44 +0000)]
Make svr4(4) version of poll(2) use the same limit of file descriptors as the
usual poll(2) does, instead of checking resource limits.

14 years agoMove functions declaration to MI code, following implementation.
Alexander Motin [Thu, 15 Jul 2010 17:49:35 +0000 (17:49 +0000)]
Move functions declaration to MI code, following implementation.

14 years ago- AcpiOsReadMemory() needs similar fixes as r209965. [1]
Jung-uk Kim [Thu, 15 Jul 2010 17:11:49 +0000 (17:11 +0000)]
- AcpiOsReadMemory() needs similar fixes as r209965. [1]
According to ACPICA User Guide and Programmer Reference, the read data must
be zero extended to fill the 32-bit return value even if the bit width of
the port is less than 32.
- Remove 64-bit read/write from AcpiOsReadMemory() and AcpiOsWriteMemory().
These functions do not support 64-bit access (yet).  Clean up style nits
and unnecessary bit masking while I am here.

Reported by: Liu, Jinsong (jinsong dot liu at intel dot com) via
Lin Ming (ming dot m dot lin at intel dot com) [1]

14 years agoFix for 64 bit compilation.
Jayachandran C. [Thu, 15 Jul 2010 16:39:17 +0000 (16:39 +0000)]
Fix for 64 bit compilation.

RMI bootloader passes argv[] and envp[] as an array of 32 bit pointers.
Convert the pointers to correct pointer type before use.

14 years agoadd some modern stuff: SMP, SCHED_ULE, PREEMPTION
Luigi Rizzo [Thu, 15 Jul 2010 16:32:59 +0000 (16:32 +0000)]
add some modern stuff: SMP, SCHED_ULE, PREEMPTION
and support for VIMAGE

14 years agoOptimize pmap_remove()'s handling of PG_G mappings. Specifically,
Alan Cox [Thu, 15 Jul 2010 16:25:51 +0000 (16:25 +0000)]
Optimize pmap_remove()'s handling of PG_G mappings.  Specifically,
instead of calling pmap_invalidate_page() for each PG_G mapping, call
pmap_invalidate_range() for each range of PG_G mappings.  In addition,
eliminate a redundant call to pmap_invalidate_page().  Both
pmap_remove_pte() and pmap_remove_page() called pmap_invalidate_page()
when the mapping had the PG_G attribute.  Now, only pmap_remove_page()
calls pmap_invalidate_page().  Altogether, these changes eliminate 53%
of the TLB shootdowns for a "buildworld" on a ZFS file system.  On
FFS, the reduction is 3%.

MFC after: 6 weeks

14 years agoremove some conditional #ifdefs (no-op on FreeBSD);
Luigi Rizzo [Thu, 15 Jul 2010 14:43:12 +0000 (14:43 +0000)]
remove some conditional #ifdefs (no-op on FreeBSD);
run the timer routine on cpu 0.

14 years agowhitespace cleanup
Luigi Rizzo [Thu, 15 Jul 2010 14:41:59 +0000 (14:41 +0000)]
whitespace cleanup

14 years agosmall portability fix to build on linux/windows
Luigi Rizzo [Thu, 15 Jul 2010 14:41:06 +0000 (14:41 +0000)]
small portability fix to build on linux/windows

14 years agowhitespace fixes
Luigi Rizzo [Thu, 15 Jul 2010 14:37:59 +0000 (14:37 +0000)]
whitespace fixes

14 years agofix a comment and final empty line
Luigi Rizzo [Thu, 15 Jul 2010 14:37:02 +0000 (14:37 +0000)]
fix a comment and final empty line

14 years agobetter printing of headers when listing flows
Luigi Rizzo [Thu, 15 Jul 2010 14:34:56 +0000 (14:34 +0000)]
better printing of headers when listing flows

14 years agoA cosmetic change - don't output empty <flags>.
Ivan Voras [Thu, 15 Jul 2010 13:46:30 +0000 (13:46 +0000)]
A cosmetic change - don't output empty <flags>.

14 years agoRelax FDT_DTS_FILE validation (and unbreak world build).
Rafal Jaworowski [Thu, 15 Jul 2010 13:21:25 +0000 (13:21 +0000)]
Relax FDT_DTS_FILE validation (and unbreak world build).

Pointed out by: kib

14 years agoHandle RUN->ASSOC->RUN transition correctly, as in not trigger a
Bernhard Schmidt [Thu, 15 Jul 2010 11:52:20 +0000 (11:52 +0000)]
Handle RUN->ASSOC->RUN transition correctly, as in not trigger a
firmware error. Convert if statements to a switch statement while
I'm here.

Tested by: Benjamin Kaduk <kaduk at mit.edu>
MFC after: 2 weeks

14 years ago- Update 6000 firmware to 9.221.4.1
Bernhard Schmidt [Thu, 15 Jul 2010 11:26:07 +0000 (11:26 +0000)]
- Update 6000 firmware to 9.221.4.1
- Add 6050 firmware

MFC after: 2 weeks

14 years agoFix FDT_DTS_FILE parsing to properly retrieve its value. This unbreaks the
Rafal Jaworowski [Thu, 15 Jul 2010 10:49:07 +0000 (10:49 +0000)]
Fix FDT_DTS_FILE parsing to properly retrieve its value. This unbreaks the
'builddtb' target.

Make the processing more robust against non-existent kernel config files
(pointed out by imp@).

14 years agoAdd support for firmware images in "type-length-value" format.
Bernhard Schmidt [Thu, 15 Jul 2010 10:37:49 +0000 (10:37 +0000)]
Add support for firmware images in "type-length-value" format.

Obtained from: OpenBSD
MFC after: 2 weeks

14 years agoDetect active chains differently to work around a firmware bug which
Bernhard Schmidt [Thu, 15 Jul 2010 09:34:00 +0000 (09:34 +0000)]
Detect active chains differently to work around a firmware bug which
would mark non-existing chains as active.

Obtained from: OpenBSD
MFC after: 1 week

14 years ago- Add new IDs for 6000 series devices.
Bernhard Schmidt [Thu, 15 Jul 2010 09:30:54 +0000 (09:30 +0000)]
- Add new IDs for 6000 series devices.
- The 6000 series WiMAX devices need a separate firmware.
- The b-gen devices are not hooked because the hardware revision type
  is not know.

Obtained from: OpenBSD
MFC after: 1 week

14 years agoFix some small whitespace nits.
Bernhard Schmidt [Thu, 15 Jul 2010 08:05:20 +0000 (08:05 +0000)]
Fix some small whitespace nits.

MFC after: 3 days

14 years agoRemove duplicate vendor:device entry.
Bernhard Schmidt [Thu, 15 Jul 2010 07:45:37 +0000 (07:45 +0000)]
Remove duplicate vendor:device entry.

MFC after: 3 days

14 years agoMove TLB definitions to tlb.h
Warner Losh [Thu, 15 Jul 2010 03:56:08 +0000 (03:56 +0000)]
Move TLB definitions to tlb.h

14 years agoThis file has been unused for a while now...
Warner Losh [Thu, 15 Jul 2010 03:36:50 +0000 (03:36 +0000)]
This file has been unused for a while now...

14 years agoUnbreak xz (liblzma) on strong-aligned architectures (and without
Marcel Moolenaar [Thu, 15 Jul 2010 03:11:04 +0000 (03:11 +0000)]
Unbreak xz (liblzma) on strong-aligned architectures (and without
emulation of misaligned memory accesses). We cannot map the unaligned
memory access functions to the ones used for aligned accesses, so do
not define TUKLIB_FAST_UNALIGNED_ACCESS.

14 years agoThis patch fixes a bug in the experimental NFSv4 server where it
Rick Macklem [Thu, 15 Jul 2010 03:02:10 +0000 (03:02 +0000)]
This patch fixes a bug in the experimental NFSv4 server where it
released a reference count on nfsv4rootfs_lock erroneously when
administrative revocation of state was done.

Submitted by: zack.kirsch at isilon.com
MFC after: 2 weeks

14 years agoSanitize CFLAGS and add WARNS?=2. The hardcoding of -Os -fPIC
Marcel Moolenaar [Thu, 15 Jul 2010 02:22:50 +0000 (02:22 +0000)]
Sanitize CFLAGS and add WARNS?=2. The hardcoding of -Os -fPIC
as compiler options resulted in an invalid executable on PowerPC.

14 years agoRemove unused stuff from cpu.h.
Warner Losh [Thu, 15 Jul 2010 01:58:20 +0000 (01:58 +0000)]
Remove unused stuff from cpu.h.
Move inappropriate stuff in cpu.h elsewhere:
{s,g}et_intr_mask -> md_var.h
num_tlbentries -> tlb.h
Remove #define clockframe trapframe and fix clock, which was the only place
this was used.
All the rest of this stuff was unused.

# we're not quite minimal yet, since we duplicate a few status register things
# here...

Inspired by: bde@

14 years agoWe don't need sys/cdefs.h for __CONCAT here.
Warner Losh [Thu, 15 Jul 2010 01:55:28 +0000 (01:55 +0000)]
We don't need sys/cdefs.h for __CONCAT here.

14 years agoRemove i386-ish sysctls. Also, make the bootinfo sysctl OID_AUTO.
Warner Losh [Thu, 15 Jul 2010 01:53:17 +0000 (01:53 +0000)]
Remove i386-ish sysctls.  Also, make the bootinfo sysctl OID_AUTO.

14 years agoGet powerpc64 install CDs a little closer: some powerpc things should be
Nathan Whitehorn [Thu, 15 Jul 2010 01:48:30 +0000 (01:48 +0000)]
Get powerpc64 install CDs a little closer: some powerpc things should be
used on powerpc64 as well (build scripts, for instance), and the MFS root
must be adjusted upward on powerpc64 as well.

This doesn't completely solve things: sysinstall stil needs some
persuasion to deal with a default kernel named GENERIC64.

14 years agoRemove one layer of indirection. No need to call cpu_throw which then
Warner Losh [Thu, 15 Jul 2010 01:47:47 +0000 (01:47 +0000)]
Remove one layer of indirection.  No need to call cpu_throw which then
calls mips_cpu_call via an obfuscated assembler call.  Instead, delete
the current cpu_throw, and rename mips_cpu_throw to cpu_throw.  This
is nicer to the cache on each context switch (since fixed jumps can be
prefected, while jumps through a register can't).  Incidentally, it
also saves about 5 or 6 instructions.

Reviewed by: jmallet@

14 years agoAdd -p option to have pom(6) print just the percentage, without extra text.
Ed Maste [Thu, 15 Jul 2010 00:16:04 +0000 (00:16 +0000)]
Add -p option to have pom(6) print just the percentage, without extra text.

PR: bin/51070
Submitted by: Matt Koivisto mkoivisto sandvine com
MFC after: 1 week

14 years agoRemove more extraneous ;s.
Ed Maste [Thu, 15 Jul 2010 00:04:14 +0000 (00:04 +0000)]
Remove more extraneous ;s.

14 years agosh: There cannot be a TNOT in simplecmd(), remove checks.
Jilles Tjoelker [Wed, 14 Jul 2010 22:31:45 +0000 (22:31 +0000)]
sh: There cannot be a TNOT in simplecmd(), remove checks.

simplecmd() only handles simple commands and function definitions, neither
of which involves the ! keyword. The initial token on entry to simplecmd()
is one of the following: TSEMI, TAND, TOR, TNL, TEOF, TWORD, TRP.

14 years agoRemove enabling Data FIFO protection with indirect memory access.
Pyun YongHyeon [Wed, 14 Jul 2010 21:47:49 +0000 (21:47 +0000)]
Remove enabling Data FIFO protection with indirect memory access.
r165114 added that code and that change ignored the same logic
committed in r135772. In addition, data FIFO protection should be
selectively enabled instead of applying to all PCIe devices.
While I'm here add BCM5785 to devices that do not require this
fix.

14 years agoRename timeevents.c to kern_clocksource.c.
Alexander Motin [Wed, 14 Jul 2010 18:43:27 +0000 (18:43 +0000)]
Rename timeevents.c to kern_clocksource.c.

Suggested by: jhb@

14 years ago- Sort list of trace points.
John Baldwin [Wed, 14 Jul 2010 18:28:46 +0000 (18:28 +0000)]
- Sort list of trace points.
- Note that 'y' (sysctl requests) is in the default set of trace points.

MFC after: 3 days

14 years agoRework the SMBIOS table walker to make it operate like other table walkers
John Baldwin [Wed, 14 Jul 2010 18:06:21 +0000 (18:06 +0000)]
Rework the SMBIOS table walker to make it operate like other table walkers
and remove a buffer overflow:
- Remove the array of per-type dispatch functions.  Instead, pass each
  structure to a single callback.  The callback should check the type of
  each table entry to take appropriate action.  This matches the behavior
  of other table walkers such as for the MP Table and MADT.
- Don't attempt to save an array of string pointers for each structure
  entry.  Instead, just skip the strings.  If this code is reused to
  provide a generic SMBIOS table walker in the future we could provide
  a method that looks up a specific string N for a given structure record
  instead of pre-populating an array of pointers.  This fixes a buffer
  overflow for structure entries with more than 20 strings.

PR: kern/148546
Reported by: Spencer Minear @ McAfee
MFC after: 3 days

14 years agoFix spelling.
Attilio Rao [Wed, 14 Jul 2010 17:46:44 +0000 (17:46 +0000)]
Fix spelling.

Sponsored by: Sandvine Incorporated
Submitted by: b.f. <bf1783 at googlemail dot com>
MFC after: 2 weeks
X-MFC: 210063

14 years ago- Document layout of KTR_STRUCT payload in a comment.
John Baldwin [Wed, 14 Jul 2010 17:38:01 +0000 (17:38 +0000)]
- Document layout of KTR_STRUCT payload in a comment.
- Simplify ktrstruct() calling convention by having ktrstruct() use
  strlen() rather than requiring the caller to hand-code the length of
  constant strings.

MFC after: 1 month

14 years agoFix the way the segments are included in the gcore outputs (with the
Attilio Rao [Wed, 14 Jul 2010 17:16:25 +0000 (17:16 +0000)]
Fix the way the segments are included in the gcore outputs (with the
default invokation):
- Right now if segments are not writable are not included. Remove this.
- Right now if a segment is mapped with NOCORE the check is not honoured.
  Change this by checking the newly added flag, from libutil,
  KVME_FLAG_NOCOREDUMP.

Besides that, add a new flag (-f) that forces a 'full' dump of all the
segments excluding just the malformed ones. This might be used very
carefully as, among the reported segments, there could be memory
mapped areas that could be vital to program execution.

Sponsored by: Sandvine Incorporated
Discussed with: kib
Reviewed by: emaste
Tested by: Sandvine Incorporated
MFC after: 2 weeks

14 years agoUpdate some release version numbers. The patch is based on the one
Benedict Reuschling [Wed, 14 Jul 2010 16:01:43 +0000 (16:01 +0000)]
Update some release version numbers. The patch is based on the one
from the PR, but the version numbers reflect the newer ones from
http://security.freebsd.org/#sup

PR:             docs/145227
Submitted by:   Glen Barber (glen dot j dot barber at gmail dot com)
Reviewed by:    cperciva
Mentored by:    jkois
MFC after:      1 week

14 years agoDo not bzero() NULL pointer on malloc() error.
Alexander Motin [Wed, 14 Jul 2010 15:14:00 +0000 (15:14 +0000)]
Do not bzero() NULL pointer on malloc() error.

Submitted by: Dmitry Luhtionov

14 years agoCorrect logic bug in aicasm's undefined register bit access detection code.
Justin T. Gibbs [Wed, 14 Jul 2010 14:31:18 +0000 (14:31 +0000)]
Correct logic bug in aicasm's undefined register bit access detection code.

The code in question verifies that all register write operations only change
bits that are defined (in the register definition file) for that effected
register.  The bug effectively disabled this checking.

o Fix the check by testing the opcode against all supported read ("and" based)
  operands.

o Add missing bit definitions to the aic7xxx and aic79xx register definition
  files so that the warning (treated as a fatal error) does not spuriously
  fire.

Reported by: Pawel Worach <pawel.worach@gmail.com>
MFC after: 1 week

14 years agoMove timeevents.c to MI code, as it is not x86-specific. I already have
Alexander Motin [Wed, 14 Jul 2010 13:31:27 +0000 (13:31 +0000)]
Move timeevents.c to MI code, as it is not x86-specific. I already have
it working on Marvell ARM SoCs, and it would be nice to unify timer code
between more platforms.

14 years agoProvide initial man pages for Flattened Device Tree support and the common
Rafal Jaworowski [Wed, 14 Jul 2010 13:03:00 +0000 (13:03 +0000)]
Provide initial man pages for Flattened Device Tree support and the common
drivers.

Sponsored by: The FreeBSD Foundation

14 years agoRemove some unneeded includes. Code now can be built on ARM.
Alexander Motin [Wed, 14 Jul 2010 10:49:14 +0000 (10:49 +0000)]
Remove some unneeded includes. Code now can be built on ARM.

14 years agoFixed cache size decoding read from a label.
Ruslan Ermilov [Wed, 14 Jul 2010 08:22:00 +0000 (08:22 +0000)]
Fixed cache size decoding read from a label.

PR: kern/144732
Submitted by: Eugene Grosbein
MFC after: 3 days

14 years agoMFp4 180933:
Pawel Jakub Dawidek [Wed, 14 Jul 2010 06:37:43 +0000 (06:37 +0000)]
MFp4 180933:

Initialize rw_count properly so that zdb(8) doesn't trigger assertion in
rw_enter():

ASSERT(rwlp->rw_count == 0);

While here, assert that rw_count is 0 when destroying the lock.

MFC after: 1 week

14 years agoUse cpuregs.h spellings over the cpu.h spellings.
Warner Losh [Wed, 14 Jul 2010 00:52:29 +0000 (00:52 +0000)]
Use cpuregs.h spellings over the cpu.h spellings.

14 years agoImport preliminary support for Atmel AT91SAM9G20 cpu, and the Hot-e HL201.
Olivier Houchard [Wed, 14 Jul 2010 00:48:53 +0000 (00:48 +0000)]
Import preliminary support for Atmel AT91SAM9G20 cpu, and the Hot-e HL201.
This fine work was done by Yohanes Nugroho <yohanes a gmail dot com>
Many thanks to John Nicholls and Thinlinx for providing sample hardware.

14 years agoRemove the unused part of cpu.h now that the rest of the tree has been
Warner Losh [Wed, 14 Jul 2010 00:47:37 +0000 (00:47 +0000)]
Remove the unused part of cpu.h now that the rest of the tree has been
transitioned to use cpuregs.h spellings.  Now we're only 4x too big,
according to the bde-ometer.

14 years agoPrefer the cpuregs.h spellings of register and bit names over cpu.h.
Warner Losh [Wed, 14 Jul 2010 00:41:22 +0000 (00:41 +0000)]
Prefer the cpuregs.h spellings of register and bit names over cpu.h.

14 years agoAdd support for packages.
Warner Losh [Tue, 13 Jul 2010 23:47:12 +0000 (23:47 +0000)]
Add support for packages.

PR: 148584
Submitted by: John Hixon

14 years agoFor the experimental NFSv4 client, make sure that attributes that
Rick Macklem [Tue, 13 Jul 2010 23:14:39 +0000 (23:14 +0000)]
For the experimental NFSv4 client, make sure that attributes that
predate the issue of a delegation are not cached once the delegation
is held. This is necessary, since cached attributes remain valid
while the delegation is held.

MFC after: 2 weeks

14 years agoRemove obsolete code that sets SHMMAXPGS to a tiny value by default
Nathan Whitehorn [Tue, 13 Jul 2010 23:10:55 +0000 (23:10 +0000)]
Remove obsolete code that sets SHMMAXPGS to a tiny value by default
on PowerPC.

14 years agoFor the experimental NFSv4 client, do not use cached attributes
Rick Macklem [Tue, 13 Jul 2010 23:07:32 +0000 (23:07 +0000)]
For the experimental NFSv4 client, do not use cached attributes
that were invalidated, even when a delegation for the file is held.

MFC after: 2 weeks

14 years agoFix a bogus comment that mentions lru lists that don't exist.
Rick Macklem [Tue, 13 Jul 2010 22:44:14 +0000 (22:44 +0000)]
Fix a bogus comment that mentions lru lists that don't exist.

Reported by: zack.kirsch at isilon.com
MFC after: 2 weeks

14 years agounion cpuprid is also unused now
Warner Losh [Tue, 13 Jul 2010 22:36:45 +0000 (22:36 +0000)]
union cpuprid is also unused now

14 years agoAdd INFO config register from mips32/64 land
Warner Losh [Tue, 13 Jul 2010 22:35:09 +0000 (22:35 +0000)]
Add INFO config register from mips32/64 land

14 years agoRemove redunant machine/cpuregs.h include.
Warner Losh [Tue, 13 Jul 2010 22:30:27 +0000 (22:30 +0000)]
Remove redunant machine/cpuregs.h include.
Also, spell things like in machine/cpuregs.h instead of machine/cpu.h.

14 years agoTemp hack to N32 kernel: turn off debugger since n32 is too weird for ddb
Warner Losh [Tue, 13 Jul 2010 22:27:19 +0000 (22:27 +0000)]
Temp hack to N32 kernel: turn off debugger since n32 is too weird for ddb

14 years agoAdd GENERIC kernel config for powerpc64.
Nathan Whitehorn [Tue, 13 Jul 2010 21:24:08 +0000 (21:24 +0000)]
Add GENERIC kernel config for powerpc64.

14 years agoConnect powerpc64 to the build. It is not presently part of make universe,
Nathan Whitehorn [Tue, 13 Jul 2010 21:19:59 +0000 (21:19 +0000)]
Connect powerpc64 to the build. It is not presently part of make universe,
which will be added soon.

Reviewed by: imp

14 years agoAdd a 4 and 7 port USB hub from NEC.
Remko Lodder [Tue, 13 Jul 2010 20:09:14 +0000 (20:09 +0000)]
Add a 4 and 7 port USB hub from NEC.

PR: 148189
MFC after: 1 week

14 years agoInitialize a variable before its use.
Jung-uk Kim [Tue, 13 Jul 2010 19:58:06 +0000 (19:58 +0000)]
Initialize a variable before its use.

14 years agoPreallocate buffers for palette and state. Do not save DAC registers as
Jung-uk Kim [Tue, 13 Jul 2010 19:48:20 +0000 (19:48 +0000)]
Preallocate buffers for palette and state.  Do not save DAC registers as
we reset DAC mode and restore palette data while we are resuming always.

14 years agoPrefer PCIR_BAR macro over BGE_PCI_BAR0.
Pyun YongHyeon [Tue, 13 Jul 2010 19:45:40 +0000 (19:45 +0000)]
Prefer PCIR_BAR macro over BGE_PCI_BAR0.

14 years agoFix error message for jumbo buffer allocation failure.
Pyun YongHyeon [Tue, 13 Jul 2010 19:42:55 +0000 (19:42 +0000)]
Fix error message for jumbo buffer allocation failure.

14 years agostyle.
Pyun YongHyeon [Tue, 13 Jul 2010 19:39:51 +0000 (19:39 +0000)]
style.

14 years agoFix a panic brought about by writing an MSR without a proper mask.
George V. Neville-Neil [Tue, 13 Jul 2010 19:37:45 +0000 (19:37 +0000)]
Fix a panic brought about by writing an MSR without a proper mask.
All of the necessary wrmsr calls are now preceded by a rdmsr
and we leave the reserved bits alone.
Document the bits in the relevant registers for future reference.

Tested by: mdf
MFC after: 1 week

14 years agoMake bge_stop_fw() static.
Pyun YongHyeon [Tue, 13 Jul 2010 19:33:46 +0000 (19:33 +0000)]
Make bge_stop_fw() static.
While I'm here use ANSI function definitions.

14 years agoFix build on architectures where PAGE_SIZE is a long (sparc64, powerpc).
Nathan Whitehorn [Tue, 13 Jul 2010 19:27:20 +0000 (19:27 +0000)]
Fix build on architectures where PAGE_SIZE is a long (sparc64, powerpc).

14 years agoDefine break value for ddb.
Warner Losh [Tue, 13 Jul 2010 17:24:30 +0000 (17:24 +0000)]
Define break value for ddb.
Use int32/intptr casts for exception vector names.
Define MIPS_SR_INT_MASK again
Change MIPS_XKPHYS_CCA_* to MIPS_CCA_* since we can use them in many contexts
Minor gratuitous whitespace churn