]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
13 years agoFor FreeBSD and Linux use awk's toupper() function.
pjd [Fri, 6 Aug 2010 21:53:50 +0000 (21:53 +0000)]
For FreeBSD and Linux use awk's toupper() function.

Suggested by: ed

13 years agoConvert file system type to upper case.
pjd [Fri, 6 Aug 2010 21:02:53 +0000 (21:02 +0000)]
Convert file system type to upper case.

13 years agoPut the early setting of the MAC type back, its
jfv [Fri, 6 Aug 2010 20:55:49 +0000 (20:55 +0000)]
Put the early setting of the MAC type back, its
removal resulted in broken code in MSIX setup.

13 years agoAdd tests for mknod(2).
pjd [Fri, 6 Aug 2010 20:51:39 +0000 (20:51 +0000)]
Add tests for mknod(2).

Submitted by: Jan Senolt <senoltj@centrum.cz>
Submitted by: Milan Cermak <Milan.Cermak@Sun.COM>
Polished by: pjd

13 years agoAdd mknod(2) support.
pjd [Fri, 6 Aug 2010 20:48:10 +0000 (20:48 +0000)]
Add mknod(2) support.

Submitted by: Jan Senolt <senoltj@centrum.cz>
Submitted by: Milan Cermak <Milan.Cermak@Sun.COM>

13 years agoSort includes.
pjd [Fri, 6 Aug 2010 20:46:26 +0000 (20:46 +0000)]
Sort includes.

13 years agoObviously the option is known as COMPAT_LINUX32 on amd64.
bschmidt [Fri, 6 Aug 2010 20:18:16 +0000 (20:18 +0000)]
Obviously the option is known as COMPAT_LINUX32 on amd64.

PR: docs/149182
Pointed out by: Fernando <fernando.apesteguia at gmail.com>
MFC after: 3 days

13 years agoRemove unneeded functions from libcompat.
ed [Fri, 6 Aug 2010 20:13:36 +0000 (20:13 +0000)]
Remove unneeded functions from libcompat.

Erwin fired up a ports build a couple of weeks ago and it seems the
following functions are not used by any of the 20k ports we have, which
makes me believe they don't have any purpose. Just remove them.

13 years agoRemove stale reference to UT_NAMESIZE from <stdio.h>.
ed [Fri, 6 Aug 2010 19:35:40 +0000 (19:35 +0000)]
Remove stale reference to UT_NAMESIZE from <stdio.h>.

Spotted by: bde@

13 years agoopen(2) returns EOPNOTSUPP when trying to open a socket.
pjd [Fri, 6 Aug 2010 19:22:42 +0000 (19:22 +0000)]
open(2) returns EOPNOTSUPP when trying to open a socket.

13 years agoTest for EACCES also when opening FIFO or directory.
pjd [Fri, 6 Aug 2010 19:20:35 +0000 (19:20 +0000)]
Test for EACCES also when opening FIFO or directory.

13 years agoTest O_RDONLY|O_RDWR flags as potentially invalid.
pjd [Fri, 6 Aug 2010 19:19:14 +0000 (19:19 +0000)]
Test O_RDONLY|O_RDWR flags as potentially invalid.

13 years agoMake description readable.
pjd [Fri, 6 Aug 2010 19:18:19 +0000 (19:18 +0000)]
Make description readable.

13 years agoAdd missing -U argument to usage.
pjd [Fri, 6 Aug 2010 19:16:48 +0000 (19:16 +0000)]
Add missing -U argument to usage.

13 years agoImplement two new syscalls: bind(2) and connect(2) for operating on UNIX
pjd [Fri, 6 Aug 2010 19:16:20 +0000 (19:16 +0000)]
Implement two new syscalls: bind(2) and connect(2) for operating on UNIX
domain sockets.

13 years agoFix whitespace nits.
bschmidt [Fri, 6 Aug 2010 18:46:27 +0000 (18:46 +0000)]
Fix whitespace nits.

PR: conf/148989
Submitted by: pluknet <pluknet at gmail.com>
MFC after: 3 days

13 years agoFix a typo introduced in r210936 which broke build.
yongari [Fri, 6 Aug 2010 18:44:07 +0000 (18:44 +0000)]
Fix a typo introduced in r210936 which broke build.

13 years agoFigure which is the IO and MEM bars- do not assume that they are in
mjacob [Fri, 6 Aug 2010 17:27:00 +0000 (17:27 +0000)]
Figure which is the IO and MEM bars- do not assume that they are in
a fixed order.

PR: 149220
Obtained from: John Baldwin
MFC after: 1 month

13 years agoRemove unnecessary casting and simplify code. We are not there yet. ;-)
jkim [Fri, 6 Aug 2010 17:21:32 +0000 (17:21 +0000)]
Remove unnecessary casting and simplify code.  We are not there yet. ;-)

13 years agoAdd links for destroy_dev_drain, destroy_dev_sched, destroy_dev_sched_cb,
stefanf [Fri, 6 Aug 2010 16:50:48 +0000 (16:50 +0000)]
Add links for destroy_dev_drain, destroy_dev_sched, destroy_dev_sched_cb,
make_dev_cred and make_dev_credf.

13 years agoCorrect argument order of acpi_restorecpu(), which was forgotten in r210804.
jkim [Fri, 6 Aug 2010 15:59:00 +0000 (15:59 +0000)]
Correct argument order of acpi_restorecpu(), which was forgotten in r210804.

13 years agoAdd a new ipi_cpu() function to the MI IPI API that can be used to send an
jhb [Fri, 6 Aug 2010 15:36:59 +0000 (15:36 +0000)]
Add a new ipi_cpu() function to the MI IPI API that can be used to send an
IPI to a specific CPU by its cpuid.  Replace calls to ipi_selected() that
constructed a mask for a single CPU with calls to ipi_cpu() instead.  This
will matter more in the future when we transition from cpumask_t to
cpuset_t for CPU masks in which case building a CPU mask is more expensive.

Submitted by: peter, sbruno
Reviewed by: rookie
Obtained from: Yahoo! (x86)
MFC after: 1 month

13 years agoConsistently use architecture specific macros.
jkim [Fri, 6 Aug 2010 15:24:37 +0000 (15:24 +0000)]
Consistently use architecture specific macros.

13 years agoAdjust the interface type in the link layer socket address for vlan(4)
jhb [Fri, 6 Aug 2010 15:15:26 +0000 (15:15 +0000)]
Adjust the interface type in the link layer socket address for vlan(4)
interfaces to be a vlan (IFT_L2VLAN) rather than an Ethernet interface
(IFT_ETHER).  The code already fixed if_type in the ifnet causing some
places to report the interface as a vlan (e.g. arp -a output) and other
places to report the interface as Ethernet (getifaddrs(3)).  Now they
should all report IFT_L2VLAN.

Reviewed by: brooks
MFC after: 1 month

13 years agoEthernet vlan(4) interfaces have valid Ethernet link layer addresses but
jhb [Fri, 6 Aug 2010 15:09:21 +0000 (15:09 +0000)]
Ethernet vlan(4) interfaces have valid Ethernet link layer addresses but
use a different interface type (IFT_L2VLAN vs IFT_ETHER).  Treat IFT_L2VLAN
interfaces like IFT_ETHER interfaces when handling link layer addresses.

Reviewed by: syrinx (bsnmpd)
MFC after: 1 week

13 years agoAdd Xen to the list of virtual vendors. In the non PV (HVM) case this fixes
csjp [Fri, 6 Aug 2010 15:04:40 +0000 (15:04 +0000)]
Add Xen to the list of virtual vendors.  In the non PV (HVM) case this fixes
the virtualization detection successfully disabling the clflush instruction.
This fixes insta-panics for XEN hvm users when the hw.clflush_disable
tunable is -1 or 0 (-1 by default).

Discussed with: jhb

13 years agoFix allocation of multiple pages, which forgot to increase page number.
jkim [Fri, 6 Aug 2010 15:04:01 +0000 (15:04 +0000)]
Fix allocation of multiple pages, which forgot to increase page number.
Particularly, it caused "vm86_addpage: overlap" panics under VirtualBox.
Add a safety check before freeing memory while I am here.

13 years agoFix typos and spelling mistakes.
joel [Fri, 6 Aug 2010 14:33:42 +0000 (14:33 +0000)]
Fix typos and spelling mistakes.

13 years agoDisable sync cache for the Transcend Jetflash V90. It is more specific
kib [Fri, 6 Aug 2010 13:45:16 +0000 (13:45 +0000)]
Disable sync cache for the Transcend Jetflash V90. It is more specific
quirk over the general one for transcend sticks.

Submitted by: Mykola Dzham <i levsha me>
MFC after: 1 week

13 years agoTypo fixes
uqs [Fri, 6 Aug 2010 12:38:30 +0000 (12:38 +0000)]
Typo fixes

PR: docs/149314
Submitted by: olgeni
MFC after: 3 days

13 years ago- Some fixes to Ukranian catalog
gabor [Fri, 6 Aug 2010 10:34:48 +0000 (10:34 +0000)]
- Some fixes to Ukranian catalog

Submitted by: avg, Alex Kozlov <spam@rm-rf.kiev.ua>

13 years agoMark /dev/zero and /dev/null as eternal.
kib [Fri, 6 Aug 2010 09:47:48 +0000 (09:47 +0000)]
Mark /dev/zero and /dev/null as eternal.

In collaboration with: pho
MFC after: 1 month

13 years agoEnable shared lookups and externed shared ops for devfs.
kib [Fri, 6 Aug 2010 09:46:53 +0000 (09:46 +0000)]
Enable shared lookups and externed shared ops for devfs.

In collaboration with: pho
MFC after: 1 month

13 years agoAdd "show cdev" ddb command.
kib [Fri, 6 Aug 2010 09:44:01 +0000 (09:44 +0000)]
Add "show cdev" ddb command.

In collaboration with: pho
MFC after: 1 month

13 years agoAdd new make_dev_p(9) flag MAKEDEV_ETERNAL to inform devfs that created
kib [Fri, 6 Aug 2010 09:42:15 +0000 (09:42 +0000)]
Add new make_dev_p(9) flag MAKEDEV_ETERNAL to inform devfs that created
cdev will never be destroyed. Propagate the flag to devfs vnodes as
VV_ETERNVALDEV. Use the flags to avoid acquiring devmtx and taking a
thread reference on such nodes.

In collaboration with: pho
MFC after: 1 month

13 years agoFix issue reported by alc :
jchandra [Fri, 6 Aug 2010 09:25:42 +0000 (09:25 +0000)]
Fix issue reported by alc :

 MIPS doesn't really need to use atomic_cmpset_int() in situations like
 this because the software dirty bit emulation in trap.c acquires
 the pmap lock.  Atomics like this appear to be a carryover from i386
 where the hardware-managed TLB might concurrently set the modified bit.

Reviewed by: alc

13 years agoEnable shared locks for the devfs vnodes. Honor the locking mode
kib [Fri, 6 Aug 2010 09:23:47 +0000 (09:23 +0000)]
Enable shared locks for the devfs vnodes. Honor the locking mode
requested by lookup(). This should be a nop at the moment.

In collaboration with: pho
MFC after: 1 month

13 years agoInitialize VV_ISTTY vnode flag on the devfs vnode creation instead of
kib [Fri, 6 Aug 2010 09:06:55 +0000 (09:06 +0000)]
Initialize VV_ISTTY vnode flag on the devfs vnode creation instead of
doing it on each open.

In collaboration with: pho
MFC after: 1 month

13 years agoFix the issue reported by alc:
jchandra [Fri, 6 Aug 2010 07:32:33 +0000 (07:32 +0000)]
Fix the issue reported by alc:

 pmap_page_wired_mappings() counts the number of pv entries for the
 specified page that have the pv entry wired flag set to TRUE.
 pmap_enter() correctly initializes this flag.  However,
 pmap_change_wiring() doesn't update the corresponding pv entry flag,
 only the PTE.  So, the count returned by pmap_page_wired_mappings()
 will sometimes be wrong.

 In the short term, the best fix would be to eliminate the pv entry
 flag and use only the PTE.  That flag is wasting non-trivial memory.

Remove pv_wired flag, and use PTE flag to count the wired mappings.

Reviewed by: alc

13 years agoRemove redundant declaration of 'pcib_driver' class from sb_zbpci.c. This
neel [Fri, 6 Aug 2010 07:03:22 +0000 (07:03 +0000)]
Remove redundant declaration of 'pcib_driver' class from sb_zbpci.c. This
causes a compilation error.

The declaration is provided by sys/dev/pci/pcib_private.h starting from r210864.

13 years agoForce commit to remind me about MFCing r210909.
pjd [Fri, 6 Aug 2010 06:04:27 +0000 (06:04 +0000)]
Force commit to remind me about MFCing r210909.

MFC after: 1 month

13 years agoUse a signed integer to hold the address of a register.
neel [Fri, 6 Aug 2010 05:30:55 +0000 (05:30 +0000)]
Use a signed integer to hold the address of a register.

This does the right thing by sign extending the address when compiling for
the n64 abi.

13 years agouint64_t is 'unsigned long' in n64 build, so compiler is unhappy if the
neel [Fri, 6 Aug 2010 05:24:41 +0000 (05:24 +0000)]
uint64_t is 'unsigned long' in n64 build, so compiler is unhappy if the
format specifier in printf is "%llu".

Use "%ju" instead.

13 years agoUpdate the arguments to yy_config_parse() to match r210883.
dougb [Fri, 6 Aug 2010 05:20:21 +0000 (05:20 +0000)]
Update the arguments to yy_config_parse() to match r210883.

Choose the more conservative option ('yes' to exit on error) to match
the equivalent code in hastd.

13 years agoReflect default Tx interrupt moderation timer value change(50ms -> 1ms).
yongari [Fri, 6 Aug 2010 00:52:23 +0000 (00:52 +0000)]
Reflect default Tx interrupt moderation timer value change(50ms -> 1ms).

13 years agoReduce Tx interrupt moderation timer from 50ms to 1ms. The default
yongari [Fri, 6 Aug 2010 00:50:16 +0000 (00:50 +0000)]
Reduce Tx interrupt moderation timer from 50ms to 1ms. The default
value resulted in poor performance for UDP packets. With this
change, UDP bulk transfer performance is more than 940Mbps.

While I'm here fix a wrong register definition.

13 years ago- Add interrupts counter for PCI devices
gonzo [Thu, 5 Aug 2010 21:31:29 +0000 (21:31 +0000)]
- Add interrupts counter for PCI devices

13 years agoDocument 'none' value for remote.
pjd [Thu, 5 Aug 2010 19:54:57 +0000 (19:54 +0000)]
Document 'none' value for remote.

Reviewed by: dougb
MFC after: 1 month

13 years agoRe-add flag register for output. Some BIOS calls actually use it to return
jkim [Thu, 5 Aug 2010 19:30:57 +0000 (19:30 +0000)]
Re-add flag register for output.  Some BIOS calls actually use it to return
success/failure status.  Oops.

13 years agoImplement configuration reload on SIGHUP. This includes:
pjd [Thu, 5 Aug 2010 19:16:31 +0000 (19:16 +0000)]
Implement configuration reload on SIGHUP. This includes:
- Load added resources.
- Stop and forget removed resources.
- Update modified resources in least intrusive way, ie. don't touch
  /dev/hast/<name> unless path to local component or provider name were
  modified.

Obtained from: Wheel Systems Sp. z o.o. http://www.wheelsystems.com
MFC after: 1 month

13 years agoDo not copy stack pointer and flags. These registers are unconditionally
jkim [Thu, 5 Aug 2010 19:12:35 +0000 (19:12 +0000)]
Do not copy stack pointer and flags.  These registers are unconditionally
destroyed from vm86_prepcall().

13 years agoPrepare configuration parsing code to be called multiple times:
pjd [Thu, 5 Aug 2010 19:08:54 +0000 (19:08 +0000)]
Prepare configuration parsing code to be called multiple times:
- Don't exit on errors if not requested.
- Don't keep configuration in global variable, but allocate memory for
  configuration.
- Call yyrestart() before yyparse() so that on error in configuration file
  we will start from the begining next time and not from the place we left of.

MFC after: 1 month

13 years agoMake control_set_role() more public. We will need it soon.
pjd [Thu, 5 Aug 2010 19:04:29 +0000 (19:04 +0000)]
Make control_set_role() more public. We will need it soon.

MFC after: 1 month

13 years agoAllow to use 'none' keywork as remote address in case second cluster node
pjd [Thu, 5 Aug 2010 19:01:57 +0000 (19:01 +0000)]
Allow to use 'none' keywork as remote address in case second cluster node
is not setup yet.

MFC after: 1 month

13 years agoReset signal handlers after fork().
pjd [Thu, 5 Aug 2010 18:58:00 +0000 (18:58 +0000)]
Reset signal handlers after fork().

MFC after: 1 month

13 years ago- Use pjdlog_exitx() to log errors and exit instead of errx().
pjd [Thu, 5 Aug 2010 18:56:24 +0000 (18:56 +0000)]
- Use pjdlog_exitx() to log errors and exit instead of errx().
- Use 'unable to' (instead of 'cannot') consistently.

MFC after: 1 month

13 years agoEnable closefrom(2) here, as we have supported it for some time now.
csjp [Thu, 5 Aug 2010 18:49:06 +0000 (18:49 +0000)]
Enable closefrom(2) here, as we have supported it for some time now.

Discussed with: mlaier
MFC after: 2 weeks

13 years agoImplement a simple native VM86 backend for X86BIOS. Now i386 uses native
jkim [Thu, 5 Aug 2010 18:48:30 +0000 (18:48 +0000)]
Implement a simple native VM86 backend for X86BIOS.  Now i386 uses native
VM86 calls instead of the real mode emulator as a backend.  VM86 has been
proven reliable for very long time and it is actually few times faster than
emulation.  Increase maximum number of page table entries per VM86 context
from 3 to 8 pages.  It was (ridiculously) low and insufficient for new VM86
backend, which shares one context globally.  Slighly rearrange and clean up
the emulator backend to accommodate new code.  The only visible change here
is stack size, which is decreased from 64K to 4K bytes to sync. with VM86.
Actually, it seems there is no need for big stack in real mode.

MFC after: 1 month

13 years agoAssert that various buffers we are large enough.
pjd [Thu, 5 Aug 2010 18:27:41 +0000 (18:27 +0000)]
Assert that various buffers we are large enough.

MFC after: 1 month

13 years agoProblem with assertion is that it logs on stderr. Add two macros:
pjd [Thu, 5 Aug 2010 18:26:38 +0000 (18:26 +0000)]
Problem with assertion is that it logs on stderr. Add two macros:
PJDLOG_ASSERT() and PJDLOG_VERIFY() that will check the given condition
and log the problem where appropriate. The difference between those
two is that PJDLOG_VERIFY() always work and PJDLOG_ASSERT() can be
turned off by defining NDEBUG.

MFC after: 1 month

13 years agoDon't point users at the old csup homepage.
joel [Thu, 5 Aug 2010 18:26:03 +0000 (18:26 +0000)]
Don't point users at the old csup homepage.

Approved by: lulf

13 years agoKeep $FreeBSD$ in __FBSDID() only for C files.
pjd [Thu, 5 Aug 2010 18:23:43 +0000 (18:23 +0000)]
Keep $FreeBSD$ in __FBSDID() only for C files.

MFC after: 1 month

13 years agoMark two more places that we won't reach.
pjd [Thu, 5 Aug 2010 18:21:45 +0000 (18:21 +0000)]
Mark two more places that we won't reach.

MFC after: 1 month

13 years agoo FreeBSD 8.1-RELEASED added.
maxim [Thu, 5 Aug 2010 18:05:37 +0000 (18:05 +0000)]
o FreeBSD 8.1-RELEASED added.

PR: misc/149118
Submitted by: pluknet

13 years agoNow that TCP will be checked last we don't need any knowledge about other
pjd [Thu, 5 Aug 2010 17:57:59 +0000 (17:57 +0000)]
Now that TCP will be checked last we don't need any knowledge about other
protocols.

MFC after: 1 month

13 years agoAdd an argument to the proto_register() function which allows protocol to
pjd [Thu, 5 Aug 2010 17:56:41 +0000 (17:56 +0000)]
Add an argument to the proto_register() function which allows protocol to
declare it is the default and be placed at the end of the queue so it is
checked last.

MFC after: 1 month

13 years agoChange the MPTable and $PIR PCI-PCI bridge drivers to inherit from the
jhb [Thu, 5 Aug 2010 17:48:37 +0000 (17:48 +0000)]
Change the MPTable and $PIR PCI-PCI bridge drivers to inherit from the
generic PCI-PCI bridge driver and only override specific methods.  This
should fix suspend/resume of PCI-PCI bridges using these drivers.

13 years agoFix a bug where endpoints bound to wildcard addresses where
tuexen [Thu, 5 Aug 2010 16:52:13 +0000 (16:52 +0000)]
Fix a bug where endpoints bound to wildcard addresses where
using addresses not announced to the peer due to address
scoping.

MFC after: 3 weeks

13 years agoFix a bunch of errors (spelling and similar).
olli [Thu, 5 Aug 2010 16:28:17 +0000 (16:28 +0000)]
Fix a bunch of errors (spelling and similar).

As des noted, the section on SCTP would benefit from
a rewrite by a native speaker (which I am not).
Any volunteers?

Approved by: des (mentor)
MFC after: 1 week

13 years ago- Retire acpi_pcib_resume(). It is has just been an alias for
jhb [Thu, 5 Aug 2010 16:10:12 +0000 (16:10 +0000)]
- Retire acpi_pcib_resume().  It is has just been an alias for
  bus_generic_resume() since the pci_link(4) driver was added.
- Change the ACPI PCI-PCI bridge driver to inherit most of its methods
  from the generic PCI-PCI bridge driver.  In particular, this will now
  restore PCI config registers for ACPI PCI-PCI bridges.

Tested by: Oleg Sharoyko  osharoiko of gmail

13 years agoAdd a daily script to the periodic framework that reports
olli [Thu, 5 Aug 2010 15:53:33 +0000 (15:53 +0000)]
Add a daily script to the periodic framework that reports
changes to the package database, i.e. any packages that
have been added, updated or deleted in the past 24 hours.
The format is intentionally simple and concise.

That information is particularly useful on servers that
are maintained by multiple administrators.  When someone
adds, updates or deletes a package, the others will see
it in the daily periodic output.

This script is disabled by default.

PR: conf/113913
Submitted by: olli
Approved by: des (mentor)
MFC after: 3 weeks

13 years ago/etc/rc.d/network_ipv6 was deprecated.
ume [Thu, 5 Aug 2010 15:11:03 +0000 (15:11 +0000)]
/etc/rc.d/network_ipv6 was deprecated.

Reported by: Alex Kozlov <spam__at__rm-rf.kiev.ua>

13 years ago/etc/rc.d/ip6fw was deprecated.
ume [Thu, 5 Aug 2010 15:07:28 +0000 (15:07 +0000)]
/etc/rc.d/ip6fw was deprecated.

Reported by: Alex Kozlov <spam__at__rm-rf.kiev.ua>
MFC after: 3 days

13 years agoFix a race between clock_intr() and tick_ticker() when updating
neel [Thu, 5 Aug 2010 04:59:54 +0000 (04:59 +0000)]
Fix a race between clock_intr() and tick_ticker() when updating
'counter_upper' and 'counter_lower_last'. The race exists because
interrupts are enabled even though tick_ticker() executes in a
critical section.

Fix a bug in clock_intr() in how it updates the cached values of
'counter_upper' and 'counter_lower_last'. They are updated only
when the COUNT register rolls over. More interestingly it will *never*
update the cached values if 'counter_lower_last' happens to be zero.

Get rid of superfluous critical section in clock_intr(). There is no
reason to do this because clock_intr() executes in hard interrupt
context.

Switch back to using 'tick_ticker()' as the cpu ticker for Sibyte.

Reviewed by: jmallett, mav

13 years agoComment out IEEE Std 1003.1-2001 conformance proclaimed too early,
ache [Thu, 5 Aug 2010 01:39:25 +0000 (01:39 +0000)]
Comment out IEEE Std 1003.1-2001 conformance proclaimed too early,
see problems described in the comment to:
http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/128933

13 years agoChange kgdb_lookup() to resolve symbols via GDB instead of via libkvm(3).
jhb [Wed, 4 Aug 2010 21:02:04 +0000 (21:02 +0000)]
Change kgdb_lookup() to resolve symbols via GDB instead of via libkvm(3).

13 years agoCopy inode birthtime to the struct stat32.
kib [Wed, 4 Aug 2010 14:38:20 +0000 (14:38 +0000)]
Copy inode birthtime to the struct stat32.

MFC after: 1 week

13 years agoFix style.
kib [Wed, 4 Aug 2010 14:35:05 +0000 (14:35 +0000)]
Fix style.

MFC after: 1 week

13 years agoAdd 3 level page tables for MIPS in n64.
jchandra [Wed, 4 Aug 2010 14:12:09 +0000 (14:12 +0000)]
Add 3 level page tables for MIPS in n64.

- 32 bit compilation will still use old 2 level page tables
- re-arrange pmap code so that adding another level is easier
- pmap code for 3 level page tables for n64
- update TLB handler to traverse 3 levels in n64

Reviewed by: jmallett

13 years agoithd_name no longer defined, use td_name. Fix compile with KTR enabled.
jchandra [Wed, 4 Aug 2010 14:03:23 +0000 (14:03 +0000)]
ithd_name no longer defined, use td_name. Fix compile with KTR enabled.

13 years agoWhitespace fix for last check-in, move empty line to below endif.
jchandra [Wed, 4 Aug 2010 10:46:17 +0000 (10:46 +0000)]
Whitespace fix for last check-in, move empty line to below endif.

13 years agoMIPS 64 bit support.
jchandra [Wed, 4 Aug 2010 10:42:06 +0000 (10:42 +0000)]
MIPS 64 bit support.

When compiled for MIPS n64 ABI
- DES_LONG should be 'unsigned int'
- BN_LLONG should be undefined
- SIXTY_FOUR_BIT_LONG should be defined.

13 years agoDo not free sc if attach failed, as it was allocated by the bus
gavin [Wed, 4 Aug 2010 07:49:40 +0000 (07:49 +0000)]
Do not free sc if attach failed, as it was allocated by the bus
infrastructure, not us.  This appears to be a leftover from an older
version of the driver.

Submitted by: avg
Tested by: Anton Shterenlikht <mexas bristol.ac.uk>
MFC after: 1 week
X-MFC-Note: To stable/8 and stable/7 only

13 years agoFix typo.
joel [Wed, 4 Aug 2010 07:47:19 +0000 (07:47 +0000)]
Fix typo.

13 years agoIn order for MAXVNODES_MAX to be an "int" on powerpc and sparc, we must
alc [Wed, 4 Aug 2010 05:09:02 +0000 (05:09 +0000)]
In order for MAXVNODES_MAX to be an "int" on powerpc and sparc, we must
cast PAGE_SIZE to an "int".  (Powerpc and sparc, unlike the other
architectures, define PAGE_SIZE as a "long".)

Submitted by: Andreas Tobler

13 years agoAdd some mutex locking on the nfsnode to the regular NFS client.
rmacklem [Wed, 4 Aug 2010 01:19:11 +0000 (01:19 +0000)]
Add some mutex locking on the nfsnode to the regular NFS client.

Reviewed by: jhb

13 years agoCatch known CPUs before using IA32_TEMPERATURE_TARGET.
delphij [Wed, 4 Aug 2010 00:25:13 +0000 (00:25 +0000)]
Catch known CPUs before using IA32_TEMPERATURE_TARGET.
This way we would have an opportunity to hide the
Tj(target) value doesn't seem right stuff if we know
it's not working there.

Add temperature value for Core2 Duo Extreme Mobile that
I have access to.

13 years agosh: Return 0 from eval if no command was given.
jilles [Tue, 3 Aug 2010 22:17:29 +0000 (22:17 +0000)]
sh: Return 0 from eval if no command was given.

This makes a difference if there is a command substitution.

To make this work, evalstring() has been changed to set exitstatus to 0 if
no command was executed (the string contained only whitespace).

Example:
  eval $(false); echo $?
should print 0.

13 years agogrdc(6): de-obfuscate the timekeeping of now/delay
uqs [Tue, 3 Aug 2010 20:56:23 +0000 (20:56 +0000)]
grdc(6): de-obfuscate the timekeeping of now/delay

- 'now' only contains current time readings
- delay is only used as parameter to nanosleep
- prev_sec contains the time_t of the previous loop iteration

Submitted by: bde
MFC after: 2 weeks

13 years agoCorrectly sort usbconfig(8) within the SEE ALSO section.
bcr [Tue, 3 Aug 2010 19:25:58 +0000 (19:25 +0000)]
Correctly sort usbconfig(8) within the SEE ALSO section.

Noticed by:     dougb
MFC after:      3 days

13 years agoFix typo.
joel [Tue, 3 Aug 2010 18:29:58 +0000 (18:29 +0000)]
Fix typo.

13 years agoSpelling fixes.
joel [Tue, 3 Aug 2010 17:40:09 +0000 (17:40 +0000)]
Spelling fixes.

13 years agoAdd support for ATI Radeon HD 4250.
olli [Tue, 3 Aug 2010 16:30:56 +0000 (16:30 +0000)]
Add support for ATI Radeon HD 4250.

PR: kern/149041
Submitted by: olli
Reviewed by: rnoland
Approved by: des (mentor)
MFC after: 1 week

13 years agoUpdate references from nonexistent usbconfig(1) to usbconfig(8).
bcr [Tue, 3 Aug 2010 16:21:48 +0000 (16:21 +0000)]
Update references from nonexistent usbconfig(1) to usbconfig(8).

PR:             docs/149221
Submitted by:   Lars Hartmann (lars at chaotika dot org)
MFC after:      3 days

13 years agoRemove an unnecessary register load.
jkim [Tue, 3 Aug 2010 16:08:58 +0000 (16:08 +0000)]
Remove an unnecessary register load.

13 years agogrdc(6): Partial backout of r210755
uqs [Tue, 3 Aug 2010 16:02:57 +0000 (16:02 +0000)]
grdc(6): Partial backout of r210755

The previous revision was flawed in numerous ways:
- the if condition would depend on garbage stack values
- grdc(6) would loop n times, not n seconds. This only makes
  a difference if it takes more than 1 second to recalculate/redraw
  the display.
- style(9) violations

The following change adds argument checking and tracks the elapsed time
between loops explicitly and will exit after roughly n seconds.

While here sort headers and remove bogus #ifndef

Submitted by: bde
MFC after: 2 weeks

13 years agoProperly set ifi_datalen for compat32 struct if_data32.
kib [Tue, 3 Aug 2010 15:40:42 +0000 (15:40 +0000)]
Properly set ifi_datalen for compat32 struct if_data32.

PR: kern/149240
Submitted by: Stef Walter <stef memberwebs com>
MFC after: 1 weeks

13 years agosavectx() has not been used for fork(2) for about 15 years. [1]
jkim [Tue, 3 Aug 2010 15:32:08 +0000 (15:32 +0000)]
savectx() has not been used for fork(2) for about 15 years. [1]
Do not clobber FPU thread's PCB as it is more harmful.  When we resume CPU,
unconditionally reload FPU state.

Pointed out by: bde [1]

13 years ago- Do not use the runtime mask when logfile is specified.
fabient [Tue, 3 Aug 2010 13:46:25 +0000 (13:46 +0000)]
- Do not use the runtime mask when logfile is specified.
- Revert the fix on rtld path that is not necessary.

MFC after: 1 week

13 years agoWhen compat32 recvmsg(2) does not need to copy out control messages, set
kib [Tue, 3 Aug 2010 11:23:44 +0000 (11:23 +0000)]
When compat32 recvmsg(2) does not need to copy out control messages, set
msg_controllen to 0.

PR: kern/149227
Submitted by: Stef Walter <stef memberwebs com>
MFC after: 1 weeks