Matthew Dillon [Fri, 3 Jul 2009 21:38:50 +0000 (14:38 -0700)]
USB - Fix mis-cast in EHCI producing garbage on console
EHCI was improperly interpreting the companion bus softc pointer,
causing it to report junk on the console.
Reported-by: Alexander Polakov <polachok@gmail.com>
Matthew Dillon [Fri, 3 Jul 2009 17:48:26 +0000 (10:48 -0700)]
frevoke - More fixes to fdfree()
* Don't replace p->p_fd until after all closef()'s have been called. Fixed
panic in kqueue's close function.
* Move softrefs check to after the closef point, but before we actually
start kfree()ing the structures. Lock the closef's normally since
p->p_fd is still valid.
Matthew Dillon [Fri, 3 Jul 2009 16:44:28 +0000 (09:44 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Fri, 3 Jul 2009 16:43:11 +0000 (09:43 -0700)]
HAMMER - Formalize the maximum number of volumes (256).
Formalize the maximum number of volumes (256) by adding a define in
hammer_disk.h and then using it.
Submitted-by: Antonio Huete Jimenez <ahuete.devel@gmail.com>
Sascha Wildner [Fri, 3 Jul 2009 12:03:54 +0000 (14:03 +0200)]
Add a wrapper script to allow "foreign" compilers into our build system.
With this commit, all CCVER values not associated with a compiler in
base will be relegated to the wrapper script which then (in conjunction
with the compilers.conf file) determines which non-base compiler to
use.
So far, defaults for clang (built from SVN) and pkgsrc/wip/gcc44 are
provided.
For more info see the compilers.conf(5) manual page and my mail on
kernel@:
http://leaf.dragonflybsd.org/mailarchive/kernel/2009-06/msg00013.html
Sascha Wildner [Fri, 3 Jul 2009 00:22:44 +0000 (02:22 +0200)]
hammer.sh: Sync with some of the recent installer changes.
* Use disklabel64(8).
* Do the 'chflags nohistory ...' after the cpdups to not have cpdup
overwrite the flags again.
Sascha Wildner [Thu, 2 Jul 2009 23:45:16 +0000 (01:45 +0200)]
installer: Use disklabel64 for the UFS only install, too.
Discussed-with: dillon, corecode
Sascha Wildner [Thu, 2 Jul 2009 21:52:49 +0000 (23:52 +0200)]
installer: Move the 'chflags nohistory ...' commands behind the cpdups.
This works around the issue that cpdup resets the flags if changed
before.
Reported-by: corecode
Dragonfly-bug: <http://bugs.dragonflybsd.org/issue1259>
Matthew Dillon [Thu, 2 Jul 2009 19:02:17 +0000 (12:02 -0700)]
file descriptor enumeration - Fix races
Add a soft refrence count, fd_softrefs, to struct filedesc. This is
used to prevent the filedesc from being ripped out from under a file
descriptor scan via the allproc scan, which is used to enumerate
file descriptors and also used by the revoke code.
Reported-by: hasso
Matthew Dillon [Thu, 2 Jul 2009 02:48:22 +0000 (19:48 -0700)]
SILI - Fix embarassing bug for non-NCQ disk accesses > 128G
Fix an embarassing bug for non-NCQ disk accesses > 128G. This only
applied to non-NCQ accesses for disks and typically will not happen
with the SILI part since it can use NCQ for everything. Non-NCQ
accesses for seek positions from 128G to 2TB were wrapping the LBA.
Matthew Dillon [Thu, 2 Jul 2009 02:46:50 +0000 (19:46 -0700)]
AHCI - Fix embarassing bug for non-NCQ disk accesses > 128G
Fix an embarassing bug for non-NCQ disk accesses > 128G. This only
applied to non-NCQ accesses for disks which effectively means disks
behind a PM, since NCQ is generally used everywhere else. Accesses
for seek positions from 128G to 2TB were wrapping the LBA.
Sascha Wildner [Wed, 1 Jul 2009 22:09:30 +0000 (00:09 +0200)]
sem_init.3: Some little mdoc cleanup.
Sascha Wildner [Wed, 1 Jul 2009 21:59:35 +0000 (23:59 +0200)]
installer: Fix debugging function call.
Matthew Dillon [Wed, 1 Jul 2009 17:46:06 +0000 (10:46 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Wed, 1 Jul 2009 17:45:23 +0000 (10:45 -0700)]
SILI - ATAPI adjustments.
Restrict the sense length to what ATAPI expects and add more legacy
initialization to the packet FIS.
This was an attempt to fix some reported issues with ATAPI, but it didn't
fix anything. However, it didn't hurt either and the adjustments are at
least as good as the old ones.
Matthew Dillon [Wed, 1 Jul 2009 17:43:15 +0000 (10:43 -0700)]
AHCI - ATAPI adjustments.
Restrict the sense length to what ATAPI expects and add more legacy
initialization to the packet FIS.
This was an attempt to fix some reported issues with ATAPI, but it didn't
fix anything. However, it didn't hurt either and the adjustments are at
least as good as the old ones.
Matthew Dillon [Wed, 1 Jul 2009 17:36:05 +0000 (10:36 -0700)]
AHCI - Move some intel-specific magic writes to intel-specific code.
It is unclear if these are needed at all but in any case they only apply
to certain Intel chipsets anyway.
Matthew Dillon [Wed, 1 Jul 2009 17:34:05 +0000 (10:34 -0700)]
pcidevs - Add entries for Intel ICH6 and ICH7 series chipsets.
Sascha Wildner [Wed, 1 Jul 2009 10:51:44 +0000 (12:51 +0200)]
Regenerate the pciconf(8) database
...from the following files:
Hart: Jan 22, 2008 (version 671)
Boemler: Jul 1, 2009
Mares: Jun 2, 2009
Matthew Dillon [Tue, 30 Jun 2009 22:04:34 +0000 (15:04 -0700)]
HAMMER VFS - Make the same changes to statfs() that were made to statvfs().
Matthew Dillon [Tue, 30 Jun 2009 19:04:40 +0000 (12:04 -0700)]
Update the sem_init(3) manpage - with additional info on the pshared impl.
Suggested-by: Hasso Tepper <hasso@crater.dragonflybsd.org>
Hasso Tepper [Tue, 30 Jun 2009 06:43:03 +0000 (09:43 +0300)]
Update the sem_init(3) manpage - shared semaphores are supported now.
It might need some more info about how shared semaphores are implemented
etc though.
Matthew Dillon [Mon, 29 Jun 2009 22:38:53 +0000 (15:38 -0700)]
libthread_xu - implement pshared for sem_init
Implement the pshared flag, creating a semaphore in shared memory
that works across fork(). No additional kernel support is needed,
our umtx* system calls work on the underlying physical memory and
thus work just fine with shared memory.
sem_alloc() uses mmap() to allocate a page of shared memory.
Currently sem_destroy() does not recover the allocated memory for
shared semaphores as it is unclear how one is supposed to do so
in a fork()ed environment.
Requested-by: Hasso Tepper <hasso@estpak.ee>
Matthew Dillon [Mon, 29 Jun 2009 15:07:58 +0000 (08:07 -0700)]
AHCI/SILI - Do manual chipid check to PMs with fake ports
* Add a switch() on the chipid for PMs which add a fake port at the
end and avoid probing the fake port. Probing the port is harmless and
typically fails in the softreset sequence, but takes time.
Sascha Wildner [Sun, 28 Jun 2009 20:09:58 +0000 (22:09 +0200)]
installer: Add UFS /boot & HAMMER ROOT support.
The installer's HAMMER install option does now recommend (and require)
a UFS /boot partition.
Also, check more explicitly for the user specifying /usr, /var, etc.
and refuse to continue if he did because these will be automatically
created as pseudo fs's.
Finally, while here, do some miscellaneous cleanup.
In-discussion-with: dillon
Sascha Wildner [Sat, 27 Jun 2009 17:31:54 +0000 (19:31 +0200)]
installer: Raise the absolute minimum disk size to 2GB.
Also adjust some minimum partition sizes and size suggestions.
While here, eliminate some dead code.
Matthew Dillon [Sun, 28 Jun 2009 18:56:01 +0000 (11:56 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Sun, 28 Jun 2009 18:55:25 +0000 (11:55 -0700)]
HAMMER VFS - Export highest supported production version with sysctl.
* Implement vfs.hammer.support_version to return the highest supported
HAMMER FS version to userland. Used by newfs_hammer.
Matthew Dillon [Sun, 28 Jun 2009 18:53:45 +0000 (11:53 -0700)]
HAMMER - newfs_hammer - format with highest production version available
* newfs_hammer now probes the HAMMER VFS and uses the highest production
version supported to format the HAMMER filesystem.
Matthew Dillon [Sun, 28 Jun 2009 18:49:54 +0000 (11:49 -0700)]
HAMMER UTIL - show command now includes inode capability field.
Matthew Dillon [Sun, 28 Jun 2009 18:21:27 +0000 (11:21 -0700)]
HAMMER VFS - Fix umount panic on buffer flush race
* Set io.waitdep on buffers during umount to ensure that the buffer
unload code waits for running buffers to complete their operation.
This fixes occassional assertion panics on umount.
Matthew Dillon [Sun, 28 Jun 2009 17:23:23 +0000 (10:23 -0700)]
HAMMER VFS - version 2 work - move directory entries to inode localization
* Set version 2 features. Directories created in a version 2 HAMMER
filesystem place their directory entries in the B-Tree's inode localization
space instead of the general space. Pre-existing directories continue
to use the old scheme.
* Amost doubles directory lookup / find / ls performance in effected
directories. However, at the moment the B-Tree must be reblocked before
the gains are seen.
Stathis Kamperis [Sun, 28 Jun 2009 12:04:53 +0000 (15:04 +0300)]
pthread_attr.3: Document pthread_attr_{get,set}_guardsize().
Adapted-from: FreeBSD
Peter Avalos [Sat, 27 Jun 2009 21:34:06 +0000 (11:34 -1000)]
Merge branch 'vendor/LIBARCHIVE'
Peter Avalos [Wed, 24 Jun 2009 21:29:09 +0000 (11:29 -1000)]
Upgrade to libarchive-2.7.0.
Sepherosa Ziehau [Sat, 27 Jun 2009 13:36:37 +0000 (21:36 +0800)]
MP table: Test 0xe0000 when searching MP float pointer structure
MultiProcessor Specification version 1.4 never mentions this location,
however, some brain-dead BIOSes put MP table there ... "Extened BIOS"
location is taken from mptable(8) in the base system.
Reported-and-Tested-by: ruse39 on EFnet #dragonflybsd
Sepherosa Ziehau [Sat, 27 Jun 2009 10:09:51 +0000 (18:09 +0800)]
IO APIC: Mask the interrupt signal before trying to clear IRR by switching
trigger mode.
Matthew Dillon [Sat, 27 Jun 2009 05:07:45 +0000 (22:07 -0700)]
udp6 - Fix multicast and other issues
* The placemarker flag was being tested inverse to what it was supposed
to be tested with.
Submitted-by: Kyle Butt <kylebutt@gmail.com>
Sepherosa Ziehau [Sat, 27 Jun 2009 03:05:19 +0000 (11:05 +0800)]
ACPI MADT: Don't use the "Initial APIC ID" in cpu_procinfo(cpuid.1)
Some BIOSes seem to reprogram LAPIC ID to a value different from
what's in cpu_procinfo.
Sepherosa Ziehau [Sat, 27 Jun 2009 01:36:42 +0000 (09:36 +0800)]
Introduce lapic enumerators, which is used to probe and config lapics
lapic enumerator implementation should provide two method:
lapic_probe() -- make sure that if this enumerator is selected, later
lapic enumeration could work. Return error code upon
failure.
lapic_enumerate() -- enumerate lapic and properly configure lapic (currently
only mapping lapic is needed).
lapic enumerator implementation could be registered by calling:
lapic_enumerator_register()
with lapic_enumerator struct. The higher the priority field, the earlier
the lapic enumerator's lapic_probe method will be invoked.
Currently two lapic enumerators are implemented and registered:
- lapic enumerator using MP table
- lapic enumerator using ACPI MADT
ACPI MADT lapic enumerator has higher priority. For testing purpose, ACPI
MADT lapic enumerator's priority could be overridden by setting tunable
hw.madt_lapic_prio to a smaller value (less than 20 currently)
Idea-from: FreeBSD
Sepherosa Ziehau [Mon, 22 Jun 2009 12:18:59 +0000 (20:18 +0800)]
Move lapic_init() into apic/mpapic.c
Sepherosa Ziehau [Mon, 22 Jun 2009 12:05:44 +0000 (20:05 +0800)]
MP probe: Get rid of cpu_apic_address
Sascha Wildner [Fri, 26 Jun 2009 23:48:34 +0000 (01:48 +0200)]
nrelease: Sync the LiveDVD kernels with some recent changes.
Sascha Wildner [Fri, 26 Jun 2009 23:41:51 +0000 (01:41 +0200)]
Make sili(4) compilable into the kernel and add it to GENERIC & LINT.
Matthew Dillon [Fri, 26 Jun 2009 23:19:38 +0000 (16:19 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Fri, 26 Jun 2009 23:19:14 +0000 (16:19 -0700)]
AHCI/SILI - Remove unused ScsiTypeArray array
Matthew Dillon [Fri, 26 Jun 2009 23:16:58 +0000 (16:16 -0700)]
SILI - Increase hard -> soft PM target delay to 5 seconds
* Unfortunately it looks like we have to seriously increase the delay
between doing a hard reset via the PM to a target and initiating
the softreset.
The Silicon Image chip appears to brick the port if its softreset
collides with the target device. There is no way to fix it other
then hard-resetting the port (which causes the PM to reset as well).
Matthew Dillon [Fri, 26 Jun 2009 22:31:55 +0000 (15:31 -0700)]
SILI - Fix assertion panic during error handling.
* The error handling code could not distinguish between the use of the
error ccb as part of a softreset sequence and the use of the error
ccb as part of a error recovery sequence and asserted in the former
case.
* Ensure that no commands are active on the port prior to initiating
a hard reset through the port multiplier.
Sascha Wildner [Fri, 26 Jun 2009 21:58:25 +0000 (23:58 +0200)]
Add Colemak keyboard layout.
Submitted-by: Geert Hendrickx <ghen@telenet.be>
Dragonfly-bug: <http://bugs.dragonflybsd.org/issue1409>
Matthew Dillon [Fri, 26 Jun 2009 16:48:11 +0000 (09:48 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Fri, 26 Jun 2009 16:47:56 +0000 (09:47 -0700)]
SILI - Add the SILI driver to the general module build
Matthew Dillon [Fri, 26 Jun 2009 16:47:25 +0000 (09:47 -0700)]
SILI - Add serial number support, strip whitespace.
* Add page inquiry emulation for DISK devices so CAM can probe the
serial number.
* Strip whitespace when reporting the mode, firmware, and serial number
from the inquiry data.
Matthew Dillon [Fri, 26 Jun 2009 16:45:54 +0000 (09:45 -0700)]
AHCI - Add serial number support, strip whitespace.
* Add page inquiry emulation for DISK devices so CAM can probe the
serial number.
* Strip whitespace when reporting the mode, firmware, and serial number
from the inquiry data.
Matthew Dillon [Fri, 26 Jun 2009 16:26:33 +0000 (09:26 -0700)]
CAM - Unconditionally print out the serial number
Do not conditionalize the printing of the serial number on bootverbose any
more. The serial number is fast becoming an important piece of information.
Sascha Wildner [Fri, 26 Jun 2009 06:46:01 +0000 (08:46 +0200)]
utf2.5: Terminate quotation.
Matthew Dillon [Fri, 26 Jun 2009 03:56:52 +0000 (20:56 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Fri, 26 Jun 2009 03:55:50 +0000 (20:55 -0700)]
SILI - Cleanup, remove broken KKASSERT and remove DIAGNOSTICs
Reported-by: Sascha Wildner <swildner@crater.dragonflybsd.org>
Matthew Dillon [Fri, 26 Jun 2009 03:55:07 +0000 (20:55 -0700)]
AHCI - Reduce console spew
Matthew Dillon [Fri, 26 Jun 2009 03:53:50 +0000 (20:53 -0700)]
nrelease - img file - use boot0cfg to set packet mode
* Use packet mode by default in the usb key image file's boot blocks.
Sascha Wildner [Fri, 26 Jun 2009 01:18:18 +0000 (03:18 +0200)]
Properly terminate some quotations in several manual pages.
Matthew Dillon [Fri, 26 Jun 2009 01:05:53 +0000 (18:05 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Fri, 26 Jun 2009 01:05:30 +0000 (18:05 -0700)]
nrelease - img build - A usb stick will now probe starting at da8
Sascha Wildner [Thu, 25 Jun 2009 23:36:47 +0000 (01:36 +0200)]
rc.conf.5: Document 'fixbootfile'.
Sascha Wildner [Sun, 14 Jun 2009 22:40:51 +0000 (00:40 +0200)]
lastcomm(1): Sync with FreeBSD's RELENG_6.
Sascha Wildner [Thu, 25 Jun 2009 22:27:32 +0000 (00:27 +0200)]
subr_diskmbr.c: Fix some warning fixes.
* No need to cast dp_size at all, just fix the format.
* Add cast for size.
Sascha Wildner [Thu, 25 Jun 2009 16:00:35 +0000 (18:00 +0200)]
Adjust hammer.sh for the new "fixbootfile" rc script.
Now that this cares about fixing kern.bootfile, there's no need to
mess in the user's /etc from outside.
Hasso Tepper [Thu, 25 Jun 2009 21:04:23 +0000 (00:04 +0300)]
We shouldn't need to drop and reaquire the lock here.
Obtained-from: FreeBSD
Hasso Tepper [Thu, 25 Jun 2009 21:00:18 +0000 (00:00 +0300)]
Some more cleanups for vblank code on Intel.
The Intel 2d driver calls modeset before reinstalling the handler on
a vt switch. This means that vblank status ends up getting cleared
after it has been setup. Restore saved values for the pipestat registers
rather than just wiping them out.
Obtained-from: FreeBSD
Hasso Tepper [Thu, 25 Jun 2009 18:44:41 +0000 (21:44 +0300)]
Initialize max_vblank_count earlier.
Small cleanup of the error paths while I'm here.
Obtained-from: FreeBSD
Hasso Tepper [Thu, 25 Jun 2009 18:37:52 +0000 (21:37 +0300)]
Keep track of the hardware counter more aggressively
... while interrupts are enabled. This should help to reduce cases
where the hardware counter reference jumps by large amounts.
Obtained-from: FreeBSD
Hasso Tepper [Thu, 25 Jun 2009 18:34:58 +0000 (21:34 +0300)]
Fix one use of atomic for refcount missed in last commit.
Obtained-from: FreeBSD
Hasso Tepper [Thu, 25 Jun 2009 18:28:47 +0000 (21:28 +0300)]
Additional vblank cleanups.
Use the vbl_lock when manipulating the refcount. Eventually I want to
convert this to use our internal refcount code. Continue to use atomic
ops for manipulating vblank count since we access it often just for
reading.
Obtained-from: FreeBSD
Hasso Tepper [Thu, 25 Jun 2009 18:21:43 +0000 (21:21 +0300)]
Ensure that we always hold the lock when executing vblank_disable_fn().
The original code in FreeBSD uses callout_init_mtx() we don't have,
therefore we lock/unlock in vblank_disable_fn itself.
Michael Neumann [Thu, 25 Jun 2009 20:24:25 +0000 (22:24 +0200)]
priv: Reduce some uses of PRISON_ROOT
Move decision whether the privilege is allowed or not into function
prison_priv_check().
Matthew Dillon [Thu, 25 Jun 2009 19:52:30 +0000 (12:52 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Thu, 25 Jun 2009 19:50:50 +0000 (12:50 -0700)]
HAMMER VFS - Take reserved space into account when reporting statvfs data
Adjust statvfs data so reserved space is taken into account, so the
filesystem starts failing modifying operations closer to when 'df' would
otherwise say that 0 free space remains.
Submitted-by: Antonio Huete Jimenez <tuxillo@quantumachine.net> (with modification)
Matthew Dillon [Thu, 25 Jun 2009 19:50:02 +0000 (12:50 -0700)]
HAMMER UTILS - Add to the newfs_hammer documentation.
Michael Neumann [Thu, 25 Jun 2009 19:36:33 +0000 (21:36 +0200)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Thu, 25 Jun 2009 19:35:14 +0000 (12:35 -0700)]
CAM - Auto unit number start based on device
If no SCSI devices are wired CAM now automatically starts any umass sim
attachments at unit #8 instead of unit #0. This reduces interference
between usb mass storage attachments and ahci/sili attachments. Otherwise
the usb mass storage can steal "da0" away from ahci/sili.
Michael Neumann [Thu, 25 Jun 2009 19:29:52 +0000 (21:29 +0200)]
priv: Introduce prison_priv_check()
One day the PRISON_ROOT flag, which when passed to priv_check_cred()
allows the privilege in a jail, will be gone and all privileges allowed
in a jail are centrally maintained in the prison_priv_check() function
instead of spreading it over all the calls.
Matthew Dillon [Thu, 25 Jun 2009 18:59:03 +0000 (11:59 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Thu, 25 Jun 2009 18:57:14 +0000 (11:57 -0700)]
CAM/UMASS - use xpt_action_async() to ensure asynchronous scsi bus scan.
Use the new xpt_action_async() function to ensure that the scsi bus is
rescanned asynchronously. XPT_SCAN_BUS already sets XPT_FC_QUEUED but
it doesn't apply to command initiation. The new function does it properly.
Stathis Kamperis [Wed, 24 Jun 2009 18:41:44 +0000 (21:41 +0300)]
nice(3): Return new nice value.
The nice(1) utility isn't affected by this change,
since it uses {get,set}priority() directly.
Dragonfly-bug: <http://bugs.dragonflybsd.org/issue1408>
Michael Neumann [Thu, 25 Jun 2009 14:22:25 +0000 (16:22 +0200)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Thu, 25 Jun 2009 14:13:08 +0000 (07:13 -0700)]
rcng - add fixbootfile to Makefile for install
Michael Neumann [Thu, 25 Jun 2009 13:42:09 +0000 (15:42 +0200)]
priv: Refactor kern_mknod and narrow privileges
Michael Neumann [Thu, 25 Jun 2009 13:15:30 +0000 (15:15 +0200)]
priv: Narrow privileges
Michael Neumann [Thu, 25 Jun 2009 12:42:01 +0000 (14:42 +0200)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Thu, 25 Jun 2009 06:09:49 +0000 (23:09 -0700)]
rcng - add rcng script to fixup kern.bootfile
In BOOT+ROOT setups the boot loader does not realize that the boot partition
will be mounted as /boot under the root partition. This RCNG script fixes
up the kern.bootfile sysctl.
Matthew Dillon [Thu, 25 Jun 2009 00:14:45 +0000 (17:14 -0700)]
libstand - remove iaddr_t, do not assume long double is a power of 2, clean up
* Remove iaddr_t. Use uintptr_t instead.
* Remove use of long double to determine alignment. Use a fixed 16 byte
alignment instead.
* General cleanup
Matthew Dillon [Wed, 24 Jun 2009 23:46:26 +0000 (16:46 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Wed, 24 Jun 2009 23:43:30 +0000 (16:43 -0700)]
AMD64 - Support -m32 by conditionalizing 32 and 64 bit typedefs.
Conditionalize the typedefs in amd64-specific headers to support -m32
compilation. This fixes the compilation of the boot loader from the
amd64 build.
Hasso Tepper [Tue, 23 Jun 2009 22:12:05 +0000 (01:12 +0300)]
Add some sysctl info so that we can see what is going on with vblanks.
Obtained-from: FreeBSD
Hasso Tepper [Tue, 23 Jun 2009 22:11:01 +0000 (01:11 +0300)]
Only release irq resources if we were actually using them.
Obtained-from: FreeBSD
Hasso Tepper [Tue, 23 Jun 2009 22:05:47 +0000 (01:05 +0300)]
Remove signals for vblank events code.
Using signals for vblank events is prone to issues. There have never
been any consumers and likely will never be. Furthermore, we have
never enabled the code for it, so just get rid of it.
Obtained-from: FreeBSD
Hasso Tepper [Tue, 23 Jun 2009 22:02:37 +0000 (01:02 +0300)]
More meaningful timeout for sleep.
Given that vblanks generally occur 60 times a second, waiting 3 seconds
seems rather excessive.
Obtained-from: FreeBSD
Hasso Tepper [Tue, 23 Jun 2009 21:58:04 +0000 (00:58 +0300)]
Don't corrupt hardware vblank counter.
vblank[crtc].last represents the hardware counter while request.sequence
represents the software counter.
Obtained-from: FreeBSD
Hasso Tepper [Tue, 23 Jun 2009 21:56:43 +0000 (00:56 +0300)]
Hold the lock while we save/restore register for suspend/resume.
Obtained-from: FreeBSD
Matthew Dillon [Wed, 24 Jun 2009 19:53:50 +0000 (12:53 -0700)]
Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
Matthew Dillon [Wed, 24 Jun 2009 19:53:32 +0000 (12:53 -0700)]
USB - Fix compiler warning.