dragonfly.git
8 years agodrm: Store PCI device information in a struct pci_dev
François Tigeot [Sun, 10 Jan 2016 10:40:42 +0000 (11:40 +0100)]
drm: Store PCI device information in a struct pci_dev

Reducing differences with Linux

8 years agodrm: Improve Linux compatibility of pci structures
François Tigeot [Sun, 10 Jan 2016 10:37:47 +0000 (11:37 +0100)]
drm: Improve Linux compatibility of pci structures

8 years agokernel/kate: Bad idea to put temporary values in the sensor value directly.
Sascha Wildner [Sun, 10 Jan 2016 09:00:46 +0000 (10:00 +0100)]
kernel/kate: Bad idea to put temporary values in the sensor value directly.

Pointed-out-by: Constantine Aleksandrovich Murenin
8 years agoncal(1): Sync with FreeBSD.
Sascha Wildner [Sun, 10 Jan 2016 08:08:23 +0000 (09:08 +0100)]
ncal(1): Sync with FreeBSD.

* Multibyte character support.

* -3 to show previous, this and next month.

* -A for months after this month.

* -B for months before this month.

* -h to disable highlighting.

* Some extensions to -m ('f' and 'p').

* Various other fixes.

Also, raise WARNS to 6.

Submitted-by: Yellow Rabbit <yrabbit@sdf.lonestar.org>
8 years agoee(1): Raise WARNS to 2.
Sascha Wildner [Sun, 10 Jan 2016 03:18:57 +0000 (04:18 +0100)]
ee(1): Raise WARNS to 2.

8 years agoee(1): Fix an uninitialized variable.
Sascha Wildner [Sun, 10 Jan 2016 03:18:30 +0000 (04:18 +0100)]
ee(1): Fix an uninitialized variable.

Since ee(1) was switched from mktemp() to mkstemp(), the file name is
no longer initialized. Initialize it to NULL at least. The message in
case of an error will still be confusing, but for now, live with it.

8 years agow(1): Use NELEM(), raise WARNS to 3.
Sascha Wildner [Sun, 10 Jan 2016 02:43:35 +0000 (03:43 +0100)]
w(1): Use NELEM(), raise WARNS to 3.

8 years agow(1): inet_addr() returns INADDR_NONE upon failure.
Sascha Wildner [Sun, 10 Jan 2016 02:41:35 +0000 (03:41 +0100)]
w(1): inet_addr() returns INADDR_NONE upon failure.

8 years agokernel/kate: CurTmp starts at -28C on Socket AM2 Revision G processors.
Sascha Wildner [Sun, 10 Jan 2016 02:16:48 +0000 (03:16 +0100)]
kernel/kate: CurTmp starts at -28C on Socket AM2 Revision G processors.

Such as on this 4850e of mine.

Taken-from: FreeBSD

8 years agodrm: Always enable MSI interrupts
François Tigeot [Sat, 9 Jan 2016 21:28:18 +0000 (22:28 +0100)]
drm: Always enable MSI interrupts

The Linux drm code doesn't contain any blacklist for hardware with
broken MSI support; there is no reason for MSI interrupts to not work
with DragonFly either.

8 years agovis(1): Fix some warnings and raise WARNS to 6.
Sascha Wildner [Sat, 9 Jan 2016 18:56:50 +0000 (19:56 +0100)]
vis(1): Fix some warnings and raise WARNS to 6.

8 years agortadvd(8)/rtadvctl(8): Raise WARNS to 6 and fix two warnings.
Sascha Wildner [Sat, 9 Jan 2016 18:44:13 +0000 (19:44 +0100)]
rtadvd(8)/rtadvctl(8): Raise WARNS to 6 and fix two warnings.

8 years agomrouted(8): Raise WARNS to 2 and fix some warnings.
Sascha Wildner [Sat, 9 Jan 2016 18:34:42 +0000 (19:34 +0100)]
mrouted(8): Raise WARNS to 2 and fix some warnings.

8 years agosliplogin(8): Fix a number of style(9) issues / warnings, raise WARNS to 3.
Sascha Wildner [Sat, 9 Jan 2016 18:00:22 +0000 (19:00 +0100)]
sliplogin(8): Fix a number of style(9) issues / warnings, raise WARNS to 3.

8 years agotzsetup(8): Raise WARNS to 6 and fix a warning.
Sascha Wildner [Sat, 9 Jan 2016 17:32:03 +0000 (18:32 +0100)]
tzsetup(8): Raise WARNS to 6 and fix a warning.

8 years agorpc.lockd(8): Clean up the Makefile a little.
Sascha Wildner [Sat, 9 Jan 2016 04:37:06 +0000 (05:37 +0100)]
rpc.lockd(8): Clean up the Makefile a little.

8 years agorpc.lockd(8): Remove an obsolete file from an older version of lockd.
Sascha Wildner [Sat, 9 Jan 2016 04:19:00 +0000 (05:19 +0100)]
rpc.lockd(8): Remove an obsolete file from an older version of lockd.

8 years agoi386 removal, part 48/x: Remove ancient bootdev parsing in sysctl(8).
Sascha Wildner [Sat, 9 Jan 2016 03:05:44 +0000 (04:05 +0100)]
i386 removal, part 48/x: Remove ancient bootdev parsing in sysctl(8).

While here, remove an i386 specific reference from isa.4's SEE ALSO.

8 years agoi386 removal, part 47/x: Remove the i386 specific spic(4) device.
Sascha Wildner [Sat, 9 Jan 2016 02:47:39 +0000 (03:47 +0100)]
i386 removal, part 47/x: Remove the i386 specific spic(4) device.

8 years agoSync ACPICA with Intel's version 20160108.
Sascha Wildner [Fri, 8 Jan 2016 18:17:15 +0000 (19:17 +0100)]
Sync ACPICA with Intel's version 20160108.

This includes the 20151218 upgrade which we had to revert again because
of a bug found by tkusumi. The only major change in 20160108 is a fix
for this bug.

20151218 changes:

* Execute "module-level code" as ACPI tables are loaded into the
  namespace during initialization and install the default operation
  region handlers for the SystemMemory, SystemIO and PCI_Config
  address spaces before any ACPI tables are loaded (in order to
  enable module-level code to access these address spaces). Also,
  several changes to the internal _REG support and to utilities
  in conjunction with these changes.

20160108 changes:

* Fixed a regression introduced in version 20151218 concerning the
  execution of so-called module-level ASL/AML code. Namespace objects
  created under a module-level If() construct were not properly/fully
  entered into the namespace and could cause an interpreter fault when
  accessed.

* Fixed a problem with the compilation of the GpioIo and GpioInt resource
  descriptors. The _PIN field name was incorrectly defined to be an array
  of 32-bit values, but the _PIN values are in fact 16 bits each. This
  would cause incorrect bit width warnings when using Word (16-bit) fields
  to access the descriptors.

Additionally, some minor fixes, features etc. were added.

For a more detailed list, please see sys/contrib/dev/acpica/changes.txt.

Tested-by: tkusumi
8 years agoroute: various small fixes
Zach Crownover [Thu, 7 Jan 2016 22:04:34 +0000 (22:04 +0000)]
route: various small fixes

* Added w and d flag to man page list of options

* Updated output from route command to list w and c in list of options
  as well

* Added documentation in man page for d option

* Removed trailing whitespace on a few lines.

8 years agoif_run: Sync with FreeBSD up to r270643 (from August 26 2014)
Imre Vadász [Thu, 7 Jan 2016 21:35:54 +0000 (22:35 +0100)]
if_run: Sync with FreeBSD up to r270643 (from August 26 2014)

Taken-From: FreeBSD (SVN r263985 r267089 r270192 r270643)

8 years agokernel/tmpfs: Use the more common 'ap' name for vops arguments.
Sascha Wildner [Thu, 7 Jan 2016 18:30:36 +0000 (19:30 +0100)]
kernel/tmpfs: Use the more common 'ap' name for vops arguments.

8 years agohammer - remove debugging kprintf
Matthew Dillon [Thu, 7 Jan 2016 18:13:07 +0000 (10:13 -0800)]
hammer - remove debugging kprintf

* Remove a debugging kprintf related to mkdir() errors which can trigger
  under nominal conditions (attempting to create a subdirectory in a parent
  which had already been rmdir'd)

8 years agohangman(6): Update to augmented version from OpenBSD.
zrj [Wed, 6 Jan 2016 10:10:23 +0000 (12:10 +0200)]
hangman(6): Update to augmented version from OpenBSD.

This implements usage of extra wordlists:
  * kernel symbols, from current booted kernel image;
  * non-default, from Elf and conformant dictonaries.

While there, free hangman(6) from the shackles of dungeon master,
since hangman now could be used as a developer tool.

8 years agoif_run: Uncomment code from FreeBSD to defer run_wme_update_cb
Imre Vadasz [Wed, 6 Jan 2016 12:53:25 +0000 (13:53 +0100)]
if_run: Uncomment code from FreeBSD to defer run_wme_update_cb

- This should avoid a frequently occuring deadlock in if_run.
- This workaround is already removed in current FreeBSD code, where the
  net80211 layer now takes care of deferred calling of the wme_update
  callback into the driver.

8 years agokernel/tmpfs: Add _PC_2_SYMLINKS support. getconf(1) adjustments.
Sascha Wildner [Wed, 6 Jan 2016 10:33:39 +0000 (11:33 +0100)]
kernel/tmpfs: Add _PC_2_SYMLINKS support. getconf(1) adjustments.

8 years agogetconf(1): Sort variable/name pairs.
Sascha Wildner [Wed, 6 Jan 2016 10:30:42 +0000 (11:30 +0100)]
getconf(1): Sort variable/name pairs.

8 years agohals_end: Fix manlint.
zrj [Wed, 6 Jan 2016 07:40:33 +0000 (09:40 +0200)]
hals_end: Fix manlint.

While there, update .Dd in colorbars(6) too.

8 years agoif_iwm: Fix IEEE80211_ADDR_COPY() usage.
Imre Vadász [Sun, 3 Jan 2016 16:10:52 +0000 (17:10 +0100)]
if_iwm: Fix IEEE80211_ADDR_COPY() usage.

Taken-From: FreeBSD

8 years agoif_iwm: Always stop LED blinking when scanning is finished
Imre Vadász [Sun, 3 Jan 2016 15:57:02 +0000 (16:57 +0100)]
if_iwm: Always stop LED blinking when scanning is finished

When state is still IEEE80211_S_RUN after scanning is done, enable the LED.
This matches the behaviour of if_iwn(4).

8 years agopiano(6): Retire. R.I.P.
zrj [Mon, 4 Jan 2016 17:15:46 +0000 (19:15 +0200)]
piano(6): Retire. R.I.P.

Support for /dev/speaker was removed back in 2005.

8 years agogames: Import colorbars and hals_end from NetBSD.
zrj [Mon, 4 Jan 2016 16:57:42 +0000 (18:57 +0200)]
games: Import colorbars and hals_end from NetBSD.

This satisfies two comes in, one comes out rule.

8 years agomdoc.local: Add NetBSD 7.0 for future reference.
zrj [Mon, 4 Jan 2016 16:38:57 +0000 (18:38 +0200)]
mdoc.local: Add NetBSD 7.0 for future reference.

8 years agotest: ,,,^..^,,,~~~~
zrj [Tue, 5 Jan 2016 13:14:37 +0000 (15:14 +0200)]
test: ,,,^..^,,,~~~~

8 years ago<unistd.h>: Add _POSIX_THREAD_ROBUST_PRIO_{INHERIT,PROTECT} definitions.
Sascha Wildner [Tue, 5 Jan 2016 05:14:07 +0000 (06:14 +0100)]
<unistd.h>: Add _POSIX_THREAD_ROBUST_PRIO_{INHERIT,PROTECT} definitions.

Also adjust sysconf(3) and getconf(1) accordingly.

8 years ago<unistd.h>: Comment some POSIX options.
Sascha Wildner [Tue, 5 Jan 2016 03:35:51 +0000 (04:35 +0100)]
<unistd.h>: Comment some POSIX options.

8 years agore: Only probe fixed PHY location
Sepherosa Ziehau [Mon, 4 Jan 2016 14:38:12 +0000 (22:38 +0800)]
re: Only probe fixed PHY location

Obtained-from: FreeBSD

8 years agore: Reduce GMII PHY read/write timeout
Sepherosa Ziehau [Mon, 4 Jan 2016 14:15:58 +0000 (22:15 +0800)]
re: Reduce GMII PHY read/write timeout

Obtained-from: FreeBSD

8 years agore: Change CFG1 and CFG5 after flip EECMD
Sepherosa Ziehau [Mon, 4 Jan 2016 13:23:40 +0000 (21:23 +0800)]
re: Change CFG1 and CFG5 after flip EECMD

Obtained-from: FreeBSD

8 years agosbin/hammer: Fix hammer(8) manpage
Tomohiro Kusumi [Sun, 3 Jan 2016 13:40:43 +0000 (22:40 +0900)]
sbin/hammer: Fix hammer(8) manpage

Add a sentence that explains PFS via PFS is not allowed.

8 years agonl(1): Sync with FreeBSD.
Sascha Wildner [Sun, 3 Jan 2016 20:55:20 +0000 (21:55 +0100)]
nl(1): Sync with FreeBSD.

* multibyte char support

* use getline()

* treat '-' as standard input

8 years agocomm(1): Sync with FreeBSD.
Sascha Wildner [Sun, 3 Jan 2016 19:07:30 +0000 (20:07 +0100)]
comm(1): Sync with FreeBSD.

* multibyte char support

* fix truncation of lines at LINE_MAX

* use getline()

* make functions static

8 years agousr.bin/uniq: Sync with FreeBSD
zrj [Wed, 11 Nov 2015 07:26:27 +0000 (09:26 +0200)]
usr.bin/uniq: Sync with FreeBSD

Few improvements, mainly wchar_t support.

Obtained-from: FreeBSD

8 years agore: Disable ASPM
Sepherosa Ziehau [Sun, 3 Jan 2016 12:12:34 +0000 (20:12 +0800)]
re: Disable ASPM

Referenced: FreeBSD

8 years agore: Update Realtek PHY power management related bits
Sepherosa Ziehau [Sun, 3 Jan 2016 06:21:08 +0000 (14:21 +0800)]
re: Update Realtek PHY power management related bits

Obtained-from: FreeBSD

8 years agokqueue: Move notifymsglist out of kqinfo
Sepherosa Ziehau [Sun, 3 Jan 2016 02:21:43 +0000 (10:21 +0800)]
kqueue: Move notifymsglist out of kqinfo

It is only used by socket code.

8 years ago<strings.h>: Fix VKERNEL64 build.
Sascha Wildner [Sat, 2 Jan 2016 14:09:00 +0000 (15:09 +0100)]
<strings.h>: Fix VKERNEL64 build.

8 years agokernel/tmpfs: Implement _PC_FILESIZEBITS.
Sascha Wildner [Sat, 2 Jan 2016 12:53:10 +0000 (13:53 +0100)]
kernel/tmpfs: Implement _PC_FILESIZEBITS.

8 years agokernel: Add flsll(). Identical to flsl().
Sascha Wildner [Sat, 2 Jan 2016 12:50:03 +0000 (13:50 +0100)]
kernel: Add flsll(). Identical to flsl().

Add FreeBSD's libkern version too, but on x86_64 this is not compiled,
because we use inline assembly.

8 years agokernel/tmpfs: Sort pathconf cases.
Sascha Wildner [Sat, 2 Jan 2016 11:23:27 +0000 (12:23 +0100)]
kernel/tmpfs: Sort pathconf cases.

8 years agoif_iwm: Make the wifi LED work with iwm(4) (2/2)
François Tigeot [Sat, 2 Jan 2016 08:16:25 +0000 (09:16 +0100)]
if_iwm: Make the wifi LED work with iwm(4) (2/2)

Add forgotten if_iwm_led.c to sys/conf/files

8 years agoif_iwm: Make the wifi LED work with iwm(4).
Imre Vadász [Fri, 1 Jan 2016 21:41:23 +0000 (22:41 +0100)]
if_iwm: Make the wifi LED work with iwm(4).

- While the wifi device is scanning or trying to associate, the LED
  should blink now.
- When associated to a wifi network, the LED will stop blinking and stay
  enabled.

Taken-From: OpenBSD

8 years agoif_iwm: Remove comments referring to Linux iwlwifi source filenames.
Imre Vadász [Fri, 1 Jan 2016 21:26:02 +0000 (22:26 +0100)]
if_iwm: Remove comments referring to Linux iwlwifi source filenames.

Taken-From: OpenBSD

8 years agodump.8: Remove overlooked .Pp
Sascha Wildner [Fri, 1 Jan 2016 12:45:32 +0000 (13:45 +0100)]
dump.8: Remove overlooked .Pp

8 years agoSplit EXIT STATUS and DIAGNOSTICS sections in manual pages.
Sascha Wildner [Fri, 1 Jan 2016 00:07:02 +0000 (01:07 +0100)]
Split EXIT STATUS and DIAGNOSTICS sections in manual pages.

Also correct some DIAGNOSTICS -> EXIT STATUS where it described only
the exit codes.

8 years agoif_iwm: Just store the struct firmware pointer, like if_iwn does.
Imre Vadász [Thu, 31 Dec 2015 22:13:12 +0000 (23:13 +0100)]
if_iwm: Just store the struct firmware pointer, like if_iwn does.

if_iwm was passing the firmware data pointer to firmware_put(9), instead
of the struct firmware pointer that was returned by firmware_get(9).

8 years agoif_iwm: Apply some changes from OpenBSD, if_iwm.c rev 1.39 -> 1.42
Imre Vadász [Wed, 30 Dec 2015 16:01:11 +0000 (17:01 +0100)]
if_iwm: Apply some changes from OpenBSD, if_iwm.c rev 1.39 -> 1.42

- Use m_defrag(9) instead of rolling our own version of it.

- Limit the number of dma segments used for tx to IWM_MAX_SCATTER - 2.
  This avoids out-of-bound accesses on the segs array in iwm_tx().

- cleanup attach message

Taken-From: OpenBSD

8 years agoBump our copyrights to 2016. Happy new year! :)
Sascha Wildner [Thu, 31 Dec 2015 22:16:00 +0000 (23:16 +0100)]
Bump our copyrights to 2016. Happy new year! :)

8 years agonewfs_hammer.8: Remove two unneeded .Pp macros.
Sascha Wildner [Thu, 31 Dec 2015 20:18:53 +0000 (21:18 +0100)]
newfs_hammer.8: Remove two unneeded .Pp macros.

8 years agobus_dma.9: Fix bus_dmamap_load_mbuf_segment documentation of maxsegs arg.
Imre Vadász [Thu, 31 Dec 2015 17:21:54 +0000 (18:21 +0100)]
bus_dma.9: Fix bus_dmamap_load_mbuf_segment documentation of maxsegs arg.

The maxsegs argument of bus_dmamap_load_mbuf_segment is allowed to be
equal to the nsegments of the dma tag, i.e. doesn't have to be "less than",
but "not larger than" the nsegments.

8 years agosbin/mount_hammer: Fix mount_hammer(8) manpage
Tomohiro Kusumi [Sat, 5 Dec 2015 12:40:23 +0000 (21:40 +0900)]
sbin/mount_hammer: Fix mount_hammer(8) manpage

8 years agosbin/newfs_hammer: Fix newfs_hammer(8) manpage
Tomohiro Kusumi [Sat, 5 Dec 2015 12:40:23 +0000 (21:40 +0900)]
sbin/newfs_hammer: Fix newfs_hammer(8) manpage

8 years agokqueue: Use wakeup_one() to wake up kqueue scanner
Sepherosa Ziehau [Thu, 31 Dec 2015 10:18:04 +0000 (18:18 +0800)]
kqueue: Use wakeup_one() to wake up kqueue scanner

It hard to imagine an application to have multiple threads to
wait on one kqueue.  However, even if this kind of application
existed, wakeup_one() would not break them (admittedly they
might suffer uneven workload distribution).

In case the old behaviour is still wanted, sysctl
kern.kq_wakeup_one can be changed to 0.

This saves us 30Kipis/s ~ 50Kipis/s in kqueue+accept test.

8 years agokqueue: Reduce file/hash list token coverage
Sepherosa Ziehau [Thu, 31 Dec 2015 09:31:39 +0000 (17:31 +0800)]
kqueue: Reduce file/hash list token coverage

They should only cover file/hash list operations.  And reduce kqueue
token coverage on termination path; hash does not need protection
there.

8 years agodrm/i915: Convert hw level to user level when reading backlight value.
Imre Vadász [Tue, 29 Dec 2015 13:21:03 +0000 (14:21 +0100)]
drm/i915: Convert hw level to user level when reading backlight value.

Values written to, and then read back from the hw.backlight_level sysctl
should now match.

8 years agolibc/nls: Sync str{error,signal}() messages with <sys/{errno,signal}.h>.
Sascha Wildner [Tue, 29 Dec 2015 19:27:09 +0000 (20:27 +0100)]
libc/nls: Sync str{error,signal}() messages with <sys/{errno,signal}.h>.

Reported-by: zrj <rimvydas.jasinskas@gmail.com>
8 years agointro.2: Document errnos 89-93.
zrj [Tue, 29 Dec 2015 07:26:49 +0000 (09:26 +0200)]
intro.2: Document errnos 89-93.

8 years agosbin/newfs_hammer: Fix newfs_hammer(8) manpage
Tomohiro Kusumi [Tue, 29 Dec 2015 19:01:19 +0000 (04:01 +0900)]
sbin/newfs_hammer: Fix newfs_hammer(8) manpage

-L is always necessary, but two or more fs with the same -L is allowed.

8 years agosbin/newfs_hammer: Fix newfs_hammer(8) manpage
Tomohiro Kusumi [Tue, 29 Dec 2015 17:39:16 +0000 (02:39 +0900)]
sbin/newfs_hammer: Fix newfs_hammer(8) manpage

newfs_hammer has calculations and upper limit for boot area,
which don't result in default 4MB, but rather likely to be 64MB.
Also this isn't used for anything just like memory log area.

8 years agokqueue: Don't check list against NULL; list is never NULL.
Sepherosa Ziehau [Tue, 29 Dec 2015 11:04:36 +0000 (19:04 +0800)]
kqueue: Don't check list against NULL; list is never NULL.

Instead, do a quick and unlocked SLIST_EMPTY check on the list.

8 years agoSync man page stat.2 with the definitions in sys/stat.h
Michael Neumann [Tue, 29 Dec 2015 10:27:58 +0000 (11:27 +0100)]
Sync man page stat.2 with the definitions in sys/stat.h

8 years agokqueue: Mark user filter MPSAFE
Sepherosa Ziehau [Tue, 29 Dec 2015 10:16:37 +0000 (18:16 +0800)]
kqueue: Mark user filter MPSAFE

8 years agokqueue: Remove unapplied comment
Sepherosa Ziehau [Tue, 29 Dec 2015 10:12:58 +0000 (18:12 +0800)]
kqueue: Remove unapplied comment

8 years agosocket: Use {get,rel}pooltoken in netmsg_so_notify()
Sepherosa Ziehau [Tue, 29 Dec 2015 09:59:04 +0000 (17:59 +0800)]
socket: Use {get,rel}pooltoken in netmsg_so_notify()

8 years agobridge: Mark callouts MPSAFE
Sepherosa Ziehau [Tue, 29 Dec 2015 09:16:51 +0000 (17:16 +0800)]
bridge: Mark callouts MPSAFE

8 years agoig_hal: Allow enabling debug at runtime
Sepherosa Ziehau [Tue, 29 Dec 2015 08:01:03 +0000 (16:01 +0800)]
ig_hal: Allow enabling debug at runtime

8 years agoBUS_SETUP_INTR.9: Mention INTR_NOPOLL flag.
Imre Vadász [Mon, 28 Dec 2015 13:57:18 +0000 (14:57 +0100)]
BUS_SETUP_INTR.9: Mention INTR_NOPOLL flag.

8 years agopsm: Remove unused struct sigio *async
Imre Vadász [Mon, 28 Dec 2015 11:30:31 +0000 (12:30 +0100)]
psm: Remove unused struct sigio *async

8 years agopsm: Remove leftover PSM_HOOKAPM #if defined check
Imre Vadász [Mon, 28 Dec 2015 11:08:59 +0000 (12:08 +0100)]
psm: Remove leftover PSM_HOOKAPM #if defined check

8 years agoagp: Remove support for Sandy Bridge hardware
François Tigeot [Mon, 28 Dec 2015 21:41:53 +0000 (22:41 +0100)]
agp: Remove support for Sandy Bridge hardware

This GPU family is now fully handled by the drm/i915 driver.

8 years agodrm/i915: Use generic drm_plane helper functions
François Tigeot [Mon, 28 Dec 2015 20:46:50 +0000 (21:46 +0100)]
drm/i915: Use generic drm_plane helper functions

* Now than drm plane helpers have been fixed, we can use the generic
  drm_plane_update function

* This should fix some mouse cursor corruption and disappearance
  issues reported by various testers

8 years agodrm: Skip vblank wait for legacy cursor updates
François Tigeot [Mon, 28 Dec 2015 20:40:43 +0000 (21:40 +0100)]
drm: Skip vblank wait for legacy cursor updates

This commit fixes an issue with the generic drm_plane_helper_update
function where the mouse cursor position was updated long after the
mouse was actually moved.

Submitted-by: Matthew Dillon <dillon@apollo.backplane.com>
8 years agokernel/acpi_cpu: Fix argument name.
Sascha Wildner [Mon, 28 Dec 2015 16:37:23 +0000 (17:37 +0100)]
kernel/acpi_cpu: Fix argument name.

8 years agoacpi/pstate: Allow user to ignore type 0x80 notify
Sepherosa Ziehau [Mon, 28 Dec 2015 15:17:48 +0000 (23:17 +0800)]
acpi/pstate: Allow user to ignore type 0x80 notify

Some BIOSes just send stupid _PPC updates which do not make
sense, so allow users to ignore the stupidity through tunable
and sysctl.

8 years agoacpi/pstate: Support type 0x80 notify
Sepherosa Ziehau [Mon, 28 Dec 2015 13:55:36 +0000 (21:55 +0800)]
acpi/pstate: Support type 0x80 notify

- Protect power domain states by per-power-domain serializer.
- Protect global P-state setting by the first usable power domain's
  serializer.

8 years agokernel/vga: Remove wrong lwkt_reltoken(). It's not taken at this point.
Sascha Wildner [Mon, 28 Dec 2015 13:41:16 +0000 (14:41 +0100)]
kernel/vga: Remove wrong lwkt_reltoken(). It's not taken at this point.

8 years agoacpi/pstate: Factor out acpi_pst_eval_{ppc,pdl}()
Sepherosa Ziehau [Mon, 28 Dec 2015 13:00:38 +0000 (21:00 +0800)]
acpi/pstate: Factor out acpi_pst_eval_{ppc,pdl}()

They will be used to handle type 0x80 notify.

8 years agotest - refactor Makefile
Matthew Dillon [Mon, 28 Dec 2015 03:18:54 +0000 (19:18 -0800)]
test - refactor Makefile

* Refactor the Makefile, also generate binaries to /tmp insetead of to the
  current directory.

8 years agotest - Add testvblank program
Matthew Dillon [Mon, 28 Dec 2015 03:14:47 +0000 (19:14 -0800)]
test - Add testvblank program

* Test program opens /dev/dri/card0 and uses DRM ioctls to test drm's
  vblank wait code.

8 years agotest - cleanup build
Matthew Dillon [Mon, 28 Dec 2015 03:12:11 +0000 (19:12 -0800)]
test - cleanup build

* Get rid of -Wall warnings, cleanup Makefile

8 years agodevfs - Fix panic on extra devfs mounts when rules are present
Matthew Dillon [Mon, 28 Dec 2015 01:01:29 +0000 (17:01 -0800)]
devfs - Fix panic on extra devfs mounts when rules are present

* Fix issue where the rules-check was being made on the root mount
  node itself.

* devfs would panic on extra devfs mounts (such as in jails) when
  /etc/defaults/devfs.conf contains any rules.

Reported-by: marino
8 years agoagp: Remove support for Ivy Bridge hardware
François Tigeot [Sun, 27 Dec 2015 08:12:50 +0000 (09:12 +0100)]
agp: Remove support for Ivy Bridge hardware

This GPU family is now fully handled by the drm/i915 driver.

8 years agodrm - Fix panic on DRM_IOCTL_WAIT_VBLANK
Matthew Dillon [Sun, 27 Dec 2015 00:11:04 +0000 (16:11 -0800)]
drm - Fix panic on DRM_IOCTL_WAIT_VBLANK

* DRM_IOCTL_WAIT_VBLANK -> drm_wait_vblank() -> DRM_WAIT_ON() -> panic

  DRM_WAIT_ON() assumed that the device was locked and tried to unlock
  and relock it around the tsleep.  However, the ioctl paths that use
  this function do not lock the device.

* Remove the unlock/relock sequence.

* Tested on haswell using ioctl call w/vblank.request.type =
   _DRM_VBLANK_RELATIVE and vblank.request.sequence = 10.  Crash reproduced
   before fix, does not occur after fix.

Reported-by: arcade@b1t.name
8 years agodrm/i915: Upgrade to Linux 4.0
François Tigeot [Sat, 26 Dec 2015 13:09:06 +0000 (14:09 +0100)]
drm/i915: Upgrade to Linux 4.0

* Basic Skylake support

* Panel Self-Refresh (PSR) now supported on Valleyview and Cherryview

* Preparations for atomic display updates

* Performance improvements on various GPU families, including Cherryview,
  Broadwell and Haswell

* GPU frequencies are now kept at a minimum of 450MHz when possible
  on Haswell and Broadwell, ensuring a minimum experience level for
  various types of workloads

* Improved reset support for gen3/4 GPUs, which should fix some OpenGL
  crashes on Core 2 and pre-2012 Atom machines

* Better sound/graphics driver synchronization for audio over hdmi support

* As usual, small bugfixes and stability improvements here and there

8 years agodrm: Fix previous commit
François Tigeot [Sat, 26 Dec 2015 09:30:29 +0000 (10:30 +0100)]
drm: Fix previous commit

8 years agodrm: Create device entries with the "video" group
François Tigeot [Sat, 26 Dec 2015 08:16:27 +0000 (09:16 +0100)]
drm: Create device entries with the "video" group

* Add a new "video" group with id 44

* Make the drm subsystem create /dev/dri/card* devices with this
  new group

* FreeBSD and many Linux distributions already do the same thing

Suggested-by: Koop Mast
Partially-obtained-from: FreeBSD

8 years agoagp: Remove support for Haswell hardware
François Tigeot [Sat, 26 Dec 2015 07:36:43 +0000 (08:36 +0100)]
agp: Remove support for Haswell hardware

This GPU family is now fully handled by the drm/i915 driver.

8 years agolibc - fix getcontext bug from last commit
Matthew Dillon [Fri, 25 Dec 2015 21:56:01 +0000 (13:56 -0800)]
libc - fix getcontext bug from last commit

* getcontext() should not save the FP state (which is all scratch),
  and was already setting the flags to indicate such, but still calling
  the xsave instruction.

  Remove the xsave instruction, fixing issues on numerous computers.

* Note that setcontext() will properly restore the FP state if the context
  contains valid FP state (i.e. the ucontext was taken from a signal handler
  argument).

Reported-by: YRabbit
8 years agodrm: Make OpenGL usable by everyone
François Tigeot [Fri, 25 Dec 2015 11:59:39 +0000 (12:59 +0100)]
drm: Make OpenGL usable by everyone

By changing permissions on /dev/dri/card* .
These devices were previously only usable by root or wheel users.