| Age | Commit message (Collapse) | Author |
|---|
|
Hey all-
I was tinkering with dccp recently and noticed that I BUG halted the
kernel when I rmmod-ed the dccp module. The bug halt occured because the page
that I passed to kfree failed the PageCompound and PageSlab test in the slub
implementation of kfree. I tracked the problem down to the following set of
events:
1) dccp, unlike all other uses of kmem_cache_create, allocates a string
dynamically when registering a slab cache. This allocated string is freed when
the cache is destroyed.
2) Normally, (1) is not an issue, but when Slub is in use, it is possible that
caches are 'merged'. This process causes multiple caches of simmilar
configuration to use the same cache data structure. When this happens, the new
name of the cache is effectively dropped.
3) (2) results in kmem_cache_name returning an ambigous value (i.e.
ccid_kmem_cache_destroy, which uses this fuction to retrieve the name pointer
for freeing), is no longer guaranteed that the string it assigned is what is
returned.
4) If such merge event occurs, ccid_kmem_cache_destroy frees the wrong pointer,
which trips over the BUG in the slub implementation of kfree (since its likely
not a slab allocation, but rather a pointer into the static string table
section.
So, what to do about this. At first blush this is pretty clearly a leak in the
information that slub owns, and as such a slub bug. Unfortunately, theres no
really good way to fix it, without exposing slub specific implementation details
to the generic slab interface. Also, even if we could fix this in slub cleanly,
I think the RCU free option would force us to do lots of string duplication, not
only in slub, but in every slab allocator. As such, I'd like to propose this
solution. Basically, I just move the storage for the kmem cache name to the
ccid_operations structure. In so doing, we don't have to do the kstrdup or
kfree when we allocate/free the various caches for dccp, and so we avoid the
problem, by storing names with static memory, rather than heap, the way all
other calls to kmem_cache_create do.
I've tested this out myself here, and it solves the problem quite well.
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
sh64 on the other hand provides both direct broken out syscalls as well
as socketcall access. As there are binaries that use both socketcall has
to stay around. The current ABI prefers direct syscalls.
It was pointed out that when sys_recvmmsg was added in, sys_accept4 was
overlooked. This takes care of wiring it up.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
|
|
sh32 at the moment only uses sys_socketcall to reach these, so unwire
recvmmsg for now. While we're at it, add it to the ignore list, as per
the s390 change.
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
|
|
* git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog:
[WATCHDOG] iTCO_wdt: Add Intel Cougar Point and PCH DeviceIDs
|
|
We can use logical flat mode if there are <= 8 logical cpu's
(irrespective of physical apic id values). This will enable simplified
and efficient IPI and device interrupt routing on such platforms.
This has been tested to work on both Intel and AMD platforms.
Exceptions like IBM summit platform which can't use logical flat mode
are addressed by using OEM platform checks.
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Chris McDermott <lcm@linux.vnet.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Chris McDermott from IBM confirmed that hurricane chipset in IBM summit
platforms doesn't support logical flat mode. Irrespective of the other
things like apic_id's, total number of logical cpu's, Linux kernel
should default to physical mode for this system.
The 32-bit kernel does so using the OEM checks for the IBM summit
platform. Add a similar OEM platform check for the 64bit kernel too.
Otherwise the linux kernel boot can hang on this platform under certain
bios/platform settings.
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Tested-by: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Chris McDermott <lcm@linux.vnet.ibm.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
ALSA: Remove warning message for invalid OSS minor ranges
ALSA: hda - Fix capture on Sony VAIO with single input
ALSA: hda - Fix mute led GPIO on HP dv-series notebooks
ALSA: use subsys_initcall for sound core instead of module_init
ALSA: hda - Fix missing capture mixer for ALC861/660 codecs
ALSA: hda - Improved MacBook (Pro) 5,1 / 5,2 support
ALSA: hda - Fix Toshiba NB20x quirk entry
|
|
* 'for-linus' of git://git.monstr.eu/linux-2.6-microblaze:
microblaze: pci_controller->arch_data really is a struct device_node *
microblaze: Add missing double apostrophe in Kconfig
microblaze: Add PT_ macros for special purpose regs
microblaze: Enable accept4 syscall
microblaze: Wire up recvmmsg syscall
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/sameo/mfd-2.6
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sameo/mfd-2.6:
mfd: Unlock mc13783 before subsystems initialisation, at probe time.
mfd: WM835x GPIO direction register is not locked
mfd: tmio_mmc hardware abstraction for CNF area
mfd: WM8350 off by one bug
mfd: Correct WM835x ISINK ramp time defines
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc
* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc:
powerpc: Move cpu hotplug driver lock from pseries to powerpc
powerpc: Move /proc/ppc64 to /proc/powerpc update
powerpc/8xx: Fix user space TLB walk in dcbX fixup
powerpc: Fix decrementer setup on 1GHz boards
powerpc/iseries: Initialise on-stack completion
powerpc/hvc: Driver build breaks with !HVC_CONSOLE
serial/pmac_zilog: Workaround problem due to interrupt on closed port
powerpc/macintosh: Make Open Firmware device id constant
powerpc: Use helpers for rlimits
powerpc: cpumask_of_node() should handle -1 as a node
powerpc/pseries: Fix dlpar compile warning without CONFIG_PROC_DEVICETREE
powerpc/pseries: Fix xics interrupt affinity
powerpc/swsusp_32: Fix TLB invalidation
powerpc/8xx: Always pin kernel instruction TLB
powerpc: 2.6.33 update of defconfigs for embedded 6xx/7xxx, 8xx, 8xxx
powerpc: Use scripts/mkuboot.sh instead of 'mkimage'
powerpc/5200: update defconfigs
|
|
* 'for-linus' of git://oss.sgi.com/xfs/xfs:
xfs: xfs_swap_extents needs to handle dynamic fork offsets
xfs: fix missing error check in xfs_rtfree_range
xfs: fix stale inode flush avoidance
xfs: Remove inode iolock held check during allocation
xfs: reclaim all inodes by background tree walks
xfs: Avoid inodes in reclaim when flushing from inode cache
xfs: reclaim inodes under a write lock
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-2.6
* 'mantis' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-2.6: (117 commits)
V4L/DVB (13851): Fix Input dependency for Mantis
V4L/DVB(13824a): mantis: Fix __devexit bad annotations
V4L/DVB (13808b): mantis: replace DMA_nnBIT_MASK to DMA_BIT_MASK(32)
V4L/DVB (13808): [Mantis/Hopper] Build update for Mantis/Hopper based cards
V4L/DVB(13808a): mantis: convert it to the new ir-core register/unregister functions
V4L/DVB (13812): [Mantis/Hopper] Update Copyright header
V4L/DVB (13811): [MB86A16] Update Copyright header
V4L/DVB (13810): [MB86A16] Use DVB_* macros
V4L/DVB (13809): Fix Checkpatch violations
V4L/DVB (13807): Fix: Free device in the device registration failure case
V4L/DVB (13806): Register and Initialize Remote control
V4L/DVB (13805): Fix: Unregister the frontend before detaching
V4L/DVB (13804): Remove unused I2C Adapter ID
V4L/DVB (13803): Remove unused dependency on CU1216
V4L/DVB (13802): [Mantis/Hopper] Fix all build related warnings
V4L/DVB (13801): [MB86A16] Use the search callback
V4L/DVB (13800): [Mantis] I2C optimization. Required delay is much lesser than 1mS.
V4L/DVB (13799): [Mantis] Unregister frontend
V4L/DVB (13798): [Mantis] Enable power for all cards, use byte mode only on relevant devices
V4L/DVB (13797): [Mantis/Hopper/TDA665x] Large overhaul,
...
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-2.6
* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-2.6: (23 commits)
V4L/DVB (13966): DVB-T regression fix for saa7134 cards
V4L/DVB (13955): cx25821: fix double unlock in medusa_video_init()
MAINTAINERS: ivtv-devel is moderated
MAINTAINERS: Andy Walls is the new ivtv maintainer
V4L/DVB (13941): rj54n1cb0c: remove compiler warning
V4L/DVB sh_mobile_ceu: don't check platform_get_irq's return value against zero
V4L/DVB mx1_camera: don't check platform_get_irq's return value against zero
V4L/DVB (13934): tda8290: Fix FM radio easy programming standard selection for TDA8295
V4L/DVB (13900): gspca - sunplus: Fix bridge exchanges.
V4L/DVB (13887): tda8290: add autodetection support for TDA8295c2
V4L/DVB (13882): gspca - stv06xx-vv6410: Ensure register STV_SCAN_RATE is zero
V4L/DVB (13880): gspca - m5602-s5k4aa: Add vflip quirk for the Amilo Xi 2428
V4L/DVB (13875): gspca - vc032x: Fix a possible crash with the vc0321 bridge.
V4L/DVB (13868): gspca - sn9c20x: Fix test of unsigned.
V4L/DVB (13858): ir-keytable: use the right header
feature-removal-schedule: Add v4l1 drivers obsoleted by gspca sub drivers
V4L/DVB (13622): gspca - ov534: Fix a compilation warning.
V4L/DVB (13834): dib8000: fix compilation if !DVB_DIB8000
V4L/DVB (13831): uvcvideo: Fix oops caused by a race condition in buffer dequeuing
V4L/DVB (13829): uvcvideo: Fix alternate setting selection in isochronous mode
...
|
|
This patch adds the Intel Cougar Point and PCH DeviceIDs for iTCO Watchdog.
Signed-off-by: Seth Heasley <seth.heasley@intel.com>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
|
|
Yasuhiro ABE <yadiary@gmail.com> reported success in sourceforge zd1211-dev list.
The device is a NEC Aterm WL54GU usb wireless stick.
The brand and retail product name
NEC, Aterm PA-WL54GU
The USB ID's (duh)
ID 0409:0248
The chip ID string
zd1211rw 1-1:1.0: zd1211b chip 0409:0248 v4810 high 00-1b-8b AL2230S_RF pa0 g--N-
The FCC ID
unknown
Signed-off-by: Hin-Tak Leung <htl10@users.sourceforge.net>
Signed-off-by: Yasuhiro ABE <yadiary@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
|
|
A long time ago, a user reported several crashes due to
data corruptions which are likely the result of a
not-100%-supported, or faulty? PCI bridge.
( http://patchwork.kernel.org/patch/53004/ )
This patch fixes entry #1.
"1. p54p_check_rx_ring - skb_over_panic: Under a ping flood
or just left running for a bit would panic with a skb_over_panic."
As described in the mail: The invalid frame length causes
skb_put to bailout and trigger a crash.
Note:
Simply dropping the frame is problematic, because if its content
contains a tx feedback we would lose some portion of the device
memory space.... And the driver/mac80211 should handle all other
invalid data.
Reported-by: Quintin Pitts <geek4linux@gmail.com>
Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
|
|
While testing the pid rate controller in mac80211_hwsim, I noticed
that once the controller reached 54 Mbit rates, it would fail to
lower the rate when necessary. The debug log shows:
1945 186786 pf_sample 50 3534 3577 50
My interpretation is that the fixed point scaling of the target
error value (pf) is incorrect: the error value of 50 compared to
a target of 14 case should result in a scaling value of
(14-50) = -36 * 256 or -9216, but instead it is (14 * 256)-50, or
3534.
Correct this by doing fixed point scaling after subtraction.
Signed-off-by: Bob Copeland <me@bobcopeland.com>
Acked-by: Stefano Brivio <stefano.brivio@polimi.it>
Acked-by: Mattias Nissler <mattias.nissler@gmx.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
|
|
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
|
|
Similar to 6000 and 1000 series, RTS/CTS is the recommended protection
mechanism for 5000 series in HT mode based on the HW design.
Using RTS/CTS will better protect the inner exchange from interference,
especially in highly-congested environment, it also prevent uCode encounter
TX FIFO underrun and other HT mode related performance issues.
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
CC: stable@kernel.org
Signed-off-by: John W. Linville <linville@tuxdriver.com>
|
|
When truncating an UBI volume, UBI should allocates a PEB-sized
buffer but does not release it, which leads to memory leaks.
This patch fixes the issue.
Reported-by: Marek Skuczynski <mareksk7@gmail.com>
Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Tested-by: Marek Skuczynski <mareksk7@gmail.com>
Cc: stable@kernel.org
|
|
we are follow powerpc change:
44ef339073f67d4abcc62ae52a5fbc069d7a4d29
Signed-off-by: Michal Simek <monstr@monstr.eu>
|
|
Signed-off-by: Michal Simek <monstr@monstr.eu>
|
|
PT_ macros are used by gdb and strace uses them too.
Signed-off-by: Michal Simek <monstr@monstr.eu>
|
|
We had wrong name in unistd.h + I wire up this syscall
in syscall table.
Signed-off-by: Michal Simek <monstr@monstr.eu>
|
|
Patch a2e2725541fad72416326798c2d7fa4dafb7d337 should
contain change in unistd.h too. The same problem
had MIPS.
Signed-off-by: Michal Simek <monstr@monstr.eu>
|
|
This reverts commit 14f0aa359365e8a93a77b71e3b840274b9b4dcb1.
That commit was needed earlier because system call restarting for
OABI (compat) required an executable stack and thus had problems
with NX. Since ab72b00734ae4d0b ("ARM: Fix signal restart issues
with NX and OABI compat") has reworked the code to not require an
executable stack anymore, we can re-enable NX support for kernels
with OABI (compat) support.
Signed-off-by: Rabin Vincent <rabin@rab.in>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
|
|
Marc reported that the BUG_ON in clockevents_notify() triggers on his
system. This happens because the kernel tries to remove an active
clock event device (used for broadcasting) from the device list.
The handling of devices which can be used as per cpu device and as a
global broadcast device is suboptimal.
The simplest solution for now (and for stable) is to check whether the
device is used as global broadcast device, but this needs to be
revisited.
[ tglx: restored the cpuweight check and massaged the changelog ]
Reported-by: Marc Dionne <marc.c.dionne@gmail.com>
Tested-by: Marc Dionne <marc.c.dionne@gmail.com>
Signed-off-by: Xiaotian Feng <dfeng@redhat.com>
LKML-Reference: <1262834564-13033-1-git-send-email-dfeng@redhat.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@kernel.org
|
|
|
|
When a card instance with a higher card number is registered, warning
messages are spewed eventually with stack traces due to the invalid minor
number for OSS device registration. For example, thinkpad-acpi registers
the card number 29 as default, and you'll see always these messages.
This is rather confusing (and worries users), thus better to return
simply the error code.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
|
|
flush(void)
Because of the include of the decompress_inflate.c file from
boot/compress/misc.c, there are different flush() defines:
In file included from arch/arm/boot/compressed/misc.c:249:
arch/arm/boot/compressed/../../../../lib/decompress_inflate.c:138:29: error: macro "flush" passed 2 arguments, but takes just 0
Fix this by removing the define of flush() in misc.c for
CONFIG_DEBUG_ICEDCC as it's already defined in mach/uncompress.h,
and that is being included unconditionally.
Also use a static inline function instead of define
for mach-mxc and mach-gemini to avoid similar bug
for those platforms.
Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
|
|
This converts the cpu_relax() to a udelay(1), which fixes up issues with
the EEPROM polling occasionally timing out.
Signed-off-by: Kuninori Morimoto <morimoto.kuninori@renesas.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
|
|
With this, mc13783 subsystems drivers can configure the mc13783 chip
reading and writing registers.
Signed-off-by: Alberto Panizzo <maramaopercheseimorto@gmail.com>
Acked-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>
|
|
No need to set the security key when writing to it.
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Cc: stable@kernel.org
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>
|
|
This patch abstracts out the CNF area code from tmio_mmc which
is not present in all hardware that can use this driver. This
is required so that we can support non-toshiba based hardware.
ASIC3 support by Philipp Zabel
Signed-off-by: Ian Molton <ian@mnementh.co.uk>
Signed-off-by: Magnus Damm <damm@opensource.se>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>
|
|
If irq == WM8350_NUM_IRQ that would put us past the end of the array.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>
|
|
The constants used to specify ISINK ramp times for WM835x had the
wrong shifts so that the on times applied to the off ramp and vice
versa. The masks for the bitfields are correct.
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Cc: stable@kernel.org
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/frederic/random-tracing into perf/core
|
|
/proc/net/rt_acct is not created if NET_CLS_ROUTE=n.
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
We can race with the unmount of an fs and the stopping of a kthread where we
will free the block group before we're done using it. The reason for this is
because we do not hold a reference on the block group while its caching, since
the allocator drops its reference once it exits or moves on to the next block
group. This patch fixes the problem by taking a reference to the block group
before we start caching and dropping it when we're done to make sure all
accesses to the block group are safe. Thanks,
Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
|
|
It is legal for btrfs_set_acl to be sent a NULL acl. This
makes sure we don't dereference it. A similar patch was sent by
Johannes Hirte <johannes.hirte@fem.tu-ilmenau.de>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
|
|
Currently orphan cleanup only ever gets triggered if we cross subvolumes during
a lookup, which means that if we just mount a plain jane fs that has orphans in
it, they will never get cleaned up. This results in panic's like these
http://www.kerneloops.org/oops.php?number=1109085
where adding an orphan entry results in -EEXIST being returned and we panic. In
order to fix this, we check to see on lookup if our root has had the orphan
cleanup done, and if not go ahead and do it. This is easily reproduceable by
running this testcase
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <string.h>
#include <unistd.h>
#include <stdio.h>
int main(int argc, char **argv)
{
char data[4096];
char newdata[4096];
int fd1, fd2;
memset(data, 'a', 4096);
memset(newdata, 'b', 4096);
while (1) {
int i;
fd1 = creat("file1", 0666);
if (fd1 < 0)
break;
for (i = 0; i < 512; i++)
write(fd1, data, 4096);
fsync(fd1);
close(fd1);
fd2 = creat("file2", 0666);
if (fd2 < 0)
break;
ftruncate(fd2, 4096 * 512);
for (i = 0; i < 512; i++)
write(fd2, newdata, 4096);
close(fd2);
i = rename("file2", "file1");
unlink("file1");
}
return 0;
}
and then pulling the power on the box, and then trying to run that test again
when the box comes back up. I've tested this locally and it fixes the problem.
Thanks to Tomas Carnecky for helping me track this down initially.
Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
|
|
Fix bug reported by Johannes Hirte. The reason of that bug
is btrfs_del_items is called after btrfs_duplicate_item and
btrfs_del_items triggers tree balance. The fix is check that
case and call btrfs_search_slot when needed.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
|
|
Stanse found 2 memory leaks in relocate_block_group and
__btrfs_map_block. cluster and multi are not freed/assigned on all
paths. Fix that.
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: linux-btrfs@vger.kernel.org
Signed-off-by: Chris Mason <chris.mason@oracle.com>
|
|
Some callers of btrfs_ordered_update_i_size can now pass in
a NULL for the ordered extent to update against. This makes
sure we properly align the offset they pass in when deciding
how much to bump the on disk i_size.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
|
|
parent 49313cdac7b34c9f7ecbb1780cfc648b1c082cd7 (v2.6.32-1-g49313cd)
commit ff48c08e1c05c67e8348ab6f8a24de8034e0e34d
Author: Jan Engelhardt <jengelh@medozas.de>
Date: Wed Dec 9 22:57:36 2009 +0100
Btrfs: fix missing last-entry in readdir(3)
When one does a 32-bit readdir(3), the last entry of a directory is
missing. This is however not due to passing a large value to filldir,
but it seems to have to do with glibc doing telldir or something
quirky. In any case, this patch fixes it in practice.
Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
|
|
On most cards the DisplayPort connector is created with 2 encoders sharing
a single SOR (for native DP, and for DVI-over-DP). The previous logic
for turning off unused encoders didn't take into account that we could
have multiple drm_encoders on a single hw encoder and ended up turning off
encoders that were actually being used still.
This patch fixes that issue. We probably want to look at something a bit
better later on, and only expose one drm_encoder per hw encoder block.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
GPU pointer to the structure is shifted right by 10 bits, so we need to
align to 1024 bytes, not 256.
Reported-by: Maarten Maathuis <madman2003@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Currently, we take down the sgdma engine without evicting all buffers
from VRAM.
The TTM device release will try to evict anything in VRAM to GART
memory, but this will fail since sgdma has already been taken down.
This causes an infinite loop in kernel mode on module unload.
It usually doesn't happen because there aren't any buffer on close.
However, if the GPU is locked up, this condition is easily triggered.
This patch fixes it in the simplest way possible by cleaning VRAM
right before cleaning SGDMA memory.
Signed-off-by: Luca Barbieri <luca@luca-barbieri.com>
Signed-off-by: Francisco Jerez <currojerez@riseup.net>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Currently Nouveau is unable to dismiss DMA_VTX_PROTECTION errors,
which results in an infinite loop in the interrupt handler.
These errors are caused both by bugs in the Gallium driver and by
user-specified index buffers with out of bounds indices.
By mmio-tracing the nVidia drivers, I found out how this is done.
On DMA_VTX_PROTECTION, The nVidia driver reads the register 0x402000,
always getting the value 4, and then writes 4 back to 0x402000.
This patch adds that logic by reading 0x402000 and writing the same
value back.
It's unclear what should happen if the value read is not 4, and
the current approach might not be the correct one.
To test this, modify mesa/progs/trivial/vbo-drawrange.c, defining
ELTOBJ to 1 and replacing indices with huge out of bounds integers.
Without this patch, the GPU and/or kernel should lock up.
With this patch, it should misrender as expected but not lock up.
The errors are still logged since they are useful for development.
This has been tested on NV49 and may not work on other cards.
To find out how things work on other cards, run the aforementioned
test using the blob with mmiotrace and grep for a read of the PGRAPH
source register.
Signed-off-by: Luca Barbieri <luca@luca-barbieri.com>
Signed-off-by: Francisco Jerez <currojerez@riseup.net>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
