Tejun reported stuttering and latency spikes on a system where random
tasks would enter direct reclaim and get stuck on dirty pages. Around
50% of memory was occupied by tmpfs backed by an SSD, and another disk
(rotating) was reading and writing at max speed to shrink a partition.
: The problem was pretty ridiculous. It's a 8gig machine w/ one ssd and 10k
: rpm harddrive and I could reliably reproduce constant stuttering every
: several seconds for as long as buffered IO was going on on the hard drive
: either with tmpfs occupying somewhere above 4gig or a test program which
: allocates about the same amount of anon memory. Although swap usage was
: zero, turning off swap also made the problem go away too.
:
: The trigger conditions seem quite plausible - high anon memory usage w/
: heavy buffered IO and swap configured - and it's highly likely that this
: is happening in the wild too. (this can happen with copying large files
: to usb sticks too, right?)
This patch (of 2):
The dirty_balance_reserve is an approximation of the fraction of free
pages that the page allocator does not make available for page cache
allocations. As a result, it has to be taken into account when
calculating the amount of "dirtyable memory", the baseline to which
dirty_background_ratio and dirty_ratio are applied.
However, currently the reserve is subtracted from the sum of free and
reclaimable pages, which is non-sensical and leads to erroneous results
when the system is dominated by unreclaimable pages and the
dirty_balance_reserve is bigger than free+reclaimable. In that case, at
least the already allocated cache should be considered dirtyable.
Fix the calculation by subtracting the reserve from the amount of free
pages, then adding the reclaimable pages on top.
[akpm@linux-foundation.org: fix CONFIG_HIGHMEM build]
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reported-by: Tejun Heo <tj@kernel.org>
Tested-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Prior to commit fe35004fbf ("mm: avoid swapping out with
swappiness==0") setting swappiness to 0, reclaim code could still evict
recently used user anonymous memory to swap even though there is a
significant amount of RAM used for page cache.
The behaviour of setting swappiness to 0 has since changed. When set,
the reclaim code does not initiate swap until the amount of free pages
and file-backed pages, is less than the high water mark in a zone.
Let's update the documentation to reflect this.
[akpm@linux-foundation.org: remove comma, per Randy]
Signed-off-by: Aaron Tomlin <atomlin@redhat.com>
Acked-by: Rik van Riel <riel@redhat.com>
Acked-by: Bryn M. Reeves <bmr@redhat.com>
Cc: Satoru Moriya <satoru.moriya@hds.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In the gen_pool_dma_alloc() the dma pointer can be NULL and while
assigning gen_pool_virt_to_phys(pool, vaddr) to dma caused the following
crash on da850 evm:
Unable to handle kernel NULL pointer dereference at virtual address 00000000
Internal error: Oops: 805 [#1] PREEMPT ARM
Modules linked in:
CPU: 0 PID: 1 Comm: swapper Tainted: G W 3.13.0-rc1-00001-g0609e45-dirty #5
task: c4830000 ti: c4832000 task.ti: c4832000
PC is at gen_pool_dma_alloc+0x30/0x3c
LR is at gen_pool_virt_to_phys+0x74/0x80
Process swapper, call trace:
gen_pool_dma_alloc+0x30/0x3c
davinci_pm_probe+0x40/0xa8
platform_drv_probe+0x1c/0x4c
driver_probe_device+0x98/0x22c
__driver_attach+0x8c/0x90
bus_for_each_dev+0x6c/0x8c
bus_add_driver+0x124/0x1d4
driver_register+0x78/0xf8
platform_driver_probe+0x20/0xa4
davinci_init_late+0xc/0x14
init_machine_late+0x1c/0x28
do_one_initcall+0x34/0x15c
kernel_init_freeable+0xe4/0x1ac
kernel_init+0x8/0xec
This patch fixes the above.
[akpm@linux-foundation.org: update kerneldoc]
Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Cc: Philipp Zabel <p.zabel@pengutronix.de>
Cc: Nicolin Chen <b42378@freescale.com>
Cc: Joe Perches <joe@perches.com>
Cc: Sachin Kamat <sachin.kamat@linaro.org>
Cc: <stable@vger.kernel.org> [3.13.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull fanotify use-after-free fixes from Jan Kara:
"Three fixes for the fanotify use after free problems guys were
reporting.
I have ended up with different lifetime rules for struct
fanotify_event_info depending on whether it is for permission event or
normal event which isn't ideal. My plan is to split these into two
different structures (as permission events need larger struct anyway)
which will make the rules trivial again. But that can wait for later
I guess (but I can add the patch to the pile if you want), now I
wanted to make -rc1 boot for these guys"
[ "These guys" being Jiri Kosina and Dave Jones that reported the slab
corruption issues due to incorrect object lifetimes ]
* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs:
fanotify: Fix use after free for permission events
fsnotify: Do not return merged event from fsnotify_add_notify_event()
fanotify: Fix use after free in mask checking
The merge of commit 7221fe4c2e ("ceph: add acl for cephfs") raced with
upstream changes in the generic POSIX ACL code (eg commit 2aeccbe957
"fs: add generic xattr_acl handlers" and others).
Some of the fallout was fixed in commit 4db658ea0c ("ceph: Fix up after
semantic merge conflict"), but it was incomplete: the set_acl
inode_operation wasn't getting set, and the prototype needed to be
adjusted a bit (it doesn't take a dentry anymore).
Signed-off-by: Sage Weil <sage@inktank.com>
Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
If either idling channels or suspending the fence were to fail, the
display would never be resumed. Also if a client fails, resume the fence
(not functionally important, but it would potentially leak memory).
See https://bugs.freedesktop.org/show_bug.cgi?id=70213
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Fixes a regression introduced by d5c1e84b3a
"drm/nouveau: hold mutex while syncing to kernel channel".
Cc: stable@vger.kernel.org #3.13
Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
The DRM uses the adjusted mode to calculate constants for vblank
timestamping. Our encoder mode_fixup (usually) replaces this data
with our backend mode information, which doesn't have the needed
data filled in already.
Reported-by: Mario Kleiner mario.kleiner.de@gmail.com
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Some DCE8 boards have a funky BlankCrtc table that results
in a timeout when trying to blank the display. The
timeout is harmless (all operations needed from the table
are complete), but wastes time and is confusing to users so
work around it.
bug:
https://bugs.freedesktop.org/show_bug.cgi?id=73420
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
This is effectively a revert of 4573388c92.
Forcing a display active when there is none causes problems with
dpm on some SI boards which results in improperly initialized
dpm state and boot failures on some boards. As for the bug commit
4573388c92 tried to address, one can manually force the state to
high for better performance when using the card as a headless compute
node until a better fix is developed.
bugs:
https://bugs.freedesktop.org/show_bug.cgi?id=73788https://bugs.freedesktop.org/show_bug.cgi?id=69395
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
cc: stable@vger.kernel.org
DCE5 and newer hardware only has 1 DAC. Use the correct
offset. This may fix display problems on certain board
configurations.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
If we are not able to properly initialize one of the gpu
engines for buffer paging, we limit vram to the size of
the cpu visible aperture. We generally either use the gfx
or dma engine to do this. Clean up the size limiting code
to only adjust the size based on what ring is selected
for buffer paging rather than making assumptions about which
engine is selected for paging.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
The hw is buggy and it's not currently used, but it's
currently still initialized by the driver. Skip the init.
Skipping init also seems to improve stability with dpm on
some r6xx asics.
bug:
https://bugs.freedesktop.org/show_bug.cgi?id=66963
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Prevent runtime suspend of non-PX GPUs. Runtime suspend is
not what we want in those cases.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Otherwise we allocate a new VMID on nearly every submit.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Remove old SPI control functions, change anti-pop init
sequence, remove some garbage from structures. The 'Apply' functions
must be called at the mixer initialization, otherwise
mixer settings sometimes will not be applied at startup.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Change the 'put' function of the high-pass filter control to use the new
SPI functions.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
First of all, we should not touch the GPIOs. They are not
for selecting the capture source, but they seems just enable
the whole audio input curcuit. The 'put' function calls the
'apply' functions to change register values. Change the order
of capture sources.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Modify the input_vol_* functions to use the new SPI routines,
There is a new applying function that will be called when
the capture source changed.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
I tried both variants: volume control and impedance selector.
In the first case one minus is that we can't change the
volume of multichannel output without additional software
volume control. However, I am using this variant for the
last three months and this seems good. All multichannel
speaker systems have internal amplifier with the
volume control included, but not all headphones have
this regulator. In the second case, my software volume
control does not save the value after reboot.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Change the order of elements in the output select control. This will
reduce the number of relay switches. Change 'put' function to call the
oxygen_update_dac_routing() function. Otherwise multichannel playback
does not work. Also there is a new function to apply settings, this
prevents from duplicating the code.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Actually CS4245 connected to the I2S channel 1 for
capture, not channel 2. Otherwise capturing and
playback does not work for CS4245.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Moving the mixer code away makes things easier. The mixer
will control the driver, so the functions of the
driver need to be non-static.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Change the function to read the data from the new shadow buffer.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
When selecting the audio output destinations (headphones,
FP headphones, multichannel output), the channel routing
should be changed depending on what destination selected.
Also unnecessary I2S channels are digitally muted. This
function called when the user selects the destination
in the ALSA mixer.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
When selecting the audio sample rate for CS4245,
the MCLK divider should also be changed.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Change CS4245 initialization: different sequence and GPIO values,
according to datasheets and reverse-engineering information.
Change cleanup/resume/suspend functions, since they use
initialization.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Add the new SPI write and read functions. The SPI read function
is used for creating initial registers dump and may be used for
debugging purposes. SPI operations are cached, so there is a new
function to manage the cache (shadow). I have to remove
the shift from the CS4245_SPI_* constants, since when
we are performing the reading, we need to shift by 8 instead
of 16.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Add additional constants to the xonar_dg.h file:
capture and playback sources. Move GPIO_* constants and the
dg struct to the header file from the xonar_dg.c file.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Add some additional information in comments and my copyright.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
When the user switches the output from stereo to multichannel
or vice versa, the driver needs to update the channel routing.
Instead of creating additional subroutines, I better export existing
oxygen_update_dac_routing symbol from the oxygen mixer
and call this function. It calls model.adjust_dac_routing()
and my function does the work.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
The Xonar DG/DGX driver needs this mask to mute unnecessary
channels.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Modify the oxygen_write_spi() function to use the newly
introduced oxygen_wait_spi() function. Change return value
from void to int, so it can return error codes. Older
drivers just ignore that return value, new drivers can
check this value. We need to wait AFTER
initiating the SPI transaction, otherwise read
operation will not work.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
The oxygen_wait_spi() function now performs waiting when the
SPI bus completes a transaction. Introduce the timeout error
checking and increase timeout to 200 from 40.
Signed-off-by: Roman Volkov <v1ron@mail.ru>
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
The driver prints IT8771F and IT8772F instead of IT8771E and IT8772E
respectively when the driver is loaded. This is a cosmetic only bug
but let's fix it.
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Add support for IT8603E.
This closes bug #57861:
https://bugzilla.kernel.org/show_bug.cgi?id=57861
[JD: Fixes and clean-ups.]
Signed-off-by: Rudolf Marek <r.marek@assembler.cz>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Move the test for res->sr_slot == NULL out of the nfs41_sequence_free_slot
helper and into the main function for efficiency.
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
The check for whether or not we sent an RPC call in nfs40_sequence_done
is insufficient to decide whether or not we are holding a session slot,
and thus should not be used to decide when to free that slot.
This patch replaces the RPC_WAS_SENT() test with the correct test for
whether or not slot == NULL.
Cc: Chuck Lever <chuck.lever@oracle.com>
Cc: stable@vger.kernel.org # 3.12+
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
When Hyper-V hypervisor leaves are present, KVM must relocate
its own leaves at 0x40000100, because Windows does not look for
Hyper-V leaves at indices other than 0x40000000. In this case,
the KVM features are at 0x40000101, but the old code would always
look at 0x40000001.
Fix by using kvm_cpuid_base(). This also requires making the
function non-inline, since kvm_cpuid_base() is static.
Fixes: 1085ba7f55
Cc: stable@vger.kernel.org
Cc: mtosatti@redhat.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
It is unnecessary to go through hypervisor_cpuid_base every time
a leaf is found (which will be every time a feature is requested
after the next patch).
Fixes: 1085ba7f55
Cc: stable@vger.kernel.org
Cc: mtosatti@redhat.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>