OSDN Git Service
Tomi Valkeinen [Mon, 1 Aug 2011 11:32:23 +0000 (14:32 +0300)]
HACK: OMAP: DSS2: clk hack for OMAP2/3
The HWMOD data for OMAP2 and 3 are currently not up to date regarding
DSS (OMAP4 HWMOD data is fine). This patch makes the DSS driver to get
the opt clocks needed for OMAP2/3 with the old clock names, thus
allowing DSS driver to use runtime PM.
The HWMOD databases should be fixes ASAP, and this patch can be reverted
after that.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Wed, 1 Jun 2011 12:56:39 +0000 (15:56 +0300)]
OMAP: DSS2: DSS: Fix context save/restore
The current method of saving and restoring the context could cause a
restore before saving, effectively "restoring" zero values to registers.
Add ctx_valid field to indicate if the saved context is valid and can be
restored.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Wed, 1 Jun 2011 12:54:06 +0000 (15:54 +0300)]
OMAP: DSS2: DISPC: Fix context save/restore
The current method of saving and restoring the context could cause a
restore before saving, effectively "restoring" zero values to registers.
Add ctx_valid field to indicate if the saved context is valid and can be
restored.
Also restructure the code to save the ctx_loss_count in save_context(),
which makes more sense than the previous method of storing new
ctx_loss_count in dispc_need_ctx_restore.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Thu, 26 May 2011 11:54:05 +0000 (14:54 +0300)]
OMAP: DSS2: Remove ctx loss count from dss.c
dss.c only saves 1 register (3 in OMAP3) so the extra overhead from
need_ctx_restore & co. is probably bigger than the time spent saving and
restoring those few registers every time.
So remove the code from dss.c and restore context every time dss has
been off.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Wed, 25 May 2011 13:22:26 +0000 (16:22 +0300)]
OMAP: DSS2: Remove unused code from display.c
oh_core variable is no longer used, so it and its initialization can be
removed.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Thu, 19 May 2011 11:12:26 +0000 (14:12 +0300)]
OMAP: DSS2: DISPC: remove finegrained clk enables/disables
dispc.c enables and disables clocks in almost every function to make
sure the clocks are enabled when the function is called. This is rather
unoptimal way to handle the problem.
With pm_runtime other components have to call dispc_runtime_get() to
enable dispc clocks before calling any other dispc functions. Thus the
finegrained clk enables/disables can be removed.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Fri, 27 May 2011 07:52:38 +0000 (10:52 +0300)]
OMAP: DSS2: Remove unused opt_clock_available
opt_clock_available() is no longer needed, so remove it.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Fri, 27 May 2011 07:52:19 +0000 (10:52 +0300)]
OMAP: DSS2: Use PM runtime & HWMOD support
Use PM runtime and HWMOD support to handle enabling and disabling of DSS
modules.
Each DSS module will have get and put functions which can be used to
enable and disable that module. The functions use pm_runtime and hwmod
opt-clocks to enable the hardware.
Acked-by: Kevin Hilman <khilman@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Wed, 18 May 2011 09:58:24 +0000 (12:58 +0300)]
OMAP: DSS2: Remove CONFIG_OMAP2_DSS_SLEEP_BEFORE_RESET
CONFIG_OMAP2_DSS_SLEEP_BEFORE_RESET is used to avoid an unclear bug at
DSS reset time. The pm runtime will handle reset in the future, and this
code has to be removed. Hopefully we won't see this error anymore.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Wed, 18 May 2011 09:37:58 +0000 (12:37 +0300)]
OMAP: DSS2: Remove core_dump_clocks
Currently dss.c does all the low level clock handling in the DSS, and
thus it contains pointers to all the clocks. This allows dss.c to dump
the clock information for all the clocks.
With pm_runtime this is no longer the case, as each submodule will
handle its clocks independently. Thus remove the core_dump_clocks
function as it cannot be used with pm_runtime.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Wed, 18 May 2011 08:33:44 +0000 (11:33 +0300)]
OMAP: DSS2: DPI: remove unneeded SYSCK enable/disable
DSI PLL requires sys_clk to function, and DPI enables sys_clk when it
wants to use DSI PLL. However, DSI PLL code already handles enabling
sys_clk, so DPI's sys_clk code is extra.
Remove the unneeded sys_clk handling from dpi.c.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Mon, 23 May 2011 12:50:47 +0000 (15:50 +0300)]
OMAP: DSS2: Use omap_pm_get_dev_context_loss_count to get ctx loss count
Initialize get_context_loss_count in the DSS board data to
omap_pm_get_dev_context_loss_count, so that omapdss driver can use it.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Mon, 23 May 2011 12:46:54 +0000 (15:46 +0300)]
OMAP: DSS2: rewrite use of context_loss_count
The function to get device's context loss count has changed from
omap_pm_get_last_off_on_transaction_id() to
omap_pm_get_dev_context_loss_count()
Change name of the function pointer in omapdss.h accordingly, and use
the term "context loss count" instead of "context id" in the code.
Restructure the context loss count functions to handle errors properly,
and ensure that context is always considered lost if an error happens.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Mon, 23 May 2011 12:39:36 +0000 (15:39 +0300)]
OMAP: DSS2: Remove clk optimization at dss init
DSS enables core clocks for the duration of initialization to avoid
unnecessary context saves and restores.
With PM runtime the clocks cannot be handled in this way, outside the
dss module drivers. Thus we need to remove the optimization.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Tue, 17 May 2011 13:10:59 +0000 (16:10 +0300)]
OMAP: DSS2: Fix init and unit sequence
The initialization order of the DSS modules is important when pm_runtime
support is implemented. Currently RFBI is initialized before DISPC,
which will cause problems with pm_runtime as RFBI uses DISPC.
The same goes for uninitialization order, and dss_uninit needs to be
called last, and dispc_uninit just before that.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Mon, 16 May 2011 10:52:51 +0000 (13:52 +0300)]
OMAP: DSS2: Clean up probe for DSS & DSI
Both dss.c and dsi.c had a probe function, which was almost a dummy one,
calling dss_init() and dsi_init().
Remove the init functions by moving the initialization code into probe
functions.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Mon, 16 May 2011 10:43:04 +0000 (13:43 +0300)]
OMAP: DSS2: Handle dpll4_m4_ck in dss_get/put_clocks
Get and put for dpll4_m4_ck was handled in dss_init/dss_exit. Move the
code to dss_get/put_clocks(), which is a better place to handle it.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Tue, 21 Jun 2011 06:35:36 +0000 (09:35 +0300)]
OMAP: DSS2: Fix FIFO threshold and burst size for OMAP4
The DMA FIFO threshold registers and burst size registers have changed
for OMAP4. The current code only handles OMAP2/3 case, and so the
values are a bit off for OMAP4. A summary of the differences between
OMAP2/3 and OMAP4:
Burst size:
OMAP2/3: 4 x 32 bits / 8 x 32 bits / 16 x 32 bits
OMAP4: 2 x 128 bits / 4 x 128 bits / 8 x 128 bits
Threshold size:
OMAP2/3: in bytes (8 bit units)
OMAP4: in 128bit units
This patch fixes the issue by creating two new helper functions in
dss_features: dss_feat_get_buffer_size_unit() and
dss_feat_get_burst_size_unit(). These return (in bytes) the unit size
for threshold registers and unit size for burst size register,
respectively, and are used to calculate correct values.
For the threshold size the usage is straightforward. However, the burst
size register has different multipliers for OMAP2/3 and OMAP4. This
patch solves the problem by defining the multipliers for the burst size
as 2x, 4x and 8x, which fit fine for the OMAP4 burst size definition
(i.e. burst size unit for OMAP4 is 128bits), but requires a slight twist
on OMAP2/3 by defining the burst size unit as 64bit.
As the driver in practice always uses the maximum burst size, and no use
case currently exists where we would want to use a smaller burst size,
this patch changes the driver to hardcode the burst size when
initializing DISPC. This makes the threshold configuration code somewhat
simpler.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Thu, 16 Jun 2011 11:34:06 +0000 (14:34 +0300)]
OMAP: DSS2: DSI: sync when disabling a display
When the panel driver calls omapdss_dsi_display_disable() it is possible
that there are still some unsent packets in the TX fifo.
Add dsi_sync_vc() calls in the beginning of
omapdss_dsi_display_disable() to make sure the TX fifos are empty.
This allows us to remove the msleep(10) hack from panel-taal.c
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Tue, 21 Jun 2011 06:34:30 +0000 (09:34 +0300)]
OMAP: DSS2: Add Color Phase Rotation support
Add Color Phase Rotation (CPR) support and sysfs files to enable CPR and
to set the CPR coefficient matrix.
CPR is enabled via manager?/cpr_enable file, and the coefficient matrix
is set via manager?/cpr_coef file. The values in cpr_coef are in the
following order:
RR RG RB GR GG GB BR BG BB
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Nishant Kamat [Mon, 20 Jun 2011 15:09:28 +0000 (20:39 +0530)]
OMAP: DSS: Minor cleanup in ovl and mgr cache structs
The overlay_cache_data and manager_cache_data structs include
the elements of omap_overlay_info and omap_overlay_manager_info
structs respectively. Include the structs instead of the individual
elements to reduce code.
Signed-off-by: Nishant Kamat <nskamat@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Fri, 27 May 2011 11:22:16 +0000 (14:22 +0300)]
OMAP: DSS2: Add new FEAT definitions for features missing from OMAP2
OMAP2 doesn't have CPR, PRELOAD nor FIR_COEF_V registers. Add new
feature definitions for those, and check the feature before accessing
those registers.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Wed, 18 May 2011 08:59:21 +0000 (11:59 +0300)]
OMAP: DSS2: Add FEAT_VENC_REQUIRES_TV_DAC_CLK
OMAP3430 requires an 96MHz clock to VENC's DAC, but no other OMAP needs
it.
Add a new feature, FEAT_VENC_REQUIRES_TV_DAC_CLK, which tells if the
clock is needed on this platform, and use that feature in venc.c to
decide if the clock needs enabling.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Tue, 31 May 2011 13:55:47 +0000 (16:55 +0300)]
OMAP: DSS2: Reset LANEx_ULPS_SIG2 bits after use
LANEx_ULPS_SIG2 bits are left on after entering ULPS. This doesn't cause
any problems currently, as DSI HW is reset when it is enabled. However,
if the reset is not done, operation fails if the bits are still set.
So reset the bits after entering ULPS to ensure operation even without
HW reset.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Tue, 31 May 2011 13:39:01 +0000 (16:39 +0300)]
OMAP: DSS2: Taal: Make driver more fault tolerant
If ULPS exit fails, and the following reset fails also, Taal driver was
left in state where it thinks DSI is enabled while it really isn't,
leading to crash.
This patch checks the return value of taal_panel_reset, and if that
fails, ulps_enabled is left true, causing the driver to retry ulps exit
later.
Also the return value of taal_wake_up is checked at taal_disable, and if
wake up fails, we'll skip the power_off. This could leave the panel into
a not-quite-valid state, but there's nothing we can do about it in that
situation.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Sat, 30 Apr 2011 11:09:53 +0000 (14:09 +0300)]
OMAP: DSS2: remove update_mode from omapdss
Remove the whole update_mode stuff from omapdss driver. If automatic
update for manual update displays is needed, it's better implemented in
higher layers.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Sat, 30 Apr 2011 13:55:12 +0000 (16:55 +0300)]
OMAP: DSS2: OMAPFB: Implement auto-update mode
Implement auto-update mode for manual-update displays. omapfb driver
uses a delayed work to update the display with a constant rate.
The update mode can be changed via OMAPFB_SET_UPDATE_MODE ioctl, which
previously called omapdss but is now handled inside omapfb, and a new
sysfs file, "update_mode".
The update interval is by default 20 times per second, but can be
changed via "auto_update_freq" module parameter. There is also a new
module parameter "auto_update", which will make omapfb start manual
update displays in auto-update mode.
This auto-update mode can be used for testing if the userspace does not
support manual update displays properly. However, it is a very
inefficient solution, and should be considered more as a hack for
testing than something that could be used as a long term solution.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Sat, 30 Apr 2011 09:13:14 +0000 (12:13 +0300)]
OMAP: DSS2: OMAPFB: Add struct to store per-display data
Create a new struct omapfb_display_data to contain omapfb's private
per-display data. Move the bpp override there.
This struct will be used to hold auto/manual update state of a display
in the following patches.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Mon, 23 May 2011 13:36:09 +0000 (16:36 +0300)]
OMAP: DSS2: DSI: Change dummy macros to inline functions
Using empty macros for performance measurement functions when DSS DEBUG
is not enabled causes an unused variable warning.
Change the empty macros to empty inline functions to remove the
warning.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Tomi Valkeinen [Mon, 23 May 2011 08:51:18 +0000 (11:51 +0300)]
OMAP: DSS2: remove extra includes from include/video/omapdss.h
omapdss.h included platform_device.h and atomic.h, neither of which is
needed by omapdss.h. Remove those includes from omapdss.h, and fix the
affected .c files which did not include platform_device.h even though
they should.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Linus Torvalds [Tue, 21 Jun 2011 03:25:46 +0000 (20:25 -0700)]
Linux 3.0-rc4
Linus Torvalds [Tue, 21 Jun 2011 03:13:49 +0000 (20:13 -0700)]
vfs: i_state needs to be 'unsigned long' for now
Commit
13e12d14e2dc ("vfs: reorganize 'struct inode' layout a bit")
moved things around a bit changed i_state to be unsigned int instead of
unsigned long. That was to help structure layout for the 64-bit case,
and shrink 'struct inode' a bit (admittedly that only happened when
spinlock debugging was on and i_flags didn't pack with i_lock).
However, Meelis Roos reports that this results in unaligned exceptions
on sprc, and it turns out that the bit-locking primitives that we use
for the I_NEW bit want to use the bitops. Which want 'unsigned long',
not 'unsigned int'.
We really should fix the bit locking code to not have that kind of
requirement, but that's a much bigger change. So for now, revert that
field back to 'unsigned long' (but keep the other re-ordering changes
from the commit that caused this).
Andi points out that we have played games with this in 'struct page', so
it's solvable with other hacks too, but since right now the struct inode
size advantage only happens with some rare config options, it's not
worth fighting.
It _would_ be worth fixing the bitlocking code, though. Especially
since there is no type safety in the bitlocking code (this never caused
any warnings, and worked fine on x86-64, because the bitlocks take a
'void *' and x86-64 doesn't care that deeply about alignment). So it's
currently a very easy problem to trigger by mistake and never notice.
Reported-by: Meelis Roos <mroos@linux.ee>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus Torvalds [Tue, 21 Jun 2011 03:12:48 +0000 (20:12 -0700)]
Merge branch 'drm-fixes' of git://git./linux/kernel/git/airlied/drm-2.6
* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
drm/radeon/kms/r6xx+: voltage fixes
drm/nouveau: drop leftover debugging
drm/radeon: avoid warnings from r600/eg irq handlers on powered off card.
drm/radeon/kms: add missing param for dce3.2 DP transmitter setup
drm/radeon/kms/atom: fix duallink on some early DCE3.2 cards
drm/nouveau: fix assumption that semaphore dmaobj is valid in x-chan sync
drm/nv50/disp: fix gamma with page flipping overlay turned on
drm/nouveau/pm: Prevent overflow in nouveau_perf_init()
drm/nouveau: fix big-endian switch
Linus Torvalds [Tue, 21 Jun 2011 03:11:34 +0000 (20:11 -0700)]
Merge branch 'msm-fix' of git://codeaurora.org/quic/kernel/davidb/linux-msm
* 'msm-fix' of git://codeaurora.org/quic/kernel/davidb/linux-msm:
msm: timer: Fix DGT rate on 8960 and 8660
msm: timer: compensate for timer shift in msm_read_timer_count
msm: timer: Fix SMP build error
Linus Torvalds [Tue, 21 Jun 2011 03:10:52 +0000 (20:10 -0700)]
Merge branch 'for-2.6.40' of git://linux-nfs.org/~bfields/linux
* 'for-2.6.40' of git://linux-nfs.org/~bfields/linux:
nfsd4: fix break_lease flags on nfsd open
nfsd: link returns nfserr_delay when breaking lease
nfsd: v4 support requires CRYPTO
nfsd: fix dependency of nfsd on auth_rpcgss
Linus Torvalds [Tue, 21 Jun 2011 03:10:18 +0000 (20:10 -0700)]
Merge git://git./linux/kernel/git/davem/net-2.6
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (40 commits)
pxa168_eth: fix race in transmit path.
ipv4, ping: Remove duplicate icmp.h include
netxen: fix race in skb->len access
sgi-xp: fix a use after free
hp100: fix an skb->len race
netpoll: copy dev name of slaves to struct netpoll
ipv4: fix multicast losses
r8169: fix static initializers.
inet_diag: fix inet_diag_bc_audit()
gigaset: call module_put before restart of if_open()
farsync: add module_put to error path in fst_open()
net: rfs: enable RFS before first data packet is received
fs_enet: fix freescale FCC ethernet dp buffer alignment
netdev: bfin_mac: fix memory leak when freeing dma descriptors
vlan: don't call ndo_vlan_rx_register on hardware that doesn't have vlan support
caif: Bugfix - XOFF removed channel from caif-mux
tun: teach the tun/tap driver to support netpoll
dp83640: drop PHY status frames in the driver.
dp83640: fix phy status frame event parsing
phylib: Allow BCM63XX PHY to be selected only on BCM63XX.
...
Linus Torvalds [Tue, 21 Jun 2011 03:09:15 +0000 (20:09 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/vfs-2.6
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6:
devcgroup_inode_permission: take "is it a device node" checks to inlined wrapper
fix comment in generic_permission()
kill obsolete comment for follow_down()
proc_sys_permission() is OK in RCU mode
reiserfs_permission() doesn't need to bail out in RCU mode
proc_fd_permission() is doesn't need to bail out in RCU mode
nilfs2_permission() doesn't need to bail out in RCU mode
logfs doesn't need ->permission() at all
coda_ioctl_permission() is safe in RCU mode
cifs_permission() doesn't need to bail out in RCU mode
bad_inode_permission() is safe from RCU mode
ubifs: dereferencing an ERR_PTR in ubifs_mount()
Alex Deucher [Mon, 20 Jun 2011 17:00:31 +0000 (13:00 -0400)]
drm/radeon/kms/r6xx+: voltage fixes
0xff01 is not an actual voltage value, but a flag
for the driver. If the power state as that value,
skip setting the voltage.
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Stephen Boyd [Thu, 21 Apr 2011 23:09:11 +0000 (23:09 +0000)]
msm: timer: Fix DGT rate on 8960 and 8660
The DGT runs at 27 MHz divided by 4 on 8660 and 8960.
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: David Brown <davidb@codeaurora.org>
Richard Cochran [Sun, 19 Jun 2011 21:48:06 +0000 (21:48 +0000)]
pxa168_eth: fix race in transmit path.
Because the socket buffer is freed in the completion interrupt, it is not
safe to access it after submitting it to the hardware.
Cc: stable@kernel.org
Cc: Sachin Sanap <ssanap@marvell.com>
Cc: Zhangfei Gao <zgao6@marvell.com>
Cc: Philip Rakity <prakity@marvell.com>
Signed-off-by: Richard Cochran <richard.cochran@omicron.at>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Jesper Juhl [Sun, 19 Jun 2011 22:31:20 +0000 (22:31 +0000)]
ipv4, ping: Remove duplicate icmp.h include
Remove the duplicate inclusion of net/icmp.h from net/ipv4/ping.c
Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet [Sun, 19 Jun 2011 20:26:15 +0000 (20:26 +0000)]
netxen: fix race in skb->len access
As soon as skb is given to hardware, TX completion can free skb under
us.
Therefore, we should update dev stats before kicking the device.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Linus Torvalds [Mon, 20 Jun 2011 16:01:33 +0000 (09:01 -0700)]
Merge branch 'stable/bug.fixes' of git://git./linux/kernel/git/konrad/xen
* 'stable/bug.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
xen/setup: Fix for incorrect xen_extra_mem_start.
xen: When calling power_off, don't call the halt function.
xen: Fix compile warning when CONFIG_SMP is not defined.
xen: support CONFIG_MAXSMP
xen: partially revert "xen: set max_pfn_mapped to the last pfn mapped"
Linus Torvalds [Mon, 20 Jun 2011 15:59:46 +0000 (08:59 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
Input: sh_keysc - 8x8 MODE_6 fix
Input: omap-keypad - add missing input_sync()
Input: evdev - try to wake up readers only if we have full packet
Input: properly assign return value of clamp() macro.
Linus Torvalds [Mon, 20 Jun 2011 15:58:53 +0000 (08:58 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/mason/btrfs-unstable
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable:
Btrfs: avoid delayed metadata items during commits
btrfs: fix uninitialized return value
btrfs: fix wrong reservation when doing delayed inode operations
btrfs: Remove unused sysfs code
btrfs: fix dereference of ERR_PTR value
Btrfs: fix relocation races
Btrfs: set no_trans_join after trying to expand the transaction
Btrfs: protect the pending_snapshots list with trans_lock
Btrfs: fix path leakage on subvol deletion
Btrfs: drop the delalloc_bytes check in shrink_delalloc
Btrfs: check the return value from set_anon_super
Linus Torvalds [Mon, 20 Jun 2011 15:58:07 +0000 (08:58 -0700)]
Merge branch 'kvm-updates/3.0' of git://git./virt/kvm/kvm
* 'kvm-updates/3.0' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
KVM: Fix register corruption in pvclock_scale_delta
KVM: MMU: fix opposite condition in mapping_level_dirty_bitmap
KVM: VMX: do not overwrite uptodate vcpu->arch.cr3 on KVM_SET_SREGS
KVM: MMU: Fix build warnings in walk_addr_generic()
Al Viro [Sun, 19 Jun 2011 17:01:04 +0000 (13:01 -0400)]
devcgroup_inode_permission: take "is it a device node" checks to inlined wrapper
inode_permission() calls devcgroup_inode_permission() and almost all such
calls are _not_ for device nodes; let's at least keep the common path
straight...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Al Viro [Sun, 19 Jun 2011 05:56:53 +0000 (01:56 -0400)]
fix comment in generic_permission()
CAP_DAC_OVERRIDE is enough for MAY_EXEC on directory, even if
no exec bits are set.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Al Viro [Fri, 17 Jun 2011 23:20:48 +0000 (19:20 -0400)]
kill obsolete comment for follow_down()
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Al Viro [Sun, 19 Jun 2011 00:42:00 +0000 (20:42 -0400)]
proc_sys_permission() is OK in RCU mode
nothing blocking there, since all instances of sysctl
->permissions() method are non-blocking - both of them,
that is.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Al Viro [Sun, 19 Jun 2011 00:37:33 +0000 (20:37 -0400)]
reiserfs_permission() doesn't need to bail out in RCU mode
nothing blocking other than generic_permission() (and
check_acl callback does bail out in RCU mode).
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Al Viro [Sun, 19 Jun 2011 00:35:23 +0000 (20:35 -0400)]
proc_fd_permission() is doesn't need to bail out in RCU mode
nothing blocking except generic_permission()
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Al Viro [Sun, 19 Jun 2011 00:21:44 +0000 (20:21 -0400)]
nilfs2_permission() doesn't need to bail out in RCU mode
Nothing blocking except for generic_permission(). Which will DTRT.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Al Viro [Sun, 19 Jun 2011 00:17:22 +0000 (20:17 -0400)]
logfs doesn't need ->permission() at all
... and never did, what with its ->permission() being what we do by default
when ->permission is NULL...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Al Viro [Sun, 19 Jun 2011 00:11:43 +0000 (20:11 -0400)]
coda_ioctl_permission() is safe in RCU mode
return (mask & MAY_EXEC) ? -EACCES : 0; is non-blocking...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Al Viro [Sun, 19 Jun 2011 00:03:36 +0000 (20:03 -0400)]
cifs_permission() doesn't need to bail out in RCU mode
nothing potentially blocking except generic_permission(), which
will DTRT
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Al Viro [Sat, 18 Jun 2011 23:59:04 +0000 (19:59 -0400)]
bad_inode_permission() is safe from RCU mode
return -EIO; is *not* a blocking operation, thank you very much.
Nick, what the hell have you been smoking?
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Dan Carpenter [Mon, 20 Jun 2011 07:10:24 +0000 (10:10 +0300)]
ubifs: dereferencing an ERR_PTR in ubifs_mount()
d251ed271d5 "ubifs: fix sget races" left out the goto from this
error path so the static checkers complain that we're dereferencing
"sb" when it's an ERR_PTR.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Tue, 7 Jun 2011 15:50:23 +0000 (11:50 -0400)]
nfsd4: fix break_lease flags on nfsd open
Thanks to Casey Bodley for pointing out that on a read open we pass 0,
instead of O_RDONLY, to break_lease, with the result that a read open is
treated like a write open for the purposes of lease breaking!
Reported-by: Casey Bodley <cbodley@citi.umich.edu>
Cc: stable@kernel.org
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Dave Airlie [Mon, 20 Jun 2011 05:25:35 +0000 (15:25 +1000)]
drm/nouveau: drop leftover debugging
this printk isn't really useful, just drop it for now.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Mon, 20 Jun 2011 02:02:38 +0000 (12:02 +1000)]
Merge branch 'drm-nouveau-fixes' of git://anongit.freedesktop.org/git/nouveau/linux-2.6 into drm-fixes
* 'drm-nouveau-fixes' of git://anongit.freedesktop.org/git/nouveau/linux-2.6:
drm/nouveau: fix assumption that semaphore dmaobj is valid in x-chan sync
drm/nv50/disp: fix gamma with page flipping overlay turned on
drm/nouveau/pm: Prevent overflow in nouveau_perf_init()
drm/nouveau: fix big-endian switch
Dave Airlie [Sat, 18 Jun 2011 03:59:51 +0000 (03:59 +0000)]
drm/radeon: avoid warnings from r600/eg irq handlers on powered off card.
Since we were calling the wptr function before checking if the IH was
even enabled, or the GPU wasn't shutdown, we'd get spam in the logs when
the GPU readback 0xffffffff. This reorders things so we return early
in the no IH and GPU shutdown cases.
Reported-and-tested-by: ManDay on #radeon
Reviewed-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Alex Deucher [Fri, 17 Jun 2011 06:11:30 +0000 (06:11 +0000)]
drm/radeon/kms: add missing param for dce3.2 DP transmitter setup
This is used during phy init to set up the phy for DP. This may
fix DP problems on DCE3.2 cards.
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Alex Deucher [Fri, 17 Jun 2011 17:13:52 +0000 (13:13 -0400)]
drm/radeon/kms/atom: fix duallink on some early DCE3.2 cards
Certain revisions of the vbios on DCE3.2 cards have a bug
in the transmitter control table which prevents duallink from
being enabled properly on some cards. The action switch statement
jumps to the wrong offset for the OUTPUT_ENABLE action. The fix
is to use the ENABLE action rather than the OUTPUT_ENABLE action
on the affected cards. In fixed version of the vbios, both
actions jump to the same offset, so the change should be safe.
Reported-and-tested-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Cc: stable@kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
Eric Dumazet [Sun, 19 Jun 2011 12:52:36 +0000 (12:52 +0000)]
sgi-xp: fix a use after free
Its illegal to dereference skb after dev_kfree_skb(skb)
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Robin Holt <holt@sgi.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet [Sun, 19 Jun 2011 12:43:33 +0000 (12:43 +0000)]
hp100: fix an skb->len race
As soon as skb is given to hardware and spinlock released, TX completion
can free skb under us. Therefore, we should update netdev stats before
spinlock release.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
David S. Miller [Sun, 19 Jun 2011 23:26:46 +0000 (16:26 -0700)]
Merge branch 'davem.r8169' of git://git./linux/kernel/git/romieu/netdev-2.6
WANG Cong [Sun, 19 Jun 2011 23:13:01 +0000 (16:13 -0700)]
netpoll: copy dev name of slaves to struct netpoll
Otherwise we will not see the name of the slave dev in error
message:
[ 388.469446] (null): doesn't support polling, aborting.
Signed-off-by: WANG Cong <amwang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Zachary Amsden [Thu, 16 Jun 2011 03:50:04 +0000 (20:50 -0700)]
KVM: Fix register corruption in pvclock_scale_delta
The 128-bit multiply in pvclock.h was missing an output constraint for
EDX which caused a register corruption to appear. Thanks to Ulrich for
diagnosing the EDX corruption and Avi for providing this fix.
Signed-off-by: Zachary Amsden <zamsden@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Steve [Fri, 17 Jun 2011 02:25:39 +0000 (10:25 +0800)]
KVM: MMU: fix opposite condition in mapping_level_dirty_bitmap
The condition is opposite, it always maps huge page for the dirty tracked page
Reported-by: Steve <stefan.bosak@gmail.com>
Signed-off-by: Steve <stefan.bosak@gmail.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Marcelo Tosatti [Mon, 6 Jun 2011 17:27:47 +0000 (14:27 -0300)]
KVM: VMX: do not overwrite uptodate vcpu->arch.cr3 on KVM_SET_SREGS
Only decache guest CR3 value if vcpu->arch.cr3 is stale.
Fixes loadvm with live guest.
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Tested-by: Markus Schade <markus.schade@gmail.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Borislav Petkov [Mon, 30 May 2011 20:11:17 +0000 (22:11 +0200)]
KVM: MMU: Fix build warnings in walk_addr_generic()
On 3.0-rc1 I get
In file included from arch/x86/kvm/mmu.c:2856:
arch/x86/kvm/paging_tmpl.h: In function ‘paging32_walk_addr_generic’:
arch/x86/kvm/paging_tmpl.h:124: warning: ‘ptep_user’ may be used uninitialized in this function
In file included from arch/x86/kvm/mmu.c:2852:
arch/x86/kvm/paging_tmpl.h: In function ‘paging64_walk_addr_generic’:
arch/x86/kvm/paging_tmpl.h:124: warning: ‘ptep_user’ may be used uninitialized in this function
caused by
6e2ca7d1802bf8ed9908435e34daa116662e7790. According to Takuya
Yoshikawa, ptep_user won't be used uninitialized so shut up gcc.
Cc: Takuya Yoshikawa <yoshikawa.takuya@oss.ntt.co.jp>
Link: http://lkml.kernel.org/r/20110530094604.GC21833@liondog.tnic
Signed-off-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
Linus Torvalds [Sun, 19 Jun 2011 16:00:18 +0000 (09:00 -0700)]
Merge branches 'perf-urgent-for-linus', 'sched-urgent-for-linus', 'timers-urgent-for-linus' and 'x86-urgent-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip
* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
tools/perf: Fix static build of perf tool
tracing: Fix regression in printk_formats file
* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
generic-ipi: Fix kexec boot crash by initializing call_single_queue before enabling interrupts
* 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
clocksource: Make watchdog robust vs. interruption
timerfd: Fix wakeup of processes when timer is cancelled on clock change
* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
x86, MAINTAINERS: Add x86 MCE people
x86, efi: Do not reserve boot services regions within reserved areas
Linus Torvalds [Sun, 19 Jun 2011 15:56:56 +0000 (08:56 -0700)]
Merge branch 'core-urgent-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip
* 'core-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
rcu: Move RCU_BOOST #ifdefs to header file
rcu: use softirq instead of kthreads except when RCU_BOOST=y
rcu: Use softirq to address performance regression
rcu: Simplify curing of load woes
Linus Torvalds [Sun, 19 Jun 2011 03:33:31 +0000 (20:33 -0700)]
Merge branch 'hwmon-for-linus' of git://git./linux/kernel/git/groeck/staging
* 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/staging:
hwmon: (s3c) Initialize sysfs attributes
hwmon: (ibmpex) Initialize sysfs attributes
hwmon: (ibmaem) Initialize sysfs attributes
hwmon: (asus_atk0110) Consolidate sysfs attribute initialization
hwmon: (coretemp) Drop unused struct members
Ingo Molnar [Sat, 18 Jun 2011 20:51:13 +0000 (22:51 +0200)]
x86, MAINTAINERS: Add x86 MCE people
Announce the new x86 MCE infrastructure maintainers.
Acked-by: Borislav Petkov <bp@amd64.org>
Acked-by: Tony Luck <tony.luck@intel.com>
Acked-by: H. Peter Anvin <hpa@linux.intel.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/n/tip-8hs7yob6wib4vblmrmbpbav4@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Maarten Lankhorst [Tue, 14 Jun 2011 17:53:09 +0000 (19:53 +0200)]
x86, efi: Do not reserve boot services regions within reserved areas
Commit
916f676f8dc started reserving boot service code since some systems
require you to keep that code around until SetVirtualAddressMap is called.
However, in some cases those areas will overlap with reserved regions.
The proper medium-term fix is to fix the bootloader to prevent the
conflicts from occurring by moving the kernel to a better position,
but the kernel should check for this possibility, and only reserve regions
which can be reserved.
Signed-off-by: Maarten Lankhorst <m.b.lankhorst@gmail.com>
Link: http://lkml.kernel.org/r/4DF7A005.1050407@gmail.com
Acked-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Eric Dumazet [Sat, 18 Jun 2011 18:59:18 +0000 (11:59 -0700)]
ipv4: fix multicast losses
Knut Tidemann found that first packet of a multicast flow was not
correctly received, and bisected the regression to commit
b23dd4fe42b4
(Make output route lookup return rtable directly.)
Special thanks to Knut, who provided a very nice bug report, including
sample programs to demonstrate the bug.
Reported-and-bisectedby: Knut Tidemann <knut.andre.tidemann@jotron.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Linus Torvalds [Wed, 8 Jun 2011 17:30:03 +0000 (10:30 -0700)]
isofs: fix bh leak in isofs_fill_super() error case
In isofs_fill_super(), when an iso_primary_descriptor is found, it is
kept in pri_bh. The error cases don't properly release it. Fix it.
Reported-and-tested-by: 김원석 <stanley.will.kim@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Magnus Damm [Sat, 18 Jun 2011 09:51:52 +0000 (02:51 -0700)]
Input: sh_keysc - 8x8 MODE_6 fix
According to the data sheet for G4, AP4 and AG5 KEYSC MODE_6 is 8x8 keys.
Bump up MAXKEYS to 64 too.
Signed-off-by: Magnus Damm <damm@opensource.se>
Reviewed-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Janusz Krzysztofik [Thu, 16 Jun 2011 04:13:55 +0000 (21:13 -0700)]
Input: omap-keypad - add missing input_sync()
Otherwise the updated evdev driver (commit
cdda911c34006f1089f3c87b1a1f,
"Input: evdev - only signal polls on full packets") no longer works on
top of omap-keypad.
Tested on Amstrad Delta.
Signed-off-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
Reviewed-by: Henrik Rydberg <rydberg@euromail.se>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Dmitry Torokhov [Sat, 18 Jun 2011 09:50:11 +0000 (02:50 -0700)]
Input: evdev - try to wake up readers only if we have full packet
We should only wake waiters on the event device when we actually post
an EV_SYN/SYN_REPORT to the queue. Otherwise we end up making waiting
threads runnable only to go right back to sleep because the device
still isn't readable.
Reported-by: Jeffrey Brown <jeffbrown@android.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Guenter Roeck [Tue, 24 May 2011 19:34:55 +0000 (12:34 -0700)]
hwmon: (s3c) Initialize sysfs attributes
Initialize dynamically allocated sysfs attributes before device_create_file()
call to suppress lockdep_init_map() warning if lockdep debugging is enabled.
Signed-off-by: Guenter Roeck <guenter.roeck@ericsson.com>
Acked-by: Jean Delvare <khali@linux-fr.org>
Cc: stable@kernel.org # 2.6.34+
Guenter Roeck [Tue, 24 May 2011 19:34:12 +0000 (12:34 -0700)]
hwmon: (ibmpex) Initialize sysfs attributes
Initialize dynamically allocated sysfs attributes before device_create_file()
call to suppress lockdep_init_map() warning if lockdep debugging is enabled.
Signed-off-by: Guenter Roeck <guenter.roeck@ericsson.com>
Acked-by: Jean Delvare <khali@linux-fr.org>
Cc: stable@kernel.org # 2.6.34+
Guenter Roeck [Tue, 24 May 2011 19:33:26 +0000 (12:33 -0700)]
hwmon: (ibmaem) Initialize sysfs attributes
Initialize dynamically allocated sysfs attributes before device_create_file()
call to suppress lockdep_init_map() warning if lockdep debugging is enabled.
Signed-off-by: Guenter Roeck <guenter.roeck@ericsson.com>
Acked-by: Jean Delvare <khali@linux-fr.org>
Cc: stable@kernel.org # 2.6.34+
Guenter Roeck [Tue, 24 May 2011 19:19:05 +0000 (12:19 -0700)]
hwmon: (asus_atk0110) Consolidate sysfs attribute initialization
Call sysfs_attr_init() from atk_init_attribute() to handle sysfs attribute
initialization in a single function.
Signed-off-by: Guenter Roeck <guenter.roeck@ericsson.com>
Cc: Luca Tettamanti <kronos.it@gmail.com>
Acked-by: Jean Delvare <khali@linux-fr.org>
Jean Delvare [Tue, 7 Jun 2011 19:14:26 +0000 (15:14 -0400)]
hwmon: (coretemp) Drop unused struct members
pdev_entry.cpu and pdev_entry.cpu_core_id aren't used anywhere in the
driver code so we can drop these struct members.
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: Guenter Roeck <guenter.roeck@ericsson.com>
Cc: Durgadoss R <durgadoss.r@intel.com>
Signed-off-by: Guenter Roeck <guenter.roeck@ericsson.com>
Ben Skeggs [Fri, 17 Jun 2011 13:41:54 +0000 (23:41 +1000)]
drm/nouveau: fix assumption that semaphore dmaobj is valid in x-chan sync
The DDX modifies DMA_SEMAPHORE on nv50 in order to implement sync-to-vblank,
things will go very wrong for cross-channel sync after this.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Thu, 16 Jun 2011 04:40:27 +0000 (14:40 +1000)]
drm/nv50/disp: fix gamma with page flipping overlay turned on
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Emil Velikov [Sat, 11 Jun 2011 12:30:32 +0000 (13:30 +0100)]
drm/nouveau/pm: Prevent overflow in nouveau_perf_init()
While parsing the perf table, there is no check if
the num of entries read from the vbios is less than
the currently allocated number.
In case of a buggy vbios this will cause overwriting
of kernel memory, causing aditional problems.
Add a simple check in order to prevent the case
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Ben Skeggs [Tue, 14 Jun 2011 00:16:17 +0000 (10:16 +1000)]
drm/nouveau: fix big-endian switch
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Linus Torvalds [Sat, 18 Jun 2011 04:15:11 +0000 (21:15 -0700)]
Merge branch 'drm-fixes' of git://git./linux/kernel/git/airlied/drm-2.6
* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
drm/i915: Fix gen6 (SNB) missed BLT ring interrupts.
Linus Torvalds [Sat, 18 Jun 2011 04:13:43 +0000 (21:13 -0700)]
Merge branch 'fixes' of git://git./linux/kernel/git/davej/cpufreq
* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq:
[CPUFREQ] powernow-k8: Don't try to transition if the pstate is incorrect
[CPUFREQ] powernow-k8: Don't notify of successful transition if we failed (vid case).
[CPUFREQ] Don't set stat->last_index to -1 if the pol->cur has incorrect value.
Linus Torvalds [Sat, 18 Jun 2011 04:01:21 +0000 (21:01 -0700)]
Merge branch 'anon_vma-locking'
* anon_vma-locking:
mm: avoid anon_vma_chain allocation under anon_vma lock
mm: avoid repeated anon_vma lock/unlock sequences in unlink_anon_vmas()
mm: avoid repeated anon_vma lock/unlock sequences in anon_vma_clone()
Linus Torvalds [Sat, 18 Jun 2011 02:05:36 +0000 (19:05 -0700)]
mm: avoid anon_vma_chain allocation under anon_vma lock
Hugh Dickins points out that lockdep (correctly) spots a potential
deadlock on the anon_vma lock, because we now do a GFP_KERNEL allocation
of anon_vma_chain while doing anon_vma_clone(). The problem is that
page reclaim will want to take the anon_vma lock of any anonymous pages
that it will try to reclaim.
So re-organize the code in anon_vma_clone() slightly: first do just a
GFP_NOWAIT allocation, which will usually work fine. But if that fails,
let's just drop the lock and re-do the allocation, now with GFP_KERNEL.
End result: not only do we avoid the locking problem, this also ends up
getting better concurrency in case the allocation does need to block.
Tim Chen reports that with all these anon_vma locking tweaks, we're now
almost back up to the spinlock performance.
Reported-and-tested-by: Hugh Dickins <hughd@google.com>
Tested-by: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Peter Zijlstra [Fri, 17 Jun 2011 11:54:23 +0000 (13:54 +0200)]
mm: avoid repeated anon_vma lock/unlock sequences in unlink_anon_vmas()
This matches the anon_vma_clone() case, and uses the same lock helper
functions. Because of the need to potentially release the anon_vma's,
it's a bit more complex, though.
We traverse the 'vma->anon_vma_chain' in two phases: the first loop gets
the anon_vma lock (with the helper function that only takes the lock
once for the whole loop), and removes any entries that don't need any
more processing.
The second phase just traverses the remaining list entries (without
holding the anon_vma lock), and does any actual freeing of the
anon_vma's that is required.
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Tested-by: Hugh Dickins <hughd@google.com>
Tested-by: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus Torvalds [Fri, 17 Jun 2011 03:44:51 +0000 (20:44 -0700)]
mm: avoid repeated anon_vma lock/unlock sequences in anon_vma_clone()
In anon_vma_clone() we traverse the vma->anon_vma_chain of the source
vma, locking the anon_vma for each entry.
But they are all going to have the same root entry, which means that
we're locking and unlocking the same lock over and over again. Which is
expensive in locked operations, but can get _really_ expensive when that
root entry sees any kind of lock contention.
In fact, Tim Chen reports a big performance regression due to this: when
we switched to use a mutex instead of a spinlock, the contention case
gets much worse.
So to alleviate this all, this commit creates a small helper function
(lock_anon_vma_root()) that can be used to take the lock just once
rather than taking and releasing it over and over again.
We still have the same "take the lock and release" it behavior in the
exit path (in unlink_anon_vmas()), but that one is a bit harder to fix
since we're actually freeing the anon_vma entries as we go, and that
will touch the lock too.
Reported-and-tested-by: Tim Chen <tim.c.chen@linux.intel.com>
Tested-by: Hugh Dickins <hughd@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Daniel J Blueman [Fri, 17 Jun 2011 18:32:19 +0000 (11:32 -0700)]
drm/i915: Fix gen6 (SNB) missed BLT ring interrupts.
The failure appeared in dmesg as:
[drm:i915_hangcheck_ring_idle] *ERROR* Hangcheck timer elapsed... blt
ring idle [waiting on
35064155, at
35064155], missed IRQ?
This works around that problem on by making the blitter command
streamer write interrupt state to the Hardware Status Page when a
MI_USER_INTERRUPT command is decoded, which appears to force the seqno
out to memory before the interrupt happens.
v1->v2: Moved to prior interrupt handler installation and RMW flags as
per feedback.
v2->v3: Removed RMW of flags (by anholt)
Cc: stable@kernel.org
Signed-off-by: Daniel J Blueman <daniel.blueman@gmail.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Tested-by: Chris Wilson <chris@chris-wilson.co.uk> [v1]
Tested-by: Eric Anholt <eric@anholt.net> [v1,v3]
(incidence of the bug with a testcase went from avg 2/1000 to
0/12651 in the latest test run (plus more for v1))
Tested-by: Kenneth Graunke <kenneth@whitecape.org> [v1]
Tested-by: Robert Hooker <robert.hooker@canonical.com> [v1]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=33394
Signed-off-by: Dave Airlie <airlied@redhat.com>
Jeff Ohlstein [Fri, 17 Jun 2011 20:55:38 +0000 (13:55 -0700)]
msm: timer: compensate for timer shift in msm_read_timer_count
Some msm targets have timers whose lower bits are unreliable. So, we
present our timers as lower frequency than they actually are, and ignore
the bottom 5 bits on such targets. This compensation was erroneously
removed from the msm_read_timer_count function, so restore it.
This was broken by
94790ec25 "msm: timer: SMP timer support for msm".
Signed-off-by: Jeff Ohlstein <johlstei@codeaurora.org>
Francois Romieu [Fri, 17 Jun 2011 20:58:54 +0000 (22:58 +0200)]
r8169: fix static initializers.
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>