Keith Packard [Wed, 3 Aug 2011 17:52:24 +0000 (10:52 -0700)]
drm/i915: Try enabling RC6 by default (again)
Jesse Barnes and I found a couple of issues where incorrect mode
setting would cause problems with RC6 enabled. We're hopeful that
fixing those will resolve the outstanding issues with a few machines
that had trouble before 3.0 with rc6.
Cc: Pekka Enberg <penberg@kernel.org> Cc: Francesco Allertsen <fallertsen@gmail.com> Cc: Ted Phelps <phelps@gnusto.com> Cc: Gu Rui <chaos.proton@gmail.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=38567
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=38332 Signed-off-by: Keith Packard <keithp@keithp.com>
Clearing the dpcd data means that if the fetch fails, any previous
data will be lost. On eDP, this is no fun as we only fetch dpcd at
init time, so the memset will destroy that the next time through.
Jesse Barnes [Mon, 1 Aug 2011 22:02:20 +0000 (15:02 -0700)]
drm/i915/dp: wait for previous AUX channel activity to clear
Before initiating a new read or write on the DP AUX channel, wait for
any outstanding activity to complete. This may happen during normal
retry behavior. If the wait fails (i.e. after 1ms the AUX channel is
still busy) dump a backtrace to make the caller easier to spot.
v2: use msleep instead, and timeout after 3ms (only ever saw 1 retry
with msleep in testing)
v3: fix backtrace check to trigger if the 3ms wait times out
drm/i915: don't use uninitialized EDID bpc values when picking pipe bpp
The EDID parser will zero out the bpc value, and the driver needs to handle
that case. In our picker, we'll just ignore 0 values as far as bpp
picking goes.
Adam Jackson [Tue, 26 Jul 2011 20:53:06 +0000 (16:53 -0400)]
drm/i915/pch: Save/restore PCH_PORT_HOTPLUG across suspend
At least on a Lenovo X220 the HPD bits of this are enabled at boot but
cleared after resume, which means plug interrupts stop working.
This also happens to fix DP displays re-lighting on resume. I'm quite
certain that's an accident: the first DP link train inevitably fails on
that machine, and it's only serendipity that we're getting multiple plug
interrupts and the second train works. But I shall take my victories
where I get them.
Signed-off-by: Adam Jackson <ajax@redhat.com> Tested-by: Keith Packard <keithp@keithp.com> Reviewed-by: Keith Packard <keithp@keithp.com> Signed-off-by: Keith Packard <keithp@keithp.com>
drm/i915: apply phase pointer override on SNB+ too
These bits moved around on SNB and above.
v2: again with the git send-email fail
v3: add macros for getting per-pipe override & enable bits
v4: enable phase sync pointer on SNB and IVB configs as well
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Keith Packard <keithp@keithp.com>
drm/i915: Add quirk to disable SSC on Sony Vaio Y2
Using the new quirk added to support disabling SSC on Lenovo U160
(#36656, commit 435793dfb8aec7b2e19f72d5bce8a22fd0b57839), also register
the Vaio as a special case and disable SSC for it.
This patch fixes #34437 on fdo bugzilla:
https://bugs.freedesktop.org/show_bug.cgi?id=34437
Signed-off-by: Michel Alexandre Salim <salimma@fedoraproject.org> Signed-off-by: Keith Packard <keithp@keithp.com>
drm/i915: provide more error output when mode sets fail
If a mode set fails we may get a message from drm_crtc_helper if we're lucky,
but it won't tell us anything about *why* we failed to set a mode. So
add a few DRM_ERRORs for the cases that shouldn't happen so we can debug
things more easily.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Keith Packard <keithp@keithp.com>
Mainly for use in debugging and benchmarking, this file allows the user
to control the max frequency used by the GPU. Frequency may still vary
based on workload (if the frequency is set to higher than the minimum)
but won't go over the newly set value.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Keith Packard <keithp@keithp.com>
Keith Packard [Fri, 29 Jul 2011 21:45:21 +0000 (14:45 -0700)]
drm/i915: Ignore GPU wedged errors while pinning scanout buffers
Failing to pin a scanout buffer will most likely lead to a black
screen, so if the GPU is wedged, then just let the pin happen and hope
that things work out OK.
Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
CB tuning is needed to handle potential process variations that might
cause clock jitter for certain PLL settings. However, we were setting
it incorrectly since we were using the wrong M value as a check (M1 when
we needed to use the whole M value). Fix it up, making my HDMI
attached display a little prettier (used to have occasional dots crawl
across the display).
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Keith Packard <keithp@keithp.com>
drm/i915: flush plane control changes on ILK+ as well
After writing to the plane control reg we need to write to the surface
reg to trigger the double buffered register latch. On previous
chipsets, writing to DSPADDR was enough, but on ILK+ DSPSURF is the reg
that triggers the double buffer latch.
v2: write DSPADDR too to cover pre-965 chipsets
v3: use flush_display_plane instead, that's what it's for
v4: send the right patch
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org> Tested-by: Keith Packard <keithp@keithp.com> Reviewed-by: Keith Packard <keithp@keithp.com> Signed-off-by: Keith Packard <keithp@keithp.com>
drm/i915: apply timing generator bug workaround on CPT and PPT
On CougarPoint and PantherPoint PCH chips, the timing generator may fail
to start after DP training completes. This is due to a bug in the
FDI autotraining detect logic (which will stall the timing generator and
re-enable it once training completes), so disable it to avoid silent DP
mode setting failures.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Keith Packard <keithp@keithp.com>
We actually *do* need to track DPMS state so that on hotplug, we don't
retrain the link until DPMS is disabled.
However, that code had avery small bug -- it wouldn't set the
dpms_mode at mode set time, and so link retraining would not actually
occur on monitor hotplug until the monitor had gone through a DPMS
off/DPMS on cycle.
Signed-off-by: Keith Packard <keithp@keithp.com> Tested-by: Andrew Lutomirski <luto@mit.edu>
Keith Packard [Tue, 26 Jul 2011 05:12:43 +0000 (22:12 -0700)]
drm/i915: DP_PIPE_ENABLED must check transcoder on CPT
Display port pipe selection on CPT is not done with a bit in the
output register, rather it is controlled by a couple of bits in the
separate transcoder register which indicate which display port output
is connected to the transcoder.
This patch replaces the simplistic macro DP_PIPE_ENABLED with the
rather more complicated function dp_pipe_enabled which checks the
output register to see if that is enabled, and then goes on to either
check the output register pipe selection bit (on non-CPT) or the
transcoder DP selection bits (on CPT).
Before this patch, any time the mode of pipe A was changed, any
display port outputs on pipe B would get disabled as
intel_disable_pch_ports would ensure that the mode setting operation
could occur on pipe A without interference from other outputs
connected to that pch port
Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org> Reviewed-by: Adam Jackson <ajax@redhat.com>
Keith Packard [Tue, 26 Jul 2011 02:50:10 +0000 (19:50 -0700)]
drm/i915: Use dp_detect_common in hotplug helper function
This uses the common dpcd reading routine, i915_dp_detect_common,
instead of open-coding a call to intel_dp_aux_native_read. Besides
reducing duplicated code, this also gains the read retries which
may be necessary when a cable is first plugged back in and the link
needs to be retrained.
Signed-off-by: Keith Packard <keithp@keithp.com> Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org> Reviewed-by: Adam Jackson <ajax@redhat.com>
Keith Packard [Thu, 28 Jul 2011 22:31:19 +0000 (15:31 -0700)]
drm/i915: Fixup for 'Hold mode_config->mutex during hotplug'
drm_helper_hpd_irq_event queues another work proc to go and deliver
the user-space event, and that function also wants to hold the config
mutex, so we shouldn't hold the mutex across the
drm_helper_hpd_irq_event call.
Adam Jackson [Thu, 21 Jul 2011 21:48:38 +0000 (17:48 -0400)]
drm/i915/dp: Don't turn CPT DP ports on too early
The docs say the port has to come on in training pattern 1; at this
point, though, ->DP is in normal mode. The intent here is to wait
until the port is in fact sending data, but that doesn't happen since
we've broken the sequence the hardware expects, and the vblank wait will
time out and kvetch in the log.
Signed-off-by: Adam Jackson <ajax@redhat.com> Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Keith Packard <keithp@keithp.com>
Adam Jackson [Thu, 21 Jul 2011 21:48:37 +0000 (17:48 -0400)]
drm/i915/dp: Explicitly disable symbol scrambling while training
The DP spec says training patterns 1 and 2 are to be sent non-scrambled,
and the GPU docs claim that happens (or at least, there's no explicit
scrambling control). But the sink may be confused if we don't
explicitly tell it what we're doing, so play it safe.
Signed-off-by: Adam Jackson <ajax@redhat.com> Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Keith Packard <keithp@keithp.com>
This is broken. The total width of scanout would then be 1593 + 4 + 4,
or 1601, which is wider than the panel itself. The hardware very
dutifully implements this, and you end up with a black 45° diagonal from
the top-left corner to the bottom edge of the screen. It's a cool
effect and all, but not what you wanted. Similar things happen for the
letterbox case.
The problem is that you have an integer number of pixels, which means
it's usually impossible to upscale equally on both axes. 1360/768 is
1.7708, 1600/900 is 1.7777. Since we're constrained on the one axis,
the other one wants to come out as an even number of pixels (the panel
is almost certainly even on both axes, and the x/y offsets will be
applied on both sides). In the math above, if 'width' comes out even,
rounding down is correct; if it's odd, you'd rather round up. So just
increment width/height in those cases.
Tested on a Lenovo T500 (Ironlake).
Signed-off-by: Adam Jackson <ajax@redhat.com> Tested-By: Daniel Manrique <daniel.manrique@canonical.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=38851 Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: stable@kernel.org Signed-off-by: Keith Packard <keithp@keithp.com>
Keith Packard [Mon, 25 Jul 2011 17:04:56 +0000 (10:04 -0700)]
drm/i915: Hold mode_config->mutex during hotplug processing
Hotplug detection is a mode setting operation and must hold the
struct_mutex or risk colliding with other mode setting operations.
In particular, the display port hotplug function attempts to re-train
the link if the monitor is supposed to be running when plugged back
in. If that happens while mode setting is underway, the link will get
scrambled, leaving it in an inconsistent state.
This is a special case -- usually the driver mode setting entry points
are covered by the upper level DRM code, but in this case the function
is invoked as a work function not under the control of DRM.
Signed-off-by: Keith Packard <keithp@keithp.com> Cc: stable@kernel.org Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
It's not clear what a sink would do if you wrote zero to this register -
which I guess would mean "I don't support any channel encodings, good
luck" - but let's not find out.
Signed-off-by: Adam Jackson <ajax@redhat.com> Reviewed-by: Keith Packard <keithp@keithp.com> Signed-off-by: Keith Packard <keithp@keithp.com>
Keith Packard [Fri, 22 Jul 2011 17:44:39 +0000 (10:44 -0700)]
drm/i915: Initialize RCS ring status page address in intel_render_ring_init_dri
Physically-addressed hardware status pages are initialized early in
the driver load process by i915_init_phys_hws. For UMS environments,
the ring structure is not initialized until the X server starts. At
that point, the entire ring structure is re-initialized with all new
values. Any values set in the ring structure (including
ring->status_page.page_addr) will be lost when the ring is
re-initialized.
This patch moves the initialization of the status_page.page_addr value
to intel_render_ring_init_dri.
Signed-off-by: Keith Packard <keithp@keithp.com> Cc: stable@kernel.org
drm/i915: Fix typo in DRM_I915_OVERLAY_PUT_IMAGE ioctl define
Because of a typo, calling ioctl with DRM_IOCTL_I915_OVERLAY_PUT_IMAGE
is broken if the macro is used directly. When using libdrm the bug is
not hit, since libdrm handles the ioctl encoding internally.
The typo also leads to the .cmd and .cmd_drv fields of the drm_ioctl
structure for DRM_I915_OVERLAY_PUT_IMAGE having inconsistent content.
Signed-off-by: Ole Henrik Jahren <olehenja@alumni.ntnu.no> Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: stable@kernel.org Signed-off-by: Keith Packard <keithp@keithp.com>
Keith Packard [Tue, 19 Jul 2011 23:21:40 +0000 (16:21 -0700)]
drm/i915: Skip GPU wait for scanout pin while wedged
Failing to pin a scanout buffer will most likely lead to a black
screen, so if the GPU is wedged, then just let the pin happen and hope
that things work out OK.
v2: Just ignore any error from i915_gem_object_wait_rendering, as
suggested by Chris Wilson
drm/i915: provide more error output when mode sets fail
If a mode set fails we may get a message from drm_crtc_helper if we're lucky,
but it won't tell us anything about *why* we failed to set a mode. So
add a few DRM_ERRORs for the cases that shouldn't happen so we can debug
things more easily.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Keith Packard <keithp@keithp.com>
Merge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip
* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
x86: Make Dell Latitude E6420 use reboot=pci
x86: Make Dell Latitude E5420 use reboot=pci
H. Peter Anvin [Thu, 21 Jul 2011 18:22:21 +0000 (11:22 -0700)]
x86: Make Dell Latitude E6420 use reboot=pci
Yet another variant of the Dell Latitude series which requires
reboot=pci.
From the E5420 bug report by Daniel J Blueman:
> The E6420 is affected also (same platform, different casing and
> features), which provides an external confirmation of the issue; I can
> submit a patch for that later or include it if you prefer:
> http://linux.koolsolutions.com/2009/08/04/howto-fix-linux-hangfreeze-during-reboots-and-restarts/
Reported-by: Daniel J Blueman <daniel.blueman@gmail.com> Signed-off-by: H. Peter Anvin <hpa@zytor.com> Cc: <stable@kernel.org>
Daniel J Blueman [Fri, 13 May 2011 01:04:59 +0000 (09:04 +0800)]
x86: Make Dell Latitude E5420 use reboot=pci
Rebooting on the Dell E5420 often hangs with the keyboard or ACPI
methods, but is reliable via the PCI method.
[ hpa: this was deferred because we believed for a long time that the
recent reshuffling of the boot priorities in commit 660e34cebf0a11d54f2d5dd8838607452355f321 fixed this platform.
Unfortunately that turned out to be incorrect. ]
vfs: drop conditional inode prefetch in __do_lookup_rcu
It seems to hurt performance in real life. Yes, the inode will be used
later, but the conditional doesn't seem to predict all that well
(negative dentries are not uncommon) and it looks like the cost of
prefetching is simply higher than depending on the cache doing the right
thing.
The compiler, at least for ix86 and m68k, validly warns that the
comparison:
next <= (loff_t)-1
is always true (and it's always true also for x86-64 and probably all
other arches - as long as pgoff_t isn't wider than loff_t). The
intention appears to be to avoid wrapping of "next", so rather than
eliminating the pointless comparison, fix the loop to indeed get exited
when "next" would otherwise wrap.
On m68k the following warning is observed:
fs/fscache/page.c: In function '__fscache_uncache_all_inode_pages':
fs/fscache/page.c:979: warning: comparison is always false due to limited range of data type
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org> Reported-by: Jan Beulich <jbeulich@novell.com> Signed-off-by: Jan Beulich <jbeulich@novell.com> Signed-off-by: David Howells <dhowells@redhat.com> Cc: Suresh Jayaraman <sjayaraman@suse.de> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: stable@kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Merge branch 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip
* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
sched: Avoid creating superfluous NUMA domains on non-NUMA systems
sched: Allow for overlapping sched_domain spans
sched: Break out cpu_power from the sched_group structure
Merge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip
* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
x86. reboot: Make Dell Latitude E6320 use reboot=pci
x86, doc only: Correct real-mode kernel header offset for init_size
x86: Disable AMD_NUMA for 32bit for now
Paul E. McKenney [Tue, 19 Jul 2011 10:25:36 +0000 (03:25 -0700)]
signal: align __lock_task_sighand() irq disabling and RCU
The __lock_task_sighand() function calls rcu_read_lock() with interrupts
and preemption enabled, but later calls rcu_read_unlock() with interrupts
disabled. It is therefore possible that this RCU read-side critical
section will be preempted and later RCU priority boosted, which means that
rcu_read_unlock() will call rt_mutex_unlock() in order to deboost itself, but
with interrupts disabled. This results in lockdep splats, so this commit
nests the RCU read-side critical section within the interrupt-disabled
region of code. This prevents the RCU read-side critical section from
being preempted, and thus prevents the attempt to deboost with interrupts
disabled.
It is quite possible that a better long-term fix is to make rt_mutex_unlock()
disable irqs when acquiring the rt_mutex structure's ->wait_lock.
Signed-off-by: Paul E. McKenney <paul.mckenney@linaro.org> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Peter Zijlstra [Tue, 19 Jul 2011 22:32:00 +0000 (15:32 -0700)]
softirq,rcu: Inform RCU of irq_exit() activity
The rcu_read_unlock_special() function relies on in_irq() to exclude
scheduler activity from interrupt level. This fails because exit_irq()
can invoke the scheduler after clearing the preempt_count() bits that
in_irq() uses to determine that it is at interrupt level. This situation
can result in failures as follows:
$task IRQ SoftIRQ
rcu_read_lock()
/* do stuff */
<preempt> |= UNLOCK_BLOCKED
rcu_read_unlock()
--t->rcu_read_lock_nesting
irq_enter();
/* do stuff, don't use RCU */
irq_exit();
sub_preempt_count(IRQ_EXIT_OFFSET);
invoke_softirq()
Ed can simply trigger this 'easy' because invoke_softirq() immediately
does a ttwu() of ksoftirqd/# instead of doing the in-place softirq stuff
first, but even without that the above happens.
Cure this by also excluding softirqs from the
rcu_read_unlock_special() handler and ensuring the force_irqthreads
ksoftirqd/# wakeup is done from full softirq context.
[ Alternatively, delaying the ->rcu_read_lock_nesting decrement
until after the special handling would make the thing more robust
in the face of interrupts as well. And there is a separate patch
for that. ]
Cc: Thomas Gleixner <tglx@linutronix.de> Reported-and-tested-by: Ed Tomlinson <edt@aei.ca> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Peter Zijlstra [Tue, 19 Jul 2011 22:07:25 +0000 (15:07 -0700)]
sched: Add irq_{enter,exit}() to scheduler_ipi()
Ensure scheduler_ipi() calls irq_{enter,exit} when it does some actual
work. Traditionally we never did any actual work from the resched IPI
and all magic happened in the return from interrupt path.
Now that we do do some work, we need to ensure irq_{enter,exit} are
called so that we don't confuse things.
This affects things like timekeeping, NO_HZ and RCU, basically
everything with a hook in irq_enter/exit.
Explicit examples of things going wrong are:
sched_clock_cpu() -- has a callback when leaving NO_HZ state to take
a new reading from GTOD and TSC. Without this
callback, time is stuck in the past.
RCU -- needs in_irq() to work in order to avoid some nasty deadlocks
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Paul E. McKenney [Mon, 18 Jul 2011 04:14:35 +0000 (21:14 -0700)]
rcu: protect __rcu_read_unlock() against scheduler-using irq handlers
The addition of RCU read-side critical sections within runqueue and
priority-inheritance lock critical sections introduced some deadlock
cycles, for example, involving interrupts from __rcu_read_unlock()
where the interrupt handlers call wake_up(). This situation can cause
the instance of __rcu_read_unlock() invoked from interrupt to do some
of the processing that would otherwise have been carried out by the
task-level instance of __rcu_read_unlock(). When the interrupt-level
instance of __rcu_read_unlock() is called with a scheduler lock held
from interrupt-entry/exit situations where in_irq() returns false,
deadlock can result.
This commit resolves these deadlocks by using negative values of
the per-task ->rcu_read_lock_nesting counter to indicate that an
instance of __rcu_read_unlock() is in flight, which in turn prevents
instances from interrupt handlers from doing any special processing.
This patch is inspired by Steven Rostedt's earlier patch that similarly
made __rcu_read_unlock() guard against interrupt-mediated recursion
(see https://lkml.org/lkml/2011/7/15/326), but this commit refines
Steven's approach to avoid the need for preemption disabling on the
__rcu_read_unlock() fastpath and to also avoid the need for manipulating
a separate per-CPU variable.
This patch avoids need for preempt_disable() by instead using negative
values of the per-task ->rcu_read_lock_nesting counter. Note that nested
rcu_read_lock()/rcu_read_unlock() pairs are still permitted, but they will
never see ->rcu_read_lock_nesting go to zero, and will therefore never
invoke rcu_read_unlock_special(), thus preventing them from seeing the
RCU_READ_UNLOCK_BLOCKED bit should it be set in ->rcu_read_unlock_special.
This patch also adds a check for ->rcu_read_unlock_special being negative
in rcu_check_callbacks(), thus preventing the RCU_READ_UNLOCK_NEED_QS
bit from being set should a scheduling-clock interrupt occur while
__rcu_read_unlock() is exiting from an outermost RCU read-side critical
section.
Of course, __rcu_read_unlock() can be preempted during the time that
->rcu_read_lock_nesting is negative. This could result in the setting
of the RCU_READ_UNLOCK_BLOCKED bit after __rcu_read_unlock() checks it,
and would also result it this task being queued on the corresponding
rcu_node structure's blkd_tasks list. Therefore, some later RCU read-side
critical section would enter rcu_read_unlock_special() to clean up --
which could result in deadlock if that critical section happened to be in
the scheduler where the runqueue or priority-inheritance locks were held.
This situation is dealt with by making rcu_preempt_note_context_switch()
check for negative ->rcu_read_lock_nesting, thus refraining from
queuing the task (and from setting RCU_READ_UNLOCK_BLOCKED) if we are
already exiting from the outermost RCU read-side critical section (in
other words, we really are no longer actually in that RCU read-side
critical section). In addition, rcu_preempt_note_context_switch()
invokes rcu_read_unlock_special() to carry out the cleanup in this case,
which clears out the ->rcu_read_unlock_special bits and dequeues the task
(if necessary), in turn avoiding needless delay of the current RCU grace
period and needless RCU priority boosting.
It is still illegal to call rcu_read_unlock() while holding a scheduler
lock if the prior RCU read-side critical section has ever had either
preemption or irqs enabled. However, the common use case is legal,
namely where then entire RCU read-side critical section executes with
irqs disabled, for example, when the scheduler lock is held across the
entire lifetime of the RCU read-side critical section.
Signed-off-by: Paul E. McKenney <paul.mckenney@linaro.org> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Peter Zijlstra [Wed, 20 Jul 2011 16:42:57 +0000 (18:42 +0200)]
sched: Avoid creating superfluous NUMA domains on non-NUMA systems
When creating sched_domains, stop when we've covered the entire
target span instead of continuing to create domains, only to
later find they're redundant and throw them away again.
This avoids single node systems from touching funny NUMA
sched_domain creation code and reduces the risks of the new
SD_OVERLAP code.
Peter Zijlstra [Fri, 15 Jul 2011 08:35:52 +0000 (10:35 +0200)]
sched: Allow for overlapping sched_domain spans
Allow for sched_domain spans that overlap by giving such domains their
own sched_group list instead of sharing the sched_groups amongst
each-other.
This is needed for machines with more than 16 nodes, because
sched_domain_node_span() will generate a node mask from the
16 nearest nodes without regard if these masks have any overlap.
Currently sched_domains have a sched_group that maps to their child
sched_domain span, and since there is no overlap we share the
sched_group between the sched_domains of the various CPUs. If however
there is overlap, we would need to link the sched_group list in
different ways for each cpu, and hence sharing isn't possible.
In order to solve this, allocate private sched_groups for each CPU's
sched_domain but have the sched_groups share a sched_group_power
structure such that we can uniquely track the power.
Reported-and-tested-by: Anton Blanchard <anton@samba.org> Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Link: http://lkml.kernel.org/n/tip-08bxqw9wis3qti9u5inifh3y@git.kernel.org Signed-off-by: Ingo Molnar <mingo@elte.hu>
Shaohua Li [Tue, 19 Jul 2011 15:49:26 +0000 (08:49 -0700)]
vmscan: fix a livelock in kswapd
I'm running a workload which triggers a lot of swap in a machine with 4
nodes. After I kill the workload, I found a kswapd livelock. Sometimes
kswapd3 or kswapd2 are keeping running and I can't access filesystem,
but most memory is free.
This looks like a regression since commit 08951e545918c159 ("mm: vmscan:
correct check for kswapd sleeping in sleeping_prematurely").
Node 2 and 3 have only ZONE_NORMAL, but balance_pgdat() will return 0
for classzone_idx. The reason is end_zone in balance_pgdat() is 0 by
default, if all zones have watermark ok, end_zone will keep 0.
Later sleeping_prematurely() always returns true. Because this is an
order 3 wakeup, and if classzone_idx is 0, both balanced_pages and
present_pages in pgdat_balanced() are 0. We add a special case here.
If a zone has no page, we think it's balanced. This fixes the livelock.
Signed-off-by: Shaohua Li <shaohua.li@intel.com> Acked-by: Mel Gorman <mgorman@suse.de> Cc: Minchan Kim <minchan.kim@gmail.com> Cc: <stable@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
A value larger than INT_MAX cannot be written to the debugfs file created
by debugfs_create_u64 or debugfs_create_x64 on 32bit machine. Because
simple_attr_write() uses simple_strtol() for the conversion.
To fix this, use simple_strtoll() instead.
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com> Cc: Greg Kroah-Hartman <gregkh@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Don't update *inode in __follow_mount_rcu() until we'd verified that
there is mountpoint there. Kudos to Hugh Dickins for catching that
one in the first place and eventually figuring out the solution (and
catching a braino in the earlier version of patch).
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Paul E. McKenney [Sat, 21 May 2011 12:57:18 +0000 (05:57 -0700)]
rcu: Streamline code produced by __rcu_read_unlock()
Given some common flag combinations, particularly -Os, gcc will inline
rcu_read_unlock_special() despite its being in an unlikely() clause.
Use noinline to prohibit this misoptimization.
In addition, move the second barrier() in __rcu_read_unlock() so that
it is not on the common-case code path. This will allow the compiler to
generate better code for the common-case path through __rcu_read_unlock().
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
The RCU_BOOST commits for TREE_PREEMPT_RCU introduced an other-task
write to a new RCU_READ_UNLOCK_BOOSTED bit in the task_struct structure's
->rcu_read_unlock_special field, but, as noted by Steven Rostedt, without
correctly synchronizing all accesses to ->rcu_read_unlock_special.
This could result in bits in ->rcu_read_unlock_special being spuriously
set and cleared due to conflicting accesses, which in turn could result
in deadlocks between the rcu_node structure's ->lock and the scheduler's
rq and pi locks. These deadlocks would result from RCU incorrectly
believing that the just-ended RCU read-side critical section had been
preempted and/or boosted. If that RCU read-side critical section was
executed with either rq or pi locks held, RCU's ensuing (incorrect)
calls to the scheduler would cause the scheduler to attempt to once
again acquire the rq and pi locks, resulting in deadlock. More complex
deadlock cycles are also possible, involving multiple rq and pi locks
as well as locks from multiple rcu_node structures.
This commit fixes synchronization by creating ->rcu_boosted field in
task_struct that is accessed and modified only when holding the ->lock
in the rcu_node structure on which the task is queued (on that rcu_node
structure's ->blkd_tasks list). This results in tasks accessing only
their own current->rcu_read_unlock_special fields, making unsynchronized
access once again legal, and keeping the rcu_read_unlock() fastpath free
of atomic instructions and memory barriers.
The reason that the rcu_read_unlock() fastpath does not need to access
the new current->rcu_boosted field is that this new field cannot
be non-zero unless the RCU_READ_UNLOCK_BLOCKED bit is set in the
current->rcu_read_unlock_special field. Therefore, rcu_read_unlock()
need only test current->rcu_read_unlock_special: if that is zero, then
current->rcu_boosted must also be zero.
This bug does not affect TINY_PREEMPT_RCU because this implementation
of RCU accesses current->rcu_read_unlock_special with irqs disabled,
thus preventing races on the !SMP systems that TINY_PREEMPT_RCU runs on.
Maybe-reported-by: Dave Jones <davej@redhat.com> Maybe-reported-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Reported-by: Steven Rostedt <rostedt@goodmis.org> Signed-off-by: Paul E. McKenney <paul.mckenney@linaro.org> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Reviewed-by: Steven Rostedt <rostedt@goodmis.org>
Paul E. McKenney [Sun, 17 Jul 2011 09:05:49 +0000 (02:05 -0700)]
rcu: decrease rcu_report_exp_rnp coupling with scheduler
PREEMPT_RCU read-side critical sections blocking an expedited grace
period invoke rcu_report_exp_rnp(). When the last such critical section
has completed, rcu_report_exp_rnp() invokes the scheduler to wake up the
task that invoked synchronize_rcu_expedited() -- needlessly holding the
root rcu_node structure's lock while doing so, thus needlessly providing
a way for RCU and the scheduler to deadlock.
This commit therefore releases the root rcu_node structure's lock before
calling wake_up().
Reported-by: Ed Tomlinson <edt@aei.ca> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Chris Wilson [Mon, 18 Jul 2011 20:11:49 +0000 (13:11 -0700)]
drm/i915: Fix unfenced alignment on pre-G33 hardware
Align unfenced buffers on older hardware to the power-of-two object
size. The docs suggest that it should be possible to align only to a
power-of-two tile height, but using the already computed fence size is
easier and always correct. We also have to make sure that we unbind
misaligned buffers upon tiling changes.
In order to prevent a repetition of this bug, we change the interface
to the alignment computation routines to force the caller to provide
the requested alignment and size of the GTT binding rather than assume
the current values on the object.
Reported-and-tested-by: Sitosfe Wheeler <sitsofe@yahoo.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=36326 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: stable@kernel.org Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Keith Packard <keithp@keithp.com>
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
pppoe: Must flush connections when MAC address changes too.
include/linux/sdla.h: remove the prototype of sdla()
tulip: dmfe: Remove old log spamming pr_debugs
Al Viro [Mon, 18 Jul 2011 17:50:40 +0000 (13:50 -0400)]
Fix cifs_get_root()
Add missing ->i_mutex, convert to lookup_one_len() instead of
(broken) open-coded analog, cope with getting something like
a//b as relative pathname. Simplify the hell out of it, while
we are there...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Reviewed-by: Jeff Layton <jlayton@redhat.com>
Joe Perches [Mon, 18 Jul 2011 17:44:44 +0000 (10:44 -0700)]
tulip: dmfe: Remove old log spamming pr_debugs
Commit 726b65ad444d ("tulip: Convert uses of KERN_DEBUG") enabled
some old previously inactive uses of pr_debug converted by
commit dde7c8ef1679 ("tulip/dmfe.c: Use dev_<level> and pr_<level>").
Remove these pr_debugs.
Signed-off-by: Joe Perches <joe@perches.com> Signed-off-by: David S. Miller <davem@davemloft.net>
si4713-i2c: avoid potential buffer overflow on si4713
While compiling it with Fedora 15, I noticed this issue:
inlined from ‘si4713_write_econtrol_string’ at drivers/media/radio/si4713-i2c.c:1065:24:
arch/x86/include/asm/uaccess_32.h:211:26: error: call to ‘copy_from_user_overflow’ declared with attribute error: copy_from_user() buffer size is not provably correct
Al Viro [Mon, 18 Jul 2011 02:27:22 +0000 (22:27 -0400)]
hppfs_lookup(): don't open-code lookup_one_len()
... and it's getting it wrong, too - missing ->d_revalidate() calls when
it's dealing with filesystem (procfs) that has non-trivial ->d_revalidate()...
* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6:
[CIFS] update cifs to version 1.74
[CIFS] update limit for snprintf in cifs_construct_tcon
cifs: Fix signing failure when server mandates signing for NTLMSSP
Merge branch 's5p-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung
* 's5p-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung:
ARM: SAMSUNG: DMA Cleanup as per sparse
ARM: SAMSUNG: Check NULL return from irq_alloc_generic_chip
projects / mv-sheeva.git / log