In theory this will have performance and power improvements. Performance
because we don't need to stall when the scanout BO is busy, and power
because we don't have to stall when the BO is busy (and the ring can
even go to sleep if the HW supports it).
v2:
squash 2 patches into 1 (me)
un-inline the enable_semaphores function (Daniel)
remove comment about SNB hangs from i915_gem_object_sync (Chris)
rename intel_enable_semaphores to i915_semaphore_is_enabled (me)
removed page flip comment; "no why" (Chris)
To address other comments from Daniel (irc):
update the comment to say 'vt-d is crap, don't enable semaphores'
- I think you misinterpreted Chris' comment, it already exists.
checking out whether we can pageflip on the render ring on ivb (didn't
work on early silicon)
- We don't want to enable workarounds for early silicon unless we have
to.
- I can't find any references in the docs about this.
optionally use it if the fb is already busy on the render ring
- This should be how the code already worked, unless I am
misunderstanding your meaning.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
By simplifying the rules to calling get_fence when writing to the
through the GTT in a tiled manner, and calling put_fence before writing
to the object through the GTT in a linear manner, the code becomes
clearer and there is less chance of making a mistake.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
[danvet: fixed up conflict with ppgtt code and spelling in a new
comment.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter wrote
First pull request for 3.5-next, slightly large than usual because new
things kept coming in since the last pull for 3.4.
Highlights:
- first batch of hw enablement for vlv (Jesse et al) and hsw (Eugeni). pci
ids are not yet added, and there's still quite a few patches to merge
(mostly modesetting). To make QA easier I've decided to merge this stuff
in pieces.
- loads of cleanups and prep patches spurred by the above. Especially vlv
is a real frankenstein chip, but also hsw is stretching our driver's
code design. Expect more to come in this area for 3.5.
- more gmbus fixes, cleanups and improvements by Daniel Kurtz. Again,
there are more patches needed (and some already queued up), but I wanted
to split this a bit for better testing.
- pwrite/pread rework and retuning. This series has been in the works for
a few months already and a lot of i-g-t tests have been created for it.
Now it's finally ready to be merged. Note that one patch in this series
touches include/pagemap.h, that patch is acked-by akpm.
- reduce mappable pressure and relocation throughput improvements from
Chris.
- mmap offset exhaustion mitigation by Chris Wilson.
- a start at figuring out which codepaths in our messy dri1/ums+gem/kms
driver we actually need to support by bailing out of unsupported case.
The driver now refuses to load without kms on gen6+ and disallows a few
ioctls that userspace never used in certain cases. More of this will
definitely come.
- More decoupling of global gtt and ppgtt.
- Improved dual-link lvds detection by Takashi Iwai.
- Shut up the compiler + plus fix the fallout (Ben)
- Inverted panel brightness handling (mostly Acer manages to break things
in this way).
- Small fixlets and adjustements and some minor things to help debugging.
Regression-wise QA reported quite a few issues on ivb, but all of them
turned out to be hw stability issues which are already fixed in
drm-intel-fixes (QA runs the nightly regression tests on -next alone,
without -fixes automatically merged in). There's still one issue open on
snb, it looks like occlusion query writes are not quite as cache coherent
as we've expected. With some of the pwrite adjustements we can now
reliably hit this. Kernel workaround for it is in the works."
* 'drm-intel-next' of git://people.freedesktop.org/~danvet/drm-intel: (101 commits)
drm/i915: VCS is not the last ring
drm/i915: Add a dual link lvds quirk for MacBook Pro 8,2
drm/i915: make quirks more verbose
drm/i915: dump the DMA fetch addr register on pre-gen6
drm/i915/sdvo: Include YRPB as an additional TV output type
drm/i915: disallow gem init ioctl on ilk
drm/i915: refuse to load on gen6+ without kms
drm/i915: extract gt interrupt handler
drm/i915: use render gen to switch ring irq functions
drm/i915: rip out old HWSTAM missed irq WA for vlv
drm/i915: open code gen6+ ring irqs
drm/i915: ring irq cleanups
drm/i915: add SFUSE_STRAP registers for digital port detection
drm/i915: add WM_LINETIME registers
drm/i915: add WRPLL clocks
drm/i915: add LCPLL control registers
drm/i915: add SSC offsets for SBI access
drm/i915: add port clock selection support for HSW
drm/i915: add S PLL control
drm/i915: add PIXCLK_GATE register
...
Conflicts:
drivers/char/agp/intel-agp.h
drivers/char/agp/intel-gtt.c
drivers/gpu/drm/i915/i915_debugfs.c
This fixes a resume regression introduced in
commit 7dd4906586
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Wed Mar 21 10:48:18 2012 +0000
drm/i915: Mark untiled BLT commands as fenced on gen2/3
which fixed fencing tracking for untiled blt commands.
A side effect of that patch was that now also untiled objects have a
non-zero obj->last_fenced_seqno to track when a fence can be set up
after a pipelined tiling change. Unfortunately this was only cleared
by the fence setup and teardown code, resulting in tons of untiled but
inactive objects with non-zero last_fenced_seqno.
Now after resume we completely reset the seqno tracking, both on the
driver side (by setting dev_priv->next_seqno = 1) and on the hw side
(by allocating a new hws page, which contains the seqnos). Hilarity
and indefinite waits ensued from the stale seqnos in
obj->last_fenced_seqno from before the suspend.
The fix is to properly clear the fencing tracking state like we
already do for the normal gpu rendering while moving objects off the
active list.
Reported-and-tested-by: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Jiri Slaby <jslaby@suse.cz>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Ums is already disabled, but on ilk we can additionally disable gem
initialization when using user mode setting. Upstream never support
ilk without kernel modesetting and not even the RHEL ilk ums backport
needs gem - that driver is based on xf86-video-intel version 2.2,
which is pre-gem.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The BLT commands on gen2/3 utilize the fence registers and so we cannot
modify any fences for the object whilst those commands are in flight.
Currently we marked tiled commands as occupying a fence, but forgot to
restrict the untiled commands from preventing a fence being assigned
before they were completed.
One side-effect is that we ten have to double check that a fence was
allocated for a fenced buffer during move-to-active.
Reported-by: Jiri Slaby <jirislaby@gmail.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43427
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47990
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Testcase: i-g-t/tests/gem_tiled_after_untiled_blt
Tested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: stable@kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The ppgtt page directory lives in a snatched part of the gtt pte
range. Which naturally gets cleared on hibernate when we pull the
power. Suspend to ram (which is what I've tested) works because
despite the fact that this is a mmio region, it is actually back by
system ram.
Fix this by moving the page directory setup code to the ppgtt init
code (which gets called on resume).
This fixes hibernate on my ivb and snb.
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Haven't seen this yet, but it doesn't hurt.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Beside helping the compiler untangle this maze they double-up as
documentation for which parts of the code aren't performance-critical
but just around to keep old (but already dead-slow) userspace from
breaking.
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The issue is that with inline clflushing the clflushing isn't properly
swizzled. Fix this by
- always clflushing entire 128 byte chunks and
- unconditionally flush before writes when swizzling a given page.
We could be clever and check whether we pwrite a partial 128 byte
chunk instead of a partial cacheline, but I've figured that's not
worth it.
Now the usual approach is to fold this into the original patch series, but
I've opted against this because
- this fixes a corner case only very old userspace relies on and
- I'd like to not invalidate all the testing the pwrite rewrite has gotten.
This fixes the regression notice by tests/gem_tiled_partial_prite_pread
from i-g-t. Unfortunately it doesn't fix the issues with partial pwrites to
tiled buffers on bit17 swizzling machines. But that is also broken without
the pwrite patches, so likely a different issue (or a problem with the
testcase).
v2: Simplify the patch by dropping the overly clever partial write
logic for swizzled pages.
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
drm/i915 wants to read/write more than one page in its fastpath
and hence needs to prefault more than PAGE_SIZE bytes.
Add new functions in filemap.h to make that possible.
Also kill a copy&pasted spurious space in both functions while at it.
v2: As suggested by Andrew Morton, add a multipage parameter to both
functions to avoid the additional branch for the pagemap.c hotpath.
My gcc 4.6 here seems to dtrt and indeed reap these branches where not
needed.
v3: Becaus I couldn't find a way around adding a uaddr += PAGE_SIZE to
the filemap.c hotpaths (that the compiler couldn't remove again),
let's go with separate new functions for the multipage use-case.
v4: Adjust comment to CodingStlye and fix spelling.
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
While moving around things, this two functions slowly grew out of any
sane bounds. So extract a few lines that do the copying and
clflushing. Also add a few comments to explain what's going on.
v2: Again do s/needs_clflush/needs_clflush_after/ in the write paths
as suggested by Chris Wilson.
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
It's around 20% faster.
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
It's too expensive to move it around just for that pwrite, especially
when we're trashing on the mappable gtt part like crazy.
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
In micro-benchmarking of the usual pwrite use-pattern of alternating
pwrites with gtt domain reads from the gpu, this yields around 30%
improvement of pwrite throughput across all buffers size. The trick is
that we can avoid clflush cachelines that we will overwrite completely
anyway.
Furthermore for partial pwrites it gives a proportional speedup on top
of the 30% percent because we only clflush back the part of the buffer
we're actually writing.
v2: Simplify the clflush-before-write logic, as suggested by Chris
Wilson.
v3: Finishing touches suggested by Chris Wilson:
- add comment to needs_clflush_before and only set this if the bo is
uncached.
- s/needs_clflush/needs_clflush_after/ in the write paths to clearly
differentiate it from needs_clflush_before.
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The pagemap.h prefault helpers do the prefaulting by simply writing
some data into every page. Hence we should not prefault when we're not
yet commited to to actually writing data to userspace. The problem is
now that
- we can't prefault while holding dev->struct_mutex for we could
deadlock with our own pagefault handler
- we need to grab dev->struct_mutex before copying to sync up with any
outsanding gpu writes.
Therefore only prefault when we're dropping the lock the first time in
the pread slowpath - at that point we're committed to the write, don't
wait on the gpu anymore and hence won't return early (with e.g.
-EINTR).
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
With the proper prefault, it's extremely unlikely that we fall back
to the gtt slowpath.
So just kill it and use the shmem_pwrite path as fallback.
To further clean up the code, move the preparatory gem calls into the
respective pwrite functions. This way the gtt_fast->shmem fallback
is much more obvious.
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This speeds up pwrite and pread from ~120 µs ro ~100 µs for
reading/writing 1mb on my snb (if the backing storage pages
are already pinned, of course).
v2: Chris Wilson pointed out a glaring page reference bug - I've
unconditionally dropped the reference. With that fixed (and the
associated reduction of dirt in dmesg) it's now even a notch faster.
v3: Unconditionaly grab a page reference when dropping
dev->struct_mutex to simplify the code-flow.
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
~120 µs instead fo ~210 µs to write 1mb on my snb. I like this.
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
No longer needed.
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This is obviously gonna slow down pread. But for a half-way realistic
micro-benchmark, it doesn't matter: Non-broken userspace reads back
data from the gpu once before the gpu again dirties it.
So all this ranged clflush tracking is just a waste of time.
No pread performance change (neglecting the dumb benchmark of
constantly reading the same data) measured.
As an added bonus, this avoids clflush on read on coherent objects.
Which means that partial preads on snb are now roughly 4x as fast.
This will be usefull for e.g. the libva encoder - when I finally get
around to fix that up.
v2: Properly sync with the gpu on LLC machines.
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
With the previous rewrite, they've become essential identical.
v2: Simplify the page_do_bit17_swizzling logic as suggested by Chris
Wilson.
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
With the previous rewrite, they've become essential identical.
v2: Simplify the page_do_bit17_swizzling logic as suggested by Chris
Wilson.
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We try to avoid writing the relocations through the uncached GTT, if the
buffer is currently in the CPU write domain and so will be flushed out to
main memory afterwards anyway. Also on SandyBridge we can safely write
to the pages in cacheable memory, so long as the buffer is LLC mapped.
In either of these cases, we therefore do not need to force the
reallocation of the buffer into the mappable region of the GTT, reducing
the aperture pressure.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
... because this is what it actually doesn now that we have the global
gtt vs. ppgtt split.
Also move it to the other global gtt functions in i915_gem_gtt.c
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
If we discard a buffer due to memory pressure, also release its alloted
mmap address space. As it may be sometime before userspace wakes up
and notices that it has buffers to purge from its cache, we may waste
valuable address space on unusable objects for a period of time.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47738
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Introduced in commit 8461d226 and 8c59967c
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
[danvet: s/fix/shut up/ in the commit msg.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Now that everything is in place, only bind to the global gtt
when actually required. Patch split-up suggested by Chris Wilson.
Reviewed-and-tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
And track the existence of such a binding similar to the aliasing
ppgtt case. Speeds up binding/unbinding in the common case where we
only need a ppgtt binding (which is accessed in a cpu coherent fashion
by the gpu) and no gloabl gtt binding (which needs uc writes for the
ptes).
This patch just puts the required tracking in place.
v2: Check that global gtt mappings exist in the error_state capture
code (with Chris Wilson's llc reloc patches batchbuffers are no longer
relocated as mappable in all situations, so this matters). Suggested
by Chris Wilson.
v3: Adapted to Chris' latest llc-reloc patches.
v4: Fix a bug in the i915 error state capture code noticed by Chris
Wilson.
Reviewed-and-tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Note that there's a functional change buried in this patch wrt the ilk
dmar workaround: We now only idle the gpu while tearing down the dmar
mappings, not while clearing the gtt. Keeping the current semantics
would have made for some really ugly code and afaik the issue is only
with the dmar unmapping that needs a fully idle gpu.
Reviewed-and-tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This error message has since been superseded by the hangcheck, and does
not add any salient information beyond that already printed by hangcheck
discovering the GPU hang that lead to i915_wait_request() bombing out in
the first place.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
By recording the location of every request in the ringbuffer, we know
that in order to retire the request the GPU must have finished reading
it and so the GPU head is now beyond the tail of the request. We can
therefore provide a conservative estimate of where the GPU is reading
from in order to avoid having to read back the ring buffer registers
when polling for space upon starting a new write into the ringbuffer.
A secondary effect is that this allows us to convert
intel_ring_buffer_wait() to use i915_wait_request() and so consolidate
upon the single function to handle the complicated task of waiting upon
the GPU. A necessary precaution is that we need to make that wait
uninterruptible to match the existing conditions as all the callers of
intel_ring_begin() have not been audited to handle ERESTARTSYS
correctly.
By using a conservative estimate for the head, and always processing all
outstanding requests first, we prevent a race condition between using
the estimate and direct reads of I915_RING_HEAD which could result in
the value of the head going backwards, and the tail overflowing once
again. We are also careful to mark any request that we skip over in
order to free space in ring as consumed which provides a
self-consistency check.
Given sufficient abuse, such as a set of unthrottled GPU bound
cairo-traces, avoiding the use of I915_RING_HEAD gives a 10-20% boost on
Sandy Bridge (i5-2520m):
firefox-paintball 18927ms -> 15646ms: 1.21x speedup
firefox-fishtank 12563ms -> 11278ms: 1.11x speedup
which is a mild consolation for the performance those traces achieved from
exploiting the buggy autoreported head.
v2: Add a few more comments and make request->tail a conservative
estimate as suggested by Daniel Vetter.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
[danvet: resolve conflicts with retirement defering and the lack of
the autoreport head removal (that will go in through -fixes).]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Currently we reserve seqnos only when we emit the request to the ring
(by bumping dev_priv->next_seqno), but start using it much earlier for
ring->oustanding_lazy_request. When 2 threads compete for the gpu and
run on two different rings (e.g. ddx on blitter vs. compositor)
hilarity ensued, especially when we get constantly interrupted while
reserving buffers.
Breakage seems to have been introduced in
commit 6f392d5486
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Sat Aug 7 11:01:22 2010 +0100
drm/i915: Use a common seqno for all rings.
This patch fixes up the seqno reservation logic by moving it into
i915_gem_next_request_seqno. The ring->add_request functions now
superflously still return the new seqno through a pointer, that will
be refactored in the next patch.
Note that with this change we now unconditionally allocate a seqno,
even when ->add_request might fail because the rings are full and the
gpu died. But this does not open up a new can of worms because we can
already leave behind an outstanding_request_seqno if e.g. the caller
gets interrupted with a signal while stalling for the gpu in the
eviciton paths. And with the bugfix we only ever have one seqno
allocated per ring (and only that ring), so there are no ordering
issues with multiple outstanding seqnos on the same ring.
v2: Keep i915_gem_get_seqno (but move it to i915_gem.c) to make it
clear that we only have one seqno counter for all rings. Suggested by
Chris Wilson.
v3: As suggested by Chris Wilson use i915_gem_next_request_seqno
instead of ring->oustanding_lazy_request to make the follow-up
refactoring more clearly correct. Also improve the commit message
with issues discussed on irc.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=45181
Tested-by: Nicolas Kalkhof nkalkhof()at()web.de
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We want to unconditionally enable ppgtt for two reasons:
- Windows uses this on snb and later.
- We need the basic hw support to work before we can think about real
per-process address spaces and other cool features we want.
But Chris Wilson was complaining all over irc and intel-gfx that this
will blow up if we don't have a module option to disable it. Hence add
one, to prevent this.
ppgtt support seems to slightly change the timings and make crashy
things slightly more or less crashy. Now in my testing and the testing
this got on troublesome snb machines, it seems to have improved things
only. But on ivb it makes quite a few crashes happen much more often,
see
https://bugs.freedesktop.org/show_bug.cgi?id=41353
Luckily Eugeni Dodonov seems to have a set of workarounds that fix
this issue.
v2: Don't try to enable ppgtt on pre-snb.
v3: Pimp commit message and make Chris Wilson less grumpy by adding a
module option.
v4: New try at making Chris Wilson happy.
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Reviewed-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This adds support to bind/unbind objects and wires it up. Objects are
only put into the ppgtt when necessary, i.e. at execbuf time.
Objects are still unconditionally put into the global gtt.
v2: Kill the quick hack and explicitly pass cache_level to ppgtt_bind
like for the global gtt function. Noticed by Chris Wilson.
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Reviewed-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
On gen5 we also need to correctly set up swizzling in the display
scanout engine, but only there. Consolidate this into the same
function.
This has a small effect on ums setups - the kernel now also sets this
bit in addition to userspace setting it. Given that this code only
runs when userspace either can't (resume, gpu reset) or explicitly
won't(gem_init) touch the hw this shouldn't have an adverse effect.
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We have to do this manually. Somebody had a Great Idea.
I've measured speed-ups just a few percent above the noise level
(below 5% for the best case), but no slowdows. Chris Wilson measured
quite a bit more (10-20% above the usual snb variance) on a more
recent and better tuned version of sna, but also recorded a few
slow-downs on benchmarks know for uglier amounts of snb-induced
variance.
v2: Incorporate Ben Widawsky's preliminary review comments and
elaborate a bit about the performance impact in the changelog.
v3: Add a comment as to why we don't need to check the 3rd memory
channel.
v4: Fixup whitespace.
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Like for shmem_pwrite_slow. The only difference is that because we
read data, we can leave the fetched cachelines in the cpu: In the case
that the object isn't in the cpu read domain anymore, the clflush for
the next cpu read domain invalidation will simply drop these
cachelines.
slow_shmem_bit17_copy is now ununsed, so kill it.
With this patch tests/gem_mmap_gtt now actually works.
v2: add __ to copy_to_user_swizzled as suggested by Chris Wilson.
v3: Fixup the swizzling logic, it swizzled the wrong pages.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=38115
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
... instead of get_user_pages, because that fails on non page-backed
user addresses like e.g. a gtt mapping of a bo.
To get there essentially copy the vfs read path into pagecache. We
can't call that right away because we have to take care of bit17
swizzling. To not deadlock with our own pagefault handler we need
to completely drop struct_mutex, reducing the atomicty-guarantees
of our userspace abi. Implications for racing with other gem ioctl:
- execbuf, pwrite, pread: Due to -EFAULT fallback to slow paths there's
already the risk of the pwrite call not being atomic, no degration.
- read/write access to mmaps: already fully racy, no degration.
- set_tiling: Calling set_tiling while reading/writing is already
pretty much undefined, now it just got a bit worse. set_tiling is
only called by libdrm on unused/new bos, so no problem.
- set_domain: When changing to the gtt domain while copying (without any
read/write access, e.g. for synchronization), we might leave unflushed
data in the cpu caches. The clflush_object at the end of pwrite_slow
takes care of this problem.
- truncating of purgeable objects: the shmem_read_mapping_page call could
reinstate backing storage for truncated objects. The check at the end
of pwrite_slow takes care of this.
v2:
- add missing intel_gtt_chipset_flush
- add __ to copy_from_user_swizzled as suggest by Chris Wilson.
v3: Fixup bit17 swizzling, it swizzled the wrong pages.
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The gtt_pwrite slowpath grabs the userspace memory with
get_user_pages. This will not work for non-page backed memory, like a
gtt mmapped gem object. Hence fall throuh to the shmem paths if we hit
-EFAULT in the gtt paths.
Now the shmem paths have exactly the same problem, but this way we
only need to rearrange the code in one write path.
v2: v1 accidentaly falls back to shmem pwrite for phys objects. Fixed.
v3: Make the codeflow around phys_pwrite cleara as suggested by Chris
Wilson.
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The original intention of comparing the bo against the mappable GTT
limits was to prevent a subsequent faulting of the bo into the GTT from
clearing the entire GTT in vain. However, that was clearly a cut'n'paste
mistake as a CPU mapping never binds the bo into the aperture. Whilst
there may be some merit to limiting the maximum size of the bo to
something that can be utilized by the GPU, that limit itself does not
belong as a safeguard to mmapping the bo, so remove the check entirely.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
With the fence accounting fixed up in the previous commit not finding
enough fences is a fatal error and userspace bug. Trashing the entire
gtt is not gonna turn up that missing fence, so don't to this by
returning another error thatn ENOSPC.
This has the added benefit that it's easier to distinguish fence
accounting errors from gtt space accounting issues.
TTM serves as precendence for the EDEADLK error code - it returns it
when the reservation code needs resources already blocked by the
current reservation.
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
In order to correctly account for reserving space in the GTT and fences
for a batch buffer, we need to independently track whether the fence is
pinned due to a fenced GPU access in the batch or whether the buffer is
pinned in the aperture. Currently we count the fenced as pinned if the
buffer has already been seen in the execbuffer. This leads to a false
accounting of available fence registers, causing frequent mass evictions.
Worse, if coupled with the change to make i915_gem_object_get_fence()
report EDADLK upon fence starvation, the batchbuffer can fail with only
one fence required...
Fixes intel-gpu-tools/tests/gem_fenced_exec_thrash
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=38735
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Tested-by: Paul Neumann <paul104x@yahoo.de>
[danvet: Resolve the functional conflict with Jesse Barnes sprite
patches, acked by Chris Wilson on irc.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Sometimes it may be the case when we idle the gpu or wait on something
we don't actually want to process the retiring list. This patch allows
callers to choose the behavior.
Reviewed-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
LLC is not SNB/IVB-specific, so we should check for it in a more generic
way.
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The waits we do here are generally so short that sleeping is a bad
idea unless we have an IRQ to wake us up. Improves regression test
performance from 18 minutes to 3.5 minutes on gen7, which is now
consistent with the previous generation.
Signed-off-by: Eric Anholt <eric@anholt.net>
Tested-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Reviewed-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Keith Packard <keithp@keithp.com>
As a workaround for IRQ synchronization issues in the gen7 BLT ring,
we want to turn the two wait functions into polling loops.
Signed-off-by: Eric Anholt <eric@anholt.net>
Tested-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Reviewed-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Keith Packard <keithp@keithp.com>
This reverts commit eb1711bb94.
It blows up the i915 seqno tracking, resulting in the
BUG_ON(seqno == 0);
in i915_wait_request() triggering, which will cause lock-ups.
See for example
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/903010https://lkml.org/lkml/2011/12/14/395
Reported-requested-and-tested-by: Dirk Hohndel <dirk@hohndel.org>
Reported-by: Richard Eames <Richard.Eames@flinders.edu.au>
Reported-by: Rocko Requin <rockorequin@hotmail.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Keith Packard <keithp@keithp.com>
Cc: Eric Anholt <eric@anholt.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The recursion loop goes retire_requests->unbind->gpu_idle->retire_reqeusts.
Every time we go through this we need a
- active object that can be retired
- and there are no other references to that object than the one from
the active list, so that it gets unbound and freed immediately.
Otherwise the recursion stops. So the recursion is only limited by the
number of objects that fit these requirements sitting in the active list
any time retire_request is called.
Issue exercised by tests/gem_unref_active_buffers from i-g-t.
There's been a decent bikeshed discussion whether it wouldn't be
better to pass around a flag, but imo this is o.k. for such a limited
case that only supports a w/a.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42180
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Chris Wilson <chris@chris-wilson>
[ickle- we built better bikesheds, but this keeps the rain off for now]
Tested-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
A call to i915_add_request() has been made in function i915_gem_busy_ioctl(). i915_add_request can fail,
so in it's exit path previously allocated memory needs to be freed.
Signed-off-by: Rakib Mullick <rakib.mullick@gmail.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
IVB supports these bits as well.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Keith Packard <keithp@keithp.com>
In preparation of to support 32 fences on Ivybdrigde.
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Keith Packard <keithp@keithp.com>
I've been seeing memory leaks on my system in the form of large
(300-400MB) GEM objects created by now-dead processes laying around
clogging up memory. I usually notice when it gets to about 1.2GB of
them. Hopefully this clears up the issue, but I just found this bug
by inspection.
Signed-off-by: Eric Anholt <eric@anholt.net>
Cc: stable@kernel.org
Signed-off-by: Keith Packard <keithp@keithp.com>
[Description from: Daniel Vetter]
I've just discussed this quickly with Chris on irc and it's probably
best to just kill the list_empty early bailout. gpu_idle isn't a
fastpath, so who cares. One candidate where we emit commands to the ring
without adding anything onto these lists is e.g. pageflip. There are
probably more.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Keith Packard <keithp@keithp.com>
We currently only round up the userspace size to the next page. We
assume that userspace hasn't made a mistake and requested a zero-length
gem object and all through our internal code we then presume that every
object is backed by at least a single page. Fix that oversight and
report EINVAL back to userspace if they try to create a zero length
object.
[danvet: This fixes tests/gem_bad_length]
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
Use the helper function already employed by the pwrite/pread
functions.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Keith Packard <keithp@keithp.com>
Various issues involved with the space character were generating
warnings in the checkpatch.pl file. This patch removes most of those
warnings.
Signed-off-by: Akshay Joshi <me@akshayjoshi.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
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
Signed-off-by: Keith Packard <keithp@keithp.com>
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>
Provide a parameter to disable hanghcheck. This is useful mostly for
developers trying to debug known problems, and probably should not be
touched by normal users.
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
Konstantin Belousov pointed out that 4697995b98 replaced the generic
i915_driver_irq_*install() functions with chipset specific routines
accessible only through driver->irq_*install(). So update the sanity
check in i915_request_wait() to match.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Keith Packard <keithp@keithp.com>
The interface to ->truncate_range is changing very slightly: once "tmpfs:
take control of its truncate_range" has been applied, this can be applied.
For now there is only a slight inefficiency while this remains unapplied,
but it will soon become essential for managing shmem's use of swap.
Change i915_gem_object_truncate() to use shmem_truncate_range() directly:
which should also spare i915 later change if we switch from
inode_operations->truncate_range to file_operations->fallocate.
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Keith Packard <keithp@keithp.com>
Cc: Dave Airlie <airlied@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Soon tmpfs will stop supporting ->readpage and read_cache_page_gfp(): once
"tmpfs: add shmem_read_mapping_page_gfp" has been applied, this patch can
be applied to ease the transition.
Make i915_gem_object_get_pages_gtt() use shmem_read_mapping_page_gfp() in
the one place it's needed; elsewhere use shmem_read_mapping_page(), with
the mapping's gfp_mask properly initialized.
Forget about __GFP_COLD: since tmpfs initializes its pages with memset,
asking for a cold page is counter-productive.
Include linux/shmem_fs.h also in drm_gem.c: with shmem_file_setup() now
declared there too, we shall remove the prototype from linux/mm.h later.
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Keith Packard <keithp@keithp.com>
Cc: Dave Airlie <airlied@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Even if the object is no longer in the GTT domain, there may still be
a user space mapping which needs to be released.
Without this fix, render-based text (mostly in firefox) would
occasionally get corrupted when the system was under load.
Signed-off-by: Keith Packard <keithp@keithp.com>
This reverts commit 4a684a4117.
Userland has always been required to set the object's domain to GTT
before using it through a GTT mapping, it's not something that the
kernel is supposed to enforce. (The pagefault support is so that we
can handle multiple mappings without userland having to pin across
them, not so that userland can use GTT after GPU domains without
telling the kernel).
Fixes 19.2% +/- 0.8% (n=6) performance regression in cairo-gl
firefox-talos-gfx on my T420 latop.
Signed-off-by: Keith Packard <keithp@keithp.com>
It seems to me that we are leaking 'user_pages' in
drivers/gpu/drm/i915/i915_gem.c::i915_gem_shmem_pread_slow() if
read_cache_page_gfp() fails.
Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Improves full-screen openarena on my laptop 20.3% +/- 4.0% (n=3)
Improves 800x600 nexuiz on my laptop 12.3% +/- 0.1% (n=3)
We have more room to improve with doing LLC caching for display using
GFDT, and in doing LLC+MLC caching, but this was an easy performance
win and incremental improvement toward those two.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The simplest and common method for ensuring scanout coherency on all
chipsets is to mark the scanout buffers as uncached (and for
userspace to remember to flush the render cache every so often).
We can improve upon this for later generations by marking scanout
objects as GFDT and only flush those cachelines when required. However,
we start simple.
[v2: Move the set to uncached above the clflush. Otherwise, we'd skip
the clflush and try to scan out data that was still sitting in the
cache.]
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We need to perform a few operations in order to move the object into the
display plane (where it can be accessed coherently by the display
engine) that are important for future safety to forbid whilst pinned. As a
result, we want to need to perform some of the operations before pinning,
but some are required once we have been bound into the GTT. So combine
the pinning performed by all the callers with set_to_display_plane(), so
this complication is contained within the single function.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
[anholt v2: Don't forget that when going from cached to uncached, we
haven't been tracking the write domain from the CPU perspective, since
we haven't needed it for GPU coherency.]
[ickle v3: We also need to make sure we relinquish any fences on older
chipsets and clear the GTT for sane domain tracking.]
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Like its siblings finish_gpu(), this function clears the object from the
GTT domain forcing it to be trigger a domain invalidation should we ever
need to use via the GTT again.
Note that the most important side-effect of finishing the GTT domain
(aside from clearing the tracking read/write domains) is that it imposes
an memory barrier so that all accesses are complete before it returns,
which is important if you intend to be modifying translation tables
shortly afterwards. The second most important side-effect is that it
tears down the GTT mappings forcing a page-fault and invalidation on
next user access to the object.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
... reincarnated from i915_gem_object_flush_gpu(). The semantic
difference is that after calling finish_gpu() the object no longer
resides in any GPU domain, and so will cause the GPU caches to be
invalidated if it is ever used again.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
A tile on gen2 has a size of 2kb, stride of 128 bytes and 16 rows.
Userspace was broken and assumed 8 rows. Chris Wilson noted that the
kernel unfortunately can't reliable check that because libdrm rounds
up the size to the next bucket.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Keith Packard <keithp@keithp.com>
Convert our open coded offset_in_page() to the common macro.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
Change each shrinker's API by consolidating the existing parameters into
shrink_control struct. This will simplify any further features added w/o
touching each file of shrinker.
[akpm@linux-foundation.org: fix build]
[akpm@linux-foundation.org: fix warning]
[kosaki.motohiro@jp.fujitsu.com: fix up new shrinker API]
[akpm@linux-foundation.org: fix xfs warning]
[akpm@linux-foundation.org: update gfs2]
Signed-off-by: Ying Han <yinghan@google.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Acked-by: Pavel Emelyanov <xemul@openvz.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Acked-by: Rik van Riel <riel@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The registers are the same as on Sandybridge. Fixes scrambled display
in X when it does software drawing to the GTT, and scans the results
out as tiled.
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Keith Packard <keithp@keithp.com>
This is once less place to miss a new INTEL_INFO(dev)->gen update now.
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Keith Packard <keithp@keithp.com>
Rely on the GPU snooping into the CPU cache for appropriately bound
objects on MI_FLUSH. Or perhaps one day we will have a cache-coherent
CPU/GPU package...
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Keith Packard <keithp@keithp.com>
... to clarify just how we use it inside the driver and remove the
confusion of the poorly matching agp_type names. We still need to
translate through agp_type for interface into the fake AGP driver.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Keith Packard <keithp@keithp.com>
We always skipped flushing the BLT ring if the request flush did not
include the RENDER domain. However, this neglects that we try to flush
the COMMAND domain after every batch and before the breadcrumb interrupt
(to make sure the batch is indeed completed prior to the interrupt
firing and so insuring CPU coherency). As a result of the missing flush,
incoherency did indeed creep in, most notable when using lots of command
buffers and so potentially rewritting an active command buffer (i.e.
the GPU was still executing from it even though the following interrupt
had already fired and the request/buffer retired).
As all ring->flush routines now have the same preconditions, de-duplicate
and move those checks up into i915_gem_flush_ring().
Fixes gem_linear_blit.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=35284
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Tested-by: mengmeng.meng@intel.com
... even though it was disabled. A mistake in the handling of fence reuse
caused us to skip the vital delay of waiting for the object to finish
rendering before changing the register. This resulted in us changing the
fence register whilst the bo was active and so causing the blits to
complete using the wrong stride or even the wrong tiling. (Visually the
effect is that small blocks of the screen look like they have been
interlaced). The fix is to wait for the GPU to finish using the memory
region pointed to by the fence before changing it.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=34584
Cc: Andy Whitcroft <apw@canonical.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
[Note for 2.6.38-stable, we need to reintroduce the interruptible passing]
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Dave Airlie <airlied@linux.ie>
When i915_gem_retire_requests_ring calls i915_gem_request_remove_from_client,
the client_list for that request may already be removed in i915_gem_release.
So we may call twice list_del(&request->client_list), resulting in an
oops like this report:
[126167.230394] BUG: unable to handle kernel paging request at 00100104
[126167.230699] IP: [<f8c2ce44>] i915_gem_retire_requests_ring+0xd4/0x240 [i915]
[126167.231042] *pdpt = 00000000314c1001 *pde = 0000000000000000
[126167.231314] Oops: 0002 [#1] SMP
[126167.231471] last sysfs file: /sys/devices/LNXSYSTM:00/device:00/PNP0C0A:00/power_supply/BAT1/current_now
[126167.231901] Modules linked in: snd_seq_dummy nls_utf8 isofs btrfs zlib_deflate libcrc32c ufs qnx4 hfsplus hfs minix ntfs vfat msdos fat jfs xfs exportfs reiserfs cryptd aes_i586 aes_generic binfmt_misc vboxnetadp vboxnetflt vboxdrv parport_pc ppdev snd_hda_codec_hdmi snd_hda_codec_conexant snd_hda_intel snd_hda_codec snd_hwdep arc4 snd_pcm snd_seq_midi snd_rawmidi snd_seq_midi_event snd_seq uvcvideo videodev snd_timer snd_seq_device joydev iwlagn iwlcore mac80211 snd cfg80211 soundcore i915 drm_kms_helper snd_page_alloc psmouse drm serio_raw i2c_algo_bit video lp parport usbhid hid sky2 sdhci_pci ahci sdhci libahci
[126167.232018]
[126167.232018] Pid: 1101, comm: Xorg Not tainted 2.6.38-6-generic-pae #34-Ubuntu Gateway MC7833U /
[126167.232018] EIP: 0060:[<f8c2ce44>] EFLAGS: 00213246 CPU: 0
[126167.232018] EIP is at i915_gem_retire_requests_ring+0xd4/0x240 [i915]
[126167.232018] EAX: 00200200 EBX: f1ac25b0 ECX: 00000040 EDX: 00100100
[126167.232018] ESI: f1a2801c EDI: e87fc060 EBP: ef4d7dd8 ESP: ef4d7db0
[126167.232018] DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
[126167.232018] Process Xorg (pid: 1101, ti=ef4d6000 task=f1ba6500 task.ti=ef4d6000)
[126167.232018] Stack:
[126167.232018] f1a28000 f1a2809c f1a28094 0058bd97 f1aa2400 f1a2801c 0058bd7b 0058bd85
[126167.232018] f1a2801c f1a28000 ef4d7e38 f8c2e995 ef4d7e30 ef4d7e60 c14d1ebc f6b3a040
[126167.232018] f1522cc0 000000db 00000000 f1ba6500 ffffffa1 00000000 00000001 f1a29214
[126167.232018] Call Trace:
Unfortunately the call trace reported was cut, but looking at debug
symbols the crash is at __list_del, when probably list_del is called
twice on the same request->client_list, as the dereferenced value is
LIST_POISON1 + 4, and by looking more at the debug symbols before
list_del call it should have being called by
i915_gem_request_remove_from_client
And as I can see in the code, it seems we indeed have the possibility
to remove a request->client_list twice, which would cause the above,
because we do list_del(&request->client_list) on both
i915_gem_request_remove_from_client and i915_gem_release
As Chris Wilson pointed out, it's indeed the case:
"(...) I had thought that the actual insertion/deletion was serialised
under the struct mutex and the intention of the spinlock was to protect
the unlocked list traversal during throttling. However, I missed that
i915_gem_release() is also called without struct mutex and so we do need
the double check for i915_gem_request_remove_from_client()."
This change does the required check to avoid the duplicate remove of
request->client_list.
Bugzilla: http://bugs.launchpad.net/bugs/733780
Cc: stable@kernel.org # 2.6.38
Signed-off-by: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
* 'intel/drm-intel-next' of ../drm-next: (755 commits)
drm/i915: Only wait on a pending flip if we intend to write to the buffer
drm/i915/dp: Sanity check eDP existence
drm/i915: Rebind the buffer if its alignment constraints changes with tiling
drm/i915: Disable GPU semaphores by default
drm/i915: Do not overflow the MMADDR write FIFO
Revert "drm/i915: fix corruptions on i8xx due to relaxed fencing"
drm/i915: Don't save/restore hardware status page address register
drm/i915: don't store the reg value for HWS_PGA
drm/i915: fix memory corruption with GM965 and >4GB RAM
Linux 2.6.38-rc7
Revert "TPM: Long default timeout fix"
drm/i915: Re-enable GPU semaphores for SandyBridge mobile
drm/i915: Replace vblank PM QoS with "Interrupt-Based AGPBUSY#"
Revert "drm/i915: Use PM QoS to prevent C-State starvation of gen3 GPU"
drm/i915: Allow relocation deltas outside of target bo
drm/i915: Silence an innocuous compiler warning for an unused variable
fs/block_dev.c: fix new kernel-doc warning
ACPI: Fix build for CONFIG_NET unset
mm: <asm-generic/pgtable.h> must include <linux/mm_types.h>
x86: Use u32 instead of long to set reset vector back to 0
...
Conflicts:
drivers/gpu/drm/i915/i915_gem.c
Early gen3 and gen2 chipset do not have the relaxed per-surface tiling
constraints of the later chipsets, so we need to check that the GTT
alignment is correct for the new tiling. If it is not, we need to
rebind.
Reported-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The code paths for modesetting are growing in complexity as we may need
to move the buffers around in order to fit the scanout in the aperture.
Therefore we face a choice as to whether to thread the interruptible status
through the entire pinning and unbinding code paths or to add a flag to
the device when we may not be interrupted by a signal. This does the
latter and so fixes a few instances of modesetting failures under stress.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Dave Airlie spotted that we had a potential bug should we ever rearrange
the drm_i915_gem_object so not the base drm_gem_object was not its first
member. He noticed that we often convert the return of
drm_gem_object_lookup() immediately into drm_i915_gem_object and then
check the result for nullity. This is only valid when the base object is
the first member and so the superobject has the same address. Play safe
instead and use the compiler to convert back to the original return
address for sanity testing.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
... so we handle that for i915_gem_fault() in the same manner as
ERESTARTSYS, or we send a SIGBUS to the faulting application.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
A lot of minor tweaks to fix the tracepoints, improve the outputting for
ftrace, and to generally make the tracepoints useful again. It is a start
and enough to begin identifying performance issues and gaps in our
coverage.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
By returning EAGAIN upon a wedged GPU before attempting to wait, we
would hit an infinite loop of repeating operation without ever
progressing. Instead this needs to be EIO so that userspace knows that
the GPU is truly wedged and not in the process of error recovery.
Similarly, we need to handle the error recovery during i915_gem_fault.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
This is just an idea that might or might not be a good idea,
it basically adds two ioctls to create a dumb and map a dumb buffer
suitable for scanout. The handle can be passed to the KMS ioctls to create
a framebuffer.
It looks to me like it would be useful in the following cases:
a) in development drivers - we can always provide a shadowfb fallback.
b) libkms users - we can clean up libkms a lot and avoid linking
to libdrm_*.
c) plymouth via libkms is a lot easier.
Userspace bits would be just calls + mmaps. We could probably
mark these handles somehow as not being suitable for acceleartion
so as top stop people who are dumber than dumb.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Instead of reporting EIO upfront in the entrance of an ioctl that may or
may not attempt to use the GPU, defer the actual detection of an invalid
ioctl to when we issue a GPU instruction. This allows us to continue to
use bo in video memory (via pread/pwrite and mmap) after the GPU has hung.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
We can only utilize the stolen portion of the GTT if we are in sole
charge of the hardware. This is only true if using GEM and KMS,
otherwise VESA continues to access stolen memory.
Reported-by: Arnd Bergmann <arnd@arndb.de>
Reported-by: Frederic Weisbecker <fweisbec@gmail.com>
Tested-by: Jiri Olsa <jolsa@redhat.com>
Tested-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
There are I915_NUM_RINGS-1 inter-ring synchronisation counters, but we
were clearing I915_NUM_RINGS of them. Oops.
Reported-by: Jiri Slaby <jirislaby@gmail.com>
Tested-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Rather than evicting an object at random, which is unlikely to alleviate
the memory pressure sufficient to allow us to continue, zap the entire
aperture. That should give the system long enough to recover and reap
some pages from the evicted objects, forestalling the allocation error
for the new object.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
In order to retire active buffers whilst no client is active, we need to
insert our own flush requests onto the ring.
This is useful for servers that queue up some rendering and then go to
sleep as it allows us to the complete processing of those requests,
potentially making that memory available again much earlier.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
We need to ensure that writes through the GTT land before any
modification to the MMIO registers and so must impose a mandatory write
barrier when flushing the GTT domain. This was revealed by relaxing the
write ordering by experimentally mapping the registers and the GATT as
write-combining.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The relative-to-general state default is useless as it means having to
rewrite the streaming kernels for each batch. Relative-to-surface is
more useful, as that stream usually needs to be rewritten for each
batch. And absolute addressing mode, vital if you start streaming
state, is also only available by adjusting the register...
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
In order to enforce the correct memory barriers for irq get/put, we need
to perform the actual counting using atomic operations.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
In order for bos to retire eventually, a request must be sent down the
ring. This is expected, for example, by occlusion queries for which mesa
will wait upon (whilst running glean) before issuing more batches and so
the normal activity upon the ring is suspended and we need to emit a
request to clear the idle ring.
Reported-by: Jinjin, Wang <jinjin.wang@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=30380
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
I'm still seeing tiling corruption of PutImage and CopyArea (I think)
under mutter on pnv, so obviously the pipelining logic is deeply flawed.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The bulk of the change is to convert the growing list of rings into an
array so that the relationship between the rings and the semaphore sync
registers can be easily computed.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
As the tracepoint is now decoupled from when the actual register is
assigned and was never complemented by detailing when the object lost
its fence, it has outlived its limited usefulness. Profiling the actual
stalls is a far more profitable venture anyway.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
As the userspace mappings are torn down on every GPU write, we prefer to
track when the buffer is activated (via a fresh i915_gem_fault). This
makes the LRU conceptually simpler. With coherent mappings, the
remaining use-case for set_domain_ioctl is GPU synchronisation.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
With this change, every batchbuffer can use all available fences (save
pinned and scanout, of course) without ever stalling the gpu!
In theory. Currently the actual pipelined update of the register is
disabled due to some stability issues. However, just the deferred update
is a significant win.
Based on a series of patches by Daniel Vetter.
The premise is that before every access to a buffer through the GTT we
have to declare whether we need a register or not. If the access is by
the GPU, a pipelined update to the register is made via the ringbuffer,
and we track the last seqno of the batches that access it. If by the
CPU we wait for the last GPU access and update the register (either
to clear or to set it for the current buffer).
One advantage of being able to pipeline changes is that we can defer the
actual updating of the fence register until we first need to access the
object through the GTT, i.e. we can eliminate the stall on set_tiling.
This is important as the userspace bo cache does not track the tiling
status of active buffers which generate frequent stalls on gen3 when
enabling tiling for an already bound buffer.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
... so that upon first use after resume we will reacquire the fence reg.
Reported-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
We don't track gpu flush request in any special way. So even with
obj->write_domain == 0, a gpu flush might be outstanding but no
yet executed. Even worse, the latest request might use the object
only for reading. So and unconditional call to object_wait_rendering
is needed for !pipelined.
Hence revert that patch fully and untangle the flushing from the
synchronization again.
Reported-by: Keith Packard <keithp@keithp.com>
Tested-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Besides the minimal improvement in reducing the execbuffer overhead, the
real benefit is clarifying a few routines.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
A number of dragons have been seen lurking within the execbuffer code.
The first step is then to isolate them from the rest and begin to
scrutinise them in depth. Suggested by Daniel Vetter.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Simply remove our accounting of objects inside the aperture, keeping
only track of what is in the aperture and its current usage. This
removes the over-complication of BUGs that were attempting to keep the
accounting correct and also removes the overhead of the accounting on
the hot-paths.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
With KMS, we can simply relinquish the fence when we idle the GPU and
reassign it upon first use.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Avoid evicting buffers that will be used later in the batch in order to
make room for the initial buffers by pinning all bound buffers in a
single pass before binding (and evicting for) fresh buffer.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
This used to check the precondition that all fences were to be located
in a mappable area, redundant now as those two parameters are combined
into one.
After pinning, we assert that the buffer is bound into the desired
region.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The pipe control object is allocated by the device for the sole use of the
render ringbuffer. Move this detail from the general code to the render
ring buffer initialisation.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Combining map_and_fenceable revealed a bug in
i915_gem_object_gtt_size() in that it always computed the appropriate
fence size for the object regardless of tiling state which caused us to
over-allocate linear buffers when binding to the GTT.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
This is required to restore gtt mappings on resume when agp is gone.
The right way to do this would be to make sturct drm_mm_node embeddable
and use the allocation list maintained by the drm memory manager. But
that's a bigger project. Getting rid of the per bo agp_mem will save
more memory than this wastes, anyway.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Currently if we hit a pagefault when applying a user relocation for the
execbuffer, we bail and return EFAULT to the application. Instead, we
need to unwind, drop the dev->struct_mutex, copy all the relocation
entries to a vmalloc array (to avoid any potential circular deadlocks
when resolving the pagefault), retake the mutex and then apply the
relocations. Afterwards, we need to again drop the lock and copy the
vmalloc array back to userspace.
v2: Incorporate feedback from Daniel Vetter.
Reported-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Commit 2549d6c2 removed the vmalloc used for temporary storage of the
relocation lists used during execbuffer. However, our use of vmalloc was
being protected by an integer overflow check which we do want to
preserve!
Reported-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Linus Torvalds found that it was rather trivial to trigger a system
freeze:
In fact, with lockdep, I don't even need to do the sysrq-d thing: it
shows the bug as it happens. It's the X server taking the same lock
recursively.
Here's the problem:
=============================================
[ INFO: possible recursive locking detected ]
2.6.37-rc2-00012-gbdbd01a #7
---------------------------------------------
Xorg/2816 is trying to acquire lock:
(&dev->struct_mutex){+.+.+.}, at: [<ffffffff812c626c>] i915_gem_fault+0x50/0x17e
but task is already holding lock:
(&dev->struct_mutex){+.+.+.}, at: [<ffffffff812c403b>] i915_mutex_lock_interruptible+0x28/0x4a
other info that might help us debug this:
2 locks held by Xorg/2816:
#0: (&dev->struct_mutex){+.+.+.}, at: [<ffffffff812c403b>] i915_mutex_lock_interruptible+0x28/0x4a
#1: (&mm->mmap_sem){++++++}, at: [<ffffffff81022d4f>] page_fault+0x156/0x37b
This recursion was introduced by rearranging the locking to avoid the
double locking on the fast path (4f27b5d and fbd5a26d) and the
introduction of the prefault to encourage the fast paths (b5e4f2b). In
order to undo the problem, we rearrange the code to perform the access
validation upfront, attempt to prefault and then fight for control of the
mutex. the best case scenario where the mutex is uncontended the
prefaulting is not wasted.
Reported-and-tested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
g33/pineview doesn't have any alignment constrains for unfenced tiled
buffers. But older chips have. Fix this.
Problem introduced in a00b10c360.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
An old and oft reported bug, is that of the GPU hanging on a
MI_WAIT_FOR_EVENT following a mode switch. The cause is that the GPU is
waiting on a scanline counter on an inactive pipe, and so waits for a
very long time until eventually the user reboots his machine.
We can prevent this either by moving the WAIT into the kernel and
thereby incurring considerable cost on every swapbuffers, or by waiting
for the GPU to retire the last batch that accesses the framebuffer
before installing a new one. As mode switches are much rarer than swap
buffers, this looks like an easy choice.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=28964
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=29252
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: stable@kernel.org
... and so prevent a potential circular reference:
[ INFO: possible circular locking dependency detected ]
2.6.37-rc1-uwe1+ #4
-------------------------------------------------------
Xorg/1401 is trying to acquire lock:
(&mm->mmap_sem){++++++}, at: [<c01e4ddb>] might_fault+0x4b/0xa0
but task is already holding lock:
(&dev->struct_mutex){+.+.+.}, at: [<f869c3ac>]
i915_mutex_lock_interruptible+0x3c/0x60 [i915]
which lock already depends on the new lock.
When the locking around the pwrite ioctl was simplified, I did not spot
that the phys path never took any locks and so we introduced this
potential circular reference.
Reported-by: Uwe Helm <uwe.helm@googlemail.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Instead of killing the process, just return no page found and reschedule
the process giving the GPU some time to (hopefully) recover.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
a00b10c360 "Only enforce fence limits inside the GTT" also
added a fenceable/mappable disdinction when binding/pinning buffers.
This only complicates the code with no pratical gain:
- In execbuffer this matters on for g33/pineview, as this is the only
chip that needs fences and has an unmappable gtt area. But fences
are only possible in the mappable part of the gtt, so need_fence
implies need_mappable. And need_mappable is only set independantly
with relocations which implies (for sane userspace) that the buffer
is untiled.
- The overlay code is only really used on i8xx, which doesn't have
unmappable gtt. And it doesn't support tiled buffers, currently.
- For all other buffers it's a bug to pass in a tiled bo.
In short, this disdinction doesn't have any practical gain.
I've also reverted mapping the overlay and context pages as possibly
unmappable. It's not worth being overtly clever here, all the big
gains from unmappable are for execbuf bos.
Also add a comment for a clever optimization that confused me
while reading the original patch by Chris Wilson.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
When merging Daniel's full-gtt patches I had a set of tweaks which I
thought I had undone. I was half right...
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=31286
Reported-by: jinjin.wang@intel.com
Reported-by: Alexey Fisher <bug-track@fisher-privat.net>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Take two passes to evict everything whilst searching for sufficient free
space to bind the batchbuffer. After searching for sufficient free space
using LRU eviction, evict everything that is purgeable and try again.
Only then if there is insufficient free space (or the GTT is too badly
fragmented) evict everything from the aperture and try one last time.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Accessing the uninitialised obj->pages instead of the local page lead to
an OOPs.
Reported-by: Xavier Chantry <chantry.xavier@gmail.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
So long as we adhere to the fence registers rules for alignment and no
overlaps (including with unfenced accesses to linear memory) and account
for the tiled access in our size allocation, we do not have to allocate
the full fenced region for the object. This allows us to fight the bloat
tiling imposed on pre-i965 chipsets and frees up RAM for real use. [Inside
the GTT we still suffer the additional alignment constraints, so it doesn't
magic allow us to render larger scenes without stalls -- we need the
expanded GTT and fence pipelining to overcome those...]
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Also spotted by Dan Carpenter.
obj->pin_count is unsigned so the BUG_ON(obj->pin_count<0) will never
trigger.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The error code is only expected during the actual pruning and not during
the first measurement (nr_to_scan == 0) pass.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
It is possible for the active list to only contain a read-only buffer so
that the ring->gpu_write_list remains entry. This leads to an
inconsistency between i915_gpu_is_active() and i915_gpu_idle() causing
an infinite spin during the shrinker and an assertion failure that
i915_gpu_idle() does indeed flush all buffers from the active lists.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
In order to force a page-fault on a GTT mapping after we start using it
from the GPU and so enforce correct CPU/GPU synchronisation, we need to
invalidate the mapping.
Pointed out by Owain G. Ainsworth.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
By using read_cache_page() for individual pages during pwrite/pread we
can eliminate an unnecessary large allocation (and immediate free) of
obj->pages. Also this eliminates any potential nesting of get/put pages,
simplifying the code and preparing the path for greater things.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Since we rarely use the mmap_offset and it is easily computable from the
obj->map_list.hash, remove it.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Eliminate the racy device unload by embedding a shrinker into each
device. Smaller, simpler code.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
More precisely: For those that _need_ to be mappable. Also add two
BUG_ONs in fault and pin to check the consistency of the mappable
flag.
Changes in v2:
- Add tracking of gtt mappable space (to notice mappable/unmappable
balancing issues).
- Improve the mappable working set tracking by tracking fault and pin
separately.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
This way we can make some more educated guesses as to why exactly
we can't use 2G apertures to their full potential ;)
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
At least the part that's currently enabled by the BIOS.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
In i915_gem_object_pin obviously unbind only if mappable is true.
This is the last part to enable gtt_mappable_end != gtt_size, which
the next patch will do.
v2: Fences on g33/pineview only work in the mappable part of the
gtt.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Like before add a parameter mappable (also to gem_object_pin) and
set it depending upon the context. Only bos that are brought into
the gtt due to an execbuffer call can be put into the unmappable
part of the gtt, everything else (especially pinned objects) need
to be put into the mappable part of the gtt.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Add a mappable parameter to i915_gem_evict_something to distinguish
the two cases (non-restricted vs. mappable gtt allocations). No
functional changes because the mappable limit is set to the end of
the gtt currently.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Preparing the ringbuffer for adding new commands can fail (a timeout
whilst waiting for the GPU to catch up and free some space). So check
for any potential error before overwriting HEAD with new commands, and
propagate that error back to the user where possible.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The ringbuffer keeps a pointer to the parent device, so we can use that
instead of passing around the pointer on the stack.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
* 'drm-core-next' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (476 commits)
vmwgfx: Implement a proper GMR eviction mechanism
drm/radeon/kms: fix r6xx/7xx 1D tiling CS checker v2
drm/radeon/kms: properly compute group_size on 6xx/7xx
drm/radeon/kms: fix 2D tile height alignment in the r600 CS checker
drm/radeon/kms/evergreen: set the clear state to the blit state
drm/radeon/kms: don't poll dac load detect.
gpu: Add Intel GMA500(Poulsbo) Stub Driver
drm/radeon/kms: MC vram map needs to be >= pci aperture size
drm/radeon/kms: implement display watermark support for evergreen
drm/radeon/kms/evergreen: add some additional safe regs v2
drm/radeon/r600: fix tiling issues in CS checker.
drm/i915: Move gpu_write_list to per-ring
drm/i915: Invalidate the to-ring, flush the old-ring when updating domains
drm/i915/ringbuffer: Write the value passed in to the tail register
agp/intel: Restore valid PTE bit for Sandybridge after bdd3072
drm/i915: Fix flushing regression from 9af90d19f
drm/i915/sdvo: Remove unused encoding member
i915: enable AVI infoframe for intel_hdmi.c [v4]
drm/i915: Fix current fb blocking for page flip
drm/i915: IS_IRONLAKE is synonymous with gen == 5
...
Fix up conflicts in
- drivers/gpu/drm/i915/{i915_gem.c, i915/intel_overlay.c}: due to the
new simplified stack-based kmap_atomic() interface
- drivers/gpu/drm/vmwgfx/vmwgfx_drv.c: added .llseek entry due to BKL
removal cleanups.
Keep the current interface but ignore the KM_type and use a stack based
approach.
The advantage is that we get rid of crappy code like:
#define __KM_PTE \
(in_nmi() ? KM_NMI_PTE : \
in_irq() ? KM_IRQ_PTE : \
KM_PTE0)
and in general can stop worrying about what context we're in and what kmap
slots might be appropriate for that.
The downside is that FRV kmap_atomic() gets more expensive.
For now we use a CPP trick suggested by Andrew:
#define kmap_atomic(page, args...) __kmap_atomic(page)
to avoid having to touch all kmap_atomic() users in a single patch.
[ not compiled on:
- mn10300: the arch doesn't actually build with highmem to begin with ]
[akpm@linux-foundation.org: coding-style fixes]
[akpm@linux-foundation.org: fix up drivers/gpu/drm/i915/intel_overlay.c]
Acked-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Chris Metcalf <cmetcalf@tilera.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: David Miller <davem@davemloft.net>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Dave Airlie <airlied@linux.ie>
Cc: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
... to prevent flush processing of an idle (or even absent) ring.
This fixes a regression during suspend from 87acb0a5.
Reported-and-tested-by: Alexey Fisher <bug-track@fisher-privat.net>
Tested-by: Peter Clifton <pcjc2@cam.ac.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
When the object has been written to by the gpu it remains on the ring
until its flush has been retired. However, when the object is moving to
the ring and the associated cache needs to be invalidated, we need to
perform the flush on the target ring, not the one it came from (which is
NULL in the reported case and so the flush was entirely absent).
Reported-by: Peter Clifton <pcjc2@cam.ac.uk>
Reported-and-tested-by: Alexey Fisher <bug-track@fisher-privat.net>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Whilst moving the code around in 9af90d19f, I dropped the or'ing in of
new write domains which would zero out the write domain for a render
target if later reused as a source later in the batch. This meant that
we might drop a required flush before reading from the render target.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=31043
Reported-by: xunx.fang@intel.com
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Based on an original patch by Zhenyu Wang, this initializes the BLT ring for
SandyBridge and enables support for user execbuffers.
Cc: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
If the userspace driver is using a constant relocation array with a
static buffer, they will pass the same relocation array back to the
kernel. So we *do* need to update the presumed offset value in those
relocations to reflect the current object so that they remain correct
with future batchbuffers and we avoid the necessity of having to suspend
execution and perform redundant relocations.
Fixes the regression introduced by 12f889c for applications using
absolute addressing on trees of buffer (i.e. the current consumers of
libdrm_intel.so).
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=30996
Reported-by: Wang, Jinjin <jinjin.wang@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
To handle retirements, we need per-ring tracking of active objects.
To handle evictions, we need global tracking of active objects.
As we enable more rings, rebuilding the global list from the individual
per-ring lists quickly grows tiresome and overly complicated. Tracking the
active objects in two lists is the lesser of two evils.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
... by always initialising the empty ringbuffer it is always then safe
to check whether it is active.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The most frequent relocation within a batchbuffer is a contiguous sequence
of vertex buffer relocations, for which we can virtually eliminate the
drm_gem_object_lookup() overhead by caching the last handle to object
translation.
In doing so we refactor the pin and relocate retry loop out of
do_execbuffer into its own helper function and so improve the error
paths.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
One of the primarily consumers of the i915 driver is X, a large signal
driven application. Frequently when writing into the buffers, there is a
pending signal which causes us not to take the interruptible lock but
then we need to take that same lock around the object unreference. By
rearranging the code to do the interruptible lock as the first check, we
can avoid the frequent additional locking around the unreference.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
... to avoid reacquiring it to drop the object reference count on
exit. Note we have to make sure we now drop (and reacquire) the lock
around acquiring the mm semaphore on the slow paths.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
After allocation a handle for the fresh object, we know that we can
safely drop the refcnt without triggering a free so we do not need the
mutex. Strangely, this mutex acquisition is the one that appears on
driver profiles.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Avoid an early eviction of the batch buffer into the uncached GTT
domain, and so do the relocation fixup in cacheable memory.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
... perform an access validation check up front instead and copy them in
on-demand, during i915_gem_object_pin_and_relocate(). As around 20% of
the CPU overhead may be spent inside vmalloc for the relocation entries
when submitting an execbuffer [for x11perf -aa10text], the savings are
considerable and result in around a 10% throughput increase [for glyphs].
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Currently, if a batch buffer refers to an object with a pending flip,
then we sleep until that pending flip is completed (unpinned and
signalled). This is so that a flip can be queued and the user can
continue rendering to the backbuffer oblivious to whether the buffer is
still pinned as the scan out. (The kernel arbitrating at the last moment
to stall the batch and wait until the buffer is unpinned and replaced as
the front buffer.)
As we only have a queue depth of 1, we can simply wait for the current
pending flip to complete and continue rendering. We can achieve this
with a single WAIT_FOR_EVENT command inserted into the ring buffer prior
to executing the batch, *without* stalling the client.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
* 'drm-intel-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/ickle/drm-intel:
drm/i915: Rephrase pwrite bounds checking to avoid any potential overflow
drm/i915: Sanity check pread/pwrite
drm/i915: Use pipe state to tell when pipe is off
drm/i915: vblank status not valid while training display port
drivers/gpu/drm/i915/i915_gem.c: Add missing error handling code
drm/i915: Fix refleak during eviction.
drm/i915: fix GMCH power reporting
Move the access control up from the fast paths, which are no longer
universally taken first, up into the caller. This then duplicates some
sanity checking along the slow paths, but is much simpler.
Tracked as CVE-2010-2962.
Reported-by: Kees Cook <kees@ubuntu.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: stable@kernel.org
Extend the error handling code with operations found in other nearby error
handling code
A simplified version of the sematic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
@r exists@
@r@
statement S1,S2,S3;
constant C1,C2,C3;
@@
*if (...)
{... S1 return -C1;}
...
*if (...)
{... when != S1
return -C2;}
...
*if (...)
{... S1 return -C3;}
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: stable@kernel.org
The return from move_to_gtt_domain() may indicate a pending signal which
needs to handled as opposed to an actual error, for instance, so report
the original return value rather than forcing an EINVAL.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
vmwgfx: Fix fb VRAM pinning failure due to fragmentation
vmwgfx: Remove initialisation of dev::devname
vmwgfx: Enable use of the vblank system
vmwgfx: vt-switch (master drop) fixes
drm/vmwgfx: Fix breakage introduced by commit "drm: block userspace under allocating buffer and having drivers overwrite it (v2)"
drm: Hold the mutex when dropping the last GEM reference (v2)
drm/gem: handlecount isn't really a kref so don't make it one.
drm: i810/i830: fix locked ioctl variant
drm/radeon/kms: add quirk for MSI K9A2GM motherboard
drm/radeon/kms: fix potential segfault in r600_ioctl_wait_idle
drm: Prune GEM vma entries
drm/radeon/kms: fix up encoder info messages for DFP6
drm/radeon: fix PCI ID 5657 to be an RV410
When the GPU is reset, the fence registers are invalidated, so release
the objects and clear them out.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Only drm/i915 does the bookkeeping that makes the information useful,
and the information maintained is driver specific, so move it out of the
core and into its single user.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Dave Airlie <airlied@redhat.com>
There were lots of places being inconsistent since handle count
looked like a kref but it really wasn't.
Fix this my just making handle count an atomic on the object,
and have it increase the normal object kref.
Now i915/radeon/nouveau drivers can drop the normal reference on
userspace object creation, and have the handle hold it.
This patch fixes a memory leak or corruption on unload, because
the driver had no way of knowing if a handle had been actually
added for this object, and the fbcon object needed to know this
to clean itself up properly.
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Dave Airlie <airlied@redhat.com>
At that point as the object is no longer in any GPU write domain it must
not be on the list, so the list_del() is redundant.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Just reschedule the retire requests again if the device is currently
busy. The request list will be pruned along other paths so will never
grow unbounded and so we can afford to miss the occasional pruning.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
With multiple rings generating requests independently, the outstanding
requests must also be track independently.
Reported-by: Wang Jinjin <jinjin.wang@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=30380
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Introduced by 48b956c5, I had thought I had already fixed this. Oh well.
Reported-by: Sitsofe Wheeler <sitsofe@yahoo.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Daniel Vetter pointed out that in this case is would be clearer and
cleaner to use a spinlock instead of a mutex to protect the per-file
request list manipulation. Make it so.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Owain Ainsworth reported an issue between the interaction of the
hangcheck and userspace immediately (and permanently) falling back to
s/w rasterisation. In order to break the mutex and begin resetting the
GPU, we must abort the current operation (usually within the wait) and
climb sufficiently far back up the call chain to drop the mutex. In his
implementation, Owain has a loop within the ioctl handler to detect the
hang and then sleep until the error handler has run. I've chosen to
return to userspace and report an EAGAIN which should trigger the
userspace ioctl handler to repeat the call (simply because it felt less
invasive...). Before hitting a wedged GPU, we then wait upon completion
of the error handler.
Reported-by: Owain G. Ainsworth <zerooa@googlemail.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Avoid cause latencies in other clients by not taking the global struct
mutex and moving the per-client request manipulation a local per-client
mutex. For example, this allows a compositor to schedule a page-flip
(through X) whilst an OpenGL application is monopolising the GPU.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
We need to drain the pending flips prior to disabling the pipe during
modeset, and these need to be done in an uninterruptible fashion.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
This is already performed with the pipelined flush, so by the time we
schedule the flush in the page-flip, the ring is NULL and we OOPs
instead.
Reported-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
A minor typo caused a single fence register to be incorrectly
programmed, resulting in occassional tiling corruption.
Reported-and-tested-by: Hans de Bruin <bruinjm@xs4all.nl>
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=18962
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: stable@kernel.org
Keep a list of pinned objects and display it via debugfs. Now all
objects that exist in the GTT are always tracked on one of the
active, flushing, inactive or pinned lists.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
If we have queued a page flip on the current fb and then request a mode
change, wait until the page flip completes before performing the new
request.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Track if the gpu requires the fence for the execution of a batch buffer
and so only wait upon the retirement of the object's last rendering
seqno if the fence is in use by the GPU.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Introduce intel_init_render_ring_buffer(), intel_init_bsd_ring_buffer
for ring initialization.
Signed-off-by: Xiang, Haihao <haihao.xiang@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Clear the GPU read domain for the inactive objects on a reset so that
they are correctly invalidated on reuse.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Owain Ainsworth noticed that the reset code failed to clear the flushing
list leaving the driver in an inconsistent state following a hung GPU.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
When flushing the GPU domains,we emit a flush on *both* rings, even
though they share a unified cache. Only emit the flush on the currently
active ring.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Change the semantics to retire any buffer older than the current seqno
rather than repeatedly calling calling the function to retire the
buffer at the head of the list matching the request seqno.
Whilst this should have no semantic impact on the implementation, Daniel
was wondering if there was a bug where we might miss a retirement and so
end up with a continually growing active list.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
With 5 places to update when adding handling for fence registers, it is
easy to overlook one or two. Correct that oversight, but fence
management should be improved before a new set of registers is added.
Bugzilla: https://bugs.freedesktop.org/show_bug?id=30199
Original patch by: Yuanhan Liu <yuanhan.liu@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: stable@kernel.org
As we currently may need to acquire a fence register during a modeset,
we need to be able to do so in an uninterruptible manner. So expose that
parameter to the callers of the fence management code.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
This ensures that we do wait upon the flushes to complete if necessary
and avoid the visual tears, whilst enabling pipelined page-flips.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
I pulled the wrong version of the patch from Daniel Vetter which was
missing the read barriers -- and the one that was causing all the trouble
was from i915_gem_object_put_fence_reg(), leading to GPU hangs on gen3.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
By reducing the hangcheck frequency we check less often, conserving
resources, and still detect a lock up quickly. On a fast machine with a
slow GPU (like a Core2 paired with a 945G) it is easy for the hangcheck to
misfire as we check too fast.
Also once hung and if we fail to completely reset the chip, we have a
nasty habit of proclaming a hang many times a second and generating a
strobe-like display.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
This spinlock only served debugging purposes in a time when we could not
be sure of the mutex ever being released upon a GPU hang. As we now
should be able rely on hangcheck to do the job for us (and that error
reporting should not itself require the struct mutex) we can kill the
incomplete attempt at protection.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
By allocating the request prior to writing to the ringbuffer, we can
abort the operation without leaving the GPU in an inconsistent state.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
... take advantage of the new implicit request issuing of
i915_wait_request.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
One caller (for the pageflip support) wants a purely pipelined flush.
Distinguish this case by a new parameter. This will also be useful
later on for pipelined fencing.
v2: Simplify the code by depending upon the implicit request emitting
of i915_wait_request.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
[ickle: And drop the non-interruptible support in the process.]
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
By moving one i915_add_request we can solely depend on the new
auto-seqno-numbering behaviour.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
i915_gem_object_move_to_active can handle zero seqno for us now.
And not emitting a request is not fatal here - we'll try to emit
a new one if we have to wait for some rendering to complete.
In case this assumption ever gets accidentally broken, there's already
a BUG_ON to catch it in i915_do_wait_request.
So just silently ignore ENOMEM here instead of screwing up the whole
drm.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
... instead of threading flush_domains through the execbuf code to
i915_add_request.
With this change 2 small cleanups are possible (likewise the majority
of the patch):
- The flush_domains parameter of i915_add_request is always 0. Drop it
and the corresponding logic.
- Ditto for the seqno param of i915_gem_process_flushing_list.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Previously I thought that one interrupt per batchbuffer should be
enough. Now tedious benchmarking showed this to be wrong.
Therefore track whether any commands have been isssued with a future
seqno (like pipelined fencing changes or flushes). If this is the case
emit a request before issueing the batchbuffer.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Now that we can move objects to the active list without already having
emitted a request, move the flushing list handling into i915_gem_flush.
This makes more sense and allows to drop a few i915_add_request calls
that are not strictly necessary.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Sometimes (like when flushing in preparation of batchbuffer execution)
we know that we'll emit a request but haven't yet done so. Allow this
case by simply taking the next seqno by default. Ensure that a request
is eventually emitted before waiting for an request by issuing it
in i915_wait_request iff this is not yet done.
Also replace one open-coded version of i915_gem_object_wait_rendering,
to prevent future code-diversion.
Chris Wilson asked me to explain and clarify what this patch does and why.
Here it goes:
Old way of moving objects onto the active list and associating them with a
reques:
1. i915_add_request + store the returned seqno somewhere
2. i915_gem_object_move_to_active (with the stored seqno as parameter)
For the current users, this is all fine. But I'd like to associate objects
(and fence regs) with the batchbuffer request deep down in the execbuf
call-chain. I thought about three ways of implementing this.
a) Don't care, just emit request when we need a new seqno. When heavily
pipelining fence reg changes, this would have caused tons of superflous
request (and corresponding irqs).
b) Thread all changed fences, objects, whatever through the execbuf-maze,
so that when we emit a request, we can store the new seqno at all the right
places.
c) Kill that seqno-threading-around business by simply storing the next
seqno, i.e. allow 2. to be done before 1. in the above sequence.
I've decided to implement c) (in this patch). The following patches are
just fall-out that resulted from this small conceptual change.
* We can handle the flushing list processing where we actually emit a flush
(i915_gem_flush and i915_retire_commands) instead of in i915_add_request.
The code makes IMHO more sense this way (and i915_add_request looses the
flush_domains parameter, obviously).
* We can avoid emitting unnecessary requests. IMHO there's no point in
emitting more than one request per batchbuffer (with or without an
corresponding irq).
* By enforcing 2. before 1. ordering in the above sequence the seqno
argument of i915_gem_object_move_to_active is redundant and can be
dropped.
v2: Now i915_wait_request issues request if it is not yet emitted.
Also introduce i915_gem_next_request_seqno(dev) just in case we ever
need to do some prep work before using a new seqno.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
[ickle: Keep i915_gem_object_set_to_display_plane() uninterruptible.]
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
ums-gem code correctly cancels the retire work (at lastclose time),
kms does not do so. Fix this by canceling the work right after ideling
the gpu.
While staring at the code I noticed that the work function is not
static. Fix this, too.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
This is the first patch to clean up module unload races due to
outstanding timers/work. Preparatory step: Thou shalt not destroy
the workqueue when new work might still get enqued.
Now error_work gets queued by the hangcheck timer and only (atomically)
reads the chip wedged status. So cancel it right after the hangcheck
timer is killed. But the hangcheck is armed by interrupts, so move
everything after irqs are disabled.
Also change a del_timer to a del_timer_sync in the ums gem code, the
hangcheck timer is self-rearming.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Sandybridge GTT has new cache control bits in PTE, which controls
graphics page cache in LLC or LLC/MLC, so we need to extend the mask
function to respect the new bits.
And set cache control to always LLC only by default on Gen6.
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Cc: stable@kernel.org
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
copy_to_user() returns the number of bytes remaining to be copied and
I'm pretty sure we want to return a negative error code here.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: stable@kernel.org
This reverts commit 86f100b136.
The kref API requires the handlecount to be initialised to one on object
creation (so that kref_get() doesn't complain upon first use) so the
dalliance in the drivers is required in order to sink the initial
floating reference.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: stable@kernel.org
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/anholt/drm-intel: (58 commits)
drm/i915,intel_agp: Add support for Sandybridge D0
drm/i915: fix render pipe control notify on sandybridge
agp/intel: set 40-bit dma mask on Sandybridge
drm/i915: Remove the conflicting BUG_ON()
drm/i915/suspend: s/IS_IRONLAKE/HAS_PCH_SPLIT/
drm/i915/suspend: Flush register writes before busy-waiting.
i915: disable DAC on Ironlake also when doing CRT load detection.
drm/i915: wait for actual vblank, not just 20ms
drm/i915: make sure eDP PLL is enabled at the right time
drm/i915: fix VGA plane disable for Ironlake+
drm/i915: eDP mode set sequence corrections
drm/i915: add panel reset workaround
drm/i915: Enable RC6 on Ironlake.
drm/i915/sdvo: Only set is_lvds if we have a valid fixed mode.
drm/i915: Set up a render context on Ironlake
drm/i915 invalidate indirect state pointers at end of ring exec
drm/i915: Wake-up wait_request() from elapsed hang-check (v2)
drm/i915: Apply i830 errata for cursor alignment
drm/i915: Only update i845/i865 CURBASE when disabled (v2)
drm/i915: FBC is updated within set_base() so remove second call in mode_set()
...
We now attempt to free "active" objects following a GPU hang as either
the GPU will be reset or the hang is permenant. In either case, the GPU
writes will not be flushed to main memory and it should be safe to
return that memory back to the system.
The BUG_ON(active) is thus overkill and can erroneously fire after a
EIO.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
This is consistent with trying to access a filename that not exist
within a directory which is a good analogy here. The main reason for the
change is that it is easy to confuse the error code of EBADF as an
performing an ioctl on an invalid file descriptor (rather than an
unknown object).
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Dave Airlie <airlied@redhat.com>
i830 requires 32bpp cursors to be aligned to 16KB, so we have to expose
the alignment parameter to i915_gem_attach_phys_object().
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
shmfs doesn't actually implement i_ops->truncate() so we were not
immedatiately releasing the backing pages when shrinking the gfx cache
under OOM. Instead use a combination of truncate_inode_pages() and
i_ops->truncate_range() as is used by shmem_delete_inode().
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
In order to reduce the penalty of fallbacks under memory pressure and to
avoid a potential immediate ping-pong of evicting a mmaped buffer, we
move the object to the tail of the inactive list when a page is freshly
faulted or the object is moved into the CPU domain.
We choose not to protect the CPU objects from casual eviction,
preferring to keep the GPU active for as long as possible.
v2: Daniel Vetter found a bug where I forgot that pinned objects are
kept off the inactive list.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
The eviction code is the gnarly underbelly of memory management, and is
clearer if kept separated from the normal domain management in GEM.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
This will be used by the eviction logic to maintain fairness between the
rings.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
This does two little changes:
- Add an alignment parameter for evict_something. It's not really great to
whack a carefully sized hole into the gtt with the wrong alignment.
Especially since the fallback path is a full evict.
- With the inactive scan stuff we need to evict more that one object, so
move the unbind call into the helper function that scans for the object
to be evicted, too. And adjust its name.
No functional changes in this patch, just preparation.
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
In order to properly track bound objects, they need to exist on one of
the inactive/active lists or be pinned. As this is a requirement, do the
work inside i915_gem_bind_to_gtt() rather than dotted around the
callsites.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
This debugging trace was useful for finding the fbcon regression on
i965, and it may prove useful again in future.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Incorporates a similar patch by Daniel Vetter, the alteration being to
report the current busy state after retiring.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
This avoids the excess flush and requests on idle rings (and spamming
the debug log ;-)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
This is required should we ever attempt to use an io-mapping where
KM_USER0 is verboten, such as inside an IRQ context.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
When creating an object, we create the handle by which it is known to
the process and which own the reference to the object. That reference to
the new handle is what we want to transfer to the process, not the lost
reference to the object; so free the local object reference *not* the
process's handle reference.
This brings i915_gem_object_create_ioctl() into line with
drm_gem_open_ioctl()
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
If we fail to flush outstanding GPU writes but return the memory to the
system, we risk corrupting memory should the GPU recovery and complete
those writes. On the other hand, if we bail early and free the object
then we have a definite use-after-free and real memory corruption.
Choose the lesser of two evils, since in order to recover from the hung
GPU we need to completely reset it, those pending writes should
never happen.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
If during the freeing of an object the unbind is interrupted by a system
call, which is quite possible if we have outstanding GPU writes that
must be flushed, the unbind is silently aborted. This still leaves the
AGP region and backing pages allocated, and perhaps more importantly,
the object remains upon the various lists exposing us to memory
corruption.
I think this is the cause behind the use-after-free, such as
Bug 15664 - Graphics hang and kernel backtrace when starting Azureus
with Compiz enabled
https://bugzilla.kernel.org/show_bug.cgi?id=15664
v2: Daniel Vetter reminded me that kernel space programming is never easy.
We cannot simply spin to clear the pending signal and so must deferred
the freeing of the object until later.
v3: Run from the top level retire requests.
v4: Tested with P(return -ERESTARTSYS)=.5 from i915_gem_do_wait_request()
v5: Rebase against Eric's for-linus tree.
v6: Refactor, split and add a comment about avoiding unbounded recursion.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
Combine the iteration over active render rings into a common function.
This is in preparation for reusing the idle function to also retire
deferred free requests.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Simple fix for error propagation along the old UMS path.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
drm/r600: fix possible NULL pointer derefernce
drm/radeon/kms: add quirk for ASUS HD 3600 board
include/linux/vgaarb.h: add missing part of include guard
drm/nouveau: Fix crashes during fbcon init on single head cards.
drm/nouveau: fix pcirom vbios shadow breakage from acpi rom patch
drm/radeon/kms: fix shared ddc harder
drm/i915: enable low power render writes on GEN3 hardware.
drm/i915: Define MI_ARB_STATE bits
vmwgfx: return -EFAULT if copy_to_user fails
fb: handle allocation failure in alloc_apertures()
drm: radeon: check kzalloc() result
drm/ttm: Fix build on architectures without AGP
drm/radeon/kms: fix gtt MC base alignment on rs4xx/rs690/rs740 asics
drm/radeon/kms: fix possible mis-detection of sideport on rs690/rs740
drm/radeon/kms: fix legacy tv-out pal mode
A lot of 945GMs have had stability issues for a long time, this manifested as X hangs, blitter engine hangs, and lots of crashes.
one such report is at:
https://bugs.freedesktop.org/show_bug.cgi?id=20560
along with numerous distro bugzillas.
This only took a week of digging and hair ripping to figure out.
Tracked down and tested on a 945GM Lenovo T60,
previously running
x11perf -copypixwin500
or
x11perf -copywinpix500
repeatedly would cause the GPU to wedge within 4 or 5 tries, with random busy bits set.
After this patch no hangs were observed.
cc: stable@kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
The current shrinker implementation requires the registered callback
to have global state to work from. This makes it difficult to shrink
caches that are not global (e.g. per-filesystem caches). Pass the shrinker
structure to the callback so that users can embed the shrinker structure
in the context the shrinker needs to operate on and get back to it in the
callback via container_of().
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
The hibernate issues that got fixed in commit 985b823b91 ("drm/i915:
fix hibernation since i915 self-reclaim fixes") turn out to have been
incomplete. Vefa Bicakci tested lots of hibernate cycles, and without
the __GFP_RECLAIMABLE flag the system eventually fails to resume.
With the flag added, Vefa can apparently hibernate forever (or until he
gets bored running his automated scripts, whichever comes first).
The reclaimable flag was there originally, and was one of the flags that
were dropped (unintentionally) by commit 4bdadb9785 ("drm/i915:
Selectively enable self-reclaim") that introduced all these problems,
but I didn't want to just blindly add back all the flags in commit
985b823b91, and it looked like __GFP_RECLAIM wasn't necessary. It
clearly was.
I still suspect that there is some subtle reason we're missing that
causes the problems, but __GFP_RECLAIMABLE is certainly not wrong to use
in this context, and is what the code historically used. And we have no
idea what the causes the corruption without it.
Reported-and-tested-by: M. Vefa Bicakci <bicave@superonline.com>
Cc: Dave Airlie <airlied@gmail.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Only ever assigned, never used.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
[glisse: I will re-add if needed for range-restricted allocations]
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Since commit 4bdadb9785 ("drm/i915:
Selectively enable self-reclaim"), we've been passing GFP_MOVABLE to the
i915 page allocator where we weren't before due to some over-eager
removal of the page mapping gfp_flags games the code used to play.
This caused hibernate on Intel hardware to result in a lot of memory
corruptions on resume. See for example
http://bugzilla.kernel.org/show_bug.cgi?id=13811
Reported-by: Evengi Golov (in bugzilla)
Signed-off-by: Dave Airlie <airlied@redhat.com>
Tested-by: M. Vefa Bicakci <bicave@superonline.com>
Cc: stable@kernel.org
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Since we now get_user_pages() outside of the mutex prior to performing
the copy, we kmap() the page inside the copy routine and so need to
perform an ordinary memcpy() and not copy_from_user().
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
As we do not have a requirement to be atomic and avoid sleeping whilst
performing the slow copy for shmem based pread and pwrite, we can use
kmap instead, thus simplifying the code.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
We can avoid an early clflush when pwriting if we use the current CPU
write domain rather than moving the object to the GTT domain for the
purposes of the pwrite. This has the advantage of not flushing the
presumably hot data that we want to upload into the bo, and of ascribing
the clflush to the execution when profiling.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
The callers expect us to cleanup any partially initialised structures
before reporting the error.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
If the object is bigger than the entire aperture, reject it early
before evicting everything in a vain attempt to find space.
v2: Use E2BIG as suggested by Owain G. Ainsworth.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: stable@kernel.org
Signed-off-by: Eric Anholt <eric@anholt.net>
This particular warning is harmless as we emit during the normal
pinning process where the batch buffer requires more fences than is
available without eviction. Only if we fail to evict enough fences does
this become a problem, so include the requested number of fences in the
ultimate *error* message.
v2: Remember to compile test even trial patches to remove warnings.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Nesting domain changes will cause confusion when trying to interpret the
tracepoints describing the sequence of changes for the object, as well
as obscuring the order of operations for the reader of the code.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
This saves a whooping 7 dwords. Zero functional changes. Because
some of the refcounts are rather tightly calculated, I've put
BUG_ONs in the code to check for overflows.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
The BSD (bit stream decoder) ring is used for accessing the BSD engine
which decodes video bitstream for H.264 and VC1 on G45+. It is
asynchronous with the render ring and has access to separate parts of
the GPU from it, though the render cache is coherent between the two.
Signed-off-by: Zou Nan hai <nanhai.zou@intel.com>
Signed-off-by: Xiang Hai hao <haihao.xiang@intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
The active list and request list move into the ringbuffer structure,
so each can track its active objects in the order they are in that
ring. The flushing list does not, as it doesn't matter which ring
caused data to end up in the render cache. Objects gain a pointer to
the ring they are active on (if any).
Signed-off-by: Zou Nan hai <nanhai.zou@intel.com>
Signed-off-by: Xiang Hai hao <haihao.xiang@intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Introduces a more complete intel_ring_buffer structure with callbacks
for setup and management of a particular ringbuffer, and converts the
render ring buffer consumers to use it.
Signed-off-by: Zou Nan hai <nanhai.zou@intel.com>
Signed-off-by: Xiang Hai hao <haihao.xiang@intel.com>
[anholt: Fixed up whitespace fail and rebased against prep patches]
Signed-off-by: Eric Anholt <eric@anholt.net>
This is preparation for supporting multiple ringbuffers on Ironlake.
The non-copy-and-paste changes are:
- de-staticing functions
- I915_GEM_GPU_DOMAINS moving to i915_drv.h to be used by both files.
- i915_gem_add_request had only half its implementation
copy-and-pasted out of the middle of it.
This lru tracks fences, not objects, so move it to where it belongs.
As a side effect, this nicely shrinks drm_i915_gem_object by two
pointers.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
Conflicts:
drivers/gpu/drm/i915/i915_dma.c
drivers/gpu/drm/i915/i915_drv.h
drivers/gpu/drm/radeon/r300.c
The BSD ringbuffer support that is landing in this branch
significantly conflicts with the Ironlake PIPE_CONTROL fix on master,
and requires it to be tested successfully anyway.
By idling the GPU and discarding everything we can when under extreme
memory pressure, the number of OOM-killer events is dramatically
reduced. For instance, this makes it possible to run
firefox-planet-gnome.trace again on my swapless 512MiB i915.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
My PIPE_CONTROL fix (just sent via Eric's tree) was buggy; I was
testing a whole set of patches together and missed a conversion to the
new HAS_PIPE_CONTROL macro, which will cause breakage on non-Ironlake
965 class chips. Fortunately, the fix is trivial and has been tested.
Be sure to use the HAS_PIPE_CONTROL macro in i915_get_gem_seqno, or
we'll end up reading the wrong graphics memory, likely causing hangs,
crashes, or worse.
Reported-by: Zdenek Kabelac <zdenek.kabelac@gmail.com>
Reported-by: Toralf Förster <toralf.foerster@gmx.de>
Tested-by: Toralf Förster <toralf.foerster@gmx.de>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Since 965, the hardware has supported the PIPE_CONTROL command, which
provides fine grained GPU cache flushing control. On recent chipsets,
this instruction is required for reliable interrupt and sequence number
reporting in the driver.
So add support for this instruction, including workarounds, on Ironlake
and Sandy Bridge hardware.
https://bugs.freedesktop.org/show_bug.cgi?id=27108
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Luckily the change is quite a little bit less invasive than I've
feared.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Thanks to the to_intel_bo helper, this change is rather trivial.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Just embed it and adjust the pointers, No other changes (that's
for later patches).
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Just preparation, no functional change.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
When drivers embed the core gem object into their own structures,
they'll have to do this. Temporarily this results in an ugly
kfree(gem_obj);
in every gem driver.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Current code is definitely crap: Largest pitch allowed spills into
the TILING_Y bit of the fence registers ... :(
I've rewritten the limits check under the assumption that 3rd gen hw
has a 3d pitch limit of 8kb (like 2nd gen). This is supported by an
otherwise totally misleading XXX comment.
This bug mostly resulted in tiling-corrupted pixmaps because the kernel
allowed too wide buffers to be tiled. Bug brought to the light by the
xf86-video-intel 2.11 release because that unconditionally enabled
tiling for pixmaps, relying on the kernel to check things. Tiling for
the framebuffer was not affected because the ddx does some additional
checks there ensure the buffer is within hw-limits.
v2: Instead of computing the value that would be written into the
hw fence registers and then checking the limits simply check whether
the stride is above the 8kb limit. To better document the hw, add
some WARN_ONs in i915_write_fence_reg like I've done for the i830
case (using the right limits).
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=27449
Tested-by: Alexander Lam <lambchop468@gmail.com>
Cc: stable@kernel.org
Signed-off-by: Eric Anholt <eric@anholt.net>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/anholt/drm-intel:
drm/i915: Ignore LVDS EDID when it is unavailabe or invalid
drm/i915: Add no_lvds entry for the Clientron U800
drm/i915: Rename many remaining uses of "output" to encoder or connector.
drm/i915: Rename intel_output to intel_encoder.
agp/intel: intel_845_driver is an agp driver!
drm/i915: introduce to_intel_bo helper
drm/i915: Disable FBC on 915GM and 945GM.
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
This is a purely cosmetic change to make changes in this area easier.
And hey, it's not only clearer and typechecked, but actually shorter,
too!
[anholt: To clarify, this is a change to let us later make
drm_i915_gem_object subclass drm_gem_object, instead of having
drm_gem_object have a pointer to i915's private data]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Dave Airlie <airlied@gmail.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
This could resolve HW deadlocks where a unit downstream of the VS is
waiting for more input, the VS has one vertex queued up but not
dispatched because it hopes to get one more vertex for 2x4 dispatch,
and software isn't handing more vertices down because it's waiting for
rendering to complete. The B-Spec says you should always have this
bit set.
Signed-off-by: Eric Anholt <eric@anholt.net>
The continue just after this call with loop around and wait for the
request just added just fine. This leads to slightly more compact code.
Signed-Off-by: Owain G. Ainsworth <oga@openbsd.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
The assumption that an object has only ever one write domain is deeply
threaded into gem (it's even encoded the the singular of the variable
name). Don't let userspace screw us over.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
Now that we have an exact gpu write domain tracking, we don't need
to move objects to the active list ourself. i915_add_request will
take care of that under all circumstances.
Idea stolen from a patch by Chris Wilson <chris@chris-wilson.co.uk>.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
We have it, so use it. This required moving the function to avoid
a forward declaration.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
The fence_list should be lru ordered for otherwise we might try
to steal a fence reg from an active object even though there are
fences from inactive objects available. lru ordering was obeyed
for gpu access everywhere save when moving dirty objects from
flushing_list to active_list.
Fixing this cause the code to indent way to much, so I've extracted
the flushing_list processing logic into its on function.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
The spaghetti logic in there tripped up my brain's code parser for a
few secs. Prevent this from happening again by extracting the fence
stealing code into a seperate functions. IMHO this slightly clears up
the code flow.
v2: Beautified according to ickle's comments.
v3: ickle forgot to flush his comment queue ... Now there's also a
we-are-paranoid BUG_ON in there.
v4: I've forgotten to switch on my brain when doing v3. Now the BUG_ON
actually checks something useful.
v5: Clean up a stale comment as noted by Eric Anholt.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
All other accesses take this spinlock, so do this here, too.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
This has a few functional changes against the old code:
* a few more unnecessary loads and stores to the drm_i915_fence_reg
objects. Also an unnecessary store to the hw fence register.
* zaps any userspace mappings before doing other flushes. Only changes
anything when userspace does racy stuff against itself.
* also flush GTT domain. This is a noop, but still try to keep the
bookkeeping correct.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
* anholt/drm-intel-next:
drm/i915: Record batch buffer following GPU error
drm/i915: give up on 8xx lid status
drm/i915: reduce some of the duplication of tiling checking
drm/i915: blow away userspace mappings before fence change
drm/i915: move a gtt flush to the correct place
agp/intel: official names for Pineview and Ironlake
drm/i915: overlay: drop superflous gpu flushes
drm/i915: overlay: nuke readback to flush wc caches
drm/i915: provide self-refresh status in debugfs
drm/i915: provide FBC status in debugfs
drm/i915: fix drps disable so unload & re-load works
drm/i915: Fix OGLC performance regression on 945
drm/i915: Deobfuscate the render p-state obfuscation
drm/i915: add dynamic performance control support for Ironlake
drm/i915: enable memory self refresh on 9xx
drm/i915: Don't reserve compatibility fence regs in KMS mode.
drm/i915: Keep MCHBAR always enabled
drm/i915: Replace open-coded eviction in i915_gem_idle()
i915_gem_object_fenceable was mostly just a repeat of the
i915_gem_object_fence_offset_ok, but also checking the size (which was
checkecd when we allowed that BO to be tiled in the first place). So
instead, export the latter function and use it in place.
Signed-Off-By: Owain G. Ainsworth <oga@openbsd.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
This aligns it with the other user of i915_gem_clear_fence_reg,
which blows away the mapping before changing the fence reg.
Only affects userspace if it races against itself when changing
tiling parameters, i.e. behaviour is undefined, anyway.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
No functional change, because gtt flushing is a no-op. Still, try
to keep the bookkeeping accurate. The if is still slightly wrong
for with execbuf2 even i915-class hw doesn't always need a fence
reg for gpu access. But that's for somewhen lateron.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
The fence start is for compatibility with UMS X Servers before fence
management. KMS X Servers only started doing tiling after fence
management appeared.
Signed-off-by: Eric Anholt <eric@anholt.net>
With the introduction of the hang-check, we can safely expect that
i915_wait_request() will always return even when the GPU hangs, and so
do not need to open code the wait in order to manually check for the
hang. Also we do not need to always evict all buffers, so only flush
the GPU (and wait for it to idle) for KMS, but continue to evict for UMS.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Mostly obvious simplifications.
The i915 pread/pwrite ioctls, intel_overlay_put_image and
nouveau_gem_new were incorrectly using the locked versions
without locking: this is also fixed in this patch.
Signed-off-by: Luca Barbieri <luca@luca-barbieri.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Before changing the status of a buffer with a pending write we will await
upon a new flush for that buffer. So we can take advantage of any flushes
posted whilst the buffer is active and pending processing by the GPU, by
clearing its write_domain and updating its last_rendering_seqno -- thus
saving a potential flush in deep queues and improves flushing behaviour
upon eviction for both GTT space and fences.
In order to reduce the time spent searching the active list for matching
write_domains, we move those to a separate list whose elements are
the buffers belong to the active/flushing list with pending writes.
Orignal patch by Chris Wilson <chris@chris-wilson.co.uk>, forward-ported
by me.
In addition to better performance, this also fixes a real bug. Before
this changes, i915_gem_evict_everything didn't work as advertised. When
the gpu was actually busy and processing request, the flush and subsequent
wait would not move active and dirty buffers to the inactive list, but
just to the flushing list. Which triggered the BUG_ON at the end of this
function. With the more tight dirty buffer tracking, all currently busy and
dirty buffers get moved to the inactive list by one i915_gem_flush operation.
I've left the BUG_ON I've used to prove this in there.
References:
Bug 25911 - 2.10.0 causes kernel oops and system hangs
http://bugs.freedesktop.org/show_bug.cgi?id=25911
Bug 26101 - [i915] xf86-video-intel 2.10.0 (and git) triggers kernel oops
within seconds after login
http://bugs.freedesktop.org/show_bug.cgi?id=26101
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Adam Lantos <hege@playma.org>
Cc: stable@kernel.org
Signed-off-by: Eric Anholt <eric@anholt.net>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/anholt/drm-intel:
drm/i915: Fix leak of relocs along do_execbuffer error path
drm/i915: slow acpi_lid_open() causes flickering - V2
drm/i915: Disable SR when more than one pipe is enabled
drm/i915: page flip support for Ironlake
drm/i915: Fix the incorrect DMI string for Samsung SX20S laptop
drm/i915: Add support for SDVO composite TV
drm/i915: don't trigger ironlake vblank interrupt at irq install
drm/i915: handle non-flip pending case when unpinning the scanout buffer
drm/i915: Fix the device info of Pineview
drm/i915: enable vblank interrupt on ironlake
drm/i915: Prevent use of uninitialized pointers along error path.
drm/i915: disable hotplug detect before Ironlake CRT detect
Following a gpu hang, we would leak the relocation buffer. So simply
earrange the error path to always free the relocation buffer.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Having missed the ENOMEM return via i915_gem_fault(), there are probably
other paths that I also missed. By not enabling NORETRY by default these
paths can run the shrinker and take memory from the system (but not from
our own inactive lists because our shrinker can not run whilst we hold
the struct mutex) and this may allow the system to survive a little longer
whilst our drivers consume all available memory.
References:
OOM killer unexpectedly called with kernel 2.6.32
http://bugzilla.kernel.org/show_bug.cgi?id=14933
v2: Pass gfp into page mapping.
v3: Use new read_cache_page_gfp() instead of open-coding.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: Eric Anholt <eric@anholt.net>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Suggested-by: Chris Wilson <chris@chris-wilson.co.uk>>
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org> (in principle)
Signed-off-by: Eric Anholt <eric@anholt.net>
When we setup buffer for display plane, we'll check any pending
required GPU flush and possible make interruptible wait for flush
complete. But that wait would be most possibly to fail in case of
signals received for X process, which will then fail modeset process
and put display engine in unconsistent state. The result could be
blank screen or CPU hang, and DDX driver would always turn on outputs
DPMS after whatever modeset fails or not.
So this one creates new helper for setup display plane buffer, and
when needing flush using uninterruptible wait for that.
This one should fix bug like https://bugs.freedesktop.org/show_bug.cgi?id=24009.
Also fixing mode switch stress test on Ironlake.
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
* 'drm-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
drm/radeon/kms: rs600: use correct mask for SW interrupt
gpu/drm/radeon/radeon_irq.c: move a dereference below a NULL test
drm/radeon/radeon_device.c: move a dereference below a NULL test
drm/radeon/radeon_fence.c: move a dereference below the NULL test
drm/radeon/radeon_connectors.c: add a NULL test before dereference
drm/radeon/kms: fix memory leak
drm/kms: Fix &&/|| confusion in drm_fb_helper_connector_parse_command_line()
drm/edid: Fix CVT width/height decode
drm/edid: Skip empty CVT codepoints
drm: remove address mask param for drm_pci_alloc()
drm/radeon/kms: add missing breaks in i2c and ss lookups
drm/radeon/kms: add primary dac adj values table
drm/radeon/kms: fallback to default connector table
drm_pci_alloc() has input of address mask for setting pci dma
mask on the device, which should be properly setup by drm driver.
And leave it as a param for drm_pci_alloc() would cause confusion
or mistake would corrupt the correct dma mask setting, as seen on
intel hw which set wrong dma mask for hw status page. So remove
it from drm_pci_alloc() function.
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
As pinning (allocating and binding GTT memory) does not actually invoke
GPU commands, it is safe, and indeed is attempted, during resumption
from suspension:
[drm:intel_init_clock_gating] *ERROR* failed to pin power context: -16
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reported-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: stable@kernel.org
Signed-off-by: Eric Anholt <eric@anholt.net>
This patch adds a new execbuf ioctl, execbuf2, for use by clients that
want to control fence register allocation more finely. The buffer
passed in to the new ioctl includes a new relocation type to indicate
whether a given object needs a fence register assigned for the command
buffer in question.
Compatibility with the existing execbuf ioctl is implemented in terms
of the new code, preserving the assumption that fence registers are
required for pre-965 rendering commands.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[ickle: Remove pre-emptive clear_fence_reg()]
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Kristian Høgsberg <krh@bitplanet.net>
[anholt: Removed dmesg spam]
Signed-off-by: Eric Anholt <eric@anholt.net>
i915_gem_object_unbind had the ordering wrong. The other user,
i915_gem_object_put_fence_reg already has the correct ordering.
Results was usually corrupted pixmaps, especially garbled font glyphs
after a suspend/resume (because this evicts everything).
I'm still waiting for the feedback from the bug-reporters, but
because this obviously fixes a bug (at least for me) I'm already
submitting it.
Bugzilla: http://bugs.freedesktop.org/show_bug.cgi?id=25406
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
CC: stable@kernel.org
* 'drm-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (189 commits)
drm/radeon/kms: fix warning about cur_placement being uninitialised.
drm/ttm: Print debug information on memory manager when eviction fails
drm: Add memory manager debug function
drm/radeon/kms: restore surface registers on resume.
drm/radeon/kms/r600/r700: fallback gracefully on ucode failure
drm/ttm: Initialize eviction placement in case the driver callback doesn't
drm/radeon/kms: cleanup structure and module if initialization fails
drm/radeon/kms: actualy set the eviction placements we choose
drm/radeon/kms: Fix NULL ptr dereference
drm/radeon/kms/avivo: add support for new pll selection algo
drm/radeon/kms/avivo: fix some bugs in the display bandwidth setup
drm/radeon/kms: fix return value from fence function.
drm/radeon: Remove tests for -ERESTART from the TTM code.
drm/ttm: Have the TTM code return -ERESTARTSYS instead of -ERESTART.
drm/radeon/kms: Convert radeon to new TTM validation API (V2)
drm/ttm: Rework validation & memory space allocation (V3)
drm: Add search/get functions to get a block in a specific range
drm/radeon/kms: fix avivo tiling regression since radeon object rework
drm/i915: Remove a debugging printk from hangcheck
drm/radeon/kms: make sure i2c id matches
...
IGD* isn't a useful name. Replace with the codenames, as sourced from
pci.ids.
Signed-off-by: Adam Jackson <ajax@redhat.com>
[anholt: Fixed up for merge with pineview/ironlake changes]
Signed-off-by: Eric Anholt <eric@anholt.net>
These are handled by the error return being propagated to user-space and
do not any add any information to the original error, so are useless.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Acked-by: Jakob Bornecrantz <jakob@vmware.com>
Acked-by: Thomas Hellström <thomas@shipmail.org>
Review-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jesse "Orange Smoothie" Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Kristian Høgsberg <krh@bitplanet.net>
Signed-off-by: Eric Anholt <eric@anholt.net>
Execbufs involve quite a bit of payload, to the extent that cache misses
show up in the profiles here, and a suspicion that some of those cachelines
may get evicted and then reloaded in the subsequent copy.
This is still abstracted like drm_calloc_large since we want to check for
size overflow, and because we want to choose between kmalloc and vmalloc
on the fly. cairo's interface for malloc-with-calloc's-args was used as
the model.
Signed-off-by: Eric Anholt <eric@anholt.net>
Replace the DRM_DEBUG with DRM_DEBUG_DRIVER in generic i915 driver.
Then the debug info can be obtained by adding the boot option of
"drm.debug=0x02".
At the same time the debug info in increase/decrease clock is also
printed by using DRM_DEBUG_DRIVER instead of DRM_DEBUG_KMS.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
As long as the gpu can keep up, neither the cpu (waiting for gpu)
nore the gpu (waiting for vblank to do an overlay flip) stalls.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
This just waits until the hw passed the current ring position with
cmd execution. This slightly changes the existing i915_wait_request
function to make uninterruptible waiting possible - no point in
returning to userspace while mucking around with the overlay, that
piece of hw is just too fragile.
Also replace a magic 0 with the symbolic constant (and kill the then
superflous comment) while I was looking at the code.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
If we trigger a tracepoint for batch buffer submission, it is a reasonable
assumption that we wish to also trace the batch buffer completion. So in
order to capture the completion events, we need to enable irqs... However,
we cannot rely on the completion event to disable the irq later, so we
defer the irq disable to the retire request.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
We increment the seqno number between submitting the batch buffer and
the flush/interrupt that demarcates its end, so the tracepoint needs to
reference the incremented value to match the completion event.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
* 'drm-intel-next' of git://git.kernel.org/pub/scm/linux/kernel/git/anholt/drm-intel: (57 commits)
drm/i915: Handle ERESTARTSYS during page fault
drm/i915: Warn before mmaping a purgeable buffer.
drm/i915: Track purged state.
drm/i915: Remove eviction debug spam
drm/i915: Immediately discard any backing storage for uneeded objects
drm/i915: Do not mis-classify clean objects as purgeable
drm/i915: Whitespace correction for madv
drm/i915: BUG_ON page refleak during unbind
drm/i915: Search harder for a reusable object
drm/i915: Clean up evict from list.
drm/i915: Add tracepoints
drm/i915: framebuffer compression for GM45+
drm/i915: split display functions by chip type
drm/i915: Skip the sanity checks if the current relocation is valid
drm/i915: Check that the relocation points to within the target
drm/i915: correct FBC update when pipe base update occurs
drm/i915: blacklist Acer AspireOne lid status
ACPI: make ACPI button funcs no-ops if not built in
drm/i915: prevent FIFO calculation overflows on 32 bits with high dotclocks
drm/i915: intel_display.c handle latency variable efficiently
...
Fix up trivial conflicts in drivers/gpu/drm/i915/{i915_dma.c|i915_drv.h}
During a page fault and rebinding the buffer there exists a window for a
signal to arrive during the i915_wait_request() and trigger a
ERESTARTSYS. This used to be handled by returning SIGBUS and thereby
killing the application. Try 'cairo-perf-trace & cairo-test-suite' and
watch X go boom!
The solution as suggested by H. Peter Anvin is to simply return NOPAGE and
leave the higher layers to spot we did not fill the page and resubmit
the page fault.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: stable@kernel.org
[anholt: Mostly squash it with another commit]
In order to correctly prevent the invalid reuse of a purged buffer, we
need to track such events and warn the user before something bad
happens.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Whilst cleaning up the patches for submission, I mis-classified non-dirty
objects as purgeable. This was causing the backing pages for those
objects to be evicted under memory-pressure, discarding valid and
unreplaceable texture data.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
As evict_something() is called by routines that do not repeatedly search
again, try harder in the initial search to find an object that matches
the request.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
First the routine attempted to unlock a mutex it did not own along the
error path.
Secondly the routine should never be called on any list but the inactive
one, since we attempt to unbind those objects, so fix the calling semantics.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
By adding tracepoint equivalents for WATCH_BUF/EXEC we are able to monitor
the lifetimes of objects, requests and significant events. These events can
then be probed using the tracing frameworks, such as systemtap and, in
particular, perf.
For example to record the stack trace for every GPU stall during a run, use
$ perf record -e i915:i915_gem_request_wait_begin -c 1 -g
And
$ perf report
to view the results.
[Updated to fix compilation issues caused.]
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Ben Gamari <bgamari@gmail.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
* 'drm-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (133 commits)
drm/vgaarb: add VGA arbitration support to the drm and kms.
drm/radeon: some r420s have a CP race with the DMA engine.
drm/radeon/r600/kms: rv670 is not DCE3
drm/radeon/kms: r420 idle after programming GA_ENHANCE
drm/radeon/kms: more fixes to rv770 suspend/resume path.
drm/radeon/kms: more alignment for rv770.c with r600.c
drm/radeon/kms: rv770 blit init called too late.
drm/radeon/kms: move around new init path code to avoid posting at init
drm/radeon/r600: fix some issues with suspend/resume.
drm/radeon/kms: disable VGA rendering engine before taking over VRAM
drm/radeon/kms: Move radeon_get_clock_info() call out of radeon_clocks_init().
drm/radeon/kms: add initial connector properties
drm/radeon/kms: Use surfaces for scanout / cursor byte swapping on big endian.
drm/radeon/kms: don't fail if we fail to init GPU acceleration
drm/r600/kms: fixup number of loops per blit calculation.
drm/radeon/kms: reprogram format in set base.
drm/radeon: avivo chips have no separate int bit for display
drm/radeon/r600: don't do interrupts
drm: fix _DRM_GEM addmap error message
drm: update crtc x/y when only fb changes
...
Fixed up trivial conflicts in firmware/Makefile due to network driver
(cxgb3) and drm (mga/r128/radeon) firmware being listed next to each
other.
If the presumed_offset as feed to userspace and returned to the kernel
from a previous execbuffer is still valid, then we do not need to rewrite
the relocation entry and may skip the offset sanity checks.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Eric noted a potential concern with the low bits not being strictly used
as part of the absolute offset (instead part of the command stream to the
GPU), but in practice that should not be an issue.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Andy Whitcroft <apw@canonical.com>
Cc: Eric Anholt <eric@anholt.net>
CC: stable@kernel.org
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Due to the necessity of having to take the struct_mutex, the i915
shrinker can not free the inactive lists if we fail to allocate memory
whilst processing a batch buffer, triggering an OOM and an ENOMEM that
is reported back to userspace. In order to fare better under such
circumstances we need to manually retry a failed allocation after
evicting inactive buffers.
To do so involves 3 steps:
1. Marking the backing shm pages as NORETRY.
2. Updating the get_pages() callers to evict something on failure and then
retry.
3. Revamping the evict something logic to be smarter about the required
buffer size and prefer to use volatile or clean inactive pages.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Similar to the madvise() concept, the application may wish to mark some
data as volatile. That is in the event of memory pressure the kernel is
free to discard such buffers safe in the knowledge that the application
can recreate them on demand, and is simply using these as a cache.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This should help GEM handle memory pressure sitatuions more gracefully.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
There is no need to store the gtt_alignment as it is either explicitly
set according to the hardware requirements (e.g. scanout) or the
minimum alignment is computed on demand.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
If we failed to set the domain, the buffer was no longer being tracked
on any list.
Cc: stable@kernel.org
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
There is a very real possibility that multiple CPUs will notice that the
GPU is wedged. This introduces all sorts of potential race conditions.
Make the wedged flag atomic to mitigate this risk.
Signed-off-by: Ben Gamari <bgamari.foss@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
We set a periodic timer to check on the GPU, resetting it every time a
batch is completed. If the timer elapses, we check acthd. If acthd
hasn't changed in two timer periods, we assume the chip is wedged.
This is implemented in such a way that it leaves the option open to
employ adaptive timer intervals in the future. One could wait until
several timer periods have elapsed before declaring the chip dead. If
the chip comes back after several periods but before the "dead"
threshold, the timer interval or dead threshold could be raised.
It is important to note that while checking for active requests, we need
to account for the fact that requests are removed from the list (i.e.
retired) in a deferred work queue handler. This means that merely
checking for an empty request_list is insufficient; the list could be
non-empty yet the GPU still idle, causing the hangcheck timer to
incorrectly mark the GPU as wedged (it took me a while to figure that
out---sigh...)
Signed-off-by: Ben Gamari <bgamari.foss@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
We'll need it in i915_irq.c for checking whether there are outstanding
requests. Also, the function really ought to return a bool, not an int.
Signed-off-by: Ben Gamari <bgamari.foss@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
i915_wait_request() only checks mm.wedged after it interacts with the
hardware, generally causing the driver to lock up waiting for a wedged
chip. Make sure we check mm.wedged as the first thing we do.
Reported-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Ben Gamari <bgamari.foss@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
drm_ht_remove_item() does not handle removing an absent item and the hlist
in particular is incorrectly initialised. The easy remedy is simply skip
calling i915_gem_free_mmap_offset() unless we have actually created the
offset and associated ht entry.
This also fixes the mishandling of a partially constructed offset which
leaves pointers initialized after freeing them along the
i915_gem_create_mmap_offset() error paths.
In particular this should fix the oops found here:
https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/415357/comments/8
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Cc: stable@kernel.org
Ever since we enabled GEM, the pre-9xx chipsets (particularly 865) have had
serious stability issues. Back in May a wbinvd was added to the DRM to
work around much of the problem. Some failure remained -- easily visible
by dragging a window around on an X -retro desktop, or by looking at bugzilla.
The chipset flush was on the right track -- hitting the right amount of
memory, and it appears to be the only way to flush on these chipsets, but the
flush page was mapped uncached. As a result, the writes trying to clear the
writeback cache ended up bypassing the cache, and not flushing anything! The
wbinvd would flush out other writeback data and often cause the data we wanted
to get flushed, but not always. By removing the setting of the page to UC
and instead just clflushing the data we write to try to flush it, we get the
desired behavior with no wbinvd.
This exports clflush_cache_range(), which was laying around and happened to
basically match the code I was otherwise going to copy from the DRM.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Brice Goglin <Brice.Goglin@ens-lyon.org>
Cc: stable@kernel.org
Otherwise, some other userland writing into its buffer may race to land
writes either after the CPU thinks it's got a coherent view, or after its
GTT entries have been redirected to point at the scratch page. Either
result is unpleasant.
Signed-off-by: Eric Anholt <eric@anholt.net>
Reinette Chatre reports a frozen system (with blinking keyboard LEDs)
when switching from graphics mode to the text console, or when
suspending (which does the same thing). With netconsole, the oops
turned out to be
BUG: unable to handle kernel NULL pointer dereference at 0000000000000084
IP: [<ffffffffa03ecaab>] i915_driver_irq_handler+0x26b/0xd20 [i915]
and it's due to the i915_gem.c code doing drm_irq_uninstall() after
having done i915_gem_idle(). And the i915_gem_idle() path will do
i915_gem_idle() ->
i915_gem_cleanup_ringbuffer() ->
i915_gem_cleanup_hws() ->
dev_priv->hw_status_page = NULL;
but if an i915 interrupt comes in after this stage, it may want to
access that hw_status_page, and gets the above NULL pointer dereference.
And since the NULL pointer dereference happens from within an interrupt,
and with the screen still in graphics mode, the common end result is
simply a silently hung machine.
Fix it by simply uninstalling the irq handler before idling rather than
after. Fixes
http://bugzilla.kernel.org/show_bug.cgi?id=13819
Reported-and-tested-by: Reinette Chatre <reinette.chatre@intel.com>
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
According to the docs, the ringbuffer is not allowed to wrap in the middle
of an instruction.
G45 PRM, Vol 1b, p101:
While the “free space” wrap may allow commands to be wrapped around the
end of the Ring Buffer, the wrap should only occur between commands.
Padding (with NOP) may be required to follow this restriction.
Do as commanded.
[Having seen bug reports where there is evidence of split commands, but
apparently the GPU has continued on merrily before a bizarre and untimely
death, this may or may not fix a few random hangs.]
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
CC: Eric Anholt <eric@anholt.net>
Signed-off-by: Eric Anholt <eric@anholt.net>
There are several sources of unnecessary power consumption on Intel
graphics systems. The first is the LVDS clock. TFTs don't suffer from
persistence issues like CRTs, and so we can reduce the LVDS refresh rate
when the screen is idle. It will be automatically upclocked when
userspace triggers graphical activity. Beyond that, we can enable memory
self refresh. This allows the memory to go into a lower power state when
the graphics are idle. Finally, we can drop some clocks on the gpu
itself. All of these things can be reenabled between frames when GPU
activity is triggered, and so there should be no user visible graphical
changes.
Signed-off-by: Jesse Barnes <jesse.barnes@intel.com>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Remember to release the local reference if we fail to wait on
the rendering.
(Also whilst in the vicinity add some whitespace so that the phasing of
the operations is clearer.)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Several functions in the GEM kernel API used int as handle type, but
user API has it __u32 which is also the intended type.
Replace int with u32.
Signed-off-by: Pekka Paalanen <pq@iki.fi>
Signed-off-by: Dave Airlie <airlied@redhat.com>
As of 52dc7d32b8, we could leave an old
linear GTT mapping in place, so that apps trying to GTT-mapped write in
tiled data wouldn't get the fence added, and garbage would get displayed.
Signed-off-by: Eric Anholt <eric@anholt.net>
As we call unmap_mapping_range() twice in identical fashion, refactor
and attempt to explain why we need to call unmap_mapping_range().
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Unitialized fence register could leads to corrupted display. Problem
encountered on MacBooks (revision 1 and 2), directly booting from EFI
or through BIOS emulation.
(bug #21710 at freedestop.org)
Signed-off-by: Grégoire Henry <henry@pps.jussieu.fr>
Signed-off-by: Eric Anholt <eric@anholt.net>
It hasn't been used in ages, and having the user tell your how much
memory is being freed at free time is a recipe for disaster even if it
was ever used.
Signed-off-by: Eric Anholt <eric@anholt.net>
The fence register value also depends upon the stride of the object, so we
need to clear the fence if that is changed as well.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
[anholt: Added 8xx and 965 paths, and renamed the confusing
i915_gem_object_tiling_ok function to i915_gem_object_fence_offset_ok]
Signed-off-by: Eric Anholt <eric@anholt.net>
With the work by Jesse Barnes to eliminate allocation of fences during
execbuffer, it becomes possible to write to the scan-out buffer with it
never acquiring a fence (simply by only ever writing to the object using
tiled GPU commands and never writing to it via the GTT). So for pre-i965
chipsets which require fenced access for tiled scan-out buffers, we need
to obtain a fence register.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
After performing an operation over the page list for a buffer retrieved by
i915_gem_object_get_pages() the pages need to be returned with
i915_gem_object_put_pages(). This was not being observed for the phys
objects which were thus leaking references to their backing pages.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
CC: Dave Airlie <airlied@gmail.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
To differentiate between encountering an out-of-memory error with running
out of space in the aperture, use ENOSPC for the later.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
The batch buffer may be shared with another read buffer, so we should not
ignore any previously set domains, but just or in the command domain (and
check that the buffer is not writable).
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
By sending a broken execbuffer (its length was not suitably aligned) I
triggered an operation upon a freed object. The invalid alignment was
discovered after updating the write_domain on the object but before the
object was placed on the active queue. So during the unwind process
following the error, the now freed object attempts to flush its
non-existent, but outstanding, GPU writes causing this use-after-free.
[drm:i915_dispatch_gem_execbuffer] *ERROR* alignment
[drm:i915_gem_execbuffer] *ERROR* dispatch failed -22
WARNING: at lib/kref.c:43 warn_slowpath_null+0x10/0x15()
Modules linked in:
Pid: 4552, comm: lt-csi-drm Not tainted 2.6.30-rc6 #423
Call Trace:
[<c0119ef3>] warn_slowpath_fmt+0x57/0x6d
[<c014de24>] ? get_pageblock_migratetype+0x18/0x1e
[<c014e8fd>] ? free_hot_page+0xa/0xc
[<c014e915>] ? __free_pages+0x16/0x1f
[<c0153ebf>] ? shmem_truncate_range+0x63e/0x656
[<c015fb2f>] ? slob_page_alloc+0x146/0x1c8
[<c0119f19>] warn_slowpath_null+0x10/0x15
[<c01f55f2>] kref_get+0x1b/0x21
[<c02605db>] i915_gem_object_move_to_active+0x1f/0x56
[<c0261302>] i915_add_request+0x156/0x19a
[<c026136e>] i915_gem_object_flush_gpu_write_domain+0x28/0x3f
[<c0261eca>] i915_gem_object_unbind+0x4a/0x124
[<c0261fd7>] i915_gem_free_object+0x33/0x9b
[<c0250d6b>] drm_gem_object_free+0x28/0x4a
[<c0250d43>] ? drm_gem_object_free+0x0/0x4a
[<c01f55ce>] kref_put+0x38/0x41
[<c0250cbf>] drm_gem_object_unreference+0x11/0x13
[<c0250d06>] drm_gem_object_handle_unreference+0x1e/0x21
[<c0250d13>] drm_gem_object_release_handle+0xa/0xe
[<c01f3e6b>] idr_for_each+0x5f/0x98
[<c0250d09>] ? drm_gem_object_release_handle+0x0/0xe
[<c0250daf>] drm_gem_release+0x22/0x34
[<c025046f>] drm_release+0x1e8/0x3c4
[<c0162d25>] __fput+0xaf/0x146
[<c0162dce>] fput+0x12/0x14
[<c01605ef>] filp_close+0x48/0x52
[<c011b182>] put_files_struct+0x57/0x9b
[<c011b1e4>] exit_files+0x1e/0x20
[<c011c6b6>] do_exit+0x16d/0x511
[<c03704ab>] ? __schedule+0x3d4/0x3e5
[<c0103f0d>] ? handle_irq+0xd/0x69
[<c011caa7>] do_group_exit+0x4d/0x73
[<c011cae0>] sys_exit_group+0x13/0x17
[<c010268c>] sysenter_do_call+0x12/0x2b
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Update interrupt handling methods for IGDNG with new registers
for display and graphics interrupt functions. As we won't use
irq-based vblank sync in dri2, so display interrupt on new chip
will be used for hotplug only in future.
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
keithp didn't like the original 20ms plan because a cooperative client could
be starved by an uncooperative client. There may even have been problems
with cooperative clients versus cooperative clients. So keithp changed
throttle to just wait for the second to last seqno emitted by that client.
It worked well, until we started getting more round-trips to the server
due to DRI2 -- the server throttles in BlockHandler, and so if you did more
than one round trip after finishing your frame, you'd end up unintentionally
syncing to the swap.
Fix this by keeping track of the client's requests, so the client can wait
when it has an outstanding request over 20ms old. This should have
non-starving behavior, good behavior in the presence of restarts, and less
waiting. Improves high-settings openarena performance on my GM45 by 50%.
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This could be triggered by a gtt mapping fault on 965 that decides to
remove the fence from another object that happens to be active currently.
Since the other object doesn't rely on the fence reg for its execution, we
don't wait for it to finish. We'll soon be not waiting on 915 most of the
time as well, so just drop the BUG_ON.
Signed-off-by: Eric Anholt <eric@anholt.net>
When a GEM object is evicted from the GTT we set it to the CPU domain,
as it might get swapped in and out or ever mmapped regularly. If the
object is mmapped through the GTT it can still get evicted in this way
by other objects requiring GTT space. When the GTT mapping is touched
again we fault it back into the GTT, but fail to set it back to the
GTT domain. This means we fail to flush any cached CPU writes to the
pages backing the object which will then happen "eventually", typically
after we write to the page through the uncached GTT mapping.
[anholt: Note that userland does do a set_domain(GTT, GTT) when starting
to access the GTT mapping. That covers getting the existing mapping of the
object synchronized if it's bound to the GTT. But set_domain(GTT, GTT)
doesn't do anything if the object is currently unbound. This fix covers the
transition to being bound for GTT mapping.]
Fixes glyph and other pixmap corruption during swapping. fd.o bug #21790
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
On the 865, but not the 855, the clflush we do appears to not actually make
it out to the hardware all the time. An easy way to safely reproduce was
X -retro, which would show that some of the blits involved in drawing the
lovely root weave didn't make it out to the hardware. Those blits are 32
bytes each, and 1-2 would be missing at various points around the screen.
Other experimentation (doing more clflush, doing more AGP chipset flush,
poking at some more device registers to maybe trigger more flushing) didn't
help. krh came up with the wbinvd as a way to successfully get all those
blits to appear.
Signed-off-by: Eric Anholt <eric@anholt.net>
The pitch field is an exponent on pre-965, so we were rejecting buffers
on 8xx that we shouldn't have. 915 got lucky in that the largest legal
value happened to match (8KB / 512 = 0x10), but 8xx has a smaller tile width.
Additionally, we programmed that bad value into the register on 8xx, so the
only pitch that would work correctly was 4096 (512-1023 pixels), while others
would probably give bad rendering or hangs.
Signed-off-by: Eric Anholt <eric@anholt.net>
fd.o bug #20473.
For some reason we never added 8xx desktop cursor support to the
kernel. This patch fixes that.
[krh: Also set the size on pre-i915 hw.]
Tested-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
For awhile now, many of the GEM code paths have allocated page or
object arrays with the slab allocator. This is nice and fast, but
won't work well if memory is fragmented, since the slab allocator works
with physically contiguous memory (i.e. order > 2 allocations are
likely to fail fairly early after booting and doing some work).
This patch works around the issue by falling back to vmalloc for
>PAGE_SIZE allocations. This is ugly, but much less work than chaining
a bunch of pages together by hand (suprisingly there's not a bunch of
generic kernel helpers for this yet afaik). vmalloc space is somewhat
precious on 32 bit kernels, but our allocations shouldn't be big enough
to cause problems, though they're routinely more than a page.
Note that this patch doesn't address the unchecked
alloc-based-on-ioctl-args in GEM; that needs to be fixed in a separate
patch.
Also, I've deliberately ignored the DRM's "area" junk. I don't think
anyone actually uses it anymore and I'm hoping it gets ripped out soon.
[Updated: removed size arg to new free function. We could unify the
free functions as well once the DRM mem tracking is ripped out.]
fd.o bug #20152 (part 1/3)
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
We might sleep here anyway so I hope an extra uncached read is ok to
add.
In #20896 we found that vbetool clobbers the IER. In KMS mode this is
particularly bad since we don't set the interrupt regs late (in
EnterVT), so we'd fail to get *any* interrupts at all after X started
(since some distros have scripts that call vbetool at X startup
apparently).
So this patch checks IER at wait_request time, and re-enables
interrupts if it's been clobbered. In a proper config this check
should never be triggered.
This is really a distro issue, but having a sanity check is nice, as
long as it doesn't have a real performance hit.
Tested-by: Mateusz Kaduk <mateusz.kaduk@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[anholt: Moved the check inside of the sleeping case to avoid perf cost]
Signed-off-by: Eric Anholt <eric@anholt.net>
regression caused by commit 5e118f4139:
i915_gem_object_move_to_inactive() should be called in task context,
as it calls fput();
Signed-off-by: Shaohua Li<shaohua.li@intel.com>
[anholt: Add more detail to the comment about the lock break that's added]
Signed-off-by: Eric Anholt <eric@anholt.net>
Save the bit 17 state of the pages when freeing the page list, and
reswizzle them if necessary when rebinding the pages (in case they were
swapped out). Since we have userland with expectations that the swizzle
enums let it pread and pwrite contents accurately, we can't expose a new
swizzle enum for bit 17 (which it would have to GTT map to handle), so we
handle it down in pread and pwrite by swizzling the copy when bit 17 of the
page address is set.
Signed-off-by: Eric Anholt <eric@anholt.net>
Otherwise, the results of our read didn't show up when we were faulting in
the page being read into (as happened with a testcase reading into a big
stack area). Likely accounts for some conformance test failures.
Signed-off-by: Eric Anholt <eric@anholt.net>
i915_gem_put_relocs_to_user returned an uninitialized value which
got returned to userspace. This caused libdrm in my setup to never
get out of a do{}while() loop retrying i915_gem_execbuffer.
result was hanging X, overheating of cpu and 2-3gb of logfile-spam.
This patch adresses the issue by
1. initializing vars in this file where necessary
2. correcting wrongly interpreted return values of copy_[from/to]_user
Signed-off-by: Florian Mickler <florian@mickler.org>
[anholt: cleanups of unnecessary changes, consistency in APIs]
Signed-off-by: Eric Anholt <eric@anholt.net>
We create a debugfs node (i915_ringbuffer_data) to expose a hex dump
of the ring buffer itself. We also expose another debugfs node
(i915_ringbuffer_info) with information on the state (i.e. head, tail
addresses) of the ringbuffer.
For batchbuffer dumping, we look at the device's active_list, dumping
each object which has I915_GEM_DOMAIN_COMMAND in its read
domains. This is all exposed through the dri/i915_batchbuffers debugfs
file with a header for each object (giving the objects gtt_offset so
that it can be matched against the offset given in the
BATCH_BUFFER_START command.
Signed-off-by: Ben Gamari <bgamari@gmail.com>
Signed-off-by: Carl Worth <cworth@cworth.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
This is a baby-step in the direction of having finer-grained
locking than the struct_mutex. Specifically, this will enable
new debugging code to read the active list for printing out
GPU state when the GPU is wedged, (while the struct_mutex is
held, of course).
Signed-off-by: Carl Worth <cworth@cworth.org>
[anholt: indentation fix]
Signed-off-by: Eric Anholt <eric@anholt.net>
Indicates something is wrong with the mapping; and apparently triggers
in current kernels.
Signed-off-by: Jesse Barnes <jbarnes@virtuosugeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
This fixes all the tiling problems with the 2d ddx. glxgears still doesn't work.
Changes:
- fix a copy&paste error in i8xx fence reg setup. It resulted in an at most a
512KB offset of the fence reg window, so was only visible sometimes.
- add tests for stride and object size constrains (also for i915 and 1965 class
hw). Userspace seems to have an of-by-one bug there, which changes the fence
size by at most 512KB due to an overflow.
- because i8xx hw is quite old (and therefore not as well-tested) I left 2 debug
WARN_ONs in the i8xx fence reg setup code to hopefully catch any further
overflows in the bit-fields. Lastly there's one small change to make the
alignment checks more consistent.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=20289
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric Anholt <eric@anholt.net>
This produced a warning on my build, not sure why super-warning-man didn't
notice this one, its much worse than the %z one.
Signed-off-by: Dave Airlie <airlied@redhat.com>
agp_chipset_flush() is for flushing the intel GMCH write cache via the
IFP, these two uses are for when we're getting the object into the cpu
READ domain, and thus should not be needed. This confused me when I was
getting my head around the code.
With thanks to airlied for helping me check my mental picture of how the
flushes and clflushes are supposed to be used.
Signed-off-by: Owain G. Ainsworth <oga@openbsd.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
This introduces allocation in the batch submission path that wasn't there
previously, but these are compatibility paths so we care about simplicity
more than performance.
kernel.org bug #12419.
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Like the GTT pwrite path fix, this uses an optimistic path and a
fallback to get_user_pages. Note that this means we have to stop using
vfs_write and roll it ourselves.
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
We've wanted this for a few consumers that touch the pages directly (such as
the following commit), which have been doing the refcounting outside of
get/put pages.
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Since the pagefault path determines that the lock order we use has to be
mmap_sem -> struct_mutex, we can't allow page faults to occur while the
struct_mutex is held. To fix this in pwrite, we first try optimistically to
see if we can copy from user without faulting. If it fails, fall back to
using get_user_pages to pin the user's memory, and map those pages
atomically when copying it to the GPU.
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
agp_chipset_flush() is for flushing the intel GMCH write cache via the
IFP, these two uses are for when we're getting the object into the cpu
READ domain, and thus should not be needed. This confused me when I was
getting my head around the code.
With thanks to airlied for helping me check my mental picture of how the
flushes and clflushes are supposed to be used.
Signed-off-by: Owain G. Ainsworth <oga@openbsd.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Once upon a time, the DRM made the distinction between the drm_map
data structure exchanged with user space and the drm_local_map used
in the kernel.
For some reasons, while the BSD port still has that "feature", the
linux part abused drm_map for kernel internal usage as the local
map only existed as a typedef of the struct drm_map.
This patch fixes it by declaring struct drm_local_map separately
(though its content is currently identical to the userspace variant),
and changing the kernel code to only use that, except when it's a
user<->kernel interface (ie. ioctl).
This allows subsequent changes to the in-kernel format
I've also replaced the use of drm_local_map_t with struct drm_local_map
in a couple of places. Mostly by accident but they are the same (the
former is a typedef of the later) and I have some remote plans and
half finished patch to completely kill the drm_local_map_t typedef
so I left those bits in.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
The last 8 fence registers sit at a different offset, so when we went to set
fence number 8 in the lower offset, we instead set PGETBL_CTL, and the GPU
got all sorts of angry at us.
fd.o bug #20567. Easily reproducible by running glxgears and killing it about
6 times.
Signed-off-by: Eric Anholt <eric@anholt.net>
The i915 also uses the fence registers for GPU access to tiled buffers so
we cannot reallocate one whilst it is on the active list. By performing a
LRU scan of the fenced buffers we also avoid waiting the possibility of
waiting on a pinned, or otherwise unusable, buffer.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
We need to check and report if there are no available fences - or else we
spin endlessly waiting for a buffer to magically unpin itself.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
As we may steal the fence register of an unpinned buffer for another,
every time we repin the buffer we need to recheck whether it needs to be
allocated a fence.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
If we wait upon a request and successfully unbind a buffer occupying a
fence register, then that slot will be freed and cause a NULL derefrence
upon rescanning.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
If userspace passes an object list with the same object appearing more
than once, we end up hitting the BUG_ON() in
i915_gem_object_set_to_gpu_domain() as it gets called a second time
for the same object.
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
dev_priv->hw_status_page can be NULL, if i915_gem_retire_requests()
is called from i915_gem_busy_ioctl().
Signed-off-by Karsten Wiese <fzu@wemgehoertderstaat.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The object is dereferenced before the NULL check. Oops.
Fixes http://bugs.freedesktop.org/show_bug.cgi?id=20235
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This ensures that the user gets the latest information from the hardware
on whether the buffer is busy, potentially reducing the working set of objects
that the user chooses.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
In the KMS case, we need to suspend/resume GEM as well. So on suspend, make
sure we idle GEM and stop any new rendering from coming in, and on resume,
re-init the framebuffer and clear the suspended flag.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The problem was that object_set_to_gpu_domain would set the new write_domains
that are getting set by this batchbuffer, then the accumulated flushes required
for all the objects in preparation for this batchbuffer were posted, and the
brand new write domain would get cleared by the flush being posted. Instead,
hang on to the new (or old if we're not changing it) value and set it after
the flush is queued.
Results from this noticably included conformance test failures from reads
shortly after writes (where the new write domain had been lost and thus not
flushed and waited on), but is a suspected cause of hangs in some apps when
a write domain is lost on a buffer that gets reused for instruction or
commmand state.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
While not strictly required, it helped while thinking about the following
change. This change should be invariant.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This fixes potential fault at fault time if the object was unreferenced
while the mapping still existed. Now, while the mmap_offset only lives
for the lifetime of the object, the object also stays alive while a vma
exists that needs it.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
If we fail to create the ringbuffer, then we need to cleanup the allocated
hws.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
A missing unpin on the error path.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
A missing unpin on the error path.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
A missing unreference and unpin after rejecting the relocation for an
invalid memory domain.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
We failed to unlock the mutex after failing to create the mmap offset.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
Lockdep warns that i915_gem_execbuffer() can trigger a page fault (which
takes mmap_sem) while holding dev->struct_mutex, while drm_vm_open()
(which is called with mmap_sem already held) takes dev->struct_mutex.
So this is a potential AB-BA deadlock.
The way that i915_gem_execbuffer() triggers a page fault is by doing
copy_to_user() when returning new buffer offsets back to userspace;
however there is no reason to hold the struct_mutex when doing this
copy, since what is being copied is the contents of an array private to
i915_gem_execbuffer() anyway. So we can fix the potential deadlock (and
get rid of the lockdep warning) by simply moving the copy_to_user()
outside of where struct_mutex is held.
This fixes <http://bugzilla.kernel.org/show_bug.cgi?id=12491>.
Reported-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
A missing unreference if the user calls pin() a second time on a pinned
buffer.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
Also spotted by Owain Ainsworth.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
The DRI people seem to have a hard time getting these right (see also
commit aeb565dfc3).
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
If we failed to allocate a new fence register we would return
VM_FAULT_SIGBUS without relinquishing the lock.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
Adds code to set up fence registers at execbuf time on pre-965 chips as
necessary. Also fixes up a few bugs in the pre-965 tile register support
(get_order != ffs). The number of fences available to the kernel defaults
to the hw limit minus 3 (for legacy X front/back/depth), but a new parameter
allows userspace to override that as needed.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
Previously, the caller would continue along without knowing that the
function failed, resulting in potential mis-rendering. Right now vm_fault
just returns SIGBUS in that case, and we may need to disable signal handling
to avoid that happening.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
We'd love to just be using PAT, but even on chips with PAT it gets disabled
sometimes due to an errata. It would probably be better to have pat_enabled
exported and only bother with this when !pat_enabled.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
drm/i915: Fix cursor physical address choice to match the 2D driver.
drm: stash AGP include under the do-we-have-AGP ifdef
drm: don't whine about not reading EDID data
drm/i915: hook up LVDS DPMS property
drm/i915: remove unnecessary debug output in KMS init
i915: fix freeing path for gem phys objects.
drm: create mode_config idr lock
drm: fix leak of device mappings since multi-master changes.
Use '%zu' to print out a size_t variable, not '%d'. Another case of the
"let's keep at least Linus' defconfig compile warningless" rule.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is an initial patch to do support for objects which needs physical
contiguous main ram, cursors and overlay registers on older chipsets.
These objects are bound on cursor bin, like pinning, and we copy
the data to/from the backing store object into the real one on attach/detach.
notes:
possible over the top in attach/detach operations.
no overlay support yet.
Signed-off-by: Dave Airlie <airlied@redhat.com>
This avoids a BUG_ON in the enter_vt path due to objects being in the GTT
when we shouldn't have ever let them be (as we're not supposed to touch the
device during that time).
This was triggered by a change in the 2D driver to use the GTT mapping of
objects after pinning them to improve software fallback performance.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
The error path for object list being null is in the second goto target.
Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
This showed up in logs where people had a hung chip, so pinning was blocked
on the chip unpinning other buffers, and the X Server took its scheduler
signal during that time.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
We haven't seen this in practice, but it was visible when looking at a bug
report from when i915_gem_evict_everything() was broken and would always
return error.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
This commit adds i915 driver support for the DRM mode setting APIs.
Currently, VGA, LVDS, SDVO DVI & VGA, TV and DVO LVDS outputs are
supported. HDMI, DisplayPort and additional SDVO output support will
follow.
Support for the mode setting code is controlled by the new 'modeset'
module option. A new config option, CONFIG_DRM_I915_KMS controls the
default behavior, and whether a PCI ID list is built into the module for
use by user level module utilities.
Note that if mode setting is enabled, user level drivers that access
display registers directly or that don't use the kernel graphics memory
manager will likely corrupt kernel graphics memory, disrupt output
configuration (possibly leading to hangs and/or blank displays), and
prevent panic/oops messages from appearing. So use caution when
enabling this code; be sure your user level code supports the new
interfaces.
A new SysRq key, 'g', provides emergency support for switching back to
the kernel's framebuffer console; which is useful for testing.
Co-authors: Dave Airlie <airlied@linux.ie>, Hong Liu <hong.liu@intel.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Use the new core GEM object mapping code to allow GTT mapping of GEM
objects on i915. The fault handler will make sure a fence register is
allocated too, if the object in question is tiled.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
These buffers don't have active rendering still occurring to them, they just
need either a flush to be emitted or a retire_requests to occur so that we
notice they're done. Return unbusy so that one of the two occurs. The two
expected consumers of this interface (OpenGL and libdrm_intel BO cache) both
want this behavior.
Signed-off-by: Eric Anholt <eric@anholt.net>
Acked-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
It's only for flushing caches appropriately for GTT access, not for actually
getting it there. Prevents potential smashing of cpu read/write domains on
unbound objects.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
If we fail to pin all of the buffers in an execbuffer request, go through
and clear the GTT and try again to see if its just a matter of fragmentation
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This eliminates the dev_set_domain function and just in-lines it
where its used, with the goal of moving the manipulation and use of
invalidate_domains and flush_domains closer together. This also
avoids calling add_request unless some domain has been flushed.
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Now that the CPU and GTT domain operations are isolated to their own
functions, the previously general-purpose set_domain function is now used
only to set GPU domains. It also has no failure cases, which is important as
this eliminates any possible interruption of the computation of new object
domains and subsequent emmission of the flushing instructions into the ring.
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This fixes several domain management bugs, including potential lack of cache
invalidation for pread, potential failure to wait for set_domain(CPU, 0),
and more, along with producing more intelligible code.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This fixes failure to flush caches in the relocation update path, and
failure to wait in the set_domain ioctl, each of which could lead to incorrect
rendering.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Otherwise, we would leave the objects in an inconsistent state, such as
write_domain == 0 but on the flushing list.
Signed-off-by: Dave Airlie <airlied@redhat.com>
obj_priv->write_domain is "write domain if the GPU went idle now", not
"write domain at this moment." By postponing the clear, we confused the
concept, required more storage, and potentially emitted more flushes than
are required.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Before we had the notion of pinning objects, we had a kludge around to make
sure all of the objects were still resident in the GTT before we committed
to executing a batch buffer. We don't need this any longer, and it sticks an
error return in the middle of object domain computations that must be
associated with a subsequent flush/invalidate emmission into the ring.
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The old code was wandering through the active list looking for pinned
buffers; there may be other pinned buffers around. Fortunately, we keep a
count of the total amount of pinned memory and can use that instead.
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Instead, just warn that bad things are happening and do our best to clean up
the mess without the GPU's help.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This could return early when reading after writing a buffer, if somebody
had already put it on the flushing list (write domains are 0, but still
active), leading to glReadPixels failure.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@linux.ie>
* 'io-mappings-for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
io mapping: clean up #ifdefs
io mapping: improve documentation
i915: use io-mapping interfaces instead of a variety of mapping kludges
resources: add io-mapping functions to dynamically map large device apertures
x86: add iomap_atomic*()/iounmap_atomic() on 32-bit using fixmaps
This will let userland know when to submit its batchbuffers, before they get
too big to fit in the aperture.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Impact: optimize/clean-up the IO mapping implementation of the i915 DRM driver
Switch the i915 device aperture mapping to the io-mapping interface, taking
advantage of the cleaner API to extend it across all of the mapping uses,
including both pwrite and relocation updates.
This dramatically improves performance on 64-bit kernels which were using
the same slow path as 32-bit non-HIGHMEM kernels prior to this patch.
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
* 'drm-next' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
drm: Avoid oops in DRM_IOCTL_RM_DRAW if a bad handle is supplied.
drm: Add 32-bit compatibility for DRM_IOCTL_UPDATE_DRAW.
drm/i915: use pipes, not planes to label vblank data
drm/i915: hold dev->struct_mutex and DRM lock during vblank ring operations
i915: Fix format string warnings on x86-64.
i915: Don't dereference HWS in /proc debug files when it isn't initialized.
i915: Enable IMR passthrough of vblank events before enabling it in pipestat.
drm: Remove two leaks of vblank reference count in error paths.
drm: fix leak of cliprects in drm_rmdraw()
i915: Disable MSI on GM965 (errata says it doesn't work)
drm: Set cliprects to NULL when changing drawable to having 0 cliprects.
i915: Protect vblank IRQ reg access with spinlock
To synchronize clip lists with the X server, the DRM lock must be held while
looking at drawable clip lists. To synchronize with other ring access, the
ring mutex must be held while inserting commands into the ring. Failure to
do the first resulted in easy visual corruption when moving windows, and the
second could have corrupted the ring with DRI2.
Grabbing the DRM lock involves using the DRM tasklet mechanism, grabbing the
ring mutex means potentially sleeping. Deal with both of these by always
running the tasklet from a work handler.
Also, protect from clip list changes since the vblank request was queued by
making sure the window has at least one rectangle while looking inside,
preventing oopses .
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
commit 9b7530cc32 ("i915: cleanup coding
horrors in i915_gem_gtt_pwrite()")
broke the i386 build for CONFIG_HIGHMEM=y.
Caught by automatic testing http://www.tglx.de/autoqa-logs/000137-0006-0001.log
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
[ My bad. It's the same patch I sent out earlier, nobody noticed then either.. ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Yes, this will probably be switched over to a cleaner model anyway, but
in the meantime I don't want to see the 'unused variable' warnings that
come from the disgusting #ifdef code. Make the special case be a nice
inlien function of its own, clean up the code, and make the warning go
away.
I wish people didn't write code that gets (valid) warnings from the
compiler, but I'll limit my fixes to code that I actually care about (in
this case just because I see the warning and it annoys me).
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
At leavevt and lastclose time, cancel any pending retire work handler
invocation, and keep the retire work handler from requeuing itself if it is
currently running.
This patch restructures i915_gem_idle to perform all of these tasks instead
of having both leavevt and lastclose call a sequence of functions.
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This should improve performance by avoiding uncached reads by the CPU (the
point of having a status page), and may improve stability. This patch only
affects G33, GM45 and G45 chips as those are the only ones using GTT-based
HWS mappings.
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
G45 appears quite sensitive to ring initialization register writes,
sometimes leaving the HEAD register with the START register contents. Check
to make sure HEAD is reset correctly when START is written, and fix it up,
screaming loudly.
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Fixes failure to map the ringbuffer when PAT tells us we don't get to do
uncached on something that's already mapped WC, or something along those lines.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
In the conversion for GEM, we had stopped using the hardware lock to protect
ring usage, since it was all internal to the DRM now. However, some paths
weren't converted to using struct_mutex to prevent multiple threads from
concurrently working on the ring, in particular between the vblank swap handler
and ioctls.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
GEM allows the creation of persistent buffer objects accessible by the
graphics device through new ioctls for managing execution of commands on the
device. The userland API is almost entirely driver-specific to ensure that
any driver building on this model can easily map the interface to individual
driver requirements.
GEM is used by the 2d driver for managing its internal state allocations and
will be used for pixmap storage to reduce memory consumption and enable
zero-copy GLX_EXT_texture_from_pixmap, and in the 3d driver is used to enable
GL_EXT_framebuffer_object and GL_ARB_pixel_buffer_object.
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
