linux-stable.git/drivers/gpu/drm/ttm, branch v3.2.99

drm/ttm: Fix accounting error when fail to get pages for pool

2017-11-26T13:51:04+00:00

commit 9afae2719273fa1d406829bf3498f82dbdba71c7 upstream.

When fail to get needed page for pool, need to put allocated pages
into pool. But current code has a miscalculation of allocated pages,
correct it.

Signed-off-by: Xiangliang.Yu 
Reviewed-by: Christian König 
Reviewed-by: Monk Liu 
Signed-off-by: Alex Deucher 
Signed-off-by: Ben Hutchings

drm/ttm: Make sure BOs being swapped out are cacheable

2017-06-05T20:13:44+00:00

commit 239ac65fa5ffab71adf66e642750f940e7241d99 upstream.

The current caching state may not be tt_cached, even though the
placement contains TTM_PL_FLAG_CACHED, because placement can contain
multiple caching flags. Trying to swap out such a BO would trip up the

	BUG_ON(ttm->caching_state != tt_cached);

in ttm_tt_swapout.

Signed-off-by: Michel Dänzer 
Reviewed-by: Thomas Hellstrom 
Reviewed-by: Christian König .
Reviewed-by: Sinclair Yeh 
Signed-off-by: Christian König 
Signed-off-by: Ben Hutchings

drm/ttm: Fixed a read/write lock imbalance

2015-12-30T02:25:59+00:00

commit 025af189fb44250206dd8a32fa4a682392af3301 upstream.

In ttm_write_lock(), the uninterruptible path should call
__ttm_write_lock() not __ttm_read_lock().  This fixes a vmwgfx hang
on F23 start up.

syeh: Extracted this from one of Thomas' internal patches.

Signed-off-by: Thomas Hellstrom 
Reviewed-by: Sinclair Yeh 
Signed-off-by: Ben Hutchings

drm/ttm: Fix possible stack overflow by recursive shrinker calls.

2014-09-13T22:41:43+00:00

commit 71336e011d1d2312bcbcaa8fcec7365024f3a95d upstream.

While ttm_dma_pool_shrink_scan() tries to take mutex before doing GFP_KERNEL
allocation, ttm_pool_shrink_scan() does not do it. This can result in stack
overflow if kmalloc() in ttm_page_pool_free() triggered recursion due to
memory pressure.

  shrink_slab()
  => ttm_pool_shrink_scan()
     => ttm_page_pool_free()
        => kmalloc(GFP_KERNEL)
           => shrink_slab()
              => ttm_pool_shrink_scan()
                 => ttm_page_pool_free()
                    => kmalloc(GFP_KERNEL)

Change ttm_pool_shrink_scan() to do like ttm_dma_pool_shrink_scan() does.

Signed-off-by: Tetsuo Handa 
Signed-off-by: Dave Airlie 
[bwh: Backported to 3.2:
 - Adjust context
 - Change return value in the contended case to follow the old shrinker
   API]   
Signed-off-by: Ben Hutchings

drm/ttm: Fix accesses through vmas with only partial coverage

2014-02-15T19:20:13+00:00

commit d386735588c3e22129c2bc6eb64fc1d37a8f805c upstream.

VMAs covering a bo but that didn't start at the same address space offset as
the bo they were mapping were incorrectly generating SEGFAULT errors in
the fault handler.

Reported-by: Joseph Dolinak 
Signed-off-by: Thomas Hellstrom 
Reviewed-by: Jakob Bornecrantz 
[bwh: Backported to 3.2: drm_vma_node_start() is open-coded;
 vma_pages() was open-coded]
Signed-off-by: Ben Hutchings

drm/ttm: Fix memory type compatibility check

2014-01-03T04:33:17+00:00

commit 59c8e66378fb78adbcd05f0d09783dde6fef282b upstream.

Also check the busy placements before deciding to move a buffer object.
Failing to do this may result in a completely unneccessary move within a
single memory type.

Signed-off-by: Thomas Hellstrom 
Reviewed-by: Jakob Bornecrantz 
[bwh: Backported to 3.2: adjust context]
Signed-off-by: Ben Hutchings

ttm: Clear the ttm page allocated from high memory zone correctly

2012-12-06T11:20:07+00:00

commit ac207ed2471150e06af0afc76e4becc701fa2733 upstream.

The TTM page can be allocated from high memory. In such case it is
wrong to use the page_address(page) as the virtual address for the high memory
page.

bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=50241

Signed-off-by: Zhao Yakui 
Reviewed-by: Thomas Hellstrom 
Signed-off-by: Dave Airlie 
[bwh: Backported to 3.2: adjust context]
Signed-off-by: Ben Hutchings

drm/ttm: Fix spinlock imbalance

2012-06-10T13:41:55+00:00

commit a8ff3ee211fccf708e1911bbc096625453ebf759 upstream.

This imbalance may cause hangs when TTM is trying to swap out a buffer
that is already on the delayed delete list.

Signed-off-by: Thomas Hellstrom 
Reviewed-by: Jakob Bornecrantz 
Signed-off-by: Dave Airlie 
Signed-off-by: Ben Hutchings

ttm: Don't return the bo reserved on error path

2011-11-22T20:06:21+00:00

An unlikely race could case a bo to be returned reserved on an error path.

Signed-off-by: Thomas Hellstrom 
Reviewed-by: Jerome Glisse 
Signed-off-by: Dave Airlie

gpu: Add export.h as required to drivers/gpu files.

2011-10-31T23:32:03+00:00

They need this to get all the EXPORT_SYMBOL variants and THIS_MODULE

Signed-off-by: Paul Gortmaker