dri-devel October 2010

dri-devel@lists.freedesktop.org

76 participants
345 discussions

[PATCH] DRM: ignore invalid EDID extensions
by Sam Tygier 01 Oct '10

01 Oct '10

Currently an invalid EDID extension will cause the whole EDID to be considered invalid. Instead just drop the extension, and return the valid base block. The base block is modified to not claim to have extensions, and update the checksum. For my EIZO S2242W the base block is fine, but the extension block is all zeros. Without this patch I get no X and no VTs. Signed-off-by: Sam Tygier <samtygier(a)yahoo.co.uk> --- drivers/gpu/drm/drm_edid.c | 9 +++++++++ 1 files changed, 9 insertions(+), 0 deletions(-) diff --git a/drivers/gpu/drm/drm_edid.c b/drivers/gpu/drm/drm_edid.c index f87bf10..5ade343 100644 --- a/drivers/gpu/drm/drm_edid.c +++ b/drivers/gpu/drm/drm_edid.c @@ -285,6 +285,15 @@ carp: dev_warn(connector->dev->dev, "%s: EDID block %d invalid.\n", drm_get_connector_name(connector), j); + /* Invalid extension, so set block[0x7e] to zero, and return + * the base block */ + block[EDID_LENGTH-1] += block[0x7e]; + block[0x7e] = 0; + new = krealloc(block, EDID_LENGTH, GFP_KERNEL); + if (!new) + goto out; + block = new; + return block; out: kfree(block); return NULL; -- 1.7.1

2 5

[PATCH] drm/ttm: Fix busy codepaths of the "Fix two race conditions" commit.
by Thomas Hellstrom 01 Oct '10

01 Oct '10

The above mentioned commit didn't queue the buffer on the delayed destroy list under some rare circumstances. It also didn't completely honor the remove_all parameter. Signed-off-by: Thomas Hellstrom <thellstrom(a)vmware.com> --- drivers/gpu/drm/ttm/ttm_bo.c | 15 +++++++++++---- 1 files changed, 11 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index b98cb2c..db809e0 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -493,6 +493,7 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, bool remove_all) int ret; spin_lock(&bo->lock); +retry: (void) ttm_bo_wait(bo, false, false, !remove_all); if (!bo->sync_obj) { @@ -501,15 +502,16 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, bool remove_all) spin_unlock(&bo->lock); spin_lock(&glob->lru_lock); - ret = ttm_bo_reserve_locked(bo, false, true, false, 0); + ret = ttm_bo_reserve_locked(bo, false, !remove_all, false, 0); /** * Someone else has the object reserved. Bail and retry. */ - if (unlikely(ret != 0)) { + if (unlikely(ret == -EBUSY)) { spin_unlock(&glob->lru_lock); - return 0; + spin_lock(&bo->lock); + goto requeue; } /** @@ -524,6 +526,11 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, bool remove_all) atomic_set(&bo->reserved, 0); wake_up_all(&bo->event_queue); spin_unlock(&glob->lru_lock); + spin_lock(&bo->lock); + if (remove_all) + goto retry; + else + goto requeue; } put_count = ttm_bo_del_from_lru(bo); @@ -540,7 +547,7 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, bool remove_all) return 0; } - +requeue: spin_lock(&glob->lru_lock); if (list_empty(&bo->ddestroy)) { void *sync_obj = bo->sync_obj; -- 1.6.2.5

1 0

[PATCH] drm/ttm: Fix missing return
by Thomas Hellstrom 01 Oct '10

01 Oct '10

The commit "drm/ttm: Fix two race conditions" was missing a return statement in a very unlikely code path. Signed-off-by: Thomas Hellstrom <thellstrom(a)vmware.com> --- drivers/gpu/drm/ttm/ttm_bo.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index b98cb2c..d6000d7 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -524,6 +524,7 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, bool remove_all) atomic_set(&bo->reserved, 0); wake_up_all(&bo->event_queue); spin_unlock(&glob->lru_lock); + return 0; } put_count = ttm_bo_del_from_lru(bo); -- 1.6.2.5

1 1

[PATCH 0/4] Urgent vmwgfx fixes.
by Thomas Hellstrom 01 Oct '10

01 Oct '10

These are fixes that should, if at all possible, make it into 2.6.36. They are extracted from the previous patch series, "vmwgfx updates". I will respin the rest of the patches for possible inclusion into drm-next. I've renamed some patches and added some comment to emphasise the fixes in them rather than what's actually being done with the code. Patch 1 fixes vt switching while 3D objects are active. Previously the device lost all state at that point. Patch 2 enables the vblank system and avoids a lot of uninitialized calls (kernel crashes) Patch 3 removes a workaround for dev::devname the current code caused a kernel crash at module unload then dev::devname was beeing kfreed when it was actually static. Patch 4 Fixes a bo pinning error causing severe rendering corruption. Thanks, Thomas

1 4

[PATCH] drm/ttm: Fix two race conditions
by Thomas Hellstrom 01 Oct '10

01 Oct '10

This fixes a race pointed out by Dave Airlie where we don't take a buffer object about to be destroyed off the LRU lists properly. It also fixes a rare case where a buffer object could be destroyed in the middle of an accelerated eviction. The patch also adds a utility function that can be used to prematurely release GPU memory space usage of an object waiting to be destroyed. For example during eviction or swapout. Signed-off-by: Thomas Hellstrom <thellstrom(a)vmware.com> --- drivers/gpu/drm/ttm/ttm_bo.c | 74 +++++++++++++++++++++++++++++++++++------ include/drm/ttm/ttm_bo_api.h | 4 ++- 2 files changed, 66 insertions(+), 12 deletions(-) diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index cb4cf7e..b98cb2c 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -442,6 +442,43 @@ out_err: } /** + * Call bo::reserved and with the lru lock held. + * Will release GPU memory type usage on destruction. + * This is the place to put in driver specific hooks. + * Will release the bo::reserved lock and the + * lru lock on exit. + */ + +static void ttm_bo_cleanup_memtype_use(struct ttm_buffer_object *bo) +{ + struct ttm_bo_global *glob = bo->glob; + + if (bo->ttm) { + + /** + * Release the lru_lock, since we don't want to have + * an atomic requirement on ttm_tt[unbind|destroy]. + */ + + spin_unlock(&glob->lru_lock); + ttm_tt_unbind(bo->ttm); + ttm_tt_destroy(bo->ttm); + bo->ttm = NULL; + spin_lock(&glob->lru_lock); + } + + if (bo->mem.mm_node) { + drm_mm_put_block(bo->mem.mm_node); + bo->mem.mm_node = NULL; + } + + atomic_set(&bo->reserved, 0); + wake_up_all(&bo->event_queue); + spin_unlock(&glob->lru_lock); +} + + +/** * If bo idle, remove from delayed- and lru lists, and unref. * If not idle, and already on delayed list, do nothing. * If not idle, and not on delayed list, put on delayed list, @@ -464,24 +501,39 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, bool remove_all) spin_unlock(&bo->lock); spin_lock(&glob->lru_lock); - put_count = ttm_bo_del_from_lru(bo); + ret = ttm_bo_reserve_locked(bo, false, true, false, 0); + + /** + * Someone else has the object reserved. Bail and retry. + */ + + if (unlikely(ret != 0)) { + spin_unlock(&glob->lru_lock); + return 0; + } - ret = ttm_bo_reserve_locked(bo, false, false, false, 0); - BUG_ON(ret); - if (bo->ttm) - ttm_tt_unbind(bo->ttm); + /** + * We can re-check for sync object without taking + * the bo::lock since setting the sync object requires + * also bo::reserved. A busy object at this point may + * be caused by another thread starting an accelerated + * eviction. + */ + + if (unlikely(bo->sync_obj)) { + atomic_set(&bo->reserved, 0); + wake_up_all(&bo->event_queue); + spin_unlock(&glob->lru_lock); + } + + put_count = ttm_bo_del_from_lru(bo); if (!list_empty(&bo->ddestroy)) { list_del_init(&bo->ddestroy); ++put_count; } - if (bo->mem.mm_node) { - drm_mm_put_block(bo->mem.mm_node); - bo->mem.mm_node = NULL; - } - spin_unlock(&glob->lru_lock); - atomic_set(&bo->reserved, 0); + ttm_bo_cleanup_memtype_use(bo); while (put_count--) kref_put(&bo->list_kref, ttm_bo_ref_bug); diff --git a/include/drm/ttm/ttm_bo_api.h b/include/drm/ttm/ttm_bo_api.h index 267a86c..2040e6c 100644 --- a/include/drm/ttm/ttm_bo_api.h +++ b/include/drm/ttm/ttm_bo_api.h @@ -246,9 +246,11 @@ struct ttm_buffer_object { atomic_t reserved; - /** * Members protected by the bo::lock + * In addition, setting sync_obj to anything else + * than NULL requires bo::reserved to be held. This allows for + * checking NULL while reserved but not holding bo::lock. */ void *sync_obj_arg; -- 1.6.2.5

2 1

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

dri-devel October 2010