Intel-GFX Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
To: intel-gfx@lists.freedesktop.org
Subject: [Intel-gfx] [PATCH 07/23] Revert "drm/i915/gem: Split eb_vma into its own allocation"
Date: Tue, 14 Jul 2020 13:45:00 +0200	[thread overview]
Message-ID: <20200714114516.3073222-7-maarten.lankhorst@linux.intel.com> (raw)
In-Reply-To: <20200714114516.3073222-1-maarten.lankhorst@linux.intel.com>

This reverts commit 0f1dd02295f35dcdcbaafcbcbbec0753884ab974.
This conflicts with the ww mutex handling, which needs to drop
the references after gpu submission anyway, because otherwise we
may risk unlocking a BO after first freeing it.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
---
 .../gpu/drm/i915/gem/i915_gem_execbuffer.c    | 124 +++++++-----------
 1 file changed, 51 insertions(+), 73 deletions(-)

diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c
index e8221b6d0a53..06a5b086fc78 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c
@@ -40,11 +40,6 @@ struct eb_vma {
 	u32 handle;
 };
 
-struct eb_vma_array {
-	struct kref kref;
-	struct eb_vma vma[];
-};
-
 enum {
 	FORCE_CPU_RELOC = 1,
 	FORCE_GTT_RELOC,
@@ -57,6 +52,7 @@ enum {
 #define __EXEC_OBJECT_NEEDS_MAP		BIT(29)
 #define __EXEC_OBJECT_NEEDS_BIAS	BIT(28)
 #define __EXEC_OBJECT_INTERNAL_FLAGS	(~0u << 28) /* all of the above */
+#define __EXEC_OBJECT_RESERVED (__EXEC_OBJECT_HAS_PIN | __EXEC_OBJECT_HAS_FENCE)
 
 #define __EXEC_HAS_RELOC	BIT(31)
 #define __EXEC_INTERNAL_FLAGS	(~0u << 31)
@@ -287,7 +283,6 @@ struct i915_execbuffer {
 	 */
 	int lut_size;
 	struct hlist_head *buckets; /** ht for relocation handles */
-	struct eb_vma_array *array;
 };
 
 static int eb_parse(struct i915_execbuffer *eb);
@@ -299,62 +294,8 @@ static inline bool eb_use_cmdparser(const struct i915_execbuffer *eb)
 		 eb->args->batch_len);
 }
 
-static struct eb_vma_array *eb_vma_array_create(unsigned int count)
-{
-	struct eb_vma_array *arr;
-
-	arr = kvmalloc(struct_size(arr, vma, count), GFP_KERNEL | __GFP_NOWARN);
-	if (!arr)
-		return NULL;
-
-	kref_init(&arr->kref);
-	arr->vma[0].vma = NULL;
-
-	return arr;
-}
-
-static inline void eb_unreserve_vma(struct eb_vma *ev)
-{
-	struct i915_vma *vma = ev->vma;
-
-	if (unlikely(ev->flags & __EXEC_OBJECT_HAS_FENCE))
-		__i915_vma_unpin_fence(vma);
-
-	if (ev->flags & __EXEC_OBJECT_HAS_PIN)
-		__i915_vma_unpin(vma);
-
-	ev->flags &= ~(__EXEC_OBJECT_HAS_PIN |
-		       __EXEC_OBJECT_HAS_FENCE);
-}
-
-static void eb_vma_array_destroy(struct kref *kref)
-{
-	struct eb_vma_array *arr = container_of(kref, typeof(*arr), kref);
-	struct eb_vma *ev = arr->vma;
-
-	while (ev->vma) {
-		eb_unreserve_vma(ev);
-		i915_vma_put(ev->vma);
-		ev++;
-	}
-
-	kvfree(arr);
-}
-
-static void eb_vma_array_put(struct eb_vma_array *arr)
-{
-	kref_put(&arr->kref, eb_vma_array_destroy);
-}
-
 static int eb_create(struct i915_execbuffer *eb)
 {
-	/* Allocate an extra slot for use by the command parser + sentinel */
-	eb->array = eb_vma_array_create(eb->buffer_count + 2);
-	if (!eb->array)
-		return -ENOMEM;
-
-	eb->vma = eb->array->vma;
-
 	if (!(eb->args->flags & I915_EXEC_HANDLE_LUT)) {
 		unsigned int size = 1 + ilog2(eb->buffer_count);
 
@@ -388,10 +329,8 @@ static int eb_create(struct i915_execbuffer *eb)
 				break;
 		} while (--size);
 
-		if (unlikely(!size)) {
-			eb_vma_array_put(eb->array);
+		if (unlikely(!size))
 			return -ENOMEM;
-		}
 
 		eb->lut_size = size;
 	} else {
@@ -502,6 +441,26 @@ eb_pin_vma(struct i915_execbuffer *eb,
 	return !eb_vma_misplaced(entry, vma, ev->flags);
 }
 
+static inline void __eb_unreserve_vma(struct i915_vma *vma, unsigned int flags)
+{
+	GEM_BUG_ON(!(flags & __EXEC_OBJECT_HAS_PIN));
+
+	if (unlikely(flags & __EXEC_OBJECT_HAS_FENCE))
+		__i915_vma_unpin_fence(vma);
+
+	__i915_vma_unpin(vma);
+}
+
+static inline void
+eb_unreserve_vma(struct eb_vma *ev)
+{
+	if (!(ev->flags & __EXEC_OBJECT_HAS_PIN))
+		return;
+
+	__eb_unreserve_vma(ev->vma, ev->flags);
+	ev->flags &= ~__EXEC_OBJECT_RESERVED;
+}
+
 static int
 eb_validate_vma(struct i915_execbuffer *eb,
 		struct drm_i915_gem_exec_object2 *entry,
@@ -944,13 +903,31 @@ eb_get_vma(const struct i915_execbuffer *eb, unsigned long handle)
 	}
 }
 
+static void eb_release_vmas(const struct i915_execbuffer *eb)
+{
+	const unsigned int count = eb->buffer_count;
+	unsigned int i;
+
+	for (i = 0; i < count; i++) {
+		struct eb_vma *ev = &eb->vma[i];
+		struct i915_vma *vma = ev->vma;
+
+		if (!vma)
+			break;
+
+		eb->vma[i].vma = NULL;
+
+		if (ev->flags & __EXEC_OBJECT_HAS_PIN)
+			__eb_unreserve_vma(vma, ev->flags);
+
+		i915_vma_put(vma);
+	}
+}
+
 static void eb_destroy(const struct i915_execbuffer *eb)
 {
 	GEM_BUG_ON(eb->reloc_cache.rq);
 
-	if (eb->array)
-		eb_vma_array_put(eb->array);
-
 	if (eb->lut_size > 0)
 		kfree(eb->buckets);
 }
@@ -2047,12 +2024,9 @@ static int eb_move_to_gpu(struct i915_execbuffer *eb)
 			err = i915_vma_move_to_active(vma, eb->request, flags);
 
 		i915_vma_unlock(vma);
-		eb_unreserve_vma(ev);
 	}
 	ww_acquire_fini(&acquire);
 
-	eb_vma_array_put(fetch_and_zero(&eb->array));
-
 	if (unlikely(err))
 		goto err_skip;
 
@@ -2348,7 +2322,6 @@ static int eb_parse(struct i915_execbuffer *eb)
 	eb->vma[eb->buffer_count].vma = i915_vma_get(shadow);
 	eb->vma[eb->buffer_count].flags = __EXEC_OBJECT_HAS_PIN;
 	eb->batch = &eb->vma[eb->buffer_count++];
-	eb->vma[eb->buffer_count].vma = NULL;
 
 	eb->trampoline = trampoline;
 	eb->batch_start_offset = 0;
@@ -2845,6 +2818,8 @@ i915_gem_do_execbuffer(struct drm_device *dev,
 		args->flags |= __EXEC_HAS_RELOC;
 
 	eb.exec = exec;
+	eb.vma = (struct eb_vma *)(exec + args->buffer_count + 1);
+	eb.vma[0].vma = NULL;
 
 	eb.invalid_flags = __EXEC_OBJECT_UNKNOWN_FLAGS;
 	reloc_cache_init(&eb.reloc_cache, eb.i915);
@@ -3021,6 +2996,8 @@ i915_gem_do_execbuffer(struct drm_device *dev,
 	if (batch->private)
 		intel_gt_buffer_pool_put(batch->private);
 err_vma:
+	if (eb.exec)
+		eb_release_vmas(&eb);
 	if (eb.trampoline)
 		i915_vma_unpin(eb.trampoline);
 	eb_unpin_engine(&eb);
@@ -3038,7 +3015,7 @@ i915_gem_do_execbuffer(struct drm_device *dev,
 
 static size_t eb_element_size(void)
 {
-	return sizeof(struct drm_i915_gem_exec_object2);
+	return sizeof(struct drm_i915_gem_exec_object2) + sizeof(struct eb_vma);
 }
 
 static bool check_buffer_count(size_t count)
@@ -3094,7 +3071,7 @@ i915_gem_execbuffer_ioctl(struct drm_device *dev, void *data,
 	/* Copy in the exec list from userland */
 	exec_list = kvmalloc_array(count, sizeof(*exec_list),
 				   __GFP_NOWARN | GFP_KERNEL);
-	exec2_list = kvmalloc_array(count, eb_element_size(),
+	exec2_list = kvmalloc_array(count + 1, eb_element_size(),
 				    __GFP_NOWARN | GFP_KERNEL);
 	if (exec_list == NULL || exec2_list == NULL) {
 		drm_dbg(&i915->drm,
@@ -3172,7 +3149,8 @@ i915_gem_execbuffer2_ioctl(struct drm_device *dev, void *data,
 	if (err)
 		return err;
 
-	exec2_list = kvmalloc_array(count, eb_element_size(),
+	/* Allocate an extra slot for use by the command parser */
+	exec2_list = kvmalloc_array(count + 1, eb_element_size(),
 				    __GFP_NOWARN | GFP_KERNEL);
 	if (exec2_list == NULL) {
 		drm_dbg(&i915->drm, "Failed to allocate exec list for %zd buffers\n",
-- 
2.27.0

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

  parent reply	other threads:[~2020-07-14 11:45 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-07-14 11:44 [Intel-gfx] [PATCH 01/23] Revert "drm/i915/gem: Async GPU relocations only" Maarten Lankhorst
2020-07-14 11:44 ` [Intel-gfx] [PATCH 02/23] drm/i915: Revert relocation chaining commits Maarten Lankhorst
2020-07-14 11:44 ` [Intel-gfx] [PATCH 03/23] Revert "drm/i915/gem: Drop relocation slowpath" Maarten Lankhorst
2020-07-14 11:44 ` [Intel-gfx] [PATCH 04/23] drm/i915: Add an implementation for i915_gem_ww_ctx locking, v2 Maarten Lankhorst
2020-07-14 11:44 ` [Intel-gfx] [PATCH 05/23] drm/i915: Remove locking from i915_gem_object_prepare_read/write Maarten Lankhorst
2020-07-14 11:44 ` [Intel-gfx] [PATCH 06/23] drm/i915: Parse command buffer earlier in eb_relocate(slow) Maarten Lankhorst
2020-07-14 11:45 ` Maarten Lankhorst [this message]
2020-07-14 11:45 ` [Intel-gfx] [PATCH 08/23] drm/i915: Use per object locking in execbuf, v12 Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 09/23] drm/i915: Use ww locking in intel_renderstate Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 10/23] drm/i915: Add ww context handling to context_barrier_task Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 11/23] drm/i915: Nuke arguments to eb_pin_engine Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 12/23] drm/i915: Pin engine before pinning all objects, v5 Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 13/23] drm/i915: Rework intel_context pinning to do everything outside of pin_mutex Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 14/23] drm/i915: Make sure execbuffer always passes ww state to i915_vma_pin Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 15/23] drm/i915: Convert i915_gem_object/client_blt.c to use ww locking as well, v2 Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 16/23] drm/i915: Kill last user of intel_context_create_request outside of selftests Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 17/23] drm/i915: Convert i915_perf to ww locking as well Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 18/23] drm/i915: Dirty hack to fix selftests locking inversion Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 19/23] drm/i915/selftests: Fix locking inversion in lrc selftest Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 20/23] drm/i915: Use ww pinning for intel_context_create_request() Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 21/23] drm/i915: Move i915_vma_lock in the selftests to avoid lock inversion, v2 Maarten Lankhorst
2020-07-14 14:40   ` [Intel-gfx] [PATCH] " Maarten Lankhorst
2020-07-15 10:49     ` [Intel-gfx] [PATCH] drm/i915: Move i915_vma_lock in the selftests to avoid lock inversion, v3 Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 22/23] drm/i915: Add ww locking to vm_fault_gtt Maarten Lankhorst
2020-07-14 11:45 ` [Intel-gfx] [PATCH 23/23] drm/i915: Add ww locking to pin_to_display_plane Maarten Lankhorst
2020-07-14 13:14 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for series starting with [01/23] Revert "drm/i915/gem: Async GPU relocations only" Patchwork
2020-07-14 13:15 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2020-07-14 13:36 ` [Intel-gfx] ✗ Fi.CI.BAT: failure " Patchwork
2020-07-14 15:22 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for series starting with [01/23] Revert "drm/i915/gem: Async GPU relocations only" (rev2) Patchwork
2020-07-14 15:24 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2020-07-14 15:44 ` [Intel-gfx] ✗ Fi.CI.BAT: failure " Patchwork
2020-07-15 11:22 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for series starting with [01/23] Revert "drm/i915/gem: Async GPU relocations only" (rev3) Patchwork
2020-07-15 11:22 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2020-07-15 11:42 ` [Intel-gfx] ✓ Fi.CI.BAT: success " Patchwork
2020-07-15 14:06 ` [Intel-gfx] ✗ Fi.CI.IGT: failure " Patchwork
  -- strict thread matches above, loose matches on Subject: below --
2020-07-03 12:21 [Intel-gfx] [PATCH 00/23] drm/i915: Use ww locking in execbuf submission Maarten Lankhorst
2020-07-03 12:22 ` [Intel-gfx] [PATCH 07/23] Revert "drm/i915/gem: Split eb_vma into its own allocation" Maarten Lankhorst

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200714114516.3073222-7-maarten.lankhorst@linux.intel.com \
    --to=maarten.lankhorst@linux.intel.com \
    --cc=intel-gfx@lists.freedesktop.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox