[Intel-gfx] [PATCH] drm/i915/gtt: Defer the free for alloc error paths

[Chris Wilson]

If we hit an error while allocating the page tables, we have to unwind
the incomplete updates, and wish to free the unused pd. However, we are
not allowed to be hoding the spinlock at that point, and so must use the
later free to defer it until after we drop the lock.

Fixes: 1d1b5490b91c ("drm/i915/gtt: Replace struct_mutex serialisation for allocation")
Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Matthew Auld <matthew.auld at intel.com>
Cc: Mika Kuoppala <mika.kuoppala at intel.com>
---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
index 9e76347e039e..1065753e86fb 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -1489,7 +1489,8 @@ static int gen8_ppgtt_alloc_pdp(struct i915_address_space *vm,
 		gen8_ppgtt_set_pdpe(pdp, vm->scratch_pd, pdpe);
 		GEM_BUG_ON(!atomic_read(&pdp->used));
 		atomic_dec(&pdp->used);
-		free_pd(vm, pd);
+		GEM_BUG_ON(alloc);
+		alloc = pd; /* defer the free to after the lock */
 	}
 	spin_unlock(&pdp->lock);
 unwind:
@@ -1558,7 +1559,8 @@ static int gen8_ppgtt_alloc_4lvl(struct i915_address_space *vm,
 	spin_lock(&pml4->lock);
 	if (atomic_dec_and_test(&pdp->used)) {
 		gen8_ppgtt_set_pml4e(pml4, vm->scratch_pdp, pml4e);
-		free_pd(vm, pdp);
+		GEM_BUG_ON(alloc);
+		alloc = pdp; /* defer the free until after the lock */
 	}
 	spin_unlock(&pml4->lock);
 unwind:
-- 
2.20.1