On 07.11.2022 17:54, Matthew Auld wrote:
Rather than getting some hard to debug uaf, add some warns to hopefully
catch issues with userfault_count being non-zero when destroying the
object. Also if we somehow add an object to lmem_userfault_list that
somehow doesn't map lmem.
References: https://gitlab.freedesktop.org/drm/intel/-/issues/7469
Signed-off-by: Matthew Auld <matthew.auld@xxxxxxxxx>
Cc: Anshuman Gupta <anshuman.gupta@xxxxxxxxx>
Cc: Rodrigo Vivi <rodrigo.vivi@xxxxxxxxx>
Cc: Andrzej Hajda <andrzej.hajda@xxxxxxxxx>
Cc: Nirmoy Das <nirmoy.das@xxxxxxxxx>
Reviewed-by: Andrzej Hajda <andrzej.hajda@xxxxxxxxx>
Regards
Andrzej
---
drivers/gpu/drm/i915/gem/i915_gem_ttm.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/drivers/gpu/drm/i915/gem/i915_gem_ttm.c b/drivers/gpu/drm/i915/gem/i915_gem_ttm.c
index 9421dc4dc98f..2c8b2d5ae903 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_ttm.c
+++ b/drivers/gpu/drm/i915/gem/i915_gem_ttm.c
@@ -1098,6 +1098,8 @@ static vm_fault_t vm_fault_ttm(struct vm_fault *vmf)
spin_lock(&to_i915(obj->base.dev)->runtime_pm.lmem_userfault_lock);
list_add(&obj->userfault_link, &to_i915(obj->base.dev)->runtime_pm.lmem_userfault_list);
spin_unlock(&to_i915(obj->base.dev)->runtime_pm.lmem_userfault_lock);
+
+ GEM_WARN_ON(!i915_ttm_cpu_maps_iomem(bo->resource));
}
if (wakeref & CONFIG_DRM_I915_USERFAULT_AUTOSUSPEND)
@@ -1180,6 +1182,8 @@ static void i915_ttm_unmap_virtual(struct drm_i915_gem_object *obj)
}
}
+ GEM_WARN_ON(obj->userfault_count);
+
ttm_bo_unmap_virtual(i915_gem_to_ttm(obj));
if (wakeref)
