Hi,
Commit message please.
Will GuC folks be reviewing this work?
Quick sanity check maybe makes sense, given data is being "sent" to
userspace directly, I am just not sure if it is worth having in
non-debug builds of i915. Though I will agree not having it in
production then largely defeats the purpose so dunno. Effective
difference if GuC load fails versus userspace libraries failing to parse
hwconfig?
On 07/02/2022 19:28, Jordan Justen wrote:
Signed-off-by: Jordan Justen <jordan.l.justen@xxxxxxxxx>
---
.../gpu/drm/i915/gt/uc/intel_guc_hwconfig.c | 26 +++++++++++++++++++
1 file changed, 26 insertions(+)
diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_hwconfig.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_hwconfig.c
index ce6088f112d4..695ef7a8f519 100644
--- a/drivers/gpu/drm/i915/gt/uc/intel_guc_hwconfig.c
+++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_hwconfig.c
@@ -71,6 +71,26 @@ static int guc_hwconfig_discover_size(struct intel_guc_hwconfig *hwconfig)
return 0;
}
+static int verify_hwconfig_blob(const struct intel_guc_hwconfig *hwconfig)
+{
+ if (hwconfig->size % 4 != 0 || hwconfig->ptr == NULL)
+ return -EINVAL;
So individual item size is minimum one u32, or zero? Document that in
patch 3?
+
+ struct drm_i915_query_hwconfig_blob_item *pos = hwconfig->ptr;
kbuild robot told you about mixing declarations and code already. :)
+ u32 remaining = (hwconfig->size / 4);
Blank line here and braces not needed.
+ while (remaining > 0) {
+ if (remaining < 2)
+ return -EINVAL;
+ if (pos->length > remaining - 2)
+ return -EINVAL;
+ remaining -= 2 + pos->length;
+ pos = (void *)&pos->data[pos->length];
+ }
+
+ DRM_INFO("hwconfig blob format appears valid\n");
Probably debug level at most.
+ return 0;
+}
+
static int guc_hwconfig_fill_buffer(struct intel_guc_hwconfig *hwconfig)
{
struct intel_guc *guc = hwconfig_to_guc(hwconfig);
@@ -91,6 +111,12 @@ static int guc_hwconfig_fill_buffer(struct intel_guc_hwconfig *hwconfig)
if (ret >= 0)
memcpy(hwconfig->ptr, vaddr, hwconfig->size);
+ if (verify_hwconfig_blob(hwconfig)) {
Merge under the "ret >= 0" branch above?
+ DRM_ERROR("Ignoring invalid hwconfig blob received from "
+ "GuC!\n");
Use drm_dbg/drm_err so the log is tied to a device in multi-gpu systems.
Also keep the string on one line as per kernel coding style guide.
Regards,
Tvrtko
+ return -EINVAL;
+ }
+
i915_vma_unpin_and_release(&vma, I915_VMA_RELEASE_MAP);
return ret;
