On 8/9/2022 5:42 PM, Juston Li wrote:
pxp will not start correctly until after mei_pxp bind completes and
intel_pxp_init_hw() is called.
This fixes a race condition during bootup where we observed a small
window for pxp commands to be sent before mei_pxp bind completed.
Changes since v1:
- check pxp_component instead of pxp_component_added (Daniele)
- pxp_component needs tee_mutex (Daniele)
- return -EAGAIN so caller knows to retry (Daniele)
Sorry for the follow up, but I have reflected on this a bit more and I
was thinking that maybe it would be better to wait a bit for the
component to be bound, instead of immediately failing the call. Do you
have a measure of how long after your failure you see the component bind
complete? The wait would make sense only if relatively short (< 1s).
Daniele
Signed-off-by: Juston Li <justonli@xxxxxxxxxx>
---
drivers/gpu/drm/i915/pxp/intel_pxp.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/drivers/gpu/drm/i915/pxp/intel_pxp.c b/drivers/gpu/drm/i915/pxp/intel_pxp.c
index 15311eaed848..8b395ebc430a 100644
--- a/drivers/gpu/drm/i915/pxp/intel_pxp.c
+++ b/drivers/gpu/drm/i915/pxp/intel_pxp.c
@@ -187,6 +187,14 @@ int intel_pxp_start(struct intel_pxp *pxp)
if (!intel_pxp_is_enabled(pxp))
return -ENODEV;
+ mutex_lock(&pxp->tee_mutex);
+ /* check if mei_pxp is bound */
+ if (!pxp->pxp_component) {
+ mutex_unlock(&pxp->tee_mutex);
+ return -EAGAIN;
+ }
+ mutex_unlock(&pxp->tee_mutex);
+
mutex_lock(&pxp->arb_mutex);
if (pxp->arb_is_valid)
