[PATCH 64/95] drm/amdgpu: fix NPA-RELEASE race in UALink exporter cleanup

From: Mukul Joshi <[email protected]>

Processing of NPA-RELEASE messages can run concurrently with
sending of NPA-REVOKEs to all the importers. This can lead to
a race condition where the interrupt handler can signal npa_done
as soon as npa_release_bitmap becomes empty even though the cleanup
worker was still sending NPA-REVOKEs to all the importers. This would
prevent the cleanup worker thread to wait for all NPA-RELEASE messages
to come before freeing the NPA addresses.
Fix this by setting the importer_bitmap for all importers before sending
any NPA-REVOKE messages.

Signed-off-by: Mukul Joshi <[email protected]>
Reviewed-by: Felix Kuehling <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.c
index e2f1829ab844c..2d1d4eae5b242 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ualink.c
@@ -2268,7 +2268,7 @@ static void amdgpu_ualink_exp_cleanup_worker(struct work_struct *work)
 	amdgpu_bo_unref(&bo);
 	exp_xa_node->bo = NULL;
 
-	/* Send NPA-REVOKE to all importers which have imported this memory */
+	/* Build the full npa_release_bitmap before sending any NPA-REVOKE. */
 	for_each_set_bit(remote_acc_id, exp_xa_node->importers_bitmap,
 				 AMDGPU_UALINK_ACCEL_MAX) {
 		imp_entry = &exp_xa_node->importer_entries[remote_acc_id];
@@ -2279,10 +2279,17 @@ static void amdgpu_ualink_exp_cleanup_worker(struct work_struct *work)
 			continue;
 		}
 
+		set_bit(remote_acc_id, exp_xa_node->npa_release_bitmap);
+	}
+
+	/* Send NPA-REVOKE to all importers which have imported this memory.
+	 * On send failure clear the bit (no response will arrive).
+	 */
+	for_each_set_bit(remote_acc_id, exp_xa_node->importers_bitmap,
+				 AMDGPU_UALINK_ACCEL_MAX) {
 		dev_dbg(adev->dev,
 			"EXP-CLEANUP: Sending NPA-REVOKE to remote:%u\n",
 			remote_acc_id);
-		set_bit(remote_acc_id, exp_xa_node->npa_release_bitmap);
 		r = amdgpu_ualink_send_npa_revoke_msg(adev, remote_acc_id, handle);
 		if (r) {
 			dev_err(adev->dev,
@@ -2292,7 +2299,8 @@ static void amdgpu_ualink_exp_cleanup_worker(struct work_struct *work)
 		}
 	}
 
-	if (!bitmap_empty(exp_xa_node->importers_bitmap,
+	/* Wait for the NPA_RELEASE to come back from all importers */
+	if (!bitmap_empty(exp_xa_node->npa_release_bitmap,
 			AMDGPU_UALINK_ACCEL_MAX)) {
 		dev_dbg(adev->dev,
 			"EXP-CLEANUP: handle:%llx:%llx NPA-RELEASE bitmap: %*pbl\n",
-- 
2.55.0