[PATCH net] net: octeontx2-af: quiesce devlink health work on teardown

[PATCH net] net: octeontx2-af: quiesce devlink health work on teardown

[Hongyan Xu]

NPA and NIX interrupt handlers queue health-reporting work whose
containers are freed by rvu_health_reporters_destroy(). Interrupt
unregistration does not drain queued work. In addition, each reporter
setup creates a workqueue in the same field, leaking the NPA queue and
leaving no teardown handle for work queued on it.

Initialize work before later allocations can fail, share one workqueue,
unregister each set of interrupts, and cancel its work before freeing
the reporters. Destroy the shared workqueue once both reporter sets are
gone.

Fixes: f1168d1e207c ("octeontx2-af: Add devlink health reporters for NPA")
Fixes: 5ed66306eab6 ("octeontx2-af: Add devlink health reporters for NIX")
Signed-off-by: Hongyan Xu <getshell@seu.edu.cn>
---
 .../marvell/octeontx2/af/rvu_devlink.c        | 46 +++++++++++++------
 1 file changed, 32 insertions(+), 14 deletions(-)

diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_devlink.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_devlink.c
index d63c3d33775a..c7c7230dc588 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_devlink.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_devlink.c
@@ -505,6 +505,11 @@ static int rvu_nix_register_reporters(struct rvu_devlink *rvu_dl)
 		return -ENOMEM;
 
 	rvu_reporters->nix_event_ctx = nix_event_context;
+	INIT_WORK(&rvu_reporters->intr_work, rvu_nix_intr_work);
+	INIT_WORK(&rvu_reporters->gen_work, rvu_nix_gen_work);
+	INIT_WORK(&rvu_reporters->err_work, rvu_nix_err_work);
+	INIT_WORK(&rvu_reporters->ras_work, rvu_nix_ras_work);
+
 	rvu_reporters->rvu_hw_nix_intr_reporter =
 		devlink_health_reporter_create(rvu_dl->dl,
 					       &rvu_hw_nix_intr_reporter_ops,
@@ -545,15 +550,11 @@ static int rvu_nix_register_reporters(struct rvu_devlink *rvu_dl)
 		return PTR_ERR(rvu_reporters->rvu_hw_nix_ras_reporter);
 	}
 
-	rvu_dl->devlink_wq = create_workqueue("rvu_devlink_wq");
+	if (!rvu_dl->devlink_wq)
+		rvu_dl->devlink_wq = create_workqueue("rvu_devlink_wq");
 	if (!rvu_dl->devlink_wq)
 		return -ENOMEM;
 
-	INIT_WORK(&rvu_reporters->intr_work, rvu_nix_intr_work);
-	INIT_WORK(&rvu_reporters->gen_work, rvu_nix_gen_work);
-	INIT_WORK(&rvu_reporters->err_work, rvu_nix_err_work);
-	INIT_WORK(&rvu_reporters->ras_work, rvu_nix_ras_work);
-
 	return 0;
 }
 
@@ -582,6 +583,13 @@ static void rvu_nix_health_reporters_destroy(struct rvu_devlink *rvu_dl)
 
 	if (!nix_reporters->rvu_hw_nix_ras_reporter)
 		return;
+
+	rvu_nix_unregister_interrupts(rvu);
+	cancel_work_sync(&nix_reporters->intr_work);
+	cancel_work_sync(&nix_reporters->gen_work);
+	cancel_work_sync(&nix_reporters->err_work);
+	cancel_work_sync(&nix_reporters->ras_work);
+
 	if (!IS_ERR_OR_NULL(nix_reporters->rvu_hw_nix_intr_reporter))
 		devlink_health_reporter_destroy(nix_reporters->rvu_hw_nix_intr_reporter);
 
@@ -594,7 +602,6 @@ static void rvu_nix_health_reporters_destroy(struct rvu_devlink *rvu_dl)
 	if (!IS_ERR_OR_NULL(nix_reporters->rvu_hw_nix_ras_reporter))
 		devlink_health_reporter_destroy(nix_reporters->rvu_hw_nix_ras_reporter);
 
-	rvu_nix_unregister_interrupts(rvu);
 	kfree(rvu_dl->rvu_nix_health_reporter->nix_event_ctx);
 	kfree(rvu_dl->rvu_nix_health_reporter);
 }
@@ -1059,6 +1066,11 @@ static int rvu_npa_register_reporters(struct rvu_devlink *rvu_dl)
 		return -ENOMEM;
 
 	rvu_reporters->npa_event_ctx = npa_event_context;
+	INIT_WORK(&rvu_reporters->intr_work, rvu_npa_intr_work);
+	INIT_WORK(&rvu_reporters->err_work, rvu_npa_err_work);
+	INIT_WORK(&rvu_reporters->gen_work, rvu_npa_gen_work);
+	INIT_WORK(&rvu_reporters->ras_work, rvu_npa_ras_work);
+
 	rvu_reporters->rvu_hw_npa_intr_reporter =
 		devlink_health_reporter_create(rvu_dl->dl,
 					       &rvu_hw_npa_intr_reporter_ops,
@@ -1099,15 +1111,11 @@ static int rvu_npa_register_reporters(struct rvu_devlink *rvu_dl)
 		return PTR_ERR(rvu_reporters->rvu_hw_npa_ras_reporter);
 	}
 
-	rvu_dl->devlink_wq = create_workqueue("rvu_devlink_wq");
+	if (!rvu_dl->devlink_wq)
+		rvu_dl->devlink_wq = create_workqueue("rvu_devlink_wq");
 	if (!rvu_dl->devlink_wq)
 		return -ENOMEM;
 
-	INIT_WORK(&rvu_reporters->intr_work, rvu_npa_intr_work);
-	INIT_WORK(&rvu_reporters->err_work, rvu_npa_err_work);
-	INIT_WORK(&rvu_reporters->gen_work, rvu_npa_gen_work);
-	INIT_WORK(&rvu_reporters->ras_work, rvu_npa_ras_work);
-
 	return 0;
 }
 
@@ -1136,6 +1144,13 @@ static void rvu_npa_health_reporters_destroy(struct rvu_devlink *rvu_dl)
 
 	if (!npa_reporters->rvu_hw_npa_ras_reporter)
 		return;
+
+	rvu_npa_unregister_interrupts(rvu);
+	cancel_work_sync(&npa_reporters->intr_work);
+	cancel_work_sync(&npa_reporters->gen_work);
+	cancel_work_sync(&npa_reporters->err_work);
+	cancel_work_sync(&npa_reporters->ras_work);
+
 	if (!IS_ERR_OR_NULL(npa_reporters->rvu_hw_npa_intr_reporter))
 		devlink_health_reporter_destroy(npa_reporters->rvu_hw_npa_intr_reporter);
 
@@ -1148,7 +1163,6 @@ static void rvu_npa_health_reporters_destroy(struct rvu_devlink *rvu_dl)
 	if (!IS_ERR_OR_NULL(npa_reporters->rvu_hw_npa_ras_reporter))
 		devlink_health_reporter_destroy(npa_reporters->rvu_hw_npa_ras_reporter);
 
-	rvu_npa_unregister_interrupts(rvu);
 	kfree(rvu_dl->rvu_npa_health_reporter->npa_event_ctx);
 	kfree(rvu_dl->rvu_npa_health_reporter);
 }
@@ -1176,6 +1190,10 @@ static void rvu_health_reporters_destroy(struct rvu *rvu)
 	rvu_dl = rvu->rvu_dl;
 	rvu_npa_health_reporters_destroy(rvu_dl);
 	rvu_nix_health_reporters_destroy(rvu_dl);
+	if (rvu_dl->devlink_wq) {
+		destroy_workqueue(rvu_dl->devlink_wq);
+		rvu_dl->devlink_wq = NULL;
+	}
 }
 
 /* Devlink Params APIs */
-- 
2.50.1.windows.1

Re: [PATCH net] net: octeontx2-af: quiesce devlink health work on teardown

[netdev-bot+sinfo]

Hi!

This is an automated message. This series looks like a fix, but its
commit messages seem to be missing some information:

 - How the issue was discovered, e.g. hit in production, hit during
   development, syzbot report, manual code inspection, LLM or static
   analysis tool scan.

 - Whether the issue was actually triggered, or is only theoretical
   (e.g. found by code inspection). If it was triggered please include
   the symptoms, like the stack trace or error messages.

 - What hardware the change was tested on. For driver fixes please
   mention the device (and if relevant firmware version) used for
   testing, or say that the change was not tested on real hardware.

Please do not repost the series just to address the above. Instead,
reply to this email with the missing information, so that reviewers
can take it into account. If the series needs another revision for
other reasons, please include the information in the commit messages
then.

The evaluation is done by an LLM so it may be wrong, if you think
that is the case please reply and explain.