public inbox for netdev@vger.kernel.org
 help / color / mirror / Atom feed
From: Eric Joyner <eric.joyner@amd.com>
To: <netdev@vger.kernel.org>
Cc: Brett Creeley <brett.creeley@amd.com>,
	Andrew Lunn <andrew+netdev@lunn.ch>,
	"David S. Miller" <davem@davemloft.net>,
	"Eric Dumazet" <edumazet@google.com>,
	Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
	Eric Joyner <eric.joyner@amd.com>
Subject: [PATCH net] pds_core: Fix potential invalid stack memory access
Date: Thu, 30 Apr 2026 20:19:44 -0700	[thread overview]
Message-ID: <20260501031944.52172-1-eric.joyner@amd.com> (raw)

pds_adminq_post() and pds_process_adminq() can run concurrently
together; either one can check if the current request is complete:
pds_adminq_post() does it after the loop inside it times out waiting
for the completion, and pds_process_adminq() when it sees the
uncompleted request on the ring.

However, since pds_adminq_post() does not do any synchronization
around checking for an incomplete command after it reaches its timeout
and marking the command as complete, there is a small window where both
pds_process_adminq() and pds_adminq_post() can execute their
if(!completion_done()) clauses when the completion_done() returns
false; if pdsc_adminq_post() finishes first and its callers return,
then pdsc_process_adminq() will attempt to access q_info->dest after
the address in there points to an invalid location.

(q_info->dest will be invalid after pdsc_adminq_post()'s call chain
exits because it is pointing to a stack variable "comp" in
pdsc_fw_rpc())

Fix this by locking around the completion_done() check and the
complete() contained in the if-statement with the same adminq_lock that
pds_process_adminq() uses, which will prevent this synchronization
issue from occurring.

Fixes: 3f77c3dfffc7 ("pds_core: make wait_context part of q_info")
Signed-off-by: Eric Joyner <eric.joyner@amd.com>
---
 drivers/net/ethernet/amd/pds_core/adminq.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/amd/pds_core/adminq.c b/drivers/net/ethernet/amd/pds_core/adminq.c
index 097bb092bdb8..6db7d29cdf5c 100644
--- a/drivers/net/ethernet/amd/pds_core/adminq.c
+++ b/drivers/net/ethernet/amd/pds_core/adminq.c
@@ -283,9 +283,15 @@ int pdsc_adminq_post(struct pdsc *pdsc,
 		__func__, jiffies_to_msecs(time_done - time_start));
 
 	/* Check the results and clear an un-completed timeout */
-	if (time_after_eq(time_done, time_limit) && !completion_done(wc)) {
-		err = -ETIMEDOUT;
-		complete(wc);
+	if (time_after_eq(time_done, time_limit)) {
+		unsigned long irqflags;
+
+		spin_lock_irqsave(&pdsc->adminq_lock, irqflags);
+		if (!completion_done(wc)) {
+			err = -ETIMEDOUT;
+			complete(wc);
+		}
+		spin_unlock_irqrestore(&pdsc->adminq_lock, irqflags);
 	}
 
 	dev_dbg(pdsc->dev, "read admin queue completion idx %d:\n", index);

base-commit: e728258debd553c95d2e70f9cd97c9fde27c7130
-- 
2.17.1


                 reply	other threads:[~2026-05-01  3:20 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260501031944.52172-1-eric.joyner@amd.com \
    --to=eric.joyner@amd.com \
    --cc=andrew+netdev@lunn.ch \
    --cc=brett.creeley@amd.com \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=kuba@kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox