From:   Mikko Perttunen <cyndis@kapsi.fi>
To:     Thierry Reding <thierry.reding@gmail.com>,
        David Airlie <airlied@gmail.com>,
        Daniel Vetter <daniel@ffwll.ch>
Cc:     Mikko Perttunen <mperttunen@nvidia.com>,
        dri-devel@lists.freedesktop.org, linux-tegra@vger.kernel.org,
        linux-kernel@vger.kernel.org
Subject: [PATCH] gpu: host1x: Don't rely on dma_fence_wait_timeout return value
Date:   Wed,  1 Mar 2023 15:51:06 +0200
Message-Id: <20230301135107.999976-1-cyndis@kapsi.fi>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

From: Mikko Perttunen <mperttunen@nvidia.com>

dma_fence_wait_timeout (along with a host of other jiffies-based
timeouting functions) returns zero both in case of timeout and when
the wait completes during the last jiffy before timeout. As such,
we can't rely on it to distinguish between success and timeout.

To prevent confusing callers by returning -EAGAIN before the timeout
period has elapsed, check if the fence got signaled again after
the wait.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
---
 drivers/gpu/host1x/syncpt.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/host1x/syncpt.c b/drivers/gpu/host1x/syncpt.c
index 2d2007760eac..f63d14a57a1d 100644
--- a/drivers/gpu/host1x/syncpt.c
+++ b/drivers/gpu/host1x/syncpt.c
@@ -248,7 +248,13 @@ int host1x_syncpt_wait(struct host1x_syncpt *sp, u32 thresh, long timeout,
 	if (value)
 		*value = host1x_syncpt_load(sp);
 
-	if (wait_err == 0)
+	/*
+	 * Don't rely on dma_fence_wait_timeout return value,
+	 * since it returns zero both on timeout and if the
+	 * wait completed with 0 jiffies left.
+	 */
+	host1x_hw_syncpt_load(sp->host, sp);
+	if (wait_err == 0 && !host1x_syncpt_is_expired(sp, thresh))
 		return -EAGAIN;
 	else if (wait_err < 0)
 		return wait_err;
-- 
2.39.2