panfrost: ci: Don't restart Weston
authorTomeu Vizoso <tomeu.vizoso@collabora.com>
Tue, 7 May 2019 10:57:58 +0000 (12:57 +0200)
committerTomeu Vizoso <tomeu.vizoso@collabora.com>
Fri, 10 May 2019 05:37:30 +0000 (07:37 +0200)
There doesn't seem to actually be any noticeably memory leaks on Weston
when running dEQP. We do seem to leak quiet a bit in the client, so we
still have to run the dEQP runner in batches.

This removes the risk of Weston not restarting properly and introducing
spurious failures.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
src/gallium/drivers/panfrost/ci/deqp-runner.sh

index 992560f8cd40ffbe83b7ed10a6d56637ab505f0e..68d9bc1fc878fd6c1935176b70d352a8af8da46f 100644 (file)
@@ -38,14 +38,7 @@ tail -f /tmp/result.txt &
 while [ -s /tmp/case-list.txt ]; do
        head -$BATCH_SIZE /tmp/case-list.txt > /tmp/next-batch.txt
        ./deqp-gles2 $DEQP_OPTIONS --deqp-log-filename=/dev/null --deqp-caselist-file=/tmp/next-batch.txt >> /tmp/result.txt
-       deqp_status=$?
-
-       kill $(pidof weston)
-       sleep 1  # Give some time for Weston to release the VT
-       weston --tty=7 &
-       sleep 1  # Give some time for Weston to start up
-
-       if [ $deqp_status -ne 0 ]; then
+       if [ $? -ne 0 ]; then
                # Continue from the subtest after the failing one
                crashed_test=$(grep "Test case" /tmp/result.txt | tail -1 | sed "s/Test case '\(.*\)'\.\./\1/")
                sed -i "0,/^$crashed_test$/d" /tmp/case-list.txt