Closed marc-hb closed 4 years ago
looks good :+1:
../sof/qemu-check.sh and maybe others invoke this script with a non-zero ${TIMEOUT} to stop qemu. This prints the following line every time; even in successful cases which can be confusing qemu-system-xtensa: terminating on signal 15 from pid 9823 (timeout)
Why'd timeout send TERM if it didn't actually timeout though? This message gets printed by qemu after receiving a shutdown request. And TERM is sent on timeout..?
It's possible to tell qemu to quit but it requires much more code than simply asking timeout
to send TERM to qemu.
This prints the following line every time; even in successful cases which can be confusing qemu-system-xtensa: terminating on signal 15 from pid 9823 (timeout)
I don't understand how this(printing timeout even in successful cases) happens? timeout cannot send a SIGTERM without the timeout happening and qemu prints qemu-system-xtensa: terminating on signal 15 from pid 9823 (timeout)
on receiving SIGTERM before quitting.
This script is not a test. The test is not here, it's in qemu-check.sh. See https://github.com/thesofproject/sof/pull/3101
When invoked by qemu-check.sh, this script runs qemu for 2 seconds and kills it after 2 seconds. That's all it does and it always prints the timeout message. Then, later, qemu-check.sh decides whether the test has been successful or not.
Replace the custom and incomplete command logging with a simpler set -x.
../sof/qemu-check.sh and maybe others invoke this script with a non-zero ${TIMEOUT} to stop qemu. This prints the following line every time; even in successful cases which can be confusing
qemu-system-xtensa: terminating on signal 15 from pid 9823 (timeout)
Logging the timeout prefix reduces this confusion
Signed-off-by: Marc Herbert marc.herbert@intel.com