qemurunner: Fix a bug with fork/exit handling

If you send this forked process a SIGTERM, it will execute all of the
parent's exit code leading to two sets of console/exit output which is
extremely confusing. Wrap the code in a try/finally to ensure we always
call os._exit() to avoid this.

I spent far too long trying to work out the crazy console output from this.

(From OE-Core rev: 652e40bfae24b8e23bbf7a7f35d900d2ab8d0f92)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

1 files changed, 9 insertions, 6 deletions

diff --git a/meta/lib/oeqa/utils/qemurunner.py b/meta/lib/oeqa/utils/qemurunner.py
index 04e0334914..c4db0cf038 100644
--- a/meta/lib/oeqa/utils/qemurunner.py
+++ b/meta/lib/oeqa/utils/qemurunner.py
@@ -267,12 +267,15 @@ class QemuRunner:
             self.monitorpipe = os.fdopen(w, "w")
         else:
             # child process
-            os.setpgrp()
-            os.close(w)
-            r = os.fdopen(r)
-            x = r.read()
-            os.killpg(os.getpgid(self.runqemu.pid), signal.SIGTERM)
-            os._exit(0)
+            try:
+                os.setpgrp()
+                os.close(w)
+                r = os.fdopen(r)
+                x = r.read()
+                os.killpg(os.getpgid(self.runqemu.pid), signal.SIGTERM)
+            finally:
+                # We must exit under all circumstances
+                os._exit(0)
 
         self.logger.debug("runqemu started, pid is %s" % self.runqemu.pid)
         self.logger.debug("waiting at most %d seconds for qemu pid (%s)" %