On 08/09/2018 11:44 AM, Paolo Bonzini wrote:
On 09/08/2018 11:20, Emanuele wrote:
Why this? Shouldn't it be:

if (g_strcmp0(old_path, path)) {
         qtest_end(); /* handles global_qtest = NULL */
         g_free(old_path); /* handles NULL */
         old_path = path;
         global_qtest = qtest_start(path);
} else ....
Yes, of course.  Though I'd have thought that my version has "just" a
memory leak.  Even better, this could call qos_invalidate_command_line.

Can you post the fixes to a separate commit on github?
Paolo, did you have a chance to test the fixes? I tried, it seems to be working fine. About valgrind, it only detects an "invalide read of size 8" in blk-hotplug test (tests/virtio-blk-test.c), but I don't know why is it doing this.
For the rest, there shouldn't be any memory leak.

Should I rebase them in the proper commits? I am ready to submit patch v3.


Reply via email to