On 2015-03-23 18:18, Tanu Kaskinen wrote: > On Mon, 2015-03-23 at 14:39 +0100, David Henningsson wrote: >> While investigating bug 89672 it was found that pa_thread_mq_done >> was called recursively. Regardless of whether the recursion should >> be stopped by other means, it seems to make sense to make >> pa_thread_mq_done more robust so that it can be called twice >> (and even recursively) without harm. >> >> BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=89672 >> Signed-off-by: David Henningsson <david.henningsson at canonical.com> >> --- >> src/pulsecore/thread-mq.c | 32 +++++++++++++++++++++++--------- >> 1 file changed, 23 insertions(+), 9 deletions(-) > > I would prefer to replace pa_thread_mq_init() and pa_thread_mq_done() > with pa_thread_mq_new() and pa_thread_mq_free(), the difference of > course being that _new() would return a newly allocated struct instead > of modifying a caller-supplied struct. That would allow the bluetooth > code to set the thread_mq to NULL when it's not used. In general, I don't mind that approach, but I wonder how well it would work in this recursion case - it seems like such an attempt would double free the thread_mq? -- David Henningsson, Canonical Ltd. https://launchpad.net/~diwic
