Dne 17. 07. 22 v 9:05 Takashi Iwai napsal(a): > Hi, > > this is a revised patch set for dropping fasync support from ALSA > core. > > The async signal itself is very difficult to use properly due to > various restrictions (e.g. you cannot perform any I/O in the context), > hence it's a feature that has been never used by real applications. > > OTOH, the real problem is that there have been quite a few syzcaller > reports indicating that fasync code path may lead to some potential > deadlocks for long time. Dropping the feature is the easiest > solution, obviously. I would probably prefer to fix the problem (deferred async kill or so). The SIGIO is just another way to wakeup the applications and there may be some corner cases, where this wakeup is usable (threaded apps). Note that we had some users (or testers) of SIGIO in past (at least there were questions about this support). Jaroslav -- Jaroslav Kysela <perex@xxxxxxxx> Linux Sound Maintainer; ALSA Project; Red Hat, Inc.