On Sun, Sep 20, 2020 at 08:03:13PM +0100, Chavdar Ivanov wrote: > On Sun, 20 Sep 2020 at 19:58, Joerg Sonnenberger <[email protected]> wrote: > > > > On Sun, Sep 20, 2020 at 02:06:42PM +0100, Chavdar Ivanov wrote: > > > I am not seeing other people report these lately, but I still continue > > > to get these hangings, e.g. when running git as part of a zsh prompt > > > function; roughly every fourth command ends up with one of these, > > > again as before, it is enough to attach with gdb to the got process > > > and quit in order for it to complete. Here is another trace, -current > > > is 9.99.73 from yesterday: > > > > Sure, nothing changed. I don't think this is really the fault of > > jemalloc, from earlier debugging the real problem is that the wait list > > got messed up. > > The weird thing is, attaching with gdb and then quitting sorts it out...
Not surprising at all. It will result in the lwp_park system call getting interrupted and the recheck will notice that there are no waiters... Joerg
