On 3/8/19 5:08 PM, Linus Torvalds wrote: > Oh well. I'm obviously not entirely happy about it, but the refcount > issues look fixable. > > So I pushed it out. Al, mind taking a look? Jens claimed in his pull > request that you've already looked at the refcounting, so I think the > only issue is whether it shares the bug with fs/aio.c that I think it > does, or whether it might be avoiding it because of doing things > slightly differently. Sorry for the late replies, was away. The refcount req business all rests on the poll command implementation, which is basically a straight port of the aio version. So I guess not surprising if it suffers from the same reference issue. I'll take a look at the patches that went into aio and apply the same fix(es) to io_uring. I'll let Al expand on the AF_UNIX socket issue. Thanks for getting this merged! I'll be on top of squashing the above two cases asap. -- Jens Axboe