wineserver sometimes crashes in fd.c:1548: set_fd_events: Assertion `poll_users[user] == fd' failed

Mihai Todor todormihai at gmail.com
Wed Feb 21 05:39:33 CST 2018


Small update on this issue: Under heavier load, besides the original
issue, we're seeing another assert happen sometimes: "wineserver:
sock.c:507: sock_get_poll_events: Assertion `sock->obj.ops ==
&sock_ops' failed."

At a quick glance, it's looking like somehow fd->user is getting freed
before the fd itself somewhere, but I'll have to spend more time
reading the code.

It's still negligible for us, but I'll keep an eye on this and update
as soon as I'm able to do more thorough digging.

Regards,
Mihai

On Sat, Feb 10, 2018 at 10:29 PM, Mihai Todor <todormihai at gmail.com> wrote:
> Hello Everyone,
>
> I am using Wine 3.0 stable and, as part of my setup, I am running
> wineserver in persistent mode forever (wineserver -p) for performance
> reasons and, sometimes, it dies with this error: "wineserver:
> fd.c:1548: set_fd_events: Assertion `poll_users[user] == fd' failed."
>
> For the time being, this is a sufficiently rare occurrence to not
> concern me (it happens once / day or so), but I am curious to learn
> what might cause it, since it looks like a bug.
>
> For a bit more detail about my setup, while wineserver is running, I
> am starting a single instance of an x86 application with an embedded
> gRPC TCP server, I am instructing this application to process some
> local files and, after it finishes, I am stopping it. This cycle
> repeats once every 10 - 20 seconds on average and there is just one
> instance of this application running at the same time. Sometimes, this
> application hangs and I terminate it forcefully by killing the process
> (bad input file / bad code / etc), which is what I am thinking might
> be the cause of the wineserver crash. However, so far, I was unable to
> reproduce this issue consistently.
>
> I am considering to spend some time adding more logging in the
> wineserver code to narrow it down, but I'd be grateful if anyone has
> any insight into what might be the cause and how to diagnose it.
>
> Thanks,
> Mihai



More information about the wine-devel mailing list