[Bug 49890] Disconnects and networking issues in applications using websockets

[WineHQ Bugzilla]

https://bugs.winehq.org/show_bug.cgi?id=49890

--- Comment #2 from David Torok <dt at zeroitlab.com> ---
(In reply to Austin English from comment #1)
> Wine version? Terminal output? Steps to reproduce?

Great questions, I'm sorry for not giving more details. This was meant as a
placeholder, until I figured out those details.

On wine versions < 5.7, I was able to reproduce this very consistently. Now I
feel silly, because I cannot reproduce this at all on 5.17 in the case of
Legends of Runeterra.

When I was testing this on wine < 5.7, the terminal output did not contain
anything related to this; however I may have missed the right debug channel.
However, when I did try to track that down (by following an unobfuscated
version of GameAssembly.dll), I came to the conclusion that the websocket
disconnect was initiated by the client because of a "timeout". I am certain
that the timeout was unreasonable, because I wrote a reverse proxy that printed
the websocket packets in real time, and the supposed timeout was shortly after
receiving packets.

As for reproduction steps, on wine versions < 5.7 staging (or any version with
my patches from #47970 applied to staging and before the ntdll PE split),
getting into a Player VS Ai game in Legends of Runeterra (
https://playruneterra.com/en-us/ , free to play however a Riot account is
needed ) surely ended with a disconnect at about the 3 minute mark from
connecting to the game. (Reproduced the issue for me basically 100% of the
time)

More testing will be needed to see if this was fixed at some point, as I cannot
reproduce this on 5.17 at this time. (To test this with Runeterra on 5.17, the
staging patch "ntdll-NtQueryVirtualMemory" needs to be rebased, which I will
try to submit separately)

-- 
Do not reply to this email, post in Bugzilla using the
above URL to reply.
You are receiving this mail because:
You are watching all bug changes.