Re: [PATCH 0/6] MR8878: ntdll: Avoid deadlock on SIGPIPE while holding virtual_mutex or fd_cache_mutex.

[Jinoh Kang (＠iamahuman)]

On Wed Sep  3 14:32:13 2025 +0000, Rémi Bernon wrote:
> > I suggested using `raise(SIGQUIT)` instead of e.g., refactoring all
>
> uninterrupted sections.
> Well, idk IMO it's easier to follow with an explicit return code
> handling rather than asynchronous signal processing, but I don't care much.
> I don't know what the argument was against !4811 but I think this would
> now be nice to have fixed or at least improved enough for the
> kernel32:sync test failure to avoid blocking !8875.

I don't have strong opinions either way, just making an observation that there is a way to avoid the refactoring which I think have better chance upstream (and also, SIGPIPE/EPIPE is asynchronous in nature already so I think that doesn't make big difference)