Re: [PATCH v2 0/3] net/9p: safely abort fd RPCs on fatal signals

[Dominique Martinet]

Ze Tan wrote on Thu, Jul 16, 2026 at 03:30:48PM +0800:
> This series fixes a hung task reported by Syzkaller when a thread blocked
> in a 9p fd RPC receives a fatal signal while another thread in the same
> group is waiting in coredump_wait(). The interrupted RPC currently enters
> the TFLUSH path and can wait indefinitely if the 9p server is no longer
> responding, preventing the thread from exiting and the coredump from
> completing.

Thanks for the patches and sorry for lack of reply

I'm sorry but I'm cutting this out for the forseeable future:
You can't just abort a request you sent without flushing, for example
if a mkdir() call is interrupted it should be failed and the next
mkdir() must succeed (if it wasn't raced) but with an aborted request
you could have the directory be created in the back of the aborted
request.

I actually had a branch with 'async flush' a few years back but there
were subtle regressions and I never had to look further into it, and I
have even less time now, so I'm sorry but unless you're very convincing
I won't take these (other maintainers may have different opinions if you
can get any of them to look, that won't be my problem then)
-- 
Dominique Martinet | Asmadeus