[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH] coroutine: drop GThread coroutine backend
From: |
Stefan Hajnoczi |
Subject: |
Re: [Qemu-devel] [PATCH] coroutine: drop GThread coroutine backend |
Date: |
Tue, 2 Feb 2016 17:29:03 +0000 |
User-agent: |
Mutt/1.5.24 (2015-08-30) |
On Mon, Feb 01, 2016 at 12:17:14PM +0000, Alex Bennée wrote:
>
> Stefan Hajnoczi <address@hidden> writes:
>
> > On Fri, Jan 29, 2016 at 04:41:41PM +0000, Alex Bennée wrote:
> >>
> >> Stefan Hajnoczi <address@hidden> writes:
> >>
> >> > The GThread coroutine backend was a portable coroutine implementation.
> >> > Over the years all platforms got their own optimized coroutine
> >> > implementations and nothing uses the GThread backend anymore.
> >> >
> >> > In fact, ./configure mentions the GThread backend doesn't work but might
> >> > be useful for debugging. Since GDB macros were added to ease debugging
> >> > of ucontext coroutines, there seems little point in keeping a broken
> >> > backend around.
> >>
> >> Except I found that I couldn't run the ThreadSanitizer without using the
> >> gthread co-routines. So while I totally agree we should dump stuff
> >> that's not used lets make sure no one else relies on it for debugging
> >> stuff as well.
> >
> > Is it still the case that ThreadSanitizer only works with gthread
> > coroutines?
>
> It certainly was very confused about what was going on with the default
> option (sigucontext IIRC?).
Okay, it looks like there is a good reason to keep the GThread backend.
If I have time I'd like to look into the reason why the GThread backend
doesn't pass "make check". I noticed that ide tests were hanging.
Stefan
signature.asc
Description: PGP signature