Re: [Qemu-devel] [PATCH 1.1] coroutine: Avoid ucontext usage on i386 Linux host

[Anthony Liguori]

On 05/09/2012 03:11 PM, Jan Kiszka wrote:
> On 2012-05-09 17:01, Anthony Liguori wrote:
> > On 05/09/2012 02:57 PM, Jan Kiszka wrote:
> > > On 2012-05-09 16:48, Anthony Liguori wrote:
> > > > On 05/09/2012 02:34 PM, Jan Kiszka wrote:
> > > > > > Can't we resort to the SIGUSR1 workaround for the time being, while
> > > > > > no RT signals are in actual use, and just have the time to let the
> > > > > > kernel side to fix the things up before some actual RTsig user will
> > > > > > emerge in qemu?  I think it is a bit more conservative approach,
> > > > > > especially having in mind the minority of users this issue affects
> > > > > > (only 32/64 mixed environment).  I'd favor for this variant, and
> > > > > > it looks like I'm the "main" 32/64bit user of qemu in this world :)
> > > > >
> > > > > Most conservative is definitely this patch, not switching to SIGUSR1,
> > > > > hoping that no other RT signal user shows up until current kernel are no
> > > > > longer in use.
> > > >
> > > > Sorry, how is using a totally different code path more conservative than using a
> > > > different signal number?
> > >
> > > If the gthread version is not safe to use, why do we fall back to it?
> >
> > It's safe, but it's significantly slower.
>
> OK. Then what about sigaltstack (once fixed)? Is it also slower?

I don't know, performance testing would need to be done.

> If not,
> can we converge over it? I would really hate staying with this time bomb
> of broken RT signals unless someone tells me we will kick out all these
> coroutines rather sooner than later.

AFAICT, neither SIGUSR1 or SIGUSR2 are used today.  We only use SIG_IPI today. 
We could easily #define SIG_IPI to SIGUSR1 unconditionally today.  We used to 
use SIGUSR2 for posix-aio but that was ages ago.  AFAICT, it's only used for 
sigaltstack now.

I don't see where this "time bomb" comes from.  I think it's perfect reasonable 
that if we end up needing more signals (after exhausting SIGUSR1/SIGUSR2) we 
simply require a fixed kernel for 32bit on 64bit.

Regards,

Anthony Liguori

> Jan