Re: [Qemu-devel] patch to swap SIGRTMIN + 1 and SIGRTMAX - 1

[Marlies Ruck]

Hi All,

You are correct, this would break Go binaries.  I was planning on
submitting a PR to the Go runtime if my patch was accepted to QEMU.  The Go
PR would ignore the failure to register a handler for SIGRTMAX - 1, just as
it currently ignores the failure on SIGRTMAX (
https://github.com/golang/go/blob/master/src/runtime/os_linux.go#L443).
Does this seem reasonable to you?  My local set up has these patches to
QEMU and Go and it allows me to run both Go guest programs, and guest
programs that use SIGRTMIN + 1.

Also, thank you Peter and Phil for your patience as this is my first time
contributing to QEMU.  If you decide to use this patch, I'm happy to resend
the patch via `git send email`.

Thanks!
Marli



On Mon, Jul 1, 2019 at 2:08 AM Peter Maydell <address@hidden>
wrote:

> On Sat, 29 Jun 2019 at 11:53, Philippe Mathieu-Daudé <address@hidden>
> wrote:
> >
> > Hi Marlies,
> >
> > On 6/29/19 1:26 AM, Marlies Ruck wrote:
> > > Hi,
> > >
> > > I just wanted to follow up since I sent this patch a week ago to make
> sure
> > > it was a received.  An ack would be appreciated.
> >
> > You did not Cc'ed the maintainer, so he likely missed it.
> > See:
> > https://wiki.qemu.org/Contribute/SubmitAPatch#CC_the_relevant_maintainer
> >
> > $ ./scripts/get_maintainer.pl -f linux-user/signal.c
> > Riku Voipio <address@hidden> (maintainer:Linux user)
> > Laurent Vivier <address@hidden> (reviewer:Linux user)
> >
> > BTW you should also look at the next paragraph in this wiki page:
> >
> https://wiki.qemu.org/Contribute/SubmitAPatch#Do_not_send_as_an_attachment
> >
> > Since you credit Peter Maydell, it would be kind to Cc him too.
>
> I have a strong suspicion that if we applied this patch it would
> break running a pile of Go binaries that currently work, which means
> it's something we need to be cautious about. (The Go runtime special
> cases the fact that when it tries to register some signals they don't
> work, but will abort on startup I think if it can't register a handler
> for SIGRTMAX-1, which will be the case after this patch.)
>
> thanks
> -- PMM
>