Alexander Graf <ag...@suse.de> wrote on 2014/07/14 18:54:27:
>
>
> On 14.07.14 18:51, Joakim Tjernlund wrote:
> > Alexander Graf <ag...@suse.de> wrote on 2014/07/14 18:34:34:
> >>
> >> On 14.07.14 18:32, Joakim Tjernlund wrote:
> >>> Alexander Graf <ag...@suse.de> wrote on 2014/07/14 18:00:35:
> >>> You think everyone feel OK with new defaults like OP ?
> >>>> Yes.
> >>> hmm, with current qemu it works to boot a LXC with just O flag.
> >>> Why would we then want to complicate things by adding OP which
> >>> then requires some version of my patch?
> >> How does current QEMU boot anything with the 0 flag? It doesn't know
how
> >> to handle it and will misinterpret the argument, no?
> > Playing some, one could possibly do both:
> >
> > - if (i > 0 && strcmp(binfmt, "-binfmt-wrapper") == 0) {
> > + execfd = qemu_getauxval(AT_EXECFD);
> > + if (execfd > 0 || i > 0 && strcmp(binfmt, "-binfmt-wrapper") ==
0) {
>
> 0 is a valid fd :). And yes, this would work, but I don't see why we
Not according to qemu:
execfd = qemu_getauxval(AT_EXECFD);
if (execfd == 0) {
execfd = open(filename, O_RDONLY);
if (execfd < 0) {
printf("Error while loading %s: %s\n", filename,
strerror(errno));
_exit(1);
}
}
> should introduce the -binfmt-wrapper logic to upstream QEMU. It's never
> been there. And the AT_EXECFD evaluation is a lot cleaner.
It is, it is just an option to those who want to use P only(for security
reasons).
Then there is a built-in fallback which is much easier to use that the
current external program.
I can do both, first the execfd and the other ontop, then you choose :)
>
> While we're at it - should we also pass the C flag to binfmt_misc?
That should be a user/distribution choise, not sure what should be default
in
QEMU.
