Re: [musl] Re: add loongarch64 port

[Christian Brauner <brauner@kernel.org>]

On Wed, Apr 13, 2022 at 10:26:06AM +0200, Arnd Bergmann wrote:
> On Wed, Apr 13, 2022 at 9:26 AM Christian Brauner <brauner@kernel.org> wrote:
> > On Sun, Apr 10, 2022 at 12:30:59PM +0200, Arnd Bergmann wrote:
> > > On Sat, Apr 9, 2022 at 3:31 PM Rich Felker <dalias@libc.org> wrote:
> > > >
> > > > Actually, if there aren't yet archs lacking SYS_clone, this API
> > > > regression may be a good argument not to drop SYS_clone on new archs
> > > > yet until there's a way for new archs to get the same behavior
> > > > (unspecified stack size).
> > >
> > > That is a good point, but it also appears that the behavior of
> > > clone3() is unintentional
> > > here, I'm fairly sure it was meant to be a drop-in replacement for clone() with
> > > additional features.
> >
> > Mostly but not in all ways. We did decide it's ok to make API
> > improvements that might break compatibility with legacy clone().
> 
> Ok
> 
> > >
> > > Not sure what the best fix for this is, as the check for size==0 was clearly
> > > intentional, but seems to prevent this from working. A special flag to ignore
> > > the size, or a magic size value like -1ull might work, but neither of them
> > > is a great interface.
> >
> > Can someone explain the use-case in a bit more detail, please?
> >
> > If it is a legitimate use-case that callers need to be able to pass a
> > stack and have no way of also passing a size then we should just remove
> > the size == 0 check for all architectures that don't have a hard
> > requirement on passing a size together with the stack pointer.
> >
> > Wdyt, Arnd?
> 
> The normal rule is that we don't define obsolete system calls in new
> architectures when an improved variant has been added, e.g. oldoldstat,
> oldstat, stat, newstat and stat64 have all been replaced by statx over
> the decades. I was expecting the same to be true for clone(), but if
> clone3() is not meant as a replacement, we can keep both around.

No, I agree with you on this and would like to only implement clone3()
on new architectures.

What I'm asking is whether removing the size == 0 check is enough to
unblock the missing behavior and whether you'd be on board with removing
the check?