Re: [PATCH] x86/asm/64: do not clear high 32 bits of syscall number when CONFIG_X86_X32=y

["Dmitry V. Levin" <ldv@altlinux.org>]

[-- Attachment #1: Type: text/plain, Size: 2021 bytes --]

On Thu, Sep 14, 2017 at 02:05:04PM -0700, Andy Lutomirski wrote:
> On Tue, Sep 12, 2017 at 3:57 PM, Dmitry V. Levin wrote:
> > Before this change, CONFIG_X86_X32=y fastpath behaviour was different
> > from slowpath:
> >
> > $ gcc -xc -Wall -O2 - <<'EOF'
> > #include <unistd.h>
> > #include <asm/unistd.h>
> > int main(void) {
> >         unsigned long nr = ~0xffffffffUL | __NR_exit;
> >         return !!syscall(nr, 42, 1, 2, 3, 4, 5);
> > }
> > EOF
> > $ ./a.out; echo \$?=$?
> > $?=42
> > $ strace -enone ./a.out
> > syscall_18446744069414584380(0x2a, 0x1, 0x2, 0x3, 0x4, 0x5) = -1 (errno 38)
> > +++ exited with 1 +++
> >
> > This change syncs CONFIG_X86_X32=y fastpath behaviour with the case
> > when CONFIG_X86_X32 is not enabled.
> 
> Do you see real brokenness here, or is it just weird?

It's definitely broken.  A syscall should be either valid or invalid
regardless of implementation peculiarities like fastpath vs slowpath.

> > Fixes: fca460f95e92 ("x32: Handle the x32 system call flag")
> > Cc: stable@vger.kernel.org
> > Signed-off-by: Dmitry V. Levin <ldv@altlinux.org>
> > ---
> >  arch/x86/entry/entry_64.S | 8 +++-----
> >  1 file changed, 3 insertions(+), 5 deletions(-)
> >
> > diff --git a/arch/x86/entry/entry_64.S b/arch/x86/entry/entry_64.S
> > index 4916725..3bab6af 100644
> > --- a/arch/x86/entry/entry_64.S
> > +++ b/arch/x86/entry/entry_64.S
> > @@ -185,12 +185,10 @@ entry_SYSCALL_64_fastpath:
> >          */
> >         TRACE_IRQS_ON
> >         ENABLE_INTERRUPTS(CLBR_NONE)
> > -#if __SYSCALL_MASK == ~0
> > -       cmpq    $__NR_syscall_max, %rax
> > -#else
> > -       andl    $__SYSCALL_MASK, %eax
> > -       cmpl    $__NR_syscall_max, %eax
> > +#if __SYSCALL_MASK != ~0
> > +       andq    $__SYSCALL_MASK, %rax
> >  #endif
> > +       cmpq    $__NR_syscall_max, %rax
> 
> I don't know much about x32 userspace, but there's an argument that
> the high bits *should* be masked off if the x32 bit is set.

Why?


-- 
ldv

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 801 bytes --]