Re: [PATCH 4.9 72/76] arm64: futex: Fix FUTEX_WAKE_OP atomic ops with non-zero result value

From: Nathan Chancellor
Date: Tue Apr 16 2019 - 12:49:16 EST


On Tue, Apr 16, 2019 at 10:13:40AM +0100, Will Deacon wrote:
> On Mon, Apr 15, 2019 at 03:01:51PM -0700, Nathan Chancellor wrote:
> > On Mon, Apr 15, 2019 at 08:44:36PM +0200, Greg Kroah-Hartman wrote:
> > > From: Will Deacon <will.deacon@xxxxxxx>
>
> [...]
>
> > > @@ -53,29 +53,29 @@
> > > static inline int
> > > arch_futex_atomic_op_inuser(int op, int oparg, int *oval, u32 __user *uaddr)
> > > {
> > > - int oldval = 0, ret, tmp;
> > > + int oldval, ret, tmp;
> > >
> > > pagefault_disable();
> > >
> > > switch (op) {
> > > case FUTEX_OP_SET:
> > > - __futex_atomic_op("mov %w0, %w4",
> > > + __futex_atomic_op("mov %w3, %w4",
> > > ret, oldval, uaddr, tmp, oparg);
> > > break;
> > > case FUTEX_OP_ADD:
> > > - __futex_atomic_op("add %w0, %w1, %w4",
> > > + __futex_atomic_op("add %w3, %w1, %w4",
> > > ret, oldval, uaddr, tmp, oparg);
> > > break;
> > > case FUTEX_OP_OR:
> > > - __futex_atomic_op("orr %w0, %w1, %w4",
> > > + __futex_atomic_op("orr %w3, %w1, %w4",
> > > ret, oldval, uaddr, tmp, oparg);
> > > break;
> > > case FUTEX_OP_ANDN:
> > > - __futex_atomic_op("and %w0, %w1, %w4",
> > > + __futex_atomic_op("and %w3, %w1, %w4",
> > > ret, oldval, uaddr, tmp, ~oparg);
> > > break;
> > > case FUTEX_OP_XOR:
> > > - __futex_atomic_op("eor %w0, %w1, %w4",
> > > + __futex_atomic_op("eor %w3, %w1, %w4",
> > > ret, oldval, uaddr, tmp, oparg);
> > > break;
> > > default:
> > >
> > >
> >
> > This causes a (false) build warning with AOSP's GCC 4.9.4 (which is
> > used to build nearly all arm64 Android kernels before 4.14):
> >
> > CC kernel/futex.o
> > ../kernel/futex.c: In function 'do_futex':
> > ../kernel/futex.c:1492:17: warning: 'oldval' may be used uninitialized in this function [-Wmaybe-uninitialized]
> > return oldval == cmparg;
> > ^
> > In file included from ../kernel/futex.c:69:0:
> > ../arch/arm64/include/asm/futex.h:56:6: note: 'oldval' was declared here
> > int oldval, ret, tmp;
> > ^
> >
> > The only reason I bring this up is Qualcomm based kernels have a Python
> > script that emulates -Werror, meaning this will be fatal for a large
> > number of kernels, when this eventually gets merged into them.
>
> Thanks. Does restoring the initial assignment of 0 suppress the bogus
> warning? If so, please could you send a patch on top for stable (assuming
> Greg is ok with the simple change for this)?
>
> Will

Yes, it does and I sure can. Greg, let me know if that is okay.