On Thu, Nov 7, 2019 at 10:46 AM Andrey Ryabinin <aryabinin@xxxxxxxxxxxxx> wrote:
> On 11/7/19 12:22 PM, Alexander Potapenko wrote:
> > On Thu, Nov 7, 2019 at 10:04 AM Arnd Bergmann <arnd@xxxxxxxx> wrote:
> >> On Thu, Nov 7, 2019 at 7:08 AM Sergey Senozhatsky
> >> <sergey.senozhatsky.work@xxxxxxxxx> wrote:
> >> The normal way to do a volatile access would be
> >> READ_ONCE()/WRITE_ONCE(), but that seems stronger than
> >> the barrier() here. I'd just stick to adding a barrier.
> > I actually like the READ_ONCE idea more, as READ_ONCE is really a
> > documented way to prevent the compiler from merging reads, which is
> > what we want here.
>
> I would rather go with -fno-builtin-bcmp or maybe even -fno-builtin if that works.
The commit message for 5f074f3e192f ("lib/string.c: implement a basic bcmp")
mentions that -fno-builtin-bcmp did not work for LTO when the global bcmp()
help was added. I don't know whether the same applies here, but my guess is
that it's the same issue.
Arnd
