On Wed, Oct 09, 2019 at 09:31:41AM -0700, Nick Desaulniers wrote: > On Wed, Oct 9, 2019 at 7:30 AM Dan Carpenter <dan.carpenter@xxxxxxxxxx> wrote: > > > > On Wed, Oct 09, 2019 at 04:21:20PM +0200, Rasmus Villemoes wrote: > > > On 09/10/2019 15.56, Dan Carpenter wrote: > > > > That's because glibc strlen is annotated with __attribute_pure__ which > > > > means it has no side effects. > > > > > > I know, except it has nothing to do with glibc headers. Just try the > > > same thing in the kernel. gcc itself knows this about __builtin_strlen() > > > etc. If anything, we could annotate some of our non-standard functions > > > (say, memchr_inv) with __pure - then we'd both get the Wunused-value in > > > the nonsense cases, and allow gcc to optimize or reorder the calls. > > > > Huh. You're right. GCC already knows. So this patch is pointless like > > you say. > > Is it? None of the functions in include/linux/string.h are currently > marked __pure today. I've already embarrassed myself with my ignorance once so I may as well keep talking now... GCC did complain about the unused result even though we don't declare them as __pure. So GCC rule must have this rule built in. We were discussing in a different thread that standard says that memcpy() pointers can't be NULL (even when we're copying zero bytes) so GCC will assume that's true. If you have: memcpy(foo, bar, 0); if (foo) *foo = 0; GCC will sometimes remove the condition. This doesn't affect the kernel because we use -fno-delete-null-pointer-checks. https://groups.google.com/forum/#!msg/syzkaller-netbsd-bugs/8B4CIKN0Xz8/wRvIUWxiAgAJ Weird, huh? regards, dan carpenter
