Messages in this thread | | | Date | Tue, 21 Apr 2015 00:08:46 +0200 | Subject | Re: [PATCH RFC] x86: enforce inlining for atomics | From | Hagen Paul Pfeifer <> |
| |
On 20 April 2015 at 23:56, Borislav Petkov <bp@alien8.de> wrote:
> Hmm, that must be config-specific as doing > > objdump -D vmlinux | grep -i "atomic_add" > > here gives me only "drm_atomic_add_affected_connectors" matches. > > It probably gets inlined here always...
Probably, the config is allyesconfig minus trace/kernel adress sanitizer and gcov related options.
> Other than that, this patch should actually even show some speedup as > we're getting rid of the stack preparation and function call overhead. > Have you done any benchmarks with it?
No, but I am sure that a benchmark should show at least theoretically better perf numbers. But the numbers should be the measurement uncertainty range. The avoided cache line miss should be the greatest perf effect.
Hagen
| |