Messages in this thread | | | Date | Wed, 22 Apr 2015 20:04:34 +0200 | From | Denys Vlasenko <> | Subject | Re: [PATCH 2/2] x86/asm/entry/32: Remove unnecessary optimization in stub32_clone |
| |
On 04/22/2015 07:10 PM, Josh Triplett wrote: > On Wed, Apr 22, 2015 at 09:54:24AM -0700, Andy Lutomirski wrote: >> On Wed, Apr 22, 2015 at 9:40 AM, Denys Vlasenko <dvlasenk@redhat.com> wrote: >>> Really swap arguments #4 and #5 in stub32_clone instead of "optimizing" >>> it into a move. >>> >>> Yes, tls_val is currently unused. Yes, on some CPUs XCHG is a little bit >>> more expensive than MOV. But a cycle or two on an expensive syscall like >>> clone() is way below noise floor, and obfuscation of logic introduced >>> by this optimization is simply not worth it. >> >> Ditto re: Josh's patch. > > I do think my two-patch HAVE_COPY_THREAD_TLS series should go in fixing > this, but I'd like to see the final version of Denys' comment added on > top of it (with an update to the type and name of the tls argument to > match the changes to sys_clone). > > Denys, would you consider submitting a patch adding your comment on top > of the two-patch series I just sent?
Okay.
| |