Messages in this thread | | | Date | Thu, 03 Dec 2009 11:13:21 +0000 | From | "Jan Beulich" <> | Subject | Re: [PATCH] x86: slightly shorten __ticket_spin_trylock() (v2) |
| |
>>> Ingo Molnar <mingo@elte.hu> 03.12.09 11:49 >>> >Looks much cleaner - and should give us pretty much the same savings as >the previous patch, right? (if not, do you have a size comparison?)
With upstream code there doesn't appear to be a difference in the generated code. I noted, however, that with the spinlock contention patch I have pending as a potential replacement for pv-ops spin locks (which adds a conditional in the trylock functions) the compiler I use no longer eliminates one of the two tests of the outcome of the locking attempt.
Jan
| |