Ben, I decided to drop this patch as I wrote a bug fix relating to this. So this patch is superseded by the later patch.
I did not investigate the potential code generation differences, but decided that we need not worry about other threads, when the documentation says that the function may only be called when no other thread has access to the rculist element. Jarno > On Jun 9, 2015, at 8:51 PM, Ben Pfaff <[email protected]> wrote: > > On Tue, Jun 09, 2015 at 05:05:57PM -0700, Jarno Rajahalme wrote: >> >>> On Jun 9, 2015, at 3:30 PM, Ben Pfaff <[email protected]> wrote: >>> >>> On Tue, Jun 09, 2015 at 03:25:08PM -0700, Jarno Rajahalme wrote: >>>> Should not use ovsrcu_set_hidden() when the pointer may have been >>>> visible to other threads already. >>>> >>>> Signed-off-by: Jarno Rajahalme <[email protected]> >>> >>> I think that rculist_poison__() is only appropriate when an rculist >>> should not be visible to any other threads. No? >> >> That was my initial thought as well, but we poison for the case other >> threads are still using it (while they shouldn't), so I thought >> ovsrcu_set_hidden() is not appropriate here after all. I doubt this >> has any practical difference either way, more about what is >> appropriate use. > > Is there a code generation difference on x86-64? If not, then I'm happy > with this change: > Acked-by: Ben Pfaff <[email protected] <mailto:[email protected]>> _______________________________________________ dev mailing list [email protected] http://openvswitch.org/mailman/listinfo/dev
