On 04/27/2016 09:22 AM, Jan Beulich wrote: >>>> On 26.04.16 at 19:23, <rcojoc...@bitdefender.com> wrote: >> On 04/26/16 19:03, George Dunlap wrote: >>> On 19/04/16 17:35, Jan Beulich wrote: >>>>>>> Razvan Cojocaru <rcojoc...@bitdefender.com> 04/19/16 1:01 PM >>> >>>>> I think this might be because the LOCK prefix should guarantee that the >>>>> instruction that follows it has exclusive use of shared memory (for both >>>>> reads and writes) but I might be misreading the docs: >>>> >>>> LOCK definitely has no effect on other than the instruction it gets applied >>>> to. >>> >>> Sorry I wasn't involved in this discussion -- what was the conclusion here? >>> >>> FWIW Andy's suggestion of using a stub seemed like the most robust >>> solution, if that could be made to work. >>> >>> If you're going to submit a patch substantially similar to this one, let >>> me know so I can review the mm bits of the original patch. >> >> I'm not really sure. >> >> Regarding this version of the patch, Jan has asked for more information >> on the performance impact, but I'm not sure how to obtain it in a >> rigorous manner. > > That was only one half, which Andrew has now answered. The > other was the not understood issue with a later variant you had > tried.
Right, there's the fact that this version (with read / write locking the whole function) works whereas just synchonizing hvmemul_cmpxchg() with a spin lock does not. It is not yet fully clear why (the part of the conversation at the very top of this email was an early attempt to elucidate it). Thanks, Razvan _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel
