Hi Andrew, >> It was discussed before. You're not the first one suggesting something like this. >> There is the comment near in_atomic() explaining well why and when your patch won't work. > Thanks for the info. > >> The easiest way of making vfree() to be safe in atomic contexts is this patch: >> http://lkml.kernel.org/r/20170330102719.13119-1-aryabinin@xxxxxxxxxxxxx >> >> But the final decision at that time was to fix caller so the call vfree from sleepable context instead: >> http://lkml.kernel.org/r/20170330152229.f2108e718114ed77acae7405@xxxxxxxxxxxxxxxxxxxx > OK, if that is the final decision, then I think Jiufei's patch that > moves kvfree() out of the locked region is the right thing to do for > this issue here. > Is that the final decision we have made? Could you please look into my patch again and give the decision? Thanks, Jiufei
