On 09/14/2018 02:08 PM, Yu-cheng Yu wrote: > On Fri, 2018-09-14 at 13:46 -0700, Dave Hansen wrote: >> On 09/14/2018 01:39 PM, Yu-cheng Yu wrote: >>> >>> With the updated ptep_set_wrprotect() below, I did MADV_WILLNEED to a shadow >>> stack of 8 MB, then 10,000 fork()'s, but could not prove it is more or less >>> efficient than the other. So can we say this is probably fine in terms of >>> efficiency? BTW, I wasn't particularly concerned about shadow stacks. Plain old memory is affected by this change too. Right? >> Well, the first fork() will do all the hard work. I don't think >> subsequent fork()s will be affected. > > Are you talking about a recent commit: > > 1b2de5d0 mm/cow: don't bother write protecting already write-protected pages > > With that, subsequent fork()s will not do all the hard work. > However, I have not done that for shadow stack PTEs (do we want to do that?). > I think the additional benefit for shadow stack is small? You're right. mprotect() doesn't use this path. But, that reminds me, can you take a quick look at change_pte_range() and double-check that it's not affected by this issue?
