On Tue, 2018-01-02 at 13:22 -0800, Andrew Morton wrote: > On Fri, 29 Dec 2017 09:55:07 +0900 Minchan Kim <minchan@xxxxxxxxxx> > wrote: > > > > > James reported a bug of swap paging-in for his testing and found it > > at rc5, soon to be -rc5. > > > > Although we can fix the specific problem at the moment, it may > > have other lurkig bugs so want to have one more cycle in -next > > before merging. > > > > This patchset reverts 23c47d2ada9f, 08fa93021d80, 8e31f339295f > > completely > > but 79b5f08fa34e partially because the swp_swap_info function that > > 79b5f08fa34e introduced is used by [1]. > > Gets a significant reject in do_swap_page(). Could you please take a > look, redo against current mainline? > > Or not. We had a bug and James fixed it. That's what -rc is > for. Why not fix the thing and proceed? My main worry was lack of testing at -rc5, since the bug could essentially be excited by pushing pages out to swap and then trying to access them again ... plus since one serious bug was discovered it wouldn't be unusual for there to be others. However, because of the IPT stuff, I think Linus is going to take 4.15 over a couple of extra -rc releases, so this is less of a problem. > There's still James's "unaccountable shutdown delay". Is that still > present? Is it possible to see whether the full revert patch fixes > it? On -rc6 it's no longer manifesting with just the bug fix applied, so it might have been a -rc5 artifact. James -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
