diff for duplicates of <20200216044551-mutt-send-email-mst@kernel.org> diff --git a/a/1.txt b/N1/1.txt index f624084..5492068 100644 --- a/a/1.txt +++ b/N1/1.txt @@ -5,15 +5,15 @@ On Fri, Feb 14, 2020 at 12:48:42PM -0800, Tyler Sanderson wrote: > ported to 4.19 where the performance regression was first introduced. > My concern with reverting to the OOM notifier is, as mst@ put it (in the other > thread): -> "when linux hits OOM all kind of error paths are being hit, latent bugs start -> triggering, latency goes up drastically." +> "when linux hits OOM all kind of error paths are being hit, latent bugs start +> triggering, latency goes up drastically." > The guest could be in a lot of pain before the OOM notifier is invoked, and it > seems like the shrinker API might allow more fine grained control of when we > deflate. > > On the other hand, I'm not totally convinced that Wei's patch is an expected > use of the shrinker/page-cache APIs, and maybe it is fragile. Needs more -> testing and scrutiny. +> testing and scrutiny. > > It seems to me like the shrinker API is the right API in the long run, perhaps > with some fixes and modifications. But maybe reverting to OOM notifier is the @@ -26,12 +26,12 @@ In that case can I see some Tested-by reports pls? > > >> There was a report that this results in undesired side effects when > >> inflating the balloon to shrink the page cache. [1] -> >>   "When inflating the balloon against page cache (i.e. no free memory -> >>    remains) vmscan.c will both shrink page cache, but also invoke the -> >>    shrinkers -- including the balloon's shrinker. So the balloon -> >>    driver allocates memory which requires reclaim, vmscan gets this -> >>    memory by shrinking the balloon, and then the driver adds the -> >>    memory back to the balloon. Basically a busy no-op." +> >> "When inflating the balloon against page cache (i.e. no free memory +> >> remains) vmscan.c will both shrink page cache, but also invoke the +> >> shrinkers -- including the balloon's shrinker. So the balloon +> >> driver allocates memory which requires reclaim, vmscan gets this +> >> memory by shrinking the balloon, and then the driver adds the +> >> memory back to the balloon. Basically a busy no-op." > >> > >> The name "deflate on OOM" makes it pretty clear when deflation should > >> happen - after other approaches to reclaim memory failed, not while diff --git a/a/content_digest b/N1/content_digest index 8c8ea34..9ac76d2 100644 --- a/a/content_digest +++ b/N1/content_digest @@ -7,13 +7,15 @@ "Subject\0Re: [PATCH v1 3/3] virtio-balloon: Switch back to OOM handler for VIRTIO_BALLOON_F_DEFLATE_ON_OOM\0" "Date\0Sun, 16 Feb 2020 04:46:12 -0500\0" "To\0Tyler Sanderson <tysand@google.com>\0" - "Cc\0linux-kernel@vger.kernel.org" + "Cc\0David Hildenbrand <david@redhat.com>" Michal Hocko <mhocko@kernel.org> + linux-kernel@vger.kernel.org linux-mm@kvack.org - Nadav Amit <namit@vmware.com> - David Rientjes <rientjes@google.com> + virtualization@lists.linux-foundation.org + Wei Wang <wei.w.wang@intel.com> Alexander Duyck <alexander.h.duyck@linux.intel.com> - " virtualization@lists.linux-foundation.org\0" + David Rientjes <rientjes@google.com> + " Nadav Amit <namit@vmware.com>\0" "\00:1\0" "b\0" "On Fri, Feb 14, 2020 at 12:48:42PM -0800, Tyler Sanderson wrote:\n" @@ -23,15 +25,15 @@ "> ported to 4.19 where the performance regression was first introduced.\n" "> My concern with reverting to the OOM notifier is, as mst@ put it (in the other\n" "> thread):\n" - "> \"when linux hits OOM\303\202\302\240all kind of error paths are being hit, latent bugs start\n" - "> triggering,\303\202\302\240latency goes up drastically.\"\n" + "> \"when linux hits OOM\302\240all kind of error paths are being hit, latent bugs start\n" + "> triggering,\302\240latency goes up drastically.\"\n" "> The guest could be in a lot of pain before the OOM notifier is invoked, and it\n" "> seems like the shrinker API might allow more fine grained control of when we\n" "> deflate.\n" "> \n" "> On the other hand, I'm not totally convinced that Wei's patch is an expected\n" "> use of the shrinker/page-cache APIs, and maybe it is fragile. Needs more\n" - "> testing\303\202\302\240and scrutiny.\n" + "> testing\302\240and scrutiny.\n" "> \n" "> It seems to me like the shrinker API is the right API in the long run, perhaps\n" "> with some fixes and modifications. But maybe reverting to OOM notifier is the\n" @@ -44,12 +46,12 @@ "> \n" "> >> There was a report that this results in undesired side effects when\n" "> >> inflating the balloon to shrink the page cache. [1]\n" - "> >>\303\202\302\240 \303\202\302\240 \303\202\302\240 \"When inflating the balloon against page cache (i.e. no free memory\n" - "> >>\303\202\302\240 \303\202\302\240 \303\202\302\240 \303\202\302\240remains) vmscan.c will both shrink page cache, but also invoke the\n" - "> >>\303\202\302\240 \303\202\302\240 \303\202\302\240 \303\202\302\240shrinkers -- including the balloon's shrinker. So the balloon\n" - "> >>\303\202\302\240 \303\202\302\240 \303\202\302\240 \303\202\302\240driver allocates memory which requires reclaim, vmscan gets this\n" - "> >>\303\202\302\240 \303\202\302\240 \303\202\302\240 \303\202\302\240memory by shrinking the balloon, and then the driver adds the\n" - "> >>\303\202\302\240 \303\202\302\240 \303\202\302\240 \303\202\302\240memory back to the balloon. Basically a busy no-op.\"\n" + "> >>\302\240 \302\240 \302\240 \"When inflating the balloon against page cache (i.e. no free memory\n" + "> >>\302\240 \302\240 \302\240 \302\240remains) vmscan.c will both shrink page cache, but also invoke the\n" + "> >>\302\240 \302\240 \302\240 \302\240shrinkers -- including the balloon's shrinker. So the balloon\n" + "> >>\302\240 \302\240 \302\240 \302\240driver allocates memory which requires reclaim, vmscan gets this\n" + "> >>\302\240 \302\240 \302\240 \302\240memory by shrinking the balloon, and then the driver adds the\n" + "> >>\302\240 \302\240 \302\240 \302\240memory back to the balloon. Basically a busy no-op.\"\n" "> >>\n" "> >> The name \"deflate on OOM\" makes it pretty clear when deflation should\n" "> >> happen - after other approaches to reclaim memory failed, not while\n" @@ -81,4 +83,4 @@ "> \n" > -cb6f8bc59ccd86a3383134dd0aa1ff94942ab0853a5eb8ce6b422c3c828e4851 +9999a1248c7b60e848b8918d6a0f2e75073d609f96ca5fa1302b770e4c9c764f
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.