From: Joe Perches <joe@perches.com>
To: Johannes Weiner <hannes@cmpxchg.org>, Michal Hocko <mhocko@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>,
Vlastimil Babka <vbabka@suse.cz>,
linux-mm@kvack.org, LKML <linux-kernel@vger.kernel.org>,
Michal Hocko <mhocko@suse.com>
Subject: Re: [PATCH] mm, sparse: do not swamp log with huge vmemmap allocation failures
Date: Mon, 06 Nov 2017 09:57:49 -0800 [thread overview]
Message-ID: <1509991069.2431.45.camel@perches.com> (raw)
In-Reply-To: <20171106173511.GA32336@cmpxchg.org>
[-- Attachment #1: Type: text/plain, Size: 3867 bytes --]
On Mon, 2017-11-06 at 12:35 -0500, Johannes Weiner wrote:
> On Mon, Nov 06, 2017 at 10:22:28AM +0100, Michal Hocko wrote:
> > From: Michal Hocko <mhocko@suse.com>
> >
> > While doing a memory hotplug tests under a heavy memory pressure we have
> > noticed too many page allocation failures when allocating vmemmap memmap
> > backed by huge page
> > [146792.281354] kworker/u3072:1: page allocation failure: order:9, mode:0x24084c0(GFP_KERNEL|__GFP_REPEAT|__GFP_ZERO)
> > [...]
> > [146792.281394] Call Trace:
> > [146792.281430] [<ffffffff81019a99>] dump_trace+0x59/0x310
> > [146792.281436] [<ffffffff81019e3a>] show_stack_log_lvl+0xea/0x170
> > [146792.281440] [<ffffffff8101abc1>] show_stack+0x21/0x40
> > [146792.281448] [<ffffffff8130f040>] dump_stack+0x5c/0x7c
> > [146792.281464] [<ffffffff8118c982>] warn_alloc_failed+0xe2/0x150
> > [146792.281471] [<ffffffff8118cddd>] __alloc_pages_nodemask+0x3ed/0xb20
> > [146792.281489] [<ffffffff811d3aaf>] alloc_pages_current+0x7f/0x100
> > [146792.281503] [<ffffffff815dfa2c>] vmemmap_alloc_block+0x79/0xb6
> > [146792.281510] [<ffffffff815dfbd3>] __vmemmap_alloc_block_buf+0x136/0x145
> > [146792.281524] [<ffffffff815dd0c5>] vmemmap_populate+0xd2/0x2b9
> > [146792.281529] [<ffffffff815dffd9>] sparse_mem_map_populate+0x23/0x30
> > [146792.281532] [<ffffffff815df88d>] sparse_add_one_section+0x68/0x18e
> > [146792.281537] [<ffffffff815d9f5a>] __add_pages+0x10a/0x1d0
> > [146792.281553] [<ffffffff8106249a>] arch_add_memory+0x4a/0xc0
> > [146792.281559] [<ffffffff815da1f9>] add_memory_resource+0x89/0x160
> > [146792.281564] [<ffffffff815da33d>] add_memory+0x6d/0xd0
> > [146792.281585] [<ffffffff813d36c4>] acpi_memory_device_add+0x181/0x251
> > [146792.281597] [<ffffffff813946e5>] acpi_bus_attach+0xfd/0x19b
> > [146792.281602] [<ffffffff81394866>] acpi_bus_scan+0x59/0x69
> > [146792.281604] [<ffffffff813949de>] acpi_device_hotplug+0xd2/0x41f
> > [146792.281608] [<ffffffff8138db67>] acpi_hotplug_work_fn+0x1a/0x23
> > [146792.281623] [<ffffffff81093cee>] process_one_work+0x14e/0x410
> > [146792.281630] [<ffffffff81094546>] worker_thread+0x116/0x490
> > [146792.281637] [<ffffffff810999ed>] kthread+0xbd/0xe0
> > [146792.281651] [<ffffffff815e4e7f>] ret_from_fork+0x3f/0x70
> >
> > and we do see many of those because essentially every the allocation
> > failes for each memory section. This is overly excessive way to tell
> > user that there is nothing to really worry about because we do have
> > a fallback mechanism to use base pages. The only downside might be a
> > performance degradation due to TLB pressure.
> >
> > This patch changes vmemmap_alloc_block to use __GFP_NOWARN and warn
> > explicitly once on the first allocation failure. This will reduce the
> > noise in the kernel log considerably, while we still have an indication
> > that a performance might be impacted.
> >
> > Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
> > Signed-off-by: Michal Hocko <mhocko@suse.com>
> > ---
> > Hi,
> > this has somehow fell of my radar completely. The patch is essentially
> > what Johannes suggested [1] so I have added his s-o-b and added the
> > changelog into it.
>
> Looks good to me.
I think it'd be better to change the ratelimit state
to something like once a minute
---
mm/page_alloc.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 82e6d2c914ab..af3f92beec04 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -3269,8 +3269,7 @@ void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...)
{
struct va_format vaf;
va_list args;
- static DEFINE_RATELIMIT_STATE(nopage_rs, DEFAULT_RATELIMIT_INTERVAL,
- DEFAULT_RATELIMIT_BURST);
+ static DEFINE_RATELIMIT_STATE(nopage_rs, HZ * 60, 1);
if ((gfp_mask & __GFP_NOWARN) || !__ratelimit(&nopage_rs))
return;
[-- Attachment #2: 1.difd --]
[-- Type: text/plain, Size: 584 bytes --]
mm/page_alloc.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 82e6d2c914ab..af3f92beec04 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -3269,8 +3269,7 @@ void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...)
{
struct va_format vaf;
va_list args;
- static DEFINE_RATELIMIT_STATE(nopage_rs, DEFAULT_RATELIMIT_INTERVAL,
- DEFAULT_RATELIMIT_BURST);
+ static DEFINE_RATELIMIT_STATE(nopage_rs, HZ * 60, 1);
if ((gfp_mask & __GFP_NOWARN) || !__ratelimit(&nopage_rs))
return;
next prev parent reply other threads:[~2017-11-06 17:57 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-11-06 9:22 [PATCH] mm, sparse: do not swamp log with huge vmemmap allocation failures Michal Hocko
2017-11-06 17:35 ` Johannes Weiner
2017-11-06 17:57 ` Joe Perches [this message]
2017-11-06 18:14 ` Khalid Aziz
2017-11-06 18:18 ` Michal Hocko
2017-11-06 20:17 ` Khalid Aziz
[not found] ` <20171106205644.29386-11-mathieu.desnoyers@efficios.com>
2017-11-07 0:37 ` [RFC PATCH for 4.15 10/14] cpu_opv: Wire up powerpc system call Nicholas Piggin
2017-11-07 0:47 ` Mathieu Desnoyers
2017-11-07 1:21 ` Nicholas Piggin
2017-11-07 9:06 ` [PATCH] mm, sparse: do not swamp log with huge vmemmap allocation failures Michal Hocko
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1509991069.2431.45.camel@perches.com \
--to=joe@perches.com \
--cc=akpm@linux-foundation.org \
--cc=hannes@cmpxchg.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mhocko@kernel.org \
--cc=mhocko@suse.com \
--cc=vbabka@suse.cz \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).