Messages in this thread | | | Date | Tue, 6 Nov 2012 14:52:20 -0800 | From | Andrew Morton <> | Subject | Re: [PATCH 2/2 v2] mm: print out information of file affected by memory error |
| |
On Tue, 6 Nov 2012 17:45:05 -0500 Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> wrote:
> > "should be" and "unlikely" aren't very reassuring things to hear! > > Emitting a million lines into syslog is pretty poor behaviour and > > should be reliably avoided. > > So capping maximum lines of messages per some duration (a hour or a day) > is a possible option. BTW, even if we don't apply this patch, the kernel > can emit million lines of messages in the above-mentioned situation because > each memory error event emits a message like "MCE 0x3f57f4: dirty LRU page > recovery: Ignored" on syslog. If it's also bad, we need to do capping > also over existing printk()s, right?
Yes, that sounds like a bug report waiting to happen.
| |