Re: RE: Broken memory management on system with no swap

From: Matthew Dillon <dillon_at_apollo.backplane.com>
Date: Mon, 21 Apr 2003 20:45:15 -0700 (PDT)
    I can't help you with gbde.  Looking at the mail archives it would
    appear that Poul answered a question from someone else in regards to
    these errors a little while back, which I have included below:

:>Also, when writing to the gbde encrypted filesystem I sporadically get
:>messages like:
:>
:>"ENOMEM 0xc26ced80 on 0xc2648b80(md1.bde)"
:
:This is when we cannot sleeplessly allocate memory in the I/O path.
:The request gets retried in that case.
:
:-- 
:Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20

						-Matt

:Matt,
:You patch, at least so far, appears to prevent processes from being
:killed due to lack of memory. Thanks! However, I am still seeing other
:memory-related error messages in the syslog:
:
:----------
:Apr 20 22:54:32 pakastelohi kernel: swap_pager_getswapspace: failed
:Apr 22 00:34:29 pakastelohi kernel: ENOMEM 0xc6bb8c00 on
:0xc6d91380(ad4s1.bde)
:Apr 22 00:34:34 pakastelohi kernel: ENOMEM 0xc6bc5300 on
:0xc6d91380(ad4s1.bde)
:Apr 22 00:48:49 pakastelohi kernel: ENOMEM 0xc65b8e00 on
:0xc6d91380(ad4s1.bde)
:Apr 22 00:52:51 pakastelohi kernel: ENOMEM 0xc6477000 on
:0xc6d91380(ad4s1.bde)
:Apr 22 00:52:51 pakastelohi kernel: ENOMEM 0xc860fc00 on
:0xc6d91380(ad4s1.bde)
:Apr 22 01:07:14 pakastelohi kernel: ENOMEM 0xc69b4900 on
:0xc6865880(ad6s1c.bde)
:Apr 22 01:07:14 pakastelohi kernel: ENOMEM 0xc85fee00 on
:0xc6865880(ad6s1c.bde)
:-----------
:
:I had reported the "swap_pager_getswapspace: failed" and GBDE "ENOMEM"
:errors in earlier, separate threads. The "swap_pager_getswapspace:
:failed" occurred most likely, though I am not positive, during the
:copying of a large file between two unencrypted partitions. The GBDE
:"ENOMEM" errors occurred while moving multiple GBs in directories
:containing files of various sizes, from a few MB to about 800MB, from
:one GBDE partition to another. (gbde.c,v 1.13)
:
:What I don't know is what the impact of these messages is: can they be
:safely ignored or are they indicative of processes failing to execute as
:requested and/or possible data corruption taking place?
:
:Thanks,
:--Lucky
Received on Mon Apr 21 2003 - 18:45:18 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:04 UTC