Re: netbsd-6 instability - vmem

[Tom Ivar Helbekkmo <tih%hamartun.priv.no@localhost>]

Dave B <spam%y2013.dberg.net@localhost> writes:

> [lockups seem to happen when using raidframe-backed swap]

I had lockup problems with a system I'm no longer running, that didn't
have an awful lot of RAM, and used raidframe for everything, including
the swap partition.  Stability seemed to be improved by making raidframe
less dependent on always being able to allocate memory quickly, thus:

--- sys/dev/raidframe/rf_driver.c       10 Dec 2012 08:36:03 -0000      1.131
+++ sys/dev/raidframe/rf_driver.c       3 Feb 2013 09:37:23 -0000
@@ -461,7 +461,7 @@
        int i;
 
        /* XXX next line needs tuning... */
-       raidPtr->numEmergencyBuffers = 10 * raidPtr->numCol;
+       raidPtr->numEmergencyBuffers = 20 * raidPtr->numCol;
 #if DEBUG
        printf("raid%d: allocating %d buffers of %d bytes.\n",
               raidPtr->raidid,
@@ -487,7 +487,7 @@
        }
 
        /* XXX next line needs tuning too... */
-       raidPtr->numEmergencyStripeBuffers = 10;
+       raidPtr->numEmergencyStripeBuffers = 20;
         for (i = 0; i < raidPtr->numEmergencyStripeBuffers; i++) {
                 tmpbuf = malloc( raidPtr->numCol * 
(raidPtr->Layout.sectorsPerStripeUnit <<
                                  raidPtr->logBytesPerSector),


-tih
-- 
It doesn't matter how beautiful your theory is, it doesn't matter how smart
you are. If it doesn't agree with experiment, it's wrong.  -Richard Feynman