[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: kernel/2731: reproducable kernel panic unless NMBCLUSTERS is increased
The following reply was made to PR kernel/2731; it has been noted by GNATS.
From: "Grudge Mason" <grudge_mason@hotmail.com>
To: art@blahonga.org, daniel@benzedrine.cx
Cc: gnats@openbsd.org
Subject: Re: kernel/2731: reproducable kernel panic unless NMBCLUSTERS is increased
Date: Fri, 07 Jun 2002 20:52:18 +0200
>Artur Grabowski wrote:
>Did you try with any other nic?
>Try this diff.
The diff seems to do the trick.
Thank you very much!
I'm in no way able to cause any harm to the kernel anymore.
I still get the "/bsd: WARNING: mclpool limit reached; increase NMBCLUSTERS"
messages though (two of 'em), but with the patch applied the kernel does not
panic anymore. It just keeps running as usual.
I tried with an fxp card (under GENERIC) which gave the same result as the
patched rtl (that is, two warning messages but everything still works as
expected).
I did a few "while [ 1 ]; do vmstat -m | grep pff; sleep 1; done"
as Daniel suggested (yepp, it has 512MB of memory. I did not change any PF
configurations such as -m frags).
Here are the results:
Realtek NIC under GENERIC with Art's patch:
pffrent 16 861 0 861 1 0 1 1 0 20
1
pffrag 32 21 0 21 1 0 1 1 0 8
1
pffrent 16 4722 0 4537 1 0 1 1 0 20
0
pffrag 32 117 0 108 1 0 1 1 0 8
0
pffrent 16 8812 0 8386 2 0 2 2 0 20
0
pffrag 32 221 0 196 1 0 1 1 0 8
0
pffrent 16 14306 0 13351 4 0 4 4 0 20
0
pffrag 32 361 0 307 1 0 1 1 0 8
0
pffrent 16 19894 0 18181 7 0 7 7 0 20
0
pffrag 32 504 0 416 1 0 1 1 0 8
0
Jun 7 10:46:06 foo /bsd: WARNING: mclpool limit reached; increase
NMBCLUSTERS
Jun 7 10:46:06 foo /bsd: WARNING: mclpool limit reached; increase
NMBCLUSTERS
pffrent 16 24921 0 22793 9 0 9 9 0 20
0
pffrag 32 630 0 523 1 0 1 1 0 8
0
pffrent 16 29194 0 27078 9 0 9 9 0 20
0
pffrag 32 734 0 627 1 0 1 1 0 8
0
pffrent 16 33336 0 31219 9 0 9 9 0 20
0
pffrag 32 836 0 728 1 0 1 1 0 8
0
pffrent 16 37570 0 35442 9 0 9 9 0 20
0
pffrag 32 939 0 831 1 0 1 1 0 8
0
pffrent 16 39969 0 37740 9 0 9 9 0 20
0
pffrag 32 1002 0 887 1 0 1 1 0 8
0
pffrent 16 39971 0 37740 9 0 9 9 0 20
0
pffrag 32 1003 0 887 1 0 1 1 0 8
0
pffrent 16 39973 0 37740 9 0 9 9 0 20
0
pffrag 32 1004 0 887 1 0 1 1 0 8
0
pffrent 16 39973 0 37740 9 0 9 9 0 20
0
pffrag 32 1004 0 887 1 0 1 1 0 8
0
Same thing but without Art's patch:
pffrent 16 2247 0 2214 1 0 1 1 0 20
0
pffrag 32 55 0 54 1 0 1 1 0 8
0
pffrent 16 6381 0 6355 1 0 1 1 0 20
0
pffrag 32 156 0 155 1 0 1 1 0 8
0
pffrent 16 10626 0 10352 2 0 2 2 0 20
0
pffrag 32 260 0 248 1 0 1 1 0 8
0
pffrent 16 14867 0 14542 2 0 2 2 0 20
0
pffrag 32 365 0 349 1 0 1 1 0 8
0
pffrent 16 18937 0 18366 3 0 3 3 0 20
0
pffrag 32 464 0 440 1 0 1 1 0 8
0
pffrent 16 23051 0 22466 3 0 3 3 0 20
0
pffrag 32 565 0 540 1 0 1 1 0 8
0
pffrent 16 27266 0 26504 4 0 4 4 0 20
0
pffrag 32 668 0 637 1 0 1 1 0 8
0
pffrent 16 31332 0 30529 4 0 4 4 0 20
0
pffrag 32 767 0 735 1 0 1 1 0 8
0
pffrent 16 35510 0 34623 4 0 4 4 0 20
0
pffrag 32 869 0 834 1 0 1 1 0 8
0
pffrent 16 39535 0 38392 5 0 5 5 0 20
0
pffrag 32 968 0 921 1 0 1 1 0 8
0
pffrent 16 43201 0 42041 5 0 5 5 0 20
0
pffrag 32 1058 0 1010 1 0 1 1 0 8
0
pffrent 16 47414 0 46264 5 0 5 5 0 20
0
pffrag 32 1161 0 1113 1 0 1 1 0 8
0
pffrent 16 51472 0 50115 6 0 6 6 0 20
0
pffrag 32 1261 0 1203 1 0 1 1 0 8
0
pffrent 16 55634 0 54230 6 0 6 6 0 20
0
pffrag 32 1363 0 1301 1 0 1 1 0 8
0
pffrent 16 59699 0 58289 6 0 6 6 0 20
0
pffrag 32 1463 0 1400 1 0 1 1 0 8
0
pffrent 16 63709 0 62124 7 0 7 7 0 20
0
pffrag 32 1561 0 1491 1 0 1 1 0 8
0
pffrent 16 67717 0 66142 7 0 7 7 0 20
0
pffrag 32 1659 0 1589 1 0 1 1 0 8
0
pffrent 16 71724 0 70119 7 0 7 7 0 20
0
pffrag 32 1756 0 1686 1 0 1 1 0 8
0
pffrent 16 75881 0 74018 8 0 8 8 0 20
0
pffrag 32 1858 0 1780 1 0 1 1 0 8
0
pffrent 16 80026 0 78200 8 0 8 8 0 20
0
pffrag 32 1959 0 1882 1 0 1 1 0 8
0
pffrent 16 84030 0 82064 8 0 8 8 0 20
0
pffrag 32 2057 0 1973 1 0 1 1 0 8
0
pffrent 16 88170 0 86180 9 0 9 9 0 20
1
pffrag 32 2158 0 2073 1 0 1 1 0 8
0
Jun 7 13:36:09 foo /bsd: WARNING: mclpool limit reached; increase
NMBCLUSTERS
Jun 7 13:36:09 foo /bsd: WARNING: mclpool limit reached; increase
NMBCLUSTERS
And then immediately followed by a panic.
On GENERIC with an fxp NIC (but still the same machine):
pffrent 16 64717 0 64717 8 0 8 8 0 20
8
pffrag 32 1582 0 1582 1 0 1 1 0 8
1
pffrent 16 64717 0 64717 8 0 8 8 0 20
8
pffrag 32 1582 0 1582 1 0 1 1 0 8
1
pffrent 16 64717 0 64717 8 0 8 8 0 20
8
pffrag 32 1582 0 1582 1 0 1 1 0 8
1
pffrent 16 66254 0 66234 8 0 8 8 0 20
7
pffrag 32 1620 0 1619 1 0 1 1 0 8
0
pffrent 16 71440 0 71400 8 0 8 8 0 20
7
pffrag 32 1746 0 1745 1 0 1 1 0 8
0
pffrent 16 77214 0 76535 8 0 8 8 0 20
5
pffrag 32 1893 0 1861 1 0 1 1 0 8
0
pffrent 16 83247 0 81965 8 0 8 8 0 20
2
pffrag 32 2045 0 1987 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
Jun 7 14:04:04 foo /bsd: WARNING: mclpool limit reached; increase
NMBCLUSTERS
Jun 7 14:04:04 foo /bsd: WARNING: mclpool limit reached; increase
NMBCLUSTERS
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
pffrag 32 2168 0 2075 1 0 1 1 0 8
0
pffrent 16 87945 0 85961 8 0 8 8 0 20
0
And everything keeps going as usual.
Just to avoid confusion... I have not experienced any problems with 3.1
GENERIC whatsoever in real life (using several different nics, (but no
realtek), and PF is enabled on all of them with different rules). This
machine was an exception and obviously Art found the sollution.
- Grudge
_________________________________________________________________
Join the world’s largest e-mail service with MSN Hotmail.
http://www.hotmail.com