Re: buffer cache patches - anybody willing to summarize?

Harald Koenig (
Wed, 6 Aug 1997 19:40:09 +0200

On Aug 05, Pete Harlan wrote:

> Jeff Wiegley <> writes:
> > But, since the main purpose of 2.1.X -> 2.2/3.0 is the inclusion of
> > proper and complete SMP support, I don't think that any SMP patches
> > should be considered for the 2.0 tree. Only if they fix a gross security
> All SMP _bugfixes_ should go into 2.0. 2.2 improves SMP performance,
> not reliability.

I'd like to second that. and if it's common sense, I'd like to ask for
help since we do have lockup problems with dual-PPro200 PCs with both
clean 2.0.x (upto .30) and with Dave's pre-2.0.31-2 which does include
Leonard's SMP lockup patch -- it didn't help for us.

these are two PCs with ELITEgroup mainboards (FX/2 or similar) with 256MB DRAM each,
NCR810, now with DEC 21140 100MBit ethernet. one PC has a ELSA 8MB S3 968 card,
the other 8MB Matrox Millenium.

both PCs lock up most likely during [high? remember: 100Mbit!] network trafic.
this happend too when we first have used 3COM 3C905 cards (not the usual boomerang lockups,
the SMP problems are completely different and I can distinguish them), and
it's the same with tulip-0.76 driver now.

could this be a some generic problem in networking code not being SMP-proof ?

I've attached a serial terminal do ttyS1 for one PC and am using SERIAL_ECHO_PORT
but I've never seen any messages when the system locked up:(

any idea what I should try to localize the problem? any SMP debugging patches
or similar? using 2.1.X stuff is not an option here, sorry...


All SCSI disks will from now on                     ___       _____
be required to send an email notice                0--,|    /OOOOOOO\
24 hours prior to complete hardware failure!      <_/  /  /OOOOOOOOOOO\
                                                    \  \/OOOOOOOOOOOOOOO\
                                                      \ OOOOOOOOOOOOOOOOO|//
Harald Koenig,                                         \/\/\/\/\/\/\/\/\/
Inst.f.Theoret.Astrophysik                              //  /     \\  \                     ^^^^^       ^^^^^