On Wednesday 23 March 2005 20:17, Jonathan S. Romero wrote:
> I am having problems with kernel stack overflow's on UML guests. Below
> is information surrounding the system causing the problems. I am
> basically looking to see if anyone notices anything blatently obvious
> about what's wrong here. I might also switch to just plain 2.6.11 to
> avoid any other external variables.
>
> I've been using 2.6.10-gentoo-r6 patched with the skas v8 patches from
> baisorblade's site as the host kernel.
>
> For the guest kernel I am using 2.4.27 (vanilla) patched with the
> following patches, in order.
>
> uml-patch-2.4.24-1base.patch.bz2
> uml-2.4.27-bs1
>
> Both from baisorblade's site. I've been running what I consider to be
> one of the most strenuous tests on a UML system, gcc compiles through
> gentoo portage.
>
> I am using the following startup script to launch the UML
>
>
> #!/bin/bash
> SYSTEM_NAME=webapps
> MEMORY_SIZE=512M
> HW_ADDRESS=FE:FD:78:CD:AC:EF
> tunctl -t uml-$SYSTEM_NAME
> brctl addif umlbr uml-$SYSTEM_NAME
> ifconfig uml-$SYSTEM_NAME up
> /uml/bin/linux ubd0=rootfs.$SYSTEM_NAME ubd1=swapfs.$SYSTEM_NAME umid=
> $SYSTEM_NAME con0=fd:0,fd:1 eth0=tuntap,uml-$SYSTEM_NAME,$HW_ADDRESS
> mem=$MEMORY_SIZE
>
> The relevant GCC options are presented below (I think I just found my
> problem):
> CFLAGS="-02 -mcpu=pentium4"
> CHOST="i386-pc-linux-gnu"
> CXXFLAGS="${CFLAGS}"
> MAKEOPTS="-j6"
>
> GCC Version:
>
> The MAKEOPTS were set to j6 because I was using distcc to help with
> kernel compilation. I dont think a system with 512M (guest) of ram can
> really handle 6 makefile branches in parallel. It must have started
> hitting the swap when it did this.
>
> Also, on my other systems I have been using mcpu=i686, this looks like
> this was a holdover from an earlier test. The host is a xeon server,
> but I dont think I should be reaching for this higher cpu target inside
> the UML until I have more stability problems locked down.
>
> Does the kernel stack order affect the size of the kernel stack? Can I
> increase this to 3 to help with some of the problems i'm noticing?
Theoretically yes, but the stack overflow panic is bogus...
Apply on top of what you have above the below patch:
http://user-mode-linux.sourceforge.net/work/current/2.4/2.4.27-1um/patches/stack-overflow
However, I'd also like to say that it seems that UML / 2.6 is more stable than
UML / 2.4
> The swapfile for this host is 512M in size, and I can confirm it is
> being configured via top.
--
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729
http://www.user-mode-linux.org/~blaisorblade
-------------------------------------------------------
This SF.net email is sponsored by Microsoft Mobile & Embedded DevCon 2005
Attend MEDC 2005 May 9-12 in Vegas. Learn more about the latest Windows
Embedded(r) & Windows Mobile(tm) platforms, applications & content. Register
by 3/29 & save $300 http://ads.osdn.com/?ad_id=6883&alloc_id=15149&op=click
_______________________________________________
User-mode-linux-user mailing list
User-mode-linux-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-user
