Re: Bug#133631: A normal user can use libapt
Hello Jason.
El dom, 17-02-2002 a las 00:00, Jason Gunthorpe escribió:
>
> On 16 Feb 2002, Ayose wrote:
>
> > How? I saw that the error was produced in msync(2), so I went to the
> > source and add "return true" at the begin of MMap::Sync() and
> > MMap::Sync(unsigned long Start,unsigned long Stop). Now, msync(2) is
> > disabled, because its code is never executed; however, apt-* works as
> > well as before :-)
>
> Well, msync is not supposed to fail. Your system is broken. Are you using
> some weird filesystem/linux version?
>
Now I'm confused: libapt doesn't fail anymore!
I'm using XFS filesystem, with low-latency patch:
# uname -a
Linux setepo 2.4.18-pre9-xfs-lowlatency #2 vie feb 8 23:16:32 GMT 2002
i686 unknown
When I started my PC and read your email, I tried to run apt-cache
dump and it crashed like before. I thought that the problem could be in
kernel, so I restarted woith another previos kernel. I saw that
apt-cache dump did *not* crashed as root!, I restarted with a third
kernel, and get *no* error in apt-cache. I thought: "definitively there
is a problem in kernel", but I restarted again with the kernel
2.4.18-pre9-xfs-lowlatency and I made several test *without* segsfault.
IMHO, there was an error in cache files that was solved when I run
another kernel, and now the newest kernel doesn't fail anymore :-)
FYI, The others kernels are patched with
a) XFS and
b) XFS + preemptive kernel (Robert M. Love)
Both of them are downloaded from oss.sgi.com cvs.
Do you have an answer to why apt was crashing?
P.S: Thanks you very much for your answers
--
Ayose Cazorla León
Debian GNU/Linux - setepo
Reply to: