Re: RSBAC version 1.1.1 released

From: Amon Ott <ao@rsbac.org>
Subject: Re: RSBAC version 1.1.1 released
Date: Fri, 23 Mar 2001 16:52:39 +0100

Next Article (by Author): Re: RSBAC version 1.1.1 now really released Amon Ott
Previous Article (by Author): Re: RSBAC version 1.1.1 released - no more Amon Ott
Top of Thread: RSBAC version 1.1.1 released Amon Ott
Articles sorted by: [Date] [Author] [Subject]

On Fre, 23 Mär 2001 Bencsath Boldizsar wrote:
> After current 'unable to handle kernel paging request' stuff I checked the
> thread's stack, and it was about 4-5k free. (some seconds delayed of
> course from the panic;-))
> 
> I tried to change kmalloc to vmalloc, the problem disappeared from that IP
> to another ;-) all IP is in the read_fd_list. Every kmalloc in the
> aci_... .c has been changed. The panic tells an IP in that function. Is it
> possible that some macro or something gets further kmalloces in that
> routine, or the vmalloc dies?
> 
> Btw I put a debug message to the first part of the read_fd_list. The first
> version (without modification, pre7) showed only 1 try to kmalloc 256
> bytes. After changing to vmalloc there were about 8-9 kmalloc'es, and
> after that somewhere died.
> 
> I'll put some other error messages to get the right point of the failure.
> I think kmalloc can be a problem, read_fd_list uses read_lock and such
> functions. Maybe not in the rsbac code itself but in other code
> (reiserfs,smb, etc) there are kmalloces that grow up too much?

Good, please go on checking the memory alloc stuff. I will concentrate on the
stack.

> other thing You wrote that you count sizes and change to vmalloc if it's
> too much. But I think You not always do this, at some places You don't add
> these numbers. 

I do not where I know there is a fixed size far below the limit. If I am wrong
here, you found a bug. ;)
  
> And again, I am a pretty bad coder and hardly ever seen a kernel code so
> sorry if I say dumb things;-)

You just became a kernel hacker doing debugging, so stop excusing and go back
to work... ;)

Amon.
-
To unsubscribe from the rsbac list, send a mail to
majordomo@rsbac.org with
unsubscribe rsbac
as single line in the body.

Go to Compuniverse LWGate Home Page.