On Mon, 12 Jun 2000, Olaf Petzold wrote: > Am Mit, 14 Jun 2000 schrieben Sie: > > On Mon, 12 Jun 2000, Olaf Petzold wrote: > > > > > I got the following message while removing the mbuff module: > > > > > > unloading mbuff > > > kmem_alloc: Bad slab magic (corrupt) (name=size-256) > > > > > > After this, the system was frozen. > > In my opinion memory corruption, either by mbuff or your code. What kernel > > /RT version (just send me your /proc/version). > > # cat /proc/version > Linux version 2.2.14-rtl2.2 (root@rtreg) (gcc version egcs-2.91.66 19990314/Linux >(egcs-1.1.2 release)) #6 Mon Apr 17 10:50:54 CEST 2000 Well, I hate to say this, because it is like shifting responsability to someone else, but up to my knowledge the "officially recommended" gcc version to compile Linux kernels is 2.7.2.3. I have for a long time been compiling 2.2.1x kernels with gcc 2.95.2 just to discover today that the long standing bug I have seen (immediate or delayed system hand when unregistering IDE interface) is probably due to gcc (kernel compiled with 2.7.2.3 works). This does not beed to be a bug in gcc, rather new gcc are optimizing stronger, and some parts of the kernel may contain assumptions no longer valid. I will of course look through mbuff code (good time to eliminate some week points) and your code, but please try to use gcc 2.7.2.3. I would like to ask what are the experiences of the others on this topic? I should also mention possible influence of flags like -DCPU= and optimizations on the layout of EXPORTED kernel structures. Best regards, -- Tomek -- [rtl] --- To unsubscribe: echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED] --- For more information on Real-Time Linux see: http://www.rtlinux.org/rtlinux/
