grsecurity forums

Juste compiled a new kernel 2.4.21, patched with grsec 1.9.11

The system isn't able to find the 128 Mo of my P1/75 (It finds just 10Mo), so i used to append="mem=128M" in the /etc/lilo.conf file.

It worked fine with my old 2.4.16 kernel, but not with the patched 2.4.21 any more.

I apologize i havn't yet read the entire doc... But i'm trying to fix a vulnerability problem on a prod server, and i'd like to have a solution as far as possible...

Thanx a lot if you can help !!

marc wrote:The system isn't able to find the 128 Mo of my P1/75 (It finds just 10Mo), so i used to append="mem=128M" in the /etc/lilo.conf file.

It worked fine with my old 2.4.16 kernel, but not with the patched 2.4.21 any more.

can you be more specific as to what does not work? kernel panic? application crashes? any logs/messages? etc...

can you be more specific as to what does not work? kernel panic? application crashes? any logs/messages? etc...

Sorry,

Nothing's wrong at runtime, except that free or top inform me that i only have 10 Mo of RAM...
And it's clear that the machine is very slow, always swapping, so it's not a bug in "free"

I had this trouble the first time i installed debian on this machine, and solve it naturally by using the append line in lilo... (and i'm perfectly sure it worked, then)
But this doesnt' work any more with 2.4.21+grsec.

Watching startup messages with dmesg, i saw that the append line was well passed as a kernel option, as it was before, with kernel 2.4.16. So it excludes a problem in lilo config, etc... However, the kernel seems to ignore it, without sending any error message... Quite amazing...

I'm not sure this trouble is linked to grsec, as i changed 2 things at the same time... The kernel, and the grsec patch. I've posted this question for the kernel too, but no constructive answer for the moment ...

I've also tried multiple compilation params, but nothing changed...

Thanx for your help !

marc wrote:Watching startup messages with dmesg, i saw that the append line was well passed as a kernel option, as it was before, with kernel 2.4.16. So it excludes a problem in lilo config, etc... However, the kernel seems to ignore it, without sending any error message... Quite amazing...

can you please post/email your .config and dmesg output (for both the working and non-working cases).