grsecurity forums

16064 execve("scripts/basic/fixdep", ["scripts/basic/fixdep", "drivers/net/e1000/.e1000_main.o."..., "drivers/net/e1000/e1000_main.o", "gcc -m32 -Wp,-MD,drivers/net/e10"...], [/* 109 vars */]) = 0
16064 uname({sys="Linux", node="localhost", ...}) = 0
16064 brk(0) = 0x805b18c
16064 brk(0x805c000) = 0x805c000
16064 old_mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb00dc000
16064 --- SIGSEGV (Segmentation fault) @ 0 (0) ---

why does 16064 get a SEGV here ? the old_mmap call seems valid...

this happens during kernel compilation with gcc
thanks for hints

Dodger wrote:why does 16064 get a SEGV here ? the old_mmap call seems valid...

probably a bug somewhere, try to reproduce it with the latest grsec test version, 2.6.11 is old and not supported anymore.

haha so now only test patches are supported? niiice

You are doing great work, theres no doubt about it.

But saying that the last stable release is out of date, sounds not good....

So what ? Do we have something more stable? Why is nothing like a 2.6.14 stable release then?

dont feel annoyed im just asking

Dodger wrote:But saying that the last stable release is out of date, sounds not good....

then i will remind you on what the PaX site says:

much like 2.6 itself, PaX for 2.6 is experimental, anything can break.

in other words, if you're expecting the slightest sense of 'stability' then you are not on 2.6. if you're on 2.6 then you already voted to be a guinea pig, so to speak.

So what ? Do we have something more stable? Why is nothing like a 2.6.14 stable release then?

PaX for 2.6.14 is more stable than it was for any previous 2.6 version (and it's normally true for any version). it will be released when all reported bugs are fixed. right now i'm waiting for feedback on the PAE/4GB+ crashes, but given how little (=zero) the affected people have reported back so far, i might as well just release it this weekend.

Hm

Pax for 2.6 is experimental ? hmmmm 2.6 is nothing thats brand-new ...

ok , thanks for pointing that