Non-Exec stack patches
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
Quoted 3 lines · Andrew Morton
Which architectures are currently making their pre-page execute permissions depend upon
VM_EXEC?
Would additional arch patches be needed for this?
Yes, they would need some straight forward minor patches e.g. in the 32bit emulation. IA64 would be a candidate I guess.
i386 could do it on NX capable CPUs with PAE kernels (but it would require backporting some fixes from x86-64). However currently it doesn't make much sense because all x86 CPUs that support NX (AMD K8 currently only) support 64bit kernels and people can as well run 64bit kernels.
Doing it on i386 would only make sense if non 64bit capable CPUs ever get NX. I heard VIA may be planning that, but so far there is nothing in their shipping CPUs, so I guess we can skip that for now.
-Andi
Hi Andi.
Quoted 7 lines
Quoted 2 lines
Which architectures are currently making their pre-page execute permissions depend upon
VM_EXEC? Would additional arch patches be needed for this?Yes, they would need some straight forward minor patches e.g. in the 32bit emulation. IA64 would be a candidate I guess.
Quoted 6 lines
i386 could do it on NX capable CPUs with PAE kernels (but it would require backporting some fixes from x86-64). However currently it doesn't make much sense because all x86 CPUs that support NX (AMD K8 currently only) support 64bit kernels and people can as well run 64bit kernels.
Doing it on i386 would only make sense if non 64bit capable CPUs ever get NX. I heard VIA may be planning that, but so far there is nothing in their shipping CPUs, so I guess we can skip that for now.
Well, there's also the case that (unknown if rumour or confirmed) there will be AthlonXPs based on the K8 core that do NOT run 64bit code.
I would THINK they would include the NX bit but that's just a guess of course.
// Stefan
Quoted 2 lines
Well, there's also the case that (unknown if rumour or confirmed) there will be AthlonXPs based on the K8 core that do NOT run 64bit code.
Yes, you're right. Some HP laptops ship with such crippled chips.
Quoted 2 lines
I would THINK they would include the NX bit but that's just a guess of course.
Most likely yes.
But who buys such crippled CPUs will have to live with that. Or do a patch. NX support is mostly hype anyways, it doesn't give you much advantage.
-Andi
Hi Andi,
Quoted 6 lines · On Wed, Mar 24, 2004 at 12:27:39PM +0100, Andi Kleen wrote: Stefan…
Quoted 2 lines
I would THINK they would include the NX bit but that's just a guess of course.
Most likely yes.
But who buys such crippled CPUs will have to live with that. Or do a patch.
It should be straightforward. Put a different value in the protection_map if such a CPU is detected. That should be all.
Quoted 1 line
NX support is mostly hype anyways, it doesn't give you much advantage.
It puts the hurdle somewhat higher and for some exploits does prevent them. You can still overwrite the return address, but you need to have put code into the data section prior to this to exploit. This may or may not possible depending on how the daemon works. Marking the data section non-executable would help further ...
It's always a tradeoff. Given the simplicity of the patch, I'm in favour of having it.
Regards,
Signature · 3 lines
Kurt Garloff <[email protected]> [Koeln, DE] Physics:Plasma modeling <[email protected]> [TU Eindhoven, NL] Linux: SUSE Labs (Head) <[email protected]> [SUSE Nuernberg, DE]