On Mon, 2014-03-31 at 21:43 +0200, Oleg Nesterov wrote:
> Hello.
>
> x86 can not handle the rip-relative jmp/call instrsuctions, the probed
> task can be killed by general protection fault. I'll describe this in
> more details when I send the fixes. Now I am sending the preparations
> which (I hope) make sense anyway, please review.
>
> My main concern is 3/7. I know absolutely nothing about instruction
> decoding, so I can only guess what, for example, OPCODE1() == 0xff or
> MODRM_REG() == 2 actually means. Please review.
>
> Ananth, David, please ack/nack the first change, it affects powerpc/arm.
>
> Oleg.
>
> arch/x86/include/asm/uprobes.h | 7 +-
> arch/x86/kernel/uprobes.c | 343 ++++++++++++++++++++--------------------
> kernel/events/uprobes.c | 23 +---
> 3 files changed, 176 insertions(+), 197 deletions(-)
>
I've reviewed all 7 patches. Aside from a couple of nits (noted
elsewhere) that Oleg inherited, it looks good so far.
Reviewed-by: Jim Keniston <[email protected]>