Messages in this thread |  | | From | "H. Peter Anvin" <> | Subject | Re: [PATCH] 2.4.0 i386 watchpoint problems | Date | 21 Sep 2000 11:36:58 -0700 |
| |
Followup to: <Pine.LNX.4.10.10009211031370.1348-100000@penguin.transmeta.com> By author: Linus Torvalds <torvalds@transmeta.com> In newsgroup: linux.dev.kernel > > Actually, even simpler approach: > > - always clear db7 after sending signal - don't test for pending or for > kernel mode at all at that point. > - re-load %db7 at the top of the for(;;) loop in do_signal(). > > That should mean that you will actually get a debug signal for ALL > triggers of the debug registers, except you'll only get one for kernel > mode. > > (For extra credit, make the patch use "eip=-1" in the siginfo structure if > the trap happened in kernel mode, so that we don't leak kernel addresses > to user space. Not that it would be much of a security issue, but it's a > beauty wart that could easily be cleared up). >
It could be (and so would even giving the signals in kernel mode)... remember that silly case of passwords crossing page boundaries?
-hpa -- <hpa@transmeta.com> at work, <hpa@zytor.com> in private! "Unix gives you enough rope to shoot yourself in the foot." http://www.zytor.com/~hpa/puzzle.txt - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/
|  |