Re: [-mm patch] ptrace: make {put,get}reg work again for gs and fs

[Frederik Deweerdt <deweerdt@free.fr>]

On Thu, Dec 21, 2006 at 11:22:05AM -0800, Jeremy Fitzhardinge wrote:
> Frederik Deweerdt wrote:
> > Following the i386 pda patches, it's not possible to set gs or fs value
> > from gdb anymore. The following patch restores the old behaviour of
> > getting and setting thread.gs of thread.fs respectively.
> > Here's a gdb session *before* the patch:
> > (gdb) info reg
> > [...]
> > fs             0x33     51
> > gs             0x33     51
> > (gdb) set $fs=0xffff
> > (gdb) info reg
> > [...]
> > fs             0x33     51
> > gs             0x33     51
> > (gdb) set $gs=0xffffffff
> > (gdb) info reg
> > [...]
> > fs             0xffff   65535
> > gs             0x33     51
> >
> > Another one *after* the patch:
> > (gdb) info reg
> > [...]
> > fs             0xd8     216
> >   
> 
> This doesn't look right.  This is the kernel's %fs, not usermode's
> (which should be 0).
> 
> > gs             0x33     51
> > (gdb) set $fs=0xffff
> > (gdb) info reg
> > [...]
> > fs             0xffff   65535
> > gs             0x33     51
> > (gdb) set $gs=0xffff
> > (gdb) info reg
> > [...]
> > fs             0xffff   65535
> > gs             0xffff   65535
> >   
> Hm.  This shouldn't be possible since this is a bad selector, but I
> guess ptrace/gdb doesn't really know that.  If you run the target (even
> single step it), these should revert to 0.
> 
Here's a third session that looks better:

(gdb) info reg
[...]
fs             0x0      0
gs             0x33     51
(gdb) set $fs=0xffff
(gdb) info reg
[...]
fs             0xffff   65535
gs             0x33     51
(gdb) set $gs=0xffff
(gdb) info reg
[...]
fs             0xffff   65535
gs             0xffff   65535
(gdb) n
Single stepping until exit from function main,
which has no line number information.
Cannot find user-level thread for LWP 10751: generic error
(gdb) set $gs=0x33
(gdb) set $fs=0
(gdb) n
Single stepping until exit from function main,
which has no line number information.
0x08048c05 in __i686.get_pc_thunk.bx ()
(gdb) info reg
[...]
fs             0x0      0
gs             0x33     51

This is a -mm1 kernel + your efl_offset fix + the attached patch.
So the problem came from putreg still saving %gs to the stack where
there's no slot for it, whereas getreg got things right.

Regards,
Frederik

Signed-off-by: Frederik Deweerdt <frederik.deweerdt@gmail.com>


diff --git a/arch/i386/kernel/ptrace.c b/arch/i386/kernel/ptrace.c
index a803a49..d8f44db 100644
--- a/arch/i386/kernel/ptrace.c
+++ b/arch/i386/kernel/ptrace.c
@@ -89,14 +89,14 @@ static int putreg(struct task_struct *child,
 	unsigned long regno, unsigned long value)
 {
 	switch (regno >> 2) {
-		case FS:
+		case GS:
 			if (value && (value & 3) != 3)
 				return -EIO;
-			child->thread.fs = value;
+			child->thread.gs = value;
 			return 0;
 		case DS:
 		case ES:
-		case GS:
+		case FS:
 			if (value && (value & 3) != 3)
 				return -EIO;
 			value &= 0xffff;
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/