Re: [PATCHv3,Hurd] Add hardware watch support

[Samuel Thibault]

Hello,

Thomas Schwinge, le Tue 16 Sep 2014 10:59:47 +0200, a écrit :
> On Tue, 16 Sep 2014 01:09:50 +0200, Samuel Thibault <samuel.thibault@gnu.org> 
> wrote:
> > Thomas Schwinge, le Tue 16 Sep 2014 00:08:01 +0200, a écrit :
> > > Do you agree that thread_get_state(i386_DEBUG_STATE) should be
> > > returning the actual DR6,
> > 
> > Indeed.
> > 
> > > and where in GNU Mach would we need to copy the DR6
> > > register into the PCB?
> > 
> > it would be user_trap(), probably, in the T_DEBUG case.
> 
> Thanks for the pointer.  Something like the following does accomplish its
> task w.r.t. GDB, but there are some TODO items.  It might help to compare
> what the Linux kernel is doing; »git grep --cached -i dr6 -- arch/x86/«
> or similar.

I believe this will be fine to only expose the known-to-be-safe
information, and clean dr6:

diff --git a/i386/i386/trap.c b/i386/i386/trap.c
index 200cbcc..661bc6a 100644
--- a/i386/i386/trap.c
+++ b/i386/i386/trap.c
@@ -395,6 +395,10 @@ printf("user trap %d error %d sub %08x\n", type, code, 
subcode);
                        return 0;
                }
 #endif /* MACH_KDB */
+               /* Make the content of the debug status register (DR6)
+                  available to user space.  */
+               thread->pcb->ims.ids.dr[6] = get_dr6() & 0x600F;
+               set_dr6(0);
                exc = EXC_BREAKPOINT;
                code = EXC_I386_SGL;
                break;

Does it fix GDB too?

Samuel