* [PATCH] Fix hppa_linux_sigtramp_frame_unwind_cache
@ 2008-09-13 23:58 John David Anglin
2008-09-14 2:44 ` Randolph Chung
0 siblings, 1 reply; 4+ messages in thread
From: John David Anglin @ 2008-09-13 23:58 UTC (permalink / raw)
To: gdb-patches; +Cc: randolph
The main fix in this patch is to remove the pad increment of scptr. As
far as I know, there is no pad field in the sigcontext struct between the
general registers and the floating point registers.
I also recorded the PSW and SAR registers. It might be a mistake to record
the PSW value, but I believe the SAR register definitely needs to be recorded.
Somewhat tested on hppa-linux but the gdb testsuite crashes 2.6.19.22.
Ok?
Dave
--
J. David Anglin dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada (613) 990-0752 (FAX: 952-6602)
2008-09-13 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
* hppa-linux-tdep.c (hppa_linux_sigtramp_frame_unwind_cache): Record
HPPA_IPSW_REGNUM and HPPA_SAR_REGNUM values. Remove pad increment.
Index: hppa-linux-tdep.c
===================================================================
RCS file: /cvs/src/src/gdb/hppa-linux-tdep.c,v
retrieving revision 1.31
diff -u -3 -p -r1.31 hppa-linux-tdep.c
--- hppa-linux-tdep.c 21 Aug 2008 13:19:18 -0000 1.31
+++ hppa-linux-tdep.c 13 Sep 2008 22:51:35 -0000
@@ -227,7 +229,8 @@ hppa_linux_sigtramp_frame_unwind_cache (
/* Skip sc_flags. */
scptr += 4;
- /* GR[0] is the psw, we don't restore that. */
+ /* GR[0] is the psw. */
+ info->saved_regs[HPPA_IPSW_REGNUM].addr = scptr;
scptr += 4;
/* General registers. */
@@ -237,9 +240,6 @@ hppa_linux_sigtramp_frame_unwind_cache (
scptr += 4;
}
- /* Pad. */
- scptr += 4;
-
/* FP regs; FP0-3 are not restored. */
scptr += (8 * 4);
@@ -262,6 +262,9 @@ hppa_linux_sigtramp_frame_unwind_cache (
info->saved_regs[HPPA_PCOQ_TAIL_REGNUM].addr = scptr;
scptr += 4;
+ info->saved_regs[HPPA_SAR_REGNUM].addr = scptr;
+ scptr += 4;
+
info->base = get_frame_register_unsigned (this_frame, HPPA_SP_REGNUM);
return info;
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [PATCH] Fix hppa_linux_sigtramp_frame_unwind_cache
2008-09-13 23:58 [PATCH] Fix hppa_linux_sigtramp_frame_unwind_cache John David Anglin
@ 2008-09-14 2:44 ` Randolph Chung
2008-09-14 3:47 ` John David Anglin
0 siblings, 1 reply; 4+ messages in thread
From: Randolph Chung @ 2008-09-14 2:44 UTC (permalink / raw)
To: John David Anglin; +Cc: gdb-patches
Hrm, this worries me.
The sigcontext structure looks like this:
struct sigcontext {
unsigned long sc_flags;
unsigned long sc_gr[32];
unsigned long long sc_fr[32];
unsigned long sc_iasq[2];
unsigned long sc_iaoq[2];
unsigned long sc_sar;
};
I am positive that when I wrote this the structure got padded between
sc_gr and sc_fr to keep the unsigned long long at a 8-byte boundary.
If this has changed now that means we've introduced a silent ABI
change....
The SAR stuff looks ok. I'm not very sure about restoring the IPSW either.
randolph
On Sat, Sep 13, 2008 at 4:57 PM, John David Anglin
<dave@hiauly1.hia.nrc.ca> wrote:
> The main fix in this patch is to remove the pad increment of scptr. As
> far as I know, there is no pad field in the sigcontext struct between the
> general registers and the floating point registers.
>
> I also recorded the PSW and SAR registers. It might be a mistake to record
> the PSW value, but I believe the SAR register definitely needs to be recorded.
>
> Somewhat tested on hppa-linux but the gdb testsuite crashes 2.6.19.22.
>
> Ok?
>
> Dave
> --
> J. David Anglin dave.anglin@nrc-cnrc.gc.ca
> National Research Council of Canada (613) 990-0752 (FAX: 952-6602)
>
> 2008-09-13 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
>
> * hppa-linux-tdep.c (hppa_linux_sigtramp_frame_unwind_cache): Record
> HPPA_IPSW_REGNUM and HPPA_SAR_REGNUM values. Remove pad increment.
>
> Index: hppa-linux-tdep.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/hppa-linux-tdep.c,v
> retrieving revision 1.31
> diff -u -3 -p -r1.31 hppa-linux-tdep.c
> --- hppa-linux-tdep.c 21 Aug 2008 13:19:18 -0000 1.31
> +++ hppa-linux-tdep.c 13 Sep 2008 22:51:35 -0000
> @@ -227,7 +229,8 @@ hppa_linux_sigtramp_frame_unwind_cache (
> /* Skip sc_flags. */
> scptr += 4;
>
> - /* GR[0] is the psw, we don't restore that. */
> + /* GR[0] is the psw. */
> + info->saved_regs[HPPA_IPSW_REGNUM].addr = scptr;
> scptr += 4;
>
> /* General registers. */
> @@ -237,9 +240,6 @@ hppa_linux_sigtramp_frame_unwind_cache (
> scptr += 4;
> }
>
> - /* Pad. */
> - scptr += 4;
> -
> /* FP regs; FP0-3 are not restored. */
> scptr += (8 * 4);
>
> @@ -262,6 +262,9 @@ hppa_linux_sigtramp_frame_unwind_cache (
> info->saved_regs[HPPA_PCOQ_TAIL_REGNUM].addr = scptr;
> scptr += 4;
>
> + info->saved_regs[HPPA_SAR_REGNUM].addr = scptr;
> + scptr += 4;
> +
> info->base = get_frame_register_unsigned (this_frame, HPPA_SP_REGNUM);
>
> return info;
>
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [PATCH] Fix hppa_linux_sigtramp_frame_unwind_cache
2008-09-14 2:44 ` Randolph Chung
@ 2008-09-14 3:47 ` John David Anglin
0 siblings, 0 replies; 4+ messages in thread
From: John David Anglin @ 2008-09-14 3:47 UTC (permalink / raw)
To: Randolph Chung; +Cc: gdb-patches
> The sigcontext structure looks like this:
>
> struct sigcontext {
> unsigned long sc_flags;
> unsigned long sc_gr[32];
> unsigned long long sc_fr[32];
> unsigned long sc_iasq[2];
> unsigned long sc_iaoq[2];
> unsigned long sc_sar;
> };
>
> I am positive that when I wrote this the structure got padded between
> sc_gr and sc_fr to keep the unsigned long long at a 8-byte boundary.
No, you are correct. The long long field will be aligned at an 8-byte
boundary. I will keep the pad.
Dave
--
J. David Anglin dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada (613) 990-0752 (FAX: 952-6602)
^ permalink raw reply [flat|nested] 4+ messages in thread
[parent not found: <no.id>]
* Re: [PATCH] Fix hppa_linux_sigtramp_frame_unwind_cache
[not found] <no.id>
@ 2008-09-15 2:24 ` John David Anglin
0 siblings, 0 replies; 4+ messages in thread
From: John David Anglin @ 2008-09-15 2:24 UTC (permalink / raw)
To: John David Anglin; +Cc: randolph, gdb-patches
> No, you are correct. The long long field will be aligned at an 8-byte
> boundary. I will keep the pad.
Committed the following after retesting.
Dave
--
J. David Anglin dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada (613) 990-0752 (FAX: 952-6602)
2008-09-14 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
* hppa-linux-tdep.c (hppa_linux_sigtramp_frame_unwind_cache): Record
HPPA_IPSW_REGNUM and HPPA_SAR_REGNUM values.
Index: hppa-linux-tdep.c
===================================================================
RCS file: /cvs/src/src/gdb/hppa-linux-tdep.c,v
retrieving revision 1.32
diff -u -3 -p -r1.32 hppa-linux-tdep.c
--- hppa-linux-tdep.c 14 Sep 2008 14:08:42 -0000 1.32
+++ hppa-linux-tdep.c 15 Sep 2008 00:50:52 -0000
@@ -227,7 +227,8 @@ hppa_linux_sigtramp_frame_unwind_cache (
/* Skip sc_flags. */
scptr += 4;
- /* GR[0] is the psw, we don't restore that. */
+ /* GR[0] is the psw. */
+ info->saved_regs[HPPA_IPSW_REGNUM].addr = scptr;
scptr += 4;
/* General registers. */
@@ -237,7 +238,7 @@ hppa_linux_sigtramp_frame_unwind_cache (
scptr += 4;
}
- /* Pad. */
+ /* Pad to long long boundary. */
scptr += 4;
/* FP regs; FP0-3 are not restored. */
@@ -262,6 +263,8 @@ hppa_linux_sigtramp_frame_unwind_cache (
info->saved_regs[HPPA_PCOQ_TAIL_REGNUM].addr = scptr;
scptr += 4;
+ info->saved_regs[HPPA_SAR_REGNUM].addr = scptr;
+
info->base = get_frame_register_unsigned (this_frame, HPPA_SP_REGNUM);
return info;
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2008-09-15 2:24 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2008-09-13 23:58 [PATCH] Fix hppa_linux_sigtramp_frame_unwind_cache John David Anglin
2008-09-14 2:44 ` Randolph Chung
2008-09-14 3:47 ` John David Anglin
[not found] <no.id>
2008-09-15 2:24 ` John David Anglin
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox