From: Yao Qi <yao@codesourcery.com>
To: <gdb-patches@sourceware.org>
Subject: [PATCH 3/4] Stop prologue analysis when past the epilogue
Date: Thu, 03 Jul 2014 06:11:00 -0000 [thread overview]
Message-ID: <1404367792-23234-4-git-send-email-yao@codesourcery.com> (raw)
In-Reply-To: <1404367792-23234-1-git-send-email-yao@codesourcery.com>
We see a fail in gdb.trace/entry-values.exp on armv4t thumb,
bt^M
#0 0x000086fc in foo (i=0, i@entry=<optimized out>, j=2, j@entry=<optimized out>)^M
#1 0x00000002 in ?? ()^M
Backtrace stopped: previous frame identical to this frame (corrupt stack?)^M
(gdb) FAIL: gdb.trace/entry-values.exp: bt (1) (pattern 1)
The fail is caused by incorrect prologue analysis, which can be illustrated by
setting a breakpoint on function foo,
(gdb) disassemble foo
Dump of assembler code for function foo:
0x000086e8 <+0>: push {r7, lr}
0x000086ea <+2>: sub sp, #8
0x000086ec <+4>: add r7, sp, #0
0x000086ee <+6>: str r0, [r7, #4]
0x000086f0 <+8>: str r1, [r7, #0]
0x000086f2 <+10>: movs r3, #0
0x000086f4 <+12>: adds r0, r3, #0
0x000086f6 <+14>: mov sp, r7
0x000086f8 <+16>: add sp, #8
0x000086fa <+18>: pop {r7}
0x000086fc <+20>: pop {r1}
0x000086fe <+22>: bx r1
End of assembler dump.
(gdb) b foo
Breakpoint 1 at 0x86fc
As we can see, GDB analyzes the prologue and skip the prologue to the last
instruction but one. The breakpoint is set within the epilogue, and GDB
skips too many instruction for prologue. This patch teaches GDB to stop
prologue analysis when goes into the epilogue. With this patch applied,
GDB is able to unwind correctly,
(gdb) bt
#0 0x000086f6 in foo (i=0, i@entry=2, j=2, j@entry=3)
#1 0x00008718 in bar (i=<optimized out>)
#2 0x00008758 in main ()
gdb:
2014-07-02 Yao Qi <yao@codesourcery.com>
* arm-tdep.c (thumb_analyze_prologue): Break the loop if
thumb_instruction_restores_sp return true.
---
gdb/arm-tdep.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/gdb/arm-tdep.c b/gdb/arm-tdep.c
index 153ef42..72beeb1 100644
--- a/gdb/arm-tdep.c
+++ b/gdb/arm-tdep.c
@@ -754,6 +754,11 @@ thumb_analyze_prologue (struct gdbarch *gdbarch,
regs[ARM_SP_REGNUM] = pv_add_constant (regs[ARM_SP_REGNUM],
-offset);
}
+ else if (thumb_instruction_restores_sp (insn))
+ {
+ /* Don't scan past the epilogue. */
+ break;
+ }
else if ((insn & 0xf800) == 0xa800) /* add Rd, sp, #imm */
regs[bits (insn, 8, 10)] = pv_add_constant (regs[ARM_SP_REGNUM],
(insn & 0xff) << 2);
--
1.9.0
next prev parent reply other threads:[~2014-07-03 6:11 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-07-03 6:11 [PATCH 0/4] Fix gdb.trace/entry-values.exp fails in thumb mode Yao Qi
2014-07-03 6:11 ` [PATCH 1/4] Restrict matching add/sub sp, #imm Yao Qi
2014-07-03 8:31 ` Will Newton
2014-07-07 1:38 ` Yao Qi
2014-07-11 13:25 ` Joel Brobecker
2014-07-11 13:49 ` Yao Qi
2014-07-03 6:11 ` [PATCH 4/4] Fix gdb.trace/entry-values.exp for thumb mode Yao Qi
2014-07-07 15:15 ` Joel Brobecker
2014-07-08 8:53 ` Yao Qi
2014-07-03 6:11 ` Yao Qi [this message]
2014-07-03 8:39 ` [PATCH 3/4] Stop prologue analysis when past the epilogue Will Newton
2014-07-11 13:39 ` Joel Brobecker
2014-07-03 6:11 ` [PATCH 2/4] Match instruction adjusts SP in thumb Yao Qi
2014-07-03 8:35 ` Will Newton
2014-07-11 13:25 ` Joel Brobecker
2014-09-24 12:56 ` Yao Qi
2014-07-11 7:34 ` [PATCH 0/4] Fix gdb.trace/entry-values.exp fails in thumb mode Yao Qi
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1404367792-23234-4-git-send-email-yao@codesourcery.com \
--to=yao@codesourcery.com \
--cc=gdb-patches@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox