* [PATCH][gdb/tdep] Fix gdb.base/large-frame.exp for aarch64
@ 2022-08-04 13:04 Tom de Vries via Gdb-patches
2022-08-04 13:17 ` Luis Machado via Gdb-patches
0 siblings, 1 reply; 2+ messages in thread
From: Tom de Vries via Gdb-patches @ 2022-08-04 13:04 UTC (permalink / raw)
To: gdb-patches
Hi,
On aarch64, I run into:
...
FAIL: gdb.base/large-frame.exp: optimize=-O0: backtrace
...
The problem is that the architecture-specific prologue analyzer fails to
handle the first two insns in the prologue properly:
...
0000000000400610 <func>:
400610: d2880210 mov x16, #0x4010
400614: cb3063ff sub sp, sp, x16
400618: a9007bfd stp x29, x30, [sp]
40061c: 910003fd mov x29, sp
400620: 910043a0 add x0, x29, #0x10
400624: 97fffff0 bl 4005e4 <blah>
...
so we get:
...
$ gdb -q -batch ./outputs/gdb.base/large-frame/large-frame-O0 -ex "b func"
Breakpoint 1 at 0x400614
...
Fix this by:
- fixing the support for the first insn to extract the immediate operand, and
- adding support for the second insn,
such that we have:
...
Breakpoint 1 at 0x400624
...
Note that we're overshooting by one insn (0x400620 is the first insn after the
prologue), but that's a pre-existing problem.
Tested on aarch64-linux.
Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29408
Any comments?
Thanks,
- Tom
[gdb/tdep] Fix gdb.base/large-frame.exp for aarch64
---
gdb/aarch64-tdep.c | 22 +++++++++++++++++++++-
1 file changed, 21 insertions(+), 1 deletion(-)
diff --git a/gdb/aarch64-tdep.c b/gdb/aarch64-tdep.c
index 8670197a888..f747ebda1ab 100644
--- a/gdb/aarch64-tdep.c
+++ b/gdb/aarch64-tdep.c
@@ -340,6 +340,20 @@ aarch64_analyze_prologue (struct gdbarch *gdbarch,
if (rn == AARCH64_SP_REGNUM && rd == AARCH64_FP_REGNUM)
seen_stack_set = true;
}
+ else if (inst.opcode->iclass == addsub_ext
+ && strcmp ("sub", inst.opcode->name) == 0)
+ {
+ unsigned rd = inst.operands[0].reg.regno;
+ unsigned rn = inst.operands[1].reg.regno;
+ unsigned rm = inst.operands[2].reg.regno;
+
+ gdb_assert (aarch64_num_of_operands (inst.opcode) == 3);
+ gdb_assert (inst.operands[0].type == AARCH64_OPND_Rd_SP);
+ gdb_assert (inst.operands[1].type == AARCH64_OPND_Rn_SP);
+ gdb_assert (inst.operands[2].type == AARCH64_OPND_Rm_EXT);
+
+ regs[rd] = pv_subtract (regs[rn], regs[rm]);
+ }
else if (inst.opcode->iclass == pcreladdr
&& inst.operands[1].type == AARCH64_OPND_ADDR_ADRP)
{
@@ -370,14 +384,20 @@ aarch64_analyze_prologue (struct gdbarch *gdbarch,
}
else if (inst.opcode->op == OP_MOVZ)
{
+ unsigned rd = inst.operands[0].reg.regno;
+
+ gdb_assert (aarch64_num_of_operands (inst.opcode) == 2);
gdb_assert (inst.operands[0].type == AARCH64_OPND_Rd);
+ gdb_assert (inst.operands[1].type == AARCH64_OPND_HALF);
+ gdb_assert (inst.operands[1].shifter.kind == AARCH64_MOD_LSL);
/* If this shows up before we set the stack, keep going. Otherwise
stop the analysis. */
if (seen_stack_set)
break;
- regs[inst.operands[0].reg.regno] = pv_unknown ();
+ regs[rd] = pv_constant (inst.operands[1].imm.value
+ << inst.operands[1].shifter.amount);
}
else if (inst.opcode->iclass == log_shift
&& strcmp (inst.opcode->name, "orr") == 0)
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH][gdb/tdep] Fix gdb.base/large-frame.exp for aarch64
2022-08-04 13:04 [PATCH][gdb/tdep] Fix gdb.base/large-frame.exp for aarch64 Tom de Vries via Gdb-patches
@ 2022-08-04 13:17 ` Luis Machado via Gdb-patches
0 siblings, 0 replies; 2+ messages in thread
From: Luis Machado via Gdb-patches @ 2022-08-04 13:17 UTC (permalink / raw)
To: Tom de Vries, gdb-patches
Hi,
Thanks for the patch. I was giving this a try before replying.
On 8/4/22 14:04, Tom de Vries wrote:
> Hi,
>
> On aarch64, I run into:
> ...
> FAIL: gdb.base/large-frame.exp: optimize=-O0: backtrace
> ...
>
> The problem is that the architecture-specific prologue analyzer fails to
> handle the first two insns in the prologue properly:
> ...
> 0000000000400610 <func>:
> 400610: d2880210 mov x16, #0x4010
> 400614: cb3063ff sub sp, sp, x16
> 400618: a9007bfd stp x29, x30, [sp]
> 40061c: 910003fd mov x29, sp
> 400620: 910043a0 add x0, x29, #0x10
> 400624: 97fffff0 bl 4005e4 <blah>
> ...
> so we get:
> ...
> $ gdb -q -batch ./outputs/gdb.base/large-frame/large-frame-O0 -ex "b func"
> Breakpoint 1 at 0x400614
> ...
>
> Fix this by:
> - fixing the support for the first insn to extract the immediate operand, and
> - adding support for the second insn,
> such that we have:
> ...
> Breakpoint 1 at 0x400624
> ...
> Note that we're overshooting by one insn (0x400620 is the first insn after the
> prologue), but that's a pre-existing problem.
>
> Tested on aarch64-linux.
>
> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29408
>
> Any comments?
>
> Thanks,
> - Tom
>
> [gdb/tdep] Fix gdb.base/large-frame.exp for aarch64
>
> ---
> gdb/aarch64-tdep.c | 22 +++++++++++++++++++++-
> 1 file changed, 21 insertions(+), 1 deletion(-)
>
> diff --git a/gdb/aarch64-tdep.c b/gdb/aarch64-tdep.c
> index 8670197a888..f747ebda1ab 100644
> --- a/gdb/aarch64-tdep.c
> +++ b/gdb/aarch64-tdep.c
> @@ -340,6 +340,20 @@ aarch64_analyze_prologue (struct gdbarch *gdbarch,
> if (rn == AARCH64_SP_REGNUM && rd == AARCH64_FP_REGNUM)
> seen_stack_set = true;
> }
> + else if (inst.opcode->iclass == addsub_ext
> + && strcmp ("sub", inst.opcode->name) == 0)
> + {
> + unsigned rd = inst.operands[0].reg.regno;
> + unsigned rn = inst.operands[1].reg.regno;
> + unsigned rm = inst.operands[2].reg.regno;
> +
> + gdb_assert (aarch64_num_of_operands (inst.opcode) == 3);
> + gdb_assert (inst.operands[0].type == AARCH64_OPND_Rd_SP);
> + gdb_assert (inst.operands[1].type == AARCH64_OPND_Rn_SP);
> + gdb_assert (inst.operands[2].type == AARCH64_OPND_Rm_EXT);
> +
> + regs[rd] = pv_subtract (regs[rn], regs[rm]);
> + }
> else if (inst.opcode->iclass == pcreladdr
> && inst.operands[1].type == AARCH64_OPND_ADDR_ADRP)
> {
> @@ -370,14 +384,20 @@ aarch64_analyze_prologue (struct gdbarch *gdbarch,
> }
> else if (inst.opcode->op == OP_MOVZ)
> {
> + unsigned rd = inst.operands[0].reg.regno;
> +
> + gdb_assert (aarch64_num_of_operands (inst.opcode) == 2);
> gdb_assert (inst.operands[0].type == AARCH64_OPND_Rd);
> + gdb_assert (inst.operands[1].type == AARCH64_OPND_HALF);
> + gdb_assert (inst.operands[1].shifter.kind == AARCH64_MOD_LSL);
>
> /* If this shows up before we set the stack, keep going. Otherwise
> stop the analysis. */
> if (seen_stack_set)
> break;
>
> - regs[inst.operands[0].reg.regno] = pv_unknown ();
> + regs[rd] = pv_constant (inst.operands[1].imm.value
> + << inst.operands[1].shifter.amount);
> }
> else if (inst.opcode->iclass == log_shift
> && strcmp (inst.opcode->name, "orr") == 0)
This looks good to me. Thanks for getting this fixed.
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2022-08-04 13:18 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-08-04 13:04 [PATCH][gdb/tdep] Fix gdb.base/large-frame.exp for aarch64 Tom de Vries via Gdb-patches
2022-08-04 13:17 ` Luis Machado via Gdb-patches
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox