diff mbox series

[v2,AArch64] Fix step-over-syscall.exp failure

Message ID 20200113174239.7819-3-luis.machado@linaro.org
State New
Headers show
Series [v2,AArch64] Fix step-over-syscall.exp failure | expand

Commit Message

Luis Machado Jan. 13, 2020, 5:42 p.m. UTC
New in v2:

- Reverted to using pc_adjust as bool/offset and added more comments to explain
how it is being used.

--

In particular, this one:

FAIL: gdb.base/step-over-syscall.exp: fork: displaced=on: check_pc_after_cross_syscall: single step over fork final pc

When ptrace fork event reporting is enabled, GDB gets a PTRACE_EVENT_FORK
event whenever the inferior executes the fork syscall.

Then the logic is that GDB needs to step the inferior yet again in order to
receive a predetermined SIGTRAP, but no execution takes place because the
signal was already queued for delivery. That means the PC should stay the same.

I noticed the aarch64 code is currently adjusting the PC in this situation,
making the inferior skip an instruction without executing it.

The following change checks if we did not execute the instruction
(pc - to == 0), making proper adjustments for such case.

Regression tested on aarch64-linux-gnu.

gdb/ChangeLog:

2020-01-13  Luis Machado  <luis.machado@linaro.org>

	* aarch64-tdep.c (struct aarch64_displaced_step_closure )
	<pc_adjust>: Adjust the documentation.
	(aarch64_displaced_step_fixup): Check if PC really moved before
	adjusting it.
---
 gdb/aarch64-tdep.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

-- 
2.17.1

Comments

Alan Hayward Jan. 21, 2020, 11:23 a.m. UTC | #1
This is ok too.

Alan.

> On 13 Jan 2020, at 17:42, Luis Machado <luis.machado@linaro.org> wrote:

> 

> New in v2:

> 

> - Reverted to using pc_adjust as bool/offset and added more comments to explain

> how it is being used.

> 

> --

> 

> In particular, this one:

> 

> FAIL: gdb.base/step-over-syscall.exp: fork: displaced=on: check_pc_after_cross_syscall: single step over fork final pc

> 

> When ptrace fork event reporting is enabled, GDB gets a PTRACE_EVENT_FORK

> event whenever the inferior executes the fork syscall.

> 

> Then the logic is that GDB needs to step the inferior yet again in order to

> receive a predetermined SIGTRAP, but no execution takes place because the

> signal was already queued for delivery. That means the PC should stay the same.

> 

> I noticed the aarch64 code is currently adjusting the PC in this situation,

> making the inferior skip an instruction without executing it.

> 

> The following change checks if we did not execute the instruction

> (pc - to == 0), making proper adjustments for such case.

> 

> Regression tested on aarch64-linux-gnu.

> 

> gdb/ChangeLog:

> 

> 2020-01-13  Luis Machado  <luis.machado@linaro.org>

> 

> 	* aarch64-tdep.c (struct aarch64_displaced_step_closure )

> 	<pc_adjust>: Adjust the documentation.

> 	(aarch64_displaced_step_fixup): Check if PC really moved before

> 	adjusting it.

> ---

> gdb/aarch64-tdep.c | 17 +++++++++++++----

> 1 file changed, 13 insertions(+), 4 deletions(-)

> 

> diff --git a/gdb/aarch64-tdep.c b/gdb/aarch64-tdep.c

> index da41e22130..6a9d34dc67 100644

> --- a/gdb/aarch64-tdep.c

> +++ b/gdb/aarch64-tdep.c

> @@ -2737,7 +2737,8 @@ struct aarch64_displaced_step_closure : public displaced_step_closure

>      is being displaced stepping.  */

>   int cond = 0;

> 

> -  /* PC adjustment offset after displaced stepping.  */

> +  /* PC adjustment offset after displaced stepping.  If 0, then we don't

> +     write the PC back, assuming the PC is already the right address.  */

>   int32_t pc_adjust = 0;

> };

> 

> @@ -3032,11 +3033,12 @@ aarch64_displaced_step_fixup (struct gdbarch *gdbarch,

> {

>   aarch64_displaced_step_closure *dsc = (aarch64_displaced_step_closure *) dsc_;

> 

> +  ULONGEST pc;

> +

> +  regcache_cooked_read_unsigned (regs, AARCH64_PC_REGNUM, &pc);

> +

>   if (dsc->cond)

>     {

> -      ULONGEST pc;

> -

> -      regcache_cooked_read_unsigned (regs, AARCH64_PC_REGNUM, &pc);

>       if (pc - to == 8)

> 	{

> 	  /* Condition is true.  */

> @@ -3052,6 +3054,13 @@ aarch64_displaced_step_fixup (struct gdbarch *gdbarch,

> 

>   if (dsc->pc_adjust != 0)

>     {

> +      /* Make sure the previous instruction was executed (that is, the PC

> +	 has changed).  If the PC didn't change, then discard the adjustment

> +	 offset.  Otherwise we may skip an instruction before its execution

> +	 took place.  */

> +      if ((pc - to) == 0)

> +	dsc->pc_adjust = 0;

> +

>       if (debug_displaced)

> 	{

> 	  debug_printf ("displaced: fixup: set PC to %s:%d\n",

> -- 

> 2.17.1

>
Luis Machado Jan. 21, 2020, 1:28 p.m. UTC | #2
Thanks. Pushed now.

On 1/21/20 8:23 AM, Alan Hayward wrote:
> This is ok too.

> 

> Alan.

> 

>> On 13 Jan 2020, at 17:42, Luis Machado <luis.machado@linaro.org> wrote:

>>

>> New in v2:

>>

>> - Reverted to using pc_adjust as bool/offset and added more comments to explain

>> how it is being used.

>>

>> --

>>

>> In particular, this one:

>>

>> FAIL: gdb.base/step-over-syscall.exp: fork: displaced=on: check_pc_after_cross_syscall: single step over fork final pc

>>

>> When ptrace fork event reporting is enabled, GDB gets a PTRACE_EVENT_FORK

>> event whenever the inferior executes the fork syscall.

>>

>> Then the logic is that GDB needs to step the inferior yet again in order to

>> receive a predetermined SIGTRAP, but no execution takes place because the

>> signal was already queued for delivery. That means the PC should stay the same.

>>

>> I noticed the aarch64 code is currently adjusting the PC in this situation,

>> making the inferior skip an instruction without executing it.

>>

>> The following change checks if we did not execute the instruction

>> (pc - to == 0), making proper adjustments for such case.

>>

>> Regression tested on aarch64-linux-gnu.

>>

>> gdb/ChangeLog:

>>

>> 2020-01-13  Luis Machado  <luis.machado@linaro.org>

>>

>> 	* aarch64-tdep.c (struct aarch64_displaced_step_closure )

>> 	<pc_adjust>: Adjust the documentation.

>> 	(aarch64_displaced_step_fixup): Check if PC really moved before

>> 	adjusting it.

>> ---

>> gdb/aarch64-tdep.c | 17 +++++++++++++----

>> 1 file changed, 13 insertions(+), 4 deletions(-)

>>

>> diff --git a/gdb/aarch64-tdep.c b/gdb/aarch64-tdep.c

>> index da41e22130..6a9d34dc67 100644

>> --- a/gdb/aarch64-tdep.c

>> +++ b/gdb/aarch64-tdep.c

>> @@ -2737,7 +2737,8 @@ struct aarch64_displaced_step_closure : public displaced_step_closure

>>       is being displaced stepping.  */

>>    int cond = 0;

>>

>> -  /* PC adjustment offset after displaced stepping.  */

>> +  /* PC adjustment offset after displaced stepping.  If 0, then we don't

>> +     write the PC back, assuming the PC is already the right address.  */

>>    int32_t pc_adjust = 0;

>> };

>>

>> @@ -3032,11 +3033,12 @@ aarch64_displaced_step_fixup (struct gdbarch *gdbarch,

>> {

>>    aarch64_displaced_step_closure *dsc = (aarch64_displaced_step_closure *) dsc_;

>>

>> +  ULONGEST pc;

>> +

>> +  regcache_cooked_read_unsigned (regs, AARCH64_PC_REGNUM, &pc);

>> +

>>    if (dsc->cond)

>>      {

>> -      ULONGEST pc;

>> -

>> -      regcache_cooked_read_unsigned (regs, AARCH64_PC_REGNUM, &pc);

>>        if (pc - to == 8)

>> 	{

>> 	  /* Condition is true.  */

>> @@ -3052,6 +3054,13 @@ aarch64_displaced_step_fixup (struct gdbarch *gdbarch,

>>

>>    if (dsc->pc_adjust != 0)

>>      {

>> +      /* Make sure the previous instruction was executed (that is, the PC

>> +	 has changed).  If the PC didn't change, then discard the adjustment

>> +	 offset.  Otherwise we may skip an instruction before its execution

>> +	 took place.  */

>> +      if ((pc - to) == 0)

>> +	dsc->pc_adjust = 0;

>> +

>>        if (debug_displaced)

>> 	{

>> 	  debug_printf ("displaced: fixup: set PC to %s:%d\n",

>> -- 

>> 2.17.1

>>

>
diff mbox series

Patch

diff --git a/gdb/aarch64-tdep.c b/gdb/aarch64-tdep.c
index da41e22130..6a9d34dc67 100644
--- a/gdb/aarch64-tdep.c
+++ b/gdb/aarch64-tdep.c
@@ -2737,7 +2737,8 @@  struct aarch64_displaced_step_closure : public displaced_step_closure
      is being displaced stepping.  */
   int cond = 0;
 
-  /* PC adjustment offset after displaced stepping.  */
+  /* PC adjustment offset after displaced stepping.  If 0, then we don't
+     write the PC back, assuming the PC is already the right address.  */
   int32_t pc_adjust = 0;
 };
 
@@ -3032,11 +3033,12 @@  aarch64_displaced_step_fixup (struct gdbarch *gdbarch,
 {
   aarch64_displaced_step_closure *dsc = (aarch64_displaced_step_closure *) dsc_;
 
+  ULONGEST pc;
+
+  regcache_cooked_read_unsigned (regs, AARCH64_PC_REGNUM, &pc);
+
   if (dsc->cond)
     {
-      ULONGEST pc;
-
-      regcache_cooked_read_unsigned (regs, AARCH64_PC_REGNUM, &pc);
       if (pc - to == 8)
 	{
 	  /* Condition is true.  */
@@ -3052,6 +3054,13 @@  aarch64_displaced_step_fixup (struct gdbarch *gdbarch,
 
   if (dsc->pc_adjust != 0)
     {
+      /* Make sure the previous instruction was executed (that is, the PC
+	 has changed).  If the PC didn't change, then discard the adjustment
+	 offset.  Otherwise we may skip an instruction before its execution
+	 took place.  */
+      if ((pc - to) == 0)
+	dsc->pc_adjust = 0;
+
       if (debug_displaced)
 	{
 	  debug_printf ("displaced: fixup: set PC to %s:%d\n",