[RFA] Submit process record and replay fourth time, 5/8

[RFA] Submit process record and replay fourth time, 5/8

[Hui Zhu]

[-- Attachment #1: Type: text/plain, Size: 930 bytes --]

This patch to add some process record and replay to infrun.c.

Code for function "use_displaced_stepping" is make sure that displaced
stepping function will disable when process record and replay target
is opened.  Because process record and replay target doesn't support
displaced stepping function.

Code for function "proceed" is call function "record_not_record_set"
to set process record and replay target doesn't record the execute
log.  Because when GDB resume the inferior, process record and replay
target doesn't need to record the memory and register store operation
of GDB.

2009-03-21  Hui Zhu  <teawater@gmail.com>

	* infrun.c (use_displaced_stepping): Return false if process
	record and replay target is used.
	(proceed): Call function "record_not_record_set" if pocess
	record and replay target is used.

	2008-12-26  Michael Snyder  <msnyder@vmware.com>
	* infrun.c (proceed): Don't call record_not_record_set.

[-- Attachment #2: 5-infrun.txt --]
[-- Type: text/plain, Size: 1595 bytes --]

---
 infrun.c |   12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

--- a/infrun.c
+++ b/infrun.c
@@ -49,6 +49,8 @@
 #include "mi/mi-common.h"
 #include "event-top.h"
 
+#include "record.h"
+
 /* Prototypes for local functions */
 
 static void signals_info (char *, int);
@@ -603,7 +605,8 @@ use_displaced_stepping (struct gdbarch *
   return (((can_use_displaced_stepping == can_use_displaced_stepping_auto
 	    && non_stop)
 	   || can_use_displaced_stepping == can_use_displaced_stepping_on)
-	  && gdbarch_displaced_step_copy_insn_p (gdbarch));
+	  && gdbarch_displaced_step_copy_insn_p (gdbarch)
+	  && current_target.to_stratum != record_stratum);
 }
 
 /* Clean out any stray displaced stepping state.  */
@@ -2108,6 +2111,10 @@ adjust_pc_after_break (struct execution_
   if (software_breakpoint_inserted_here_p (breakpoint_pc)
       || (non_stop && moribund_breakpoint_here_p (breakpoint_pc)))
     {
+      struct cleanup *old_cleanups = NULL;
+      if (current_target.to_stratum == record_stratum)
+	old_cleanups = record_gdb_operation_disable_set ();
+
       /* When using hardware single-step, a SIGTRAP is reported for both
 	 a completed single-step and a software breakpoint.  Need to
 	 differentiate between the two, as the latter needs adjusting
@@ -2131,6 +2138,9 @@ adjust_pc_after_break (struct execution_
 	  || !currently_stepping (ecs->event_thread)
 	  || ecs->event_thread->prev_pc == breakpoint_pc)
 	regcache_write_pc (regcache, breakpoint_pc);
+
+      if (current_target.to_stratum == record_stratum)
+	do_cleanups (old_cleanups);
     }
 }
 

Re: [RFA] Submit process record and replay fourth time, 5/8

[Hui Zhu]

[-- Attachment #1: Type: text/plain, Size: 1154 bytes --]

Update follow "3-record_target.txt".

Thanks,
Hui

On Sun, Mar 22, 2009 at 00:02, Hui Zhu <teawater@gmail.com> wrote:
> This patch to add some process record and replay to infrun.c.
>
> Code for function "use_displaced_stepping" is make sure that displaced
> stepping function will disable when process record and replay target
> is opened.  Because process record and replay target doesn't support
> displaced stepping function.
>
> Code for function "proceed" is call function "record_not_record_set"
> to set process record and replay target doesn't record the execute
> log.  Because when GDB resume the inferior, process record and replay
> target doesn't need to record the memory and register store operation
> of GDB.
>
> 2009-03-21  Hui Zhu  <teawater@gmail.com>
>
>        * infrun.c (use_displaced_stepping): Return false if process
>        record and replay target is used.
>        (proceed): Call function "record_not_record_set" if pocess
>        record and replay target is used.
>
>        2008-12-26  Michael Snyder  <msnyder@vmware.com>
>        * infrun.c (proceed): Don't call record_not_record_set.
>

[-- Attachment #2: 5-infrun.txt --]
[-- Type: text/plain, Size: 1564 bytes --]

---
 infrun.c |   12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

--- a/infrun.c
+++ b/infrun.c
@@ -49,6 +49,8 @@
 #include "mi/mi-common.h"
 #include "event-top.h"
 
+#include "record.h"
+
 /* Prototypes for local functions */
 
 static void signals_info (char *, int);
@@ -603,7 +605,8 @@ use_displaced_stepping (struct gdbarch *
   return (((can_use_displaced_stepping == can_use_displaced_stepping_auto
 	    && non_stop)
 	   || can_use_displaced_stepping == can_use_displaced_stepping_on)
-	  && gdbarch_displaced_step_copy_insn_p (gdbarch));
+	  && gdbarch_displaced_step_copy_insn_p (gdbarch)
+	  && current_target.to_stratum != record_stratum);
 }
 
 /* Clean out any stray displaced stepping state.  */
@@ -2109,6 +2112,10 @@ adjust_pc_after_break (struct execution_
   if (software_breakpoint_inserted_here_p (breakpoint_pc)
       || (non_stop && moribund_breakpoint_here_p (breakpoint_pc)))
     {
+      struct cleanup *old_cleanups = NULL;
+      if (current_target.to_stratum == record_stratum)
+	old_cleanups = record_gdb_operation_disable_set ();
+
       /* When using hardware single-step, a SIGTRAP is reported for both
 	 a completed single-step and a software breakpoint.  Need to
 	 differentiate between the two, as the latter needs adjusting
@@ -2132,6 +2139,9 @@ adjust_pc_after_break (struct execution_
 	  || !currently_stepping (ecs->event_thread)
 	  || ecs->event_thread->prev_pc == breakpoint_pc)
 	regcache_write_pc (regcache, breakpoint_pc);
+
+      if (old_cleanups)
+	do_cleanups (old_cleanups);
     }
 }
 

Re: [RFA] Submit process record and replay fourth time, 5/8

[Hui Zhu]

Add RECORD_IS_USED.
---
 infrun.c |   11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

--- a/infrun.c
+++ b/infrun.c
@@ -48,6 +48,7 @@
 #include "gdb_assert.h"
 #include "mi/mi-common.h"
 #include "event-top.h"
+#include "record.h"

 /* Prototypes for local functions */

@@ -603,7 +604,8 @@ use_displaced_stepping (struct gdbarch *
   return (((can_use_displaced_stepping == can_use_displaced_stepping_auto
 	    && non_stop)
 	   || can_use_displaced_stepping == can_use_displaced_stepping_on)
-	  && gdbarch_displaced_step_copy_insn_p (gdbarch));
+	  && gdbarch_displaced_step_copy_insn_p (gdbarch)
+	  && !RECORD_IS_USED);
 }

 /* Clean out any stray displaced stepping state.  */
@@ -2130,6 +2132,10 @@ adjust_pc_after_break (struct execution_
   if (software_breakpoint_inserted_here_p (breakpoint_pc)
       || (non_stop && moribund_breakpoint_here_p (breakpoint_pc)))
     {
+      struct cleanup *old_cleanups = NULL;
+      if (RECORD_IS_USED)
+	old_cleanups = record_gdb_operation_disable_set ();
+
       /* When using hardware single-step, a SIGTRAP is reported for both
 	 a completed single-step and a software breakpoint.  Need to
 	 differentiate between the two, as the latter needs adjusting
@@ -2153,6 +2159,9 @@ adjust_pc_after_break (struct execution_
 	  || !currently_stepping (ecs->event_thread)
 	  || ecs->event_thread->prev_pc == breakpoint_pc)
 	regcache_write_pc (regcache, breakpoint_pc);
+
+      if (old_cleanups)
+	do_cleanups (old_cleanups);
     }
 }

Re: [RFA] Submit process record and replay fourth time, 5/8

[Pedro Alves]

On Wednesday 15 April 2009 18:04:48, Hui Zhu wrote:

> 
>  /* Clean out any stray displaced stepping state.  */
> @@ -2130,6 +2132,10 @@ adjust_pc_after_break (struct execution_
>    if (software_breakpoint_inserted_here_p (breakpoint_pc)
>        || (non_stop && moribund_breakpoint_here_p (breakpoint_pc)))
>      {
> +      struct cleanup *old_cleanups = NULL;
> +      if (RECORD_IS_USED)
> +	old_cleanups = record_gdb_operation_disable_set ();
> +
>        /* When using hardware single-step, a SIGTRAP is reported for both
>  	 a completed single-step and a software breakpoint.  Need to
>  	 differentiate between the two, as the latter needs adjusting
> @@ -2153,6 +2159,9 @@ adjust_pc_after_break (struct execution_
>  	  || !currently_stepping (ecs->event_thread)
>  	  || ecs->event_thread->prev_pc == breakpoint_pc)
>  	regcache_write_pc (regcache, breakpoint_pc);
> +
> +      if (old_cleanups)
> +	do_cleanups (old_cleanups);
>      }
>  }

Is this in the last version?  This still has the problem with the
NULL cleanup I've just explained.

-- 
Pedro Alves

Re: [RFA] Submit process record and replay fourth time, 5/8

[Hui Zhu]

>
> Is this in the last version?  This still has the problem with the
> NULL cleanup I've just explained.
>

The following fix it, this is the new version:

---
 infrun.c |   11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

--- a/infrun.c
+++ b/infrun.c
@@ -48,6 +48,7 @@
 #include "gdb_assert.h"
 #include "mi/mi-common.h"
 #include "event-top.h"
+#include "record.h"

 /* Prototypes for local functions */

@@ -603,7 +604,8 @@ use_displaced_stepping (struct gdbarch *
   return (((can_use_displaced_stepping == can_use_displaced_stepping_auto
 	    && non_stop)
 	   || can_use_displaced_stepping == can_use_displaced_stepping_on)
-	  && gdbarch_displaced_step_copy_insn_p (gdbarch));
+	  && gdbarch_displaced_step_copy_insn_p (gdbarch)
+	  && !RECORD_IS_USED);
 }

 /* Clean out any stray displaced stepping state.  */
@@ -2130,6 +2132,10 @@ adjust_pc_after_break (struct execution_
   if (software_breakpoint_inserted_here_p (breakpoint_pc)
       || (non_stop && moribund_breakpoint_here_p (breakpoint_pc)))
     {
+      struct cleanup *old_cleanups = NULL;
+      if (RECORD_IS_USED)
+	old_cleanups = record_gdb_operation_disable_set ();
+
       /* When using hardware single-step, a SIGTRAP is reported for both
 	 a completed single-step and a software breakpoint.  Need to
 	 differentiate between the two, as the latter needs adjusting
@@ -2153,6 +2159,9 @@ adjust_pc_after_break (struct execution_
 	  || !currently_stepping (ecs->event_thread)
 	  || ecs->event_thread->prev_pc == breakpoint_pc)
 	regcache_write_pc (regcache, breakpoint_pc);
+
+      if (RECORD_IS_USED)
+	do_cleanups (old_cleanups);
     }
 }

Re: [RFA] Submit process record and replay fourth time, 5/8

[Pedro Alves]

On Wednesday 22 April 2009 10:08:53, Hui Zhu wrote:
> >
> > Is this in the last version?  This still has the problem with the
> > NULL cleanup I've just explained.
> >
> 
> The following fix it, this is the new version:

Thanks, this is approved.  (ahem, again, ChangeLog entry.)


-- 
Pedro Alves

Re: [RFA] Submit process record and replay fourth time, 5/8

[Hui Zhu]

2009-04-28  Hui Zhu  <teawater@gmail.com>

       * infrun.c (use_displaced_stepping): Return false if process
       record and replay target is used.
       (proceed): Call function "record_not_record_set" if pocess
       record and replay target is used.

       2008-12-26  Michael Snyder  <msnyder@vmware.com>
       * infrun.c (proceed): Don't call record_not_record_set.

Thanks,
Hui

On Tue, Apr 28, 2009 at 06:22, Pedro Alves <pedro@codesourcery.com> wrote:
> On Wednesday 22 April 2009 10:08:53, Hui Zhu wrote:
>> >
>> > Is this in the last version?  This still has the problem with the
>> > NULL cleanup I've just explained.
>> >
>>
>> The following fix it, this is the new version:
>
> Thanks, this is approved.  (ahem, again, ChangeLog entry.)
>
>
> --
> Pedro Alves
>