diff mbox

[0/2] GDB process record and reverse debugging improvements for arm*-linux*

Message ID 52929058.1050202@linaro.org
State Accepted
Headers show

Commit Message

Omair Javaid Nov. 24, 2013, 11:48 p.m. UTC
On Mon 11 Nov 2013 01:24:06 PM PKT, Omair Javaid wrote:
> On Fri 08 Nov 2013 01:18:45 PM PKT, oza Pawandeep wrote:
>> Hi,
>>
>> The previous support for arm instruction and thumb instruction for
>> record and reply was added by me.
>> but there some reverse test case failing...
>> I had issues with employer disclaimer, so could not proceed.
>> I have got the disclaimer now and is up-to date with FSF.
>>
>> some of the queries to you regarding thumb2 support ?
>>
>> 1) did you bring in co processor and dsp instruction support for thumb2 ?
>>
>> 2) can you please run gdb-rever test suite and see if your code passes
>> the same ?
>>
>>
>> Regards,
>> Oza.
>>
>>
>> On Fri, Nov 8, 2013 at 12:55 PM, Eli Zaretskii <eliz@gnu.org> wrote:
>>>
>>>> Date: Fri, 08 Nov 2013 08:18:35 +0500
>>>> From: Omair Javaid <omair.javaid@linaro.org>
>>>> CC: gdb-patches@sourceware.org, Patch Tracking <patches@linaro.org>
>>>>
>>>> --- gdb/NEWS  11 Oct 2013 13:51:26 -0000      1.622
>>>> +++ gdb/NEWS  7 Nov 2013 23:58:26 -0000
>>>> @@ -3,6 +3,10 @@
>>>
>>> This part is OK, but please make sure to have 2 spaces between
>>> sentences
>>>
>>>> +* Improved support of process record-replay and reverse debugging on
>>>> +  arm*-linux* targets. Support for thumb32 and syscall instructions
>>>                          ^^
>>> here.
>>>
>>> Thanks.
>>
>
> Hello Oza,
>
> I really appreciate your work on arm process record and I am trying to 
> improve on the work you have done previously. I have sent 3 patches 
> upstream which include basic support for Thumb32 instructions and 
> system call instructions along with some fixes.
>
> There were around 1500 failures in gdb.reverse on arm that have been 
> significantly reduced. Here is a current status of gdb.reverse test 
> suite on arm targets:
>
> arm-native configuration:
> # of expected passes		2253
> # of unexpected failures	71
>
> arm-remote configuration:
> # of expected passes		2142
> # of unexpected failures	122
>
> I am currently working on patches to fix above unexpected failures and 
> also working on supporting co-processor and other unsupported 
> instruction types.
>
> --
> Omair.

After incorporating all suggestions I am posting a final version of all 
three patches posted earlier. Looking for a go ahead for commit.

== PATCH 0 ==

This patch adds a news entry about improved process record-replay on
arm*-linux* targets.
It also adds linux-record.o to gdb_target_obs in configure.tgt and
enables gdb.reverse testsuite for arm*-linux* targets.

gdb:

2013-11-08  Omair Javaid  <omair.javaid@linaro.org>

	* NEWS: Add note on improved process record-replay on arm*-linux*
	targets.
	* configure.tgt: Add linux-record.o to gdb_target_obs.

gdb/testsuite:

2013-10-24  Omair Javaid  <omair.javaid@linaro.org>

	* lib/gdb.exp: Enable gdb.reverse testsuite for arm*-linux* targets.


---
 gdb/NEWS                  |    4 ++++
 gdb/configure.tgt         |    2 +-
 gdb/testsuite/lib/gdb.exp |    6 ++++--
 3 files changed, 9 insertions(+), 3 deletions(-)

--

Comments

Eli Zaretskii Nov. 25, 2013, 3:40 a.m. UTC | #1
> Date: Mon, 25 Nov 2013 04:48:40 +0500
> From: Omair Javaid <omair.javaid@linaro.org>
> CC: Yao Qi <yao@codesourcery.com>, 
>  "gdb-patches@sourceware.org" <gdb-patches@sourceware.org>,
>  patches <patches@linaro.org>, Eli Zaretskii <eliz@gnu.org>
> 
> This patch adds a news entry about improved process record-replay on
> arm*-linux* targets.
> It also adds linux-record.o to gdb_target_obs in configure.tgt and
> enables gdb.reverse testsuite for arm*-linux* targets.
> 
> gdb:
> 
> 2013-11-08  Omair Javaid  <omair.javaid@linaro.org>
> 
> 	* NEWS: Add note on improved process record-replay on arm*-linux*
> 	targets.

The NEWS part is OK.

Thanks.
Omair Javaid Dec. 17, 2013, 10:22 a.m. UTC | #2
On Mon 25 Nov 2013 08:40:14 AM PKT, Eli Zaretskii wrote:
>> Date: Mon, 25 Nov 2013 04:48:40 +0500
>> From: Omair Javaid <omair.javaid@linaro.org>
>> CC: Yao Qi <yao@codesourcery.com>,
>>  "gdb-patches@sourceware.org" <gdb-patches@sourceware.org>,
>>  patches <patches@linaro.org>, Eli Zaretskii <eliz@gnu.org>
>>
>> This patch adds a news entry about improved process record-replay on
>> arm*-linux* targets.
>> It also adds linux-record.o to gdb_target_obs in configure.tgt and
>> enables gdb.reverse testsuite for arm*-linux* targets.
>>
>> gdb:
>>
>> 2013-11-08  Omair Javaid  <omair.javaid@linaro.org>
>>
>> 	* NEWS: Add note on improved process record-replay on arm*-linux*
>> 	targets.
>
> The NEWS part is OK.
>
> Thanks.
>

Ping! Looking for maintainer's approval for arm process record/replay 
improvement patches.
Pedro Alves Dec. 20, 2013, 12:38 p.m. UTC | #3
On 11/24/2013 11:48 PM, Omair Javaid wrote:

> After incorporating all suggestions I am posting a final version of all 
> three patches posted earlier. Looking for a go ahead for commit.

All three emails in the series have the same subject apart from
the numbers, which make it confusing.  It'd be good if you made
sure the subjects were actually different, and more specific,
in the git commit log.

Hmm, I'm also surprised to now find that this "0/2" email
actually contains a patch.  That makes it actually 3 patches
in the series.  This is not the standard de facto way to
send a series.  The best way is for 0 to be the series intro,
with no patch, and then have the patches be replies to the
intro mail.  (I'd suggest addressing the comments I just
sent to patches #1 and #2, and resend the whole series with
git send-mail, which does that for you automatically.)

> 
> == PATCH 0 ==
> 
> This patch adds a news entry about improved process record-replay on
> arm*-linux* targets.
> It also adds linux-record.o to gdb_target_obs in configure.tgt and
> enables gdb.reverse testsuite for arm*-linux* targets.

These could be two separate patches, with NEWS actually being the
last patch in the whole series.

> gdb/testsuite:
> 
> 2013-10-24  Omair Javaid  <omair.javaid@linaro.org>
> 
> 	* lib/gdb.exp: Enable gdb.reverse testsuite for arm*-linux* targets.

Write:

 	* lib/gdb.exp (supports_process_record): Return true for arm*-linux*.

Otherwise looks good.
diff mbox

Patch

diff --git a/gdb/NEWS b/gdb/NEWS
index 10834df..bcf3e09 100644
--- a/gdb/NEWS
+++ b/gdb/NEWS
@@ -3,6 +3,10 @@ 
 
 *** Changes since GDB 7.6
 
+* Improved support of process record-replay and reverse debugging on
+  arm*-linux* targets.  Support for thumb32 and syscall instructions
+  recording has been added.
+
 * GDB now supports Fission DWP file format version 2.
   http://gcc.gnu.org/wiki/DebugFission
 
diff --git a/gdb/configure.tgt b/gdb/configure.tgt
index ea0faf1..5410f72 100644
--- a/gdb/configure.tgt
+++ b/gdb/configure.tgt
@@ -89,7 +89,7 @@  arm*-wince-pe | arm*-*-mingw32ce*)
 arm*-*-linux*)
 	# Target: ARM based machine running GNU/Linux
 	gdb_target_obs="arm-tdep.o arm-linux-tdep.o glibc-tdep.o \
-			solib-svr4.o symfile-mem.o linux-tdep.o"
+			solib-svr4.o symfile-mem.o linux-tdep.o linux-record.o"
 	build_gdbserver=yes
 	;;
 arm*-*-netbsd* | arm*-*-knetbsd*-gnu)
diff --git a/gdb/testsuite/lib/gdb.exp b/gdb/testsuite/lib/gdb.exp
index bde4e48..047aefb 100644
--- a/gdb/testsuite/lib/gdb.exp
+++ b/gdb/testsuite/lib/gdb.exp
@@ -1774,7 +1774,8 @@  proc supports_process_record {} {
 	return [target_info gdb,use_precord]
     }
 
-    if { [istarget "x86_64-*-linux*"] || [istarget "i\[34567\]86-*-linux*"] } {
+    if { [istarget "arm*-*-linux*"] || [istarget "x86_64-*-linux*"]
+         || [istarget "i\[34567\]86-*-linux*"] } {
 	return 1
     }
 
@@ -1789,7 +1790,8 @@  proc supports_reverse {} {
 	return [target_info gdb,can_reverse]
     }
 
-    if { [istarget "x86_64-*-linux*"] || [istarget "i\[34567\]86-*-linux*"] } {
+    if { [istarget "arm*-*-linux*"] || [istarget "x86_64-*-linux*"]
+         || [istarget "i\[34567\]86-*-linux*"] } {
 	return 1
     }