[PATCH] Change message when reaching end of reverse history.
Metzger, Markus T
markus.t.metzger@intel.com
Thu Apr 4 06:22:14 GMT 2024
Hello,
I see no reason for changing the existing portion. The message is short enough:
No more reverse-execution history. Recording will continue.
Eli, are you OK with that change?
Regards,
Markus.
From: Alex Chronopoulos <achronop@gmail.com>
Sent: Wednesday, April 3, 2024 11:15 PM
To: Guinevere Larsen <blarsen@redhat.com>
Cc: Metzger, Markus T <markus.t.metzger@intel.com>; Eli Zaretskii (eliz@gnu.org) <eliz@gnu.org>; gdb-patches@sourceware.org
Subject: Re: [PATCH] Change message when reaching end of reverse history.
Thank you, Guinevere.
My understanding is that you are referring to the forward case. Does it mean we keep the backward case as is?
"No more history" sounds good to me. If that were the case, would the suggested forward message become: "No more history, recording will continue"?
On Wed, Apr 3, 2024 at 7:54 PM Guinevere Larsen <blarsen@redhat.com<mailto:blarsen@redhat.com>> wrote:
On 3/14/24 14:26, Alex Chronopoulos wrote:
> I think "reached beginning of ~" makes sense, but that wouldn't fix the
> confusion about whether GDB can continue executing (and recording) on
> its own.
>
> On the other hand, I worry just saying "switching to recording mode"
> might seem too arbitrary for users, if they don't know the context of
> the commit message.
That's correct. The current issue in the forward case is that execution can
continue. GDB will halt at the beginning of the reverse history because
emulation cannot proceed. However, execution can resume in the 'normal' mode,
and recording can also continue. I was attempting to inform the user that a
different type of execution would be possible to continue debugging. Backward
execution does not encounter the same issue because reaching the end of history
signifies the end of the reverse execution. I am unsure how to convey all these
points with fewer words unless we opt for less specificity. Please suggest
something in the preferred direction.
Markus, Alex,
Sorry about the last 2 weeks of silence, I had 2 weeks of vacation I had to take.
Back to the complicated wording, I asked a couple (non GDB developer) friends to get end users perspectives. One said that "switching to recording mode" was very confusing, since they thought GDB was already in recording mode. After some brainstorming, we came up with "recording will continue" as a way to express that the program can still execute forward
We can continue using the old "No more reverse-execution history." to explain why record more, or we can shorten it to "no more history", "no more recorded history", or some other variation.
What do you think?
Thank you in advance,
Alex
On Thu, Mar 14, 2024 at 11:26 AM Guinevere Larsen <blarsen@redhat.com<mailto:blarsen@redhat.com>> wrote:
On 14/03/2024 07:37, Metzger, Markus T wrote:
> Hello Alex,
>
> I like the idea, but I find the wording a bit verbose. Can we find something shorter?
> E.g. "reached beginning of reverse-execution history" vs "reached end of ~".
> Not sure if that's clear enough that we're recording again. We could also omit
> the "reached end of ~" part and just say "switching to recording mode".
I think "reached beginning of ~" makes sense, but that wouldn't fix the
confusion about whether GDB can continue executing (and recording) on
its own.
On the other hand, I worry just saying "switching to recording mode"
might seem too arbitrary for users, if they don't know the context of
the commit message.
I wonder if good documentation can offset the second issue, though...
>
>> if (uiout->is_mi_like_p ())
>> uiout->field_string ("reason", async_reason_lookup
>> (EXEC_ASYNC_NO_HISTORY));
>> + else if (execution_direction == EXEC_FORWARD)
>> + uiout->text ("\nNo more reverse-execution history for emulation. "
>> + "Going forward will continue executing and recording.\n");
>> else
>> - uiout->text ("\nNo more reverse-execution history.\n");
>> + {
>> + gdb_assert (execution_direction == EXEC_REVERSE);
>> + uiout->text ("\nNo more reverse-execution history.\n");
>> + }
> We should probably give MI a similar indication, although that would likely
> break existing MI consumers. Not sure how opt-in is handled in MI.
I don't know either, but relatively easy change would be to add a new
field to the mi response to show direction when recording is enabled, so
mi consumers could decide how to handle a no-history stop.
Or we could announce the change on GDB 15 release and hold off on
merging this after 15 branched? Not sure if this is the right process,
though.
I had a suggestion after FOSDEM to have GDB indicate the direction of
execution for every command when recording is enabled, and adding that
to MI would be a step in that direction (again, leaving it up to MI
consumers on how to report it to users). what do you think?
>
> regards,
> Markus.
>
>> -----Original Message-----
>> From: Alex Chronopoulos <achronop@gmail.com<mailto:achronop@gmail.com>>
>> Sent: Wednesday, March 13, 2024 9:49 PM
>> To: gdb-patches@sourceware.org<mailto:gdb-patches@sourceware.org>
>> Cc: Alex Chronopoulos <achronop@gmail.com<mailto:achronop@gmail.com>>
>> Subject: [PATCH] Change message when reaching end of reverse history.
>>
>> In a record session, when we move backward, GDB switches from normal
>> execution to simulation. Moving forward again, the emulation continues
>> until the end of the reverse history. When the end is reached, the
>> execution stops, and a warning message is shown. This message has been
>> modified to indicate that the forward emulation has reached the end, but
>> the execution can continue as normal, and the recording will also continue.
>>
>> Before this patch, the warning message shown in that case was the same as
>> in the reverse case. This meant that when the end of history was reached in
>> either backward or forward emulation, the same message was displayed:
>>
>> "No more reverse-execution history."
>>
>> This message remains for backward emulation. However, in forward
>> emulation,
>> it has been modified to:
>>
>> "No more reverse-execution history for emulation. Going forward will
>> continue executing and recording."
>>
>> The reason for this change is that the initial message was deceiving,
>> making the user believe that forward debugging could not continue.
>>
>> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=31224
>> ---
>> gdb/NEWS | 5 ++++
>> gdb/infrun.c | 8 ++++-
>> gdb/testsuite/gdb.btrace/non-stop.exp | 30 ++++++++++++-------
>> gdb/testsuite/gdb.reverse/break-precsave.exp | 4 +--
>> gdb/testsuite/gdb.reverse/break-reverse.exp | 2 +-
>> .../gdb.reverse/machinestate-precsave.exp | 2 +-
>> 6 files changed, 36 insertions(+), 15 deletions(-)
>>
>> diff --git a/gdb/NEWS b/gdb/NEWS
>> index 2638b3e0d9c..f2e85776a53 100644
>> --- a/gdb/NEWS
>> +++ b/gdb/NEWS
>> @@ -13,6 +13,11 @@
>> the background, resulting in faster startup. This can be controlled
>> using "maint set dwarf synchronous".
>>
>> +* In a record session, when a forward emulation reaches the end of the
>> reverse
>> + history, the warning message has been changed to indicate that the end of
>> the
>> + history has been reached. It also specifies that the forward execution can
>> + continue, and the recording will also continue.
>> +
>> * Changed commands
>>
>> disassemble
>> diff --git a/gdb/infrun.c b/gdb/infrun.c
>> index bbb98f6dcdb..e129eb0582f 100644
>> --- a/gdb/infrun.c
>> +++ b/gdb/infrun.c
>> @@ -9244,8 +9244,14 @@ print_no_history_reason (struct ui_out *uiout)
>> {
>> if (uiout->is_mi_like_p ())
>> uiout->field_string ("reason", async_reason_lookup
>> (EXEC_ASYNC_NO_HISTORY));
>> + else if (execution_direction == EXEC_FORWARD)
>> + uiout->text ("\nNo more reverse-execution history for emulation. "
>> + "Going forward will continue executing and recording.\n");
>> else
>> - uiout->text ("\nNo more reverse-execution history.\n");
>> + {
>> + gdb_assert (execution_direction == EXEC_REVERSE);
>> + uiout->text ("\nNo more reverse-execution history.\n");
>> + }
>> }
>>
>> /* Print current location without a level number, if we have changed
>> diff --git a/gdb/testsuite/gdb.btrace/non-stop.exp
>> b/gdb/testsuite/gdb.btrace/non-stop.exp
>> index 62c940e4cd6..7ce3008c120 100644
>> --- a/gdb/testsuite/gdb.btrace/non-stop.exp
>> +++ b/gdb/testsuite/gdb.btrace/non-stop.exp
>> @@ -79,7 +79,7 @@ proc gdb_cont_to_bp_line { line threads nthreads } {
>> $nthreads
>> }
>>
>> -proc gdb_cont_to_no_history { threads cmd nthreads } {
>> +proc gdb_cont_to_no_history_backward { threads cmd nthreads } {
>> gdb_cont_to $threads $cmd \
>> [multi_line \
>> "No more reverse-execution history\." \
>> @@ -89,6 +89,16 @@ proc gdb_cont_to_no_history { threads cmd nthreads }
>> {
>> $nthreads
>> }
>>
>> +proc gdb_cont_to_no_history_forward { threads cmd nthreads } {
>> + gdb_cont_to $threads $cmd \
>> + [multi_line \
>> + "No more reverse-execution history for emulation\. Going forward
>> will continue executing and recording\." \
>> + "\[^\\\r\\\n\]*" \
>> + "\[^\\\r\\\n\]*" \
>> + ] \
>> + $nthreads
>> +}
>> +
>> # trace the code between the two breakpoints
>> with_test_prefix "prepare" {
>> gdb_cont_to_bp_line "$srcfile:$bp_1" all 2
>> @@ -176,14 +186,14 @@ with_test_prefix "reverse-step" {
>> with_test_prefix "continue" {
>> with_test_prefix "thread 1" {
>> with_test_prefix "continue" {
>> - gdb_cont_to_no_history 1 "continue" 1
>> + gdb_cont_to_no_history_forward 1 "continue" 1
>> gdb_test "thread apply 1 info record" \
>> ".*Recorded \[0-9\]+ instructions \[^\\\r\\\n\]*"
>> gdb_test "thread apply 2 info record" \
>> ".*Replay in progress\. At instruction 5\."
>> }
>> with_test_prefix "reverse-continue" {
>> - gdb_cont_to_no_history 1 "reverse-continue" 1
>> + gdb_cont_to_no_history_backward 1 "reverse-continue" 1
>> gdb_test "thread apply 1 info record" \
>> ".*Replay in progress\. At instruction 1\."
>> gdb_test "thread apply 2 info record" \
>> @@ -193,14 +203,14 @@ with_test_prefix "continue" {
>>
>> with_test_prefix "thread 2" {
>> with_test_prefix "continue" {
>> - gdb_cont_to_no_history 2 "continue" 1
>> + gdb_cont_to_no_history_forward 2 "continue" 1
>> gdb_test "thread apply 1 info record" \
>> ".*Replay in progress\. At instruction 1\."
>> gdb_test "thread apply 2 info record" \
>> ".*Recorded \[0-9\]+ instructions \[^\\\r\\\n\]*"
>> }
>> with_test_prefix "reverse-continue" {
>> - gdb_cont_to_no_history 2 "reverse-continue" 1
>> + gdb_cont_to_no_history_backward 2 "reverse-continue" 1
>> gdb_test "thread apply 1 info record" \
>> ".*Replay in progress\. At instruction 1\."
>> gdb_test "thread apply 2 info record" \
>> @@ -215,8 +225,8 @@ with_test_prefix "no progress" {
>> gdb_test "thread apply 1 record goto end" ".*"
>> gdb_test "thread apply 2 record goto begin" ".*"
>>
>> - gdb_cont_to_no_history 1 "continue" 1
>> - gdb_cont_to_no_history 1 "step" 1
>> + gdb_cont_to_no_history_forward 1 "continue" 1
>> + gdb_cont_to_no_history_forward 1 "step" 1
>> gdb_test "thread apply 1 info record" \
>> ".*Recorded \[0-9\]+ instructions \[^\\\r\\\n\]*"
>> gdb_test "thread apply 2 info record" \
>> @@ -227,8 +237,8 @@ with_test_prefix "no progress" {
>> gdb_test "thread apply 1 record goto begin" ".*"
>> gdb_test "thread apply 2 record goto end" ".*"
>>
>> - gdb_cont_to_no_history 2 "continue" 1
>> - gdb_cont_to_no_history 2 "step" 1
>> + gdb_cont_to_no_history_forward 2 "continue" 1
>> + gdb_cont_to_no_history_forward 2 "step" 1
>> gdb_test "thread apply 1 info record" \
>> ".*Replay in progress\. At instruction 1\."
>> gdb_test "thread apply 2 info record" \
>> @@ -238,7 +248,7 @@ with_test_prefix "no progress" {
>> with_test_prefix "all" {
>> gdb_test "thread apply all record goto begin" ".*"
>>
>> - gdb_cont_to_no_history all "continue" 2
>> + gdb_cont_to_no_history_forward all "continue" 2
>> gdb_test "thread apply 1 info record" \
>> ".*Recorded \[0-9\]+ instructions \[^\\\r\\\n\]*"
>> gdb_test "thread apply 2 info record" \
>> diff --git a/gdb/testsuite/gdb.reverse/break-precsave.exp
>> b/gdb/testsuite/gdb.reverse/break-precsave.exp
>> index b9d94681247..6d3b95933d4 100644
>> --- a/gdb/testsuite/gdb.reverse/break-precsave.exp
>> +++ b/gdb/testsuite/gdb.reverse/break-precsave.exp
>> @@ -73,7 +73,7 @@ proc precsave_tests {} {
>> -re ".*Breakpoint $decimal,.*$srcfile:$end_location.*$gdb_prompt $"
>> {
>> pass "go to end of main forward"
>> }
>> - -re "No more reverse-execution history.* end of main .*$gdb_prompt
>> $" {
>> + -re "No more reverse-execution history for emulation. Going forward
>> will continue executing and recording.* end of main .*$gdb_prompt $" {
>> pass "go to end of main forward"
>> }
>> }
>> @@ -103,7 +103,7 @@ proc precsave_tests {} {
>> -re ".*Breakpoint $decimal,.*$srcfile:$end_location.*$gdb_prompt $"
>> {
>> pass "end of record log"
>> }
>> - -re "No more reverse-execution history.* end of main .*$gdb_prompt
>> $" {
>> + -re "No more reverse-execution history for emulation. Going forward
>> will continue executing and recording.* end of main .*$gdb_prompt $" {
>> pass "end of record log"
>> }
>> }
>> diff --git a/gdb/testsuite/gdb.reverse/break-reverse.exp
>> b/gdb/testsuite/gdb.reverse/break-reverse.exp
>> index 1dd327ca654..bf74d6f7d07 100644
>> --- a/gdb/testsuite/gdb.reverse/break-reverse.exp
>> +++ b/gdb/testsuite/gdb.reverse/break-reverse.exp
>> @@ -80,7 +80,7 @@ gdb_test_multiple "continue" "end of record log" {
>> -re ".*Breakpoint $decimal,.*$srcfile:$end_location.*$gdb_prompt $" {
>> pass "end of record log"
>> }
>> - -re "No more reverse-execution history.* end of main .*$gdb_prompt $" {
>> + -re "No more reverse-execution history for emulation. Going forward will
>> continue executing and recording.* end of main .*$gdb_prompt $" {
>> pass "end of record log"
>> }
>> }
>> diff --git a/gdb/testsuite/gdb.reverse/machinestate-precsave.exp
>> b/gdb/testsuite/gdb.reverse/machinestate-precsave.exp
>> index 19c5934bfdf..693c304bc18 100644
>> --- a/gdb/testsuite/gdb.reverse/machinestate-precsave.exp
>> +++ b/gdb/testsuite/gdb.reverse/machinestate-precsave.exp
>> @@ -85,7 +85,7 @@ gdb_test_multiple "continue" "go to end of main
>> forward" {
>> -re ".*Breakpoint $decimal,.*$srcfile:$endmain.*$gdb_prompt $" {
>> pass "go to end of main forward"
>> }
>> - -re "No more reverse-execution history.* end main .*$gdb_prompt $" {
>> + -re "No more reverse-execution history for emulation. Going forward will
>> continue executing and recording.* end main .*$gdb_prompt $" {
>> pass "go to end of main forward"
>> }
>> }
>> --
>> 2.42.0
> Intel Deutschland GmbH
> Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
> Tel: +49 89 99 8853-0, www.intel.de<http://www.intel.de> <http://www.intel.de>
> Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
> Chairperson of the Supervisory Board: Nicole Lau
> Registered Office: Munich
> Commercial Register: Amtsgericht Muenchen HRB 186928
>
--
Cheers,
Guinevere Larsen
She/Her/Hers
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://sourceware.org/pipermail/gdb-patches/attachments/20240404/c960ac16/attachment-0001.htm>
More information about the Gdb-patches
mailing list