[PATCH] gdb, testsuite, btrace: relax unneeded stepi expected output

[PATCH] gdb, testsuite, btrace: relax unneeded stepi expected output

[Markus Metzger]

In gdb.btrace/reconnect.exp, we test that we can disconnect and reconnect
again to a GDB session that is recording with the btrace recording format.
It does not really matter what we are recording.

The test assumed that stepping from _start will bring us into an area
without debug information.  This is not correct on all systems.

Relax the expected output to also support systems where we do have debug
information for that code.

gdb/testsuite/ChangeLog:
2021-03-08  Markus Metzger  <markus.t.metzger@intel.com>

	* gdb.btrace/reconnect.exp: Relax expected stepi output.
---
 gdb/testsuite/gdb.btrace/reconnect.exp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/gdb/testsuite/gdb.btrace/reconnect.exp b/gdb/testsuite/gdb.btrace/reconnect.exp
index 2e61d3591f3..26578706eaf 100644
--- a/gdb/testsuite/gdb.btrace/reconnect.exp
+++ b/gdb/testsuite/gdb.btrace/reconnect.exp
@@ -51,7 +51,7 @@ gdb_target_cmd $gdbserver_protocol $gdbserver_gdbport
 # Create a record, check, reconnect
 with_test_prefix "first" {
   gdb_test_no_output "record btrace" "record btrace enable"
-  gdb_test "stepi 19" "($hex in .* from .*|$hex\t$decimal.*)"
+  gdb_test "stepi 19" ".*"
 
   gdb_test "info record" [multi_line \
     "Active record target: .*" \
-- 
2.29.2

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

Re: [PATCH] gdb, testsuite, btrace: relax unneeded stepi expected output

[Tom Tromey]

>>>>> "Markus" == Markus Metzger via Gdb-patches <gdb-patches@sourceware.org> writes:

Markus> gdb/testsuite/ChangeLog:
Markus> 2021-03-08  Markus Metzger  <markus.t.metzger@intel.com>

Markus> 	* gdb.btrace/reconnect.exp: Relax expected stepi output.

This seems fine to me.  Thank you.

Tom

Re: [PATCH] gdb, testsuite, btrace: relax unneeded stepi expected output

[Luis Machado]

On 3/16/21 6:26 AM, Markus Metzger via Gdb-patches wrote:
> In gdb.btrace/reconnect.exp, we test that we can disconnect and reconnect
> again to a GDB session that is recording with the btrace recording format.
> It does not really matter what we are recording.
> 
> The test assumed that stepping from _start will bring us into an area
> without debug information.  This is not correct on all systems.
> 
> Relax the expected output to also support systems where we do have debug
> information for that code.
> 
> gdb/testsuite/ChangeLog:
> 2021-03-08  Markus Metzger  <markus.t.metzger@intel.com>
> 
> 	* gdb.btrace/reconnect.exp: Relax expected stepi output.
> ---
>   gdb/testsuite/gdb.btrace/reconnect.exp | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/gdb/testsuite/gdb.btrace/reconnect.exp b/gdb/testsuite/gdb.btrace/reconnect.exp
> index 2e61d3591f3..26578706eaf 100644
> --- a/gdb/testsuite/gdb.btrace/reconnect.exp
> +++ b/gdb/testsuite/gdb.btrace/reconnect.exp
> @@ -51,7 +51,7 @@ gdb_target_cmd $gdbserver_protocol $gdbserver_gdbport
>   # Create a record, check, reconnect
>   with_test_prefix "first" {
>     gdb_test_no_output "record btrace" "record btrace enable"
> -  gdb_test "stepi 19" "($hex in .* from .*|$hex\t$decimal.*)"
> +  gdb_test "stepi 19" ".*"
>   
>     gdb_test "info record" [multi_line \
>       "Active record target: .*" \
> 

Investigating some racy tests under make check-read1, isn't there a 
better matching pattern that we can use here to prevent future 
non-determinism?

RE: [PATCH] gdb, testsuite, btrace: relax unneeded stepi expected output

[Metzger, Markus T]

Hello Luis,

>> @@ -51,7 +51,7 @@ gdb_target_cmd $gdbserver_protocol $gdbserver_gdbport
>>   # Create a record, check, reconnect
>>   with_test_prefix "first" {
>>     gdb_test_no_output "record btrace" "record btrace enable"
>> -  gdb_test "stepi 19" "($hex in .* from .*|$hex\t$decimal.*)"
>> +  gdb_test "stepi 19" ".*"
>>
>>     gdb_test "info record" [multi_line \
>>       "Active record target: .*" \
>>
>
>Investigating some racy tests under make check-read1, isn't there a
>better matching pattern that we can use here to prevent future
>non-determinism?

The test now ignores all output since it doesn't really matter where we end up
after stepping.  This should be the most future-proof pattern.  Were you looking
for a less lax pattern?

I ran gdb.btrace with check-read1 on IA and didn't find any issues.

Regards,
Markus.
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

Re: [PATCH] gdb, testsuite, btrace: relax unneeded stepi expected output

[Luis Machado]

Hi,

On 4/12/21 5:22 AM, Metzger, Markus T wrote:
> Hello Luis,
> 
>>> @@ -51,7 +51,7 @@ gdb_target_cmd $gdbserver_protocol $gdbserver_gdbport
>>>    # Create a record, check, reconnect
>>>    with_test_prefix "first" {
>>>      gdb_test_no_output "record btrace" "record btrace enable"
>>> -  gdb_test "stepi 19" "($hex in .* from .*|$hex\t$decimal.*)"
>>> +  gdb_test "stepi 19" ".*"
>>>
>>>      gdb_test "info record" [multi_line \
>>>        "Active record target: .*" \
>>>
>>
>> Investigating some racy tests under make check-read1, isn't there a
>> better matching pattern that we can use here to prevent future
>> non-determinism?
> 
> The test now ignores all output since it doesn't really matter where we end up
> after stepping.  This should be the most future-proof pattern.  Were you looking
> for a less lax pattern?

Yeah, I was looking for a more meaningful kind of test other than 
expecting anything.

In any case, I was worried about possible races in the test...

> 
> I ran gdb.btrace with check-read1 on IA and didn't find any issues.

... and since you verified it with read1, I guess it should be fine.

Thanks,
Luis

RE: [PATCH] gdb, testsuite, btrace: relax unneeded stepi expected output

[Metzger, Markus T]

Hello Luis,

>>>> @@ -51,7 +51,7 @@ gdb_target_cmd $gdbserver_protocol
>$gdbserver_gdbport
>>>>    # Create a record, check, reconnect
>>>>    with_test_prefix "first" {
>>>>      gdb_test_no_output "record btrace" "record btrace enable"
>>>> -  gdb_test "stepi 19" "($hex in .* from .*|$hex\t$decimal.*)"
>>>> +  gdb_test "stepi 19" ".*"
>>>>
>>>>      gdb_test "info record" [multi_line \
>>>>        "Active record target: .*" \
>>>>
>>>
>>> Investigating some racy tests under make check-read1, isn't there a
>>> better matching pattern that we can use here to prevent future
>>> non-determinism?
>>
>> The test now ignores all output since it doesn't really matter where we end up
>> after stepping.  This should be the most future-proof pattern.  Were you
>looking
>> for a less lax pattern?
>
>Yeah, I was looking for a more meaningful kind of test other than
>expecting anything.

The meaningful part is just outside of this hunk:

  gdb_test "info record" [multi_line \
    "Active record target: .*" \
    "Recorded 19 instructions in .+ functions \\(. gaps\\) for thread 1 \\(Thread .*\\)."
  ]

We check that we actually stepped exactly those 19 instructions.  The test doesn't
care where this brought us.

Regards,
Markus.

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

Re: [PATCH] gdb, testsuite, btrace: relax unneeded stepi expected output

[Luis Machado]

On 4/12/21 8:09 AM, Metzger, Markus T wrote:
> Hello Luis,
> 
>>>>> @@ -51,7 +51,7 @@ gdb_target_cmd $gdbserver_protocol
>> $gdbserver_gdbport
>>>>>     # Create a record, check, reconnect
>>>>>     with_test_prefix "first" {
>>>>>       gdb_test_no_output "record btrace" "record btrace enable"
>>>>> -  gdb_test "stepi 19" "($hex in .* from .*|$hex\t$decimal.*)"
>>>>> +  gdb_test "stepi 19" ".*"
>>>>>
>>>>>       gdb_test "info record" [multi_line \
>>>>>         "Active record target: .*" \
>>>>>
>>>>
>>>> Investigating some racy tests under make check-read1, isn't there a
>>>> better matching pattern that we can use here to prevent future
>>>> non-determinism?
>>>
>>> The test now ignores all output since it doesn't really matter where we end up
>>> after stepping.  This should be the most future-proof pattern.  Were you
>> looking
>>> for a less lax pattern?
>>
>> Yeah, I was looking for a more meaningful kind of test other than
>> expecting anything.
> 
> The meaningful part is just outside of this hunk:
> 
>    gdb_test "info record" [multi_line \
>      "Active record target: .*" \
>      "Recorded 19 instructions in .+ functions \\(. gaps\\) for thread 1 \\(Thread .*\\)."
>    ]
> 
> We check that we actually stepped exactly those 19 instructions.  The test doesn't
> care where this brought us.

Indeed. Thanks for making it clear.