Re: [PATCH v2 2/2] gdb/testsuite: Add gdb.base/memops-watchpoint.exp

[Thiago Jung Bauermann <thiago.bauermann@linaro.org>]

Hello Luis,

Thanks for reviewing the patches!

Luis Machado <luis.machado@arm.com> writes:

> On 4/21/24 23:26, Thiago Jung Bauermann wrote:
>> +setup_kfail breakpoints/31665 arm*-*-linux*
>> +gdb_test "continue" \
>> +    [multi_line \
>> +	 "Continuing\\." \
>> +	 "" \
>> +	 "(Hardware w|W)atchpoint ${decimal}: -location a\\\[28\\\]" \
>> +	 "" \
>> +	 "Old value = 104 'h'" \
>> +	 "New value = 0 '\\\\000'" \
>> +	 ".*memset.* \\(\\) at .*:$decimal" \
>> +	 ".*"] \
>> +    "continue until memset watchpoint hits"
>> +
>> +setup_kfail breakpoints/31665 arm*-*-linux*
>> +gdb_test "continue" \
>> +    [multi_line \
>> +	 "Continuing\\." \
>> +	 "" \
>> +	 "(Hardware w|W)atchpoint ${decimal}: -location b\\\[28\\\]" \
>> +	 "" \
>> +	 "Old value = 101 'e'" \
>> +	 "New value = 114 'r'" \
>> +	 ".*memcpy.* \\(\\) at .*:$decimal" \
>> +	".*"] \
>> +    "continue until memcpy watchpoint hits"
>> +
>> +# Note: Some architectures use memcpy for memmove.
>> +setup_kfail breakpoints/31665 arm*-*-linux*
>> +gdb_test "continue" \
>> +    [multi_line \
>> +	 "Continuing\\." \
>> +	 "" \
>> +	 "(Hardware w|W)atchpoint ${decimal}: -location c\\\[28\\\]" \
>> +	 "" \
>> +	 "Old value = 100 'd'" \
>> +	 "New value = 114 'r'" \
>> +	 ".*(memmove|memcpy).* \\(\\) at .*:$decimal" \
>> +	".*"] \
>> +    "continue until memmove watchpoint hits"
>
> For this test, on aarch64 I see full passes, but on 32-bit Arm I see
> 3 passes and 3 kfail's.
>
> Is that the expected outcome?

Yes, the testcase uncovered a bug in in hardware watchpoints with 32-bit
Arm (the testcase fully passes when the testcase is modified to use
software watchpoints) so I opened PR breakpoints/31665 to track it.

I didn't have the chance to investigate the problem yet, but it's in my
TODO list.

--
Thiago