Regression on gdb.trace/{actions,change-loc,pending}.exp on native-{,extended-}gdbserver (was: Re: [PATCH 2/2] "catch catch/throw/rethrow", breakpoint -> catchpoint)

[Simon Marchi]

On 2019-08-02 7:08 p.m., Sergio Durigan Junior wrote:
> On Tuesday, July 09 2019, Pedro Alves wrote:
> 
>> On 7/9/19 5:22 PM, Andrew Burgess wrote:
>>> This looks good to me.
>>
>> On 7/9/19 5:07 PM, Tom Tromey wrote:
>>> This looks great to me.
>>
>> Thanks.  I pushed it in with a couple additional & obvious tweaks:
>>
>> - A tweak to gdb.mi/mi-catch-cpp-exceptions.exp, to explicitly
>>   check for type="catchpoint".
>>
>> - A tweak to the manual to adjust the MI -catch-* examples.
> 
> Hi Pedro,
> 
> BuildBot has caught these regressions when testing this patch on
> native-{,extended-}gdbserver:
> 
>   https://sourceware.org/ml/gdb-testers/2019-q3/msg00438.html
> 
>   PASS -> FAIL: gdb.trace/actions.exp: ctf: tracepoint on gdb_asm_test
>   PASS -> FAIL: gdb.trace/actions.exp: ctf: tracepoint on gdb_c_test
>   new FAIL: gdb.trace/actions.exp: ctf: tracepoint on gdb_recursion_test 0
>   PASS -> FAIL: gdb.trace/actions.exp: tfile: tracepoint on gdb_asm_test
>   PASS -> FAIL: gdb.trace/actions.exp: tfile: tracepoint on gdb_c_test
>   new FAIL: gdb.trace/actions.exp: tfile: tracepoint on gdb_recursion_test 0
>   PASS -> FAIL: gdb.trace/change-loc.exp: 1 trace: continue to marker 2
>   PASS -> FAIL: gdb.trace/change-loc.exp: 1 trace: continue to marker 3
>   PASS -> FAIL: gdb.trace/change-loc.exp: 1 trace: tfind frame 0
>   PASS -> FAIL: gdb.trace/change-loc.exp: 1 trace: tracepoint with two locations - installed
>   PASS -> FAIL: gdb.trace/change-loc.exp: 1 trace: tracepoint with two locations - pending
>   PASS -> FAIL: gdb.trace/change-loc.exp: 2 trace: continue to marker 2
>   PASS -> FAIL: gdb.trace/change-loc.exp: 2 trace: continue to marker 3
>   PASS -> FAIL: gdb.trace/change-loc.exp: 2 trace: tfind frame 2
>   PASS -> FAIL: gdb.trace/change-loc.exp: 2 trace: tracepoint with two locations - installed
>   PASS -> FAIL: gdb.trace/change-loc.exp: 2 trace: tracepoint with two locations - pending
>   PASS -> FAIL: gdb.trace/pending.exp: ftrace installed_in_trace: continue to marker 2
>   new FAIL: gdb.trace/pending.exp: ftrace installed_in_trace: tfind test frame 0
>   PASS -> FAIL: gdb.trace/pending.exp: trace installed_in_trace: continue to marker 2
>   PASS -> FAIL: gdb.trace/pending.exp: trace installed_in_trace: tfind test frame 0

I wanted to run the gdb.trace tests to make sure that this patch here
doesn't cause any regression related to unavailable registers:

  https://sourceware.org/pipermail/gdb-patches/2021-May/179222.html

I see many failures, including these ones.  I'll give it a quick look,
see if it's something obvious.

Simon