Re: [PATCH 1/2] Check trace data from trace file

[Pedro Alves <palves at redhat dot com>]

On 02/27/2013 01:53 AM, Yao Qi wrote:
> On 02/27/2013 02:44 AM, Pedro Alves wrote:> On 02/25/2013 03:08 AM, Yao Qi wrote:
>>>> Here is the same patch with spaces ignored.  It is easier to read.
>>>> Note that I modified a matched pattern "PC register is not
>>>> available" for tfile target, as the regcache is not available when
>>>> switch to tfind mode (no regcache has been read out of 'R' block yet).
>> Hmmm.  Not sure.  Shouldn't something have caught that exception?  IOW,
>> could that be a bug?
>>
>> I'm having difficulty figuring it out, because when I run the test
>> I don't see that output anywhere in the gdb.log:
>>
>>   (gdb) printf "x %d x\n", $tracepoint
>>   x 2 x
>>   (gdb) PASS: gdb.trace/report.exp: tfile: 9.1: find frame for TP 2
>>   tdump
>>   Data collected at tracepoint 2, trace frame 0:
>>   (gdb) PASS: gdb.trace/report.exp: tfile: 9.1: tdump, nothing collected
>>
>> I wonder what's different in our environments.
>>
> 
> Yeah, it shows up in one build server.  I tried the test on my laptop
> (Fedora 16) and the other build server, it is OK.  Then, I get rid of
> the pattern to match "PC register is not available".  I'll check it
> on that build server.

Thanks.

> The changelog entry is completed as you suggested (wo/ "top level").

I'd prefer keeping it actually.  It's there to disambiguate with
the case of the new line of text being a continuation of the previous
context, that happens to end up on a new line just because that's how the
text flowed.  Like:

 	(foo_function_foo): New blah foo bar yum foo foo foo.
        Call use_collected_data once on the live target.

What called "use_collected_data"?  foo_function_foo, or the top level?

Otherwise this version is OK.  Thanks.

-- 
Pedro Alves