This is the mail archive of the gdb@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Ada testsuite failures


> (gdb) p empty
> $1 = {<text variable, no debug info>} 0x2aaaaad3d590 <gnat__sockets__empty>
> 
> Isn't that interesting - that's something completely different.  There
> are zero references to the string "empty" in the debug info for
> null_record, so it is definitely a debugging information problem.

Yes and no. The test isn't representative of "real code". It will
be more productive to enhance the testcase a bit, I was very naive
when I wrote this code - it must have been the reduced testcase
I have been using to track down the issue, the code we have in our
testsuite is a bit better that way.

> Looking at the test I'm not surprised.  Empty is unused; adding
> something that references it is probably sufficient.

Yes, I agree. Could you try the following patch?

2006-01-02  Joel Brobecker  <brobecker@adacore.com>

	* gdb.ada/null_record/bar.ads (Do_Nothing): Add extra parameter.
	* gdb.ada/null_record/bar.adb (Do_Nothing): Likewise.
	* gdb.ada/null_record/null_record.adb: Fake use of type Empty
	to prevent this type from being optimized out.

That should help,
-- 
Joel


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]