This is the mail archive of the mailing list for the glibc 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: Add tests that backtrace and backtrace_symbols produce correctresults

On Sat, 19 Jan 2013, Andreas Schwab wrote:

> "Joseph S. Myers" <> writes:
> > This code should not be called for the do_test frame; if it is, you have 
> > some other problem with the frames before do_test.  The code deliberately 
> > does not check for a function name for the do_test frame.
> But do_test isn't the outermost frame.

That's irrelevant.  The tests work from the innermost frame (the caller of 
backtrace) outwards, knowing what each frame in turn should be and 
stopping just before they get to the frame that should be do_test.  If 
they try to look at the name of the do_test frame, that indicates a 
problem further within the backtrace.

As far as I can tell, the only issue you have mentioned that does not look 
like an incorrect backtrace on some platform or a purely hypothetical 
issue arising from a misunderstanding of the tests is the issue with the 
build directory name.  Certainly you haven't given any actual backtraces 
showing failures on any platform, but if you did then it would be easy to 
see (by comparison with x86 / x86_64, for example) what's wrong with those 
backtraces from the point of view of the tests.

Joseph S. Myers

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