Bug#775744: libapp-stacktrace-perl: FTBFS: Failed 1/4 test programs. 1/6 subtests failed.

[Niko Tyni]

On Sun, Dec 04, 2016 at 03:10:03AM +0100, Axel Beckert wrote:
> Chris Lamb wrote:
> > libapp-stacktrace-perl fails to build from source in unstable/amd64 too:

> Just having read
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=814929#10, it sounds
> as if Niko's observation about longer addresses could also have
> triggered the FTBFS you ran into, too, although it's likely
> different code involved. (If so, I'd clone this bug to track the
> recent FTBFS on probably all architectures separately and downgrade
> the old one to important again.)
> 
> Niko: What do you think?

I expect it's the same bug as the mips64el one that was originally
reported here, but there may be other issues on 32-bit architectures.

I don't think it's worth fiddling with clones and severities, just fixing
the immediate one and seeing how it fares after that seems fine. But
YMMV of course.

The problem itself seems to be that the gdb macros in
 lib/App/Stacktrace/perl_backtrace_raw.txt
cast pointers to the C 'int' type, which is broken on 64-bit
architectures. For instance, in 'perl_backtrace_5_12_x':
    set $interpreter = (int) Perl_get_context()

I suppose it used to work on amd64 because perl used low memory addresses
so things didn't overflow, but that has changed with the latest perl
upload, presumably due to gcc PIE changes.

Hope this helps,
-- 
Niko Tyni   ntyni at debian.org