This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC 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]

Call for Testers: VTA debug info is usable


I've just installed a number of patches to
var-tracking-assignments-branch and
var-tracking-assignments-4_3-branch, which bring them to passing
compare-debug for all object files in both the bootstrapped compiler
and the target libraries on x86_64-linux-gnu, i686-linux-gnu and
ppc64-linux-gnu with -O1, -O2 and -O3 in BOOT_CFLAGS.

I'm very happy to report that I could debug all of the most recent
errors using a stage3 compiler built with -O3 -g
(-fvar-tracking-assignments defaults to enabled with -g in the
branches), along with GDB as in Fedora development, with support for
debugging inlined functions (thanks, Jan!)

Debug info still doesn't account for all possible locations of
variables, and it still doesn't support constants in location lists,
but it's possible to inspect most live variables otherwise.

Having completed the post-merge stabilization, I'll now go back to the
project of creating guality testcases, along the lines proposed at
http://gcc.gnu.org/ml/gcc/2008-07/msg00591.html

At some point I'll proceed to implement meaningful is_stmt dwarf
markers for .loc directives, which should enable the implementation of
the final stage of guality tests, avoiding any interference with the
generated code.
http://gcc.gnu.org/ml/gcc-patches/2008-08/msg00757.html

I also have some thoughts on introducing an option for GCC to verify
that a translation unit produces the same code when compiled with and
without debug info, to be used by make check and by vta testers.

Meanwhile, I'm ready to start tackling reports about:

- bootstrap-debug, bootstrap4-debug or any compare-debug failures on
your favorite native or cross platforms;

- compilation errors with vta enabled; and

- unmet debug info expectations, such as variables that must have
available values at certain points in the program, but that don't,
according to debug info.

Thanks in advance,

-- 
Alexandre Oliva         http://www.lsd.ic.unicamp.br/~oliva/
Free Software Evangelist  oliva@{lsd.ic.unicamp.br, gnu.org}
FSFLA Board Member       ÂSÃ Libre! => http://www.fsfla.org/
Red Hat Compiler Engineer   aoliva@{redhat.com, gcc.gnu.org}


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