[0/12] RFC: support for 'dwz -m'

[0/12] RFC: support for 'dwz -m'

[Tom Tromey]

I'm posting this series as an RFC.  I intend to check it in, just not
yet -- it is too large, IMO, to go in just before 7.5, and also we're
hoping to get some official DWARF constants assigned beforehand (see
patch #1).  Finally, this series needs a couple of links to
documentation, which I think isn't ready yet.

Jakub recently posted about his new DWARF compression tool, 'dwz':

    http://gcc.gnu.org/ml/gcc/2012-04/msg00686.html

The 'multifile' branch of the dwz repository implements support for a
multi-file mode.  The idea behind this mode is that commonalities
between a set of files are extracted into a new '.dwz' file, that is
then referenced by the rewritten original files.

We intend to use this when creating debuginfo RPMs.  It yields a nice
compression boost for some packages.

This patch series implements support for the needed DWARF extensions in
gdb.  Part of the series comes from my earlier minsym sharing series.
Some of these are really just plain cleanups that I probably should have
put in long ago.

The bulk of the real work is in patches #9 and #10.

I'd like to thank Doug for his Fission work, which made this much
simpler than it otherwise would have been.

I've regression tested the series using plain 'dwz -m', but not
multifile compression in conjunction with .gdb_index (the dwz change
here isn't ready yet).  I will do this testing before committing.

Tom

Re: [0/12] RFC: support for 'dwz -m'

[Tom Tromey]

>>>>> "Tom" == Tom Tromey <tromey@redhat.com> writes:

Tom> I'm posting this series as an RFC.

Oops, I forgot I had a final patch in this series to add a bit of info
to 'maint print stats' that I did not intend to submit.

All those notes saying NN/12 are wrong and should read NN/11.

thanks,
Tom

Re: [0/12] RFC: support for 'dwz -m'

[Tom Tromey]

>>>>> "Tom" == Tom Tromey <tromey@redhat.com> writes:

Tom> I'm posting this series as an RFC.  I intend to check it in, just not
Tom> yet -- it is too large, IMO, to go in just before 7.5, and also we're
Tom> hoping to get some official DWARF constants assigned beforehand (see
Tom> patch #1).  Finally, this series needs a couple of links to
Tom> documentation, which I think isn't ready yet.

I'm going to post a new version of this series.

I dropped patch #1, because Jakub put the equivalent patch into gcc and
src already.  It has a link to the DWARF proposal:

    http://www.dwarfstd.org/ShowIssue.php?issue=120604.1&type=open

Of course, because of this, the patch numbers all change, leading to
general confusion.  I plan to handle this by sending follow-ups to the
most recent version of the patches.

Also, I'm only planning to repost the patches that have changed -- just
the last 3.

Tom> I've regression tested the series using plain 'dwz -m', but not
Tom> multifile compression in conjunction with .gdb_index (the dwz change
Tom> here isn't ready yet).  I will do this testing before committing.

I did this and there are no regressions as compared to a test suite run
using just .gdb_index.  (.gdb_index has a regression against a plain
test run.)

Tom