This is the mail archive of the
gdb-prs@sources.redhat.com
mailing list for the GDB project.
Re: java/1039: regression: cannot set breakpoint on jmisc.main(java.lang.String[])
- From: David Carlton <carlton at math dot stanford dot edu>
- To: nobody at sources dot redhat dot com
- Cc: gdb-prs at sources dot redhat dot com,
- Date: 28 Feb 2003 22:38:01 -0000
- Subject: Re: java/1039: regression: cannot set breakpoint on jmisc.main(java.lang.String[])
- Reply-to: David Carlton <carlton at math dot stanford dot edu>
The following reply was made to PR java/1039; it has been noted by GNATS.
From: David Carlton <carlton at math dot stanford dot edu>
To: GNATS Filer <gdb-gnats at sources dot redhat dot com>
Cc: Michael Elizabeth Chastain <mec at shout dot net>
Subject: Re: java/1039: regression: cannot set breakpoint on jmisc.main(java.lang.String[])
Date: 28 Feb 2003 14:31:42 -0800
(I hope I got that subject right for GNATS to pick up this reply.)
At Michael's prompting (since I've been messing with symtabs
recently), I've been looking into this one. I confess, I haven't been
running gdb.java/jmisc*.exp for a little while, because the version of
binutils that I have installed doesn't always get along well with the
version of gcj that I have installed.
But I did poke into it a bit with an older binutils. The thing is,
though, this test has never worked for me: checking through my logs
since August of last year, GDB has never been able to set a breakpoint
at jmisc.main(java.lang.String[]). So it's hard for me to diagnose it
too well.
I'm using gcj 2.95.3. Michael doesn't test this with that version of
gcj, because of the binutils problem. So that's probably the reason
for our difference in the test results. Unfortunately, for various
reasons (disk space, largely), I'm not too likely to install a newer
version of gcj on this machine any time soon, though I might soon have
a newer gcj on a different machine.
Given the time period where Michael's seeing breakage, one possibility
is that it's an unexpected interaction with the demangling partial
symbols patch. That's just a wild stab, though: there are certainly
other possibilities. But it does fit in well with something I just
noticed: I happened to notice yesterday that lookup_partial_symbol has
the bug that it doesn't give you any way of specifying a linkage name
to match on. (Sigh.) So maybe this test is tickling that bug; maybe
this lookup more or less accidentally worked right before partial
symbols were demangled, but now it fails. I'll prepare a patch and
e-mail it to Michael.
In the mean time, there is a workaround for this specific issue, in
that 'break jmisc.main' works. It's only when you specify the type of
the argument that you run into problems.