This is the mail archive of the gcc-patches@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]

Re: [PATCH] Fix PR debug/14114 [DWARF2 guru needed]


Jason Merrill wrote:

On Thu, 04 Mar 2004 16:23:00 -0800, Zack Weinberg <zack@codesourcery.com> wrote:


Eric Botcazou <ebotcazou@libertysurf.fr> writes:


If the C front end didn't contain a broken data structure, it wouldn't
be passing the same tree twice, yes.  But the brokenness of that data
structure is *unrelated* to the basic design question of whose
responsibility it is to keep track of whether debug information has
already been generated for a given tree.

I think it's the front-end responsability, you think it's the back-end responsability. Let's have a third opinion. OK?

Yes, by all means. Would one of the DWARF2 gurus like to comment?


I think it's appropriate to expect that a single decl only be passed to
dwarf2out_decl once in a compilation.  That's certainly been the
convention, modulo this bug.

I'm not going to get tangled up in this particular design argument, if I can help it.


But, I am going to approve the patch *for 3.4 only*. There, we all agree that (a) the C front end is slightly busted, and (b) the patch looks safe to me.

We don't actually have to settle the original argument, probably, because Zack's c-decl.c fixes will go into 3.5, and then this case will never happen.

--
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com


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