This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [Patch, fortran] PR40011 - Problems with -fwhole-file
Richard Guenther wrote:
>>>> The following fails at any level of optimization [...]
>>>>
>>> As written, I think the behaviour is OK as the local variable is no
>>> longer available after returning from the procedure.
> I see no change in the generated code with -O3 -fwhole-file before
> or after the patch. I don't see how we could optimize the conditions
> as we have [...]
> where j is uninitialized (the stores to it have been optimized away
> before inlining). Whether j.this != &base we cannot tell.
>
Thanks for the check - I expected something like that. Presumably 4.4
did not optimize away the store to j and thus it worked.
* * *
Remains:
a) Some fine-comb work plus changelogs
b) Finding out what does wrong with derived->backend_decl being not
record_type, cf. PR 40011
Answer by Richard:
12:24 < richi> Tobias__: easy fix
12:25 < richi> s/cgraph_finalize_function (...,
false)/cgraph_finalize_function (..., true)/g
[cgraphunit.c: "If NESTED is true, then our caller cannot stand to have
the garbage collector run at the moment."]
12:25 < richi> of course we only get to the next ICE ...
c) Finding out why the for static one can currently not inline without
running into trouble
My suggestion is to concentrate on (a) - with some work on (b) - and get
the whole stuff committed. Then one can finish the remaining bits of (b)
and start then looking at (c).
Tobias