This is the mail archive of the
mailing list for the GCC project.
Re: [tree-ssa] bootstrap problem on powerpc-apple-darwin6.6
- From: law at redhat dot com
- To: Andrew Pinski <pinskia at physics dot uc dot edu>
- Cc: jason at redhat dot com (Jason Merrill), dberlin at dberlin dot org (Daniel Berlin), gcc at gcc dot gnu dot org
- Date: Thu, 21 Aug 2003 21:45:23 -0600
- Subject: Re: [tree-ssa] bootstrap problem on powerpc-apple-darwin6.6
- Reply-to: law at redhat dot com
In message <200308220001.h7M013m28429@tin.geop.uc.edu>, Andrew Pinski writes:
>> On Thu, 21 Aug 2003 16:53:04 -0400, Daniel Berlin <email@example.com> wro
>> > This stripping you do undoes what gimple_boolify_expr did to convert it t
>> > a boolean_type.
>> > Then invert_truthop aborts when it tries to invert a non-boolean_type tre
>> > So either
>> > 1. The conversion isn't useless, but gimplify_boolify_expr is not
>> > generating the conversion in a way that makes it seem un-useless.
>> > 2. The conversion isn't useless, and the above is not working like it
>> > should.
>> I think the conversion isn't useless, because BOOLEAN_TYPE has a (much)
>> smaller range of values, which matters to invert_truthop, at least, and
>> should be useful to optimizers.
>Would this patch be acceptable?
>It gets my by the part where I was failing.
> * tree-ssa.c (tree_ssa_useless_type_conversion): Types are only equal
> if they have the same min and max values.
So, two types can have the same mode, but different min/max values? I guess
that makes a certain amount of sense.
Ugh. I learn more about trees every day... Some things are good, some things
This is probably going to cause some missed optimizations. I guess we'll
find another way to deal with them.