[RFC] avoiding too narrow register classes in reload

[Alexandre Oliva]

On Dec 14, 2004, Richard Henderson <rth@redhat.com> wrote:

> On Tue, Dec 14, 2004 at 07:20:10PM -0200, Alexandre Oliva wrote:
>> I realize the patch introduces some possibly-significant semantic
>> changes, in that NO_REGS is now regarded as a SMALL_REGISTER_CLASS_P

> I think that is indeed a surprising result.

>> Does it make sense to narrow the
>> predicate to hold only for non-empty classes, and adjust the
>> one-before-last hunk, that uses the negated predicate in a way that
>> requires it to hold for empty classes?

> Please.

Here's the revised patch.  Bootstrapped on x86_64-linux-gnu,
regtesting will still take a while to complete.  Ok to install if no
regressions show up?

-------------- next part --------------
A non-text attachment was scrubbed...
Name: reload-small-regclass-likely-spilled-take2.patch
Type: text/x-patch
Size: 4232 bytes
Desc: not available
URL: <https://gcc.gnu.org/pipermail/gcc/attachments/20041215/5949e007/attachment.bin>
-------------- next part --------------

-- 
Alexandre Oliva             http://www.ic.unicamp.br/~oliva/
Red Hat Compiler Engineer   aoliva@{redhat.com, gcc.gnu.org}
Free Software Evangelist  oliva@{lsd.ic.unicamp.br, gnu.org}