[RFC] avoiding too narrow register classes in reload

[Alexandre Oliva]

On Dec 14, 2004, bernds_cb1@t-online.de (Bernd Schmidt) wrote:

>> This is what causes the problem I'm running into: if this alternative
>> is selected (and it will often be), we end up forced to use one of the
>> first two argument registers for the reload, and this will fail if the
>> reload is needed in the middle of a function call sequence.

> What does the insn that it's trying to reload look like?

  (set (mem addr) (pseudo))

addr is a legitimate address, that doesn't require any reloads.

-- 
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}