This is the mail archive of the 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] rtx_cost of SUBREG

Hi Eric,

On Sat, 22 Jan 2005, Eric Botcazou wrote:
> Granted, the real problem is elsewhere (bad interaction between
> apply_distributive_law and simplify_associative_operation, I can give
> the gory details if you want) but I think tweaking the generic cost
> function at the very end of stage3 is dangerous.

Could you outline the gory details?  The clock is ticking on the 48 hour
rule, but in stage3 its clearly better to address the "real problem"
that was exposed by this change rather than leave it latent to potentially
cause problems elsewhere before a release.

There's also the compromise suggested in my review of Mark's patch that
allows the backends to override the code of REG and SUBREG themselves.
This would allow the sparc (in 64-bit mode) to specify a cost for a
SUBREG and avoid the problem.  But you'll appreciate this is also just
a (target-specific) workaround as the cost of the SUBREGs in the RTL
you give really should be the same as the cost of a REG.

> Please revert the patch.

I'm building a tree to sparc64-sun-solaris2.8 to investigate the issue.
Whatever the outcome, this should be resolved by Monday morning.
Sorry for the inconvenience,


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