Re: [PATCH 3/4, MIPS] Move clear_upper32* patterns to the and patterns

[Richard Sandiford <rdsandiford at googlemail dot com>]

Adam Nemet <anemet@caviumnetworks.com> writes:
> Richard Henderson writes:
>> How about merging all the patterns?  E.g.
>
> I tried working out the details, this is where it stands.  It passes
> ext-[4567].c tests from the MIPS testsuite but I haven't done any further
> testing.

Hmm.  If we do this, we should do it for all combinations (i.e. have
patterns for TARGET_MIPS16 || !ISA_HAS_EXT_DEXT too).  And MIPS16 has
the added complication that ZEB and ZEH are only for GENERATE_MIPS16E;
shifts are needed otherwise.  All in all, I don't take back what I said
about it not being easy to do this cleanly. ;)

I was half-way through a patch that added support for the other cases,
but it just doesn't feel right.  It seems like we're going to great
lengths to work around a deficiency in the IR (that we can't use
zero_extend for some zero extensions.)  Maybe that's just something
we have to live with, but let me think about this a bit first?

(I know that's famous last words, what with the still-unresolved
$gp/long-branch issue ;(.)

Richard