[i386] : Rewrite LEA handling (was:Re: PATCH [10/n] X32: Support x32 LEA insns)

On Mon, Jul 25, 2011 at 3:58 AM, H.J. Lu <hjl.tools@gmail.com> wrote:

>> You are not fixing the core of the problem... this is why you need so
>> much hacks and kludges at various places (some w.r.t. -fPIC already
>> existed, see the patch). Above, you correctly identified the problem,
>> so let's avoid gen_lowpart on SImode operands by not calling it
>> anymore.
>>
>> Attached patch effectively rewrites LEA handling. The trick is, that
>> instead of using Pmode operations in addresses, we use either SImode
>> or DImode operations to calculate the address on 64bit targets. Up to
>> now, address calculations strictly used Pmode, so SImode on 32bit
>> targets and DImode on 64bit targets. Recent patches to
>> ix86_decompose_address and ix86_legitimate_address_p relaxed this
>> requirement.
>>
>> Attached patch changes LEA patterns and LEA splitters to accept
>> addresses, calculated with either SImode or DImode operations.This
>> means, that on x64 targets, we don't use gen_lowpart on SImode
>> operands anymore. Since symbol references on x32 are in SImode, this
>> solves the problem. The patch also avoids generating SImode subregs of
>> DImode addresses and DImode zero_extends of SImode addresses, since
>> LEA insn does this for us automatically.
>>
>> Please also note the change to ix86_print_operand_address. To avoid
>> addr32 prefixes, we can force registers in DImode on 64bit targets
>> without any problems. On x32, we can investigate, if this change
>> avoids unnecessary LEAs (for PR 49781, patched gcc genrates 6 vs. 8).
>
> The testcase won't compile since PIC doesn't work:

Well, I did say that -fPIC did not work.

>> Patch was bootstrapped and regression tested on x86_64-pc-linux-gnu
>> {,-m32} with no regressions. H.J., can you please test it on x32?
>
> On x32, it failed:
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49832
>
>> BTW: -fPIC is not yet implemented on trunk and still fails there with
>> an (unrelated) error, I didn't check x32 branch.
>>
>
> This could be:
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49833

Attached patch implements -fpic handling for x32. In x32 mode, we now
use x86_64_general_operand and corresponding "e" constraints for adds
in SImode, since it looks that invalid addresses can only be generated
through adds. This avoids a whole bunch of new predicates and
constraints.

2011-07-25  Uros Bizjak  <ubizjak@gmail.com>

	PR target/47381
	PR target/49832
	PR target/49833
	* config/i386/i386.md (add_operand): New mode attribute.
	(*movdi_internal_rex64): Remove mode from pic_32bit_operand check.
	(*movsi_internal): Ditto.  Use "e" constraint in alternative 2.
	(*lea_1): Use SWI48 mode iterator.
	(*lea_1_zext): New insn pattern.
	(add<mode>3): Use <add_operand> predicate for operand 2.
	(*add<mode>_1): Use <add_operand> predicate for operand 2.  Use "le"
	constraint for alternative 2.
	(addsi_1_zext): Use addsi_operand predicate for operand 2.  Use "le"
	constraint for alternative 2.
	(add->lea splitter): Check operand modes in insn constraint.  Extend
	operands less than SImode wide to SImode.
	(add->lea zext splitter): Do not extend operands to DImode.
	(*lea_general_1): Handle only QImode and HImode operands.
	(*lea_general_2): Ditto.
	(*lea_general_3): Ditto.
	(*lea_general_1_zext): Remove.
	(*lea_general_2_zext): Ditto.
	(*lea_general_3_zext): Ditto.
	(*lea_general_4): Check operand modes in insn constraint.  Extend
	operands less than SImode wide to SImode.
	(ashift->lea splitter): Ditto.
	* config/i386/i386.c (ix86_print_operand_address): Print address
	registers with 'q' modifier on 64bit targets.
	* config/i386/predicates.md (pic_32bit_opreand): Define as special
	predicate.  Reject non-SI and non-DI modes.
	(addsi_operand): New predicate.

Uros.

[i386] : Rewrite LEA handling (was:Re: PATCH [10/n] X32: Support x32 LEA insns)

Commit Message

Comments

Patch