llvm-project/llvm/lib/Target/ARM
Jim Grosbach 1c171b121a Explicitly disallow predication in Thumb1 assembly.
llvm-svn: 138562
2011-08-25 17:23:55 +00:00
..
AsmParser Explicitly disallow predication in Thumb1 assembly. 2011-08-25 17:23:55 +00:00
Disassembler Perform more thorough checking of t2IT mask parameters, which fixes all remaining crashers when disassembling the entire 16-bit instruction space. 2011-08-24 22:40:22 +00:00
InstPrinter When printing Thumb1 NOP ('mov r8, r8'), make sure to print the predicate. 2011-08-24 20:06:14 +00:00
MCTargetDesc ARM asm backend initialize isThumbMode based on target triple. 2011-08-24 22:27:35 +00:00
TargetInfo Move TargetRegistry and TargetSelect from Target to Support where they belong. 2011-08-24 18:08:43 +00:00
ARM.h Code clean up. 2011-07-25 20:18:48 +00:00
ARM.td Rename attribute 'thumb' to a more descriptive 'thumb-mode'. 2011-07-07 19:05:12 +00:00
ARMAsmPrinter.cpp Move TargetRegistry and TargetSelect from Target to Support where they belong. 2011-08-24 18:08:43 +00:00
ARMAsmPrinter.h Use TableGen'erated pseudo lowering for ARM. 2011-07-08 17:40:42 +00:00
ARMBaseInstrInfo.cpp Clean up Thumb load/store multiple definitions. 2011-08-23 17:41:15 +00:00
ARMBaseInstrInfo.h Implement isLoadFromStackSlotPostFE and isStoreToStackSlotPostFE for ARM. 2011-08-08 21:45:32 +00:00
ARMBaseRegisterInfo.cpp Thumb1 ADD/SUB SP instructions are predicable in Thumb2 mode. 2011-08-24 17:46:13 +00:00
ARMBaseRegisterInfo.h Thumb parsing diagnostics for low-reg requirements on ADD and MOV. 2011-08-16 22:20:01 +00:00
ARMBuildAttrs.h Rework some .ARM.attribute work for improved gcc compatibility. 2011-02-07 00:49:53 +00:00
ARMCallingConv.h In the calling convention logic, ValVT is always a legal type, 2010-11-04 10:49:57 +00:00
ARMCallingConv.td ARM byval support. Will be enabled by another patch to the FE. <rdar://problem/7662569> 2011-04-20 16:47:52 +00:00
ARMCodeEmitter.cpp ARM clean up the imm_sr operand class representation. 2011-08-17 21:51:27 +00:00
ARMConstantIslandPass.cpp Sink ARMMCExpr and ARMAddressingModes into MC layer. First step to separate ARM MC code from target. 2011-07-20 23:34:39 +00:00
ARMConstantPoolValue.cpp land David Blaikie's patch to de-constify Type, with a few tweaks. 2011-07-18 04:54:35 +00:00
ARMConstantPoolValue.h Nuke a FIXME. No need to be fancier here, as ARM handles constant pools 2010-11-29 23:09:20 +00:00
ARMELFWriterInfo.cpp Move the ARM reloc constants to Support/ELF.h 2010-11-23 19:40:36 +00:00
ARMELFWriterInfo.h Move the ARM reloc constants to Support/ELF.h 2010-11-23 19:40:36 +00:00
ARMExpandPseudoInsts.cpp Remove the VMOVQQ pseudo instruction. 2011-08-20 00:52:40 +00:00
ARMFastISel.cpp [SU]XT[BH] are only available on ARMv6 and up. 2011-08-23 20:53:08 +00:00
ARMFrameLowering.cpp ARM refactor indexed store instructions. 2011-08-05 20:35:44 +00:00
ARMFrameLowering.h In Thumb2 mode, lower frame indix references to: 2011-04-22 01:42:52 +00:00
ARMGlobalMerge.cpp Convert ConstantExpr::getGetElementPtr and 2011-07-21 14:31:17 +00:00
ARMHazardRecognizer.cpp - Rename TargetInstrDesc, TargetOperandInfo to MCInstrDesc and MCOperandInfo and 2011-06-28 19:10:37 +00:00
ARMHazardRecognizer.h Various bits of framework needed for precise machine-level selection 2010-12-24 05:03:26 +00:00
ARMISelDAGToDAG.cpp Thumb1 ADD/SUB SP instructions are predicable in Thumb2 mode. 2011-08-24 17:46:13 +00:00
ARMISelLowering.cpp ARM expansion of pre-indexed store pseudos should maintain memoperands. 2011-08-12 21:02:34 +00:00
ARMISelLowering.h land David Blaikie's patch to de-constify Type, with a few tweaks. 2011-07-18 04:54:35 +00:00
ARMInstrFormats.td Fix Thumb2 decoding of CPS instructions to mirror ARM decoding of the same instructions. 2011-08-23 17:45:18 +00:00
ARMInstrInfo.cpp Split am2offset into register addend and immediate addend forms, necessary for allowing the fixed-length disassembler to distinguish between SBFX and STR_PRE. 2011-07-26 20:54:26 +00:00
ARMInstrInfo.h Eliminate ARM::MOVi2pieces. Just use MOVi32imm and expand it to either movi+orr or movw+movt depending on the subtarget. 2010-11-12 23:03:38 +00:00
ARMInstrInfo.td Clean up predicates on ARM target instruction aliases. 2011-08-22 18:04:24 +00:00
ARMInstrNEON.td Remove the VMOVQQ pseudo instruction. 2011-08-20 00:52:40 +00:00
ARMInstrThumb.td Thumb parsing and encoding for SUB (SP minu immediate). 2011-08-24 21:42:27 +00:00
ARMInstrThumb2.td Perform more thorough checking of t2IT mask parameters, which fixes all remaining crashers when disassembling the entire 16-bit instruction space. 2011-08-24 22:40:22 +00:00
ARMInstrVFP.td Provide operand encoding information for half-precision VCVT instructions. Found by randomized testing. 2011-08-22 21:34:00 +00:00
ARMJITInfo.cpp Update comment to match my recent change. 2011-02-02 17:29:40 +00:00
ARMJITInfo.h Materialize GA addresses with movw + movt pairs for Darwin in PIC mode. e.g. 2011-01-17 08:03:18 +00:00
ARMLoadStoreOptimizer.cpp ARM STR_POST_IMM offset encoding fix in load/store optimizer. 2011-08-12 22:20:41 +00:00
ARMMCInstLower.cpp Sink ARMMCExpr and ARMAddressingModes into MC layer. First step to separate ARM MC code from target. 2011-07-20 23:34:39 +00:00
ARMMachineFunctionInfo.h Remove unused bitvectors that record ARM callee-saved registers. 2011-02-14 23:40:38 +00:00
ARMPerfectShuffle.h In r131488 I misunderstood how VREV works. It splits the vector in half and splits each half. Therefore, the real problem was that we were using a VREV64 for a 4xi16, when we should have been using a VREV32. 2011-05-18 06:42:21 +00:00
ARMRegisterInfo.cpp Trim a few unneeded includes. 2011-04-18 21:35:54 +00:00
ARMRegisterInfo.h
ARMRegisterInfo.td Thumb1 ADD/SUB SP instructions are predicable in Thumb2 mode. 2011-08-24 17:46:13 +00:00
ARMRelocations.h Add support to MOVimm32 using movt/movw for ARM JIT 2010-05-25 08:42:45 +00:00
ARMSchedule.td Sorry, several patches in one. 2011-01-20 08:34:58 +00:00
ARMScheduleA8.td Sorry, several patches in one. 2011-01-20 08:34:58 +00:00
ARMScheduleA9.td Improvements for the Cortex-A9 scheduling itineraries. 2011-04-19 18:11:36 +00:00
ARMScheduleV6.td Sorry, several patches in one. 2011-01-20 08:34:58 +00:00
ARMSelectionDAGInfo.cpp land David Blaikie's patch to de-constify Type, with a few tweaks. 2011-07-18 04:54:35 +00:00
ARMSelectionDAGInfo.h Sink ARMMCExpr and ARMAddressingModes into MC layer. First step to separate ARM MC code from target. 2011-07-20 23:34:39 +00:00
ARMSubtarget.cpp Next round of MC refactoring. This patch factor MC table instantiations, MC 2011-07-14 20:59:42 +00:00
ARMSubtarget.h Rewrite comment in English. 2011-07-07 19:09:06 +00:00
ARMTargetMachine.cpp Hide -global-merge option. 2011-08-25 01:22:49 +00:00
ARMTargetMachine.h - Move CodeModel from a TargetMachine global option to MCCodeGenInfo. 2011-07-20 07:51:56 +00:00
ARMTargetObjectFile.cpp Some first rudimentary support for ARM EHABI: print exception table in "text mode". 2011-03-05 18:43:15 +00:00
ARMTargetObjectFile.h Second set of ARM/MC/ELF changes. 2010-10-11 23:01:44 +00:00
CMakeLists.txt Don't generate the old-style disassembler in CMake builds either. 2011-08-09 21:36:11 +00:00
MLxExpansionPass.cpp - Rename TargetInstrDesc, TargetOperandInfo to MCInstrDesc and MCOperandInfo and 2011-06-28 19:10:37 +00:00
Makefile Don't continue generating the old-style decoder file. 2011-08-09 21:30:29 +00:00
NEONMoveFix.cpp Transfer implicit operands in NEONMoveFixPass. 2011-07-29 00:27:35 +00:00
README-Thumb.txt Add another note taken from the gcc bugzilla. 2011-01-12 09:06:19 +00:00
README-Thumb2.txt
README.txt Add an entry. 2011-07-13 01:33:00 +00:00
Thumb1FrameLowering.cpp Thumb1 ADD/SUB SP instructions are predicable in Thumb2 mode. 2011-08-24 17:46:13 +00:00
Thumb1FrameLowering.h Fix Target/ARM/Thumb1FrameLowering.h header guard. 2011-04-24 15:47:01 +00:00
Thumb1InstrInfo.cpp Refact ARM Thumb1 tMOVr instruction family. 2011-06-30 23:38:17 +00:00
Thumb1InstrInfo.h Move callee-saved regs spills / reloads to TFI 2010-11-27 23:05:03 +00:00
Thumb1RegisterInfo.cpp Thumb1 ADD/SUB SP instructions are predicable in Thumb2 mode. 2011-08-24 17:46:13 +00:00
Thumb1RegisterInfo.h Add a TRI::getLargestLegalSuperClass hook to provide an upper limit on register class inflation. 2011-04-26 18:52:33 +00:00
Thumb2ITBlockPass.cpp Refact ARM Thumb1 tMOVr instruction family. 2011-06-30 23:38:17 +00:00
Thumb2InstrInfo.cpp Thumb1 ADD/SUB SP instructions are predicable in Thumb2 mode. 2011-08-24 17:46:13 +00:00
Thumb2InstrInfo.h Making use of VFP / NEON floating point multiply-accumulate / subtraction is 2010-12-05 22:04:16 +00:00
Thumb2RegisterInfo.cpp Trim a few unneeded includes. 2011-04-18 21:35:54 +00:00
Thumb2RegisterInfo.h Implement frame unwinding information emission for Thumb1. Not finished yet because there is no way given the constpool index to examine the actual entry: the reason is clones inserted by constant island pass, which are not tracked at all! The only connection is done during asmprinting time via magic label names which is really gross and needs to be eventually fixed. 2011-03-05 18:43:50 +00:00
Thumb2SizeReduction.cpp Thumb1 ADD/SUB SP instructions are predicable in Thumb2 mode. 2011-08-24 17:46:13 +00:00

README.txt

//===---------------------------------------------------------------------===//
// Random ideas for the ARM backend.
//===---------------------------------------------------------------------===//

Reimplement 'select' in terms of 'SEL'.

* We would really like to support UXTAB16, but we need to prove that the
  add doesn't need to overflow between the two 16-bit chunks.

* Implement pre/post increment support.  (e.g. PR935)
* Implement smarter constant generation for binops with large immediates.

A few ARMv6T2 ops should be pattern matched: BFI, SBFX, and UBFX

Interesting optimization for PIC codegen on arm-linux:
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43129

//===---------------------------------------------------------------------===//

Crazy idea:  Consider code that uses lots of 8-bit or 16-bit values.  By the
time regalloc happens, these values are now in a 32-bit register, usually with
the top-bits known to be sign or zero extended.  If spilled, we should be able
to spill these to a 8-bit or 16-bit stack slot, zero or sign extending as part
of the reload.

Doing this reduces the size of the stack frame (important for thumb etc), and
also increases the likelihood that we will be able to reload multiple values
from the stack with a single load.

//===---------------------------------------------------------------------===//

The constant island pass is in good shape.  Some cleanups might be desirable,
but there is unlikely to be much improvement in the generated code.

1.  There may be some advantage to trying to be smarter about the initial
placement, rather than putting everything at the end.

2.  There might be some compile-time efficiency to be had by representing
consecutive islands as a single block rather than multiple blocks.

3.  Use a priority queue to sort constant pool users in inverse order of
    position so we always process the one closed to the end of functions
    first. This may simply CreateNewWater.

//===---------------------------------------------------------------------===//

Eliminate copysign custom expansion. We are still generating crappy code with
default expansion + if-conversion.

//===---------------------------------------------------------------------===//

Eliminate one instruction from:

define i32 @_Z6slow4bii(i32 %x, i32 %y) {
        %tmp = icmp sgt i32 %x, %y
        %retval = select i1 %tmp, i32 %x, i32 %y
        ret i32 %retval
}

__Z6slow4bii:
        cmp r0, r1
        movgt r1, r0
        mov r0, r1
        bx lr
=>

__Z6slow4bii:
        cmp r0, r1
        movle r0, r1
        bx lr

//===---------------------------------------------------------------------===//

Implement long long "X-3" with instructions that fold the immediate in.  These
were disabled due to badness with the ARM carry flag on subtracts.

//===---------------------------------------------------------------------===//

More load / store optimizations:
1) Better representation for block transfer? This is from Olden/power:

	fldd d0, [r4]
	fstd d0, [r4, #+32]
	fldd d0, [r4, #+8]
	fstd d0, [r4, #+40]
	fldd d0, [r4, #+16]
	fstd d0, [r4, #+48]
	fldd d0, [r4, #+24]
	fstd d0, [r4, #+56]

If we can spare the registers, it would be better to use fldm and fstm here.
Need major register allocator enhancement though.

2) Can we recognize the relative position of constantpool entries? i.e. Treat

	ldr r0, LCPI17_3
	ldr r1, LCPI17_4
	ldr r2, LCPI17_5

   as
	ldr r0, LCPI17
	ldr r1, LCPI17+4
	ldr r2, LCPI17+8

   Then the ldr's can be combined into a single ldm. See Olden/power.

Note for ARM v4 gcc uses ldmia to load a pair of 32-bit values to represent a
double 64-bit FP constant:

	adr	r0, L6
	ldmia	r0, {r0-r1}

	.align 2
L6:
	.long	-858993459
	.long	1074318540

3) struct copies appear to be done field by field 
instead of by words, at least sometimes:

struct foo { int x; short s; char c1; char c2; };
void cpy(struct foo*a, struct foo*b) { *a = *b; }

llvm code (-O2)
        ldrb r3, [r1, #+6]
        ldr r2, [r1]
        ldrb r12, [r1, #+7]
        ldrh r1, [r1, #+4]
        str r2, [r0]
        strh r1, [r0, #+4]
        strb r3, [r0, #+6]
        strb r12, [r0, #+7]
gcc code (-O2)
        ldmia   r1, {r1-r2}
        stmia   r0, {r1-r2}

In this benchmark poor handling of aggregate copies has shown up as
having a large effect on size, and possibly speed as well (we don't have
a good way to measure on ARM).

//===---------------------------------------------------------------------===//

* Consider this silly example:

double bar(double x) {  
  double r = foo(3.1);
  return x+r;
}

_bar:
        stmfd sp!, {r4, r5, r7, lr}
        add r7, sp, #8
        mov r4, r0
        mov r5, r1
        fldd d0, LCPI1_0
        fmrrd r0, r1, d0
        bl _foo
        fmdrr d0, r4, r5
        fmsr s2, r0
        fsitod d1, s2
        faddd d0, d1, d0
        fmrrd r0, r1, d0
        ldmfd sp!, {r4, r5, r7, pc}

Ignore the prologue and epilogue stuff for a second. Note 
	mov r4, r0
	mov r5, r1
the copys to callee-save registers and the fact they are only being used by the
fmdrr instruction. It would have been better had the fmdrr been scheduled
before the call and place the result in a callee-save DPR register. The two
mov ops would not have been necessary.

//===---------------------------------------------------------------------===//

Calling convention related stuff:

* gcc's parameter passing implementation is terrible and we suffer as a result:

e.g.
struct s {
  double d1;
  int s1;
};

void foo(struct s S) {
  printf("%g, %d\n", S.d1, S.s1);
}

'S' is passed via registers r0, r1, r2. But gcc stores them to the stack, and
then reload them to r1, r2, and r3 before issuing the call (r0 contains the
address of the format string):

	stmfd	sp!, {r7, lr}
	add	r7, sp, #0
	sub	sp, sp, #12
	stmia	sp, {r0, r1, r2}
	ldmia	sp, {r1-r2}
	ldr	r0, L5
	ldr	r3, [sp, #8]
L2:
	add	r0, pc, r0
	bl	L_printf$stub

Instead of a stmia, ldmia, and a ldr, wouldn't it be better to do three moves?

* Return an aggregate type is even worse:

e.g.
struct s foo(void) {
  struct s S = {1.1, 2};
  return S;
}

	mov	ip, r0
	ldr	r0, L5
	sub	sp, sp, #12
L2:
	add	r0, pc, r0
	@ lr needed for prologue
	ldmia	r0, {r0, r1, r2}
	stmia	sp, {r0, r1, r2}
	stmia	ip, {r0, r1, r2}
	mov	r0, ip
	add	sp, sp, #12
	bx	lr

r0 (and later ip) is the hidden parameter from caller to store the value in. The
first ldmia loads the constants into r0, r1, r2. The last stmia stores r0, r1,
r2 into the address passed in. However, there is one additional stmia that
stores r0, r1, and r2 to some stack location. The store is dead.

The llvm-gcc generated code looks like this:

csretcc void %foo(%struct.s* %agg.result) {
entry:
	%S = alloca %struct.s, align 4		; <%struct.s*> [#uses=1]
	%memtmp = alloca %struct.s		; <%struct.s*> [#uses=1]
	cast %struct.s* %S to sbyte*		; <sbyte*>:0 [#uses=2]
	call void %llvm.memcpy.i32( sbyte* %0, sbyte* cast ({ double, int }* %C.0.904 to sbyte*), uint 12, uint 4 )
	cast %struct.s* %agg.result to sbyte*		; <sbyte*>:1 [#uses=2]
	call void %llvm.memcpy.i32( sbyte* %1, sbyte* %0, uint 12, uint 0 )
	cast %struct.s* %memtmp to sbyte*		; <sbyte*>:2 [#uses=1]
	call void %llvm.memcpy.i32( sbyte* %2, sbyte* %1, uint 12, uint 0 )
	ret void
}

llc ends up issuing two memcpy's (the first memcpy becomes 3 loads from
constantpool). Perhaps we should 1) fix llvm-gcc so the memcpy is translated
into a number of load and stores, or 2) custom lower memcpy (of small size) to
be ldmia / stmia. I think option 2 is better but the current register
allocator cannot allocate a chunk of registers at a time.

A feasible temporary solution is to use specific physical registers at the
lowering time for small (<= 4 words?) transfer size.

* ARM CSRet calling convention requires the hidden argument to be returned by
the callee.

//===---------------------------------------------------------------------===//

We can definitely do a better job on BB placements to eliminate some branches.
It's very common to see llvm generated assembly code that looks like this:

LBB3:
 ...
LBB4:
...
  beq LBB3
  b LBB2

If BB4 is the only predecessor of BB3, then we can emit BB3 after BB4. We can
then eliminate beq and and turn the unconditional branch to LBB2 to a bne.

See McCat/18-imp/ComputeBoundingBoxes for an example.

//===---------------------------------------------------------------------===//

Pre-/post- indexed load / stores:

1) We should not make the pre/post- indexed load/store transform if the base ptr
is guaranteed to be live beyond the load/store. This can happen if the base
ptr is live out of the block we are performing the optimization. e.g.

mov r1, r2
ldr r3, [r1], #4
...

vs.

ldr r3, [r2]
add r1, r2, #4
...

In most cases, this is just a wasted optimization. However, sometimes it can
negatively impact the performance because two-address code is more restrictive
when it comes to scheduling.

Unfortunately, liveout information is currently unavailable during DAG combine
time.

2) Consider spliting a indexed load / store into a pair of add/sub + load/store
   to solve #1 (in TwoAddressInstructionPass.cpp).

3) Enhance LSR to generate more opportunities for indexed ops.

4) Once we added support for multiple result patterns, write indexed loads
   patterns instead of C++ instruction selection code.

5) Use VLDM / VSTM to emulate indexed FP load / store.

//===---------------------------------------------------------------------===//

Implement support for some more tricky ways to materialize immediates.  For
example, to get 0xffff8000, we can use:

mov r9, #&3f8000
sub r9, r9, #&400000

//===---------------------------------------------------------------------===//

We sometimes generate multiple add / sub instructions to update sp in prologue
and epilogue if the inc / dec value is too large to fit in a single immediate
operand. In some cases, perhaps it might be better to load the value from a
constantpool instead.

//===---------------------------------------------------------------------===//

GCC generates significantly better code for this function.

int foo(int StackPtr, unsigned char *Line, unsigned char *Stack, int LineLen) {
    int i = 0;

    if (StackPtr != 0) {
       while (StackPtr != 0 && i < (((LineLen) < (32768))? (LineLen) : (32768)))
          Line[i++] = Stack[--StackPtr];
        if (LineLen > 32768)
        {
            while (StackPtr != 0 && i < LineLen)
            {
                i++;
                --StackPtr;
            }
        }
    }
    return StackPtr;
}

//===---------------------------------------------------------------------===//

This should compile to the mlas instruction:
int mlas(int x, int y, int z) { return ((x * y + z) < 0) ? 7 : 13; }

//===---------------------------------------------------------------------===//

At some point, we should triage these to see if they still apply to us:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19598
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18560
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27016

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11831
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11826
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11825
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11824
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11823
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11820
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10982

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10242
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9831
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9760
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9759
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9703
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9702
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9663

http://www.inf.u-szeged.hu/gcc-arm/
http://citeseer.ist.psu.edu/debus04linktime.html

//===---------------------------------------------------------------------===//

gcc generates smaller code for this function at -O2 or -Os:

void foo(signed char* p) {
  if (*p == 3)
     bar();
   else if (*p == 4)
    baz();
  else if (*p == 5)
    quux();
}

llvm decides it's a good idea to turn the repeated if...else into a
binary tree, as if it were a switch; the resulting code requires -1 
compare-and-branches when *p<=2 or *p==5, the same number if *p==4
or *p>6, and +1 if *p==3.  So it should be a speed win
(on balance).  However, the revised code is larger, with 4 conditional 
branches instead of 3.

More seriously, there is a byte->word extend before
each comparison, where there should be only one, and the condition codes
are not remembered when the same two values are compared twice.

//===---------------------------------------------------------------------===//

More LSR enhancements possible:

1. Teach LSR about pre- and post- indexed ops to allow iv increment be merged
   in a load / store.
2. Allow iv reuse even when a type conversion is required. For example, i8
   and i32 load / store addressing modes are identical.


//===---------------------------------------------------------------------===//

This:

int foo(int a, int b, int c, int d) {
  long long acc = (long long)a * (long long)b;
  acc += (long long)c * (long long)d;
  return (int)(acc >> 32);
}

Should compile to use SMLAL (Signed Multiply Accumulate Long) which multiplies 
two signed 32-bit values to produce a 64-bit value, and accumulates this with 
a 64-bit value.

We currently get this with both v4 and v6:

_foo:
        smull r1, r0, r1, r0
        smull r3, r2, r3, r2
        adds r3, r3, r1
        adc r0, r2, r0
        bx lr

//===---------------------------------------------------------------------===//

This:
        #include <algorithm>
        std::pair<unsigned, bool> full_add(unsigned a, unsigned b)
        { return std::make_pair(a + b, a + b < a); }
        bool no_overflow(unsigned a, unsigned b)
        { return !full_add(a, b).second; }

Should compile to:

_Z8full_addjj:
	adds	r2, r1, r2
	movcc	r1, #0
	movcs	r1, #1
	str	r2, [r0, #0]
	strb	r1, [r0, #4]
	mov	pc, lr

_Z11no_overflowjj:
	cmn	r0, r1
	movcs	r0, #0
	movcc	r0, #1
	mov	pc, lr

not:

__Z8full_addjj:
        add r3, r2, r1
        str r3, [r0]
        mov r2, #1
        mov r12, #0
        cmp r3, r1
        movlo r12, r2
        str r12, [r0, #+4]
        bx lr
__Z11no_overflowjj:
        add r3, r1, r0
        mov r2, #1
        mov r1, #0
        cmp r3, r0
        movhs r1, r2
        mov r0, r1
        bx lr

//===---------------------------------------------------------------------===//

Some of the NEON intrinsics may be appropriate for more general use, either
as target-independent intrinsics or perhaps elsewhere in the ARM backend.
Some of them may also be lowered to target-independent SDNodes, and perhaps
some new SDNodes could be added.

For example, maximum, minimum, and absolute value operations are well-defined
and standard operations, both for vector and scalar types.

The current NEON-specific intrinsics for count leading zeros and count one
bits could perhaps be replaced by the target-independent ctlz and ctpop
intrinsics.  It may also make sense to add a target-independent "ctls"
intrinsic for "count leading sign bits".  Likewise, the backend could use
the target-independent SDNodes for these operations.

ARMv6 has scalar saturating and halving adds and subtracts.  The same
intrinsics could possibly be used for both NEON's vector implementations of
those operations and the ARMv6 scalar versions.

//===---------------------------------------------------------------------===//

ARM::MOVCCr is commutable (by flipping the condition). But we need to implement
ARMInstrInfo::commuteInstruction() to support it.

//===---------------------------------------------------------------------===//

Split out LDR (literal) from normal ARM LDR instruction. Also consider spliting
LDR into imm12 and so_reg forms. This allows us to clean up some code. e.g.
ARMLoadStoreOptimizer does not need to look at LDR (literal) and LDR (so_reg)
while ARMConstantIslandPass only need to worry about LDR (literal).

//===---------------------------------------------------------------------===//

Constant island pass should make use of full range SoImm values for LEApcrel.
Be careful though as the last attempt caused infinite looping on lencod.

//===---------------------------------------------------------------------===//

Predication issue. This function:   

extern unsigned array[ 128 ];
int     foo( int x ) {
  int     y;
  y = array[ x & 127 ];
  if ( x & 128 )
     y = 123456789 & ( y >> 2 );
  else
     y = 123456789 & y;
  return y;
}

compiles to:

_foo:
	and r1, r0, #127
	ldr r2, LCPI1_0
	ldr r2, [r2]
	ldr r1, [r2, +r1, lsl #2]
	mov r2, r1, lsr #2
	tst r0, #128
	moveq r2, r1
	ldr r0, LCPI1_1
	and r0, r2, r0
	bx lr

It would be better to do something like this, to fold the shift into the
conditional move:

	and r1, r0, #127
	ldr r2, LCPI1_0
	ldr r2, [r2]
	ldr r1, [r2, +r1, lsl #2]
	tst r0, #128
	movne r1, r1, lsr #2
	ldr r0, LCPI1_1
	and r0, r1, r0
	bx lr

it saves an instruction and a register.

//===---------------------------------------------------------------------===//

It might be profitable to cse MOVi16 if there are lots of 32-bit immediates
with the same bottom half.

//===---------------------------------------------------------------------===//

Robert Muth started working on an alternate jump table implementation that
does not put the tables in-line in the text.  This is more like the llvm
default jump table implementation.  This might be useful sometime.  Several
revisions of patches are on the mailing list, beginning at:
http://lists.cs.uiuc.edu/pipermail/llvmdev/2009-June/022763.html

//===---------------------------------------------------------------------===//

Make use of the "rbit" instruction.

//===---------------------------------------------------------------------===//

Take a look at test/CodeGen/Thumb2/machine-licm.ll. ARM should be taught how
to licm and cse the unnecessary load from cp#1.

//===---------------------------------------------------------------------===//

The CMN instruction sets the flags like an ADD instruction, while CMP sets
them like a subtract. Therefore to be able to use CMN for comparisons other
than the Z bit, we'll need additional logic to reverse the conditionals
associated with the comparison. Perhaps a pseudo-instruction for the comparison,
with a post-codegen pass to clean up and handle the condition codes?
See PR5694 for testcase.

//===---------------------------------------------------------------------===//

Given the following on armv5:
int test1(int A, int B) {
  return (A&-8388481)|(B&8388480);
}

We currently generate:
	ldr	r2, .LCPI0_0
	and	r0, r0, r2
	ldr	r2, .LCPI0_1
	and	r1, r1, r2
	orr	r0, r1, r0
	bx	lr

We should be able to replace the second ldr+and with a bic (i.e. reuse the
constant which was already loaded).  Not sure what's necessary to do that.

//===---------------------------------------------------------------------===//

The code generated for bswap on armv4/5 (CPUs without rev) is less than ideal:

int a(int x) { return __builtin_bswap32(x); }

a:
	mov	r1, #255, 24
	mov	r2, #255, 16
	and	r1, r1, r0, lsr #8
	and	r2, r2, r0, lsl #8
	orr	r1, r1, r0, lsr #24
	orr	r0, r2, r0, lsl #24
	orr	r0, r0, r1
	bx	lr

Something like the following would be better (fewer instructions/registers):
	eor     r1, r0, r0, ror #16
	bic     r1, r1, #0xff0000
	mov     r1, r1, lsr #8
	eor     r0, r1, r0, ror #8
	bx	lr

A custom Thumb version would also be a slight improvement over the generic
version.

//===---------------------------------------------------------------------===//

Consider the following simple C code:

void foo(unsigned char *a, unsigned char *b, int *c) {
 if ((*a | *b) == 0) *c = 0;
}

currently llvm-gcc generates something like this (nice branchless code I'd say):

       ldrb    r0, [r0]
       ldrb    r1, [r1]
       orr     r0, r1, r0
       tst     r0, #255
       moveq   r0, #0
       streq   r0, [r2]
       bx      lr

Note that both "tst" and "moveq" are redundant.

//===---------------------------------------------------------------------===//

When loading immediate constants with movt/movw, if there are multiple
constants needed with the same low 16 bits, and those values are not live at
the same time, it would be possible to use a single movw instruction, followed
by multiple movt instructions to rewrite the high bits to different values.
For example:

  volatile store i32 -1, i32* inttoptr (i32 1342210076 to i32*), align 4,
  !tbaa
!0
  volatile store i32 -1, i32* inttoptr (i32 1342341148 to i32*), align 4,
  !tbaa
!0

is compiled and optimized to:

    movw    r0, #32796
    mov.w    r1, #-1
    movt    r0, #20480
    str    r1, [r0]
    movw    r0, #32796    @ <= this MOVW is not needed, value is there already
    movt    r0, #20482
    str    r1, [r0]

//===---------------------------------------------------------------------===//

Improve codegen for select's:
if (x != 0) x = 1
if (x == 1) x = 1

ARM codegen used to look like this:
       mov     r1, r0
       cmp     r1, #1
       mov     r0, #0
       moveq   r0, #1

The naive lowering select between two different values. It should recognize the
test is equality test so it's more a conditional move rather than a select:
       cmp     r0, #1
       movne   r0, #0

Currently this is a ARM specific dag combine. We probably should make it into a
target-neutral one.