X-Git-Url: http://demsky.eecs.uci.edu/git/?a=blobdiff_plain;f=lib%2FTarget%2FARM%2FREADME.txt;h=57dc6cb88bed483d42db746dc1a68a154a22f714;hb=67297cd9563d6385570be219098faf8d0fe4c0de;hp=a6f26a5dfe1954d7f3fc66158391aa10959d0fd3;hpb=d5d2baec2609da3ade9ca205e87c88d35e9e6976;p=oota-llvm.git diff --git a/lib/Target/ARM/README.txt b/lib/Target/ARM/README.txt index a6f26a5dfe1..57dc6cb88be 100644 --- a/lib/Target/ARM/README.txt +++ b/lib/Target/ARM/README.txt @@ -10,6 +10,11 @@ Reimplement 'select' in terms of 'SEL'. * 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 @@ -71,26 +76,6 @@ were disabled due to badness with the ARM carry flag on subtracts. //===---------------------------------------------------------------------===// -We currently compile abs: -int foo(int p) { return p < 0 ? -p : p; } - -into: - -_foo: - rsb r1, r0, #0 - cmn r0, #1 - movgt r1, r0 - mov r0, r1 - bx lr - -This is very, uh, literal. This could be a 3 operation sequence: - t = (p sra 31); - res = (p xor t)-t - -Which would be better. This occurs in png decode. - -//===---------------------------------------------------------------------===// - More load / store optimizations: 1) Better representation for block transfer? This is from Olden/power: @@ -516,11 +501,6 @@ 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) @@ -605,3 +585,149 @@ 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. + +//===---------------------------------------------------------------------===// + +Optimize unnecessary checks for zero with __builtin_clz/ctz. Those builtins +are specified to be undefined at zero, so portable code must check for zero +and handle it as a special case. That is unnecessary on ARM where those +operations are implemented in a way that is well-defined for zero. For +example: + +int f(int x) { return x ? __builtin_clz(x) : sizeof(int)*8; } + +should just be implemented with a CLZ instruction. Since there are other +targets, e.g., PPC, that share this behavior, it would be best to implement +this in a target-independent way: we should probably fold that (when using +"undefined at zero" semantics) to set the "defined at zero" bit and have +the code generator expand out the right code. + + +//===---------------------------------------------------------------------===// + +Clean up the test/MC/ARM files to have more robust register choices. + +R0 should not be used as a register operand in the assembler tests as it's then +not possible to distinguish between a correct encoding and a missing operand +encoding, as zero is the default value for the binary encoder. +e.g., + add r0, r0 // bad + add r3, r5 // good + +Register operands should be distinct. That is, when the encoding does not +require two syntactical operands to refer to the same register, two different +registers should be used in the test so as to catch errors where the +operands are swapped in the encoding. +e.g., + subs.w r1, r1, r1 // bad + subs.w r1, r2, r3 // good +