1 Target Independent Opportunities:
3 //===---------------------------------------------------------------------===//
5 With the recent changes to make the implicit def/use set explicit in
6 machineinstrs, we should change the target descriptions for 'call' instructions
7 so that the .td files don't list all the call-clobbered registers as implicit
8 defs. Instead, these should be added by the code generator (e.g. on the dag).
10 This has a number of uses:
12 1. PPC32/64 and X86 32/64 can avoid having multiple copies of call instructions
13 for their different impdef sets.
14 2. Targets with multiple calling convs (e.g. x86) which have different clobber
15 sets don't need copies of call instructions.
16 3. 'Interprocedural register allocation' can be done to reduce the clobber sets
19 //===---------------------------------------------------------------------===//
21 Make the PPC branch selector target independant
23 //===---------------------------------------------------------------------===//
25 Get the C front-end to expand hypot(x,y) -> llvm.sqrt(x*x+y*y) when errno and
26 precision don't matter (ffastmath). Misc/mandel will like this. :) This isn't
27 safe in general, even on darwin. See the libm implementation of hypot for
28 examples (which special case when x/y are exactly zero to get signed zeros etc
31 //===---------------------------------------------------------------------===//
33 Solve this DAG isel folding deficiency:
51 The problem is the store's chain operand is not the load X but rather
52 a TokenFactor of the load X and load Y, which prevents the folding.
54 There are two ways to fix this:
56 1. The dag combiner can start using alias analysis to realize that y/x
57 don't alias, making the store to X not dependent on the load from Y.
58 2. The generated isel could be made smarter in the case it can't
59 disambiguate the pointers.
61 Number 1 is the preferred solution.
63 This has been "fixed" by a TableGen hack. But that is a short term workaround
64 which will be removed once the proper fix is made.
66 //===---------------------------------------------------------------------===//
68 On targets with expensive 64-bit multiply, we could LSR this:
75 for (i = ...; ++i, tmp+=tmp)
78 This would be a win on ppc32, but not x86 or ppc64.
80 //===---------------------------------------------------------------------===//
82 Shrink: (setlt (loadi32 P), 0) -> (setlt (loadi8 Phi), 0)
84 //===---------------------------------------------------------------------===//
86 Reassociate should turn: X*X*X*X -> t=(X*X) (t*t) to eliminate a multiply.
88 //===---------------------------------------------------------------------===//
90 Interesting? testcase for add/shift/mul reassoc:
92 int bar(int x, int y) {
93 return x*x*x+y+x*x*x*x*x*y*y*y*y;
95 int foo(int z, int n) {
96 return bar(z, n) + bar(2*z, 2*n);
99 Reassociate should handle the example in GCC PR16157.
101 //===---------------------------------------------------------------------===//
103 These two functions should generate the same code on big-endian systems:
105 int g(int *j,int *l) { return memcmp(j,l,4); }
106 int h(int *j, int *l) { return *j - *l; }
108 this could be done in SelectionDAGISel.cpp, along with other special cases,
111 //===---------------------------------------------------------------------===//
113 It would be nice to revert this patch:
114 http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20060213/031986.html
116 And teach the dag combiner enough to simplify the code expanded before
117 legalize. It seems plausible that this knowledge would let it simplify other
120 //===---------------------------------------------------------------------===//
122 For vector types, TargetData.cpp::getTypeInfo() returns alignment that is equal
123 to the type size. It works but can be overly conservative as the alignment of
124 specific vector types are target dependent.
126 //===---------------------------------------------------------------------===//
128 We should produce an unaligned load from code like this:
130 v4sf example(float *P) {
131 return (v4sf){P[0], P[1], P[2], P[3] };
134 //===---------------------------------------------------------------------===//
136 Add support for conditional increments, and other related patterns. Instead
141 je LBB16_2 #cond_next
152 //===---------------------------------------------------------------------===//
154 Combine: a = sin(x), b = cos(x) into a,b = sincos(x).
156 Expand these to calls of sin/cos and stores:
157 double sincos(double x, double *sin, double *cos);
158 float sincosf(float x, float *sin, float *cos);
159 long double sincosl(long double x, long double *sin, long double *cos);
161 Doing so could allow SROA of the destination pointers. See also:
162 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687
164 This is now easily doable with MRVs. We could even make an intrinsic for this
165 if anyone cared enough about sincos.
167 //===---------------------------------------------------------------------===//
169 Turn this into a single byte store with no load (the other 3 bytes are
172 define void @test(i32* %P) {
174 %tmp14 = or i32 %tmp, 3305111552
175 %tmp15 = and i32 %tmp14, 3321888767
176 store i32 %tmp15, i32* %P
180 //===---------------------------------------------------------------------===//
182 dag/inst combine "clz(x)>>5 -> x==0" for 32-bit x.
188 int t = __builtin_clz(x);
198 //===---------------------------------------------------------------------===//
200 quantum_sigma_x in 462.libquantum contains the following loop:
202 for(i=0; i<reg->size; i++)
204 /* Flip the target bit of each basis state */
205 reg->node[i].state ^= ((MAX_UNSIGNED) 1 << target);
208 Where MAX_UNSIGNED/state is a 64-bit int. On a 32-bit platform it would be just
209 so cool to turn it into something like:
211 long long Res = ((MAX_UNSIGNED) 1 << target);
213 for(i=0; i<reg->size; i++)
214 reg->node[i].state ^= Res & 0xFFFFFFFFULL;
216 for(i=0; i<reg->size; i++)
217 reg->node[i].state ^= Res & 0xFFFFFFFF00000000ULL
220 ... which would only do one 32-bit XOR per loop iteration instead of two.
222 It would also be nice to recognize the reg->size doesn't alias reg->node[i], but
225 //===---------------------------------------------------------------------===//
227 This should be optimized to one 'and' and one 'or', from PR4216:
229 define i32 @test_bitfield(i32 %bf.prev.low) nounwind ssp {
231 %bf.prev.lo.cleared10 = or i32 %bf.prev.low, 32962 ; <i32> [#uses=1]
232 %0 = and i32 %bf.prev.low, -65536 ; <i32> [#uses=1]
233 %1 = and i32 %bf.prev.lo.cleared10, 40186 ; <i32> [#uses=1]
234 %2 = or i32 %1, %0 ; <i32> [#uses=1]
238 //===---------------------------------------------------------------------===//
240 This isn't recognized as bswap by instcombine (yes, it really is bswap):
242 unsigned long reverse(unsigned v) {
244 t = v ^ ((v << 16) | (v >> 16));
246 v = (v << 24) | (v >> 8);
250 //===---------------------------------------------------------------------===//
252 These idioms should be recognized as popcount (see PR1488):
254 unsigned countbits_slow(unsigned v) {
256 for (c = 0; v; v >>= 1)
260 unsigned countbits_fast(unsigned v){
263 v &= v - 1; // clear the least significant bit set
267 BITBOARD = unsigned long long
268 int PopCnt(register BITBOARD a) {
276 unsigned int popcount(unsigned int input) {
277 unsigned int count = 0;
278 for (unsigned int i = 0; i < 4 * 8; i++)
279 count += (input >> i) & i;
283 //===---------------------------------------------------------------------===//
285 These should turn into single 16-bit (unaligned?) loads on little/big endian
288 unsigned short read_16_le(const unsigned char *adr) {
289 return adr[0] | (adr[1] << 8);
291 unsigned short read_16_be(const unsigned char *adr) {
292 return (adr[0] << 8) | adr[1];
295 //===---------------------------------------------------------------------===//
297 -instcombine should handle this transform:
298 icmp pred (sdiv X / C1 ), C2
299 when X, C1, and C2 are unsigned. Similarly for udiv and signed operands.
301 Currently InstCombine avoids this transform but will do it when the signs of
302 the operands and the sign of the divide match. See the FIXME in
303 InstructionCombining.cpp in the visitSetCondInst method after the switch case
304 for Instruction::UDiv (around line 4447) for more details.
306 The SingleSource/Benchmarks/Shootout-C++/hash and hash2 tests have examples of
309 //===---------------------------------------------------------------------===//
311 viterbi speeds up *significantly* if the various "history" related copy loops
312 are turned into memcpy calls at the source level. We need a "loops to memcpy"
315 //===---------------------------------------------------------------------===//
319 typedef unsigned U32;
320 typedef unsigned long long U64;
321 int test (U32 *inst, U64 *regs) {
324 int r1 = (temp >> 20) & 0xf;
325 int b2 = (temp >> 16) & 0xf;
326 effective_addr2 = temp & 0xfff;
327 if (b2) effective_addr2 += regs[b2];
328 b2 = (temp >> 12) & 0xf;
329 if (b2) effective_addr2 += regs[b2];
330 effective_addr2 &= regs[4];
331 if ((effective_addr2 & 3) == 0)
336 Note that only the low 2 bits of effective_addr2 are used. On 32-bit systems,
337 we don't eliminate the computation of the top half of effective_addr2 because
338 we don't have whole-function selection dags. On x86, this means we use one
339 extra register for the function when effective_addr2 is declared as U64 than
340 when it is declared U32.
342 PHI Slicing could be extended to do this.
344 //===---------------------------------------------------------------------===//
346 LSR should know what GPR types a target has. This code:
348 volatile short X, Y; // globals
352 for (i = 0; i < N; i++) { X = i; Y = i*4; }
355 produces two near identical IV's (after promotion) on PPC/ARM:
365 add r2, r2, #1 <- [0,+,1]
366 sub r0, r0, #1 <- [0,-,1]
370 LSR should reuse the "+" IV for the exit test.
373 //===---------------------------------------------------------------------===//
375 Tail call elim should be more aggressive, checking to see if the call is
376 followed by an uncond branch to an exit block.
378 ; This testcase is due to tail-duplication not wanting to copy the return
379 ; instruction into the terminating blocks because there was other code
380 ; optimized out of the function after the taildup happened.
381 ; RUN: llvm-as < %s | opt -tailcallelim | llvm-dis | not grep call
383 define i32 @t4(i32 %a) {
385 %tmp.1 = and i32 %a, 1 ; <i32> [#uses=1]
386 %tmp.2 = icmp ne i32 %tmp.1, 0 ; <i1> [#uses=1]
387 br i1 %tmp.2, label %then.0, label %else.0
389 then.0: ; preds = %entry
390 %tmp.5 = add i32 %a, -1 ; <i32> [#uses=1]
391 %tmp.3 = call i32 @t4( i32 %tmp.5 ) ; <i32> [#uses=1]
394 else.0: ; preds = %entry
395 %tmp.7 = icmp ne i32 %a, 0 ; <i1> [#uses=1]
396 br i1 %tmp.7, label %then.1, label %return
398 then.1: ; preds = %else.0
399 %tmp.11 = add i32 %a, -2 ; <i32> [#uses=1]
400 %tmp.9 = call i32 @t4( i32 %tmp.11 ) ; <i32> [#uses=1]
403 return: ; preds = %then.1, %else.0, %then.0
404 %result.0 = phi i32 [ 0, %else.0 ], [ %tmp.3, %then.0 ],
409 //===---------------------------------------------------------------------===//
411 Tail recursion elimination should handle:
416 return 2 * pow2m1 (n - 1) + 1;
419 Also, multiplies can be turned into SHL's, so they should be handled as if
420 they were associative. "return foo() << 1" can be tail recursion eliminated.
422 //===---------------------------------------------------------------------===//
424 Argument promotion should promote arguments for recursive functions, like
427 ; RUN: llvm-as < %s | opt -argpromotion | llvm-dis | grep x.val
429 define internal i32 @foo(i32* %x) {
431 %tmp = load i32* %x ; <i32> [#uses=0]
432 %tmp.foo = call i32 @foo( i32* %x ) ; <i32> [#uses=1]
436 define i32 @bar(i32* %x) {
438 %tmp3 = call i32 @foo( i32* %x ) ; <i32> [#uses=1]
442 //===---------------------------------------------------------------------===//
444 "basicaa" should know how to look through "or" instructions that act like add
445 instructions. For example in this code, the x*4+1 is turned into x*4 | 1, and
446 basicaa can't analyze the array subscript, leading to duplicated loads in the
449 void test(int X, int Y, int a[]) {
451 for (i=2; i<1000; i+=4) {
452 a[i+0] = a[i-1+0]*a[i-2+0];
453 a[i+1] = a[i-1+1]*a[i-2+1];
454 a[i+2] = a[i-1+2]*a[i-2+2];
455 a[i+3] = a[i-1+3]*a[i-2+3];
459 BasicAA also doesn't do this for add. It needs to know that &A[i+1] != &A[i].
461 //===---------------------------------------------------------------------===//
463 We should investigate an instruction sinking pass. Consider this silly
479 je LBB1_2 # cond_true
487 The PIC base computation (call+popl) is only used on one path through the
488 code, but is currently always computed in the entry block. It would be
489 better to sink the picbase computation down into the block for the
490 assertion, as it is the only one that uses it. This happens for a lot of
491 code with early outs.
493 Another example is loads of arguments, which are usually emitted into the
494 entry block on targets like x86. If not used in all paths through a
495 function, they should be sunk into the ones that do.
497 In this case, whole-function-isel would also handle this.
499 //===---------------------------------------------------------------------===//
501 Investigate lowering of sparse switch statements into perfect hash tables:
502 http://burtleburtle.net/bob/hash/perfect.html
504 //===---------------------------------------------------------------------===//
506 We should turn things like "load+fabs+store" and "load+fneg+store" into the
507 corresponding integer operations. On a yonah, this loop:
512 for (b = 0; b < 10000000; b++)
513 for (i = 0; i < 256; i++)
517 is twice as slow as this loop:
522 for (b = 0; b < 10000000; b++)
523 for (i = 0; i < 256; i++)
524 a[i] ^= (1ULL << 63);
527 and I suspect other processors are similar. On X86 in particular this is a
528 big win because doing this with integers allows the use of read/modify/write
531 //===---------------------------------------------------------------------===//
533 DAG Combiner should try to combine small loads into larger loads when
534 profitable. For example, we compile this C++ example:
536 struct THotKey { short Key; bool Control; bool Shift; bool Alt; };
537 extern THotKey m_HotKey;
538 THotKey GetHotKey () { return m_HotKey; }
540 into (-O3 -fno-exceptions -static -fomit-frame-pointer):
545 movb _m_HotKey+3, %cl
546 movb _m_HotKey+4, %dl
547 movb _m_HotKey+2, %ch
562 movzwl _m_HotKey+4, %edx
566 The LLVM IR contains the needed alignment info, so we should be able to
567 merge the loads and stores into 4-byte loads:
569 %struct.THotKey = type { i16, i8, i8, i8 }
570 define void @_Z9GetHotKeyv(%struct.THotKey* sret %agg.result) nounwind {
572 %tmp2 = load i16* getelementptr (@m_HotKey, i32 0, i32 0), align 8
573 %tmp5 = load i8* getelementptr (@m_HotKey, i32 0, i32 1), align 2
574 %tmp8 = load i8* getelementptr (@m_HotKey, i32 0, i32 2), align 1
575 %tmp11 = load i8* getelementptr (@m_HotKey, i32 0, i32 3), align 2
577 Alternatively, we should use a small amount of base-offset alias analysis
578 to make it so the scheduler doesn't need to hold all the loads in regs at
581 //===---------------------------------------------------------------------===//
583 We should add an FRINT node to the DAG to model targets that have legal
584 implementations of ceil/floor/rint.
586 //===---------------------------------------------------------------------===//
591 long long input[8] = {1,1,1,1,1,1,1,1};
595 We currently compile this into a memcpy from a global array since the
596 initializer is fairly large and not memset'able. This is good, but the memcpy
597 gets lowered to load/stores in the code generator. This is also ok, except
598 that the codegen lowering for memcpy doesn't handle the case when the source
599 is a constant global. This gives us atrocious code like this:
604 movl _C.0.1444-"L1$pb"+32(%eax), %ecx
606 movl _C.0.1444-"L1$pb"+20(%eax), %ecx
608 movl _C.0.1444-"L1$pb"+36(%eax), %ecx
610 movl _C.0.1444-"L1$pb"+44(%eax), %ecx
612 movl _C.0.1444-"L1$pb"+40(%eax), %ecx
614 movl _C.0.1444-"L1$pb"+12(%eax), %ecx
616 movl _C.0.1444-"L1$pb"+4(%eax), %ecx
628 //===---------------------------------------------------------------------===//
630 http://llvm.org/PR717:
632 The following code should compile into "ret int undef". Instead, LLVM
633 produces "ret int 0":
642 //===---------------------------------------------------------------------===//
644 The loop unroller should partially unroll loops (instead of peeling them)
645 when code growth isn't too bad and when an unroll count allows simplification
646 of some code within the loop. One trivial example is:
652 for ( nLoop = 0; nLoop < 1000; nLoop++ ) {
661 Unrolling by 2 would eliminate the '&1' in both copies, leading to a net
662 reduction in code size. The resultant code would then also be suitable for
663 exit value computation.
665 //===---------------------------------------------------------------------===//
667 We miss a bunch of rotate opportunities on various targets, including ppc, x86,
668 etc. On X86, we miss a bunch of 'rotate by variable' cases because the rotate
669 matching code in dag combine doesn't look through truncates aggressively
670 enough. Here are some testcases reduces from GCC PR17886:
672 unsigned long long f(unsigned long long x, int y) {
673 return (x << y) | (x >> 64-y);
675 unsigned f2(unsigned x, int y){
676 return (x << y) | (x >> 32-y);
678 unsigned long long f3(unsigned long long x){
680 return (x << y) | (x >> 64-y);
682 unsigned f4(unsigned x){
684 return (x << y) | (x >> 32-y);
686 unsigned long long f5(unsigned long long x, unsigned long long y) {
687 return (x << 8) | ((y >> 48) & 0xffull);
689 unsigned long long f6(unsigned long long x, unsigned long long y, int z) {
692 return (x << 8) | ((y >> 48) & 0xffull);
694 return (x << 16) | ((y >> 40) & 0xffffull);
696 return (x << 24) | ((y >> 32) & 0xffffffull);
698 return (x << 32) | ((y >> 24) & 0xffffffffull);
700 return (x << 40) | ((y >> 16) & 0xffffffffffull);
704 On X86-64, we only handle f2/f3/f4 right. On x86-32, a few of these
705 generate truly horrible code, instead of using shld and friends. On
706 ARM, we end up with calls to L___lshrdi3/L___ashldi3 in f, which is
707 badness. PPC64 misses f, f5 and f6. CellSPU aborts in isel.
709 //===---------------------------------------------------------------------===//
711 We do a number of simplifications in simplify libcalls to strength reduce
712 standard library functions, but we don't currently merge them together. For
713 example, it is useful to merge memcpy(a,b,strlen(b)) -> strcpy. This can only
714 be done safely if "b" isn't modified between the strlen and memcpy of course.
716 //===---------------------------------------------------------------------===//
718 Reassociate should turn things like:
720 int factorial(int X) {
721 return X*X*X*X*X*X*X*X;
724 into llvm.powi calls, allowing the code generator to produce balanced
725 multiplication trees.
727 //===---------------------------------------------------------------------===//
729 We generate a horrible libcall for llvm.powi. For example, we compile:
732 double f(double a) { return std::pow(a, 4); }
738 movsd 16(%esp), %xmm0
741 call L___powidf2$stub
749 movsd 16(%esp), %xmm0
757 //===---------------------------------------------------------------------===//
759 We compile this program: (from GCC PR11680)
760 http://gcc.gnu.org/bugzilla/attachment.cgi?id=4487
762 Into code that runs the same speed in fast/slow modes, but both modes run 2x
763 slower than when compile with GCC (either 4.0 or 4.2):
765 $ llvm-g++ perf.cpp -O3 -fno-exceptions
767 1.821u 0.003s 0:01.82 100.0% 0+0k 0+0io 0pf+0w
769 $ g++ perf.cpp -O3 -fno-exceptions
771 0.821u 0.001s 0:00.82 100.0% 0+0k 0+0io 0pf+0w
773 It looks like we are making the same inlining decisions, so this may be raw
774 codegen badness or something else (haven't investigated).
776 //===---------------------------------------------------------------------===//
778 We miss some instcombines for stuff like this:
780 void foo (unsigned int a) {
781 /* This one is equivalent to a >= (3 << 2). */
786 A few other related ones are in GCC PR14753.
788 //===---------------------------------------------------------------------===//
790 Divisibility by constant can be simplified (according to GCC PR12849) from
791 being a mulhi to being a mul lo (cheaper). Testcase:
793 void bar(unsigned n) {
798 I think this basically amounts to a dag combine to simplify comparisons against
799 multiply hi's into a comparison against the mullo.
801 //===---------------------------------------------------------------------===//
803 Better mod/ref analysis for scanf would allow us to eliminate the vtable and a
804 bunch of other stuff from this example (see PR1604):
814 std::scanf("%d", &t.val);
815 std::printf("%d\n", t.val);
818 //===---------------------------------------------------------------------===//
820 Instcombine will merge comparisons like (x >= 10) && (x < 20) by producing (x -
821 10) u< 10, but only when the comparisons have matching sign.
823 This could be converted with a similiar technique. (PR1941)
825 define i1 @test(i8 %x) {
826 %A = icmp uge i8 %x, 5
827 %B = icmp slt i8 %x, 20
832 //===---------------------------------------------------------------------===//
834 These functions perform the same computation, but produce different assembly.
836 define i8 @select(i8 %x) readnone nounwind {
837 %A = icmp ult i8 %x, 250
838 %B = select i1 %A, i8 0, i8 1
842 define i8 @addshr(i8 %x) readnone nounwind {
843 %A = zext i8 %x to i9
844 %B = add i9 %A, 6 ;; 256 - 250 == 6
846 %D = trunc i9 %C to i8
850 //===---------------------------------------------------------------------===//
854 f (unsigned long a, unsigned long b, unsigned long c)
856 return ((a & (c - 1)) != 0) || ((b & (c - 1)) != 0);
859 f (unsigned long a, unsigned long b, unsigned long c)
861 return ((a & (c - 1)) != 0) | ((b & (c - 1)) != 0);
863 Both should combine to ((a|b) & (c-1)) != 0. Currently not optimized with
864 "clang -emit-llvm-bc | opt -std-compile-opts".
866 //===---------------------------------------------------------------------===//
869 #define PMD_MASK (~((1UL << 23) - 1))
870 void clear_pmd_range(unsigned long start, unsigned long end)
872 if (!(start & ~PMD_MASK) && !(end & ~PMD_MASK))
875 The expression should optimize to something like
876 "!((start|end)&~PMD_MASK). Currently not optimized with "clang
877 -emit-llvm-bc | opt -std-compile-opts".
879 //===---------------------------------------------------------------------===//
883 foo (unsigned int a, unsigned int b)
885 if (a <= 7 && b <= 7)
888 Should combine to "(a|b) <= 7". Currently not optimized with "clang
889 -emit-llvm-bc | opt -std-compile-opts".
891 //===---------------------------------------------------------------------===//
897 return (n >= 0 ? 1 : -1);
899 Should combine to (n >> 31) | 1. Currently not optimized with "clang
900 -emit-llvm-bc | opt -std-compile-opts | llc".
902 //===---------------------------------------------------------------------===//
905 int test(int a, int b)
912 Should combine to "a <= b". Currently not optimized with "clang
913 -emit-llvm-bc | opt -std-compile-opts | llc".
915 //===---------------------------------------------------------------------===//
919 if (variable == 4 || variable == 6)
922 This should optimize to "if ((variable | 2) == 6)". Currently not
923 optimized with "clang -emit-llvm-bc | opt -std-compile-opts | llc".
925 //===---------------------------------------------------------------------===//
927 unsigned int f(unsigned int i, unsigned int n) {++i; if (i == n) ++i; return
929 unsigned int f2(unsigned int i, unsigned int n) {++i; i += i == n; return i;}
930 These should combine to the same thing. Currently, the first function
931 produces better code on X86.
933 //===---------------------------------------------------------------------===//
936 #define abs(x) x>0?x:-x
939 return (abs(x)) >= 0;
941 This should optimize to x == INT_MIN. (With -fwrapv.) Currently not
942 optimized with "clang -emit-llvm-bc | opt -std-compile-opts".
944 //===---------------------------------------------------------------------===//
948 rotate_cst (unsigned int a)
950 a = (a << 10) | (a >> 22);
955 minus_cst (unsigned int a)
964 mask_gt (unsigned int a)
966 /* This is equivalent to a > 15. */
971 rshift_gt (unsigned int a)
973 /* This is equivalent to a > 23. */
977 All should simplify to a single comparison. All of these are
978 currently not optimized with "clang -emit-llvm-bc | opt
981 //===---------------------------------------------------------------------===//
984 int c(int* x) {return (char*)x+2 == (char*)x;}
985 Should combine to 0. Currently not optimized with "clang
986 -emit-llvm-bc | opt -std-compile-opts" (although llc can optimize it).
988 //===---------------------------------------------------------------------===//
990 int a(unsigned char* b) {return *b > 99;}
991 There's an unnecessary zext in the generated code with "clang
992 -emit-llvm-bc | opt -std-compile-opts".
994 //===---------------------------------------------------------------------===//
996 int a(unsigned b) {return ((b << 31) | (b << 30)) >> 31;}
997 Should be combined to "((b >> 1) | b) & 1". Currently not optimized
998 with "clang -emit-llvm-bc | opt -std-compile-opts".
1000 //===---------------------------------------------------------------------===//
1002 unsigned a(unsigned x, unsigned y) { return x | (y & 1) | (y & 2);}
1003 Should combine to "x | (y & 3)". Currently not optimized with "clang
1004 -emit-llvm-bc | opt -std-compile-opts".
1006 //===---------------------------------------------------------------------===//
1008 unsigned a(unsigned a) {return ((a | 1) & 3) | (a & -4);}
1009 Should combine to "a | 1". Currently not optimized with "clang
1010 -emit-llvm-bc | opt -std-compile-opts".
1012 //===---------------------------------------------------------------------===//
1014 int a(int a, int b, int c) {return (~a & c) | ((c|a) & b);}
1015 Should fold to "(~a & c) | (a & b)". Currently not optimized with
1016 "clang -emit-llvm-bc | opt -std-compile-opts".
1018 //===---------------------------------------------------------------------===//
1020 int a(int a,int b) {return (~(a|b))|a;}
1021 Should fold to "a|~b". Currently not optimized with "clang
1022 -emit-llvm-bc | opt -std-compile-opts".
1024 //===---------------------------------------------------------------------===//
1026 int a(int a, int b) {return (a&&b) || (a&&!b);}
1027 Should fold to "a". Currently not optimized with "clang -emit-llvm-bc
1028 | opt -std-compile-opts".
1030 //===---------------------------------------------------------------------===//
1032 int a(int a, int b, int c) {return (a&&b) || (!a&&c);}
1033 Should fold to "a ? b : c", or at least something sane. Currently not
1034 optimized with "clang -emit-llvm-bc | opt -std-compile-opts".
1036 //===---------------------------------------------------------------------===//
1038 int a(int a, int b, int c) {return (a&&b) || (a&&c) || (a&&b&&c);}
1039 Should fold to a && (b || c). Currently not optimized with "clang
1040 -emit-llvm-bc | opt -std-compile-opts".
1042 //===---------------------------------------------------------------------===//
1044 int a(int x) {return x | ((x & 8) ^ 8);}
1045 Should combine to x | 8. Currently not optimized with "clang
1046 -emit-llvm-bc | opt -std-compile-opts".
1048 //===---------------------------------------------------------------------===//
1050 int a(int x) {return x ^ ((x & 8) ^ 8);}
1051 Should also combine to x | 8. Currently not optimized with "clang
1052 -emit-llvm-bc | opt -std-compile-opts".
1054 //===---------------------------------------------------------------------===//
1056 int a(int x) {return (x & 8) == 0 ? -1 : -9;}
1057 Should combine to (x | -9) ^ 8. Currently not optimized with "clang
1058 -emit-llvm-bc | opt -std-compile-opts".
1060 //===---------------------------------------------------------------------===//
1062 int a(int x) {return (x & 8) == 0 ? -9 : -1;}
1063 Should combine to x | -9. Currently not optimized with "clang
1064 -emit-llvm-bc | opt -std-compile-opts".
1066 //===---------------------------------------------------------------------===//
1068 int a(int x) {return ((x | -9) ^ 8) & x;}
1069 Should combine to x & -9. Currently not optimized with "clang
1070 -emit-llvm-bc | opt -std-compile-opts".
1072 //===---------------------------------------------------------------------===//
1074 unsigned a(unsigned a) {return a * 0x11111111 >> 28 & 1;}
1075 Should combine to "a * 0x88888888 >> 31". Currently not optimized
1076 with "clang -emit-llvm-bc | opt -std-compile-opts".
1078 //===---------------------------------------------------------------------===//
1080 unsigned a(char* x) {if ((*x & 32) == 0) return b();}
1081 There's an unnecessary zext in the generated code with "clang
1082 -emit-llvm-bc | opt -std-compile-opts".
1084 //===---------------------------------------------------------------------===//
1086 unsigned a(unsigned long long x) {return 40 * (x >> 1);}
1087 Should combine to "20 * (((unsigned)x) & -2)". Currently not
1088 optimized with "clang -emit-llvm-bc | opt -std-compile-opts".
1090 //===---------------------------------------------------------------------===//
1092 This was noticed in the entryblock for grokdeclarator in 403.gcc:
1094 %tmp = icmp eq i32 %decl_context, 4
1095 %decl_context_addr.0 = select i1 %tmp, i32 3, i32 %decl_context
1096 %tmp1 = icmp eq i32 %decl_context_addr.0, 1
1097 %decl_context_addr.1 = select i1 %tmp1, i32 0, i32 %decl_context_addr.0
1099 tmp1 should be simplified to something like:
1100 (!tmp || decl_context == 1)
1102 This allows recursive simplifications, tmp1 is used all over the place in
1103 the function, e.g. by:
1105 %tmp23 = icmp eq i32 %decl_context_addr.1, 0 ; <i1> [#uses=1]
1106 %tmp24 = xor i1 %tmp1, true ; <i1> [#uses=1]
1107 %or.cond8 = and i1 %tmp23, %tmp24 ; <i1> [#uses=1]
1111 //===---------------------------------------------------------------------===//
1113 Store sinking: This code:
1115 void f (int n, int *cond, int *res) {
1118 for (i = 0; i < n; i++)
1120 *res ^= 234; /* (*) */
1123 On this function GVN hoists the fully redundant value of *res, but nothing
1124 moves the store out. This gives us this code:
1126 bb: ; preds = %bb2, %entry
1127 %.rle = phi i32 [ 0, %entry ], [ %.rle6, %bb2 ]
1128 %i.05 = phi i32 [ 0, %entry ], [ %indvar.next, %bb2 ]
1129 %1 = load i32* %cond, align 4
1130 %2 = icmp eq i32 %1, 0
1131 br i1 %2, label %bb2, label %bb1
1134 %3 = xor i32 %.rle, 234
1135 store i32 %3, i32* %res, align 4
1138 bb2: ; preds = %bb, %bb1
1139 %.rle6 = phi i32 [ %3, %bb1 ], [ %.rle, %bb ]
1140 %indvar.next = add i32 %i.05, 1
1141 %exitcond = icmp eq i32 %indvar.next, %n
1142 br i1 %exitcond, label %return, label %bb
1144 DSE should sink partially dead stores to get the store out of the loop.
1146 Here's another partial dead case:
1147 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12395
1149 //===---------------------------------------------------------------------===//
1151 Scalar PRE hoists the mul in the common block up to the else:
1153 int test (int a, int b, int c, int g) {
1163 It would be better to do the mul once to reduce codesize above the if.
1164 This is GCC PR38204.
1166 //===---------------------------------------------------------------------===//
1168 GCC PR37810 is an interesting case where we should sink load/store reload
1169 into the if block and outside the loop, so we don't reload/store it on the
1190 We now hoist the reload after the call (Transforms/GVN/lpre-call-wrap.ll), but
1191 we don't sink the store. We need partially dead store sinking.
1193 //===---------------------------------------------------------------------===//
1195 [PHI TRANSLATE GEPs]
1197 GCC PR37166: Sinking of loads prevents SROA'ing the "g" struct on the stack
1198 leading to excess stack traffic. This could be handled by GVN with some crazy
1199 symbolic phi translation. The code we get looks like (g is on the stack):
1203 %9 = getelementptr %struct.f* %g, i32 0, i32 0
1204 store i32 %8, i32* %9, align bel %bb3
1206 bb3: ; preds = %bb1, %bb2, %bb
1207 %c_addr.0 = phi %struct.f* [ %g, %bb2 ], [ %c, %bb ], [ %c, %bb1 ]
1208 %b_addr.0 = phi %struct.f* [ %b, %bb2 ], [ %g, %bb ], [ %b, %bb1 ]
1209 %10 = getelementptr %struct.f* %c_addr.0, i32 0, i32 0
1210 %11 = load i32* %10, align 4
1212 %11 is fully redundant, an in BB2 it should have the value %8.
1214 GCC PR33344 is a similar case.
1216 //===---------------------------------------------------------------------===//
1218 [PHI TRANSLATE INDEXED GEPs] PR5313
1220 Load redundancy elimination for simple loop. This loop:
1222 void append_text(const char* text,unsigned char * const io) {
1227 Compiles to have a fully redundant load in the loop (%2):
1229 define void @append_text(i8* nocapture %text, i8* nocapture %io) nounwind {
1231 %0 = load i8* %text, align 1 ; <i8> [#uses=1]
1232 %1 = icmp eq i8 %0, 0 ; <i1> [#uses=1]
1233 br i1 %1, label %return, label %bb
1235 bb: ; preds = %bb, %entry
1236 %indvar = phi i32 [ 0, %entry ], [ %tmp, %bb ] ; <i32> [#uses=2]
1237 %text_addr.04 = getelementptr i8* %text, i32 %indvar ; <i8*> [#uses=1]
1238 %2 = load i8* %text_addr.04, align 1 ; <i8> [#uses=1]
1239 store i8 %2, i8* %io, align 1
1240 %tmp = add i32 %indvar, 1 ; <i32> [#uses=2]
1241 %scevgep = getelementptr i8* %text, i32 %tmp ; <i8*> [#uses=1]
1242 %3 = load i8* %scevgep, align 1 ; <i8> [#uses=1]
1243 %4 = icmp eq i8 %3, 0 ; <i1> [#uses=1]
1244 br i1 %4, label %return, label %bb
1246 return: ; preds = %bb, %entry
1250 //===---------------------------------------------------------------------===//
1252 There are many load PRE testcases in testsuite/gcc.dg/tree-ssa/loadpre* in the
1253 GCC testsuite. There are many pre testcases as ssa-pre-*.c
1255 //===---------------------------------------------------------------------===//
1257 There are some interesting cases in testsuite/gcc.dg/tree-ssa/pred-comm* in the
1258 GCC testsuite. For example, predcom-1.c is:
1260 for (i = 2; i < 1000; i++)
1261 fib[i] = (fib[i-1] + fib[i - 2]) & 0xffff;
1263 which compiles into:
1265 bb1: ; preds = %bb1, %bb1.thread
1266 %indvar = phi i32 [ 0, %bb1.thread ], [ %0, %bb1 ]
1267 %i.0.reg2mem.0 = add i32 %indvar, 2
1268 %0 = add i32 %indvar, 1 ; <i32> [#uses=3]
1269 %1 = getelementptr [1000 x i32]* @fib, i32 0, i32 %0
1270 %2 = load i32* %1, align 4 ; <i32> [#uses=1]
1271 %3 = getelementptr [1000 x i32]* @fib, i32 0, i32 %indvar
1272 %4 = load i32* %3, align 4 ; <i32> [#uses=1]
1273 %5 = add i32 %4, %2 ; <i32> [#uses=1]
1274 %6 = and i32 %5, 65535 ; <i32> [#uses=1]
1275 %7 = getelementptr [1000 x i32]* @fib, i32 0, i32 %i.0.reg2mem.0
1276 store i32 %6, i32* %7, align 4
1277 %exitcond = icmp eq i32 %0, 998 ; <i1> [#uses=1]
1278 br i1 %exitcond, label %return, label %bb1
1285 instead of handling this as a loop or other xform, all we'd need to do is teach
1286 load PRE to phi translate the %0 add (i+1) into the predecessor as (i'+1+1) =
1287 (i'+2) (where i' is the previous iteration of i). This would find the store
1290 predcom-2.c is apparently the same as predcom-1.c
1291 predcom-3.c is very similar but needs loads feeding each other instead of
1293 predcom-4.c seems the same as the rest.
1296 //===---------------------------------------------------------------------===//
1298 Other simple load PRE cases:
1299 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35287 [LPRE crit edge splitting]
1301 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34677 (licm does this, LPRE crit edge)
1302 llvm-gcc t2.c -S -o - -O0 -emit-llvm | llvm-as | opt -mem2reg -simplifycfg -gvn | llvm-dis
1304 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16799 [BITCAST PHI TRANS]
1306 //===---------------------------------------------------------------------===//
1308 Type based alias analysis:
1309 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14705
1311 //===---------------------------------------------------------------------===//
1313 A/B get pinned to the stack because we turn an if/then into a select instead
1314 of PRE'ing the load/store. This may be fixable in instcombine:
1315 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37892
1317 struct X { int i; };
1331 //===---------------------------------------------------------------------===//
1333 Interesting missed case because of control flow flattening (should be 2 loads):
1334 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26629
1335 With: llvm-gcc t2.c -S -o - -O0 -emit-llvm | llvm-as |
1336 opt -mem2reg -gvn -instcombine | llvm-dis
1337 we miss it because we need 1) GEP PHI TRAN, 2) CRIT EDGE 3) MULTIPLE DIFFERENT
1338 VALS PRODUCED BY ONE BLOCK OVER DIFFERENT PATHS
1340 //===---------------------------------------------------------------------===//
1342 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19633
1343 We could eliminate the branch condition here, loading from null is undefined:
1345 struct S { int w, x, y, z; };
1346 struct T { int r; struct S s; };
1347 void bar (struct S, int);
1348 void foo (int a, struct T b)
1356 //===---------------------------------------------------------------------===//
1358 simplifylibcalls should do several optimizations for strspn/strcspn:
1360 strcspn(x, "") -> strlen(x)
1363 strspn(x, "") -> strlen(x)
1364 strspn(x, "a") -> strchr(x, 'a')-x
1366 strcspn(x, "a") -> inlined loop for up to 3 letters (similarly for strspn):
1368 size_t __strcspn_c3 (__const char *__s, int __reject1, int __reject2,
1370 register size_t __result = 0;
1371 while (__s[__result] != '\0' && __s[__result] != __reject1 &&
1372 __s[__result] != __reject2 && __s[__result] != __reject3)
1377 This should turn into a switch on the character. See PR3253 for some notes on
1380 456.hmmer apparently uses strcspn and strspn a lot. 471.omnetpp uses strspn.
1382 //===---------------------------------------------------------------------===//
1384 "gas" uses this idiom:
1385 else if (strchr ("+-/*%|&^:[]()~", *intel_parser.op_string))
1387 else if (strchr ("<>", *intel_parser.op_string)
1389 Those should be turned into a switch.
1391 //===---------------------------------------------------------------------===//
1393 252.eon contains this interesting code:
1395 %3072 = getelementptr [100 x i8]* %tempString, i32 0, i32 0
1396 %3073 = call i8* @strcpy(i8* %3072, i8* %3071) nounwind
1397 %strlen = call i32 @strlen(i8* %3072) ; uses = 1
1398 %endptr = getelementptr [100 x i8]* %tempString, i32 0, i32 %strlen
1399 call void @llvm.memcpy.i32(i8* %endptr,
1400 i8* getelementptr ([5 x i8]* @"\01LC42", i32 0, i32 0), i32 5, i32 1)
1401 %3074 = call i32 @strlen(i8* %endptr) nounwind readonly
1403 This is interesting for a couple reasons. First, in this:
1405 %3073 = call i8* @strcpy(i8* %3072, i8* %3071) nounwind
1406 %strlen = call i32 @strlen(i8* %3072)
1408 The strlen could be replaced with: %strlen = sub %3072, %3073, because the
1409 strcpy call returns a pointer to the end of the string. Based on that, the
1410 endptr GEP just becomes equal to 3073, which eliminates a strlen call and GEP.
1412 Second, the memcpy+strlen strlen can be replaced with:
1414 %3074 = call i32 @strlen([5 x i8]* @"\01LC42") nounwind readonly
1416 Because the destination was just copied into the specified memory buffer. This,
1417 in turn, can be constant folded to "4".
1419 In other code, it contains:
1421 %endptr6978 = bitcast i8* %endptr69 to i32*
1422 store i32 7107374, i32* %endptr6978, align 1
1423 %3167 = call i32 @strlen(i8* %endptr69) nounwind readonly
1425 Which could also be constant folded. Whatever is producing this should probably
1426 be fixed to leave this as a memcpy from a string.
1428 Further, eon also has an interesting partially redundant strlen call:
1430 bb8: ; preds = %_ZN18eonImageCalculatorC1Ev.exit
1431 %682 = getelementptr i8** %argv, i32 6 ; <i8**> [#uses=2]
1432 %683 = load i8** %682, align 4 ; <i8*> [#uses=4]
1433 %684 = load i8* %683, align 1 ; <i8> [#uses=1]
1434 %685 = icmp eq i8 %684, 0 ; <i1> [#uses=1]
1435 br i1 %685, label %bb10, label %bb9
1438 %686 = call i32 @strlen(i8* %683) nounwind readonly
1439 %687 = icmp ugt i32 %686, 254 ; <i1> [#uses=1]
1440 br i1 %687, label %bb10, label %bb11
1442 bb10: ; preds = %bb9, %bb8
1443 %688 = call i32 @strlen(i8* %683) nounwind readonly
1445 This could be eliminated by doing the strlen once in bb8, saving code size and
1446 improving perf on the bb8->9->10 path.
1448 //===---------------------------------------------------------------------===//
1450 I see an interesting fully redundant call to strlen left in 186.crafty:InputMove
1452 %movetext11 = getelementptr [128 x i8]* %movetext, i32 0, i32 0
1455 bb62: ; preds = %bb55, %bb53
1456 %promote.0 = phi i32 [ %169, %bb55 ], [ 0, %bb53 ]
1457 %171 = call i32 @strlen(i8* %movetext11) nounwind readonly align 1
1458 %172 = add i32 %171, -1 ; <i32> [#uses=1]
1459 %173 = getelementptr [128 x i8]* %movetext, i32 0, i32 %172
1462 br i1 %or.cond, label %bb65, label %bb72
1464 bb65: ; preds = %bb62
1465 store i8 0, i8* %173, align 1
1468 bb72: ; preds = %bb65, %bb62
1469 %trank.1 = phi i32 [ %176, %bb65 ], [ -1, %bb62 ]
1470 %177 = call i32 @strlen(i8* %movetext11) nounwind readonly align 1
1472 Note that on the bb62->bb72 path, that the %177 strlen call is partially
1473 redundant with the %171 call. At worst, we could shove the %177 strlen call
1474 up into the bb65 block moving it out of the bb62->bb72 path. However, note
1475 that bb65 stores to the string, zeroing out the last byte. This means that on
1476 that path the value of %177 is actually just %171-1. A sub is cheaper than a
1479 This pattern repeats several times, basically doing:
1484 where it is "obvious" that B = A-1.
1486 //===---------------------------------------------------------------------===//
1488 186.crafty contains this interesting pattern:
1490 %77 = call i8* @strstr(i8* getelementptr ([6 x i8]* @"\01LC5", i32 0, i32 0),
1492 %phitmp648 = icmp eq i8* %77, getelementptr ([6 x i8]* @"\01LC5", i32 0, i32 0)
1493 br i1 %phitmp648, label %bb70, label %bb76
1495 bb70: ; preds = %OptionMatch.exit91, %bb69
1496 %78 = call i32 @strlen(i8* %30) nounwind readonly align 1 ; <i32> [#uses=1]
1500 if (strstr(cststr, P) == cststr) {
1504 The strstr call would be significantly cheaper written as:
1507 if (memcmp(P, str, strlen(P)))
1510 This is memcmp+strlen instead of strstr. This also makes the strlen fully
1513 //===---------------------------------------------------------------------===//
1515 186.crafty also contains this code:
1517 %1906 = call i32 @strlen(i8* getelementptr ([32 x i8]* @pgn_event, i32 0,i32 0))
1518 %1907 = getelementptr [32 x i8]* @pgn_event, i32 0, i32 %1906
1519 %1908 = call i8* @strcpy(i8* %1907, i8* %1905) nounwind align 1
1520 %1909 = call i32 @strlen(i8* getelementptr ([32 x i8]* @pgn_event, i32 0,i32 0))
1521 %1910 = getelementptr [32 x i8]* @pgn_event, i32 0, i32 %1909
1523 The last strlen is computable as 1908-@pgn_event, which means 1910=1908.
1525 //===---------------------------------------------------------------------===//
1527 186.crafty has this interesting pattern with the "out.4543" variable:
1529 call void @llvm.memcpy.i32(
1530 i8* getelementptr ([10 x i8]* @out.4543, i32 0, i32 0),
1531 i8* getelementptr ([7 x i8]* @"\01LC28700", i32 0, i32 0), i32 7, i32 1)
1532 %101 = call@printf(i8* ... @out.4543, i32 0, i32 0)) nounwind
1534 It is basically doing:
1536 memcpy(globalarray, "string");
1537 printf(..., globalarray);
1539 Anyway, by knowing that printf just reads the memory and forward substituting
1540 the string directly into the printf, this eliminates reads from globalarray.
1541 Since this pattern occurs frequently in crafty (due to the "DisplayTime" and
1542 other similar functions) there are many stores to "out". Once all the printfs
1543 stop using "out", all that is left is the memcpy's into it. This should allow
1544 globalopt to remove the "stored only" global.
1546 //===---------------------------------------------------------------------===//
1550 define inreg i32 @foo(i8* inreg %p) nounwind {
1552 %tmp1 = ashr i8 %tmp0, 5
1553 %tmp2 = sext i8 %tmp1 to i32
1557 could be dagcombine'd to a sign-extending load with a shift.
1558 For example, on x86 this currently gets this:
1564 while it could get this:
1569 //===---------------------------------------------------------------------===//
1573 int test(int x) { return 1-x == x; } // --> return false
1574 int test2(int x) { return 2-x == x; } // --> return x == 1 ?
1576 Always foldable for odd constants, what is the rule for even?
1578 //===---------------------------------------------------------------------===//
1580 PR 3381: GEP to field of size 0 inside a struct could be turned into GEP
1581 for next field in struct (which is at same address).
1583 For example: store of float into { {{}}, float } could be turned into a store to
1586 //===---------------------------------------------------------------------===//
1589 double foo(double a) { return sin(a); }
1591 This compiles into this on x86-64 Linux:
1602 //===---------------------------------------------------------------------===//
1604 The arg promotion pass should make use of nocapture to make its alias analysis
1605 stuff much more precise.
1607 //===---------------------------------------------------------------------===//
1609 The following functions should be optimized to use a select instead of a
1610 branch (from gcc PR40072):
1612 char char_int(int m) {if(m>7) return 0; return m;}
1613 int int_char(char m) {if(m>7) return 0; return m;}
1615 //===---------------------------------------------------------------------===//
1617 int func(int a, int b) { if (a & 0x80) b |= 0x80; else b &= ~0x80; return b; }
1621 define i32 @func(i32 %a, i32 %b) nounwind readnone ssp {
1623 %0 = and i32 %a, 128 ; <i32> [#uses=1]
1624 %1 = icmp eq i32 %0, 0 ; <i1> [#uses=1]
1625 %2 = or i32 %b, 128 ; <i32> [#uses=1]
1626 %3 = and i32 %b, -129 ; <i32> [#uses=1]
1627 %b_addr.0 = select i1 %1, i32 %3, i32 %2 ; <i32> [#uses=1]
1631 However, it's functionally equivalent to:
1633 b = (b & ~0x80) | (a & 0x80);
1635 Which generates this:
1637 define i32 @func(i32 %a, i32 %b) nounwind readnone ssp {
1639 %0 = and i32 %b, -129 ; <i32> [#uses=1]
1640 %1 = and i32 %a, 128 ; <i32> [#uses=1]
1641 %2 = or i32 %0, %1 ; <i32> [#uses=1]
1645 This can be generalized for other forms:
1647 b = (b & ~0x80) | (a & 0x40) << 1;
1649 //===---------------------------------------------------------------------===//
1651 These two functions produce different code. They shouldn't:
1655 uint8_t p1(uint8_t b, uint8_t a) {
1656 b = (b & ~0xc0) | (a & 0xc0);
1660 uint8_t p2(uint8_t b, uint8_t a) {
1661 b = (b & ~0x40) | (a & 0x40);
1662 b = (b & ~0x80) | (a & 0x80);
1666 define zeroext i8 @p1(i8 zeroext %b, i8 zeroext %a) nounwind readnone ssp {
1668 %0 = and i8 %b, 63 ; <i8> [#uses=1]
1669 %1 = and i8 %a, -64 ; <i8> [#uses=1]
1670 %2 = or i8 %1, %0 ; <i8> [#uses=1]
1674 define zeroext i8 @p2(i8 zeroext %b, i8 zeroext %a) nounwind readnone ssp {
1676 %0 = and i8 %b, 63 ; <i8> [#uses=1]
1677 %.masked = and i8 %a, 64 ; <i8> [#uses=1]
1678 %1 = and i8 %a, -128 ; <i8> [#uses=1]
1679 %2 = or i8 %1, %0 ; <i8> [#uses=1]
1680 %3 = or i8 %2, %.masked ; <i8> [#uses=1]
1684 //===---------------------------------------------------------------------===//
1686 IPSCCP does not currently propagate argument dependent constants through
1687 functions where it does not not all of the callers. This includes functions
1688 with normal external linkage as well as templates, C99 inline functions etc.
1689 Specifically, it does nothing to:
1691 define i32 @test(i32 %x, i32 %y, i32 %z) nounwind {
1693 %0 = add nsw i32 %y, %z
1696 %3 = add nsw i32 %1, %2
1700 define i32 @test2() nounwind {
1702 %0 = call i32 @test(i32 1, i32 2, i32 4) nounwind
1706 It would be interesting extend IPSCCP to be able to handle simple cases like
1707 this, where all of the arguments to a call are constant. Because IPSCCP runs
1708 before inlining, trivial templates and inline functions are not yet inlined.
1709 The results for a function + set of constant arguments should be memoized in a
1712 //===---------------------------------------------------------------------===//