1
Target Independent Opportunities:
3
//===---------------------------------------------------------------------===//
5
Dead argument elimination should be enhanced to handle cases when an argument is
6
dead to an externally visible function. Though the argument can't be removed
7
from the externally visible function, the caller doesn't need to pass it in.
8
For example in this testcase:
10
void foo(int X) __attribute__((noinline));
11
void foo(int X) { sideeffect(); }
12
void bar(int A) { foo(A+1); }
16
define void @bar(i32 %A) nounwind ssp {
17
%0 = add nsw i32 %A, 1 ; <i32> [#uses=1]
18
tail call void @foo(i32 %0) nounwind noinline ssp
22
The add is dead, we could pass in 'i32 undef' instead. This occurs for C++
23
templates etc, which usually have linkonce_odr/weak_odr linkage, not internal
26
//===---------------------------------------------------------------------===//
28
With the recent changes to make the implicit def/use set explicit in
29
machineinstrs, we should change the target descriptions for 'call' instructions
30
so that the .td files don't list all the call-clobbered registers as implicit
31
defs. Instead, these should be added by the code generator (e.g. on the dag).
33
This has a number of uses:
35
1. PPC32/64 and X86 32/64 can avoid having multiple copies of call instructions
36
for their different impdef sets.
37
2. Targets with multiple calling convs (e.g. x86) which have different clobber
38
sets don't need copies of call instructions.
39
3. 'Interprocedural register allocation' can be done to reduce the clobber sets
42
//===---------------------------------------------------------------------===//
44
Make the PPC branch selector target independant
46
//===---------------------------------------------------------------------===//
48
Get the C front-end to expand hypot(x,y) -> llvm.sqrt(x*x+y*y) when errno and
49
precision don't matter (ffastmath). Misc/mandel will like this. :) This isn't
50
safe in general, even on darwin. See the libm implementation of hypot for
51
examples (which special case when x/y are exactly zero to get signed zeros etc
54
//===---------------------------------------------------------------------===//
56
Solve this DAG isel folding deficiency:
74
The problem is the store's chain operand is not the load X but rather
75
a TokenFactor of the load X and load Y, which prevents the folding.
77
There are two ways to fix this:
79
1. The dag combiner can start using alias analysis to realize that y/x
80
don't alias, making the store to X not dependent on the load from Y.
81
2. The generated isel could be made smarter in the case it can't
82
disambiguate the pointers.
84
Number 1 is the preferred solution.
86
This has been "fixed" by a TableGen hack. But that is a short term workaround
87
which will be removed once the proper fix is made.
89
//===---------------------------------------------------------------------===//
91
On targets with expensive 64-bit multiply, we could LSR this:
98
for (i = ...; ++i, tmp+=tmp)
101
This would be a win on ppc32, but not x86 or ppc64.
103
//===---------------------------------------------------------------------===//
105
Shrink: (setlt (loadi32 P), 0) -> (setlt (loadi8 Phi), 0)
107
//===---------------------------------------------------------------------===//
109
Reassociate should turn things like:
111
int factorial(int X) {
112
return X*X*X*X*X*X*X*X;
115
into llvm.powi calls, allowing the code generator to produce balanced
116
multiplication trees.
118
First, the intrinsic needs to be extended to support integers, and second the
119
code generator needs to be enhanced to lower these to multiplication trees.
121
//===---------------------------------------------------------------------===//
123
Interesting? testcase for add/shift/mul reassoc:
125
int bar(int x, int y) {
126
return x*x*x+y+x*x*x*x*x*y*y*y*y;
128
int foo(int z, int n) {
129
return bar(z, n) + bar(2*z, 2*n);
132
This is blocked on not handling X*X*X -> powi(X, 3) (see note above). The issue
133
is that we end up getting t = 2*X s = t*t and don't turn this into 4*X*X,
134
which is the same number of multiplies and is canonical, because the 2*X has
135
multiple uses. Here's a simple example:
137
define i32 @test15(i32 %X1) {
138
%B = mul i32 %X1, 47 ; X1*47
144
//===---------------------------------------------------------------------===//
146
Reassociate should handle the example in GCC PR16157:
148
extern int a0, a1, a2, a3, a4; extern int b0, b1, b2, b3, b4;
149
void f () { /* this can be optimized to four additions... */
150
b4 = a4 + a3 + a2 + a1 + a0;
151
b3 = a3 + a2 + a1 + a0;
156
This requires reassociating to forms of expressions that are already available,
157
something that reassoc doesn't think about yet.
160
//===---------------------------------------------------------------------===//
162
This function: (derived from GCC PR19988)
163
double foo(double x, double y) {
164
return ((x + 0.1234 * y) * (x + -0.1234 * y));
170
mulsd LCPI1_1(%rip), %xmm1
171
mulsd LCPI1_0(%rip), %xmm2
178
Reassociate should be able to turn it into:
180
double foo(double x, double y) {
181
return ((x + 0.1234 * y) * (x - 0.1234 * y));
184
Which allows the multiply by constant to be CSE'd, producing:
187
mulsd LCPI1_0(%rip), %xmm1
194
This doesn't need -ffast-math support at all. This is particularly bad because
195
the llvm-gcc frontend is canonicalizing the later into the former, but clang
196
doesn't have this problem.
198
//===---------------------------------------------------------------------===//
200
These two functions should generate the same code on big-endian systems:
202
int g(int *j,int *l) { return memcmp(j,l,4); }
203
int h(int *j, int *l) { return *j - *l; }
205
this could be done in SelectionDAGISel.cpp, along with other special cases,
208
//===---------------------------------------------------------------------===//
210
It would be nice to revert this patch:
211
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20060213/031986.html
213
And teach the dag combiner enough to simplify the code expanded before
214
legalize. It seems plausible that this knowledge would let it simplify other
217
//===---------------------------------------------------------------------===//
219
For vector types, TargetData.cpp::getTypeInfo() returns alignment that is equal
220
to the type size. It works but can be overly conservative as the alignment of
221
specific vector types are target dependent.
223
//===---------------------------------------------------------------------===//
225
We should produce an unaligned load from code like this:
227
v4sf example(float *P) {
228
return (v4sf){P[0], P[1], P[2], P[3] };
231
//===---------------------------------------------------------------------===//
233
Add support for conditional increments, and other related patterns. Instead
238
je LBB16_2 #cond_next
249
//===---------------------------------------------------------------------===//
251
Combine: a = sin(x), b = cos(x) into a,b = sincos(x).
253
Expand these to calls of sin/cos and stores:
254
double sincos(double x, double *sin, double *cos);
255
float sincosf(float x, float *sin, float *cos);
256
long double sincosl(long double x, long double *sin, long double *cos);
258
Doing so could allow SROA of the destination pointers. See also:
259
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17687
261
This is now easily doable with MRVs. We could even make an intrinsic for this
262
if anyone cared enough about sincos.
264
//===---------------------------------------------------------------------===//
266
quantum_sigma_x in 462.libquantum contains the following loop:
268
for(i=0; i<reg->size; i++)
270
/* Flip the target bit of each basis state */
271
reg->node[i].state ^= ((MAX_UNSIGNED) 1 << target);
274
Where MAX_UNSIGNED/state is a 64-bit int. On a 32-bit platform it would be just
275
so cool to turn it into something like:
277
long long Res = ((MAX_UNSIGNED) 1 << target);
279
for(i=0; i<reg->size; i++)
280
reg->node[i].state ^= Res & 0xFFFFFFFFULL;
282
for(i=0; i<reg->size; i++)
283
reg->node[i].state ^= Res & 0xFFFFFFFF00000000ULL
286
... which would only do one 32-bit XOR per loop iteration instead of two.
288
It would also be nice to recognize the reg->size doesn't alias reg->node[i], but
291
//===---------------------------------------------------------------------===//
293
This isn't recognized as bswap by instcombine (yes, it really is bswap):
295
unsigned long reverse(unsigned v) {
297
t = v ^ ((v << 16) | (v >> 16));
299
v = (v << 24) | (v >> 8);
303
Neither is this (very standard idiom):
307
return (((n) << 24) | (((n) & 0xff00) << 8)
308
| (((n) >> 8) & 0xff00) | ((n) >> 24));
311
//===---------------------------------------------------------------------===//
315
These idioms should be recognized as popcount (see PR1488):
317
unsigned countbits_slow(unsigned v) {
319
for (c = 0; v; v >>= 1)
323
unsigned countbits_fast(unsigned v){
326
v &= v - 1; // clear the least significant bit set
330
BITBOARD = unsigned long long
331
int PopCnt(register BITBOARD a) {
339
unsigned int popcount(unsigned int input) {
340
unsigned int count = 0;
341
for (unsigned int i = 0; i < 4 * 8; i++)
342
count += (input >> i) & i;
346
This is a form of idiom recognition for loops, the same thing that could be
347
useful for recognizing memset/memcpy.
349
//===---------------------------------------------------------------------===//
351
These should turn into single 16-bit (unaligned?) loads on little/big endian
354
unsigned short read_16_le(const unsigned char *adr) {
355
return adr[0] | (adr[1] << 8);
357
unsigned short read_16_be(const unsigned char *adr) {
358
return (adr[0] << 8) | adr[1];
361
//===---------------------------------------------------------------------===//
363
-instcombine should handle this transform:
364
icmp pred (sdiv X / C1 ), C2
365
when X, C1, and C2 are unsigned. Similarly for udiv and signed operands.
367
Currently InstCombine avoids this transform but will do it when the signs of
368
the operands and the sign of the divide match. See the FIXME in
369
InstructionCombining.cpp in the visitSetCondInst method after the switch case
370
for Instruction::UDiv (around line 4447) for more details.
372
The SingleSource/Benchmarks/Shootout-C++/hash and hash2 tests have examples of
375
//===---------------------------------------------------------------------===//
379
viterbi speeds up *significantly* if the various "history" related copy loops
380
are turned into memcpy calls at the source level. We need a "loops to memcpy"
383
//===---------------------------------------------------------------------===//
387
SingleSource/Benchmarks/Misc/dt.c shows several interesting optimization
388
opportunities in its double_array_divs_variable function: it needs loop
389
interchange, memory promotion (which LICM already does), vectorization and
390
variable trip count loop unrolling (since it has a constant trip count). ICC
391
apparently produces this very nice code with -ffast-math:
393
..B1.70: # Preds ..B1.70 ..B1.69
394
mulpd %xmm0, %xmm1 #108.2
395
mulpd %xmm0, %xmm1 #108.2
396
mulpd %xmm0, %xmm1 #108.2
397
mulpd %xmm0, %xmm1 #108.2
399
cmpl $131072, %edx #108.2
400
jb ..B1.70 # Prob 99% #108.2
402
It would be better to count down to zero, but this is a lot better than what we
405
//===---------------------------------------------------------------------===//
409
typedef unsigned U32;
410
typedef unsigned long long U64;
411
int test (U32 *inst, U64 *regs) {
414
int r1 = (temp >> 20) & 0xf;
415
int b2 = (temp >> 16) & 0xf;
416
effective_addr2 = temp & 0xfff;
417
if (b2) effective_addr2 += regs[b2];
418
b2 = (temp >> 12) & 0xf;
419
if (b2) effective_addr2 += regs[b2];
420
effective_addr2 &= regs[4];
421
if ((effective_addr2 & 3) == 0)
426
Note that only the low 2 bits of effective_addr2 are used. On 32-bit systems,
427
we don't eliminate the computation of the top half of effective_addr2 because
428
we don't have whole-function selection dags. On x86, this means we use one
429
extra register for the function when effective_addr2 is declared as U64 than
430
when it is declared U32.
432
PHI Slicing could be extended to do this.
434
//===---------------------------------------------------------------------===//
436
LSR should know what GPR types a target has from TargetData. This code:
438
volatile short X, Y; // globals
442
for (i = 0; i < N; i++) { X = i; Y = i*4; }
445
produces two near identical IV's (after promotion) on PPC/ARM:
455
add r2, r2, #1 <- [0,+,1]
456
sub r0, r0, #1 <- [0,-,1]
460
LSR should reuse the "+" IV for the exit test.
462
//===---------------------------------------------------------------------===//
464
Tail call elim should be more aggressive, checking to see if the call is
465
followed by an uncond branch to an exit block.
467
; This testcase is due to tail-duplication not wanting to copy the return
468
; instruction into the terminating blocks because there was other code
469
; optimized out of the function after the taildup happened.
470
; RUN: llvm-as < %s | opt -tailcallelim | llvm-dis | not grep call
472
define i32 @t4(i32 %a) {
474
%tmp.1 = and i32 %a, 1 ; <i32> [#uses=1]
475
%tmp.2 = icmp ne i32 %tmp.1, 0 ; <i1> [#uses=1]
476
br i1 %tmp.2, label %then.0, label %else.0
478
then.0: ; preds = %entry
479
%tmp.5 = add i32 %a, -1 ; <i32> [#uses=1]
480
%tmp.3 = call i32 @t4( i32 %tmp.5 ) ; <i32> [#uses=1]
483
else.0: ; preds = %entry
484
%tmp.7 = icmp ne i32 %a, 0 ; <i1> [#uses=1]
485
br i1 %tmp.7, label %then.1, label %return
487
then.1: ; preds = %else.0
488
%tmp.11 = add i32 %a, -2 ; <i32> [#uses=1]
489
%tmp.9 = call i32 @t4( i32 %tmp.11 ) ; <i32> [#uses=1]
492
return: ; preds = %then.1, %else.0, %then.0
493
%result.0 = phi i32 [ 0, %else.0 ], [ %tmp.3, %then.0 ],
498
//===---------------------------------------------------------------------===//
500
Tail recursion elimination should handle:
505
return 2 * pow2m1 (n - 1) + 1;
508
Also, multiplies can be turned into SHL's, so they should be handled as if
509
they were associative. "return foo() << 1" can be tail recursion eliminated.
511
//===---------------------------------------------------------------------===//
513
Argument promotion should promote arguments for recursive functions, like
516
; RUN: llvm-as < %s | opt -argpromotion | llvm-dis | grep x.val
518
define internal i32 @foo(i32* %x) {
520
%tmp = load i32* %x ; <i32> [#uses=0]
521
%tmp.foo = call i32 @foo( i32* %x ) ; <i32> [#uses=1]
525
define i32 @bar(i32* %x) {
527
%tmp3 = call i32 @foo( i32* %x ) ; <i32> [#uses=1]
531
//===---------------------------------------------------------------------===//
533
We should investigate an instruction sinking pass. Consider this silly
549
je LBB1_2 # cond_true
557
The PIC base computation (call+popl) is only used on one path through the
558
code, but is currently always computed in the entry block. It would be
559
better to sink the picbase computation down into the block for the
560
assertion, as it is the only one that uses it. This happens for a lot of
561
code with early outs.
563
Another example is loads of arguments, which are usually emitted into the
564
entry block on targets like x86. If not used in all paths through a
565
function, they should be sunk into the ones that do.
567
In this case, whole-function-isel would also handle this.
569
//===---------------------------------------------------------------------===//
571
Investigate lowering of sparse switch statements into perfect hash tables:
572
http://burtleburtle.net/bob/hash/perfect.html
574
//===---------------------------------------------------------------------===//
576
We should turn things like "load+fabs+store" and "load+fneg+store" into the
577
corresponding integer operations. On a yonah, this loop:
582
for (b = 0; b < 10000000; b++)
583
for (i = 0; i < 256; i++)
587
is twice as slow as this loop:
592
for (b = 0; b < 10000000; b++)
593
for (i = 0; i < 256; i++)
594
a[i] ^= (1ULL << 63);
597
and I suspect other processors are similar. On X86 in particular this is a
598
big win because doing this with integers allows the use of read/modify/write
601
//===---------------------------------------------------------------------===//
603
DAG Combiner should try to combine small loads into larger loads when
604
profitable. For example, we compile this C++ example:
606
struct THotKey { short Key; bool Control; bool Shift; bool Alt; };
607
extern THotKey m_HotKey;
608
THotKey GetHotKey () { return m_HotKey; }
610
into (-O3 -fno-exceptions -static -fomit-frame-pointer):
615
movb _m_HotKey+3, %cl
616
movb _m_HotKey+4, %dl
617
movb _m_HotKey+2, %ch
632
movzwl _m_HotKey+4, %edx
636
The LLVM IR contains the needed alignment info, so we should be able to
637
merge the loads and stores into 4-byte loads:
639
%struct.THotKey = type { i16, i8, i8, i8 }
640
define void @_Z9GetHotKeyv(%struct.THotKey* sret %agg.result) nounwind {
642
%tmp2 = load i16* getelementptr (@m_HotKey, i32 0, i32 0), align 8
643
%tmp5 = load i8* getelementptr (@m_HotKey, i32 0, i32 1), align 2
644
%tmp8 = load i8* getelementptr (@m_HotKey, i32 0, i32 2), align 1
645
%tmp11 = load i8* getelementptr (@m_HotKey, i32 0, i32 3), align 2
647
Alternatively, we should use a small amount of base-offset alias analysis
648
to make it so the scheduler doesn't need to hold all the loads in regs at
651
//===---------------------------------------------------------------------===//
653
We should add an FRINT node to the DAG to model targets that have legal
654
implementations of ceil/floor/rint.
656
//===---------------------------------------------------------------------===//
661
long long input[8] = {1,1,1,1,1,1,1,1};
665
We currently compile this into a memcpy from a global array since the
666
initializer is fairly large and not memset'able. This is good, but the memcpy
667
gets lowered to load/stores in the code generator. This is also ok, except
668
that the codegen lowering for memcpy doesn't handle the case when the source
669
is a constant global. This gives us atrocious code like this:
674
movl _C.0.1444-"L1$pb"+32(%eax), %ecx
676
movl _C.0.1444-"L1$pb"+20(%eax), %ecx
678
movl _C.0.1444-"L1$pb"+36(%eax), %ecx
680
movl _C.0.1444-"L1$pb"+44(%eax), %ecx
682
movl _C.0.1444-"L1$pb"+40(%eax), %ecx
684
movl _C.0.1444-"L1$pb"+12(%eax), %ecx
686
movl _C.0.1444-"L1$pb"+4(%eax), %ecx
698
//===---------------------------------------------------------------------===//
700
http://llvm.org/PR717:
702
The following code should compile into "ret int undef". Instead, LLVM
703
produces "ret int 0":
712
//===---------------------------------------------------------------------===//
714
The loop unroller should partially unroll loops (instead of peeling them)
715
when code growth isn't too bad and when an unroll count allows simplification
716
of some code within the loop. One trivial example is:
722
for ( nLoop = 0; nLoop < 1000; nLoop++ ) {
731
Unrolling by 2 would eliminate the '&1' in both copies, leading to a net
732
reduction in code size. The resultant code would then also be suitable for
733
exit value computation.
735
//===---------------------------------------------------------------------===//
737
We miss a bunch of rotate opportunities on various targets, including ppc, x86,
738
etc. On X86, we miss a bunch of 'rotate by variable' cases because the rotate
739
matching code in dag combine doesn't look through truncates aggressively
740
enough. Here are some testcases reduces from GCC PR17886:
742
unsigned long long f(unsigned long long x, int y) {
743
return (x << y) | (x >> 64-y);
745
unsigned f2(unsigned x, int y){
746
return (x << y) | (x >> 32-y);
748
unsigned long long f3(unsigned long long x){
750
return (x << y) | (x >> 64-y);
752
unsigned f4(unsigned x){
754
return (x << y) | (x >> 32-y);
756
unsigned long long f5(unsigned long long x, unsigned long long y) {
757
return (x << 8) | ((y >> 48) & 0xffull);
759
unsigned long long f6(unsigned long long x, unsigned long long y, int z) {
762
return (x << 8) | ((y >> 48) & 0xffull);
764
return (x << 16) | ((y >> 40) & 0xffffull);
766
return (x << 24) | ((y >> 32) & 0xffffffull);
768
return (x << 32) | ((y >> 24) & 0xffffffffull);
770
return (x << 40) | ((y >> 16) & 0xffffffffffull);
774
On X86-64, we only handle f2/f3/f4 right. On x86-32, a few of these
775
generate truly horrible code, instead of using shld and friends. On
776
ARM, we end up with calls to L___lshrdi3/L___ashldi3 in f, which is
777
badness. PPC64 misses f, f5 and f6. CellSPU aborts in isel.
779
//===---------------------------------------------------------------------===//
781
We do a number of simplifications in simplify libcalls to strength reduce
782
standard library functions, but we don't currently merge them together. For
783
example, it is useful to merge memcpy(a,b,strlen(b)) -> strcpy. This can only
784
be done safely if "b" isn't modified between the strlen and memcpy of course.
786
//===---------------------------------------------------------------------===//
788
We compile this program: (from GCC PR11680)
789
http://gcc.gnu.org/bugzilla/attachment.cgi?id=4487
791
Into code that runs the same speed in fast/slow modes, but both modes run 2x
792
slower than when compile with GCC (either 4.0 or 4.2):
794
$ llvm-g++ perf.cpp -O3 -fno-exceptions
796
1.821u 0.003s 0:01.82 100.0% 0+0k 0+0io 0pf+0w
798
$ g++ perf.cpp -O3 -fno-exceptions
800
0.821u 0.001s 0:00.82 100.0% 0+0k 0+0io 0pf+0w
802
It looks like we are making the same inlining decisions, so this may be raw
803
codegen badness or something else (haven't investigated).
805
//===---------------------------------------------------------------------===//
807
We miss some instcombines for stuff like this:
809
void foo (unsigned int a) {
810
/* This one is equivalent to a >= (3 << 2). */
815
A few other related ones are in GCC PR14753.
817
//===---------------------------------------------------------------------===//
819
Divisibility by constant can be simplified (according to GCC PR12849) from
820
being a mulhi to being a mul lo (cheaper). Testcase:
822
void bar(unsigned n) {
827
This is equivalent to the following, where 2863311531 is the multiplicative
828
inverse of 3, and 1431655766 is ((2^32)-1)/3+1:
829
void bar(unsigned n) {
830
if (n * 2863311531U < 1431655766U)
834
The same transformation can work with an even modulo with the addition of a
835
rotate: rotate the result of the multiply to the right by the number of bits
836
which need to be zero for the condition to be true, and shrink the compare RHS
837
by the same amount. Unless the target supports rotates, though, that
838
transformation probably isn't worthwhile.
840
The transformation can also easily be made to work with non-zero equality
841
comparisons: just transform, for example, "n % 3 == 1" to "(n-1) % 3 == 0".
843
//===---------------------------------------------------------------------===//
845
Better mod/ref analysis for scanf would allow us to eliminate the vtable and a
846
bunch of other stuff from this example (see PR1604):
856
std::scanf("%d", &t.val);
857
std::printf("%d\n", t.val);
860
//===---------------------------------------------------------------------===//
862
These functions perform the same computation, but produce different assembly.
864
define i8 @select(i8 %x) readnone nounwind {
865
%A = icmp ult i8 %x, 250
866
%B = select i1 %A, i8 0, i8 1
870
define i8 @addshr(i8 %x) readnone nounwind {
871
%A = zext i8 %x to i9
872
%B = add i9 %A, 6 ;; 256 - 250 == 6
874
%D = trunc i9 %C to i8
878
//===---------------------------------------------------------------------===//
882
f (unsigned long a, unsigned long b, unsigned long c)
884
return ((a & (c - 1)) != 0) || ((b & (c - 1)) != 0);
887
f (unsigned long a, unsigned long b, unsigned long c)
889
return ((a & (c - 1)) != 0) | ((b & (c - 1)) != 0);
891
Both should combine to ((a|b) & (c-1)) != 0. Currently not optimized with
892
"clang -emit-llvm-bc | opt -std-compile-opts".
894
//===---------------------------------------------------------------------===//
897
#define PMD_MASK (~((1UL << 23) - 1))
898
void clear_pmd_range(unsigned long start, unsigned long end)
900
if (!(start & ~PMD_MASK) && !(end & ~PMD_MASK))
903
The expression should optimize to something like
904
"!((start|end)&~PMD_MASK). Currently not optimized with "clang
905
-emit-llvm-bc | opt -std-compile-opts".
907
//===---------------------------------------------------------------------===//
911
if (variable == 4 || variable == 6)
914
This should optimize to "if ((variable | 2) == 6)". Currently not
915
optimized with "clang -emit-llvm-bc | opt -std-compile-opts | llc".
917
//===---------------------------------------------------------------------===//
919
unsigned int f(unsigned int i, unsigned int n) {++i; if (i == n) ++i; return
921
unsigned int f2(unsigned int i, unsigned int n) {++i; i += i == n; return i;}
922
These should combine to the same thing. Currently, the first function
923
produces better code on X86.
925
//===---------------------------------------------------------------------===//
928
#define abs(x) x>0?x:-x
931
return (abs(x)) >= 0;
933
This should optimize to x == INT_MIN. (With -fwrapv.) Currently not
934
optimized with "clang -emit-llvm-bc | opt -std-compile-opts".
936
//===---------------------------------------------------------------------===//
940
rotate_cst (unsigned int a)
942
a = (a << 10) | (a >> 22);
947
minus_cst (unsigned int a)
956
mask_gt (unsigned int a)
958
/* This is equivalent to a > 15. */
963
rshift_gt (unsigned int a)
965
/* This is equivalent to a > 23. */
969
All should simplify to a single comparison. All of these are
970
currently not optimized with "clang -emit-llvm-bc | opt
973
//===---------------------------------------------------------------------===//
976
int c(int* x) {return (char*)x+2 == (char*)x;}
977
Should combine to 0. Currently not optimized with "clang
978
-emit-llvm-bc | opt -std-compile-opts" (although llc can optimize it).
980
//===---------------------------------------------------------------------===//
982
int a(unsigned b) {return ((b << 31) | (b << 30)) >> 31;}
983
Should be combined to "((b >> 1) | b) & 1". Currently not optimized
984
with "clang -emit-llvm-bc | opt -std-compile-opts".
986
//===---------------------------------------------------------------------===//
988
unsigned a(unsigned x, unsigned y) { return x | (y & 1) | (y & 2);}
989
Should combine to "x | (y & 3)". Currently not optimized with "clang
990
-emit-llvm-bc | opt -std-compile-opts".
992
//===---------------------------------------------------------------------===//
994
int a(int a, int b, int c) {return (~a & c) | ((c|a) & b);}
995
Should fold to "(~a & c) | (a & b)". Currently not optimized with
996
"clang -emit-llvm-bc | opt -std-compile-opts".
998
//===---------------------------------------------------------------------===//
1000
int a(int a,int b) {return (~(a|b))|a;}
1001
Should fold to "a|~b". Currently not optimized with "clang
1002
-emit-llvm-bc | opt -std-compile-opts".
1004
//===---------------------------------------------------------------------===//
1006
int a(int a, int b) {return (a&&b) || (a&&!b);}
1007
Should fold to "a". Currently not optimized with "clang -emit-llvm-bc
1008
| opt -std-compile-opts".
1010
//===---------------------------------------------------------------------===//
1012
int a(int a, int b, int c) {return (a&&b) || (!a&&c);}
1013
Should fold to "a ? b : c", or at least something sane. Currently not
1014
optimized with "clang -emit-llvm-bc | opt -std-compile-opts".
1016
//===---------------------------------------------------------------------===//
1018
int a(int a, int b, int c) {return (a&&b) || (a&&c) || (a&&b&&c);}
1019
Should fold to a && (b || c). Currently not optimized with "clang
1020
-emit-llvm-bc | opt -std-compile-opts".
1022
//===---------------------------------------------------------------------===//
1024
int a(int x) {return x | ((x & 8) ^ 8);}
1025
Should combine to x | 8. Currently not optimized with "clang
1026
-emit-llvm-bc | opt -std-compile-opts".
1028
//===---------------------------------------------------------------------===//
1030
int a(int x) {return x ^ ((x & 8) ^ 8);}
1031
Should also combine to x | 8. Currently not optimized with "clang
1032
-emit-llvm-bc | opt -std-compile-opts".
1034
//===---------------------------------------------------------------------===//
1036
int a(int x) {return (x & 8) == 0 ? -1 : -9;}
1037
Should combine to (x | -9) ^ 8. Currently not optimized with "clang
1038
-emit-llvm-bc | opt -std-compile-opts".
1040
//===---------------------------------------------------------------------===//
1042
int a(int x) {return (x & 8) == 0 ? -9 : -1;}
1043
Should combine to x | -9. Currently not optimized with "clang
1044
-emit-llvm-bc | opt -std-compile-opts".
1046
//===---------------------------------------------------------------------===//
1048
int a(int x) {return ((x | -9) ^ 8) & x;}
1049
Should combine to x & -9. Currently not optimized with "clang
1050
-emit-llvm-bc | opt -std-compile-opts".
1052
//===---------------------------------------------------------------------===//
1054
unsigned a(unsigned a) {return a * 0x11111111 >> 28 & 1;}
1055
Should combine to "a * 0x88888888 >> 31". Currently not optimized
1056
with "clang -emit-llvm-bc | opt -std-compile-opts".
1058
//===---------------------------------------------------------------------===//
1060
unsigned a(char* x) {if ((*x & 32) == 0) return b();}
1061
There's an unnecessary zext in the generated code with "clang
1062
-emit-llvm-bc | opt -std-compile-opts".
1064
//===---------------------------------------------------------------------===//
1066
unsigned a(unsigned long long x) {return 40 * (x >> 1);}
1067
Should combine to "20 * (((unsigned)x) & -2)". Currently not
1068
optimized with "clang -emit-llvm-bc | opt -std-compile-opts".
1070
//===---------------------------------------------------------------------===//
1072
This was noticed in the entryblock for grokdeclarator in 403.gcc:
1074
%tmp = icmp eq i32 %decl_context, 4
1075
%decl_context_addr.0 = select i1 %tmp, i32 3, i32 %decl_context
1076
%tmp1 = icmp eq i32 %decl_context_addr.0, 1
1077
%decl_context_addr.1 = select i1 %tmp1, i32 0, i32 %decl_context_addr.0
1079
tmp1 should be simplified to something like:
1080
(!tmp || decl_context == 1)
1082
This allows recursive simplifications, tmp1 is used all over the place in
1083
the function, e.g. by:
1085
%tmp23 = icmp eq i32 %decl_context_addr.1, 0 ; <i1> [#uses=1]
1086
%tmp24 = xor i1 %tmp1, true ; <i1> [#uses=1]
1087
%or.cond8 = and i1 %tmp23, %tmp24 ; <i1> [#uses=1]
1091
//===---------------------------------------------------------------------===//
1095
Store sinking: This code:
1097
void f (int n, int *cond, int *res) {
1100
for (i = 0; i < n; i++)
1102
*res ^= 234; /* (*) */
1105
On this function GVN hoists the fully redundant value of *res, but nothing
1106
moves the store out. This gives us this code:
1108
bb: ; preds = %bb2, %entry
1109
%.rle = phi i32 [ 0, %entry ], [ %.rle6, %bb2 ]
1110
%i.05 = phi i32 [ 0, %entry ], [ %indvar.next, %bb2 ]
1111
%1 = load i32* %cond, align 4
1112
%2 = icmp eq i32 %1, 0
1113
br i1 %2, label %bb2, label %bb1
1116
%3 = xor i32 %.rle, 234
1117
store i32 %3, i32* %res, align 4
1120
bb2: ; preds = %bb, %bb1
1121
%.rle6 = phi i32 [ %3, %bb1 ], [ %.rle, %bb ]
1122
%indvar.next = add i32 %i.05, 1
1123
%exitcond = icmp eq i32 %indvar.next, %n
1124
br i1 %exitcond, label %return, label %bb
1126
DSE should sink partially dead stores to get the store out of the loop.
1128
Here's another partial dead case:
1129
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12395
1131
//===---------------------------------------------------------------------===//
1133
Scalar PRE hoists the mul in the common block up to the else:
1135
int test (int a, int b, int c, int g) {
1145
It would be better to do the mul once to reduce codesize above the if.
1146
This is GCC PR38204.
1148
//===---------------------------------------------------------------------===//
1152
GCC PR37810 is an interesting case where we should sink load/store reload
1153
into the if block and outside the loop, so we don't reload/store it on the
1174
We now hoist the reload after the call (Transforms/GVN/lpre-call-wrap.ll), but
1175
we don't sink the store. We need partially dead store sinking.
1177
//===---------------------------------------------------------------------===//
1179
[LOAD PRE CRIT EDGE SPLITTING]
1181
GCC PR37166: Sinking of loads prevents SROA'ing the "g" struct on the stack
1182
leading to excess stack traffic. This could be handled by GVN with some crazy
1183
symbolic phi translation. The code we get looks like (g is on the stack):
1187
%9 = getelementptr %struct.f* %g, i32 0, i32 0
1188
store i32 %8, i32* %9, align bel %bb3
1190
bb3: ; preds = %bb1, %bb2, %bb
1191
%c_addr.0 = phi %struct.f* [ %g, %bb2 ], [ %c, %bb ], [ %c, %bb1 ]
1192
%b_addr.0 = phi %struct.f* [ %b, %bb2 ], [ %g, %bb ], [ %b, %bb1 ]
1193
%10 = getelementptr %struct.f* %c_addr.0, i32 0, i32 0
1194
%11 = load i32* %10, align 4
1196
%11 is partially redundant, an in BB2 it should have the value %8.
1198
GCC PR33344 and PR35287 are similar cases.
1201
//===---------------------------------------------------------------------===//
1205
There are many load PRE testcases in testsuite/gcc.dg/tree-ssa/loadpre* in the
1206
GCC testsuite, ones we don't get yet are (checked through loadpre25):
1208
[CRIT EDGE BREAKING]
1209
loadpre3.c predcom-4.c
1211
[PRE OF READONLY CALL]
1214
[TURN SELECT INTO BRANCH]
1215
loadpre14.c loadpre15.c
1217
actually a conditional increment: loadpre18.c loadpre19.c
1220
//===---------------------------------------------------------------------===//
1223
There are many PRE testcases in testsuite/gcc.dg/tree-ssa/ssa-pre-*.c in the
1226
//===---------------------------------------------------------------------===//
1228
There are some interesting cases in testsuite/gcc.dg/tree-ssa/pred-comm* in the
1229
GCC testsuite. For example, we get the first example in predcom-1.c, but
1230
miss the second one:
1235
__attribute__ ((noinline))
1236
void count_averages(int n) {
1238
for (i = 1; i < n; i++)
1239
avg[i] = (((unsigned long) fib[i - 1] + fib[i] + fib[i + 1]) / 3) & 0xffff;
1242
which compiles into two loads instead of one in the loop.
1244
predcom-2.c is the same as predcom-1.c
1246
predcom-3.c is very similar but needs loads feeding each other instead of
1250
//===---------------------------------------------------------------------===//
1254
Type based alias analysis:
1255
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14705
1257
We should do better analysis of posix_memalign. At the least it should
1258
no-capture its pointer argument, at best, we should know that the out-value
1259
result doesn't point to anything (like malloc). One example of this is in
1260
SingleSource/Benchmarks/Misc/dt.c
1262
//===---------------------------------------------------------------------===//
1264
A/B get pinned to the stack because we turn an if/then into a select instead
1265
of PRE'ing the load/store. This may be fixable in instcombine:
1266
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=37892
1268
struct X { int i; };
1282
//===---------------------------------------------------------------------===//
1284
Interesting missed case because of control flow flattening (should be 2 loads):
1285
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26629
1286
With: llvm-gcc t2.c -S -o - -O0 -emit-llvm | llvm-as |
1287
opt -mem2reg -gvn -instcombine | llvm-dis
1288
we miss it because we need 1) CRIT EDGE 2) MULTIPLE DIFFERENT
1289
VALS PRODUCED BY ONE BLOCK OVER DIFFERENT PATHS
1291
//===---------------------------------------------------------------------===//
1293
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19633
1294
We could eliminate the branch condition here, loading from null is undefined:
1296
struct S { int w, x, y, z; };
1297
struct T { int r; struct S s; };
1298
void bar (struct S, int);
1299
void foo (int a, struct T b)
1307
//===---------------------------------------------------------------------===//
1309
simplifylibcalls should do several optimizations for strspn/strcspn:
1311
strcspn(x, "") -> strlen(x)
1314
strspn(x, "") -> strlen(x)
1315
strspn(x, "a") -> strchr(x, 'a')-x
1317
strcspn(x, "a") -> inlined loop for up to 3 letters (similarly for strspn):
1319
size_t __strcspn_c3 (__const char *__s, int __reject1, int __reject2,
1321
register size_t __result = 0;
1322
while (__s[__result] != '\0' && __s[__result] != __reject1 &&
1323
__s[__result] != __reject2 && __s[__result] != __reject3)
1328
This should turn into a switch on the character. See PR3253 for some notes on
1331
456.hmmer apparently uses strcspn and strspn a lot. 471.omnetpp uses strspn.
1333
//===---------------------------------------------------------------------===//
1335
"gas" uses this idiom:
1336
else if (strchr ("+-/*%|&^:[]()~", *intel_parser.op_string))
1338
else if (strchr ("<>", *intel_parser.op_string)
1340
Those should be turned into a switch.
1342
//===---------------------------------------------------------------------===//
1344
252.eon contains this interesting code:
1346
%3072 = getelementptr [100 x i8]* %tempString, i32 0, i32 0
1347
%3073 = call i8* @strcpy(i8* %3072, i8* %3071) nounwind
1348
%strlen = call i32 @strlen(i8* %3072) ; uses = 1
1349
%endptr = getelementptr [100 x i8]* %tempString, i32 0, i32 %strlen
1350
call void @llvm.memcpy.i32(i8* %endptr,
1351
i8* getelementptr ([5 x i8]* @"\01LC42", i32 0, i32 0), i32 5, i32 1)
1352
%3074 = call i32 @strlen(i8* %endptr) nounwind readonly
1354
This is interesting for a couple reasons. First, in this:
1356
%3073 = call i8* @strcpy(i8* %3072, i8* %3071) nounwind
1357
%strlen = call i32 @strlen(i8* %3072)
1359
The strlen could be replaced with: %strlen = sub %3072, %3073, because the
1360
strcpy call returns a pointer to the end of the string. Based on that, the
1361
endptr GEP just becomes equal to 3073, which eliminates a strlen call and GEP.
1363
Second, the memcpy+strlen strlen can be replaced with:
1365
%3074 = call i32 @strlen([5 x i8]* @"\01LC42") nounwind readonly
1367
Because the destination was just copied into the specified memory buffer. This,
1368
in turn, can be constant folded to "4".
1370
In other code, it contains:
1372
%endptr6978 = bitcast i8* %endptr69 to i32*
1373
store i32 7107374, i32* %endptr6978, align 1
1374
%3167 = call i32 @strlen(i8* %endptr69) nounwind readonly
1376
Which could also be constant folded. Whatever is producing this should probably
1377
be fixed to leave this as a memcpy from a string.
1379
Further, eon also has an interesting partially redundant strlen call:
1381
bb8: ; preds = %_ZN18eonImageCalculatorC1Ev.exit
1382
%682 = getelementptr i8** %argv, i32 6 ; <i8**> [#uses=2]
1383
%683 = load i8** %682, align 4 ; <i8*> [#uses=4]
1384
%684 = load i8* %683, align 1 ; <i8> [#uses=1]
1385
%685 = icmp eq i8 %684, 0 ; <i1> [#uses=1]
1386
br i1 %685, label %bb10, label %bb9
1389
%686 = call i32 @strlen(i8* %683) nounwind readonly
1390
%687 = icmp ugt i32 %686, 254 ; <i1> [#uses=1]
1391
br i1 %687, label %bb10, label %bb11
1393
bb10: ; preds = %bb9, %bb8
1394
%688 = call i32 @strlen(i8* %683) nounwind readonly
1396
This could be eliminated by doing the strlen once in bb8, saving code size and
1397
improving perf on the bb8->9->10 path.
1399
//===---------------------------------------------------------------------===//
1401
I see an interesting fully redundant call to strlen left in 186.crafty:InputMove
1403
%movetext11 = getelementptr [128 x i8]* %movetext, i32 0, i32 0
1406
bb62: ; preds = %bb55, %bb53
1407
%promote.0 = phi i32 [ %169, %bb55 ], [ 0, %bb53 ]
1408
%171 = call i32 @strlen(i8* %movetext11) nounwind readonly align 1
1409
%172 = add i32 %171, -1 ; <i32> [#uses=1]
1410
%173 = getelementptr [128 x i8]* %movetext, i32 0, i32 %172
1413
br i1 %or.cond, label %bb65, label %bb72
1415
bb65: ; preds = %bb62
1416
store i8 0, i8* %173, align 1
1419
bb72: ; preds = %bb65, %bb62
1420
%trank.1 = phi i32 [ %176, %bb65 ], [ -1, %bb62 ]
1421
%177 = call i32 @strlen(i8* %movetext11) nounwind readonly align 1
1423
Note that on the bb62->bb72 path, that the %177 strlen call is partially
1424
redundant with the %171 call. At worst, we could shove the %177 strlen call
1425
up into the bb65 block moving it out of the bb62->bb72 path. However, note
1426
that bb65 stores to the string, zeroing out the last byte. This means that on
1427
that path the value of %177 is actually just %171-1. A sub is cheaper than a
1430
This pattern repeats several times, basically doing:
1435
where it is "obvious" that B = A-1.
1437
//===---------------------------------------------------------------------===//
1439
186.crafty also contains this code:
1441
%1906 = call i32 @strlen(i8* getelementptr ([32 x i8]* @pgn_event, i32 0,i32 0))
1442
%1907 = getelementptr [32 x i8]* @pgn_event, i32 0, i32 %1906
1443
%1908 = call i8* @strcpy(i8* %1907, i8* %1905) nounwind align 1
1444
%1909 = call i32 @strlen(i8* getelementptr ([32 x i8]* @pgn_event, i32 0,i32 0))
1445
%1910 = getelementptr [32 x i8]* @pgn_event, i32 0, i32 %1909
1447
The last strlen is computable as 1908-@pgn_event, which means 1910=1908.
1449
//===---------------------------------------------------------------------===//
1451
186.crafty has this interesting pattern with the "out.4543" variable:
1453
call void @llvm.memcpy.i32(
1454
i8* getelementptr ([10 x i8]* @out.4543, i32 0, i32 0),
1455
i8* getelementptr ([7 x i8]* @"\01LC28700", i32 0, i32 0), i32 7, i32 1)
1456
%101 = call@printf(i8* ... @out.4543, i32 0, i32 0)) nounwind
1458
It is basically doing:
1460
memcpy(globalarray, "string");
1461
printf(..., globalarray);
1463
Anyway, by knowing that printf just reads the memory and forward substituting
1464
the string directly into the printf, this eliminates reads from globalarray.
1465
Since this pattern occurs frequently in crafty (due to the "DisplayTime" and
1466
other similar functions) there are many stores to "out". Once all the printfs
1467
stop using "out", all that is left is the memcpy's into it. This should allow
1468
globalopt to remove the "stored only" global.
1470
//===---------------------------------------------------------------------===//
1474
define inreg i32 @foo(i8* inreg %p) nounwind {
1476
%tmp1 = ashr i8 %tmp0, 5
1477
%tmp2 = sext i8 %tmp1 to i32
1481
could be dagcombine'd to a sign-extending load with a shift.
1482
For example, on x86 this currently gets this:
1488
while it could get this:
1493
//===---------------------------------------------------------------------===//
1497
int test(int x) { return 1-x == x; } // --> return false
1498
int test2(int x) { return 2-x == x; } // --> return x == 1 ?
1500
Always foldable for odd constants, what is the rule for even?
1502
//===---------------------------------------------------------------------===//
1504
PR 3381: GEP to field of size 0 inside a struct could be turned into GEP
1505
for next field in struct (which is at same address).
1507
For example: store of float into { {{}}, float } could be turned into a store to
1510
//===---------------------------------------------------------------------===//
1513
double foo(double a) { return sin(a); }
1515
This compiles into this on x86-64 Linux:
1526
//===---------------------------------------------------------------------===//
1528
The arg promotion pass should make use of nocapture to make its alias analysis
1529
stuff much more precise.
1531
//===---------------------------------------------------------------------===//
1533
The following functions should be optimized to use a select instead of a
1534
branch (from gcc PR40072):
1536
char char_int(int m) {if(m>7) return 0; return m;}
1537
int int_char(char m) {if(m>7) return 0; return m;}
1539
//===---------------------------------------------------------------------===//
1541
int func(int a, int b) { if (a & 0x80) b |= 0x80; else b &= ~0x80; return b; }
1545
define i32 @func(i32 %a, i32 %b) nounwind readnone ssp {
1547
%0 = and i32 %a, 128 ; <i32> [#uses=1]
1548
%1 = icmp eq i32 %0, 0 ; <i1> [#uses=1]
1549
%2 = or i32 %b, 128 ; <i32> [#uses=1]
1550
%3 = and i32 %b, -129 ; <i32> [#uses=1]
1551
%b_addr.0 = select i1 %1, i32 %3, i32 %2 ; <i32> [#uses=1]
1555
However, it's functionally equivalent to:
1557
b = (b & ~0x80) | (a & 0x80);
1559
Which generates this:
1561
define i32 @func(i32 %a, i32 %b) nounwind readnone ssp {
1563
%0 = and i32 %b, -129 ; <i32> [#uses=1]
1564
%1 = and i32 %a, 128 ; <i32> [#uses=1]
1565
%2 = or i32 %0, %1 ; <i32> [#uses=1]
1569
This can be generalized for other forms:
1571
b = (b & ~0x80) | (a & 0x40) << 1;
1573
//===---------------------------------------------------------------------===//
1575
These two functions produce different code. They shouldn't:
1579
uint8_t p1(uint8_t b, uint8_t a) {
1580
b = (b & ~0xc0) | (a & 0xc0);
1584
uint8_t p2(uint8_t b, uint8_t a) {
1585
b = (b & ~0x40) | (a & 0x40);
1586
b = (b & ~0x80) | (a & 0x80);
1590
define zeroext i8 @p1(i8 zeroext %b, i8 zeroext %a) nounwind readnone ssp {
1592
%0 = and i8 %b, 63 ; <i8> [#uses=1]
1593
%1 = and i8 %a, -64 ; <i8> [#uses=1]
1594
%2 = or i8 %1, %0 ; <i8> [#uses=1]
1598
define zeroext i8 @p2(i8 zeroext %b, i8 zeroext %a) nounwind readnone ssp {
1600
%0 = and i8 %b, 63 ; <i8> [#uses=1]
1601
%.masked = and i8 %a, 64 ; <i8> [#uses=1]
1602
%1 = and i8 %a, -128 ; <i8> [#uses=1]
1603
%2 = or i8 %1, %0 ; <i8> [#uses=1]
1604
%3 = or i8 %2, %.masked ; <i8> [#uses=1]
1608
//===---------------------------------------------------------------------===//
1610
IPSCCP does not currently propagate argument dependent constants through
1611
functions where it does not not all of the callers. This includes functions
1612
with normal external linkage as well as templates, C99 inline functions etc.
1613
Specifically, it does nothing to:
1615
define i32 @test(i32 %x, i32 %y, i32 %z) nounwind {
1617
%0 = add nsw i32 %y, %z
1620
%3 = add nsw i32 %1, %2
1624
define i32 @test2() nounwind {
1626
%0 = call i32 @test(i32 1, i32 2, i32 4) nounwind
1630
It would be interesting extend IPSCCP to be able to handle simple cases like
1631
this, where all of the arguments to a call are constant. Because IPSCCP runs
1632
before inlining, trivial templates and inline functions are not yet inlined.
1633
The results for a function + set of constant arguments should be memoized in a
1636
//===---------------------------------------------------------------------===//
1638
The libcall constant folding stuff should be moved out of SimplifyLibcalls into
1639
libanalysis' constantfolding logic. This would allow IPSCCP to be able to
1640
handle simple things like this:
1642
static int foo(const char *X) { return strlen(X); }
1643
int bar() { return foo("abcd"); }
1645
//===---------------------------------------------------------------------===//
1647
InstCombine should use SimplifyDemandedBits to remove the or instruction:
1649
define i1 @test(i8 %x, i8 %y) {
1651
%B = icmp ugt i8 %A, 3
1655
Currently instcombine calls SimplifyDemandedBits with either all bits or just
1656
the sign bit, if the comparison is obviously a sign test. In this case, we only
1657
need all but the bottom two bits from %A, and if we gave that mask to SDB it
1658
would delete the or instruction for us.
1660
//===---------------------------------------------------------------------===//
1662
functionattrs doesn't know much about memcpy/memset. This function should be
1663
marked readnone rather than readonly, since it only twiddles local memory, but
1664
functionattrs doesn't handle memset/memcpy/memmove aggressively:
1666
struct X { int *p; int *q; };
1673
p = __builtin_memcpy (&x, &y, sizeof (int *));
1677
//===---------------------------------------------------------------------===//
1679
Missed instcombine transformation:
1680
define i1 @a(i32 %x) nounwind readnone {
1682
%cmp = icmp eq i32 %x, 30
1683
%sub = add i32 %x, -30
1684
%cmp2 = icmp ugt i32 %sub, 9
1685
%or = or i1 %cmp, %cmp2
1688
This should be optimized to a single compare. Testcase derived from gcc.
1690
//===---------------------------------------------------------------------===//
1692
Missed instcombine transformation:
1694
void a(int x) { if (((1<<x)&8)==0) b(); }
1696
The shift should be optimized out. Testcase derived from gcc.
1698
//===---------------------------------------------------------------------===//
1700
Missed instcombine or reassociate transformation:
1701
int a(int a, int b) { return (a==12)&(b>47)&(b<58); }
1703
The sgt and slt should be combined into a single comparison. Testcase derived
1706
//===---------------------------------------------------------------------===//
1708
Missed instcombine transformation:
1709
define i32 @a(i32 %x) nounwind readnone {
1711
%rem = srem i32 %x, 32
1712
%shl = shl i32 1, %rem
1716
The srem can be transformed to an and because if x is negative, the shift is
1717
undefined. Testcase derived from gcc.
1719
//===---------------------------------------------------------------------===//
1721
Missed instcombine/dagcombine transformation:
1722
define i32 @a(i32 %x, i32 %y) nounwind readnone {
1724
%mul = mul i32 %y, -8
1725
%sub = sub i32 %x, %mul
1729
Should compile to something like x+y*8, but currently compiles to an
1730
inefficient result. Testcase derived from gcc.
1732
//===---------------------------------------------------------------------===//
1734
Missed instcombine/dagcombine transformation:
1735
define void @lshift_lt(i8 zeroext %a) nounwind {
1737
%conv = zext i8 %a to i32
1738
%shl = shl i32 %conv, 3
1739
%cmp = icmp ult i32 %shl, 33
1740
br i1 %cmp, label %if.then, label %if.end
1743
tail call void @bar() nounwind
1749
declare void @bar() nounwind
1751
The shift should be eliminated. Testcase derived from gcc.
1753
//===---------------------------------------------------------------------===//
1755
These compile into different code, one gets recognized as a switch and the
1756
other doesn't due to phase ordering issues (PR6212):
1758
int test1(int mainType, int subType) {
1761
else if (mainType == 9)
1763
else if (mainType == 11)
1768
int test2(int mainType, int subType) {
1778
//===---------------------------------------------------------------------===//
1780
The following test case (from PR6576):
1782
define i32 @mul(i32 %a, i32 %b) nounwind readnone {
1784
%cond1 = icmp eq i32 %b, 0 ; <i1> [#uses=1]
1785
br i1 %cond1, label %exit, label %bb.nph
1786
bb.nph: ; preds = %entry
1787
%tmp = mul i32 %b, %a ; <i32> [#uses=1]
1789
exit: ; preds = %entry
1793
could be reduced to:
1795
define i32 @mul(i32 %a, i32 %b) nounwind readnone {
1797
%tmp = mul i32 %b, %a
1801
//===---------------------------------------------------------------------===//
1803
We should use DSE + llvm.lifetime.end to delete dead vtable pointer updates.
1806
Another interesting case is that something related could be used for variables
1807
that go const after their ctor has finished. In these cases, globalopt (which
1808
can statically run the constructor) could mark the global const (so it gets put
1809
in the readonly section). A testcase would be:
1812
using namespace std;
1813
const complex<char> should_be_in_rodata (42,-42);
1814
complex<char> should_be_in_data (42,-42);
1815
complex<char> should_be_in_bss;
1817
Where we currently evaluate the ctors but the globals don't become const because
1818
the optimizer doesn't know they "become const" after the ctor is done. See
1819
GCC PR4131 for more examples.
1821
//===---------------------------------------------------------------------===//
1826
return x > 1 ? x : 1;
1829
LLVM emits a comparison with 1 instead of 0. 0 would be equivalent
1830
and cheaper on most targets.
1832
LLVM prefers comparisons with zero over non-zero in general, but in this
1833
case it choses instead to keep the max operation obvious.
1835
//===---------------------------------------------------------------------===//
1837
Take the following testcase on x86-64 (similar testcases exist for all targets
1840
define void @a(i64* nocapture %s, i64* nocapture %t, i64 %a, i64 %b,
1843
%0 = zext i64 %a to i128 ; <i128> [#uses=1]
1844
%1 = zext i64 %b to i128 ; <i128> [#uses=1]
1845
%2 = add i128 %1, %0 ; <i128> [#uses=2]
1846
%3 = zext i64 %c to i128 ; <i128> [#uses=1]
1847
%4 = shl i128 %3, 64 ; <i128> [#uses=1]
1848
%5 = add i128 %4, %2 ; <i128> [#uses=1]
1849
%6 = lshr i128 %5, 64 ; <i128> [#uses=1]
1850
%7 = trunc i128 %6 to i64 ; <i64> [#uses=1]
1851
store i64 %7, i64* %s, align 8
1852
%8 = trunc i128 %2 to i64 ; <i64> [#uses=1]
1853
store i64 %8, i64* %t, align 8
1873
The generated SelectionDAG has an ADD of an ADDE, where both operands of the
1874
ADDE are zero. Replacing one of the operands of the ADDE with the other operand
1875
of the ADD, and replacing the ADD with the ADDE, should give the desired result.
1877
(That said, we are doing a lot better than gcc on this testcase. :) )
1879
//===---------------------------------------------------------------------===//
1881
Switch lowering generates less than ideal code for the following switch:
1882
define void @a(i32 %x) nounwind {
1884
switch i32 %x, label %if.end [
1885
i32 0, label %if.then
1886
i32 1, label %if.then
1887
i32 2, label %if.then
1888
i32 3, label %if.then
1889
i32 5, label %if.then
1892
tail call void @foo() nounwind
1899
Generated code on x86-64 (other platforms give similar results):
1912
The movl+movl+btq+jb could be simplified to a cmpl+jne.
1914
Or, if we wanted to be really clever, we could simplify the whole thing to
1915
something like the following, which eliminates a branch:
1922
//===---------------------------------------------------------------------===//
1923
Given a branch where the two target blocks are identical ("ret i32 %b" in
1924
both), simplifycfg will simplify them away. But not so for a switch statement:
1926
define i32 @f(i32 %a, i32 %b) nounwind readnone {
1928
switch i32 %a, label %bb3 [
1933
bb: ; preds = %entry, %entry
1936
bb3: ; preds = %entry
1939
//===---------------------------------------------------------------------===//
added
removed
libclamav/c++/llvm/lib/Target/README.txt
