6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-12-13 20:32:21 +00:00
Code Issues Projects Releases Wiki Activity
release_35
llvm-6502/test/Analysis/ScalarEvolution/trip-count3.ll

81 lines
3.4 KiB
LLVM
Raw Permalink Normal View History

FileCheckize some of the testcases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@187756 91177308-0d34-0410-b5e6-96231b3b80d8
2013-08-05 23:43:18 +00:00
; RUN: opt < %s -scalar-evolution -analyze | FileCheck %s
Fix ScalarEvolution's backedge-taken count computations to check for overflow when computing a integer division to round up. Thanks to Nick Lewycky for noticing this! git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@73862 91177308-0d34-0410-b5e6-96231b3b80d8
2009-06-21 23:46:38 +00:00
; ScalarEvolution can't compute a trip count because it doesn't know if
; dividing by the stride will have a remainder. This could theoretically
; be teaching it how to use a more elaborate trip count computation.
Extend ScalarEvolution's getBackedgeTakenCount to be able to compute an upper-bound value for the trip count, in addition to the actual trip count. Use this to allow getZeroExtendExpr and getSignExtendExpr to fold casts in more cases. This may eventually morph into a more general value-range analysis capability; there are certainly plenty of places where more complete value-range information would allow more folding. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70509 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-30 20:47:05 +00:00
Rewrite SCEV's backedge taken count computation. Patch by Michele Scandale! Rewrite of the functions used to compute the backedge taken count of a loop on LT and GT comparisons. I decided to split the handling of LT and GT cases becasue the trick "a > b == -a < -b" in some cases prevents the trip count computation due to the multiplication by -1 on the two operands of the comparison. This issue comes from the conservative computation of value range of SCEVs: taking the negative SCEV of an expression that have a small positive range (e.g. [0,31]), we would have a SCEV with a fullset as value range. Indeed, in the new rewritten function I tried to better handle the maximum backedge taken count computation when MAX/MIN expression are used to handle the cases where no entry guard is found. Some test have been modified in order to check the new value correctly (I manually check them and reasoning on possible overflow the new values seem correct). I finally added a new test case related to the multiplication by -1 issue on GT comparisons. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@194116 91177308-0d34-0410-b5e6-96231b3b80d8
2013-11-06 02:08:26 +00:00
; CHECK: Loop %bb3.i: backedge-taken count is ((64 + (-64 smax (-1 + (-1 * %0))) + %0) /u 64)
; CHECK: Loop %bb3.i: max backedge-taken count is 33554431
FileCheckize some of the testcases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@187756 91177308-0d34-0410-b5e6-96231b3b80d8
2013-08-05 23:43:18 +00:00
rip out a ton of intrinsic modernization logic from AutoUpgrade.cpp, which is for pre-2.9 bitcode files. We keep x86 unaligned loads, movnt, crc32, and the target indep prefetch change. As usual, updating the testsuite is a PITA. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@133337 91177308-0d34-0410-b5e6-96231b3b80d8
2011-06-18 06:05:24 +00:00
%struct.FILE = type { i32, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, %struct._IO_marker*, %struct.FILE*, i32, i32, i64, i16, i8, [1 x i8], i8*, i64, i8*, i8*, i8*, i8*, i64, i32, [20 x i8] }
%struct.SHA_INFO = type { [5 x i32], i32, i32, [16 x i32] }
%struct._IO_marker = type { %struct._IO_marker*, %struct.FILE*, i32 }
@_2E_str = external constant [26 x i8]
@stdin = external global %struct.FILE*
@_2E_str1 = external constant [3 x i8]
@_2E_str12 = external constant [30 x i8]
Extend ScalarEvolution's getBackedgeTakenCount to be able to compute an upper-bound value for the trip count, in addition to the actual trip count. Use this to allow getZeroExtendExpr and getSignExtendExpr to fold casts in more cases. This may eventually morph into a more general value-range analysis capability; there are certainly plenty of places where more complete value-range information would allow more folding. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70509 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-30 20:47:05 +00:00
declare void @sha_init(%struct.SHA_INFO* nocapture) nounwind
declare fastcc void @sha_transform(%struct.SHA_INFO* nocapture) nounwind
declare void @sha_print(%struct.SHA_INFO* nocapture) nounwind
declare i32 @printf(i8* nocapture, ...) nounwind
declare void @sha_final(%struct.SHA_INFO* nocapture) nounwind
declare void @sha_update(%struct.SHA_INFO* nocapture, i8* nocapture, i32) nounwind
declare i64 @fread(i8* noalias nocapture, i64, i64, %struct.FILE* noalias nocapture) nounwind
declare i32 @main(i32, i8** nocapture) nounwind
declare noalias %struct.FILE* @fopen(i8* noalias nocapture, i8* noalias nocapture) nounwind
declare i32 @fclose(%struct.FILE* nocapture) nounwind
declare void @sha_stream(%struct.SHA_INFO* nocapture, %struct.FILE* nocapture) nounwind
define void @sha_stream_bb3_2E_i(%struct.SHA_INFO* %sha_info, i8* %data1, i32, i8** %buffer_addr.0.i.out, i32* %count_addr.0.i.out) nounwind {
newFuncRoot:
rip out a ton of intrinsic modernization logic from AutoUpgrade.cpp, which is for pre-2.9 bitcode files. We keep x86 unaligned loads, movnt, crc32, and the target indep prefetch change. As usual, updating the testsuite is a PITA. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@133337 91177308-0d34-0410-b5e6-96231b3b80d8
2011-06-18 06:05:24 +00:00
br label %bb3.i
sha_update.exit.exitStub: ; preds = %bb3.i
store i8* %buffer_addr.0.i, i8** %buffer_addr.0.i.out
store i32 %count_addr.0.i, i32* %count_addr.0.i.out
ret void
bb2.i: ; preds = %bb3.i
%1 = getelementptr %struct.SHA_INFO* %sha_info, i64 0, i32 3
%2 = bitcast [16 x i32]* %1 to i8*
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %2, i8* %buffer_addr.0.i, i64 64, i32 1, i1 false)
%3 = getelementptr %struct.SHA_INFO* %sha_info, i64 0, i32 3, i64 0
%4 = bitcast i32* %3 to i8*
br label %codeRepl
codeRepl: ; preds = %bb2.i
call void @sha_stream_bb3_2E_i_bb1_2E_i_2E_i(i8* %4)
br label %byte_reverse.exit.i
byte_reverse.exit.i: ; preds = %codeRepl
call fastcc void @sha_transform(%struct.SHA_INFO* %sha_info) nounwind
%5 = getelementptr i8* %buffer_addr.0.i, i64 64
%6 = add i32 %count_addr.0.i, -64
br label %bb3.i
bb3.i: ; preds = %byte_reverse.exit.i, %newFuncRoot
%buffer_addr.0.i = phi i8* [ %data1, %newFuncRoot ], [ %5, %byte_reverse.exit.i ]
%count_addr.0.i = phi i32 [ %0, %newFuncRoot ], [ %6, %byte_reverse.exit.i ]
%7 = icmp sgt i32 %count_addr.0.i, 63
br i1 %7, label %bb2.i, label %sha_update.exit.exitStub
Extend ScalarEvolution's getBackedgeTakenCount to be able to compute an upper-bound value for the trip count, in addition to the actual trip count. Use this to allow getZeroExtendExpr and getSignExtendExpr to fold casts in more cases. This may eventually morph into a more general value-range analysis capability; there are certainly plenty of places where more complete value-range information would allow more folding. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@70509 91177308-0d34-0410-b5e6-96231b3b80d8
2009-04-30 20:47:05 +00:00
}
declare void @sha_stream_bb3_2E_i_bb1_2E_i_2E_i(i8*) nounwind
rip out a ton of intrinsic modernization logic from AutoUpgrade.cpp, which is for pre-2.9 bitcode files. We keep x86 unaligned loads, movnt, crc32, and the target indep prefetch change. As usual, updating the testsuite is a PITA. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@133337 91177308-0d34-0410-b5e6-96231b3b80d8
2011-06-18 06:05:24 +00:00
declare void @llvm.memset.p0i8.i64(i8* nocapture, i8, i64, i32, i1) nounwind
declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i32, i1) nounwind
Reference in New Issue Copy Permalink