RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-28 14:32:51 +01:00
Code Issues Projects Releases Wiki Activity
fbd5dd7f96
llvm-mirror/test/CodeGen/X86/2009-06-02-RewriterBug.ll

363 lines
18 KiB
LLVM
Raw Normal View History

Eliminate more uses of llvm-as and llvm-dis. llvm-svn: 81290
2009-09-09 01:54:48 +02:00
; RUN: llc < %s -mtriple=x86_64-undermydesk-freebsd8.0 -relocation-model=pic -disable-fp-elim
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
; PR4225
define void @sha256_block1(i32* nocapture %arr, i8* nocapture %in, i64 %num) nounwind {
entry:
br i1 undef, label %while.end, label %bb.nph
bb.nph: ; preds = %entry
br label %while.body
while.body: ; preds = %for.end, %bb.nph
%indvar2787 = phi i64 [ 0, %bb.nph ], [ %indvar.next2788, %for.end ] ; <i64> [#uses=2]
%tmp2791 = mul i64 %indvar2787, 44 ; <i64> [#uses=0]
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction One of several parallel first steps to remove the target type of pointers, replacing them with a single opaque pointer type. This adds an explicit type parameter to the gep instruction so that when the first parameter becomes an opaque pointer type, the type to gep through is still available to the instructions. * This doesn't modify gep operators, only instructions (operators will be handled separately) * Textual IR changes only. Bitcode (including upgrade) and changing the in-memory representation will be in separate changes. * geps of vectors are transformed as: getelementptr <4 x float*> %x, ... ->getelementptr float, <4 x float*> %x, ... Then, once the opaque pointer type is introduced, this will ultimately look like: getelementptr float, <4 x ptr> %x with the unambiguous interpretation that it is a vector of pointers to float. * address spaces remain on the pointer, not the type: getelementptr float addrspace(1)* %x ->getelementptr float, float addrspace(1)* %x Then, eventually: getelementptr float, ptr addrspace(1) %x Importantly, the massive amount of test case churn has been automated by same crappy python code. I had to manually update a few test cases that wouldn't fit the script's model (r228970,r229196,r229197,r229198). The python script just massages stdin and writes the result to stdout, I then wrapped that in a shell script to handle replacing files, then using the usual find+xargs to migrate all the files. update.py: import fileinput import sys import re ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") def conv(match, line): if not match: return line line = match.groups()[0] if len(match.groups()[5]) == 0: line += match.groups()[2] line += match.groups()[3] line += ", " line += match.groups()[1] line += "\n" return line for line in sys.stdin: if line.find("getelementptr ") == line.find("getelementptr inbounds"): if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("): line = conv(re.match(ibrep, line), line) elif line.find("getelementptr ") != line.find("getelementptr ("): line = conv(re.match(normrep, line), line) sys.stdout.write(line) apply.sh: for name in "$@" do python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name" rm -f "$name.tmp" done The actual commands: From llvm/src: find test/ -name *.ll | xargs ./apply.sh From llvm/src/tools/clang: find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}" From llvm/src/tools/polly: find test/ -name *.ll | xargs ./apply.sh After that, check-all (with llvm, clang, clang-tools-extra, lld, compiler-rt, and polly all checked out). The extra 'rm' in the apply.sh script is due to a few files in clang's test suite using interesting unicode stuff that my python script was throwing exceptions on. None of those files needed to be migrated, so it seemed sufficient to ignore those cases. Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7636 llvm-svn: 230786
2015-02-27 20:29:02 +01:00
%ctg22996 = getelementptr i8, i8* %in, i64 0 ; <i8*> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv = zext i32 undef to i64 ; <i64> [#uses=1]
%conv11 = zext i32 undef to i64 ; <i64> [#uses=1]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%tmp18 = load i32, i32* undef ; <i32> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv19 = zext i32 %tmp18 to i64 ; <i64> [#uses=1]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%tmp30 = load i32, i32* undef ; <i32> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv31 = zext i32 %tmp30 to i64 ; <i64> [#uses=4]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%ptrincdec3065 = load i8, i8* null ; <i8> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv442709 = zext i8 %ptrincdec3065 to i64 ; <i64> [#uses=1]
%shl45 = shl i64 %conv442709, 16 ; <i64> [#uses=1]
%conv632707 = zext i8 undef to i64 ; <i64> [#uses=1]
%or = or i64 %shl45, 0 ; <i64> [#uses=1]
%or55 = or i64 %or, %conv632707 ; <i64> [#uses=1]
%or64 = or i64 %or55, 0 ; <i64> [#uses=1]
%shr85 = lshr i64 %conv31, 25 ; <i64> [#uses=0]
%add = add i64 %conv11, 1508970993 ; <i64> [#uses=1]
%add95 = add i64 %add, 0 ; <i64> [#uses=1]
%add98 = add i64 %add95, 0 ; <i64> [#uses=1]
%add99 = add i64 %add98, %or64 ; <i64> [#uses=1]
%add134 = add i64 %add99, 0 ; <i64> [#uses=4]
store i32 undef, i32* undef
%shl187 = shl i64 %add134, 21 ; <i64> [#uses=0]
%and203 = and i64 %add134, %conv31 ; <i64> [#uses=1]
%xor208 = xor i64 0, %and203 ; <i64> [#uses=1]
%add212 = add i64 0, %xor208 ; <i64> [#uses=1]
%add213 = add i64 %add212, 0 ; <i64> [#uses=1]
%add248 = add i64 %add213, 0 ; <i64> [#uses=3]
%conv2852690 = zext i8 undef to i64 ; <i64> [#uses=1]
%or277 = or i64 0, %conv2852690 ; <i64> [#uses=1]
%or286 = or i64 %or277, 0 ; <i64> [#uses=1]
%neg319 = xor i64 %add248, 4294967295 ; <i64> [#uses=1]
%and321 = and i64 %neg319, %conv31 ; <i64> [#uses=1]
%xor322 = xor i64 %and321, 0 ; <i64> [#uses=1]
%add314 = add i64 %conv, 2870763221 ; <i64> [#uses=1]
%add323 = add i64 %add314, %or286 ; <i64> [#uses=1]
%add326 = add i64 %add323, %xor322 ; <i64> [#uses=1]
%add327 = add i64 %add326, 0 ; <i64> [#uses=2]
%add362 = add i64 %add327, %conv19 ; <i64> [#uses=4]
%add365 = add i64 0, %add327 ; <i64> [#uses=3]
%shl409 = shl i64 %add362, 26 ; <i64> [#uses=0]
%and431 = and i64 %add362, %add248 ; <i64> [#uses=1]
%neg433 = xor i64 %add362, -1 ; <i64> [#uses=1]
%and435 = and i64 %add134, %neg433 ; <i64> [#uses=1]
%xor436 = xor i64 %and431, %and435 ; <i64> [#uses=1]
%add428 = add i64 %conv31, 3624381080 ; <i64> [#uses=1]
%add437 = add i64 %add428, 0 ; <i64> [#uses=1]
%add440 = add i64 %add437, %xor436 ; <i64> [#uses=1]
%add441 = add i64 %add440, 0 ; <i64> [#uses=1]
%shl443 = shl i64 %add365, 30 ; <i64> [#uses=1]
%and445 = lshr i64 %add365, 2 ; <i64> [#uses=1]
%shr446 = and i64 %and445, 1073741823 ; <i64> [#uses=1]
%or447 = or i64 %shr446, %shl443 ; <i64> [#uses=1]
%xor461 = xor i64 0, %or447 ; <i64> [#uses=1]
%add473 = add i64 %xor461, 0 ; <i64> [#uses=1]
%add479 = add i64 %add473, %add441 ; <i64> [#uses=3]
%conv4932682 = zext i8 undef to i64 ; <i64> [#uses=1]
%shl494 = shl i64 %conv4932682, 16 ; <i64> [#uses=1]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%ptrincdec4903012 = load i8, i8* null ; <i8> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv5032681 = zext i8 %ptrincdec4903012 to i64 ; <i64> [#uses=1]
%shl504 = shl i64 %conv5032681, 8 ; <i64> [#uses=1]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%ptrincdec5003009 = load i8, i8* null ; <i8> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv5132680 = zext i8 %ptrincdec5003009 to i64 ; <i64> [#uses=1]
%or495 = or i64 %shl494, 0 ; <i64> [#uses=1]
%or505 = or i64 %or495, %conv5132680 ; <i64> [#uses=1]
%or514 = or i64 %or505, %shl504 ; <i64> [#uses=1]
store i32 undef, i32* undef
%or540 = or i64 undef, 0 ; <i64> [#uses=0]
%add542 = add i64 %add134, 310598401 ; <i64> [#uses=1]
%add551 = add i64 %add542, %or514 ; <i64> [#uses=1]
%add554 = add i64 %add551, 0 ; <i64> [#uses=1]
%add555 = add i64 %add554, 0 ; <i64> [#uses=1]
%or561 = or i64 undef, undef ; <i64> [#uses=1]
%or567 = or i64 undef, undef ; <i64> [#uses=1]
%and572 = lshr i64 %add479, 22 ; <i64> [#uses=1]
%shr573 = and i64 %and572, 1023 ; <i64> [#uses=1]
%or574 = or i64 %shr573, 0 ; <i64> [#uses=1]
%xor568 = xor i64 %or567, %or574 ; <i64> [#uses=1]
%xor575 = xor i64 %xor568, %or561 ; <i64> [#uses=1]
%add587 = add i64 %xor575, 0 ; <i64> [#uses=1]
%add593 = add i64 %add587, %add555 ; <i64> [#uses=1]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%ptrincdec6043000 = load i8, i8* null ; <i8> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv6172676 = zext i8 %ptrincdec6043000 to i64 ; <i64> [#uses=1]
%shl618 = shl i64 %conv6172676, 8 ; <i64> [#uses=1]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%ptrincdec6142997 = load i8, i8* %ctg22996 ; <i8> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv6272675 = zext i8 %ptrincdec6142997 to i64 ; <i64> [#uses=1]
%or619 = or i64 0, %conv6272675 ; <i64> [#uses=1]
%or628 = or i64 %or619, %shl618 ; <i64> [#uses=1]
%add656 = add i64 %add248, 607225278 ; <i64> [#uses=1]
%add665 = add i64 %add656, %or628 ; <i64> [#uses=1]
%add668 = add i64 %add665, 0 ; <i64> [#uses=1]
%add669 = add i64 %add668, 0 ; <i64> [#uses=1]
%and699 = and i64 %add479, %add365 ; <i64> [#uses=1]
%xor700 = xor i64 0, %and699 ; <i64> [#uses=1]
%add701 = add i64 0, %xor700 ; <i64> [#uses=1]
%add707 = add i64 %add701, %add669 ; <i64> [#uses=4]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%ptrincdec6242994 = load i8, i8* null ; <i8> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv7122673 = zext i8 %ptrincdec6242994 to i64 ; <i64> [#uses=1]
%shl713 = shl i64 %conv7122673, 24 ; <i64> [#uses=1]
%conv7412670 = zext i8 undef to i64 ; <i64> [#uses=1]
%or723 = or i64 0, %shl713 ; <i64> [#uses=1]
%or733 = or i64 %or723, %conv7412670 ; <i64> [#uses=1]
%or742 = or i64 %or733, 0 ; <i64> [#uses=2]
%conv743 = trunc i64 %or742 to i32 ; <i32> [#uses=1]
store i32 %conv743, i32* undef
%xor762 = xor i64 undef, 0 ; <i64> [#uses=0]
%add770 = add i64 %add362, 1426881987 ; <i64> [#uses=1]
%add779 = add i64 %add770, %or742 ; <i64> [#uses=1]
%add782 = add i64 %add779, 0 ; <i64> [#uses=1]
%add783 = add i64 %add782, 0 ; <i64> [#uses=1]
%shl785 = shl i64 %add707, 30 ; <i64> [#uses=1]
%and787 = lshr i64 %add707, 2 ; <i64> [#uses=1]
%shr788 = and i64 %and787, 1073741823 ; <i64> [#uses=1]
%or789 = or i64 %shr788, %shl785 ; <i64> [#uses=1]
%shl791 = shl i64 %add707, 19 ; <i64> [#uses=0]
%xor803 = xor i64 0, %or789 ; <i64> [#uses=1]
%and813 = and i64 %add593, %add479 ; <i64> [#uses=1]
%xor814 = xor i64 0, %and813 ; <i64> [#uses=1]
%add815 = add i64 %xor803, %xor814 ; <i64> [#uses=1]
%add821 = add i64 %add815, %add783 ; <i64> [#uses=1]
%add1160 = add i64 0, %add707 ; <i64> [#uses=0]
%add1157 = add i64 undef, undef ; <i64> [#uses=0]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%ptrincdec11742940 = load i8, i8* null ; <i8> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv11872651 = zext i8 %ptrincdec11742940 to i64 ; <i64> [#uses=1]
%shl1188 = shl i64 %conv11872651, 8 ; <i64> [#uses=1]
%or1198 = or i64 0, %shl1188 ; <i64> [#uses=1]
store i32 undef, i32* undef
%add1226 = add i64 %or1198, 3248222580 ; <i64> [#uses=1]
%add1235 = add i64 %add1226, 0 ; <i64> [#uses=1]
%add1238 = add i64 %add1235, 0 ; <i64> [#uses=1]
%add1239 = add i64 %add1238, 0 ; <i64> [#uses=1]
br label %for.cond
for.cond: ; preds = %for.body, %while.body
%add821.pn = phi i64 [ %add821, %while.body ], [ undef, %for.body ] ; <i64> [#uses=0]
%add1239.pn = phi i64 [ %add1239, %while.body ], [ 0, %for.body ] ; <i64> [#uses=0]
br i1 undef, label %for.end, label %for.body
for.body: ; preds = %for.cond
br label %for.cond
for.end: ; preds = %for.cond
%indvar.next2788 = add i64 %indvar2787, 1 ; <i64> [#uses=1]
br i1 undef, label %while.end, label %while.body
while.end: ; preds = %for.end, %entry
ret void
}
define void @sha256_block2(i32* nocapture %arr, i8* nocapture %in, i64 %num) nounwind {
entry:
br i1 undef, label %while.end, label %bb.nph
bb.nph: ; preds = %entry
[opaque pointer type] Add textual IR support for explicit type parameter to getelementptr instruction One of several parallel first steps to remove the target type of pointers, replacing them with a single opaque pointer type. This adds an explicit type parameter to the gep instruction so that when the first parameter becomes an opaque pointer type, the type to gep through is still available to the instructions. * This doesn't modify gep operators, only instructions (operators will be handled separately) * Textual IR changes only. Bitcode (including upgrade) and changing the in-memory representation will be in separate changes. * geps of vectors are transformed as: getelementptr <4 x float*> %x, ... ->getelementptr float, <4 x float*> %x, ... Then, once the opaque pointer type is introduced, this will ultimately look like: getelementptr float, <4 x ptr> %x with the unambiguous interpretation that it is a vector of pointers to float. * address spaces remain on the pointer, not the type: getelementptr float addrspace(1)* %x ->getelementptr float, float addrspace(1)* %x Then, eventually: getelementptr float, ptr addrspace(1) %x Importantly, the massive amount of test case churn has been automated by same crappy python code. I had to manually update a few test cases that wouldn't fit the script's model (r228970,r229196,r229197,r229198). The python script just massages stdin and writes the result to stdout, I then wrapped that in a shell script to handle replacing files, then using the usual find+xargs to migrate all the files. update.py: import fileinput import sys import re ibrep = re.compile(r"(^.*?[^%\w]getelementptr inbounds )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") normrep = re.compile( r"(^.*?[^%\w]getelementptr )(((?:<\d* x )?)(.*?)(| addrspace\(\d\)) *\*(|>)(?:$| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$))") def conv(match, line): if not match: return line line = match.groups()[0] if len(match.groups()[5]) == 0: line += match.groups()[2] line += match.groups()[3] line += ", " line += match.groups()[1] line += "\n" return line for line in sys.stdin: if line.find("getelementptr ") == line.find("getelementptr inbounds"): if line.find("getelementptr inbounds") != line.find("getelementptr inbounds ("): line = conv(re.match(ibrep, line), line) elif line.find("getelementptr ") != line.find("getelementptr ("): line = conv(re.match(normrep, line), line) sys.stdout.write(line) apply.sh: for name in "$@" do python3 `dirname "$0"`/update.py < "$name" > "$name.tmp" && mv "$name.tmp" "$name" rm -f "$name.tmp" done The actual commands: From llvm/src: find test/ -name *.ll | xargs ./apply.sh From llvm/src/tools/clang: find test/ -name *.mm -o -name *.m -o -name *.cpp -o -name *.c | xargs -I '{}' ../../apply.sh "{}" From llvm/src/tools/polly: find test/ -name *.ll | xargs ./apply.sh After that, check-all (with llvm, clang, clang-tools-extra, lld, compiler-rt, and polly all checked out). The extra 'rm' in the apply.sh script is due to a few files in clang's test suite using interesting unicode stuff that my python script was throwing exceptions on. None of those files needed to be migrated, so it seemed sufficient to ignore those cases. Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7636 llvm-svn: 230786
2015-02-27 20:29:02 +01:00
%arrayidx5 = getelementptr i32, i32* %arr, i64 1 ; <i32*> [#uses=1]
%arrayidx9 = getelementptr i32, i32* %arr, i64 2 ; <i32*> [#uses=2]
%arrayidx13 = getelementptr i32, i32* %arr, i64 3 ; <i32*> [#uses=2]
%arrayidx25 = getelementptr i32, i32* %arr, i64 6 ; <i32*> [#uses=1]
%arrayidx29 = getelementptr i32, i32* %arr, i64 7 ; <i32*> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
br label %while.body
while.body: ; preds = %for.end, %bb.nph
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%tmp3 = load i32, i32* %arr ; <i32> [#uses=2]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv = zext i32 %tmp3 to i64 ; <i64> [#uses=1]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%tmp10 = load i32, i32* %arrayidx9 ; <i32> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv11 = zext i32 %tmp10 to i64 ; <i64> [#uses=1]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%tmp14 = load i32, i32* %arrayidx13 ; <i32> [#uses=3]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv15 = zext i32 %tmp14 to i64 ; <i64> [#uses=2]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%tmp18 = load i32, i32* undef ; <i32> [#uses=2]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv19 = zext i32 %tmp18 to i64 ; <i64> [#uses=1]
%conv23 = zext i32 undef to i64 ; <i64> [#uses=1]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%tmp26 = load i32, i32* %arrayidx25 ; <i32> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv27 = zext i32 %tmp26 to i64 ; <i64> [#uses=1]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%tmp30 = load i32, i32* %arrayidx29 ; <i32> [#uses=2]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv31 = zext i32 %tmp30 to i64 ; <i64> [#uses=5]
%shl72 = shl i64 %conv31, 26 ; <i64> [#uses=1]
%shr = lshr i64 %conv31, 6 ; <i64> [#uses=1]
%or74 = or i64 %shl72, %shr ; <i64> [#uses=1]
%shr85 = lshr i64 %conv31, 25 ; <i64> [#uses=0]
%xor87 = xor i64 0, %or74 ; <i64> [#uses=1]
%and902706 = and i32 %tmp30, %tmp3 ; <i32> [#uses=1]
%and90 = zext i32 %and902706 to i64 ; <i64> [#uses=1]
%xor94 = xor i64 0, %and90 ; <i64> [#uses=1]
%add = add i64 %conv11, 1508970993 ; <i64> [#uses=1]
%add95 = add i64 %add, %xor94 ; <i64> [#uses=1]
%add98 = add i64 %add95, %xor87 ; <i64> [#uses=1]
%add99 = add i64 %add98, 0 ; <i64> [#uses=2]
%xor130 = zext i32 undef to i64 ; <i64> [#uses=1]
%add134 = add i64 %add99, %conv27 ; <i64> [#uses=2]
%add131 = add i64 %xor130, 0 ; <i64> [#uses=1]
%add137 = add i64 %add131, %add99 ; <i64> [#uses=5]
%conv1422700 = zext i8 undef to i64 ; <i64> [#uses=1]
%shl143 = shl i64 %conv1422700, 24 ; <i64> [#uses=1]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%ptrincdec1393051 = load i8, i8* undef ; <i8> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv1512699 = zext i8 %ptrincdec1393051 to i64 ; <i64> [#uses=1]
%shl152 = shl i64 %conv1512699, 16 ; <i64> [#uses=1]
%conv1712697 = zext i8 undef to i64 ; <i64> [#uses=1]
%or153 = or i64 %shl152, %shl143 ; <i64> [#uses=1]
%or163 = or i64 %or153, %conv1712697 ; <i64> [#uses=1]
%or172 = or i64 %or163, 0 ; <i64> [#uses=1]
%and203 = and i64 %add134, %conv31 ; <i64> [#uses=1]
%xor208 = xor i64 0, %and203 ; <i64> [#uses=1]
%add200 = add i64 0, 2453635748 ; <i64> [#uses=1]
%add209 = add i64 %add200, %or172 ; <i64> [#uses=1]
%add212 = add i64 %add209, %xor208 ; <i64> [#uses=1]
%add213 = add i64 %add212, 0 ; <i64> [#uses=2]
%shl228 = shl i64 %add137, 10 ; <i64> [#uses=1]
%and230 = lshr i64 %add137, 22 ; <i64> [#uses=1]
%shr231 = and i64 %and230, 1023 ; <i64> [#uses=1]
%or232 = or i64 %shr231, %shl228 ; <i64> [#uses=1]
%xor226 = xor i64 0, %or232 ; <i64> [#uses=1]
%xor233 = xor i64 %xor226, 0 ; <i64> [#uses=1]
%and2362695 = zext i32 undef to i64 ; <i64> [#uses=1]
%xor240 = and i64 %add137, %and2362695 ; <i64> [#uses=1]
%and2432694 = and i32 %tmp18, %tmp14 ; <i32> [#uses=1]
%and243 = zext i32 %and2432694 to i64 ; <i64> [#uses=1]
%xor244 = xor i64 %xor240, %and243 ; <i64> [#uses=1]
%add248 = add i64 %add213, %conv23 ; <i64> [#uses=2]
%add245 = add i64 %xor233, %xor244 ; <i64> [#uses=1]
%add251 = add i64 %add245, %add213 ; <i64> [#uses=1]
%conv2752691 = zext i8 undef to i64 ; <i64> [#uses=1]
%shl276 = shl i64 %conv2752691, 8 ; <i64> [#uses=0]
%and317 = and i64 %add248, %add134 ; <i64> [#uses=1]
%neg319 = xor i64 %add248, 4294967295 ; <i64> [#uses=1]
%and321 = and i64 %neg319, %conv31 ; <i64> [#uses=1]
%xor322 = xor i64 %and321, %and317 ; <i64> [#uses=1]
%add314 = add i64 %conv, 2870763221 ; <i64> [#uses=1]
%add323 = add i64 %add314, 0 ; <i64> [#uses=1]
%add326 = add i64 %add323, %xor322 ; <i64> [#uses=1]
%add327 = add i64 %add326, 0 ; <i64> [#uses=2]
%and3502689 = xor i64 %add137, %conv15 ; <i64> [#uses=1]
%xor354 = and i64 %add251, %and3502689 ; <i64> [#uses=1]
%and357 = and i64 %add137, %conv15 ; <i64> [#uses=1]
%xor358 = xor i64 %xor354, %and357 ; <i64> [#uses=1]
%add362 = add i64 %add327, %conv19 ; <i64> [#uses=1]
%add359 = add i64 0, %xor358 ; <i64> [#uses=1]
%add365 = add i64 %add359, %add327 ; <i64> [#uses=1]
%add770 = add i64 %add362, 1426881987 ; <i64> [#uses=1]
%add779 = add i64 %add770, 0 ; <i64> [#uses=1]
%add782 = add i64 %add779, 0 ; <i64> [#uses=1]
%add783 = add i64 %add782, 0 ; <i64> [#uses=2]
%add818 = add i64 %add783, %add365 ; <i64> [#uses=1]
%add821 = add i64 0, %add783 ; <i64> [#uses=1]
store i32 undef, i32* undef
%add1046 = add i64 undef, undef ; <i64> [#uses=1]
%add1160 = add i64 undef, undef ; <i64> [#uses=1]
store i32 0, i32* undef
%add1235 = add i64 0, %add818 ; <i64> [#uses=1]
%add1238 = add i64 %add1235, 0 ; <i64> [#uses=1]
%add1239 = add i64 %add1238, 0 ; <i64> [#uses=1]
br label %for.cond
for.cond: ; preds = %for.body, %while.body
%h.0 = phi i64 [ undef, %while.body ], [ %add2035, %for.body ] ; <i64> [#uses=1]
%g.0 = phi i64 [ %add1046, %while.body ], [ undef, %for.body ] ; <i64> [#uses=1]
%f.0 = phi i64 [ %add1160, %while.body ], [ undef, %for.body ] ; <i64> [#uses=1]
%add821.pn = phi i64 [ %add821, %while.body ], [ undef, %for.body ] ; <i64> [#uses=0]
%add1239.pn2648 = phi i64 [ %add1239, %while.body ], [ undef, %for.body ] ; <i64> [#uses=0]
%d.0 = phi i64 [ undef, %while.body ], [ %add2038, %for.body ] ; <i64> [#uses=2]
br i1 undef, label %for.end, label %for.body
for.body: ; preds = %for.cond
%conv1390 = zext i32 undef to i64 ; <i64> [#uses=1]
%add1375 = add i64 0, %h.0 ; <i64> [#uses=1]
%add1384 = add i64 %add1375, 0 ; <i64> [#uses=1]
%add1391 = add i64 %add1384, %conv1390 ; <i64> [#uses=1]
%add1392 = add i64 %add1391, 0 ; <i64> [#uses=2]
%or1411 = or i64 0, undef ; <i64> [#uses=1]
%xor1405 = xor i64 0, %or1411 ; <i64> [#uses=1]
%xor1412 = xor i64 %xor1405, 0 ; <i64> [#uses=1]
%add1427 = add i64 %add1392, %d.0 ; <i64> [#uses=1]
%add1424 = add i64 %xor1412, 0 ; <i64> [#uses=1]
%add1430 = add i64 %add1424, %add1392 ; <i64> [#uses=5]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%tmp1438 = load i32, i32* undef ; <i32> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv1439 = zext i32 %tmp1438 to i64 ; <i64> [#uses=4]
%shl1441 = shl i64 %conv1439, 25 ; <i64> [#uses=1]
%shr1444 = lshr i64 %conv1439, 7 ; <i64> [#uses=1]
%or1445 = or i64 %shl1441, %shr1444 ; <i64> [#uses=1]
%shr1450 = lshr i64 %conv1439, 18 ; <i64> [#uses=1]
%or1451 = or i64 0, %shr1450 ; <i64> [#uses=1]
%shr1454 = lshr i64 %conv1439, 3 ; <i64> [#uses=1]
%xor1452 = xor i64 %or1451, %shr1454 ; <i64> [#uses=1]
%xor1455 = xor i64 %xor1452, %or1445 ; <i64> [#uses=1]
%conv1464 = zext i32 undef to i64 ; <i64> [#uses=4]
%shl1466 = shl i64 %conv1464, 15 ; <i64> [#uses=1]
%shr1469 = lshr i64 %conv1464, 17 ; <i64> [#uses=1]
%or1470 = or i64 %shl1466, %shr1469 ; <i64> [#uses=1]
%shr1475 = lshr i64 %conv1464, 19 ; <i64> [#uses=1]
%or1476 = or i64 0, %shr1475 ; <i64> [#uses=1]
%shr1479 = lshr i64 %conv1464, 10 ; <i64> [#uses=1]
%xor1477 = xor i64 %or1476, %shr1479 ; <i64> [#uses=1]
%xor1480 = xor i64 %xor1477, %or1470 ; <i64> [#uses=1]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%tmp1499 = load i32, i32* null ; <i32> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv1500 = zext i32 %tmp1499 to i64 ; <i64> [#uses=1]
%add1491 = add i64 %conv1500, 0 ; <i64> [#uses=1]
%add1501 = add i64 %add1491, %xor1455 ; <i64> [#uses=1]
%add1502 = add i64 %add1501, %xor1480 ; <i64> [#uses=1]
%conv1504 = and i64 %add1502, 4294967295 ; <i64> [#uses=1]
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%tmp1541 = load i32, i32* undef ; <i32> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv1542 = zext i32 %tmp1541 to i64 ; <i64> [#uses=1]
%add1527 = add i64 %conv1542, %g.0 ; <i64> [#uses=1]
%add1536 = add i64 %add1527, 0 ; <i64> [#uses=1]
%add1543 = add i64 %add1536, %conv1504 ; <i64> [#uses=1]
%add1544 = add i64 %add1543, 0 ; <i64> [#uses=1]
%shl1546 = shl i64 %add1430, 30 ; <i64> [#uses=1]
%and1548 = lshr i64 %add1430, 2 ; <i64> [#uses=1]
%shr1549 = and i64 %and1548, 1073741823 ; <i64> [#uses=1]
%or1550 = or i64 %shr1549, %shl1546 ; <i64> [#uses=1]
%shl1552 = shl i64 %add1430, 19 ; <i64> [#uses=1]
%or1556 = or i64 0, %shl1552 ; <i64> [#uses=1]
%shl1559 = shl i64 %add1430, 10 ; <i64> [#uses=1]
%or1563 = or i64 0, %shl1559 ; <i64> [#uses=1]
%xor1557 = xor i64 %or1556, %or1563 ; <i64> [#uses=1]
%xor1564 = xor i64 %xor1557, %or1550 ; <i64> [#uses=1]
%add1576 = add i64 %xor1564, 0 ; <i64> [#uses=1]
%add1582 = add i64 %add1576, %add1544 ; <i64> [#uses=3]
store i32 undef, i32* undef
[opaque pointer type] Add textual IR support for explicit type parameter to load instruction Essentially the same as the GEP change in r230786. A similar migration script can be used to update test cases, though a few more test case improvements/changes were required this time around: (r229269-r229278) import fileinput import sys import re pat = re.compile(r"((?:=|:|^)\s*load (?:atomic )?(?:volatile )?(.*?))(| addrspace\(\d+\) *)\*($| *(?:%|@|null|undef|blockaddress|getelementptr|addrspacecast|bitcast|inttoptr|\[\[[a-zA-Z]|\{\{).*$)") for line in sys.stdin: sys.stdout.write(re.sub(pat, r"\1, \2\3*\4", line)) Reviewers: rafael, dexonsmith, grosser Differential Revision: http://reviews.llvm.org/D7649 llvm-svn: 230794
2015-02-27 22:17:42 +01:00
%tmp1693 = load i32, i32* undef ; <i32> [#uses=1]
Fix for PR4225: When rewriter reuse a value in a physical register , it clear the register kill operand marker and its kill ops information. However, the cleared operand may be a def of a super-register. Clear the kill ops info for the super-register's sub-registers as well. llvm-svn: 72758
2009-06-03 11:00:27 +02:00
%conv1694 = zext i32 %tmp1693 to i64 ; <i64> [#uses=1]
%add1679 = add i64 %conv1694, %f.0 ; <i64> [#uses=1]
%add1688 = add i64 %add1679, 0 ; <i64> [#uses=1]
%add1695 = add i64 %add1688, 0 ; <i64> [#uses=1]
%add1696 = add i64 %add1695, 0 ; <i64> [#uses=1]
%shl1698 = shl i64 %add1582, 30 ; <i64> [#uses=0]
%shl1704 = shl i64 %add1582, 19 ; <i64> [#uses=0]
%add1734 = add i64 0, %add1696 ; <i64> [#uses=1]
%add1983 = add i64 0, %add1427 ; <i64> [#uses=1]
%add1992 = add i64 %add1983, 0 ; <i64> [#uses=1]
%add1999 = add i64 %add1992, 0 ; <i64> [#uses=1]
%add2000 = add i64 %add1999, 0 ; <i64> [#uses=2]
%and2030 = and i64 %add1734, %add1582 ; <i64> [#uses=1]
%xor2031 = xor i64 0, %and2030 ; <i64> [#uses=1]
%add2035 = add i64 %add2000, %add1430 ; <i64> [#uses=1]
%add2032 = add i64 0, %xor2031 ; <i64> [#uses=1]
%add2038 = add i64 %add2032, %add2000 ; <i64> [#uses=1]
store i32 0, i32* undef
br label %for.cond
for.end: ; preds = %for.cond
store i32 undef, i32* %arrayidx5
store i32 undef, i32* %arrayidx9
%d.02641 = trunc i64 %d.0 to i32 ; <i32> [#uses=1]
%conv2524 = add i32 %tmp14, %d.02641 ; <i32> [#uses=1]
store i32 %conv2524, i32* %arrayidx13
%exitcond2789 = icmp eq i64 undef, %num ; <i1> [#uses=1]
br i1 %exitcond2789, label %while.end, label %while.body
while.end: ; preds = %for.end, %entry
ret void
}
Reference in New Issue Copy Permalink