maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/X86/tailcall-64.ll

246 lines
6.6 KiB
LLVM
Raw Normal View History

Fix an illegal tailcall opt where the callee returns a double via xmm while caller returns x86_fp80 via st0. rdar://12229511 llvm-svn: 164588
2012-09-25 13:32:34 +08:00
; RUN: llc -mtriple=x86_64-apple-macosx -mcpu=core2 < %s | FileCheck %s
add some simple 64-bit tail call tests. llvm-svn: 157797
2012-06-01 13:03:31 +08:00
declare i64 @testi()
define i64 @test_trivial() {
%A = tail call i64 @testi()
ret i64 %A
}
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: test_trivial:
add some simple 64-bit tail call tests. llvm-svn: 157797
2012-06-01 13:03:31 +08:00
; CHECK: jmp _testi ## TAILCALL
define i64 @test_noop_bitcast() {
%A = tail call i64 @testi()
%B = bitcast i64 %A to i64
ret i64 %B
}
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: test_noop_bitcast:
add some simple 64-bit tail call tests. llvm-svn: 157797
2012-06-01 13:03:31 +08:00
; CHECK: jmp _testi ## TAILCALL
enhance getNoopInput to know about vector<->vector bitcasts of legal types, as well as int<->ptr casts. This allows us to tailcall functions with some trivial casts between the call and return (i.e. because the return types disagree). llvm-svn: 157798
2012-06-01 13:16:33 +08:00
; Tail call shouldn't be blocked by no-op inttoptr.
define i8* @test_inttoptr() {
%A = tail call i64 @testi()
%B = inttoptr i64 %A to i8*
ret i8* %B
}
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: test_inttoptr:
enhance getNoopInput to know about vector<->vector bitcasts of legal types, as well as int<->ptr casts. This allows us to tailcall functions with some trivial casts between the call and return (i.e. because the return types disagree). llvm-svn: 157798
2012-06-01 13:16:33 +08:00
; CHECK: jmp _testi ## TAILCALL
declare <4 x float> @testv()
define <4 x i32> @test_vectorbitcast() {
%A = tail call <4 x float> @testv()
%B = bitcast <4 x float> %A to <4 x i32>
ret <4 x i32> %B
}
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: test_vectorbitcast:
enhance getNoopInput to know about vector<->vector bitcasts of legal types, as well as int<->ptr casts. This allows us to tailcall functions with some trivial casts between the call and return (i.e. because the return types disagree). llvm-svn: 157798
2012-06-01 13:16:33 +08:00
; CHECK: jmp _testv ## TAILCALL
enhance the logic for looking through tailcalls to look through transparent casts in multiple-return value scenarios, like what happens on X86-64 when returning small structs. llvm-svn: 157800
2012-06-01 13:29:15 +08:00
declare { i64, i64 } @testp()
define {i64, i64} @test_pair_trivial() {
%A = tail call { i64, i64} @testp()
ret { i64, i64} %A
}
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: test_pair_trivial:
enhance the logic for looking through tailcalls to look through transparent casts in multiple-return value scenarios, like what happens on X86-64 when returning small structs. llvm-svn: 157800
2012-06-01 13:29:15 +08:00
; CHECK: jmp _testp ## TAILCALL
Allow tail call opportunity detection through nested and/or multiple iterations of extractelement/insertelement indirection llvm-svn: 179924
2013-04-20 12:27:51 +08:00
define {i64, i64} @test_pair_notail() {
%A = tail call i64 @testi()
%b = insertvalue {i64, i64} undef, i64 %A, 0
%c = insertvalue {i64, i64} %b, i64 %A, 1
enhance the logic for looking through tailcalls to look through transparent casts in multiple-return value scenarios, like what happens on X86-64 when returning small structs. llvm-svn: 157800
2012-06-01 13:29:15 +08:00
Allow tail call opportunity detection through nested and/or multiple iterations of extractelement/insertelement indirection llvm-svn: 179924
2013-04-20 12:27:51 +08:00
ret { i64, i64} %c
}
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: test_pair_notail:
Allow tail call opportunity detection through nested and/or multiple iterations of extractelement/insertelement indirection llvm-svn: 179924
2013-04-20 12:27:51 +08:00
; CHECK-NOT: jmp _testi
enhance the logic for looking through tailcalls to look through transparent casts in multiple-return value scenarios, like what happens on X86-64 when returning small structs. llvm-svn: 157800
2012-06-01 13:29:15 +08:00
Minor renaming of tests (for consistency with an in-development patch) llvm-svn: 179954
2013-04-21 00:21:26 +08:00
define {i64, i64} @test_pair_extract_trivial() {
enhance the logic for looking through tailcalls to look through transparent casts in multiple-return value scenarios, like what happens on X86-64 when returning small structs. llvm-svn: 157800
2012-06-01 13:29:15 +08:00
%A = tail call { i64, i64} @testp()
%x = extractvalue { i64, i64} %A, 0
%y = extractvalue { i64, i64} %A, 1
%b = insertvalue {i64, i64} undef, i64 %x, 0
%c = insertvalue {i64, i64} %b, i64 %y, 1
ret { i64, i64} %c
}
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: test_pair_extract_trivial:
enhance the logic for looking through tailcalls to look through transparent casts in multiple-return value scenarios, like what happens on X86-64 when returning small structs. llvm-svn: 157800
2012-06-01 13:29:15 +08:00
; CHECK: jmp _testp ## TAILCALL
Minor renaming of tests (for consistency with an in-development patch) llvm-svn: 179954
2013-04-21 00:21:26 +08:00
define {i64, i64} @test_pair_extract_notail() {
Allow tail call opportunity detection through nested and/or multiple iterations of extractelement/insertelement indirection llvm-svn: 179924
2013-04-20 12:27:51 +08:00
%A = tail call { i64, i64} @testp()
%x = extractvalue { i64, i64} %A, 0
%y = extractvalue { i64, i64} %A, 1
%b = insertvalue {i64, i64} undef, i64 %y, 0
%c = insertvalue {i64, i64} %b, i64 %x, 1
ret { i64, i64} %c
}
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: test_pair_extract_notail:
Allow tail call opportunity detection through nested and/or multiple iterations of extractelement/insertelement indirection llvm-svn: 179924
2013-04-20 12:27:51 +08:00
; CHECK-NOT: jmp _testp
Minor renaming of tests (for consistency with an in-development patch) llvm-svn: 179954
2013-04-21 00:21:26 +08:00
define {i8*, i64} @test_pair_extract_conv() {
enhance the logic for looking through tailcalls to look through transparent casts in multiple-return value scenarios, like what happens on X86-64 when returning small structs. llvm-svn: 157800
2012-06-01 13:29:15 +08:00
%A = tail call { i64, i64} @testp()
%x = extractvalue { i64, i64} %A, 0
%y = extractvalue { i64, i64} %A, 1
%x1 = inttoptr i64 %x to i8*
%b = insertvalue {i8*, i64} undef, i8* %x1, 0
%c = insertvalue {i8*, i64} %b, i64 %y, 1
ret { i8*, i64} %c
}
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: test_pair_extract_conv:
enhance the logic for looking through tailcalls to look through transparent casts in multiple-return value scenarios, like what happens on X86-64 when returning small structs. llvm-svn: 157800
2012-06-01 13:29:15 +08:00
; CHECK: jmp _testp ## TAILCALL
Minor renaming of tests (for consistency with an in-development patch) llvm-svn: 179954
2013-04-21 00:21:26 +08:00
define {i64, i64} @test_pair_extract_multiple() {
Allow tail call opportunity detection through nested and/or multiple iterations of extractelement/insertelement indirection llvm-svn: 179924
2013-04-20 12:27:51 +08:00
%A = tail call { i64, i64} @testp()
%x = extractvalue { i64, i64} %A, 0
%y = extractvalue { i64, i64} %A, 1
%b = insertvalue {i64, i64} undef, i64 %x, 0
%c = insertvalue {i64, i64} %b, i64 %y, 1
%x1 = extractvalue { i64, i64} %b, 0
%y1 = extractvalue { i64, i64} %c, 1
%d = insertvalue {i64, i64} undef, i64 %x1, 0
%e = insertvalue {i64, i64} %b, i64 %y1, 1
ret { i64, i64} %e
}
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: test_pair_extract_multiple:
Allow tail call opportunity detection through nested and/or multiple iterations of extractelement/insertelement indirection llvm-svn: 179924
2013-04-20 12:27:51 +08:00
; CHECK: jmp _testp ## TAILCALL
Minor renaming of tests (for consistency with an in-development patch) llvm-svn: 179954
2013-04-21 00:21:26 +08:00
define {i64, i64} @test_pair_extract_undef() {
Allow tail call opportunity detection through nested and/or multiple iterations of extractelement/insertelement indirection llvm-svn: 179924
2013-04-20 12:27:51 +08:00
%A = tail call { i64, i64} @testp()
%x = extractvalue { i64, i64} %A, 0
%b = insertvalue {i64, i64} undef, i64 %x, 0
ret { i64, i64} %b
}
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: test_pair_extract_undef:
Allow tail call opportunity detection through nested and/or multiple iterations of extractelement/insertelement indirection llvm-svn: 179924
2013-04-20 12:27:51 +08:00
; CHECK: jmp _testp ## TAILCALL
declare { i64, { i32, i32 } } @testn()
define {i64, {i32, i32}} @test_nest() {
%A = tail call { i64, { i32, i32 } } @testn()
%x = extractvalue { i64, { i32, i32}} %A, 0
%y = extractvalue { i64, { i32, i32}} %A, 1
%y1 = extractvalue { i32, i32} %y, 0
%y2 = extractvalue { i32, i32} %y, 1
%b = insertvalue {i64, {i32, i32}} undef, i64 %x, 0
%c1 = insertvalue {i32, i32} undef, i32 %y1, 0
%c2 = insertvalue {i32, i32} %c1, i32 %y2, 1
%c = insertvalue {i64, {i32, i32}} %b, {i32, i32} %c2, 1
ret { i64, { i32, i32}} %c
}
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: test_nest:
Allow tail call opportunity detection through nested and/or multiple iterations of extractelement/insertelement indirection llvm-svn: 179924
2013-04-20 12:27:51 +08:00
; CHECK: jmp _testn ## TAILCALL
%struct.A = type { i32 }
%struct.B = type { %struct.A, i32 }
declare %struct.B* @testu()
define %struct.A* @test_upcast() {
entry:
%A = tail call %struct.B* @testu()
[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
2015-02-28 03:29:02 +08:00
%x = getelementptr inbounds %struct.B, %struct.B* %A, i32 0, i32 0
Allow tail call opportunity detection through nested and/or multiple iterations of extractelement/insertelement indirection llvm-svn: 179924
2013-04-20 12:27:51 +08:00
ret %struct.A* %x
}
enhance the logic for looking through tailcalls to look through transparent casts in multiple-return value scenarios, like what happens on X86-64 when returning small structs. llvm-svn: 157800
2012-06-01 13:29:15 +08:00
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: test_upcast:
Allow tail call opportunity detection through nested and/or multiple iterations of extractelement/insertelement indirection llvm-svn: 179924
2013-04-20 12:27:51 +08:00
; CHECK: jmp _testu ## TAILCALL
enhance the logic for looking through tailcalls to look through transparent casts in multiple-return value scenarios, like what happens on X86-64 when returning small structs. llvm-svn: 157800
2012-06-01 13:29:15 +08:00
testcase for PR13006, thanks to Duncan for filing it. llvm-svn: 157824
2012-06-02 02:19:46 +08:00
; PR13006
define { i64, i64 } @crash(i8* %this) {
%c = tail call { i64, i64 } @testp()
%mrv7 = insertvalue { i64, i64 } %c, i64 undef, 1
ret { i64, i64 } %mrv7
}
Fix the TCRETURNmi64 bug differently. Add a PatFrag to match X86tcret using 6 fixed registers or less. This avoids folding loads into TCRETURNmi64 using 7 or more volatile registers. <rdar://problem/12282281> llvm-svn: 163819
2012-09-14 02:31:27 +08:00
; Check that we can fold an indexed load into a tail call instruction.
; CHECK: fold_indexed_load
; CHECK: leaq (%rsi,%rsi,4), %[[RAX:r..]]
Win64: Put a REX_W prefix on all TAILJMP* instructions MSDN's x64 software conventions page says that this is one of the fixed list of legal epilogues: https://msdn.microsoft.com/en-us/library/tawsa7cb.aspx Presumably this is how the unwinder distinguishes epilogue jumps from in-function control flow. Also normalize the way we place "## TAILCALL" comments on such jumps. llvm-svn: 227611
2015-01-31 05:03:31 +08:00
; CHECK: jmpq *16(%{{r..}},%[[RAX]],8) ## TAILCALL
Fix the TCRETURNmi64 bug differently. Add a PatFrag to match X86tcret using 6 fixed registers or less. This avoids folding loads into TCRETURNmi64 using 7 or more volatile registers. <rdar://problem/12282281> llvm-svn: 163819
2012-09-14 02:31:27 +08:00
%struct.funcs = type { i32 (i8*, i32*, i32)*, i32 (i8*)*, i32 (i8*)*, i32 (i8*, i32)*, i32 }
@func_table = external global [0 x %struct.funcs]
define void @fold_indexed_load(i8* %mbstr, i64 %idxprom) nounwind uwtable ssp {
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
2015-02-28 03:29:02 +08:00
%dsplen = getelementptr inbounds [0 x %struct.funcs], [0 x %struct.funcs]* @func_table, i64 0, i64 %idxprom, i32 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
2015-02-28 05:17:42 +08:00
%x1 = load i32 (i8*)*, i32 (i8*)** %dsplen, align 8
Fix the TCRETURNmi64 bug differently. Add a PatFrag to match X86tcret using 6 fixed registers or less. This avoids folding loads into TCRETURNmi64 using 7 or more volatile registers. <rdar://problem/12282281> llvm-svn: 163819
2012-09-14 02:31:27 +08:00
%call = tail call i32 %x1(i8* %mbstr) nounwind
ret void
}
enhance the logic for looking through tailcalls to look through transparent casts in multiple-return value scenarios, like what happens on X86-64 when returning small structs. llvm-svn: 157800
2012-06-01 13:29:15 +08:00
Fix the TCRETURNmi64 bug differently. Add a PatFrag to match X86tcret using 6 fixed registers or less. This avoids folding loads into TCRETURNmi64 using 7 or more volatile registers. <rdar://problem/12282281> llvm-svn: 163819
2012-09-14 02:31:27 +08:00
; <rdar://problem/12282281> Fold an indexed load into the tail call instruction.
; Calling a varargs function with 6 arguments requires 7 registers (%al is the
; vector count for varargs functions). This leaves %r11 as the only available
; scratch register.
;
; It is not possible to fold an indexed load into TCRETURNmi64 in that case.
;
; typedef int (*funcptr)(void*, ...);
; extern const funcptr funcs[];
; int f(int n) {
; return funcs[n](0, 0, 0, 0, 0, 0);
; }
;
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: rdar12282281
Win64: Put a REX_W prefix on all TAILJMP* instructions MSDN's x64 software conventions page says that this is one of the fixed list of legal epilogues: https://msdn.microsoft.com/en-us/library/tawsa7cb.aspx Presumably this is how the unwinder distinguishes epilogue jumps from in-function control flow. Also normalize the way we place "## TAILCALL" comments on such jumps. llvm-svn: 227611
2015-01-31 05:03:31 +08:00
; CHECK: jmpq *%r11 ## TAILCALL
Fix the TCRETURNmi64 bug differently. Add a PatFrag to match X86tcret using 6 fixed registers or less. This avoids folding loads into TCRETURNmi64 using 7 or more volatile registers. <rdar://problem/12282281> llvm-svn: 163819
2012-09-14 02:31:27 +08:00
@funcs = external constant [0 x i32 (i8*, ...)*]
define i32 @rdar12282281(i32 %n) nounwind uwtable ssp {
entry:
%idxprom = sext i32 %n to i64
[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
2015-02-28 03:29:02 +08:00
%arrayidx = getelementptr inbounds [0 x i32 (i8*, ...)*], [0 x i32 (i8*, ...)*]* @funcs, i64 0, i64 %idxprom
[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
2015-02-28 05:17:42 +08:00
%0 = load i32 (i8*, ...)*, i32 (i8*, ...)** %arrayidx, align 8
Fix the TCRETURNmi64 bug differently. Add a PatFrag to match X86tcret using 6 fixed registers or less. This avoids folding loads into TCRETURNmi64 using 7 or more volatile registers. <rdar://problem/12282281> llvm-svn: 163819
2012-09-14 02:31:27 +08:00
%call = tail call i32 (i8*, ...)* %0(i8* null, i32 0, i32 0, i32 0, i32 0, i32 0) nounwind
ret i32 %call
}
Fix an illegal tailcall opt where the callee returns a double via xmm while caller returns x86_fp80 via st0. rdar://12229511 llvm-svn: 164588
2012-09-25 13:32:34 +08:00
define x86_fp80 @fp80_call(x86_fp80 %x) nounwind {
entry:
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: fp80_call:
Fix an illegal tailcall opt where the callee returns a double via xmm while caller returns x86_fp80 via st0. rdar://12229511 llvm-svn: 164588
2012-09-25 13:32:34 +08:00
; CHECK: jmp _fp80_callee
%call = tail call x86_fp80 @fp80_callee(x86_fp80 %x) nounwind
ret x86_fp80 %call
}
declare x86_fp80 @fp80_callee(x86_fp80)
; rdar://12229511
define x86_fp80 @trunc_fp80(x86_fp80 %x) nounwind {
entry:
Start using CHECK-LABEL in some tests. llvm-svn: 186163
2013-07-12 22:54:12 +08:00
; CHECK-LABEL: trunc_fp80
Fix an illegal tailcall opt where the callee returns a double via xmm while caller returns x86_fp80 via st0. rdar://12229511 llvm-svn: 164588
2012-09-25 13:32:34 +08:00
; CHECK: callq _trunc
; CHECK-NOT: jmp _trunc
; CHECK: ret
%conv = fptrunc x86_fp80 %x to double
%call = tail call double @trunc(double %conv) nounwind readnone
%conv1 = fpext double %call to x86_fp80
ret x86_fp80 %conv1
}
declare double @trunc(double) nounwind readnone