6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-12-15 20:29:48 +00:00
Code Issues Projects Releases Wiki Activity
c58b8f0b65
llvm-6502/test/CodeGen/PowerPC/subreg-postra.ll

169 lines
6.9 KiB
LLVM
Raw Normal View History

[PowerPC] Don't use a non-allocatable register to implement the 'cc' alias GCC accepts 'cc' as an alias for 'cr0', and we need to do the same when processing inline asm constraints. This had previously been implemented using a non-allocatable register, named 'cc', that was listed as an alias of 'cr0', but the infrastructure does not seem to support this properly (neither the register allocator nor the scheduler properly accounts for the alias). Instead, we can just process this as a naming alias inside of the inline asm constraint-processing code, so we'll do that instead. There are two regression tests, one where the post-RA scheduler did the wrong thing with the non-allocatable alias, and one where the register allocator did the wrong thing. Fixes PR21742. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223708 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-08 22:54:22 +00:00
; RUN: llc -mcpu=pwr7 < %s | FileCheck %s
target datalayout = "E-m:e-i64:64-n32:64"
target triple = "powerpc64-unknown-linux-gnu"
; Function Attrs: nounwind
define void @jbd2_journal_commit_transaction(i32* %journal) #0 {
entry:
br i1 undef, label %do.body, label %if.then5
if.then5: ; preds = %entry
unreachable
do.body: ; preds = %entry
br i1 undef, label %do.body.i, label %trace_jbd2_start_commit.exit
do.body.i: ; preds = %do.body
unreachable
trace_jbd2_start_commit.exit: ; preds = %do.body
br i1 undef, label %do.body.i1116, label %trace_jbd2_commit_locking.exit
do.body.i1116: ; preds = %trace_jbd2_start_commit.exit
br i1 undef, label %if.end.i1123, label %do.body5.i1122
do.body5.i1122: ; preds = %do.body.i1116
unreachable
if.end.i1123: ; preds = %do.body.i1116
br label %trace_jbd2_commit_locking.exit
trace_jbd2_commit_locking.exit: ; preds = %if.end.i1123, %trace_jbd2_start_commit.exit
br i1 undef, label %spin_unlock.exit1146, label %if.then.i.i.i.i1144
if.then.i.i.i.i1144: ; preds = %trace_jbd2_commit_locking.exit
unreachable
spin_unlock.exit1146: ; preds = %trace_jbd2_commit_locking.exit
br i1 undef, label %spin_unlock.exit1154, label %if.then.i.i.i.i1152
if.then.i.i.i.i1152: ; preds = %spin_unlock.exit1146
br label %spin_unlock.exit1154
spin_unlock.exit1154: ; preds = %if.then.i.i.i.i1152, %spin_unlock.exit1146
br i1 undef, label %do.body.i1159, label %trace_jbd2_commit_flushing.exit
do.body.i1159: ; preds = %spin_unlock.exit1154
unreachable
trace_jbd2_commit_flushing.exit: ; preds = %spin_unlock.exit1154
br i1 undef, label %for.end.i, label %for.body.lr.ph.i
for.body.lr.ph.i: ; preds = %trace_jbd2_commit_flushing.exit
br i1 undef, label %spin_unlock.exit.i, label %if.then.i.i.i.i.i
if.then.i.i.i.i.i: ; preds = %for.body.lr.ph.i
unreachable
spin_unlock.exit.i: ; preds = %for.body.lr.ph.i
unreachable
for.end.i: ; preds = %trace_jbd2_commit_flushing.exit
br i1 undef, label %journal_submit_data_buffers.exit, label %if.then.i.i.i.i31.i
if.then.i.i.i.i31.i: ; preds = %for.end.i
unreachable
journal_submit_data_buffers.exit: ; preds = %for.end.i
br i1 undef, label %if.end103, label %if.then102
if.then102: ; preds = %journal_submit_data_buffers.exit
unreachable
if.end103: ; preds = %journal_submit_data_buffers.exit
br i1 undef, label %do.body.i1182, label %trace_jbd2_commit_logging.exit
do.body.i1182: ; preds = %if.end103
unreachable
trace_jbd2_commit_logging.exit: ; preds = %if.end103
br i1 undef, label %for.end.i1287, label %for.body.i1277
for.body.i1277: ; preds = %trace_jbd2_commit_logging.exit
unreachable
for.end.i1287: ; preds = %trace_jbd2_commit_logging.exit
br i1 undef, label %journal_finish_inode_data_buffers.exit, label %if.then.i.i.i.i84.i
if.then.i.i.i.i84.i: ; preds = %for.end.i1287
unreachable
journal_finish_inode_data_buffers.exit: ; preds = %for.end.i1287
br i1 undef, label %if.end256, label %if.then249
if.then249: ; preds = %journal_finish_inode_data_buffers.exit
unreachable
if.end256: ; preds = %journal_finish_inode_data_buffers.exit
br i1 undef, label %do.body277, label %if.then260
if.then260: ; preds = %if.end256
br label %do.body277
do.body277: ; preds = %if.then260, %if.end256
br label %while.body318
while.body318: ; preds = %wait_on_buffer.exit, %do.body277
%tobool.i1295 = icmp eq i64 undef, 0
br i1 %tobool.i1295, label %wait_on_buffer.exit, label %if.then.i1296
if.then.i1296: ; preds = %while.body318
unreachable
wait_on_buffer.exit: ; preds = %while.body318
br i1 undef, label %do.body378, label %while.body318
do.body378: ; preds = %wait_on_buffer.exit
br i1 undef, label %while.end418, label %while.body392.lr.ph
while.body392.lr.ph: ; preds = %do.body378
br label %while.body392
while.body392: ; preds = %wait_on_buffer.exit1319, %while.body392.lr.ph
[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 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@230794 91177308-0d34-0410-b5e6-96231b3b80d8
2015-02-27 21:17:42 +00:00
%0 = load i8*, i8** undef, align 8
[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 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@230786 91177308-0d34-0410-b5e6-96231b3b80d8
2015-02-27 19:29:02 +00:00
%add.ptr399 = getelementptr inbounds i8, i8* %0, i64 -72
[PowerPC] Don't use a non-allocatable register to implement the 'cc' alias GCC accepts 'cc' as an alias for 'cr0', and we need to do the same when processing inline asm constraints. This had previously been implemented using a non-allocatable register, named 'cc', that was listed as an alias of 'cr0', but the infrastructure does not seem to support this properly (neither the register allocator nor the scheduler properly accounts for the alias). Instead, we can just process this as a naming alias inside of the inline asm constraint-processing code, so we'll do that instead. There are two regression tests, one where the post-RA scheduler did the wrong thing with the non-allocatable alias, and one where the register allocator did the wrong thing. Fixes PR21742. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223708 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-08 22:54:22 +00:00
%b_state.i.i1314 = bitcast i8* %add.ptr399 to i64*
%tobool.i1316 = icmp eq i64 undef, 0
br i1 %tobool.i1316, label %wait_on_buffer.exit1319, label %if.then.i1317
if.then.i1317: ; preds = %while.body392
unreachable
wait_on_buffer.exit1319: ; preds = %while.body392
[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 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@230794 91177308-0d34-0410-b5e6-96231b3b80d8
2015-02-27 21:17:42 +00:00
%1 = load volatile i64, i64* %b_state.i.i1314, align 8
[PowerPC] Don't use a non-allocatable register to implement the 'cc' alias GCC accepts 'cc' as an alias for 'cr0', and we need to do the same when processing inline asm constraints. This had previously been implemented using a non-allocatable register, named 'cc', that was listed as an alias of 'cr0', but the infrastructure does not seem to support this properly (neither the register allocator nor the scheduler properly accounts for the alias). Instead, we can just process this as a naming alias inside of the inline asm constraint-processing code, so we'll do that instead. There are two regression tests, one where the post-RA scheduler did the wrong thing with the non-allocatable alias, and one where the register allocator did the wrong thing. Fixes PR21742. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223708 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-08 22:54:22 +00:00
%conv.i.i1322 = and i64 %1, 1
%lnot404 = icmp eq i64 %conv.i.i1322, 0
%.err.4 = select i1 %lnot404, i32 -5, i32 undef
%2 = call i64 asm sideeffect "1:.long 0x7c0000a8 $| ((($0) & 0x1f) << 21) $| (((0) & 0x1f) << 16) $| ((($3) & 0x1f) << 11) $| (((0) & 0x1) << 0) \0Aandc $0,$0,$2\0Astdcx. $0,0,$3\0Abne- 1b\0A", "=&r,=*m,r,r,*m,~{cc},~{memory}"(i64* %b_state.i.i1314, i64 262144, i64* %b_state.i.i1314, i64* %b_state.i.i1314) #1
[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 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@230786 91177308-0d34-0410-b5e6-96231b3b80d8
2015-02-27 19:29:02 +00:00
%prev.i.i.i1325 = getelementptr inbounds i8, i8* %0, i64 8
[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 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@230794 91177308-0d34-0410-b5e6-96231b3b80d8
2015-02-27 21:17:42 +00:00
%3 = load i32*, i32** null, align 8
[PowerPC] Don't use a non-allocatable register to implement the 'cc' alias GCC accepts 'cc' as an alias for 'cr0', and we need to do the same when processing inline asm constraints. This had previously been implemented using a non-allocatable register, named 'cc', that was listed as an alias of 'cr0', but the infrastructure does not seem to support this properly (neither the register allocator nor the scheduler properly accounts for the alias). Instead, we can just process this as a naming alias inside of the inline asm constraint-processing code, so we'll do that instead. There are two regression tests, one where the post-RA scheduler did the wrong thing with the non-allocatable alias, and one where the register allocator did the wrong thing. Fixes PR21742. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@223708 91177308-0d34-0410-b5e6-96231b3b80d8
2014-12-08 22:54:22 +00:00
store i32* %3, i32** undef, align 8
call void @__brelse(i32* undef) #1
br i1 undef, label %while.end418, label %while.body392
; CHECK-LABEL: @jbd2_journal_commit_transaction
; CHECK: andi.
; CHECK: cror [[REG:[0-9]+]], 1, 1
; CHECK: stdcx.
; CHECK: isel {{[0-9]+}}, {{[0-9]+}}, {{[0-9]+}}, [[REG]]
while.end418: ; preds = %wait_on_buffer.exit1319, %do.body378
%err.4.lcssa = phi i32 [ undef, %do.body378 ], [ %.err.4, %wait_on_buffer.exit1319 ]
br i1 undef, label %if.end421, label %if.then420
if.then420: ; preds = %while.end418
call void @jbd2_journal_abort(i32* %journal, i32 signext %err.4.lcssa) #1
br label %if.end421
if.end421: ; preds = %if.then420, %while.end418
unreachable
}
declare void @jbd2_journal_abort(i32*, i32 signext)
declare void @__brelse(i32*)
attributes #0 = { nounwind }
attributes #1 = { nounwind }
Reference in New Issue Copy Permalink