6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-12-18 10:31:57 +00:00
Code Issues Projects Releases Wiki Activity
fcc897cc45
llvm-6502/test/CodeGen/AArch64/fast-isel-int-ext2.ll

440 lines
9.2 KiB
LLVM
Raw Normal View History

Change the fast-isel-abort option from bool to int to enable "levels" Summary: Currently fast-isel-abort will only abort for regular instructions, and just warn for function calls, terminators, function arguments. There is already fast-isel-abort-args but nothing for calls and terminators. This change turns the fast-isel-abort options into an integer option, so that multiple levels of strictness can be defined. This will help no being surprised when the "abort" option indeed does not abort, and enables the possibility to write test that verifies that no intrinsics are forgotten by fast-isel. Reviewers: resistor, echristo Subscribers: jfb, llvm-commits Differential Revision: http://reviews.llvm.org/D7941 From: Mehdi Amini <mehdi.amini@apple.com> git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@230775 91177308-0d34-0410-b5e6-96231b3b80d8
2015-02-27 18:32:11 +00:00
; RUN: llc -mtriple=aarch64-apple-darwin -fast-isel -fast-isel-abort=1 -aarch64-atomic-cfg-tidy=false -disable-cgp-branch-opts -verify-machineinstrs < %s | FileCheck %s
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
;
; Test folding of the sign-/zero-extend into the load instruction.
;
; Unscaled
define i32 @load_unscaled_zext_i8_to_i32(i64 %a) {
; CHECK-LABEL: load_unscaled_zext_i8_to_i32
; CHECK: ldurb w0, [x0, #-8]
; CHECK-NOT: uxtb
%1 = sub i64 %a, 8
%2 = inttoptr i64 %1 to i8*
[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 i8, i8* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = zext i8 %3 to i32
ret i32 %4
}
define i32 @load_unscaled_zext_i16_to_i32(i64 %a) {
; CHECK-LABEL: load_unscaled_zext_i16_to_i32
; CHECK: ldurh w0, [x0, #-8]
; CHECK-NOT: uxth
%1 = sub i64 %a, 8
%2 = inttoptr i64 %1 to i16*
[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 i16, i16* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = zext i16 %3 to i32
ret i32 %4
}
define i64 @load_unscaled_zext_i8_to_i64(i64 %a) {
; CHECK-LABEL: load_unscaled_zext_i8_to_i64
; CHECK: ldurb w0, [x0, #-8]
; CHECK-NOT: uxtb
%1 = sub i64 %a, 8
%2 = inttoptr i64 %1 to i8*
[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 i8, i8* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = zext i8 %3 to i64
ret i64 %4
}
define i64 @load_unscaled_zext_i16_to_i64(i64 %a) {
; CHECK-LABEL: load_unscaled_zext_i16_to_i64
; CHECK: ldurh w0, [x0, #-8]
; CHECK-NOT: uxth
%1 = sub i64 %a, 8
%2 = inttoptr i64 %1 to i16*
[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 i16, i16* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = zext i16 %3 to i64
ret i64 %4
}
define i64 @load_unscaled_zext_i32_to_i64(i64 %a) {
; CHECK-LABEL: load_unscaled_zext_i32_to_i64
; CHECK: ldur w0, [x0, #-8]
; CHECK-NOT: uxtw
%1 = sub i64 %a, 8
%2 = inttoptr i64 %1 to i32*
[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* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = zext i32 %3 to i64
ret i64 %4
}
define i32 @load_unscaled_sext_i8_to_i32(i64 %a) {
; CHECK-LABEL: load_unscaled_sext_i8_to_i32
; CHECK: ldursb w0, [x0, #-8]
; CHECK-NOT: sxtb
%1 = sub i64 %a, 8
%2 = inttoptr i64 %1 to i8*
[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 i8, i8* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = sext i8 %3 to i32
ret i32 %4
}
define i32 @load_unscaled_sext_i16_to_i32(i64 %a) {
; CHECK-LABEL: load_unscaled_sext_i16_to_i32
; CHECK: ldursh w0, [x0, #-8]
; CHECK-NOT: sxth
%1 = sub i64 %a, 8
%2 = inttoptr i64 %1 to i16*
[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 i16, i16* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = sext i16 %3 to i32
ret i32 %4
}
define i64 @load_unscaled_sext_i8_to_i64(i64 %a) {
; CHECK-LABEL: load_unscaled_sext_i8_to_i64
; CHECK: ldursb x0, [x0, #-8]
; CHECK-NOT: sxtb
%1 = sub i64 %a, 8
%2 = inttoptr i64 %1 to i8*
[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 i8, i8* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = sext i8 %3 to i64
ret i64 %4
}
define i64 @load_unscaled_sext_i16_to_i64(i64 %a) {
; CHECK-LABEL: load_unscaled_sext_i16_to_i64
; CHECK: ldursh x0, [x0, #-8]
; CHECK-NOT: sxth
%1 = sub i64 %a, 8
%2 = inttoptr i64 %1 to i16*
[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 i16, i16* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = sext i16 %3 to i64
ret i64 %4
}
define i64 @load_unscaled_sext_i32_to_i64(i64 %a) {
; CHECK-LABEL: load_unscaled_sext_i32_to_i64
; CHECK: ldursw x0, [x0, #-8]
; CHECK-NOT: sxtw
%1 = sub i64 %a, 8
%2 = inttoptr i64 %1 to i32*
[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* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = sext i32 %3 to i64
ret i64 %4
}
; Register
define i32 @load_register_zext_i8_to_i32(i64 %a, i64 %b) {
; CHECK-LABEL: load_register_zext_i8_to_i32
; CHECK: ldrb w0, [x0, x1]
; CHECK-NOT: uxtb
%1 = add i64 %a, %b
%2 = inttoptr i64 %1 to i8*
[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 i8, i8* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = zext i8 %3 to i32
ret i32 %4
}
define i32 @load_register_zext_i16_to_i32(i64 %a, i64 %b) {
; CHECK-LABEL: load_register_zext_i16_to_i32
; CHECK: ldrh w0, [x0, x1]
; CHECK-NOT: uxth
%1 = add i64 %a, %b
%2 = inttoptr i64 %1 to i16*
[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 i16, i16* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = zext i16 %3 to i32
ret i32 %4
}
define i64 @load_register_zext_i8_to_i64(i64 %a, i64 %b) {
; CHECK-LABEL: load_register_zext_i8_to_i64
; CHECK: ldrb w0, [x0, x1]
; CHECK-NOT: uxtb
%1 = add i64 %a, %b
%2 = inttoptr i64 %1 to i8*
[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 i8, i8* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = zext i8 %3 to i64
ret i64 %4
}
define i64 @load_register_zext_i16_to_i64(i64 %a, i64 %b) {
; CHECK-LABEL: load_register_zext_i16_to_i64
; CHECK: ldrh w0, [x0, x1]
; CHECK-NOT: uxth
%1 = add i64 %a, %b
%2 = inttoptr i64 %1 to i16*
[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 i16, i16* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = zext i16 %3 to i64
ret i64 %4
}
define i64 @load_register_zext_i32_to_i64(i64 %a, i64 %b) {
; CHECK-LABEL: load_register_zext_i32_to_i64
; CHECK: ldr w0, [x0, x1]
; CHECK-NOT: uxtw
%1 = add i64 %a, %b
%2 = inttoptr i64 %1 to i32*
[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* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = zext i32 %3 to i64
ret i64 %4
}
define i32 @load_register_sext_i8_to_i32(i64 %a, i64 %b) {
; CHECK-LABEL: load_register_sext_i8_to_i32
; CHECK: ldrsb w0, [x0, x1]
; CHECK-NOT: sxtb
%1 = add i64 %a, %b
%2 = inttoptr i64 %1 to i8*
[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 i8, i8* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = sext i8 %3 to i32
ret i32 %4
}
define i32 @load_register_sext_i16_to_i32(i64 %a, i64 %b) {
; CHECK-LABEL: load_register_sext_i16_to_i32
; CHECK: ldrsh w0, [x0, x1]
; CHECK-NOT: sxth
%1 = add i64 %a, %b
%2 = inttoptr i64 %1 to i16*
[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 i16, i16* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = sext i16 %3 to i32
ret i32 %4
}
define i64 @load_register_sext_i8_to_i64(i64 %a, i64 %b) {
; CHECK-LABEL: load_register_sext_i8_to_i64
; CHECK: ldrsb x0, [x0, x1]
; CHECK-NOT: sxtb
%1 = add i64 %a, %b
%2 = inttoptr i64 %1 to i8*
[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 i8, i8* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = sext i8 %3 to i64
ret i64 %4
}
define i64 @load_register_sext_i16_to_i64(i64 %a, i64 %b) {
; CHECK-LABEL: load_register_sext_i16_to_i64
; CHECK: ldrsh x0, [x0, x1]
; CHECK-NOT: sxth
%1 = add i64 %a, %b
%2 = inttoptr i64 %1 to i16*
[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 i16, i16* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = sext i16 %3 to i64
ret i64 %4
}
define i64 @load_register_sext_i32_to_i64(i64 %a, i64 %b) {
; CHECK-LABEL: load_register_sext_i32_to_i64
; CHECK: ldrsw x0, [x0, x1]
; CHECK-NOT: sxtw
%1 = add i64 %a, %b
%2 = inttoptr i64 %1 to i32*
[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* %2
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%4 = sext i32 %3 to i64
ret i64 %4
}
; Extend
define i32 @load_extend_zext_i8_to_i32(i64 %a, i32 %b) {
; CHECK-LABEL: load_extend_zext_i8_to_i32
; CHECK: ldrb w0, [x0, w1, sxtw]
; CHECK-NOT: uxtb
%1 = sext i32 %b to i64
%2 = add i64 %a, %1
%3 = inttoptr i64 %2 to i8*
[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
%4 = load i8, i8* %3
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%5 = zext i8 %4 to i32
ret i32 %5
}
define i32 @load_extend_zext_i16_to_i32(i64 %a, i32 %b) {
; CHECK-LABEL: load_extend_zext_i16_to_i32
; CHECK: ldrh w0, [x0, w1, sxtw]
; CHECK-NOT: uxth
%1 = sext i32 %b to i64
%2 = add i64 %a, %1
%3 = inttoptr i64 %2 to i16*
[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
%4 = load i16, i16* %3
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%5 = zext i16 %4 to i32
ret i32 %5
}
define i64 @load_extend_zext_i8_to_i64(i64 %a, i32 %b) {
; CHECK-LABEL: load_extend_zext_i8_to_i64
; CHECK: ldrb w0, [x0, w1, sxtw]
; CHECK-NOT: uxtb
%1 = sext i32 %b to i64
%2 = add i64 %a, %1
%3 = inttoptr i64 %2 to i8*
[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
%4 = load i8, i8* %3
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%5 = zext i8 %4 to i64
ret i64 %5
}
define i64 @load_extend_zext_i16_to_i64(i64 %a, i32 %b) {
; CHECK-LABEL: load_extend_zext_i16_to_i64
; CHECK: ldrh w0, [x0, w1, sxtw]
; CHECK-NOT: uxth
%1 = sext i32 %b to i64
%2 = add i64 %a, %1
%3 = inttoptr i64 %2 to i16*
[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
%4 = load i16, i16* %3
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%5 = zext i16 %4 to i64
ret i64 %5
}
define i64 @load_extend_zext_i32_to_i64(i64 %a, i32 %b) {
; CHECK-LABEL: load_extend_zext_i32_to_i64
; CHECK: ldr w0, [x0, w1, sxtw]
; CHECK-NOT: uxtw
%1 = sext i32 %b to i64
%2 = add i64 %a, %1
%3 = inttoptr i64 %2 to i32*
[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
%4 = load i32, i32* %3
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%5 = zext i32 %4 to i64
ret i64 %5
}
define i32 @load_extend_sext_i8_to_i32(i64 %a, i32 %b) {
; CHECK-LABEL: load_extend_sext_i8_to_i32
; CHECK: ldrsb w0, [x0, w1, sxtw]
; CHECK-NOT: sxtb
%1 = sext i32 %b to i64
%2 = add i64 %a, %1
%3 = inttoptr i64 %2 to i8*
[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
%4 = load i8, i8* %3
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%5 = sext i8 %4 to i32
ret i32 %5
}
define i32 @load_extend_sext_i16_to_i32(i64 %a, i32 %b) {
; CHECK-LABEL: load_extend_sext_i16_to_i32
; CHECK: ldrsh w0, [x0, w1, sxtw]
; CHECK-NOT: sxth
%1 = sext i32 %b to i64
%2 = add i64 %a, %1
%3 = inttoptr i64 %2 to i16*
[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
%4 = load i16, i16* %3
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%5 = sext i16 %4 to i32
ret i32 %5
}
define i64 @load_extend_sext_i8_to_i64(i64 %a, i32 %b) {
; CHECK-LABEL: load_extend_sext_i8_to_i64
; CHECK: ldrsb x0, [x0, w1, sxtw]
; CHECK-NOT: sxtb
%1 = sext i32 %b to i64
%2 = add i64 %a, %1
%3 = inttoptr i64 %2 to i8*
[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
%4 = load i8, i8* %3
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%5 = sext i8 %4 to i64
ret i64 %5
}
define i64 @load_extend_sext_i16_to_i64(i64 %a, i32 %b) {
; CHECK-LABEL: load_extend_sext_i16_to_i64
; CHECK: ldrsh x0, [x0, w1, sxtw]
; CHECK-NOT: sxth
%1 = sext i32 %b to i64
%2 = add i64 %a, %1
%3 = inttoptr i64 %2 to i16*
[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
%4 = load i16, i16* %3
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%5 = sext i16 %4 to i64
ret i64 %5
}
define i64 @load_extend_sext_i32_to_i64(i64 %a, i32 %b) {
; CHECK-LABEL: load_extend_sext_i32_to_i64
; CHECK: ldrsw x0, [x0, w1, sxtw]
; CHECK-NOT: sxtw
%1 = sext i32 %b to i64
%2 = add i64 %a, %1
%3 = inttoptr i64 %2 to i32*
[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
%4 = load i32, i32* %3
[FastISel][AArch64] Fix sign-/zero-extend folding when SelectionDAG is involved. Sign-/zero-extend folding depended on the load and the integer extend to be both selected by FastISel. This cannot always be garantueed and SelectionDAG might interfer. This commit adds additonal checks to load and integer extend lowering to catch this. Related to rdar://problem/18495928. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219716 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-14 20:36:02 +00:00
br label %bb2
bb2:
%5 = sext i32 %4 to i64
ret i64 %5
}
Reference in New Issue Copy Permalink