AArch64: Implement arraytranslateTRTO255 #7499
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -6400,10 +6400,86 @@ OMR::ARM64::TreeEvaluator::arraytranslateAndTestEvaluator(TR::Node *node, TR::Co | |
|
|
||
| TR::Register * | ||
| OMR::ARM64::TreeEvaluator::arraytranslateEvaluator(TR::Node *node, TR::CodeGenerator *cg) | ||
| { | ||
| // tree looks as follows: | ||
| // arraytranslate | ||
| // (0) input ptr | ||
| // (1) output ptr | ||
| // (2) translation table (dummy) | ||
| // (3) stop character (terminal character, either 0xff00ff00 (ISO8859) or 0xff80ff80 (ASCII) | ||
| // (4) input length (in elements) | ||
| // (5) stopping char (dummy) | ||
|
There was a problem hiding this comment. Are the dummy nodes used by other architectures? Just wondering why they're dummy on AArch64. There was a problem hiding this comment. It is different between p/x/aarch64 and z. See the code below: omr/compiler/optimizer/ValuePropagationCommon.cpp Lines 2535 to 2565 in 205c540 The code for z from line 2543 generates a non-dummy node for child(2) at line 2562, while the non-z path generates a dummy There was a problem hiding this comment. An example of the |
||
| // | ||
| // Number of translated elements is returned | ||
|
|
||
| TR::Compilation *comp = cg->comp(); | ||
|
|
||
| TR_ASSERT_FATAL(!node->isSourceByteArrayTranslate(), "Source is byte[] for arraytranslate"); | ||
| TR_ASSERT_FATAL(node->isTargetByteArrayTranslate(), "Target is char[] for arraytranslate"); | ||
| TR_ASSERT_FATAL(node->getChild(3)->getOpCodeValue() == TR::iconst && node->getChild(3)->getInt() == 0x0ff00ff00, "Non-ISO8859 stop character for arraytranslate"); | ||
|
There was a problem hiding this comment. What is child(3) used for? It does not appear to be used in this evaluator, so is this assert necessary? There was a problem hiding this comment. child(3) is used for the Do you want to also review |
||
|
|
||
| static bool verboseArrayTranslate = (feGetEnv("TR_verboseArrayTranslate") != NULL); | ||
| if (verboseArrayTranslate) | ||
| { | ||
| fprintf(stderr, "arrayTranslateTRTO255: %s @ %s\n", | ||
| comp->signature(), | ||
| comp->getHotnessName(comp->getMethodHotness()) | ||
| ); | ||
| } | ||
|
|
||
| TR::Register *inputReg = cg->gprClobberEvaluate(node->getChild(0)); | ||
| TR::Register *outputReg = cg->gprClobberEvaluate(node->getChild(1)); | ||
| TR::Register *inputLenReg = cg->gprClobberEvaluate(node->getChild(4)); | ||
| TR::Register *outputLenReg = cg->allocateRegister(); | ||
|
There was a problem hiding this comment. Children 2,3,5 still need to be evaluated even if they are "dummy". There was a problem hiding this comment. Arraytranslate evaluators of p and x platforms leave those dummy children unevaluated.
Reference counts of all the child nodes are decremented. |
||
|
|
||
| int numDeps = 10; | ||
|
|
||
| TR::RegisterDependencyConditions *deps = new (cg->trHeapMemory()) TR::RegisterDependencyConditions(1, numDeps, cg->trMemory()); | ||
|
|
||
| deps->addPreCondition(inputReg, TR::RealRegister::x0); | ||
|
|
||
| deps->addPostCondition(outputLenReg, TR::RealRegister::x0); | ||
| deps->addPostCondition(outputReg, TR::RealRegister::x1); | ||
| deps->addPostCondition(inputLenReg, TR::RealRegister::x2); | ||
|
There was a problem hiding this comment. Register |
||
|
|
||
| // Clobbered by the helper | ||
| TR::Register *clobberedReg; | ||
| deps->addPostCondition(clobberedReg = cg->allocateRegister(), TR::RealRegister::x4); | ||
| cg->stopUsingRegister(clobberedReg); | ||
| deps->addPostCondition(clobberedReg = cg->allocateRegister(), TR::RealRegister::x5); | ||
| cg->stopUsingRegister(clobberedReg); | ||
| deps->addPostCondition(clobberedReg = cg->allocateRegister(), TR::RealRegister::x6); | ||
| cg->stopUsingRegister(clobberedReg); | ||
|
|
||
| deps->addPostCondition(clobberedReg = cg->allocateRegister(TR_VRF), TR::RealRegister::v0); | ||
| cg->stopUsingRegister(clobberedReg); | ||
| deps->addPostCondition(clobberedReg = cg->allocateRegister(TR_VRF), TR::RealRegister::v1); | ||
| cg->stopUsingRegister(clobberedReg); | ||
| deps->addPostCondition(clobberedReg = cg->allocateRegister(TR_VRF), TR::RealRegister::v2); | ||
| cg->stopUsingRegister(clobberedReg); | ||
|
|
||
| // Array Translate helper call | ||
| TR_RuntimeHelper helper = TR_ARM64arrayTranslateTRTO255; | ||
| TR::SymbolReference *helperSym = cg->symRefTab()->findOrCreateRuntimeHelper(helper); | ||
| uintptr_t addr = reinterpret_cast<uintptr_t>(helperSym->getMethodAddress()); | ||
| generateImmSymInstruction(cg, TR::InstOpCode::bl, node, addr, deps, helperSym, NULL); | ||
|
There was a problem hiding this comment. What you've provided here is essentially a stub that calls a helper that must be implemented by any downstream project that wants to support arraytranslate opcodes. You should either have a comment at the beginning of this evaluator explaining this, or you should consider moving the entire evaluator into the downstream project itself. There was a problem hiding this comment. I moved the helper implementation to OMR as discussed yesterday. |
||
|
|
||
| for (uint32_t i = 0; i < node->getNumChildren(); i++) | ||
| cg->decReferenceCount(node->getChild(i)); | ||
|
|
||
| if (inputReg != node->getChild(0)->getRegister()) | ||
| cg->stopUsingRegister(inputReg); | ||
|
|
||
| if (outputReg != node->getChild(1)->getRegister()) | ||
| cg->stopUsingRegister(outputReg); | ||
|
|
||
| if (inputLenReg != node->getChild(4)->getRegister()) | ||
| cg->stopUsingRegister(inputLenReg); | ||
|
|
||
| cg->machine()->setLinkRegisterKilled(true); | ||
| node->setRegister(outputLenReg); | ||
| return outputLenReg; | ||
| } | ||
|
|
||
| TR::Register * | ||
| OMR::ARM64::TreeEvaluator::arraysetEvaluator(TR::Node *node, TR::CodeGenerator *cg) | ||
|
|
||
File renamed without changes.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,130 @@ | ||
| /******************************************************************************* | ||
| * Copyright IBM Corp. and others 2024 | ||
| * | ||
| * This program and the accompanying materials are made available under | ||
| * the terms of the Eclipse Public License 2.0 which accompanies this | ||
| * distribution and is available at https://www.eclipse.org/legal/epl-2.0/ | ||
| * or the Apache License, Version 2.0 which accompanies this distribution | ||
| * and is available at https://www.apache.org/licenses/LICENSE-2.0. | ||
| * | ||
| * This Source Code may also be made available under the following Secondary | ||
| * Licenses when the conditions for such availability set forth in the | ||
| * Eclipse Public License, v. 2.0 are satisfied: GNU General Public License, | ||
| * version 2 with the GNU Classpath Exception [1] and GNU General Public | ||
| * License, version 2 with the OpenJDK Assembly Exception [2]. | ||
| * | ||
| * [1] https://www.gnu.org/software/classpath/license.html | ||
| * [2] https://openjdk.org/legal/assembly-exception.html | ||
| * | ||
| * SPDX-License-Identifier: EPL-2.0 OR Apache-2.0 OR GPL-2.0-only WITH Classpath-exception-2.0 OR GPL-2.0-only WITH OpenJDK-assembly-exception-1.0 | ||
| *******************************************************************************/ | ||
|
|
||
| .file "ARM64ArrayTranslate.s" | ||
|
|
||
| #include "aarch64/runtime/arm64asmdefs.inc" | ||
|
|
||
| .globl FUNC_LABEL(__arrayTranslateTRTO255) | ||
|
|
||
| .text | ||
| .align 2 | ||
|
|
||
| // ---- | ||
| // arrayTranslateTRTO255 | ||
| // ---- | ||
| // TO stands for Two bytes to One byte | ||
| // | ||
| // uint16 input[]; | ||
| // uint8 output[]; | ||
| // int32 len; | ||
| // | ||
| // int32 i = 0; | ||
| // for (i < len) { | ||
| // uint16 ch = input[i]; | ||
| // if (ch > 0xFF) break; | ||
| // output[i] = ch & 0xFF; | ||
| // i++; | ||
| // } | ||
| // return i; | ||
| // | ||
| // in: x0: input | ||
| // x1: output | ||
| // x2: len | ||
| // out: x0: num of translated elements | ||
| // trash: x4-x6, v0-v2 | ||
|
|
||
| FUNC_LABEL(__arrayTranslateTRTO255): | ||
| // preserve output address | ||
| mov x6, x1 | ||
| cmp w2, #16 | ||
| b.cc atTRTO255_15 | ||
| lsr w4, w2, #4 | ||
| atTRTO255_16Loop: | ||
| // load 16 elements | ||
| ldp q0, q1, [x0] | ||
| // collect upper 8 bits | ||
| uzp2 v2.16b, v0.16b, v1.16b | ||
| // fail when any one of them is non-zero | ||
| umaxp v2.4s, v2.4s, v2.4s | ||
| mov x5, v2.D[0] | ||
| cbnz x5, atTRTO255_Fail | ||
| // collect lower 8 bits | ||
| uzp1 v2.16b, v0.16b, v1.16b | ||
| add x0, x0, #32 | ||
| subs w4, w4, #1 | ||
| // store 16 elements | ||
| str q2, [x1], #16 | ||
| b.ne atTRTO255_16Loop | ||
| atTRTO255_15: | ||
| // 15 elements or less remaining | ||
| tst w2, #8 | ||
| b.eq atTRTO255_7 | ||
| // load 8 elements | ||
| ldr q0, [x0] | ||
| // collect upper 8 bits | ||
| trn2 v2.16b, v0.16b, v0.16b | ||
| // fail when any one of them is non-zero | ||
| umaxp v2.4s, v2.4s, v2.4s | ||
| mov x5, v2.D[0] | ||
| cbnz x5, atTRTO255_Fail | ||
| // collect lower 8 bits | ||
| xtn v2.8b, v0.8h | ||
| add x0, x0, #16 | ||
| // store 8 elements | ||
| str d2, [x1], #8 | ||
| atTRTO255_7: | ||
| // 7 elements or less remaining | ||
| tst w2, #4 | ||
| b.eq atTRTO255_3 | ||
| // load 4 elements | ||
| ldr d0, [x0] | ||
| // collect upper 8 bits | ||
| trn2 v2.8b, v0.8b, v0.8b | ||
| // fail when any one of them is non-zero | ||
| mov x5, v2.D[0] | ||
| cbnz x5, atTRTO255_Fail | ||
| // collect lower 8 bits | ||
| xtn v2.8b, v0.8h | ||
| add x0, x0, #8 | ||
| // store 4 elements | ||
| str s2, [x1], #4 | ||
| atTRTO255_3: | ||
| // 3 elements or less remaining | ||
| ands w4, w2, #3 | ||
| atTRTO255_1Loop: | ||
| b.eq atTRTO255_Done | ||
| ldrh w5, [x0], #2 | ||
| cmp w5, #256 | ||
| b.cs atTRTO255_Done | ||
| subs w4, w4, #1 | ||
| strb w5, [x1], #1 | ||
| b atTRTO255_1Loop | ||
| atTRTO255_Fail: | ||
| ldrh w5, [x0], #2 | ||
| cmp w5, #256 | ||
| b.cs atTRTO255_Done | ||
| strb w5, [x1], #1 | ||
| b atTRTO255_Fail | ||
| atTRTO255_Done: | ||
| // number of translated elements | ||
| sub x0, x1, x6 | ||
| ret |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't think support for this opcode should be enabled in OMR when there isn't an implementation for it in OMR (the helper implementation is missing). In the absence of that, this logic should be enabled in the downstream projects that do provide the helper.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Both p and x platforms do the same: Implement the helper functions in the downstream project, and enable the feature in OMR.
On the other hand, z inlines the vectorized code instead of calling helper functions.
I can move the implementation of the AArch64 helper functions to OMR. In that case, maybe we want to move the helper functions for p and x to OMR for consistency.
[p]
omr/compiler/p/codegen/OMRCodeGenerator.cpp
Lines 250 to 262 in 205c540
[x]
omr/compiler/x/codegen/OMRCodeGenerator.cpp
Lines 424 to 447 in 205c540