Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

8301313: RISC-V: C2: assert(false) failed: bad AD file due to missing match rule #12295

Closed
wants to merge 1 commit into from

Conversation

feilongjiang
Copy link
Member

@feilongjiang feilongjiang commented Jan 30, 2023

The RISC-V architecture does not support conditional move instructions or flags register for now. So we set ConditionalMoveLimit to 0 for this port to avoid the generation of the CMove node by C2. But turns out this could not avoid all CMove nodes generation: some CMove generations are not guarded by this parameter. Therefore, we still added several match rules in AD file for CMove nodes like CMoveI_CmpI/CMoveL_CmpL etc. to cover those scenarios:

// Conditional Move Instructions
instruct cmovI_cmpI(iRegINoSp dst, iRegI src, iRegI op1, iRegI op2, cmpOp cop) %{
match(Set dst (CMoveI (Binary cop (CmpI op1 op2)) (Binary dst src)));
ins_cost(ALU_COST + BRANCH_COST);
format %{
"CMove $dst, ($op1 $cop $op2), $dst, $src\t#@cmovI_cmpI\n\t"
%}
ins_encode %{
__ enc_cmove($cop$$cmpcode,
as_Register($op1$$reg), as_Register($op2$$reg),
as_Register($dst$$reg), as_Register($src$$reg));
%}
ins_pipe(pipe_class_compare);
%}
instruct cmovI_cmpU(iRegINoSp dst, iRegI src, iRegI op1, iRegI op2, cmpOpU cop) %{
match(Set dst (CMoveI (Binary cop (CmpU op1 op2)) (Binary dst src)));
ins_cost(ALU_COST + BRANCH_COST);
format %{
"CMove $dst, ($op1 $cop $op2), $dst, $src\t#@cmovI_cmpU\n\t"
%}
ins_encode %{
__ enc_cmove($cop$$cmpcode | C2_MacroAssembler::unsigned_branch_mask,
as_Register($op1$$reg), as_Register($op2$$reg),
as_Register($dst$$reg), as_Register($src$$reg));
%}
ins_pipe(pipe_class_compare);
%}
instruct cmovI_cmpL(iRegINoSp dst, iRegI src, iRegL op1, iRegL op2, cmpOp cop) %{
match(Set dst (CMoveI (Binary cop (CmpL op1 op2)) (Binary dst src)));
ins_cost(ALU_COST + BRANCH_COST);
format %{
"CMove $dst, ($op1 $cop $op2), $dst, $src\t#@cmovI_cmpL\n\t"
%}
ins_encode %{
__ enc_cmove($cop$$cmpcode,
as_Register($op1$$reg), as_Register($op2$$reg),
as_Register($dst$$reg), as_Register($src$$reg));
%}
ins_pipe(pipe_class_compare);
%}
instruct cmovL_cmpL(iRegLNoSp dst, iRegL src, iRegL op1, iRegL op2, cmpOp cop) %{
match(Set dst (CMoveL (Binary cop (CmpL op1 op2)) (Binary dst src)));
ins_cost(ALU_COST + BRANCH_COST);
format %{
"CMove $dst, ($op1 $cop $op2), $dst, $src\t#@cmovL_cmpL\n\t"
%}
ins_encode %{
__ enc_cmove($cop$$cmpcode,
as_Register($op1$$reg), as_Register($op2$$reg),
as_Register($dst$$reg), as_Register($src$$reg));
%}
ins_pipe(pipe_class_compare);
%}
instruct cmovL_cmpUL(iRegLNoSp dst, iRegL src, iRegL op1, iRegL op2, cmpOpU cop) %{
match(Set dst (CMoveL (Binary cop (CmpUL op1 op2)) (Binary dst src)));
ins_cost(ALU_COST + BRANCH_COST);
format %{
"CMove $dst, ($op1 $cop $op2), $dst, $src\t#@cmovL_cmpUL\n\t"
%}
ins_encode %{
__ enc_cmove($cop$$cmpcode | C2_MacroAssembler::unsigned_branch_mask,
as_Register($op1$$reg), as_Register($op2$$reg),
as_Register($dst$$reg), as_Register($src$$reg));
%}
ins_pipe(pipe_class_compare);
%}
instruct cmovI_cmpUL(iRegINoSp dst, iRegI src, iRegL op1, iRegL op2, cmpOpU cop) %{
match(Set dst (CMoveI (Binary cop (CmpUL op1 op2)) (Binary dst src)));
ins_cost(ALU_COST + BRANCH_COST);
format %{
"CMove $dst, ($op1 $cop $op2), $dst, $src\t#@cmovI_cmpUL\n\t"
%}
ins_encode %{
__ enc_cmove($cop$$cmpcode | C2_MacroAssembler::unsigned_branch_mask,
as_Register($op1$$reg), as_Register($op2$$reg),
as_Register($dst$$reg), as_Register($src$$reg));
%}
ins_pipe(pipe_class_compare);
%}

In this case, we observed that CMoveL was generated in PhaseIdealLoop::transform_long_range_checks, which is expected to match cmoveL_cmpL node:

// L_clamp = Q_min < 0 ? 0 : Q_min
Node* Q_min_cmp = new CmpLNode(Q_min, long_zero);
register_new_node(Q_min_cmp, entry_control);
Node* Q_min_bool = new BoolNode(Q_min_cmp, BoolTest::lt);
register_new_node(Q_min_bool, entry_control);
Node* L_clamp = new CMoveLNode(Q_min_bool, Q_min, long_zero, TypeLong::LONG);
register_new_node(L_clamp, entry_control);
// (This could also be coded bitwise as L_clamp = Q_min & ~(Q_min>>63).)

But PhaseIdealLoop::optimize after PhaseIdealLoop::transform_long_range_checks may replace CmpLNode with CmpINode in CmpLNode::Ideal:

Node *CmpLNode::Ideal( PhaseGVN *phase, bool can_reshape ) {
const TypeLong *t2 = phase->type(in(2))->isa_long();
if (Opcode() == Op_CmpL && in(1)->Opcode() == Op_ConvI2L && t2 && t2->is_con()) {
const jlong con = t2->get_con();
if (con >= min_jint && con <= max_jint) {
return new CmpINode(in(1)->in(1), phase->intcon((jint)con));
}
}
return NULL;
}

There is no match rule in riscv.ad for CMoveL with CmpI, which result in the bad AD file crash.

With this fix, we should be able to cover all other CMove scenarios not guarded by ConditionalMoveLimit.

Testing:

  • jdk_foreign with -XX:-TieredCompilation (linux-riscv64, fastdebug)
  • Tier1~3 on Unmatched board (linux-riscv64, release)

Progress

  • Change must be properly reviewed (1 review required, with at least 1 Reviewer)
  • Change must not contain extraneous whitespace
  • Commit message must refer to an issue

Issue

  • JDK-8301313: RISC-V: C2: assert(false) failed: bad AD file due to missing match rule

Reviewers

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk pull/12295/head:pull/12295
$ git checkout pull/12295

Update a local copy of the PR:
$ git checkout pull/12295
$ git pull https://git.openjdk.org/jdk pull/12295/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 12295

View PR using the GUI difftool:
$ git pr show -t 12295

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/12295.diff

@feilongjiang feilongjiang marked this pull request as draft January 30, 2023 13:58
@bridgekeeper
Copy link

bridgekeeper bot commented Jan 30, 2023

👋 Welcome back fjiang! A progress list of the required criteria for merging this PR into master will be added to the body of your pull request. There are additional pull request commands available for use with this pull request.

@openjdk
Copy link

openjdk bot commented Jan 30, 2023

@feilongjiang The following label will be automatically applied to this pull request:

  • hotspot-compiler

When this pull request is ready to be reviewed, an "RFR" email will be sent to the corresponding mailing list. If you would like to change these labels, use the /label pull request command.

@openjdk openjdk bot added the hotspot-compiler hotspot-compiler-dev@openjdk.org label Jan 30, 2023
@feilongjiang feilongjiang marked this pull request as ready for review January 30, 2023 14:45
@openjdk openjdk bot added the rfr Pull request is ready for review label Jan 30, 2023
@mlbridge
Copy link

mlbridge bot commented Jan 30, 2023

Webrevs

Copy link
Member

@RealFYang RealFYang left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks reasonable.

@openjdk
Copy link

openjdk bot commented Jan 31, 2023

@feilongjiang This change now passes all automated pre-integration checks.

ℹ️ This project also has non-automated pre-integration requirements. Please see the file CONTRIBUTING.md for details.

After integration, the commit message for the final commit will be:

8301313: RISC-V: C2: assert(false) failed: bad AD file due to missing match rule

Reviewed-by: fyang, yadongwang

You can use pull request commands such as /summary, /contributor and /issue to adjust it as needed.

At the time when this comment was updated there had been 83 new commits pushed to the master branch:

  • 4c9de87: 8301655: Problemlist jdk/jdk/nio/zipfs/TestLocOffsetFromZip64EF.java on Linux
  • 04278e6: 8301564: Non-C-heap allocated ResourceHashtable keys and values must have trivial destructor
  • b00b70c: 8286907: keytool should warn about weak PBE algorithms
  • ee0f5b5: 8301392: Port fdlibm log1p to Java
  • f696785: 8300869: Make use of the Double.toString(double) algorithm in java.util.Formatter
  • cf6b9eb: 8301637: ThreadLocalRandom.current().doubles().parallel() contention
  • c647ae6: 8301149: Parallel: Refactor MutableNUMASpace::update_layout
  • de57733: 8301644: com/sun/jdi/JdbStopThreadTest.java fails after JDK-8300811
  • 930ec00: 8301636: Minor cleanup in CommentHelper and DocPretty
  • 725d57b: 8301659: Resolve initialization reordering issues on Windows for libawt and libsaproc
  • ... and 73 more: https://git.openjdk.org/jdk/compare/ebb84ad70d3295d9a429904fcdacdb8ecd1bf434...master

As there are no conflicts, your changes will automatically be rebased on top of these commits when integrating. If you prefer to avoid this automatic rebasing, please check the documentation for the /integrate command for further details.

As you do not have Committer status in this project an existing Committer must agree to sponsor your change. Possible candidates are the reviewers of this PR (@RealFYang) but any other Committer may sponsor as well.

➡️ To flag this PR as ready for integration with the above commit message, type /integrate in a new comment. (Afterwards, your sponsor types /sponsor in a new comment to perform the integration).

@openjdk openjdk bot added the ready Pull request is ready to be integrated label Jan 31, 2023
@yadongw
Copy link
Contributor

yadongw commented Feb 2, 2023

LGTM. But with ConditionalMoveLimit of 0, can we guarantee all the CMoves covered by these match rules?

@feilongjiang
Copy link
Member Author

feilongjiang commented Feb 2, 2023

LGTM. But with ConditionalMoveLimit of 0, can we guarantee all the CMoves covered by these match rules?

Yes, this pr added CMoveL_CmpI and CMoveL_CmpU, combined with existing CMove instructs, we should be able to cover all other CMove scenarios not guarded by ConditionalMoveLimit.

@feilongjiang
Copy link
Member Author

@RealFYang @yadongw -- Thanks!
/integrate

@openjdk openjdk bot added the sponsor Pull request is ready to be sponsored label Feb 3, 2023
@openjdk
Copy link

openjdk bot commented Feb 3, 2023

@feilongjiang
Your change (at version a296e72) is now ready to be sponsored by a Committer.

@RealFYang
Copy link
Member

/sponsor

@openjdk
Copy link

openjdk bot commented Feb 3, 2023

Going to push as commit 3ad6aef.
Since your change was applied there have been 83 commits pushed to the master branch:

  • 4c9de87: 8301655: Problemlist jdk/jdk/nio/zipfs/TestLocOffsetFromZip64EF.java on Linux
  • 04278e6: 8301564: Non-C-heap allocated ResourceHashtable keys and values must have trivial destructor
  • b00b70c: 8286907: keytool should warn about weak PBE algorithms
  • ee0f5b5: 8301392: Port fdlibm log1p to Java
  • f696785: 8300869: Make use of the Double.toString(double) algorithm in java.util.Formatter
  • cf6b9eb: 8301637: ThreadLocalRandom.current().doubles().parallel() contention
  • c647ae6: 8301149: Parallel: Refactor MutableNUMASpace::update_layout
  • de57733: 8301644: com/sun/jdi/JdbStopThreadTest.java fails after JDK-8300811
  • 930ec00: 8301636: Minor cleanup in CommentHelper and DocPretty
  • 725d57b: 8301659: Resolve initialization reordering issues on Windows for libawt and libsaproc
  • ... and 73 more: https://git.openjdk.org/jdk/compare/ebb84ad70d3295d9a429904fcdacdb8ecd1bf434...master

Your commit was automatically rebased without conflicts.

@openjdk openjdk bot added the integrated Pull request has been integrated label Feb 3, 2023
@openjdk openjdk bot closed this Feb 3, 2023
@openjdk openjdk bot removed ready Pull request is ready to be integrated rfr Pull request is ready for review sponsor Pull request is ready to be sponsored labels Feb 3, 2023
@openjdk
Copy link

openjdk bot commented Feb 3, 2023

@RealFYang @feilongjiang Pushed as commit 3ad6aef.

💡 You may see a message that your pull request was closed with unmerged commits. This can be safely ignored.

@feilongjiang feilongjiang deleted the fix_cmove_bad_ad branch February 3, 2023 06:06
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
hotspot-compiler hotspot-compiler-dev@openjdk.org integrated Pull request has been integrated
3 participants