RFR: 8325467: Support methods with many arguments in C2 [v32]

Tue Sep 23 14:35:08 UTC 2025

On Fri, 19 Sep 2025 16:02:35 GMT, Daniel Lundén <dlunden at openjdk.org> wrote:

>> If a method has a large number of parameters, we currently bail out from C2 compilation.
>> 
>> ### Changeset
>> 
>> Allowing C2 compilation of methods with a large number of parameters requires fundamental changes to the register mask data structure, used in many places in C2. In particular, register masks currently have a statically determined size and cannot represent arbitrary numbers of stack slots. This is needed if we want to compile methods with arbitrary numbers of parameters. Register mask operations are present in performance-sensitive parts of C2, which further complicates changes.
>> 
>> Changes:
>> - Add functionality to dynamically grow/extend register masks. I experimented with a number of design choices to achieve this. To keep the common case (normal number of method parameters) quick and also to avoid more intrusive changes to the current `RegMask` interface, I decided to leave the "base" statically allocated memory for masks unchanged and only use dynamically allocated memory in the rare cases where it is needed.
>> - Generalize the "chunk"-logic from `PhaseChaitin::Select()` to allow arbitrary-sized chunks, and also move most of the logic into register mask methods to separate concerns and to make the `PhaseChaitin::Select()` code more readable.
>> - Remove all `can_represent` checks and bailouts.
>> - Performance tuning. A particularly important change is the early-exit optimization in `RegMask::overlap`, used in the performance-sensitive method `PhaseChaitin::interfere_with_live`.
>> - Add a new test case `TestManyMethodArguments.java` and extend an old test `TestNestedSynchronize.java`.
>> 
>> ### Testing
>> 
>> - [GitHub Actions](https://github.com/dlunde/jdk/actions/runs/10178060450)
>> - `tier1` to `tier4` (and additional Oracle-internal testing) on Windows x64, Linux x64, Linux aarch64, macOS x64, and macOS aarch64.
>> - Standard performance benchmarking. No observed conclusive overall performance degradation/improvement.
>> - Specific benchmarking of C2 compilation time. The changes increase C2 compilation time by, approximately and on average, 1% for methods that could also be compiled before this changeset (see the figure below). The reason for the degradation is further checks required in performance-sensitive code (in particular `PhaseChaitin::remove_bound_register_from_interfering_live_ranges`). I have tried optimizing in various ways, but changes I found that lead to improvement also lead to less readable code (and are, in my opinion, no...
>
> Daniel Lundén has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Increase timeout for TestMethodArguments.java

Looks good to me, thanks for the good work done here, Daniel! I only have a couple of minor suggestions.

src/hotspot/share/opto/regmask.hpp line 89:

> 87: 
> 88:   // RM_SIZE_IN_INTS_MIN, but in number of machine words
> 89:   static const unsigned int RM_SIZE_IN_WORDS_MIN =

This appears to be unused, consider removing.

src/hotspot/share/opto/regmask.hpp line 198:

> 196:   // (for a made-up platform with 10 registers and 4-bit
> 197:   // words) that has been extended with two additional words to represent more
> 198:   // stack locations:

Suggestion:

  // (for a made-up platform with 10 registers and 4-bit words) that has been
  // extended with two additional words to represent more stack locations:

src/hotspot/share/opto/regmask.hpp line 417:

> 415: 
> 416:   // A constructor only used by the ADLC output.  All mask fields are filled
> 417:   // in directly.  Calls to this look something like RM(1,2,3,4);

Consider updating this comment after the introduction of the `infinite_stack` parameter:

Suggestion:

  // in directly.  Calls to this look something like RM(0xc0, 0x0, 0x0, false);

test/hotspot/jtreg/compiler/arguments/TestMaxMethodArguments.java line 63:

> 61: 
> 62:     public static int test(int x1, int x2, int x3, int x4, int x5, int x6, int x7, int x8, int x9, int x10, int x11, int x12, int x13, int x14, int x15, int x16, int x17, int x18, int x19, int x20, int x21, int x22, int x23, int x24, int x25, int x26, int x27, int x28, int x29, int x30, int x31, int x32, int x33, int x34, int x35, int x36, int x37, int x38, int x39, int x40, int x41, int x42, int x43, int x44, int x45, int x46, int x47, int x48, int x49, int x50, int x51, int x52, int x53, int x54, int x55, int x56, int x57, int x58, int x59, int x60, int x61, int x62, int x63, int x64, int x65, int x66, int x67, int x68, int x69, int x70, int x71, int x72, int x73, int x74, int x75, int x76, int x77, int x78, int x79, int x80, int x81, int x82, int x83, int x84, int x85, int x86, int x87, int x88, int x89, int x90, int x91, int x92, int x93, int x94, int x95, int x96, int x97, int x98, int x99, int x100, int x101, int x102, int x103, int x104, int x105, int x106, int x107, int
  x108, int x109, int x110, int x111, int x112, int x113, int x114, int x115, int x116, int x117, int x118, int x119, int x120, int x121, int x122, int x123, int x124, int x125, int x126, int x127, int x128, int x129, int x130, int x131, int x132, int x133, int x134, int x135, int x136, int x137, int x138, int x139, int x140, int x141, int x142, int x143, int x144, int x145, int x146, int x147, int x148, int x149, int x150, int x151, int x152, int x153, int x154, int x155, int x156, int x157, int x158, int x159, int x160, int x161, int x162, int x163, int x164, int x165, int x166, int x167, int x168, int x169, int x170, int x171, int x172, int x173, int x174, int x175, int x176, int x177, int x178, int x179, int x180, int x181, int x182, int x183, int x184, int x185, int x186, int x187, int x188, int x189, int x190, int x191, int x192, int x193, int x194, int x195, int x196, int x197, int x198, int x199, int x200, int x201, int x202, int x203, int x204, int x205, int x206, int x207, 
 int x208, int x209, int x210, int x211, int x212, int x213, int x214, int x215, int x216, int x217, int x218, int x219, int x220, int x221, int x222, int x223, int x224, int x225, int x226, int x227, int x228, int x229, int x230, int x231, int x232, int x233, int x234, int x235, int x236, int x237, int x238, int x239, int x240, int x241, int x242, int x243, int x244, int x245, int x246, int x247, int x248, int x249, int x250, int x251, int x252, int x253, int x254, int x255) throws TestException {
> 63:         // Exceptions after every definition of a temporary forces the

Suggestion:

        // Exceptions after every definition of a temporary force the

-------------

Marked as reviewed by rcastanedalo (Reviewer).

PR Review: https://git.openjdk.org/jdk/pull/20404#pullrequestreview-3258293347
PR Review Comment: https://git.openjdk.org/jdk/pull/20404#discussion_r2372522663
PR Review Comment: https://git.openjdk.org/jdk/pull/20404#discussion_r2372525679
PR Review Comment: https://git.openjdk.org/jdk/pull/20404#discussion_r2372529815
PR Review Comment: https://git.openjdk.org/jdk/pull/20404#discussion_r2372531549