Skip to content

GH-135904: Implement assembler optimization for AArch64. #139855

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.

Sign up for GitHub

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

Merged
markshannon merged 5 commits into from
Oct 17, 2025
Merged

GH-135904: Implement assembler optimization for AArch64. #139855

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
4abdaa1
Implement assembler optimization for AArch64.
markshannon Oct 2, 2025
b39e875
AArch64: Don't optimize 14 bit jumps to avoid failures when patching.…
markshannon Oct 10, 2025
87015f3
Add new relocation to match statement for Windows
markshannon Oct 10, 2025
3cc7eed
Merge branch 'main' into optimize-jit-cfg-arm
markshannon Oct 15, 2025
80ea07a
Mach-O does not support 19 bit branches
markshannon Oct 15, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Mach-O does not support 19 bit branches
  • Loading branch information
@markshannon
markshannon committed Oct 15, 2025
commit 80ea07aa5cc131370c18aadae84b2ac5fd02076e
16 changes: 0 additions & 16 deletions Python/jit.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -334,7 +334,6 @@ patch_aarch64_21rx(unsigned char *location, uint64_t value)
patch_aarch64_21r(location, value);
}


// 21-bit relative branch.
void
patch_aarch64_19r(unsigned char *location, uint64_t value)
Expand All @@ -350,21 +349,6 @@ patch_aarch64_19r(unsigned char *location, uint64_t value)
set_bits(loc32, 5, value, 2, 19);
}

// 16-bit relative branch.
void
patch_aarch64_14r(unsigned char *location, uint64_t value)
{
uint32_t *loc32 = (uint32_t *)location;
assert(IS_AARCH64_TEST_AND_BRANCH(*loc32));
value -= (uintptr_t)location;
// Check that we're not out of range of 16 signed bits:
assert((int64_t)value >= -(1 << 15));
assert((int64_t)value < (1 << 15));
// Since instructions are 4-byte aligned, only use 14 bits:
assert(get_bits(value, 0, 2) == 0);
set_bits(loc32, 5, value, 2, 14);
}

// 28-bit relative branch.
void
patch_aarch64_26r(unsigned char *location, uint64_t value)
Expand Down
10 changes: 9 additions & 1 deletion Tools/jit/_optimizers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -311,8 +311,16 @@ def run(self) -> None:
self.path.write_text(self._body())


# Mach-O does not support the 19 bit branch locations needed for branch reordering
class OptimizerAArch64_MachO(Optimizer): # pylint: disable = too-few-public-methods
"""aarch64-apple-darwin"""

# https://developer.arm.com/documentation/ddi0602/2025-03/Base-Instructions/B--Branch-
_re_jump = re.compile(r"\s*b\s+(?P<target>[\w.]+)")


class OptimizerAArch64(Optimizer): # pylint: disable = too-few-public-methods
"""aarch64-apple-darwin/aarch64-pc-windows-msvc/aarch64-unknown-linux-gnu"""
"""aarch64-pc-windows-msvc/aarch64-unknown-linux-gnu"""

_branches = _AARCH64_BRANCHES
_re_branch = re.compile(
Expand Down
2 changes: 1 addition & 1 deletion Tools/jit/_targets.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -565,7 +565,7 @@ def get_target(host: str) -> _COFF32 | _COFF64 | _ELF | _MachO:
if re.fullmatch(r"aarch64-apple-darwin.*", host):
host = "aarch64-apple-darwin"
condition = "defined(__aarch64__) && defined(__APPLE__)"
optimizer = _optimizers.OptimizerAArch64
optimizer = _optimizers.OptimizerAArch64_MachO
target = _MachO(host, condition, optimizer=optimizer)
elif re.fullmatch(r"aarch64-pc-windows-msvc", host):
host = "aarch64-pc-windows-msvc"
Expand Down
Loading