Skip to content

GH-117512: Allow 64-bit JIT operands on 32-bit platforms #117527

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
merged 2 commits into from
Apr 6, 2024
Merged

GH-117512: Allow 64-bit JIT operands on 32-bit platforms #117527

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f0797f9
Allow 64-bit operands on 32-bit platforms
brandtbucher Apr 3, 2024
4ae7401
Clean up some types
brandtbucher Apr 5, 2024
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
Next Next commit
Allow 64-bit operands on 32-bit platforms
  • Loading branch information
@brandtbucher
brandtbucher committed Apr 3, 2024
commit f0797f97042fbeaaa5db13784de0d6ac6481cff1
6 changes: 6 additions & 0 deletions Python/jit.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -416,7 +416,13 @@ _PyJIT_Compile(_PyExecutorObject *executor, const _PyUOpInstruction *trace, size
patches[HoleValue_DATA] = (uint64_t)data;
patches[HoleValue_EXECUTOR] = (uint64_t)executor;
patches[HoleValue_OPARG] = instruction->oparg;
#if SIZEOF_VOID_P == 8
patches[HoleValue_OPERAND] = instruction->operand;
#else
assert(SIZEOF_VOID_P == 4);
patches[HoleValue_OPERAND_HI] = instruction->operand >> 32;
Copy link
Member

Choose a reason for hiding this comment

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

Should patches become uintptr_t instead of uint64_t? (on all platforms).
Given many of the casts are of the form (uint64_t)some_pointer, changing them to (uintptr_t)some_pointer avoids unnecessary widening on 32 bit platforms.

patches[HoleValue_OPERAND_LO] = instruction->operand & UINT32_MAX;
#endif
switch (instruction->format) {
case UOP_FORMAT_TARGET:
patches[HoleValue_TARGET] = instruction->target;
Expand Down
5 changes: 4 additions & 1 deletion Tools/jit/_stencils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,8 +27,11 @@ class HoleValue(enum.Enum):
GOT = enum.auto()
# The current uop's oparg (exposed as _JIT_OPARG):
OPARG = enum.auto()
# The current uop's operand (exposed as _JIT_OPERAND):
# The current uop's operand on 64-bit platforms (exposed as _JIT_OPERAND):
OPERAND = enum.auto()
# The current uop's operand on 32-bit platforms (exposed as _JIT_OPERAND_HI and _JIT_OPERAND_LO):
OPERAND_HI = enum.auto()
OPERAND_LO = enum.auto()
# The current uop's target (exposed as _JIT_TARGET):
TARGET = enum.auto()
# The base address of the machine code for the jump target (exposed as _JIT_JUMP_TARGET):
Expand Down
7 changes: 7 additions & 0 deletions Tools/jit/template.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,7 +87,14 @@ _JIT_ENTRY(_PyInterpreterFrame *frame, PyObject **stack_pointer, PyThreadState *
int opcode = _JIT_OPCODE;
// Other stuff we need handy:
PATCH_VALUE(uint16_t, _oparg, _JIT_OPARG)
#if SIZEOF_VOID_P == 8
PATCH_VALUE(uint64_t, _operand, _JIT_OPERAND)
#else
assert(SIZEOF_VOID_P == 4);
PATCH_VALUE(uint32_t, _operand_hi, _JIT_OPERAND_HI)
PATCH_VALUE(uint32_t, _operand_lo, _JIT_OPERAND_LO)
uint64_t _operand = ((uint64_t)_operand_hi << 32) | _operand_lo;
#endif
PATCH_VALUE(uint32_t, _target, _JIT_TARGET)
PATCH_VALUE(uint16_t, _exit_index, _JIT_EXIT_INDEX)

Expand Down