gh-131798: JIT: optimize _LOAD_COMMON_CONSTANT

[trdthg]

Try to learn and make a small contribution :D

• Issue: Better uop coverage in the JIT optimizer #131798

[bedevere-app]

Most changes to Python require a NEWS entry. Add one using the blurb_it web app or the blurb command-line tool.

If this change has little impact on Python users, wait for a maintainer to apply the skip news label instead.

[python-cla-bot]

All commit authors signed the Contributor License Agreement.

[Fidget-Spinner]

Thanks, this LGTM! Please remove the news entry, we don't need it as it's not a user facing code.

[bedevere-app]

Most changes to Python require a NEWS entry. Add one using the blurb_it web app or the blurb command-line tool.

If this change has little impact on Python users, wait for a maintainer to apply the skip news label instead.

[Fidget-Spinner]

_LOAD_COMMON_CONSTANT x86_64 dissasembly in jit_stencils-x86_64-unknown-linux-gnu.h:

 // 0000000000000000 <_JIT_ENTRY>:
// 0: 49 8b 47 10 movq 0x10(%r15), %rax
// 4: 48 b9 00 00 00 00 00 00 00 00 movabsq $0x0, %rcx
// 0000000000000006: R_X86_64_64 _JIT_OPARG
// e: 0f b7 c9 movzwl %cx, %ecx
// 11: 48 8b bc c8 60 68 03 00 movq 0x36860(%rax,%rcx,8), %rdi
// 19: 8b 07 movl (%rdi), %eax
// 1b: 85 c0 testl %eax, %eax
// 1d: 78 06 js 0x25 <_JIT_ENTRY+0x25>
// 1f: ff c0 incl %eax
// 21: 89 07 movl %eax, (%rdi)
// 23: eb 04 jmp 0x29 <_JIT_ENTRY+0x29>
// 25: 48 83 cf 01 orq $0x1, %rdi

I think this is prior to our CFG optimizations in the backend, so it's not 100% accurate, but pretty close.

_LOAD_CONST_INLINE_BORROW:

 // // _LOAD_CONST_INLINE_BORROW_r01.o: file format elf64-x86-64
// // Disassembly of section .text:
// // 0000000000000000 <_JIT_ENTRY>:
// 0: 48 bf 00 00 00 00 00 00 00 00 movabsq $0x0, %rdi
// 0000000000000002: R_X86_64_64 _JIT_OPERAND0
// a: 48 83 cf 01 orq $0x1, %rdi

This is a really nice reduction. Good job!

[Fidget-Spinner]

We generally don't force push, because all PRs in CPython are squash-merged. So just push normal commits, thanks!

[Fidget-Spinner]

Once again, LGTM. Thanks for this!