LibWasm: Try really hard to avoid touching the value stack

Browse source

This commit adds a register allocator, with 8 available "register"
slots.
In testing with various random blobs, this moves anywhere from 30% to
74% of value accesses into predefined slots, and is about a ~20% perf
increase end-to-end.

To actually make this usable, a few structural changes were also made:
- we no longer do one instruction per interpret call
- trapping is an (unlikely) exit condition
- the label and frame stacks are replaced with linked lists with a huge
  node cache size, as we only need to touch the last element and
  push/pop is very frequent.

...

This commit is contained in:

Ali Mohammad Pur 2025-08-02 20:30:44 +02:00 • committed by Ali Mohammad Pur

parent 998454028c

commit 0e5ecef848

Notes: github-actions[bot] 2025-08-08 10:55:57 +00:00

Author: https://github.com/alimpfard
Commit: 0e5ecef848
Pull-request: https://github.com/LadybirdBrowser/ladybird/pull/5060
Reviewed-by: https://github.com/Hendiadyoin1
Reviewed-by: https://github.com/R-Goc
Reviewed-by: https://github.com/awesomekling
Reviewed-by: https://github.com/gmta

9 changed files with 3217 additions and 1824 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download patch file Download diff file Expand all files Collapse all files

3904

Libraries/LibWasm/AbstractMachine/BytecodeInterpreter.cpp

View file

File diff suppressed because it is too large Load diff