Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
* ⚡ Add the folowing optimizations (1) - ♻️ Simplify `restore_cpu_core` assembly a bit - ⚡ Unwind becomes a jump table * ⚡ Jump table working. May have bugs * ⚡ [[gnu::always_inline]] Many functions - ⚡ Carve out a branch for decoding uleb128 call-site information, the most common and encouraged encoding. - 🐛 Add `lr` and `pc` to clobber list to ensure that the compiler does not attempt to use the `lr` register to hold the pointer to the virtual_cpu structure - ⚡ Move index entry definitions to the header * 🐛 Get multi levels demo to work * ⚡ Many fixes and improvements * 🚧 Turn on destructors (currently broken) * About this code I was trying to make a version of the `enter_function()` function that has less branches. Trading space for speed. * ⚡ Add enter_function2() `enter_function2()` will be a replacement for `enter_function()`. This function correctly decodes the call sites using their encoding values and decodes the action table and type table correctly as well. Although the type table parsing is still assumed to be pcrel. This should be fixed at some point. * 💩 unwind_frame2 The idea of unwind_frame2 was to evaluate the instructions in place vs putting them into order in the instruction_t. The problem is word boundaries which requires logic to determine if we are at the end of the instructions. * 🔥 Remove some uneeded unwind code * ⚡ Cache & Use the personality pointer * 🗑️ Remove old unneeded code * ⚡ Remove usage of personality in create_instructions_from_entry * 📝 Add todo w/ github issues everywhere in the code - Remove a few `#if 0` code - Slightly fix up issues with CI
- Loading branch information