merry comments

Results 76 comments of


                                            merry

Build Errors? "error: no match for ‘operator=’"

[Boost ticket #12802](https://svn.boost.org/trac/boost/ticket/12802). Regression in Boost 1.63.

ARM64 Backend

~~Reminder: Handle fpcr_controlled in FPVector.~~ Done.

FRECPE improvements

Might be worth converting singles to double to avoid the need for denormal detection in that case.

spin_lock: Add `waitpkg`-based locking

`umonitor` probably needs to be within the spinloop, but i'm not sure here. (i can imagine a context switch invalidating monitor state.) i can also imagine senarios depending on where...

spin_lock: Add `waitpkg`-based locking

reminder to self: use of `eax:edx` in context where they may not be avaliable for use

decoder/arm: Make CLZ required encoding bits more relaxed

Note to self: Bits 8-11 and 16-19 are SBO. TODO: This breaks tests because unicorn doesn't recognise these encodings.

InvalidateCacheRange does not update fast dispatch table

Thanks for digging into this. > `Unpatch` is not virtual `Unpatch` is virtual. > since `EmitX64::InvalidateBasicBlocks` only calls it if the address has entries in `patch_information` This is the logic...

InvalidateCacheRange does not update fast dispatch table

Reproduced in a small test case, unfortunately the bug seems to be more complex than above. Looking into this. ```cpp TEST_CASE("ensure fast dispatch entry is cleared even when a block...

InvalidateCacheRange does not update fast dispatch table

Found a second long-standing bug in the fast-dispatch lookup code.