Re: [Qemu-trivial] [PATCH 1/1] tci: eliminate UB due to unaligned reads

[Anatoly Trosinenko]

> So.  Why do you want to use TCI instead of a native TCG backend?

Frankly speaking, personally I just have a strange experiment on porting QEMU to _javascript_. :) I used the TCI bytecode as some intermediate patchable form for rarely executing BBs and for (re)generating asm.js from it when required. I used a Python script to generate wrappers with exactly 10 arguments around helper functions. In fact, it may be worth for me to create WebAssembly TCG backend and interpret **that** bytecode if required.

TCI may still be useful for someone else running QEMU on unsupported host, though.

2018-03-03 17:13 GMT+03:00 Richard Henderson <address@hidden>:

On 03/03/2018 06:07 AM, Anatoly Trosinenko wrote:
> Can rewriting TCI in such a way that every operation is aligned at 4- or even
> 8-byte boundary fix the situation or are there some more serious problems?

With the current TCI, there are also problems with calls to helper functions.
The only portable way to do this is to use a library such as libffi.

I once rewrote TCI completely in order to address both problems, but that only
brought questions as to why TCI is useful at all.

So.  Why do you want to use TCI instead of a native TCG backend?


r~

--

Best regards,

Anatoly