Re: Observations from a bytecode validation feasibility experiment

Luiz Henrique de Figueiredo <[email protected]>
Newsgroups gmane.comp.lang.lua.general
Message-ID <CABt16qngkwZAwk1R3Br3FDUa22nhZ=kAw4Un_apKw62A7qmKPA@mail.gmail.com>
> My main conclusion is that it's
> gotten a lot harder, and a complete verifier would likely be as big as
> (if not larger than) the luaV_execute core loop.

A complete verifier is a hopeless task (halting problem etc.).
We dropped even the simple integrity test in Lua 4.0, way back in 2000.

> Many 5.1/5.2 validators relied on luaP_opmodes to handle most opcodes
> generically, with a switch/case only covering the more complex cases.

We did that for bytecode listing in luac.c, but it was not clear.
Since Lua 5.4 luac.c uses a complete switch; it's much easier to
maintain even if there is much repetition.

> It's clear that bytecode validation isn't a priority for the community,

It's an impossible task.
Our take is that if you don't trust bytecode, don't allow Lua to load it.
--lhf

-- 
You received this message because you are subscribed to the Google Groups "lua-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To view this discussion visit https://groups.google.com/d/msgid/lua-l/CABt16qngkwZAwk1R3Br3FDUa22nhZ%3DkAw4Un_apKw62A7qmKPA%40mail.gmail.com.
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.