Re: LuaJIT optimization bug?

[Mike Pall <mikelu-1108@...>]

Pierre Chapuis wrote:
> It's actually worse than I thought... It works with 3, 6 or
> 19 successive assert(true) statements. It doesn't work for any
> other number of assert(true) statements between 0 and 20.

Different alignment of bytecode may cause a different set of
traces to be generated. Apparently one variant is buggy, but
another isn't. That doesn't say much about the problem, though.

I've just fixed a different problem reported by another user. This
also involved bit operations (or rather the conversions caused by
them). Not sure your problem is related. Do a git pull and try
again -- maybe we're lucky.

--Mike