[CRIU] Re: [RFC 0/4] Bring ability to multiple c/r of a program

[Pavel Emelyanov]

On 07/20/2012 09:33 PM, Cyrill Gorcunov wrote:
> Guys, as you know we've a bit of problem c/r'ing program
> being previously checkpointed, such as
> 
>  - resident restorer code get accumulated over c/r sessions

Yes, this is the problem.

>  - problem with zero-size VMA rest from clone() stack

What is it?

> both are addressed in this series. Please review, comments are welcome.
> 
> 

IMO idea with uuids is not nice. I'd do it the other way:

1. We need two restorer.o codes -- one big blob which we have now,
   the other one small which does only two things (see below)
2. Big blob:
   a. prepares everything it does now except for the sigreturn stuff
   b. puts sigreturn frame on target task's stack end
   c. loads the small blob
   d. jumps into it
3. Small blob:
   It looks like zeroed out page with small portion of code at the
   beginning which does:
   a. calls munmap on address somewhere on stack (big blob vma)
   b. calls sigreturn with sigframe in stack (prepared by big blob)

With this after restore we have only one vma left which looks like
tons of zero and several bytes of code. On 2nd restore we can just memcmp
this VMA and do not re-load it again if it matches.

Thanks,
Pavel