[RFC v14][PATCH 31/54] powerpc: checkpoint/restart implementation

Serge E. Hallyn serue at us.ibm.com
Wed Apr 29 08:49:35 PDT 2009


Quoting Nathan Lynch (ntl at pobox.com):
> Which brings me to the subject of tree management... it's rather
> difficult for interested parties to follow development of a tree that is
> frequently rewritten.  It would be much easier to base work on a linear
> "append-only" branch.  The guesswork involved in tracking down
> regressions in C/R function would be reduced because bisection would
> work.  And we would have an accurate history of the changes made over
> time.  The cost would be that the checkpoint/restart work would not have
> an easily-reviewable native form, but I think it would be possible to
> generate comprehensible diffs for review since the majority of the code
> is in self-contained files.

It would make this set much easier for us to review (and bisect).  Late
last night I decided re-reviewing the patches just isn't going to work
for me.  I guess I'll just review the checkpoint/*.c files individually,
and generate one big c/r diff for the rest of the kernel tree.

-serge


More information about the Containers mailing list