On Fri, Apr 10, 2009 at 08:06:55AM -0700, Linus Torvalds wrote: > On Fri, 10 Apr 2009, Alexey Dobriyan wrote: > > > > This is to show how we see C/R and to provoke discussion on number of > > important issues (mounts, ...). > > My only initial reaction is that I absolutely hate the naming (not to say > I love the code - just to say that I didn't even look at it, because I got > hung up on the name). > > "cr"? It could be anything. I realize that to _you_ that is meaningful, > but to somebody less specifically interested in checkpoint-restore 'cr' > means 'carriage return' or just doesn't really say anything at all. Well, in OpenVZ everything is in kernel/cpt/ and prefixed with "cpt_" and "rst_". And I think "cr_" is super nice prefix: it's short, it's C-like, it reminds about restart part. Eventually, C/R will become standard in-kernel thing everyone should be at least aware of, so it's like learning what "vma" means. > That goes both for file naming (kernel/cr/xyzzy.c) and to a lesser degree > for function naming too. I also don't think it makes sense to have > something like kernel/cr/cr-x86_32.c or kernel/cr/cr-tty.c - maybe that is > good right now, but I sure hope that the long-term goal is to have these > things in the code that will need to change them when the code gets > updated (ie arch/x86/kernel and drivers/char/) In the long run, yes, C/R should be moved closer to core code it tries to checkpoint. Right now, however, doing "make kernel/cr/" is much quicker and C/R can not do much, so it's unclear how exactly splitting should be done. _______________________________________________ Containers mailing list Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/containers
