Re: [RFC v2][PATCH 2/9] General infrastructure for checkpoint restart

[Oren Laadan]

Louis Rilling wrote:
> On Wed, Aug 20, 2008 at 11:04:13PM -0400, Oren Laadan wrote:
> > Add those interfaces, as well as helpers needed to easily manage the
> > file format. The code is roughly broken out as follows:
> >
> > ckpt/sys.c - user/kernel data transfer, as well as setup of the
> > checkpoint/restart context (a per-checkpoint data structure for
> > housekeeping)
> >
> > ckpt/checkpoint.c - output wrappers and basic checkpoint handling
> >
> > ckpt/restart.c - input wrappers and basic restart handling
> >
> > Patches to add the per-architecture support as well as the actual
> > work to do the memory checkpoint follow in subsequent patches.
>
> [...]
>
> > diff --git a/checkpoint/sys.c b/checkpoint/sys.c
> > new file mode 100644
> > index 0000000..2891c48
> > --- /dev/null
> > +++ b/checkpoint/sys.c
>
> [...]
>
> > +/*
> > + * helpers to manage CR contexts: allocated for each checkpoint and/or
> > + * restart operation, and persists until the operation is completed.
> > + */
> > +
> > +static atomic_t cr_ctx_count;	/* unique checkpoint identifier */
>
> I thought we agreed that this counter should be per-container. Perhaps add a
> TODO here?

True.

> > +
> > +void cr_ctx_free(struct cr_ctx *ctx)
> > +{
> > +
> > +	if (ctx->file)
> > +		fput(ctx->file);
> > +	if (ctx->vfsroot)
> > +		path_put(ctx->vfsroot);
> > +
> > +	free_pages((unsigned long) ctx->tbuf, CR_TBUF_ORDER);
> > +	free_pages((unsigned long) ctx->hbuf, CR_HBUF_ORDER);
> > +
> > +	kfree(ctx);
> > +}
> > +
> > +struct cr_ctx *cr_ctx_alloc(pid_t pid, struct file *file, unsigned long flags)
> > +{
> > +	struct cr_ctx *ctx;
> > +
> > +	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
> > +	if (!ctx)
> > +		return NULL;
> > +
> > +	ctx->tbuf = (void *) __get_free_pages(GFP_KERNEL, CR_TBUF_ORDER);
> > +	ctx->hbuf = (void *) __get_free_pages(GFP_KERNEL, CR_HBUF_ORDER);
> > +	if (!ctx->tbuf || !ctx->hbuf)
> > +		goto nomem;
> > +
> > +	ctx->pid = pid;
> > +	ctx->flags = flags;
> > +
> > +	ctx->file = file;
> > +	get_file(file);
> > +
> > +	/* assume checkpointer is in container's root vfs */
>
> I'm a bit puzzled by this assumption. I would say: either this is a
> self-checkpoint (only current process), or this is a container checkpoint. In
> the latter case, I expect that in the general case the checkpointer lives
> outside the container (and the interface of sys_checkpoint() below confirms
> this), so it's root fs is probably not the container's one.
>
> Does it differ from what you're planning?

You are correct. We lack infrastructure for what I'd call "container-object",
and the patchset does not yet tackle a container and multiple tasks, so this
is an interim solution. Will add a FIXME comment.

Thanks,

Oren.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/