Re: suspend2 merge (was Re: [Suspend2-devel] Re: CFS and suspend2: hang in atomic copy)

[Ingo Molnar]

* Christoph Hellwig <hch@xxxxxxxxxxxxx> wrote:

> > The interface isn't even 64/32-bit compatible...
> 
> It's not .  And it's one of the worst interface I've seen lately. Did 
> anyone actually review this crap before it went in?  I completely 
> agree with Linus that these kind of boundaries that lead to horribly 
> complex ioctl interface are totally wrong.

it's a bit hard to see the point of it anyway: the resume binary (much 
of the focus of the ioctls) fundamentally lives as an 'initrd binary' - 
and most of the stuff that wants to execute in an initrd is 
fundamentally tied to the kernel anyway.

Perhaps we should allow "in-kernel userspace" that would be allowed to 
grow ad-hoc interfaces and linking that would only be compatible with 
the kernel they are embedded into: e.g. the klibc stuff in linux/usr/* 
could link to the kernel (via whatever method) and just be in essence 
another type of kernel code - but happening to execute in user-space, 
having access to the normal user-space facilities and being able to link 
to (GPL) user-space libraries. Perhaps this would bridge the "i want to 
tinker in user-space because it's technically easier/cleaner there" and 
"fine but that needs formalized ABIs for your connection to 
kernel-space" gap.

> Now suspend2 wasn't exactly nice either when I last reviewed it, but 
> we should probably give it another attempt if we can sort out a proper 
> incremental merge.

yeah, it still has quite a bit of work left, but it looked fundamentally 
split-uppable.

	Ingo
-
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/