[Devel] Re: [RFC][cr][PATCH 2/6] Checkpoint file-locks

[Serge E. Hallyn]

Quoting Matt Helsley (matthltc at us.ibm.com):
> On Tue, May 04, 2010 at 06:44:39PM -0700, sukadev at linux.vnet.ibm.com wrote:
> > From: Sukadev Bhattiprolu <sukadev at linux.vnet.ibm.com>
> > +int
> > +checkpoint_file_locks(struct ckpt_ctx *ctx, struct files_struct *files,
> > +		struct file *file, int fd)
> > +{
> > +	int rc;
> > +	struct inode *inode;
> > +	struct file_lock **lockpp;
> > +	struct file_lock *lockp;
> > +	struct file_lock last_lock;
> > +
> > +	lock_kernel();
> 
> Eep. What are the current standards as far as adding "new" uses of the BKL?
> Arnd/anti-BKL-ninjas might be good folks to Cc on the next round if this
> is still here.

I'd say
	1. look at Arnd's tree just to be ready to switch to it's
	   locking method
	   (see http://git.kernel.org/gitweb.cgi?p=linux/kernel/git/arnd/playground.git;a=commit;h=8dd5597e27d8c055376719434de6fa630da1b9f7)
	   Heck, maybe even use 'lock_flocks()' and just #define it to
	   lock_kernel for now.
	2. put this code straight into fs/locks.c, so that the related
	   uses of lock_kernel() are localized.
_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers