Re: [PATCH 11/22] xfs: refactor special inode roll out of xfs_dir_ialloc

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, Jan 17, 2019 at 06:24:27AM -0800, Christoph Hellwig wrote:
> > diff --git a/fs/xfs/libxfs/xfs_inode_util.h b/fs/xfs/libxfs/xfs_inode_util.h
> > index 5a1d98d1546d..ee274d74b8d4 100644
> > --- a/fs/xfs/libxfs/xfs_inode_util.h
> > +++ b/fs/xfs/libxfs/xfs_inode_util.h
> > @@ -82,6 +82,12 @@ struct xfs_ialloc_ops {
> >  
> >  	/* Do any final setup needed before we return the inode. */
> >  	void (*setup)(struct xfs_inode *ip);
> > +
> > +	/*
> > +	 * Roll the transaction between allocating a new ichunk and
> > +	 * initializing a new inode core.
> > +	 */
> > +	int (*ichunk_roll)(struct xfs_trans **tpp);
> 
> Sorry, but this whole idea to add gracious indirect calls is just
> backwards.  They do have a non-trivial cost, and we should rather
> get rid of pointless indirect calls instead of adding more.

I built all this indirect call stuff so that mkfs protofile code would
be able to set parameters ... but seeing as Eric said he'll deprecate
all that, I think it's no longer necessary.

--D



[Index of Archives]     [XFS Filesystem Development (older mail)]     [Linux Filesystem Development]     [Linux Audio Users]     [Yosemite Trails]     [Linux Kernel]     [Linux RAID]     [Linux SCSI]


  Powered by Linux