On 24/02/26 12:05PM, Jonathan Cameron wrote:
> On Fri, 23 Feb 2024 11:41:46 -0600
> John Groves <John@xxxxxxxxxx> wrote:
>
> > This function should be called by fs-dax file systems after opening the
> > devdax device. This adds holder_operations.
> >
> > This function serves the same role as fs_dax_get_by_bdev(), which dax
> > file systems call after opening the pmem block device.
> >
> > Signed-off-by: John Groves <john@xxxxxxxxxx>
>
> A few trivial comments form a first read to get my head around this.
>
> Yeah, it is only an RFC, but who doesn't like tidy code? :)
Hope your eyes don't burn too much ;)
>
>
> > ---
> > drivers/dax/super.c | 38 ++++++++++++++++++++++++++++++++++++++
> > include/linux/dax.h | 5 +++++
> > 2 files changed, 43 insertions(+)
> >
> > diff --git a/drivers/dax/super.c b/drivers/dax/super.c
> > index f4b635526345..fc96362de237 100644
> > --- a/drivers/dax/super.c
> > +++ b/drivers/dax/super.c
> > @@ -121,6 +121,44 @@ void fs_put_dax(struct dax_device *dax_dev, void *holder)
> > EXPORT_SYMBOL_GPL(fs_put_dax);
> > #endif /* CONFIG_BLOCK && CONFIG_FS_DAX */
> >
> > +#if IS_ENABLED(CONFIG_DEV_DAX_IOMAP)
> > +
> > +/**
> > + * fs_dax_get()
>
> Smells like kernel doc but fairly sure it needs a short description.
> Have you sanity checked for warnings when running scripts/kerneldoc on it?
Right, and there were other cases. Randy pointed one out, and I've already
gone through and "fixed" them.
>
> > + *
> > + * fs-dax file systems call this function to prepare to use a devdax device for fsdax.
> Trivial but lines too long. Keep under 80 chars unless there is a strong
> readability arguement for not doing so.
I was under the impression the "kids these days" have a 100 column standard.
But I will go through and limit line to 80 except where it gets too awkward.
>
>
> > + * This is like fs_dax_get_by_bdev(), but the caller already has struct dev_dax (and there
> > + * is no bdev). The holder makes this exclusive.
>
> Not familiar with this area: what does exclusive mean here?
The holder_ops are set via cmpxchg, in such a way that if there are already
holder_ops, the call to fs_dax_get() will fail. (as it should)
>
> > + *
> > + * @dax_dev: dev to be prepared for fs-dax usage
> > + * @holder: filesystem or mapped device inside the dax_device
> > + * @hops: operations for the inner holder
> > + *
> > + * Returns: 0 on success, -1 on failure
>
> Why not return < 0 and use somewhat useful return values?
Good idea, will do.
>
> > + */
> > +int fs_dax_get(
> > + struct dax_device *dax_dev,
> > + void *holder,
> > + const struct dax_holder_operations *hops)
>
> Match local style for indents - it's a bit inconsistent but probably...
>
> int fs_dax_get(struct dad_device *dev_dax, void *holder,
> const struct dax_holder_operations *hops)
Done
>
> > +{
> > + /* dax_dev->ops should have been populated by devm_create_dev_dax() */
> > + if (WARN_ON(!dax_dev->ops))
> > + return -1;
> > +
> > + if (!dax_dev || !dax_alive(dax_dev) || !igrab(&dax_dev->inode))
>
> You dereferenced dax_dev on the line above so check is too late or
> unnecessary
Good catch, thank you!
>
> > + return -1;
> > +
> > + if (cmpxchg(&dax_dev->holder_data, NULL, holder)) {
> > + pr_warn("%s: holder_data already set\n", __func__);
>
> Perhaps nicer to use a pr_fmt() deal with the func name if you need it.
> or make it pr_debug and let dynamic debug control formatting if anyone
> wants the function name.
Sounds good.
>
> > + return -1;
> > + }
> > + dax_dev->holder_ops = hops;
> > +
> > + return 0;
> > +}
> > +EXPORT_SYMBOL_GPL(fs_dax_get);
> > +#endif /* DEV_DAX_IOMAP */
> > +
> > enum dax_device_flags {
> > /* !alive + rcu grace period == no new operations / mappings */
> > DAXDEV_ALIVE,
> > diff --git a/include/linux/dax.h b/include/linux/dax.h
> > index b463502b16e1..e973289bfde3 100644
> > --- a/include/linux/dax.h
> > +++ b/include/linux/dax.h
> > @@ -57,7 +57,12 @@ struct dax_holder_operations {
> >
> > #if IS_ENABLED(CONFIG_DAX)
> > struct dax_device *alloc_dax(void *private, const struct dax_operations *ops);
> > +
> > +#if IS_ENABLED(CONFIG_DEV_DAX_IOMAP)
> > +int fs_dax_get(struct dax_device *dax_dev, void *holder, const struct dax_holder_operations *hops);
> line wrap < 80 chars
Roger that
>
> > +#endif
> > void *dax_holder(struct dax_device *dax_dev);
> > +struct dax_device *inode_dax(struct inode *inode);
>
> Unrelated change?
Kinda, but I'm not sure there is a better home for this one. Patch 18,
which is a famfs patch, calls inode_dax(). It was already exported but not
prototyped in dax.h.
Mixing it in with other dev_dax_iomap content seems better than mixing it
with famfs content. Could make it a separate patch, but I was trying to
some old docs that said keep patch sets <=15 - which I deemed impossible here.
What say others?
>
> > void put_dax(struct dax_device *dax_dev);
> > void kill_dax(struct dax_device *dax_dev);
> > void dax_write_cache(struct dax_device *dax_dev, bool wc);
>
Thanks Jonathan!
