Message-ID: <CAPcyv4gVpK2US=-FhZYccKN-9sVa9WC4k5TD+WNH0bBkjwhE2w@mail.gmail.com>
Date:   Fri, 20 Aug 2021 09:06:22 -0700
From:   Dan Williams <dan.j.williams@...el.com>
To:     Shiyang Ruan <ruansy.fnst@...itsu.com>
Cc:     Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        linux-xfs <linux-xfs@...r.kernel.org>,
        Linux NVDIMM <nvdimm@...ts.linux.dev>,
        Linux MM <linux-mm@...ck.org>,
        linux-fsdevel <linux-fsdevel@...r.kernel.org>,
        device-mapper development <dm-devel@...hat.com>,
        "Darrick J. Wong" <djwong@...nel.org>, david <david@...morbit.com>,
        Christoph Hellwig <hch@....de>,
        Alasdair Kergon <agk@...hat.com>,
        Mike Snitzer <snitzer@...hat.com>
Subject: Re: [PATCH RESEND v6 2/9] dax: Introduce holder for dax_device

On Fri, Jul 30, 2021 at 3:02 AM Shiyang Ruan <ruansy.fnst@...itsu.com> wrote:
>
> To easily track filesystem from a pmem device, we introduce a holder for
> dax_device structure, and also its operation.  This holder is used to
> remember who is using this dax_device:
>  - When it is the backend of a filesystem, the holder will be the
>    superblock of this filesystem.
>  - When this pmem device is one of the targets in a mapped device, the
>    holder will be this mapped device.  In this case, the mapped device
>    has its own dax_device and it will follow the first rule.  So that we
>    can finally track to the filesystem we needed.
>
> The holder and holder_ops will be set when filesystem is being mounted,
> or an target device is being activated.
>
> Signed-off-by: Shiyang Ruan <ruansy.fnst@...itsu.com>
> ---
>  drivers/dax/super.c | 46 +++++++++++++++++++++++++++++++++++++++++++++
>  include/linux/dax.h | 17 +++++++++++++++++
>  2 files changed, 63 insertions(+)
>
> diff --git a/drivers/dax/super.c b/drivers/dax/super.c
> index 5fa6ae9dbc8b..00c32dfa5665 100644
> --- a/drivers/dax/super.c
> +++ b/drivers/dax/super.c
> @@ -214,6 +214,8 @@ enum dax_device_flags {
>   * @cdev: optional character interface for "device dax"
>   * @host: optional name for lookups where the device path is not available
>   * @private: dax driver private data
> + * @holder_rwsem: prevent unregistration while holder_ops is in progress
> + * @holder_data: holder of a dax_device: could be filesystem or mapped device
>   * @flags: state and boolean properties
>   */
>  struct dax_device {
> @@ -222,8 +224,11 @@ struct dax_device {
>         struct cdev cdev;
>         const char *host;
>         void *private;
> +       struct rw_semaphore holder_rwsem;

Given the rarity of notification failures and the infrequency of
registration events I think it would be ok for this to be a global
lock rather than per-device. In fact there is already a global dax
lock, see dax_read_lock(). Let's convert that from srcu to rwsem and
add a dax_write_lock().

> +       void *holder_data;
>         unsigned long flags;
>         const struct dax_operations *ops;
> +       const struct dax_holder_operations *holder_ops;
>  };
>
>  static ssize_t write_cache_show(struct device *dev,
> @@ -373,6 +378,25 @@ int dax_zero_page_range(struct dax_device *dax_dev, pgoff_t pgoff,
>  }
>  EXPORT_SYMBOL_GPL(dax_zero_page_range);
>
> +int dax_holder_notify_failure(struct dax_device *dax_dev, loff_t offset,
> +                             size_t size, void *data)
> +{
> +       int rc;
> +
> +       if (!dax_dev)
> +               return -ENXIO;

There also needs to be a dax_dev->alive check, which is only valid to
be checked under dax_read_lock().

Who would ever pass NULL to this function?

> +
> +       if (!dax_dev->holder_data)
> +               return -EOPNOTSUPP;
> +
> +       down_read(&dax_dev->holder_rwsem);
> +       rc = dax_dev->holder_ops->notify_failure(dax_dev, offset,
> +                                                        size, data);
> +       up_read(&dax_dev->holder_rwsem);



> +       return rc;
> +}
> +EXPORT_SYMBOL_GPL(dax_holder_notify_failure);
> +
>  #ifdef CONFIG_ARCH_HAS_PMEM_API
>  void arch_wb_cache_pmem(void *addr, size_t size);
>  void dax_flush(struct dax_device *dax_dev, void *addr, size_t size)
> @@ -603,6 +627,7 @@ struct dax_device *alloc_dax(void *private, const char *__host,
>         dax_add_host(dax_dev, host);
>         dax_dev->ops = ops;
>         dax_dev->private = private;
> +       init_rwsem(&dax_dev->holder_rwsem);
>         if (flags & DAXDEV_F_SYNC)
>                 set_dax_synchronous(dax_dev);
>
> @@ -624,6 +649,27 @@ void put_dax(struct dax_device *dax_dev)
>  }
>  EXPORT_SYMBOL_GPL(put_dax);
>
> +void dax_set_holder(struct dax_device *dax_dev, void *holder,
> +               const struct dax_holder_operations *ops)
> +{
> +       if (!dax_dev)

Same questions about NULL dax dev and ->alive checking.

> +               return;
> +       down_write(&dax_dev->holder_rwsem);
> +       dax_dev->holder_data = holder;
> +       dax_dev->holder_ops = ops;
> +       up_write(&dax_dev->holder_rwsem);
> +}
> +EXPORT_SYMBOL_GPL(dax_set_holder);
> +
> +void *dax_get_holder(struct dax_device *dax_dev)
> +{
> +       if (!dax_dev)
> +               return NULL;

Where is this API used? This result is not valid unless the caller is
holding the read lock.

> +
> +       return dax_dev->holder_data;
> +}
> +EXPORT_SYMBOL_GPL(dax_get_holder);
> +
>  /**
>   * dax_get_by_host() - temporary lookup mechanism for filesystem-dax
>   * @host: alternate name for the device registered by a dax driver
> diff --git a/include/linux/dax.h b/include/linux/dax.h
> index b52f084aa643..6f4b5c97ceb0 100644
> --- a/include/linux/dax.h
> +++ b/include/linux/dax.h
> @@ -38,10 +38,17 @@ struct dax_operations {
>         int (*zero_page_range)(struct dax_device *, pgoff_t, size_t);
>  };
>
> +struct dax_holder_operations {
> +       int (*notify_failure)(struct dax_device *, loff_t, size_t, void *);
> +};
> +
>  extern struct attribute_group dax_attribute_group;
>
>  #if IS_ENABLED(CONFIG_DAX)
>  struct dax_device *dax_get_by_host(const char *host);
> +void dax_set_holder(struct dax_device *dax_dev, void *holder,
> +               const struct dax_holder_operations *ops);
> +void *dax_get_holder(struct dax_device *dax_dev);
>  struct dax_device *alloc_dax(void *private, const char *host,
>                 const struct dax_operations *ops, unsigned long flags);
>  void put_dax(struct dax_device *dax_dev);
> @@ -77,6 +84,14 @@ static inline struct dax_device *dax_get_by_host(const char *host)
>  {
>         return NULL;
>  }
> +static inline void dax_set_holder(struct dax_device *dax_dev, void *holder,
> +               const struct dax_holder_operations *ops)
> +{
> +}
> +static inline void *dax_get_holder(struct dax_device *dax_dev)
> +{
> +       return NULL;
> +}
>  static inline struct dax_device *alloc_dax(void *private, const char *host,
>                 const struct dax_operations *ops, unsigned long flags)
>  {
> @@ -226,6 +241,8 @@ size_t dax_copy_to_iter(struct dax_device *dax_dev, pgoff_t pgoff, void *addr,
>                 size_t bytes, struct iov_iter *i);
>  int dax_zero_page_range(struct dax_device *dax_dev, pgoff_t pgoff,
>                         size_t nr_pages);
> +int dax_holder_notify_failure(struct dax_device *dax_dev, loff_t offset,
> +               size_t size, void *data);
>  void dax_flush(struct dax_device *dax_dev, void *addr, size_t size);
>
>  ssize_t dax_iomap_rw(struct kiocb *iocb, struct iov_iter *iter,
> --
> 2.32.0
>
>
>

Powered by blists - more mailing lists