Dave Jiang <[email protected]> writes:

> With address_space_operations missing for device dax, namely the
> .set_page_dirty, we hit a kernel warning when running destructive
> ndctl unit test: make TESTS=device-dax check
>
> WARNING: CPU: 3 PID: 7380 at fs/buffer.c:581 __set_page_dirty+0xb1/0xc0
>
> Setting address_space_operations to noop_set_page_dirty and
> noop_invalidatepage for device dax to prevent fallback to
> __set_page_dirty_buffers() and block_invalidatepage() respectively.
>
> Fixes: 2232c6382a ("device-dax: Enable page_mapping()")
>
> Reported-by: Vishal Verma <[email protected]>
> Suggested-by: Dan Williams <[email protected]>
> Signed-off-by: Dave Jiang <[email protected]>

I looked over the other aops, and I think this covers the fallback
paths.

Acked-by: Jeff Moyer <[email protected]>

> ---
>  drivers/dax/device.c |    6 ++++++
>  1 file changed, 6 insertions(+)
>
> diff --git a/drivers/dax/device.c b/drivers/dax/device.c
> index 6fd46083e629..079f6811eca6 100644
> --- a/drivers/dax/device.c
> +++ b/drivers/dax/device.c
> @@ -534,6 +534,11 @@ static unsigned long dax_get_unmapped_area(struct file 
> *filp,
>       return current->mm->get_unmapped_area(filp, addr, len, pgoff, flags);
>  }
>  
> +static const struct address_space_operations dev_dax_aops = {
> +     .set_page_dirty         = noop_set_page_dirty,
> +     .invalidatepage         = noop_invalidatepage,
> +};
> +
>  static int dax_open(struct inode *inode, struct file *filp)
>  {
>       struct dax_device *dax_dev = inode_dax(inode);
> @@ -543,6 +548,7 @@ static int dax_open(struct inode *inode, struct file 
> *filp)
>       dev_dbg(&dev_dax->dev, "trace\n");
>       inode->i_mapping = __dax_inode->i_mapping;
>       inode->i_mapping->host = __dax_inode;
> +     inode->i_mapping->a_ops = &dev_dax_aops;
>       filp->f_mapping = inode->i_mapping;
>       filp->f_wb_err = filemap_sample_wb_err(filp->f_mapping);
>       filp->private_data = dev_dax;
>
> _______________________________________________
> Linux-nvdimm mailing list
> [email protected]
> https://lists.01.org/mailman/listinfo/linux-nvdimm
_______________________________________________
Linux-nvdimm mailing list
[email protected]
https://lists.01.org/mailman/listinfo/linux-nvdimm

Reply via email to