Re: linux-next: manual merge of the block tree with the mmc-fixes tree

[Ulf Hansson <ulf.hansson@linaro.org>]

On Fri, 9 Jul 2021 at 02:32, Stephen Rothwell <sfr@canb.auug.org.au> wrote:
>
> Hi all,
>
> Today's linux-next merge of the block tree got a conflict in:
>
>   drivers/mmc/core/block.c
>
> between commit:
>
>   5c0777665b3e ("mmc: core: Use kref in place of struct mmc_blk_data::usage")
>
> from the mmc-fixes tree and commits:
>
>   249cda3325e0 ("mmc: remove an extra blk_{get,put}_queue pair")
>   607d968a5769 ("mmc: switch to blk_mq_alloc_disk")
>
> from the block tree.
>
> I fixed it up (see below) and can carry the fix as necessary. This
> is now fixed as far as linux-next is concerned, but any non trivial
> conflicts should be mentioned to your upstream maintainer when your tree
> is submitted for merging.  You may also want to consider cooperating
> with the maintainer of the conflicting tree to minimise any particularly
> complex conflicts.
>
> --
> Cheers,
> Stephen Rothwell

Stephen, thanks for the heads up!

Jens, I noticed that you sent the PR with the mmc commits as of
yesterday. Assuming Linus will pull it before rc1, I will rebase and
fix up the conflict from my fixes branch on top. No action needed from
your side.

Kind regards
Uffe

>
> diff --cc drivers/mmc/core/block.c
> index d7b5c5ab75fa,9890a1532cb0..000000000000
> --- a/drivers/mmc/core/block.c
> +++ b/drivers/mmc/core/block.c
> @@@ -195,26 -196,18 +195,25 @@@ static inline int mmc_get_devidx(struc
>         return devidx;
>   }
>
>  -static void mmc_blk_put(struct mmc_blk_data *md)
>  +static void mmc_blk_kref_release(struct kref *ref)
>   {
>  -      mutex_lock(&open_lock);
>  -      md->usage--;
>  -      if (md->usage == 0) {
>  -              int devidx = mmc_get_devidx(md->disk);
>  +      struct mmc_blk_data *md = container_of(ref, struct mmc_blk_data, kref);
>  +      int devidx;
>
>  -              ida_simple_remove(&mmc_blk_ida, devidx);
>  -              put_disk(md->disk);
>  -              kfree(md);
>  -      }
>  +      devidx = mmc_get_devidx(md->disk);
> -       blk_put_queue(md->queue.queue);
>  +      ida_simple_remove(&mmc_blk_ida, devidx);
>  +
>  +      mutex_lock(&open_lock);
>  +      md->disk->private_data = NULL;
>         mutex_unlock(&open_lock);
>  +
>  +      put_disk(md->disk);
>  +      kfree(md);
>  +}
>  +
>  +static void mmc_blk_put(struct mmc_blk_data *md)
>  +{
>  +      kref_put(&md->kref, mmc_blk_kref_release);
>   }
>
>   static ssize_t power_ro_lock_show(struct device *dev,
> @@@ -2334,27 -2327,11 +2333,11 @@@ static struct mmc_blk_data *mmc_blk_all
>
>         INIT_LIST_HEAD(&md->part);
>         INIT_LIST_HEAD(&md->rpmbs);
>  -      md->usage = 1;
>  +      kref_init(&md->kref);
> -
> -       ret = mmc_init_queue(&md->queue, card);
> -       if (ret)
> -               goto err_putdisk;
> -
>         md->queue.blkdata = md;
>
> -       /*
> -        * Keep an extra reference to the queue so that we can shutdown the
> -        * queue (i.e. call blk_cleanup_queue()) while there are still
> -        * references to the 'md'. The corresponding blk_put_queue() is in
> -        * mmc_blk_put().
> -        */
> -       if (!blk_get_queue(md->queue.queue)) {
> -               mmc_cleanup_queue(&md->queue);
> -               ret = -ENODEV;
> -               goto err_putdisk;
> -       }
> -
>         md->disk->major = MMC_BLOCK_MAJOR;
> +       md->disk->minors = perdev_minors;
>         md->disk->first_minor = devidx * perdev_minors;
>         md->disk->fops = &mmc_bdops;
>         md->disk->private_data = md;