David,

  Gentle ping on this patch?

Thanks, Anand


On 17/11/2021 11:29, Anand Jain wrote:

On 17/11/2021 01:16, David Sterba wrote:

quoted

On Thu, Oct 21, 2021 at 11:31:17PM +0800, Anand Jain wrote:

quoted

In the case of the seed device, the fsid can be different from the 
mounted
sprout fsid.  The userland has to read the device superblock to know the
fsid but, that idea fails if the device is missing. So add a sysfs
interface devinfo/<devid>/fsid to show the fsid of the device.

For example:
  $ cd /sys/fs/btrfs/b10b02a5-f9de-4276-b9e8-2bfd09a578a8

  $ cat devinfo/1/fsid
  c44d771f-639d-4df3-99ec-5bc7ad2af93b
  $ cat  devinfo/3/fsid
  b10b02a5-f9de-4276-b9e8-2bfd09a578a8

 From user perspective, it's another fsid, one is in the path, so I'm
wondering if it should be named like read_fsid or sprout_fsid or if the
seed/sprout information should be put into another directory completely.

I am viewing it as fsid as per the device's sb. This fsid matches with
the blkid(8) output. A path to the device's fsid will help to script.
So I am not voting for sprout_fsid because it does not exist in the
most common non-sprouted fsid.

Now to show whether a device is seed or sprout, the user friendly
approach will be like for example:
      /sys/fs/btrfs/<fsid>/devinfo/<devid>/device_state
to show all that apply, for example:
      SEED|SPROUT|READ_ONLY|WRITABLE|ZONED|METADATA_ONLY|\
      DATA_ONLY|READ_PREFERRED|REPLACE_TGT|REPLACE_SRC|\
      SCRUB_RUNNING

However, per kernel general rule of thumb one value per sys-fs file,
so it should be for example:
      /sys/fs/btrfs/<fsid>/devinfo/<devid>/seed
0
      /sys/fs/btrfs/<fsid>/devinfo/<devid>/sprout
1

So based on this my patch "btrfs: introduce new device-state
read_preferred" in the mailing list already proposed
      /sys/fs/btrfs/<fsid>/devinfo/<devid>/read_preferred

So to summarize,  I can rename fsid to read_fsid. And, progressively
in a separate patch, we can add <>/seed <>/sprout sysfs files as above.
What do you think?