[PATCH] fs: fat: allow FS and device to have mismatching sector sizes

[Simon Glass]

Hi Caleb,

On Tue, 12 Nov 2024 at 21:50, Caleb Connolly <caleb.connolly at linaro.org> wrote:
>
> The Qualcomm Robotics RB1 and RB2 boards feature MMC storage provisioned
> with a 4k sector size GPT, with factory partitions formatted with
> filesystems using 512 byte sector sizes.
>
> Whilst certainly unusual, and really not ideal, this appears to work
> fine in practise. So demote this failure condition to a debug log.
>

Can you please take a look at [1] - this patch was not applied, but I
hit a similar problem.

> Signed-off-by: Caleb Connolly <caleb.connolly at linaro.org>
> ---
>  fs/fat/fat.c | 7 +++----
>  1 file changed, 3 insertions(+), 4 deletions(-)
>
> diff --git a/fs/fat/fat.c b/fs/fat/fat.c
> index e2570e816768..a84ca301cad1 100644
> --- a/fs/fat/fat.c
> +++ b/fs/fat/fat.c
> @@ -645,13 +645,12 @@ static int get_fs_info(fsdata *mydata)
>         mydata->rootdir_sect = mydata->fat_sect + mydata->fatlength * bs.fats;
>
>         mydata->sect_size = get_unaligned_le16(bs.sector_size);
>         mydata->clust_size = bs.cluster_size;
> -       if (mydata->sect_size != cur_part_info.blksz) {
> -               log_err("FAT sector size mismatch (fs=%u, dev=%lu)\n",
> +       if (mydata->sect_size != cur_part_info.blksz)
> +               log_debug("FAT sector size mismatch (fs=%u, dev=%lu)\n",
>                         mydata->sect_size, cur_part_info.blksz);
> -               return -1;
> -       }
> +
>         if (mydata->clust_size == 0) {
>                 log_err("FAT cluster size not set\n");
>                 return -1;
>         }
> --
> 2.47.0
>

[1] https://patchwork.ozlabs.org/project/uboot/patch/20230730111516.v2.1.Ia13846500fab3d5a1d5573db11a040d233994fa6@changeid/