Re: [PATCH 1/8] xfs: ensure sync write errors are returned

[Christoph Hellwig <hch@xxxxxxxxxxxxx>]

> +             if (ret == -ENOSPC && !enospc) {
> +                     ssize_t ret2 = xfs_flush_pages(ip, 0, -1, 0, FI_NONE);
> +                     if (ret2) {
> +                             ret = ret2;
>                               goto out_unlock_internal;
> +                     }
>                       enospc = 1;
>                       goto write_retry;
>               }

I don't think there is any reason for the ret2 variale here.  If it's
non-zero we assign it to ret, and if it's zero we overwrite it again
after jumping to the write_retry label.