Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20200827142724.15032-1-tiwai@suse.de>
In-Reply-To: <20200827142724.15032-1-tiwai@suse.de>
From:   David Wysochanski <dwysocha@redhat.com>
Date:   Wed, 14 Oct 2020 09:48:32 -0400
Message-ID: <CALF+zOn5Pzuf1zv1_bR_MbX8oJ5UZ-FpCd8KkkMzWOE9n3p7EQ@mail.gmail.com>
Subject: Re: [Linux-cachefs] [PATCH] fs/cachefiles: Drop superfluous readpages
 aops NULL check
To:     Takashi Iwai <tiwai@suse.de>
Cc:     David Howells <dhowells@redhat.com>, linux-cachefs@redhat.com,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, Aug 27, 2020 at 11:10 AM Takashi Iwai <tiwai@suse.de> wrote:
>
> After the recent actions to convert readpages aops to readahead, the
> NULL checks of readpages aops in cachefiles_read_or_alloc_page() may
> hit falsely.  More badly, it's an ASSERT() call, and this panics.
>
> Drop the superfluous NULL checks for fixing this regression.
>
> BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=208883
> BugLink: https://bugzilla.opensuse.org/show_bug.cgi?id=1175245
> Cc: <stable@vger.kernel.org> # v5.8
> Signed-off-by: Takashi Iwai <tiwai@suse.de>
> ---
>  fs/cachefiles/rdwr.c | 2 --
>  1 file changed, 2 deletions(-)
>
> diff --git a/fs/cachefiles/rdwr.c b/fs/cachefiles/rdwr.c
> index 3080cda9e824..5b4cee71fa32 100644
> --- a/fs/cachefiles/rdwr.c
> +++ b/fs/cachefiles/rdwr.c
> @@ -412,7 +412,6 @@ int cachefiles_read_or_alloc_page(struct fscache_retrieval *op,
>
>         inode = d_backing_inode(object->backer);
>         ASSERT(S_ISREG(inode->i_mode));
> -       ASSERT(inode->i_mapping->a_ops->readpages);
>
>         /* calculate the shift required to use bmap */
>         shift = PAGE_SHIFT - inode->i_sb->s_blocksize_bits;
> @@ -712,7 +711,6 @@ int cachefiles_read_or_alloc_pages(struct fscache_retrieval *op,
>
>         inode = d_backing_inode(object->backer);
>         ASSERT(S_ISREG(inode->i_mode));
> -       ASSERT(inode->i_mapping->a_ops->readpages);
>
>         /* calculate the shift required to use bmap */
>         shift = PAGE_SHIFT - inode->i_sb->s_blocksize_bits;
> --
> 2.16.4
>
> --
> Linux-cachefs mailing list
> Linux-cachefs@redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cachefs
>

Agree this should go in as fscache (I am testing with NFS) becomes
unusable without it after 5.8.

It may have flown under the radar due to the in-progress FS-Cache rewrite.