Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
Message-ID: <54ade5e4-d277-1831-06f5-da482c76601f@redhat.com>
Date:   Thu, 5 May 2022 14:57:17 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.8.0
Subject: Re: [PATCH 2/3] fbdev/simplefb: Cleanup fb_info in .fb_destroy rather
 than .remove
Content-Language: en-US
To:     linux-kernel@vger.kernel.org, dri-devel@lists.freedesktop.org,
        linux-fbdev@vger.kernel.org, Helge Deller <deller@gmx.de>,
        Hans de Goede <hdegoede@redhat.com>
References: <20220504215151.55082-1-javierm@redhat.com>
 <20220504215722.56970-1-javierm@redhat.com>
 <YnPInuayAYQa1jmz@phenom.ffwll.local>
From:   Javier Martinez Canillas <javierm@redhat.com>
In-Reply-To: <YnPInuayAYQa1jmz@phenom.ffwll.local>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

Hello Daniel,

On 5/5/22 14:52, Daniel Vetter wrote:
> On Wed, May 04, 2022 at 11:57:22PM +0200, Javier Martinez Canillas wrote:
>> The driver is calling framebuffer_release() in its .remove callback, but
>> this will cause the struct fb_info to be freed too early. Since it could
>> be that a reference is still hold to it if user-space opened the fbdev.
>>
>> This would lead to a use-after-free error if the framebuffer device was
>> unregistered but later a user-space process tries to close the fbdev fd.
>>
>> The correct thing to do is to only unregister the framebuffer in the
>> driver's .remove callback, but do any cleanup in the fb_ops.fb_destroy.
>>
>> Suggested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
>> Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
> 
> I think this should have a Fixes: line for the patch from Thomas which
> changed the remove_conflicting_fb code:
> 
> 27599aacbaef ("fbdev: Hot-unplug firmware fb devices on forced removal")
>

Ok.
 
> I think we should also mention that strictly speaking the code flow is now
> wrong, because hw cleanup (like iounmap) should be done from ->remove
> while sw cleanup (like calling framebuffer_release()) is the only thing
> that should be done from ->fb_destroy. But the current code matches what
> was happening before 27599aacbaef so more minimal "fix"
>

Yes, I tried to make it as small as possible. Thomas pointed out that vesafb
has the same issue and I included in v2. I had move things around more there
though.
 
> With those details added Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
>
> Same for the next patch.

Thanks. I'll post a v3 adding what you suggested but probably not today.

-- 
Best regards,

Javier Martinez Canillas
Linux Engineering
Red Hat