linux-kernel - Re: [PATCH v2 2/2] drivers: gpio: add virtio-gpio guest driver

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20201206075131-mutt-send-email-mst@kernel.org>
Date:   Mon, 7 Dec 2020 08:52:20 -0500
From:   "Michael S. Tsirkin" <mst@...hat.com>
To:     "Enrico Weigelt, metux IT consult" <lkml@...ux.net>
Cc:     Jason Wang <jasowang@...hat.com>,
        "Enrico Weigelt, metux IT consult" <info@...ux.net>,
        linux-kernel@...r.kernel.org, corbet@....net,
        linus.walleij@...aro.org, bgolaszewski@...libre.com,
        linux-doc@...r.kernel.org, linux-gpio@...r.kernel.org,
        virtualization@...ts.linux-foundation.org,
        linux-riscv@...ts.infradead.org, stefanha@...hat.com,
        msuchanek@...e.de
Subject: Re: [PATCH v2 2/2] drivers: gpio: add virtio-gpio guest driver

On Sat, Dec 05, 2020 at 09:05:16PM +0100, Enrico Weigelt, metux IT consult wrote:
> On 05.12.20 20:32, Michael S. Tsirkin wrote:
> 
> Hi,
> 
> > It seems a bit of a mess, at this point I'm not entirely sure when
> > should drivers select VIRTIO and when depend on it.
> 
> if VIRTIO just enables something that could be seen as library
> functions, then select should be right, IMHO.
> 
> > The text near it says:
> > 
> > # SPDX-License-Identifier: GPL-2.0-only
> > config VIRTIO
> >         tristate
> 
> oh, wait, doesn't have an menu text, so we can't even explicitly enable
> it (not shown in menu) - only implicitly. Which means that some other
> option must select it, in order to become availe at all, and in order
> to make others depending on it becoming available.
> 
> IMHO, therefore select is the correct approach.
> 
> 
> >         help
> >           This option is selected by any driver which implements the virtio
> >           bus, such as CONFIG_VIRTIO_PCI, CONFIG_VIRTIO_MMIO, CONFIG_RPMSG
> >           or CONFIG_S390_GUEST.
> > 
> > Which seems clear enough and would indicate drivers for devices *behind*
> > the bus should not select VIRTIO and thus presumably should "depend on" it.
> > This is violated in virtio console and virtio fs drivers.
> 
> See above: NAK. because it can't even be enabled directly (by the user).
> If it wasn't meant otherwise, we'd have to add an menu text.


The point is that user enables one of the bindings.
That in turn enables drivers. If we merely select VIRTIO
there's a chance user won't remember to select any bindings
and will be surprised not to see any devices.



> > For console it says:
> > 
> > commit 9f30eb29c514589e16f2999ea070598583d1f6ec
> > Author: Michal Suchanek <msuchanek@...e.de>
> > Date:   Mon Aug 31 18:58:50 2020 +0200
> > 
> >     char: virtio: Select VIRTIO from VIRTIO_CONSOLE.
> >     
> >     Make it possible to have virtio console built-in when
> >     other virtio drivers are modular.
> >     
> >     Signed-off-by: Michal Suchanek <msuchanek@...e.de>
> >     Reviewed-by: Amit Shah <amit@...nel.org>
> >     Link: https://lore.kernel.org/r/20200831165850.26163-1-msuchanek@suse.de
> >     Signed-off-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
> > 
> > which seems kind of bogus - why do we care about allowing a builtin
> > virtio console driver if the pci virtio bus driver is a module?
> > There won't be any devices on the bus to attach to ...
> 
> When using other transports ?

Any transport selects VIRTIO so if you enable that, you get
VIRTIO and thus it's enough to depend on it.

> In my current project, eg. I'm using mmio - my kernel has pci completely
> disabled.
> 
> > I am inclined to fix console and virtio fs to depend on VIRTIO:
> > select is harder to use correctly ...
> 
> I don't thinkt that would be good - instead everybody should just select
> VIRTIO, never depend on it (maybe depend on VIRTIO_MENU instead)

GPU depends on VIRTIO and on VIRTIO_MENU ... which seems even messier
...

> 
> --mtx
> 
> -- 
> ---
> Hinweis: unverschlüsselte E-Mails können leicht abgehört und manipuliert
> werden ! Für eine vertrauliche Kommunikation senden Sie bitte ihren
> GPG/PGP-Schlüssel zu.
> ---
> Enrico Weigelt, metux IT consult
> Free software and Linux embedded engineering
> info@...ux.net -- +49-151-27565287