linux-kernel - Re: [PATCH v3 09/15] platform: goldfish: pipe: Move goldfish_pipe to goldfish_pipe

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] [day] [month] [year] [list]

Message-ID: <CAOGAQeo-4GF9uMe5WSosX9njOporqhkQyhPvzpG7mmNdq5Gh1w@mail.gmail.com>
Date:   Tue, 16 Oct 2018 14:02:01 -0700
From:   Roman Kiryanov <rkir@...gle.com>
To:     gregkh@...uxfoundation.org
Cc:     linux-kernel@...r.kernel.org, Todd Kjos <tkjos@...gle.com>
Subject: Re: [PATCH v3 09/15] platform: goldfish: pipe: Move goldfish_pipe to goldfish_pipe_v2

> > I don't want to mix v1 and v2.
>
> Why?  What is the difference?
> Why do you even need/want 2 different drivers here

In old emulators we implemented the device differently, so we need a
different driver.

>  for something that really should be done with the virtio
> interface instead?

We are already looking into virtio. But we did not find a userspace
API that virtio provides to
replace our pipe. The closest we found is virtqueue, but it is a
kernel thing. So we will
need goldfish_pipe_v3.

> > I don't see how userspace could be affected, unless we refer to
> > __FILE__ inside the driver for some important things. I believe we
> > don't.
>
> How does your userspace know what module to automatically load?  That's
> what usually breaks if you do not have proper platform binding happening
> automatically.

I am not sure if I understand your question.

Userspace opens the device by its name. The device name is hardcoded:

#define DEVICE_NAME "goldfish_pipe"

https://android.googlesource.com/kernel/goldfish/+/android-goldfish-4.14-dev/drivers/platform/goldfish/goldfish_pipe.h#19

I still don't understand how kernel filenames are involved (and
whether should they) into userpspace.