Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Sat, 6 Aug 2022 13:46:16 -0700
From:   Jack Rosenthal <jrosenth@chromium.org>
To:     Stephen Boyd <swboyd@chromium.org>
Cc:     chrome-platform@lists.linux.dev, linux-kernel@vger.kernel.org,
        Tzung-Bi Shih <tzungbi@kernel.org>,
        Guenter Roeck <groeck@chromium.org>,
        Julius Werner <jwerner@chromium.org>
Subject: Re: [PATCH v7] firmware: google: Implement cbmem in sysfs driver
Message-ID: <Yu7TGDYTyzFLoVe4@chromium.org>
References: <YuuBrUmiRYoaSmPw@google.com>
 <20220804142856.306032-1-jrosenth@chromium.org>
 <CAE-0n51UcauP1ojoskV9_2+CFPoJrs08jGHVPCZdSA_oXX1zUw@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CAE-0n51UcauP1ojoskV9_2+CFPoJrs08jGHVPCZdSA_oXX1zUw@mail.gmail.com>
Precedence: bulk

On 2022-08-04 at 15:59 -0500, Stephen Boyd wrote:
> I suspect this won't work because the driver can be unbound, and thus
> the mapping can be destroyed by devm, but then the kobject and sysfs
> file like 'mem' can be held by userspace after the driver is unbound.
> Userspace access after that time will try to read/write unmapped memory.

When the driver is unbound, read()/write() will return -ENODEV, and
mmap() isn't supported.

> Why not make this in the bus driver (coreboot-table.c)? In fact, most
> things could probably be created there instead of in a 'driver' that
> really isn't doing much 'driving' at all.

In v8, I moved the coreboot_kobj to coreboot-table.c.  I imagine in the
future we could move some of the other things (e.g., "vpd") under this
object for consistency.  So now the structure is:

   /sys/firmware/coreboot/cbmem/<id>/...