From: Vincent MAILHOL <mailhol.vincent@wanadoo.fr>
To: Alan Stern <stern@rowland.harvard.edu>
Cc: Andrew Lunn <andrew@lunn.ch>,
linux-can@vger.kernel.org, Marc Kleine-Budde <mkl@pengutronix.de>,
linux-kernel@vger.kernel.org,
Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
netdev@vger.kernel.org, linux-usb@vger.kernel.org,
Saeed Mahameed <saeed@kernel.org>, Jiri Pirko <jiri@nvidia.com>,
Lukas Magel <lukas.magel@posteo.net>
Subject: Re: [PATCH v4 2/6] can: etas_es58x: add devlink support
Date: Tue, 29 Nov 2022 08:30:34 +0900 [thread overview]
Message-ID: <CAMZ6RqKuGC8eyMb45x_oVBTQ0M707JhL-oUs34+8c267cKovAw@mail.gmail.com> (raw)
In-Reply-To: <Y4TYzgOczlegG7OK@rowland.harvard.edu>
On Tue. 29 Nov. 2022 at 00:50, Alan Stern <stern@rowland.harvard.edu> wrote:
> On Mon, Nov 28, 2022 at 02:32:23PM +0900, Vincent MAILHOL wrote:
> > On Mon. 28 Nov. 2022 at 10:34, Vincent MAILHOL
> > <mailhol.vincent@wanadoo.fr> wrote:
> > > On Mon. 28 Nov. 2022 at 00:41, Alan Stern <stern@rowland.harvard.edu> wrote:
> > > > On Sun, Nov 27, 2022 at 02:10:32PM +0900, Vincent MAILHOL wrote:
> > > > > > Should devlink_free() be after usb_set_inftdata()?
> > > > >
> > > > > A look at
> > > > > $ git grep -W "usb_set_intfdata(.*NULL)"
> > > > >
> > > > > shows that the two patterns (freeing before or after
> > > > > usb_set_intfdata()) coexist.
> > > > >
> > > > > You are raising an important question here. usb_set_intfdata() does
> > > > > not have documentation that freeing before it is risky. And the
> > > > > documentation of usb_driver::disconnect says that:
> > > > > "@disconnect: Called when the interface is no longer accessible,
> > > > > usually because its device has been (or is being) disconnected
> > > > > or the driver module is being unloaded."
> > > > > Ref: https://elixir.bootlin.com/linux/v6.1-rc6/source/include/linux/usb.h#L1130
> > > > >
> > > > > So the interface no longer being accessible makes me assume that the
> > > > > order does not matter. If it indeed matters, then this is a foot gun
> > > > > and there is some clean-up work waiting for us on many drivers.
> > > > >
> > > > > @Greg, any thoughts on whether or not the order of usb_set_intfdata()
> > > > > and resource freeing matters or not?
> > > >
> > > > In fact, drivers don't have to call usb_set_intfdata(NULL) at all; the
> > > > USB core does it for them after the ->disconnect() callback returns.
> > >
> > > Interesting. This fact is widely unknown, cf:
> > > $ git grep "usb_set_intfdata(.*NULL)" | wc -l
> > > 215
> > >
> > > I will do some clean-up later on, at least for the CAN USB drivers.
> > >
> > > > But if a driver does make the call, it should be careful to ensure that
> > > > the call happens _after_ the driver is finished using the interface-data
> > > > pointer. For example, after all outstanding URBs have completed, if the
> > > > completion handlers will need to call usb_get_intfdata().
> > >
> > > ACK. I understand that it should be called *after* the completion of
> > > any ongoing task.
> > >
> > > My question was more on:
> > >
> > > devlink_free(priv_to_devlink(es58x_dev));
> > > usb_set_intfdata(intf, NULL);
> > >
> > > VS.
> > >
> > > usb_set_intfdata(intf, NULL);
> > > devlink_free(priv_to_devlink(es58x_dev));
> > >
> > > From your comments, I understand that both are fine.
> >
> > Do we agree that the usb-skeleton is doing it wrong?
> > https://elixir.bootlin.com/linux/latest/source/drivers/usb/usb-skeleton.c#L567
> > usb_set_intfdata(interface, NULL) is called before deregistering the
> > interface and terminating the outstanding URBs!
>
> Going through the usb-skeleton.c source code, you will find that
> usb_get_intfdata() is called from only a few routines:
>
> skel_open()
> skel_disconnect()
> skel_suspend()
> skel_pre_reset()
> skel_post_reset()
>
> Of those, all but the first are called only by the USB core and they are
> mutually exclusive with disconnect processing (except for
> skel_disconnect() itself, of course). So they don't matter.
>
> The first, skel_open(), can be called as a result of actions by the
> user, so the driver needs to ensure that this can't happen after it
> clears the interface-data pointer. The user can open the device file at
> any time before the minor number is given back, so it is not proper to
> call usb_set_intfdata(interface, NULL) before usb_deregister_dev() --
> but the driver does exactly this!
>
> (Well, it's not quite that bad. skel_open() does check whether the
> interface-data pointer value it gets from usb_get_intfdata() is NULL.
> But it's still a race.)
>
> So yes, the current code is wrong. And in fact, it will still be wrong
> even after the usb_set_intfdata(interface, NULL) line is removed,
> because there is no synchronization between skel_open() and
> skel_disconnect().
ACK. I did not look outside of skel_disconnect(). Regardless, I think
that removing the usb_set_intdata(interface, NULL) is still one step
in the good direction despite the other synchronisation issues. I sent
a patch for that which Greg already pick-up:
https://git.kernel.org/gregkh/usb/c/c568f8bb41a4
>It is possible for skel_disconnect() to run to
> completion and the USB core to clear the interface-data pointer all
> while skel_open() is running. The driver needs a static private mutex
> to synchronize opens with unregistrations. (This is a general
> phenomenon, true of all drivers that have a user interface such as a
> device file.)
>
> The driver _does_ have a per-instance mutex, dev->io_mutex, to
> synchronize I/O with disconnects. But that's separate from
> synchronizing opens with unregistrations, because at open time the
> driver doesn't yet know the address of the private data structure or
> even if the structure is still allocated. So obviously it can't use a
> mutex that is embedded within the private data structure for this
> purpose.
ACK. However, I have other priorities, I will not invest more time to
dig in the usb-skeleton.c
Thank you for the answer! That was a long but interesting diversion
from the initial topic :)
Yours sincerely,
Vincent Mailhol
next prev parent reply other threads:[~2022-11-28 23:30 UTC|newest]
Thread overview: 53+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <20221104073659.414147-1-mailhol.vincent@wanadoo.fr>
2022-11-13 4:01 ` [PATCH v3 0/3] can: etas_es58x: report firmware, bootloader and hardware version Vincent Mailhol
2022-11-13 4:01 ` [PATCH v3 1/3] USB: core: export usb_cache_string() Vincent Mailhol
2022-11-22 16:00 ` Greg Kroah-Hartman
2022-11-13 4:01 ` [PATCH v3 2/3] can: etas_es58x: export firmware, bootloader and hardware versions in sysfs Vincent Mailhol
2022-11-15 22:47 ` Saeed Mahameed
2022-11-16 0:36 ` Vincent MAILHOL
2022-11-16 23:20 ` Saeed Mahameed
2022-11-13 4:01 ` [PATCH v3 3/3] can: etas_es58x: report firmware-version through ethtool Vincent Mailhol
2022-11-13 16:48 ` [PATCH v3 0/3] can: etas_es58x: report firmware, bootloader and hardware version Andrew Lunn
2022-11-14 16:49 ` Vincent MAILHOL
2022-11-14 17:08 ` Andrew Lunn
2022-11-26 16:22 ` [PATCH v4 0/6] " Vincent Mailhol
2022-11-26 16:22 ` [PATCH v4 1/6] USB: core: export usb_cache_string() Vincent Mailhol
2022-11-26 16:22 ` [PATCH v4 2/6] can: etas_es58x: add devlink support Vincent Mailhol
2022-11-26 16:51 ` Andrew Lunn
2022-11-27 5:10 ` Vincent MAILHOL
2022-11-27 15:36 ` Alan Stern
2022-11-28 1:34 ` Vincent MAILHOL
2022-11-28 5:32 ` Vincent MAILHOL
2022-11-28 15:50 ` Alan Stern
2022-11-28 23:30 ` Vincent MAILHOL [this message]
2022-11-28 13:42 ` Andrew Lunn
2022-11-28 14:29 ` Vincent MAILHOL
2022-11-26 16:22 ` [PATCH v4 3/6] can: etas_es58x: export product information through devlink_ops::info_get() Vincent Mailhol
2022-11-26 17:16 ` Andrew Lunn
2022-11-27 3:42 ` Vincent MAILHOL
2022-11-27 4:31 ` Vincent MAILHOL
2022-11-27 15:07 ` Andrew Lunn
2022-11-28 1:21 ` Vincent MAILHOL
2022-11-28 13:43 ` Andrew Lunn
2022-11-28 13:47 ` Andrew Lunn
2022-11-28 14:43 ` Vincent MAILHOL
2022-11-28 22:27 ` Jakub Kicinski
2022-11-28 23:17 ` Vincent MAILHOL
2022-11-26 16:22 ` [PATCH v4 4/6] can: etas_es58x: remove es58x_get_product_info() Vincent Mailhol
2022-11-28 13:44 ` Andrew Lunn
2022-11-28 14:36 ` Vincent MAILHOL
2022-11-26 16:22 ` [PATCH v4 5/6] can: etas_es58x: report the firmware version through ethtool Vincent Mailhol
2022-11-28 13:44 ` Andrew Lunn
2022-11-28 22:29 ` Jakub Kicinski
2022-11-29 17:12 ` Vincent MAILHOL
2022-11-30 2:31 ` Jakub Kicinski
2022-11-26 16:22 ` [PATCH v4 6/6] Documentation: devlink: add devlink documentation for the etas_es58x driver Vincent Mailhol
2022-11-30 17:46 ` [PATCH v5 0/7] can: etas_es58x: report firmware, bootloader and hardware version Vincent Mailhol
2022-11-30 17:46 ` [PATCH v5 1/7] can: etas_es58x: add devlink support Vincent Mailhol
2022-11-30 17:46 ` [PATCH v5 2/7] can: etas_es58x: add devlink port support Vincent Mailhol
2022-11-30 17:46 ` [PATCH v5 3/7] USB: core: export usb_cache_string() Vincent Mailhol
2022-11-30 17:46 ` [PATCH v5 4/7] net: devlink: add DEVLINK_INFO_VERSION_GENERIC_FW_BOOTLOADER Vincent Mailhol
2022-11-30 17:46 ` [PATCH v5 5/7] can: etas_es58x: export product information through devlink_ops::info_get() Vincent Mailhol
2022-11-30 17:46 ` [PATCH v5 6/7] can: etas_es58x: remove es58x_get_product_info() Vincent Mailhol
2022-11-30 17:46 ` [PATCH v5 7/7] Documentation: devlink: add devlink documentation for the etas_es58x driver Vincent Mailhol
2022-12-02 12:27 ` Marc Kleine-Budde
2022-12-02 13:15 ` Vincent MAILHOL
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CAMZ6RqKuGC8eyMb45x_oVBTQ0M707JhL-oUs34+8c267cKovAw@mail.gmail.com \
--to=mailhol.vincent@wanadoo.fr \
--cc=andrew@lunn.ch \
--cc=gregkh@linuxfoundation.org \
--cc=jiri@nvidia.com \
--cc=linux-can@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-usb@vger.kernel.org \
--cc=lukas.magel@posteo.net \
--cc=mkl@pengutronix.de \
--cc=netdev@vger.kernel.org \
--cc=saeed@kernel.org \
--cc=stern@rowland.harvard.edu \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).