From: Oliver Neukum <oneukum@suse.com>
To: Marek Vasut <marex@denx.de>, netdev@vger.kernel.org
Cc: "David S . Miller" <davem@davemloft.net>,
Florian Fainelli <f.fainelli@gmail.com>,
Andrew Lunn <andrew@lunn.ch>,
Nisar Sayed <Nisar.Sayed@microchip.com>,
Woojung Huh <Woojung.Huh@microchip.com>,
linux-usb@vger.kernel.org
Subject: [02/19] usbnet: smsc95xx: Stop propagation of in_pm
Date: Mon, 07 Jan 2019 12:02:50 +0100 [thread overview]
Message-ID: <1546858970.3037.21.camel@suse.com> (raw)
On Do, 2019-01-03 at 02:10 +0100, Marek Vasut wrote:
> The information whether the SMSC95xx is in_pm or not can be derived from
> the usbdev->suspend_count. First thing called in smsc95xx_suspend() is
> usbnet_suspend(), which increments the usbdev->suspend_count and since
> then the driver only calls _nopm() functions and functions with in_pm
> set to 1. The smsc95xx_resume() does the inverse, it calls functions
> with _nopm() suffix and with in_pm set to 1 until usbnet_resume(), which
> sets the usbdev->suspend_count to 0.
Hi,
unfortunately I am forced to disagree in the strongest terms.
The logic behind this patch is wrong. The "in_pm" parameter
exists because some function need to be called in the code paths
needed to implement power management. Under those circumstances
they must not take a pm reference to keep the device awake, lest
the driver deadlock.
(For example __smsc95xx_read_reg)
However from other code paths precisely that reference must be taken
in order to make sure that the driver do not try to communicate with
a suspended device.
If you check the suspend counter, you will omit the necessary getting
of a reference precisely when it is needed. The driver will never
dedalock, but you will cause numerous races.
I must suggest to simply drop this part and redo the series.
Regards
Oliver
WARNING: multiple messages have this Message-ID (diff)
From: Oliver Neukum <oneukum@suse.com>
To: Marek Vasut <marex@denx.de>, netdev@vger.kernel.org
Cc: "David S . Miller" <davem@davemloft.net>,
Florian Fainelli <f.fainelli@gmail.com>,
Andrew Lunn <andrew@lunn.ch>,
Nisar Sayed <Nisar.Sayed@microchip.com>,
Woojung Huh <Woojung.Huh@microchip.com>,
linux-usb@vger.kernel.org
Subject: Re: [PATCH 02/19] usbnet: smsc95xx: Stop propagation of in_pm
Date: Mon, 07 Jan 2019 12:02:50 +0100 [thread overview]
Message-ID: <1546858970.3037.21.camel@suse.com> (raw)
In-Reply-To: <20190103011040.25974-3-marex@denx.de>
On Do, 2019-01-03 at 02:10 +0100, Marek Vasut wrote:
> The information whether the SMSC95xx is in_pm or not can be derived from
> the usbdev->suspend_count. First thing called in smsc95xx_suspend() is
> usbnet_suspend(), which increments the usbdev->suspend_count and since
> then the driver only calls _nopm() functions and functions with in_pm
> set to 1. The smsc95xx_resume() does the inverse, it calls functions
> with _nopm() suffix and with in_pm set to 1 until usbnet_resume(), which
> sets the usbdev->suspend_count to 0.
Hi,
unfortunately I am forced to disagree in the strongest terms.
The logic behind this patch is wrong. The "in_pm" parameter
exists because some function need to be called in the code paths
needed to implement power management. Under those circumstances
they must not take a pm reference to keep the device awake, lest
the driver deadlock.
(For example __smsc95xx_read_reg)
However from other code paths precisely that reference must be taken
in order to make sure that the driver do not try to communicate with
a suspended device.
If you check the suspend counter, you will omit the necessary getting
of a reference precisely when it is needed. The driver will never
dedalock, but you will cause numerous races.
I must suggest to simply drop this part and redo the series.
Regards
Oliver
next reply other threads:[~2019-01-07 11:02 UTC|newest]
Thread overview: 65+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-01-07 11:02 Oliver Neukum [this message]
2019-01-07 11:02 ` [PATCH 02/19] usbnet: smsc95xx: Stop propagation of in_pm Oliver Neukum
-- strict thread matches above, loose matches on Subject: below --
2019-01-07 11:50 [02/19] " Marek Vasut
2019-01-07 11:50 ` [PATCH 02/19] " Marek Vasut
2019-01-07 11:05 [07/19] usbnet: smsc95xx: Split the reset function Oliver Neukum
2019-01-07 11:05 ` [PATCH 07/19] " Oliver Neukum
2019-01-04 18:02 [10/19] usbnet: smsc95xx: Replace smsc95xx_mdio_read() with phy_read() Heiner Kallweit
2019-01-04 18:02 ` [PATCH 10/19] " Heiner Kallweit
2019-01-04 15:57 [10/19] " Andrew Lunn
2019-01-04 15:57 ` [PATCH 10/19] " Andrew Lunn
2019-01-04 14:58 [10/19] " Marek Vasut
2019-01-04 14:58 ` [PATCH 10/19] " Marek Vasut
2019-01-04 13:24 [10/19] " Andrew Lunn
2019-01-04 13:24 ` [PATCH 10/19] " Andrew Lunn
2019-01-04 2:19 [12/19] usbnet: smsc95xx: Replace ad-hoc PHY functions with generic ones Marek Vasut
2019-01-04 2:19 ` [PATCH 12/19] " Marek Vasut
2019-01-04 2:19 [10/19] usbnet: smsc95xx: Replace smsc95xx_mdio_read() with phy_read() Marek Vasut
2019-01-04 2:19 ` [PATCH 10/19] " Marek Vasut
2019-01-04 2:18 [09/19] usbnet: smsc95xx: Connect to phydev Marek Vasut
2019-01-04 2:18 ` [PATCH 09/19] " Marek Vasut
2019-01-03 13:39 [10/19] usbnet: smsc95xx: Replace smsc95xx_mdio_read() with phy_read() Andrew Lunn
2019-01-03 13:39 ` [PATCH 10/19] " Andrew Lunn
2019-01-03 13:28 [12/19] usbnet: smsc95xx: Replace ad-hoc PHY functions with generic ones Andrew Lunn
2019-01-03 13:28 ` [PATCH 12/19] " Andrew Lunn
2019-01-03 13:22 [09/19] usbnet: smsc95xx: Connect to phydev Andrew Lunn
2019-01-03 13:22 ` [PATCH 09/19] " Andrew Lunn
2019-01-03 1:10 [19/19] usbnet: smsc95xx: Use phy bit operations Marek Vasut
2019-01-03 1:10 ` [PATCH 19/19] " Marek Vasut
2019-01-03 1:10 [18/19] usbnet: smsc95xx: Remove all of the carrier checking code Marek Vasut
2019-01-03 1:10 ` [PATCH 18/19] " Marek Vasut
2019-01-03 1:10 [17/19] usbnet: smsc95xx: Replace generic_mii_ioctl() Marek Vasut
2019-01-03 1:10 ` [PATCH 17/19] " Marek Vasut
2019-01-03 1:10 [16/19] usbnet: smsc95xx: Replace mii_check_media() Marek Vasut
2019-01-03 1:10 ` [PATCH 16/19] " Marek Vasut
2019-01-03 1:10 [15/19] usbnet: smsc95xx: Replace mii_ethtool_gset() Marek Vasut
2019-01-03 1:10 ` [PATCH 15/19] " Marek Vasut
2019-01-03 1:10 [14/19] usbnet: smsc95xx: Replace mii_nway_restart() Marek Vasut
2019-01-03 1:10 ` [PATCH 14/19] " Marek Vasut
2019-01-03 1:10 [13/19] usbnet: smsc95xx: Replace smsc95xx_link_ok_nopm() Marek Vasut
2019-01-03 1:10 ` [PATCH 13/19] " Marek Vasut
2019-01-03 1:10 [12/19] usbnet: smsc95xx: Replace ad-hoc PHY functions with generic ones Marek Vasut
2019-01-03 1:10 ` [PATCH 12/19] " Marek Vasut
2019-01-03 1:10 [11/19] usbnet: smsc95xx: Replace smsc95xx_mdio_write() with phy_write() Marek Vasut
2019-01-03 1:10 ` [PATCH 11/19] " Marek Vasut
2019-01-03 1:10 [10/19] usbnet: smsc95xx: Replace smsc95xx_mdio_read() with phy_read() Marek Vasut
2019-01-03 1:10 ` [PATCH 10/19] " Marek Vasut
2019-01-03 1:10 [09/19] usbnet: smsc95xx: Connect to phydev Marek Vasut
2019-01-03 1:10 ` [PATCH 09/19] " Marek Vasut
2019-01-03 1:10 [08/19] usbnet: smsc95xx: Register MII bus Marek Vasut
2019-01-03 1:10 ` [PATCH 08/19] " Marek Vasut
2019-01-03 1:10 [07/19] usbnet: smsc95xx: Split the reset function Marek Vasut
2019-01-03 1:10 ` [PATCH 07/19] " Marek Vasut
2019-01-03 1:10 [06/19] usbnet: smsc95xx: Remove __smsc95xx_mdio_{read,write}() Marek Vasut
2019-01-03 1:10 ` [PATCH 06/19] " Marek Vasut
2019-01-03 1:10 [05/19] usbnet: smsc95xx: Remove smsc95xx_{read,write}_nopm() Marek Vasut
2019-01-03 1:10 ` [PATCH 05/19] " Marek Vasut
2019-01-03 1:10 [04/19] usbnet: smsc95xx: Remove __smsc95xx_{read,write}_reg() Marek Vasut
2019-01-03 1:10 ` [PATCH 04/19] " Marek Vasut
2019-01-03 1:10 [03/19] usbnet: smsc95xx: Remove smsc95xx_{read,write}_reg_nopm() Marek Vasut
2019-01-03 1:10 ` [PATCH 03/19] " Marek Vasut
2019-01-03 1:10 [02/19] usbnet: smsc95xx: Stop propagation of in_pm Marek Vasut
2019-01-03 1:10 ` [PATCH 02/19] " Marek Vasut
2019-01-03 1:10 [01/19] usbnet: smsc95xx: Fix memory leak in smsc95xx_bind Marek Vasut
2019-01-03 1:10 ` [PATCH 01/19] " Marek Vasut
2019-01-03 1:10 [PATCH 00/19] usbnet: smsc95xx: Convert to phydev Marek Vasut
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=1546858970.3037.21.camel@suse.com \
--to=oneukum@suse.com \
--cc=Nisar.Sayed@microchip.com \
--cc=Woojung.Huh@microchip.com \
--cc=andrew@lunn.ch \
--cc=davem@davemloft.net \
--cc=f.fainelli@gmail.com \
--cc=linux-usb@vger.kernel.org \
--cc=marex@denx.de \
--cc=netdev@vger.kernel.org \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.