From: Matthew Garrett <mjg59@srcf.ucam.org>
To: David Brownell <david-b@pacbell.net>
Cc: linux-kernel@vger.kernel.org, gregkh@suse.de
Subject: Re: [PATCH 1/2] Fix /sys/device/.../power/state
Date: Thu, 21 Dec 2006 04:06:48 +0000 [thread overview]
Message-ID: <20061221040648.GA1499@srcf.ucam.org> (raw)
In-Reply-To: <200612201904.28681.david-b@pacbell.net>
On Wed, Dec 20, 2006 at 07:04:28PM -0800, David Brownell wrote:
> On Wednesday 20 December 2006 5:29 pm, Matthew Garrett wrote:
> > I dislike that.
>
> Tough noogies, as they say. In a tradeoff between correctness and your
> personal taste (or even mine, sigh!), the normal tradeoff is in favor
> of correctness.
But it's not correct - the test prohibits suspending devices even if
it would be safe to do so.
> > We're asking to suspend an individual device - whether
> > the bus supports devices that need to suspend with interrupts disabled
> > is irrelevent, it's the device that we care about. We should just make
> > it necessary for every bus to support this method until the interface is
> > removed.
>
> But you _didn't_ do anything to "make it necessary". Which means that
> your patch *WILL* cause bugs whenever a driver uses those calls, and
> courtesy of your patch userspace tries to suspend that device ...
New patch attached.
> > We shouldn't remove interfaces that userland uses until there's been a
> > replacement for long enough that userland can switch over.
>
> Userland can stop using this **TODAY** and just "ifdown", so that
> argument seems weak. For all your examples, the userland interface
> is already available.
No. The proposed interface is currently implemented by three network
drivers out of over seventy. Once the rest (or even the majority of the
rest) are converted to support that, I'd have no objections to the
removal being scheduled.
Signed-off-by: Matthew Garrett <mjg59@srcf.ucam.org>
diff --git a/Documentation/feature-removal-schedule.txt b/Documentation/feature-removal-schedule.txt
index 30f3c8c..8a91689 100644
--- a/Documentation/feature-removal-schedule.txt
+++ b/Documentation/feature-removal-schedule.txt
@@ -9,7 +9,8 @@ be removed from this file.
What: /sys/devices/.../power/state
dev->power.power_state
dpm_runtime_{suspend,resume)()
-When: July 2007
+ bus->pm_has_noirq_stage()
+When: Once alternative functionality has been implemented
Why: Broken design for runtime control over driver power states, confusing
driver-internal runtime power management with: mechanisms to support
system-wide sleep state transitions; event codes that distinguish
diff --git a/Documentation/power/devices.txt b/Documentation/power/devices.txt
index d0e79d5..345cca4 100644
--- a/Documentation/power/devices.txt
+++ b/Documentation/power/devices.txt
@@ -79,6 +79,7 @@ struct bus_type {
int (*resume_early)(struct device *dev);
int (*resume)(struct device *dev);
+ int (*pm_has_noirq_stage)(struct device *dev);
};
Bus drivers implement those methods as appropriate for the hardware and
@@ -236,6 +237,10 @@ The phases are seen by driver notifications issued in this order:
may stay partly usable until this late. This "late" call may also
help when coping with hardware that behaves badly.
+ If a bus implements the suspend_late method, it must also provide a
+ pm_has_noirq_stage function in order to determine whether devices
+ may be suspended during runtime.
+
The pm_message_t parameter is currently used to refine those semantics
(described later).
@@ -348,7 +353,9 @@ The phases are seen by driver notifications issued in this order:
won't be supported on busses that require IRQs in order to
interact with devices.
- This reverses the effects of bus.suspend_late().
+ This reverses the effects of bus.suspend_late(). As with suspend_late,
+ if a bus implements this function it must provide a pm_has_noirq_stage
+ function.
2 bus.resume(dev) is called next. This may be morphed into a device
driver call with bus-specific parameters; implementations may sleep.
diff --git a/drivers/base/platform.c b/drivers/base/platform.c
index f9c903b..6bf1218 100644
--- a/drivers/base/platform.c
+++ b/drivers/base/platform.c
@@ -597,6 +597,16 @@ static int platform_resume(struct device * dev)
return ret;
}
+static int platform_pm_has_noirq_stage(struct device * dev)
+{
+ int ret = 0;
+ struct platform_driver *drv = to_platform_driver(dev->driver);
+
+ if (dev->driver && (drv->resume_early || drv->suspend_late))
+ ret = 1;
+ return ret;
+}
+
struct bus_type platform_bus_type = {
.name = "platform",
.dev_attrs = platform_dev_attrs,
@@ -606,6 +616,7 @@ struct bus_type platform_bus_type = {
.suspend_late = platform_suspend_late,
.resume_early = platform_resume_early,
.resume = platform_resume,
+ .pm_has_noirq_stage = platform_pm_has_noirq_stage,
};
EXPORT_SYMBOL_GPL(platform_bus_type);
diff --git a/drivers/base/power/sysfs.c b/drivers/base/power/sysfs.c
index 2d47517..c4ce060 100644
--- a/drivers/base/power/sysfs.c
+++ b/drivers/base/power/sysfs.c
@@ -46,7 +46,8 @@ static ssize_t state_store(struct device * dev, struct device_attribute *attr, c
int error = -EINVAL;
/* disallow incomplete suspend sequences */
- if (dev->bus && (dev->bus->suspend_late || dev->bus->resume_early))
+ if (dev->bus && dev->bus->pm_has_noirq_stage
+ && dev->bus->pm_has_noirq_stage(dev))
return error;
state.event = PM_EVENT_SUSPEND;
diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
index e5ae3a0..c0e4e7a 100644
--- a/drivers/pci/pci-driver.c
+++ b/drivers/pci/pci-driver.c
@@ -351,6 +351,17 @@ static int pci_device_resume(struct device * dev)
return error;
}
+static int pci_device_pm_has_noirq_stage(struct device * dev)
+{
+ int error = 0;
+ struct pci_dev * pci_dev = to_pci_dev(dev);
+ struct pci_driver * drv = pci_dev->driver;
+
+ if (drv && (drv->resume_early || drv->suspend_late))
+ error = 1;
+ return error;
+}
+
static int pci_device_resume_early(struct device * dev)
{
int error = 0;
@@ -569,6 +580,7 @@ struct bus_type pci_bus_type = {
.suspend_late = pci_device_suspend_late,
.resume_early = pci_device_resume_early,
.resume = pci_device_resume,
+ .pm_has_noirq_stage = pci_device_pm_has_noirq_stage,
.shutdown = pci_device_shutdown,
.dev_attrs = pci_dev_attrs,
};
diff --git a/include/linux/device.h b/include/linux/device.h
index 49ab53c..1c663c4 100644
--- a/include/linux/device.h
+++ b/include/linux/device.h
@@ -59,6 +59,7 @@ struct bus_type {
int (*suspend)(struct device * dev, pm_message_t state);
int (*suspend_late)(struct device * dev, pm_message_t state);
int (*resume_early)(struct device * dev);
+ int (*pm_has_noirq_stage)(struct device * dev);
int (*resume)(struct device * dev);
};
--
Matthew Garrett | mjg59@srcf.ucam.org
next prev parent reply other threads:[~2006-12-21 4:07 UTC|newest]
Thread overview: 119+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-12-19 18:52 Changes to sysfs PM layer break userspace Matthew Garrett
2006-12-19 19:34 ` Arjan van de Ven
2006-12-19 19:44 ` Matthew Garrett
2006-12-19 20:03 ` Arjan van de Ven
2006-12-19 20:08 ` Matthew Garrett
2006-12-19 20:23 ` Arjan van de Ven
2006-12-19 20:32 ` Matthew Garrett
2006-12-19 20:55 ` Arjan van de Ven
2006-12-21 0:08 ` Kyle Moffett
2006-12-19 21:34 ` David Brownell
2006-12-20 0:25 ` Matthew Garrett
2006-12-20 3:59 ` Changes to " David Brownell
2006-12-20 4:26 ` Matthew Garrett
2006-12-20 5:14 ` David Brownell
2006-12-20 5:34 ` Greg KH
2006-12-20 5:52 ` Matthew Garrett
2006-12-20 7:50 ` Arjan van de Ven
2006-12-20 12:53 ` Network drivers that don't suspend on interface down Matthew Garrett
2006-12-20 13:38 ` Arjan van de Ven
2006-12-20 14:31 ` Matthew Garrett
2006-12-20 15:51 ` Arjan van de Ven
2006-12-20 22:49 ` Stephen Hemminger
2006-12-20 23:37 ` Rick Jones
2006-12-19 23:51 ` Stephen Hemminger
2006-12-21 0:11 ` Francois Romieu
2006-12-20 0:26 ` Stephen Hemminger
2006-12-21 11:18 ` Francois Romieu
2006-12-21 1:12 ` Matthew Garrett
2006-12-21 2:05 ` Michael Wu
2006-12-21 2:18 ` Matthew Garrett
2006-12-21 2:38 ` Daniel Drake
2006-12-21 2:45 ` Matthew Garrett
2006-12-21 3:08 ` Daniel Drake
2006-12-21 3:25 ` Matthew Garrett
2006-12-21 3:37 ` Dan Williams
2006-12-21 3:29 ` Dan Williams
2006-12-21 3:14 ` Dan Williams
2006-12-21 13:14 ` jamal
2006-12-21 2:29 ` Daniel Drake
2006-12-21 2:10 ` Jesse Brandeburg
2006-12-21 8:54 ` Arjan van de Ven
2006-12-22 1:03 ` Herbert Xu
2006-12-23 8:54 ` Pavel Machek
2006-12-20 15:27 ` Olivier Galibert
2006-12-20 15:34 ` Arjan van de Ven
2006-12-20 16:40 ` Olivier Galibert
2006-12-20 17:21 ` Arjan van de Ven
2006-12-20 20:40 ` Benny Amorsen
2006-12-20 21:49 ` Arjan van de Ven
2006-12-20 21:15 ` Stefan Rompf
2006-12-20 14:00 ` Jiri Benc
2006-12-20 18:12 ` Dan Williams
2006-12-21 1:15 ` Matthew Garrett
2006-12-21 1:57 ` Michael Wu
2006-12-21 2:20 ` Matthew Garrett
2006-12-21 3:02 ` Dan Williams
2006-12-21 3:06 ` Dan Williams
2006-12-21 3:14 ` Matthew Garrett
2006-12-21 3:32 ` Dan Williams
2006-12-21 13:19 ` Sven-Haegar Koch
2006-12-21 17:16 ` Dan Williams
2006-12-21 18:27 ` Valdis.Kletnieks
2006-12-22 1:25 ` Matt Domsch
2006-12-20 16:04 ` Maciej W. Rozycki
2006-12-22 21:09 ` Changes to PM layer break userspace Pavel Machek
2006-12-24 7:02 ` David Brownell
2006-12-28 13:31 ` Alan
2006-12-28 16:04 ` Arjan van de Ven
2006-12-29 5:27 ` David Brownell
2006-12-20 2:15 ` Changes to sysfs " Andrew Morton
2006-12-20 2:35 ` Randy Dunlap
2006-12-20 3:49 ` Andrew Morton
2006-12-20 3:29 ` David Brownell
2006-12-21 3:51 ` Andrew Morton
2006-12-21 4:56 ` David Brownell
2006-12-21 5:02 ` Andrew Morton
2006-12-21 7:05 ` David Brownell
2006-12-21 8:27 ` Arjan van de Ven
2006-12-22 20:44 ` Pavel Machek
2006-12-23 14:02 ` Stefan Seyfried
2006-12-19 21:22 ` David Brownell
2006-12-19 22:57 ` Matthew Garrett
2006-12-19 23:36 ` Changes to " David Brownell
2006-12-20 0:09 ` Matthew Garrett
2006-12-20 3:19 ` David Brownell
2006-12-20 3:43 ` Matthew Garrett
2006-12-20 4:15 ` David Brownell
2006-12-20 4:56 ` [PATCH 1/2] Fix /sys/device/.../power/state Matthew Garrett
2006-12-20 21:18 ` David Brownell
2006-12-21 1:29 ` Matthew Garrett
2006-12-21 3:04 ` David Brownell
2006-12-21 4:06 ` Matthew Garrett [this message]
2006-12-21 4:51 ` David Brownell
2007-01-25 5:00 ` [PATCH] Fix /sys/device/.../power/state regression Matthew Garrett
2007-01-26 19:59 ` Andrew Morton
2007-01-26 20:42 ` Greg KH
2007-01-27 1:26 ` Matthew Garrett
2007-01-27 17:19 ` Pavel Machek
2007-01-26 20:41 ` Greg KH
2007-01-26 21:56 ` David Brownell
2007-01-26 22:19 ` Greg KH
2007-01-26 23:15 ` Matthew Garrett
2007-01-27 0:42 ` David Brownell
2007-01-27 0:56 ` Andrew Morton
2007-01-27 2:40 ` David Brownell
2007-01-27 17:38 ` Pavel Machek
2007-01-27 22:42 ` Matthew Garrett
2007-01-31 10:34 ` Pavel Machek
2007-01-27 1:19 ` Matthew Garrett
2007-01-27 3:02 ` David Brownell
2007-01-29 17:36 ` Stephen Hemminger
2007-01-27 13:17 ` Pavel Machek
2007-01-30 18:43 ` Eric Piel
2007-01-30 18:47 ` Oliver Neukum
2007-01-30 19:01 ` Eric Piel
2007-01-31 10:55 ` Jiri Kosina
2007-01-31 11:03 ` Oliver Neukum
2006-12-20 4:56 ` [PATCH 2/2] Update feature-removal-schedule.txt Matthew Garrett
2006-12-22 20:47 ` Changes to PM layer break userspace Pavel Machek
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=20061221040648.GA1499@srcf.ucam.org \
--to=mjg59@srcf.ucam.org \
--cc=david-b@pacbell.net \
--cc=gregkh@suse.de \
--cc=linux-kernel@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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox