[PATCH v4 0/22] On-demand device probing
3954d
Revision v4 of 6 in this series.
6 messages,
3 authors,
2015-09-09 · open the first message on its own page
Hello,
I have a problem with the panel on my Tegra Chromebook taking longer
than expected to be ready during boot (Stéphane Marchesin reported what
is basically the same issue in [0]), and have looked into ordered
probing as a better way of solving this than moving nodes around in the
DT or playing with initcall levels and linking order.
While reading the thread [1] that Alexander Holler started with his
series to make probing order deterministic, it occurred to me that it
should be possible to achieve the same by probing devices as they are
referenced by other devices.
This basically reuses the information that is already implicit in the
probe() implementations, saving us from refactoring existing drivers or
adding information to DTBs.
During review of v1 of this series Linus Walleij suggested that it
should be the device driver core to make sure that dependencies are
ready before probing a device. I gave this idea a try [2] but Mark Brown
pointed out to the logic duplication between the resource acquisition
and dependency discovery code paths (though I think it's fairly minor).
To address that code duplication I experimented with Arnd's devm_probe
[3] concept of having drivers declare their dependencies instead of
acquiring them during probe, and while it worked [4], I don't think we
end up winning anything when compared to just probing devices on-demand
from resource getters.
One remaining objection is to the "sprinkling" of calls to
of_device_probe() in the resource getters of each subsystem, but I think
it's the right thing to do given that the storage of resources is
currently subsystem-specific.
We could avoid the above by moving resource storage into the core, but I
don't think there's a compelling case for that.
I have tested this on boards with Tegra, iMX.6, Exynos, Rockchip and
OMAP SoCs, and these patches were enough to eliminate all the deferred
probes (except one in PandaBoard because omap_dma_system doesn't have a
firmware node as of yet).
Have submitted a branch [5] with only these patches on top of thursday's
linux-next to kernelci.org and I don't see any issues that could be
caused by them. For some reason it currently has more passes than the
version of -next it's based on!
With this series I get the kernel to output to the panel in 0.5s,
instead of 2.8s.
Regards,
Tomeu
[0] http://lists.freedesktop.org/archives/dri-devel/2014-August/066527.html
[1] https://lkml.org/lkml/2014/5/12/452
[2] https://lkml.org/lkml/2015/6/17/305
[3] http://article.gmane.org/gmane.linux.ports.arm.kernel/277689
[4] https://lkml.org/lkml/2015/7/21/441a
[5] https://git.collabora.com/cgit/user/tomeu/linux.git/log/?h=on-demand-probes-v6
[6] http://kernelci.org/boot/all/job/collabora/kernel/v4.2-11902-g25d80c927f8b/
[7] http://kernelci.org/boot/all/job/next/kernel/next-20150903/
Changes in v4:
- Added bus.pre_probe callback so the probes of Primecell devices can be
deferred if their device IDs cannot be yet read because of the clock
driver not having probed when they are registered. Maybe this goes
overboard and the matching information should be in the DT if there is
one.
- Rename of_platform_probe to of_device_probe
- Use device_node.device instead of device_node.platform_dev
- Take a reference to the regulator's device to prevent dangling
pointers
- Add Kconfig DELAY_DEVICE_PROBES to allow disabling delayed probing in
machines with initcalls that depend on devices probing at a given time.
- Start processing deferred probes in device_initcall_sync
- Also defer probes of AMBA devices registered from the DT as they can
also request resources.
Changes in v3:
- Set and use device_node.platform_dev instead of reversing the logic to
find the platform device that encloses a device node.
- Drop the fwnode API to probe firmware nodes and add OF-only API for
now. I think this same scheme could be used for machines with ACPI,
but I haven't been able to find one that had to defer its probes because
of the device probe order.
- Avoid unlocking the regulator device's mutex if we don't have a device
Changes in v2:
- Acquire regulator device lock before returning from regulator_dev_lookup()
Tomeu Vizoso (22):
driver core: Add pre_probe callback to bus_type
ARM: amba: Move reading of periphid to pre_probe()
of/platform: Point to struct device from device node
of: add function to allow probing a device from a OF node
gpio: Probe GPIO drivers on demand
gpio: Probe pinctrl devices on demand
regulator: core: Reduce critical area in _regulator_get
regulator: core: Probe regulators on demand
drm: Probe panels on demand
drm/tegra: Probe dpaux devices on demand
i2c: core: Probe i2c adapters and devices on demand
pwm: Probe PWM chip devices on demand
backlight: Probe backlight devices on demand
usb: phy: Probe phy devices on demand
clk: Probe clk providers on demand
pinctrl: Probe pinctrl devices on demand
phy: core: Probe phy providers on demand
dma: of: Probe DMA controllers on demand
power-supply: Probe power supplies on demand
driver core: Allow deferring probes until late init
driver core: Start processing deferred probes earlier
of/platform: Defer probes of registered devices
drivers/amba/bus.c | 78 ++++++++++++++++++-------------------
drivers/base/Kconfig | 18 +++++++++
drivers/base/dd.c | 35 ++++++++++++++++-
drivers/clk/clk.c | 3 ++
drivers/dma/of-dma.c | 3 ++
drivers/gpio/gpiolib-of.c | 5 +++
drivers/gpu/drm/drm_panel.c | 3 ++
drivers/gpu/drm/tegra/dpaux.c | 3 ++
drivers/i2c/i2c-core.c | 4 ++
drivers/of/device.c | 58 +++++++++++++++++++++++++++
drivers/of/platform.c | 26 +++++++------
drivers/phy/phy-core.c | 3 ++
drivers/pinctrl/devicetree.c | 3 ++
drivers/power/power_supply_core.c | 3 ++
drivers/pwm/core.c | 3 ++
drivers/regulator/core.c | 58 ++++++++++++++++++---------
drivers/usb/phy/phy.c | 3 ++
drivers/video/backlight/backlight.c | 3 ++
include/linux/device.h | 6 +++
include/linux/of.h | 1 +
include/linux/of_device.h | 3 ++
21 files changed, 251 insertions(+), 71 deletions(-)
--
2.4.3
When looking up a backlight device through its OF node, probe it if it
hasn't already.
The goal is to reduce deferred probes to a minimum, as it makes it very
cumbersome to find out why a device failed to probe, and can introduce
very big delays in when a critical device is probed.
Signed-off-by: Tomeu Vizoso <redacted>
---
Changes in v4: None
Changes in v3: None
Changes in v2: None
drivers/video/backlight/backlight.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drivers/video/backlight/backlight.c b/drivers/video/backlight/backlight.c
index bddc8b17a4d8..9bcdc16eacdf 100644
--- a/drivers/video/backlight/backlight.c
+++ b/drivers/video/backlight/backlight.c @@ -16,6 +16,7 @@
#include <linux/err.h>
#include <linux/fb.h>
#include <linux/slab.h>
+ #include <linux/of_device.h>
#ifdef CONFIG_PMAC_BACKLIGHT
#include <asm/backlight.h> @@ -559,6 +560,8 @@ struct backlight_device *of_find_backlight_by_node(struct device_node *node)
{
struct device * dev ;
+ of_device_probe ( node ); + dev = class_find_device ( backlight_class , NULL , node , of_parent_match );
return dev ? to_backlight_device ( dev ) : NULL ; --
2.4.3
On Mon, Sep 7, 2015 at 7:23 AM, Tomeu Vizoso [off-list ref] wrote: Hello,
I have a problem with the panel on my Tegra Chromebook taking longer
than expected to be ready during boot (Stéphane Marchesin reported what
is basically the same issue in [0]), and have looked into ordered
probing as a better way of solving this than moving nodes around in the
DT or playing with initcall levels and linking order.
While reading the thread [1] that Alexander Holler started with his
series to make probing order deterministic, it occurred to me that it
should be possible to achieve the same by probing devices as they are
referenced by other devices.
This basically reuses the information that is already implicit in the
probe() implementations, saving us from refactoring existing drivers or
adding information to DTBs.
During review of v1 of this series Linus Walleij suggested that it
should be the device driver core to make sure that dependencies are
ready before probing a device. I gave this idea a try [2] but Mark Brown
pointed out to the logic duplication between the resource acquisition
and dependency discovery code paths (though I think it's fairly minor).
To address that code duplication I experimented with Arnd's devm_probe
[3] concept of having drivers declare their dependencies instead of
acquiring them during probe, and while it worked [4], I don't think we
end up winning anything when compared to just probing devices on-demand
from resource getters.
One remaining objection is to the "sprinkling" of calls to
of_device_probe() in the resource getters of each subsystem, but I think
it's the right thing to do given that the storage of resources is
currently subsystem-specific.
We could avoid the above by moving resource storage into the core, but I
don't think there's a compelling case for that.
I have tested this on boards with Tegra, iMX.6, Exynos, Rockchip and
OMAP SoCs, and these patches were enough to eliminate all the deferred
probes (except one in PandaBoard because omap_dma_system doesn't have a
firmware node as of yet).
Have submitted a branch [5] with only these patches on top of thursday's
linux-next to kernelci.org and I don't see any issues that could be
caused by them. For some reason it currently has more passes than the
version of -next it's based on!
With this series I get the kernel to output to the panel in 0.5s,
instead of 2.8s.
Regards,
Tomeu
[0] http://lists.freedesktop.org/archives/dri-devel/2014-August/066527.html
[1] https://lkml.org/lkml/2014/5/12/452
[2] https://lkml.org/lkml/2015/6/17/305
[3] http://article.gmane.org/gmane.linux.ports.arm.kernel/277689
[4] https://lkml.org/lkml/2015/7/21/441a
[5] https://git.collabora.com/cgit/user/tomeu/linux.git/log/?h=on-demand-probes-v6
[6] http://kernelci.org/boot/all/job/collabora/kernel/v4.2-11902-g25d80c927f8b/
[7] http://kernelci.org/boot/all/job/next/kernel/next-20150903/
Changes in v4:
- Added bus.pre_probe callback so the probes of Primecell devices can be
deferred if their device IDs cannot be yet read because of the clock
driver not having probed when they are registered. Maybe this goes
overboard and the matching information should be in the DT if there is
one.
Seems overboard to me or at least a separate problem. Most clocks have
to be setup before the driver model simply because timers depend on
clocks usually.
Rob
On 7 September 2015 at 22:50, Rob Herring [off-list ref] wrote: On Mon, Sep 7, 2015 at 7:23 AM, Tomeu Vizoso [off-list ref] wrote: quoted Hello,
I have a problem with the panel on my Tegra Chromebook taking longer
than expected to be ready during boot (Stéphane Marchesin reported what
is basically the same issue in [0]), and have looked into ordered
probing as a better way of solving this than moving nodes around in the
DT or playing with initcall levels and linking order.
While reading the thread [1] that Alexander Holler started with his
series to make probing order deterministic, it occurred to me that it
should be possible to achieve the same by probing devices as they are
referenced by other devices.
This basically reuses the information that is already implicit in the
probe() implementations, saving us from refactoring existing drivers or
adding information to DTBs.
During review of v1 of this series Linus Walleij suggested that it
should be the device driver core to make sure that dependencies are
ready before probing a device. I gave this idea a try [2] but Mark Brown
pointed out to the logic duplication between the resource acquisition
and dependency discovery code paths (though I think it's fairly minor).
To address that code duplication I experimented with Arnd's devm_probe
[3] concept of having drivers declare their dependencies instead of
acquiring them during probe, and while it worked [4], I don't think we
end up winning anything when compared to just probing devices on-demand
from resource getters.
One remaining objection is to the "sprinkling" of calls to
of_device_probe() in the resource getters of each subsystem, but I think
it's the right thing to do given that the storage of resources is
currently subsystem-specific.
We could avoid the above by moving resource storage into the core, but I
don't think there's a compelling case for that.
I have tested this on boards with Tegra, iMX.6, Exynos, Rockchip and
OMAP SoCs, and these patches were enough to eliminate all the deferred
probes (except one in PandaBoard because omap_dma_system doesn't have a
firmware node as of yet).
Have submitted a branch [5] with only these patches on top of thursday's
linux-next to kernelci.org and I don't see any issues that could be
caused by them. For some reason it currently has more passes than the
version of -next it's based on!
With this series I get the kernel to output to the panel in 0.5s,
instead of 2.8s.
Regards,
Tomeu
[0] http://lists.freedesktop.org/archives/dri-devel/2014-August/066527.html
[1] https://lkml.org/lkml/2014/5/12/452
[2] https://lkml.org/lkml/2015/6/17/305
[3] http://article.gmane.org/gmane.linux.ports.arm.kernel/277689
[4] https://lkml.org/lkml/2015/7/21/441a
[5] https://git.collabora.com/cgit/user/tomeu/linux.git/log/?h=on-demand-probes-v6
[6] http://kernelci.org/boot/all/job/collabora/kernel/v4.2-11902-g25d80c927f8b/
[7] http://kernelci.org/boot/all/job/next/kernel/next-20150903/
Changes in v4:
- Added bus.pre_probe callback so the probes of Primecell devices can be
deferred if their device IDs cannot be yet read because of the clock
driver not having probed when they are registered. Maybe this goes
overboard and the matching information should be in the DT if there is
one.
Seems overboard to me or at least a separate problem.
It's a separate problem but this was preventing the series from
working on a few boards.
Most clocks have
to be setup before the driver model simply because timers depend on
clocks usually.
Yes, but in this case the apb clocks for the primecell devices are
implemented in a normal platform driver (vexpress_osc_driver), instead
of using CLK_OF_DECLARE.
Regards,
Tomeu
Rob
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
On 09/08/2015 02:30 AM, Tomeu Vizoso wrote: On 7 September 2015 at 22:50, Rob Herring [off-list ref] wrote: quoted On Mon, Sep 7, 2015 at 7:23 AM, Tomeu Vizoso [off-list ref] wrote: quoted Hello,
I have a problem with the panel on my Tegra Chromebook taking longer
than expected to be ready during boot (Stéphane Marchesin reported what
is basically the same issue in [0]), and have looked into ordered
probing as a better way of solving this than moving nodes around in the
DT or playing with initcall levels and linking order.
While reading the thread [1] that Alexander Holler started with his
series to make probing order deterministic, it occurred to me that it
should be possible to achieve the same by probing devices as they are
referenced by other devices.
This basically reuses the information that is already implicit in the
probe() implementations, saving us from refactoring existing drivers or
adding information to DTBs.
During review of v1 of this series Linus Walleij suggested that it
should be the device driver core to make sure that dependencies are
ready before probing a device. I gave this idea a try [2] but Mark Brown
pointed out to the logic duplication between the resource acquisition
and dependency discovery code paths (though I think it's fairly minor).
To address that code duplication I experimented with Arnd's devm_probe
[3] concept of having drivers declare their dependencies instead of
acquiring them during probe, and while it worked [4], I don't think we
end up winning anything when compared to just probing devices on-demand
from resource getters.
One remaining objection is to the "sprinkling" of calls to
of_device_probe() in the resource getters of each subsystem, but I think
it's the right thing to do given that the storage of resources is
currently subsystem-specific.
We could avoid the above by moving resource storage into the core, but I
don't think there's a compelling case for that.
I have tested this on boards with Tegra, iMX.6, Exynos, Rockchip and
OMAP SoCs, and these patches were enough to eliminate all the deferred
probes (except one in PandaBoard because omap_dma_system doesn't have a
firmware node as of yet).
Have submitted a branch [5] with only these patches on top of thursday's
linux-next to kernelci.org and I don't see any issues that could be
caused by them. For some reason it currently has more passes than the
version of -next it's based on!
With this series I get the kernel to output to the panel in 0.5s,
instead of 2.8s.
Regards,
Tomeu
[0] http://lists.freedesktop.org/archives/dri-devel/2014-August/066527.html
[1] https://lkml.org/lkml/2014/5/12/452
[2] https://lkml.org/lkml/2015/6/17/305
[3] http://article.gmane.org/gmane.linux.ports.arm.kernel/277689
[4] https://lkml.org/lkml/2015/7/21/441a
[5] https://git.collabora.com/cgit/user/tomeu/linux.git/log/?h=on-demand-probes-v6
[6] http://kernelci.org/boot/all/job/collabora/kernel/v4.2-11902-g25d80c927f8b/
[7] http://kernelci.org/boot/all/job/next/kernel/next-20150903/
Changes in v4:
- Added bus.pre_probe callback so the probes of Primecell devices can be
deferred if their device IDs cannot be yet read because of the clock
driver not having probed when they are registered. Maybe this goes
overboard and the matching information should be in the DT if there is
one.
Seems overboard to me or at least a separate problem.
It's a separate problem but this was preventing the series from
working on a few boards.
What is the failure? Not booting? Fixing not working would certainly not
be overboard.
quoted Most clocks have
to be setup before the driver model simply because timers depend on
clocks usually.
Yes, but in this case the apb clocks for the primecell devices are
implemented in a normal platform driver (vexpress_osc_driver), instead
of using CLK_OF_DECLARE.
Okay.
Rob
On 9 September 2015 at 03:33, Rob Herring [off-list ref] wrote: On 09/08/2015 02:30 AM, Tomeu Vizoso wrote: quoted On 7 September 2015 at 22:50, Rob Herring [off-list ref] wrote: quoted On Mon, Sep 7, 2015 at 7:23 AM, Tomeu Vizoso [off-list ref] wrote: quoted Hello,
I have a problem with the panel on my Tegra Chromebook taking longer
than expected to be ready during boot (Stéphane Marchesin reported what
is basically the same issue in [0]), and have looked into ordered
probing as a better way of solving this than moving nodes around in the
DT or playing with initcall levels and linking order.
While reading the thread [1] that Alexander Holler started with his
series to make probing order deterministic, it occurred to me that it
should be possible to achieve the same by probing devices as they are
referenced by other devices.
This basically reuses the information that is already implicit in the
probe() implementations, saving us from refactoring existing drivers or
adding information to DTBs.
During review of v1 of this series Linus Walleij suggested that it
should be the device driver core to make sure that dependencies are
ready before probing a device. I gave this idea a try [2] but Mark Brown
pointed out to the logic duplication between the resource acquisition
and dependency discovery code paths (though I think it's fairly minor).
To address that code duplication I experimented with Arnd's devm_probe
[3] concept of having drivers declare their dependencies instead of
acquiring them during probe, and while it worked [4], I don't think we
end up winning anything when compared to just probing devices on-demand
from resource getters.
One remaining objection is to the "sprinkling" of calls to
of_device_probe() in the resource getters of each subsystem, but I think
it's the right thing to do given that the storage of resources is
currently subsystem-specific.
We could avoid the above by moving resource storage into the core, but I
don't think there's a compelling case for that.
I have tested this on boards with Tegra, iMX.6, Exynos, Rockchip and
OMAP SoCs, and these patches were enough to eliminate all the deferred
probes (except one in PandaBoard because omap_dma_system doesn't have a
firmware node as of yet).
Have submitted a branch [5] with only these patches on top of thursday's
linux-next to kernelci.org and I don't see any issues that could be
caused by them. For some reason it currently has more passes than the
version of -next it's based on!
With this series I get the kernel to output to the panel in 0.5s,
instead of 2.8s.
Regards,
Tomeu
[0] http://lists.freedesktop.org/archives/dri-devel/2014-August/066527.html
[1] https://lkml.org/lkml/2014/5/12/452
[2] https://lkml.org/lkml/2015/6/17/305
[3] http://article.gmane.org/gmane.linux.ports.arm.kernel/277689
[4] https://lkml.org/lkml/2015/7/21/441a
[5] https://git.collabora.com/cgit/user/tomeu/linux.git/log/?h=on-demand-probes-v6
[6] http://kernelci.org/boot/all/job/collabora/kernel/v4.2-11902-g25d80c927f8b/
[7] http://kernelci.org/boot/all/job/next/kernel/next-20150903/
Changes in v4:
- Added bus.pre_probe callback so the probes of Primecell devices can be
deferred if their device IDs cannot be yet read because of the clock
driver not having probed when they are registered. Maybe this goes
overboard and the matching information should be in the DT if there is
one.
Seems overboard to me or at least a separate problem.
It's a separate problem but this was preventing the series from
working on a few boards.
What is the failure? Not booting? Fixing not working would certainly not
be overboard.
On the device I was testing on (qemu's vexpress-a15 machine) the
machine booted and I was able to open a ssh session, but serial was
broken among other AMBA devices:
/memory-controller@2b0a0000
/memory-controller@7ffd0000
/dma@7ffb0000
/smb/motherboard/iofpga@3,00000000/sysctl@020000
/smb/motherboard/iofpga@3,00000000/aaci@040000
/smb/motherboard/iofpga@3,00000000/mmci@050000
/smb/motherboard/iofpga@3,00000000/kmi@060000
/smb/motherboard/iofpga@3,00000000/kmi@070000
/smb/motherboard/iofpga@3,00000000/uart@090000
/smb/motherboard/iofpga@3,00000000/uart@0a0000
/smb/motherboard/iofpga@3,00000000/uart@0b0000
/smb/motherboard/iofpga@3,00000000/uart@0c0000
/smb/motherboard/iofpga@3,00000000/wdt@0f0000
/smb/motherboard/iofpga@3,00000000/timer@110000
/smb/motherboard/iofpga@3,00000000/timer@120000
/smb/motherboard/iofpga@3,00000000/rtc@170000
/smb/motherboard/iofpga@3,00000000/clcd@1f0000
Another way of avoiding this particular problem would be not delaying
the probe of devices in the configuration bus, by doing something like
this:
diff --git a/drivers/bus/vexpress-config.c b/drivers/bus/vexpress-config.c
index 6575c0fe6a4e..eda293869cd3 100644
--- a/drivers/bus/vexpress-config.c
+++ b/drivers/bus/vexpress-config.c @@ -181,7 +181,7 @@ static int vexpress_config_populate(struct device_node *node)
if (WARN_ON(!parent))
return -ENODEV;
- return of_platform_populate(node, NULL, NULL, parent);
+ return of_platform_populate_early(node, NULL, NULL, parent);
}
static int __init vexpress_config_init(void)
But I think this would be papering over the underlying issue and it
would be better to have proper explicit dependencies.
Regards,
Tomeu
quoted quoted Most clocks have
to be setup before the driver model simply because timers depend on
clocks usually.
Yes, but in this case the apb clocks for the primecell devices are
implemented in a normal platform driver (vexpress_osc_driver), instead
of using CLK_OF_DECLARE.
Okay.
Rob
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/