dm: core: Add DM_FLAG_PROBE_AFTER_BIND flag
PATCH 1/3 dm: core: Add DM_FLAG_PROBE_AFTER_BIND flag
Introduce DM_FLAG_PROBE_AFTER_BIND flag, which can be set by driver or uclass in .bind(), to indicate such driver instance should be probe()d once binding of all devices is complete.
This is useful in case the driver determines that hardware initialization is mandatory on boot, and such initialization happens only in probe(). This also solves the inability to call device_probe() from .bind().
--- drivers/core/root.c | 24 +++++++++++++++++++++++- include/dm/device.h | 3 +++ 2 files changed, 26 insertions(+), 1 deletion(-)
drivers/core/root.c +23 -1
@@ -361,6 +361,28 @@ void *dm_priv_to_rw(void *priv) }#endif +static int dm_probe_devices (struct udevice *dev,bool pre_reloc_only) +{ +u32 mask =DM_FLAG_PROBE_AFTER_BIND ; +u32 flags =dev_get_flags (dev); +struct udevice *child; +int ret; + +if (pre_reloc_only) + mask |=DM_FLAG_PRE_RELOC ; + +if ((flags & mask) == mask) { + ret =device_probe (dev); +if (ret) +return ret; + } + +list_for_each_entry (child, &dev->child_head, sibling_node) +dm_probe_devices (child, pre_reloc_only); + +return 0 ; +} +/** * dm_scan() - Scan tables to bind devices *
@@ -393,7 +415,7 @@ static int dm_scan(bool pre_reloc_only)if (ret)return ret; -return 0 ; +return dm_probe_devices (gd->dm_root, pre_reloc_only); }int dm_init_and_scan (bool pre_reloc_only)
include/dm/device.h +3 -0
@@ -80,6 +80,9 @@ struct driver_info;*/ #define DM_FLAG_VITAL (1 << 14) +/* Device must be probed after it was bound */ +#define DM_FLAG_PROBE_AFTER_BIND (1 << 15) +/* * One or multiple of these flags are passed to device_remove() so that * a selective device removal as specified by the remove-stage and the
Signature · 1 line
2.35.1
PATCH 2/3 led: Mark device instance with DM_FLAG_PROBE_AFTER_BIND
Calling device_probe() from uclass .post_bind() callback has all kinds of odd side-effects, e.g. device instances not being available just yet. Make use of the DM_FLAG_PROBE_AFTER_BIND instead, mark device instances which need to be probe()d in order to configure the LED default state with this flag and let the DM core do the device_probe() at the right time instead.
--- drivers/led/led-uclass.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/led/led-uclass.c +3 -1
@@ -98,7 +98,9 @@ static int led_post_bind(struct udevice *dev)* In case the LED has default-state DT property, trigger * probe() to configure its default state during startup. */ -return device_probe (dev); +dev_or_flags (dev,DM_FLAG_PROBE_AFTER_BIND ); + +return 0 ; }static int led_post_probe (struct udevice *dev)
Signature · 1 line
2.35.1
Re: [PATCH 2/3] led: Mark device instance with DM_FLAG_PROBE_AFTER_BIND
Hi Marek
Quoted 23 lines · Marek Vasut
Calling
device_probe()from uclass .post_bind() callback has all kinds of odd side-effects, e.g. device instances not being available just yet. Make use of theDM_FLAG_PROBE_AFTER_BINDinstead, mark device instances which need to be probe()d in order to configure the LED default state with this flag and let the DM core do thedevice_probe()at the right time instead.Fixes 72675b063b6 ("led: Configure LED default-state on boot")Signed-off-by Marek Vasut <[email protected]>Cc Patrice Chotard <[email protected]>Cc Patrick Delaunay <[email protected]>Cc Sean Anderson <[email protected]>Cc Simon Glass <[email protected]>Cc Steven Lawrance <[email protected]>--- drivers/led/led-uclass.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)drivers/led/led-uclass.c +0
-0
drivers/led/led-uclass.c static int led_post_bind(struct udevice *dev) · line 106
Quoted 10 lines
drivers/led/led-uclass.c +3
-1@@ -98,7 +98,9 @@ static int led_post_bind(struct udevice *dev)* In case the LED has default-state DT property, trigger * probe() to configure its default state during startup. */ -return device_probe (dev); +dev_or_flags (dev,DM_FLAG_PROBE_AFTER_BIND ); + +return 0 ; }static int led_post_probe (struct udevice *dev)
Tested on stm32mp157c-dk2 board
Thanks
Patrice
Re: [PATCH 2/3] led: Mark device instance with DM_FLAG_PROBE_AFTER_BIND
In reply to Marek VasutQuoted 16 lines · Marek Vasut
Calling
device_probe()from uclass .post_bind() callback has all kinds of odd side-effects, e.g. device instances not being available just yet. Make use of theDM_FLAG_PROBE_AFTER_BINDinstead, mark device instances which need to be probe()d in order to configure the LED default state with this flag and let the DM core do thedevice_probe()at the right time instead.Fixes 72675b063b6 ("led: Configure LED default-state on boot")Signed-off-by Marek Vasut <[email protected]>Cc Patrice Chotard <[email protected]>Cc Patrick Delaunay <[email protected]>Cc Sean Anderson <[email protected]>Cc Simon Glass <[email protected]>Cc Steven Lawrance <[email protected]>Reviewed-by Patrice Chotard <[email protected]>Tested-by Patrice Chotard <[email protected]>
Quoted 3 lines
--- drivers/led/led-uclass.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
This breaks all of the sandbox tests, which perhaps need another update for what you've changed here? I already had to tweak them once in 72675b063b6e ("led: Configure LED default-state on boot"). But my perhaps incorrect read then was that the dts / test weren't quite right to start with. Perhaps that's still the case however?
Signature · 1 line
Tom
Re: [PATCH 2/3] led: Mark device instance with DM_FLAG_PROBE_AFTER_BIND
Quoted 22 lines · Tom Rini
Quoted 16 lines · Marek Vasut
Calling
device_probe()from uclass .post_bind() callback has all kinds of odd side-effects, e.g. device instances not being available just yet. Make use of theDM_FLAG_PROBE_AFTER_BINDinstead, mark device instances which need to be probe()d in order to configure the LED default state with this flag and let the DM core do thedevice_probe()at the right time instead.Fixes 72675b063b6 ("led: Configure LED default-state on boot")Signed-off-by Marek Vasut <[email protected]>Cc Patrice Chotard <[email protected]>Cc Patrick Delaunay <[email protected]>Cc Sean Anderson <[email protected]>Cc Simon Glass <[email protected]>Cc Steven Lawrance <[email protected]>Reviewed-by Patrice Chotard <[email protected]>Tested-by Patrice Chotard <[email protected]>Quoted 3 lines
--- drivers/led/led-uclass.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
Quoted 5 lines
This breaks all of the sandbox tests, which perhaps need another update for what you've changed here? I already had to tweak them once in
72675b063b6e("led: Configure LED default-state on boot"). But my perhaps incorrect read then was that the dts / test weren't quite right to start with. Perhaps that's still the case however?
Pick these two test fixes:
[PATCH 1/2] test: dm: led: Fix LED enumeration
[PATCH 2/2] test: dm: pinmux: Get LED2 udevice in the pinmux test
Re: [PATCH 2/3] led: Mark device instance with DM_FLAG_PROBE_AFTER_BIND
In reply to Marek VasutQuoted 7 lines · Marek Vasut
Calling
device_probe()from uclass .post_bind() callback has all kinds of odd side-effects, e.g. device instances not being available just yet. Make use of theDM_FLAG_PROBE_AFTER_BINDinstead, mark device instances which need to be probe()d in order to configure the LED default state with this flag and let the DM core do thedevice_probe()at the right time instead.
Quoted 9 lines
Fixes 72675b063b6 ("led: Configure LED default-state on boot")Signed-off-by Marek Vasut <[email protected]>Cc Patrice Chotard <[email protected]>Cc Patrick Delaunay <[email protected]>Cc Sean Anderson <[email protected]>Cc Simon Glass <[email protected]>Cc Steven Lawrance <[email protected]>Reviewed-by Patrice Chotard <[email protected]>Tested-by Patrice Chotard <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
PATCH 3/3 led: gpio: Check device compatible string to determine the top level node
In reply to Marek VasutSince 2d1deaf88ed ("led: gpio: Drop duplicate OF "label" property parsing"), all LED nodes have some sort of label. Use device_is_compatible(..."leds-gpio") to determine whether this is a top-level node, since it is only the top level node which is compatible with "leds-gpio", the GPIO LEDs subnodes are not.
--- drivers/led/led_gpio.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/led/led_gpio.c +1 -2
@@ -57,12 +57,11 @@ static enum led_state_t gpio_led_get_state(struct udevice *dev)static int led_gpio_probe (struct udevice *dev) { -struct led_uc_plat *uc_plat =dev_get_uclass_plat (dev);struct led_gpio_priv *priv =dev_get_priv (dev);int ret;/* Ignore the top-level LED node */ -if (!uc_plat->label) +if (device_is_compatible (dev,"gpio-leds" ))return 0 ; ret =gpio_request_by_name (dev,"gpios" ,0 , &priv->gpio,GPIOD_IS_OUT );
Signature · 1 line
2.35.1
Re: [PATCH 3/3] led: gpio: Check device compatible string to determine the top level node
Hi Marek
Quoted 25 lines · Marek Vasut
Since
2d1deaf88ed("led: gpio: Drop duplicate OF "label" property parsing"), all LED nodes have some sort of label. Usedevice_is_compatible(..."leds-gpio")to determine whether this is a top-level node, since it is only the top level node which is compatible with "leds-gpio", the GPIO LEDs subnodes are not.Fixes 2d1deaf88ed ("led: gpio: Drop duplicate OF "label" property parsing")Signed-off-by Marek Vasut <[email protected]>Cc Patrice Chotard <[email protected]>Cc Patrick Delaunay <[email protected]>Cc Sean Anderson <[email protected]>Cc Simon Glass <[email protected]>Cc Steven Lawrance <[email protected]>--- drivers/led/led_gpio.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)drivers/led/led_gpio.c +0
-0@@ -57,12 +57,11 @@ static enum led_state_t gpio_led_get_state(struct udevice *dev)static int led_gpio_probe (struct udevice *dev) {
drivers/led/led_gpio.c static enum led_state_t gpio_led_get_state(struct udevice *dev) · line 67
Quoted 10 lines
drivers/led/led_gpio.c +1
-2-struct led_uc_plat *uc_plat =dev_get_uclass_plat (dev);struct led_gpio_priv *priv =dev_get_priv (dev);int ret;/* Ignore the top-level LED node */ -if (!uc_plat->label) +if (device_is_compatible (dev,"gpio-leds" ))return 0 ; ret =gpio_request_by_name (dev,"gpios" ,0 , &priv->gpio,GPIOD_IS_OUT );
Tested on stm32mp157c-dk2 board
Thanks
Patrice
Re: [PATCH 3/3] led: gpio: Check device compatible string to determine the top level node
In reply to Marek VasutQuoted 6 lines · Marek Vasut
Since
2d1deaf88ed("led: gpio: Drop duplicate OF "label" property parsing"), all LED nodes have some sort of label. Usedevice_is_compatible(..."leds-gpio")to determine whether this is a top-level node, since it is only the top level node which is compatible with "leds-gpio", the GPIO LEDs subnodes are not.
Quoted 9 lines
Fixes 2d1deaf88ed ("led: gpio: Drop duplicate OF "label" property parsing")Signed-off-by Marek Vasut <[email protected]>Cc Patrice Chotard <[email protected]>Cc Patrick Delaunay <[email protected]>Cc Sean Anderson <[email protected]>Cc Simon Glass <[email protected]>Cc Steven Lawrance <[email protected]>Reviewed-by Patrice Chotard <[email protected]>Tested-by Patrice Chotard <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom
Hi Marek
Quoted 66 lines · Marek Vasut
Introduce
DM_FLAG_PROBE_AFTER_BINDflag, which can be set by driver or uclass in .bind(), to indicate such driver instance should be probe()d once binding of all devices is complete.This is useful in case the driver determines that hardware initialization is mandatory on boot, and such initialization happens only in
probe(). This also solves the inability to calldevice_probe()from .bind().Signed-off-by Marek Vasut <[email protected]>Cc Patrice Chotard <[email protected]>Cc Patrick Delaunay <[email protected]>Cc Sean Anderson <[email protected]>Cc Simon Glass <[email protected]>Cc Steven Lawrance <[email protected]>--- drivers/core/root.c | 24 +++++++++++++++++++++++- include/dm/device.h | 3 +++ 2 files changed, 26 insertions(+), 1 deletion(-)drivers/core/root.c +23
-1@@ -361,6 +361,28 @@ void *dm_priv_to_rw(void *priv) }#endif +static int dm_probe_devices (struct udevice *dev,bool pre_reloc_only) +{ +u32 mask =DM_FLAG_PROBE_AFTER_BIND ; +u32 flags =dev_get_flags (dev); +struct udevice *child; +int ret; + +if (pre_reloc_only) + mask |=DM_FLAG_PRE_RELOC ; + +if ((flags & mask) == mask) { + ret =device_probe (dev); +if (ret) +return ret; + } + +list_for_each_entry (child, &dev->child_head, sibling_node) +dm_probe_devices (child, pre_reloc_only); + +return 0 ; +} +/** * dm_scan() - Scan tables to bind devices * @@ -393,7 +415,7 @@ static int dm_scan(bool pre_reloc_only)if (ret)return ret; -return 0 ; +return dm_probe_devices (gd->dm_root, pre_reloc_only); }int dm_init_and_scan (bool pre_reloc_only)include/dm/device.h +0
-0
include/dm/device.h struct driver_info; · line 88
Quoted 9 lines
include/dm/device.h +3
-0@@ -80,6 +80,9 @@ struct driver_info;*/ #define DM_FLAG_VITAL (1 << 14) +/* Device must be probed after it was bound */ +#define DM_FLAG_PROBE_AFTER_BIND (1 << 15) +/* * One or multiple of these flags are passed to device_remove() so that * a selective device removal as specified by the remove-stage and the
Tested on stm32mp157c-dk2 board
Thanks
Patrice
Quoted 7 lines · Marek Vasut
Introduce
DM_FLAG_PROBE_AFTER_BINDflag, which can be set by driver or uclass in .bind(), to indicate such driver instance should be probe()d once binding of all devices is complete.This is useful in case the driver determines that hardware initialization is mandatory on boot, and such initialization happens only in
probe(). This also solves the inability to calldevice_probe()from .bind().
So why not add an init_leds function to init_sequence_r?
--Sean
Quoted 10 lines · Sean Anderson
Quoted 7 lines · Marek Vasut
Introduce
DM_FLAG_PROBE_AFTER_BINDflag, which can be set by driver or uclass in .bind(), to indicate such driver instance should be probe()d once binding of all devices is complete.This is useful in case the driver determines that hardware initialization is mandatory on boot, and such initialization happens only in
probe(). This also solves the inability to calldevice_probe()from .bind().So why not add an
init_ledsfunction toinit_sequence_r?
Because this flag is generic solution and not hack specific to LEDs.
Quoted 6 lines · Marek Vasut
Quoted 5 lines · Sean Anderson
Quoted 4 lines · Marek Vasut
Introduce
DM_FLAG_PROBE_AFTER_BINDflag, which can be set by driver or uclass in .bind(), to indicate such driver instance should be probe()d once binding of all devices is complete.
Quoted 7 lines
Quoted 5 lines
Quoted 3 lines
This is useful in case the driver determines that hardware initialization is mandatory on boot, and such initialization happens only in
probe(). This also solves the inability to calldevice_probe()from .bind().So why not add an
init_ledsfunction toinit_sequence_r?Because this flag is generic solution and not hack specific to LEDs.
Isn't init_sequence_r the generic solution? Every device gets probed that way (or directly as a result of a command).
--Sean
Quoted 11 lines · Sean Anderson
Quoted 10 lines · Marek Vasut
Quoted 9 lines · Sean Anderson
Quoted 7 lines · Marek Vasut
Introduce
DM_FLAG_PROBE_AFTER_BINDflag, which can be set by driver or uclass in .bind(), to indicate such driver instance should be probe()d once binding of all devices is complete.This is useful in case the driver determines that hardware initialization is mandatory on boot, and such initialization happens only in
probe(). This also solves the inability to calldevice_probe()from .bind().
Quoted 5 lines
Quoted 3 lines
Quoted 1 line
So why not add an
init_ledsfunction toinit_sequence_r?Because this flag is generic solution and not hack specific to LEDs.
Isn't
init_sequence_rthe generic solution?
No, init_sequence_r is just a list of functions that gets called early on during boot to bring U-Boot up. One of the functions is initr_dm(), DM init, which brings up the driver model. The init_sequence_r has nothing to do with driver binding or probing though, that's the DM job.
Quoted 1 line
Every device gets probed that way (or directly as a result of a command).
No device probe gets triggered via init_sequence_r. There are many ad-hoc device_probe() calls in the various init_sequence_r function implementation, but that's all there because there is no better way to implement that within the DM framework so far. Now there is, with this flag, so those workarounds can also be cleaned up.
Quoted 16 lines · Marek Vasut
Quoted 15 lines · Sean Anderson
Quoted 13 lines · Marek Vasut
Quoted 10 lines · Sean Anderson
Quoted 7 lines · Marek Vasut
Introduce
DM_FLAG_PROBE_AFTER_BINDflag, which can be set by driver or uclass in .bind(), to indicate such driver instance should be probe()d once binding of all devices is complete.This is useful in case the driver determines that hardware initialization is mandatory on boot, and such initialization happens only in
probe(). This also solves the inability to calldevice_probe()from .bind().So why not add an
init_ledsfunction toinit_sequence_r?Because this flag is generic solution and not hack specific to LEDs.
Quoted 8 lines
Quoted 1 line
Isn't
init_sequence_rthe generic solution?No,
init_sequence_ris just a list of functions that gets called early on during boot to bring U-Boot up. One of the functions isinitr_dm(), DM init, which brings up the driver model. Theinit_sequence_rhas nothing to do with driver binding or probing though, that's the DM job.Quoted 1 line
Every device gets probed that way (or directly as a result of a command).
No device probe gets triggered via
init_sequence_r.
There are many ad-hocdevice_probe()calls in the variousinit_sequence_rfunction implementation,
Every device in U-Boot is probed in one of three ways
- As a dependency of another device (clocks, pinmuxes, gpios, etc.)
- Directly on the command line (most boot devices)
- From an init_sequence (timers, watchdogs, serial, PCI, gpio hogs, etc.)
The goal of course being to minimize the amount of probed devices. This of course requires having some things at the "root" of the tree. In Linux, everything bound is probed until nothing more can be probed. In that sense, under Linux everything is the "root" of the tree. But for U-Boot the second two options specify the "roots."
Some of the things in init_sequence are ad-hoc, especially the non-DM stuff. LEDs are pretty ad-hoc, last I checked, but that's mainly because there are 2-3 separate non-DM ways to control LEDs.
Quoted 1 line
but that's all there because there is no better way to implement that within the DM framework so far. Now there is, with this flag, so those workarounds can also be cleaned up.
This *is* within the DM framework. Adding a flag which is set on bound devices has the same effect as dm_mux_init does when called from initr_dm_devices. The only difference is implementation. So why is this better?
--Sean
Quoted 23 lines · Sean Anderson
Quoted 22 lines · Marek Vasut
Quoted 11 lines · Sean Anderson
Quoted 10 lines · Marek Vasut
Quoted 9 lines · Sean Anderson
Quoted 7 lines · Marek Vasut
Introduce
DM_FLAG_PROBE_AFTER_BINDflag, which can be set by driver or uclass in .bind(), to indicate such driver instance should be probe()d once binding of all devices is complete.This is useful in case the driver determines that hardware initialization is mandatory on boot, and such initialization happens only in
probe(). This also solves the inability to calldevice_probe()from .bind().Quoted 5 lines
Quoted 3 lines
Quoted 1 line
So why not add an
init_ledsfunction toinit_sequence_r?Because this flag is generic solution and not hack specific to LEDs.
Isn't
init_sequence_rthe generic solution?No,
init_sequence_ris just a list of functions that gets called early on during boot to bring U-Boot up. One of the functions isinitr_dm(), DM init, which brings up the driver model. Theinit_sequence_rhas nothing to do with driver binding or probing though, that's the DM job.Quoted 1 line
Every device gets probed that way (or directly as a result of a command).
Quoted 8 lines
Quoted 2 lines
No device probe gets triggered via
init_sequence_r.
There are many ad-hocdevice_probe()calls in the variousinit_sequence_rfunction implementation,Every device in U-Boot is probed in one of three ways
- As a dependency of another device (clocks, pinmuxes, gpios, etc.)
- Directly on the command line (most boot devices)
- From aninit_sequence(timers, watchdogs, serial, PCI, gpio hogs, etc.)
In short, the DM does lazy initialization of devices to boot quickly and waste few resources.
Quoted 1 line
The goal of course being to minimize the amount of probed devices.
This was never the goal. The probe() is called when needed, that's all.
Quoted 7 lines
This of course requires having
some things at the "root" of the tree. In Linux, everything bound is probed until nothing more can be probed. In that sense, under Linux everything is the "root" of the tree. But for U-Boot the second two options specify the "roots."Some of the things in
init_sequenceare ad-hoc, especially the non-DM stuff. LEDs are pretty ad-hoc, last I checked, but that's mainly because there are 2-3 separate non-DM ways to control LEDs.
I'm not sure I understand this part, but maybe see
doc/develop/driver-model/design.rst
that should clarify the design decisions behind the DM.
Quoted 3 lines
Quoted 1 line
but that's all there because there is no better way to implement that within the DM framework so far. Now there is, with this flag, so those workarounds can also be cleaned up.
This *is* within the DM framework.
No, init_sequence_r/init_sequence_f is U-Boot init sequence. The DM init is only one step in that init sequence. Please do not conflate those two concepts.
Quoted 2 lines
Adding a flag which is set on bound devices has the same effect as
dm_mux_initdoes when called frominitr_dm_devices.
This cannot scale, you would have to add a huge list of such functions to init_sequence_r/init_sequence_f probably for each uclass, which would each do some sort of ad-hoc device_probe(), outside of DM. And neither of those functions would have easy access to the context of each driver instance, so they would have to work around that somehow (that's usually done by wasting cycles on traversing DT again or other such heuristics).
Quoted 2 lines
The only difference is implementation. So
why is this better?
See above. Also, this implementation is within the context of DM, i.e. driver/uclass bind call sets this flag for driver instances which for some reason need to be probe()d. The reason can be inferred from the driver instance context, e.g. because the DT node matching the driver instance contains some DT property, this flag is set, and the DM core probes the driver instead of only binding it. No need for any ad-hoc workarounds in init code.
Quoted 8 lines · Marek Vasut
Introduce
DM_FLAG_PROBE_AFTER_BINDflag, which can be set by driver or uclass in .bind(), to indicate such driver instance should be probe()d once binding of all devices is complete.This is useful in case the driver determines that hardware initialization is mandatory on boot, and such initialization happens only in
probe(). This also solves the inability to calldevice_probe()from .bind().
Quoted 8 lines
Signed-off-by Marek Vasut <[email protected]>Cc Patrice Chotard <[email protected]>Cc Patrick Delaunay <[email protected]>Cc Sean Anderson <[email protected]>Cc Simon Glass <[email protected]>Cc Steven Lawrance <[email protected]>Reviewed-by Patrice Chotard <[email protected]>Tested-by Patrice Chotard <[email protected]>
Applied to u-boot/master, thanks!
Signature · 1 line
Tom