Age | Commit message (Collapse) | Author |
|
Calling dev_pm_domain_attach()/dev_pm_domain_detach() in bus driver
probe/remove functions can affect system behavior when the drivers
attached to the bus use devres-managed resources. Since devres actions
may need to access device registers, calling dev_pm_domain_detach() too
early, i.e., before these actions complete, can cause failures on some
systems. One such example is Renesas RZ/G3S SoC-based platforms.
If the device clocks are managed via PM domains, invoking
dev_pm_domain_detach() in the bus driver's remove function removes the
device's clocks from the PM domain, preventing any subsequent
pm_runtime_resume*() calls from enabling those clocks.
The second argument of dev_pm_domain_attach() specifies whether the PM
domain should be powered on during attachment. Likewise, the second
argument of dev_pm_domain_detach() indicates whether the domain should be
powered off during detachment.
Upcoming changes address the issue described above (initially for the
platform bus only) by deferring the call to dev_pm_domain_detach() until
after devres_release_all() in device_unbind_cleanup(). The detach_power_off
field in struct dev_pm_info stores the detach power off info from the
second argument of dev_pm_domain_attach().
Because there are cases where the device's PM domain power-on/off behavior
must be conditional (e.g., in i2c_device_probe()), the patch introduces
PD_FLAG_ATTACH_POWER_ON and PD_FLAG_DETACH_POWER_OFF flags to be passed
to dev_pm_domain_attach().
Finally, dev_pm_domain_attach() and its users are updated to use the newly
introduced PD_FLAG_ATTACH_POWER_ON and PD_FLAG_DETACH_POWER_OFF macros.
This change is preparatory.
Signed-off-by: Claudiu Beznea <claudiu.beznea.uj@bp.renesas.com>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Acked-by: Wolfram Sang <wsa+renesas@sang-engineering.com> # I2C
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Link: https://patch.msgid.link/20250703112708.1621607-2-claudiu.beznea.uj@bp.renesas.com
[ rjw: Changelog adjustments ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
|
|
In the current implementation, the return value of dev_pm_domain_attach()
is not checked. This can lead to silent failures if the function fails,
as the code would continue execution and return 0, ignoring the error.
This patch adds a check for the return value of dev_pm_domain_attach().
If the function fails, an error message is logged using dev_err_probe(),
and the error is propagated to the existing error handling path `err`,
which ensures proper cleanup by calling clk_notifier_unregister().
Signed-off-by: Wentao Liang <vulab@iscas.ac.cn>
Link: https://lore.kernel.org/r/20250211092017.562-1-vulab@iscas.ac.cn
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
|
|
After commit 0edb555a65d1 ("platform: Make platform_driver::remove()
return void") .remove() is (again) the right callback to implement for
platform drivers.
Convert all clk drivers to use .remove(), with the eventual goal to drop
struct platform_driver::remove_new(). As .remove() and .remove_new() have
the same prototypes, conversion is done by just changing the structure
member name in the driver initializer.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@baylibre.com>
Link: https://lore.kernel.org/r/20240909144026.870565-2-u.kleine-koenig@baylibre.com
Acked-by: Geert Uytterhoeven <geert+renesas@glider.be> # renesas
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
|
|
The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.
Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20230312161512.2715500-23-u.kleine-koenig@pengutronix.de
|
|
The dev_err_probe() function prints an error message if the error
code is not -EPROBE_DEFER. If we know the error code in is -ENODEV
then there is no reason to check.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Link: https://lore.kernel.org/r/YJotlJBJ1CVAgvMT@mwanda
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
|
|
dev_err_probe() returns the error code passed as second parameter. Also if
the error code is -EPROBE_DEFER dev_err_probe() is silent, so there is no
need to check for this value before calling dev_err_probe().
Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Link: https://lore.kernel.org/r/20210427164522.2886825-1-uwe@kleine-koenig.org
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
|
|
Add a driver for the SDX55 APCS clock controller. It is part of the APCS
hardware block, which among other things implements also a combined mux
and half integer divider functionality. The APCS clock controller has 3
parent clocks:
1. Board XO
2. Fixed rate GPLL0
3. A7 PLL
This is required for enabling CPU frequency scaling on SDX55-based
platforms.
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Link: https://lore.kernel.org/r/20210118041156.50016-6-manivannan.sadhasivam@linaro.org
[sboyd@kernel.org: Fix unused ret in probe by hardcoding it]
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
|