Both irqnum and irqarray properties reflect the same thing:
the number of bits and bytes for interrupts at this
chipset. E. g.:
irqnum = 8 x irqarray
This can be seen by the way pending interrupts are handled:
/* During probe time */
pmic->irqs = devm_kzalloc(dev, pmic->irqnum * sizeof(int), GFP_KERNEL);
/* While handling IRQs */
for (i = 0; i < pmic->irqarray; i++) {
pending = hi6421_spmi_pmic_read(pmic, (i + pmic->irq_addr));
pending &= 0xff;
for_each_set_bit(offset, &pending, 8)
generic_handle_irq(pmic->irqs[offset + i * 8]);
}
Going further, there are some logic at the driver which assumes
that irqarray is 2:
/* solve powerkey order */
if ((i == HISI_IRQ_KEY_NUM) &&
((pending & HISI_IRQ_KEY_VALUE) == HISI_IRQ_KEY_VALUE)) {
generic_handle_irq(pmic->irqs[HISI_IRQ_KEY_DOWN]);
generic_handle_irq(pmic->irqs[HISI_IRQ_KEY_UP]);
pending &= (~HISI_IRQ_KEY_VALUE);
}
As HISI_IRQ_KEY_DOWN and HISI_IRQ_KEY_UP are fixed values
and don't depend on irqnum/irqarray.
The IRQ addr and mask addr seem to be also fixed, based on some
comments at the OF parsing code. So, get rid of them too,
removing the of parsing function completely.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/e231244e42cb5b56240705cac2f987e11a078038.1597762400.git.mchehab+huawei@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Both the SPMI controller and the SPMI PMIC driver
depends on the SPMI bus support.
The dependency for the regulator is also wrong:
it should depends on the SPMI version of the HiSilicon 6421,
and not on the normal one.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/1b7e23500e7449593393115cc0954af441b0c730.1597762400.git.mchehab+huawei@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently, an array is used to store both vsel and enable
settings, mixing registers, masks and bit settings.
Change it in order to have one separate property for each.
This makes easier to understand the contents of the DT
file, and to describe it at the Documentation/.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/fd72215b0a7da55dd727c2d7bd01c047a46cdd2e.1597647359.git.mchehab+huawei@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Using dev_dbg() is not too nice, as, instead of printing the
name of the regulator, it prints "regulator.<number>", making
harder to associate what is happening with each ldo line.
So, add a debug-specific macro, which will print the rdev's
name, just like the regulator core.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/1690b8531b23910aa915bd2725410b926022c481.1597647359.git.mchehab+huawei@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Instead of implementing a custom set of properties, set
valid_modes_mask based on having or not a mask for enabling
the eco_mode.
This makes the code clearer, and remove some uneeded props
from DT.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/0f4ac5a4167bbca428c8507b6992acd8b3ebabd1.1597647359.git.mchehab+huawei@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There are several fields on this struct that can be removed,
as they already exists at struct regulator_desc.
Remove them, cleaning up the code in the process.
While here, rename it to hi6421v600_regulator_info, in order
to better match the driver's name.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/648d48f11368a9869d760c2bd54bfbc3feb4f44c.1597647359.git.mchehab+huawei@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There are several OF properties that aren't used by Hikey 970,
and some are not even used inside the driver.
So, drop them, as as this makes easier to document what's
actually used.
If latter needed, those could be re-added later.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/332f96c178b81bf1e9908a1da2127f043909ae0c.1597647359.git.mchehab+huawei@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There are some checks there which could make sense for
downstream builds, but doesn't make much sense for
upstream ones. They came from the official Hikey970 tree
from Linaro, but even there, the commented-out code is not
set via other Kconfig vars.
So, let's just get rid of that. If needed later, this
patch can be (partially?) reversed.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/ecbef801f6c32ba0850ad9e5c534a4304807df3b.1597647359.git.mchehab+huawei@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Right now, driver is printing some messages as:
[ 33.833026] (NULL device *): spmi_read_cmd: id:0 addr:0x17, read value: 00
This is because dev_foo() are not using a device with a name
set. Change the logic for it to print it right.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/f647600d6e709f3dc3f92a64f1788f739c924330.1597647359.git.mchehab+huawei@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add a MODULE_DEVICE_TABLE() to the driver.
Also, the current logic calls platform_set_drvdata(pdev, NULL)
if the driver succeeds loading.
While here, remove the .owner, as it is not needed upstream
anymore.
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/122bd1c02f325094f4830ac79a86430fbe01e3d5.1597647359.git.mchehab+huawei@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently emxx_udc.h defines bit values using local macros. Use the
standard one instead.
Also, combine bit values with bitwise-or rather than addition, as
suggested by Coccinelle.
Signed-off-by: Alex Dewar <alex.dewar90@gmail.com>
Link: https://lore.kernel.org/r/20200818134922.409195-1-alex.dewar90@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There are various spelling mistakes in comments and error messages.
Fix these.
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Reviewed-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200806104701.46123-1-colin.king@canonical.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
According to the TODO 5 GHz code should be removed.
- find and remove remaining code valid only for 5 GHz. Most of the obvious
ones have been removed, but things like channel > 14 still exist.
Remove code for channels > 14 from rtw_get_center_ch().
Signed-off-by: Michael Straube <straube.linux@gmail.com>
Link: https://lore.kernel.org/r/20200813065053.13883-1-straube.linux@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
'rtl8192_irq_rx_tasklet()' is a tasklet initialized in
'rtl8192_init_priv_task()'.
>From this function it is possible to allocate some memory with the
GFP_KERNEL flag, which is not allowed in the atomic context of a tasklet.
Use GFP_ATOMIC instead.
The call chain is:
rtl8192_irq_rx_tasklet (in r8192U_core.c)
--> rtl8192_rx_nomal (in r8192U_core.c)
--> ieee80211_rx (in ieee80211/ieee80211_rx.c)
--> RxReorderIndicatePacket (in ieee80211/ieee80211_rx.c)
Fixes: 79a5ccd972 ("staging: rtl8192u: fix large frame size compiler warning")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/20200813173458.758284-1-christophe.jaillet@wanadoo.fr
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The type cast
padapter = (struct adapter *)rtw_netdev_priv(dev);
do nothing because type of rtw_netdev_priv() result
is (struct adapter *).
Signed-off-by: Ivan Safonov <insafonov@gmail.com>
Link: https://lore.kernel.org/r/20200815203306.121039-1-insafonov@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Mauro Carvalho Chehab
Mayulong
Alex Dewar
Colin Ian King
Michael Straube
Christophe JAILLET
Ivan Safonov