linux-stable.git/drivers/pinctrl, branch v5.4.277

pinctrl: core: handle radix_tree_insert() errors in pinctrl_register_one_pin()

2024-05-25T14:17:16+00:00

commit ecfe9a015d3e1e46504d5b3de7eef1f2d186194a upstream.

pinctrl_register_one_pin() doesn't check the result of radix_tree_insert()
despite they both may return a negative error code.  Linus Walleij said he
has copied the radix tree code from kernel/irq/ where the functions calling
radix_tree_insert() are *void* themselves; I think it makes more sense to
propagate the errors from radix_tree_insert() upstream if we can do that...

Found by Linux Verification Center (linuxtesting.org) with the Svace static
analysis tool.

Signed-off-by: Sergey Shtylyov 
Link: https://lore.kernel.org/r/20230719202253.13469-3-s.shtylyov@omp.ru
Signed-off-by: Linus Walleij 
Cc: "Hemdan, Hagar Gamal Halim" 
Signed-off-by: Greg Kroah-Hartman

pinctrl: mediatek: paris: Fix PIN_CONFIG_INPUT_SCHMITT_ENABLE readback

2024-05-17T09:43:56+00:00

commit 08f66a8edd08f6f7cfa769c81634b29a2b123908 upstream.

In the generic pin config library, readback of some options are handled
differently compared to the setting of those options: the argument value
is used to convey enable/disable of an option in the set path, but
success or -EINVAL is used to convey if an option is enabled or disabled
in the debugfs readback path.

PIN_CONFIG_INPUT_SCHMITT_ENABLE is one such option. Fix the readback of
the option in the mediatek-paris library, so that the debugfs dump is
not showing "input schmitt enabled" for pins that don't have it enabled.

Fixes: 1bea6afbc842 ("pinctrl: mediatek: Refine mtk_pinconf_get()")
Signed-off-by: Chen-Yu Tsai 
Reviewed-by: AngeloGioacchino Del Regno 
Message-ID: <20240327091336.3434141-2-wenst@chromium.org>
Signed-off-by: Linus Walleij 
Signed-off-by: Greg Kroah-Hartman

pinctrl: mediatek: remove set but not used variable 'e'

2024-05-17T09:43:56+00:00

commit 86ecb7d6853c77711c14cb6600179196f179ee2d upstream.

drivers/pinctrl/mediatek/pinctrl-mtk-common-v2.c: In function mtk_hw_pin_field_lookup:
drivers/pinctrl/mediatek/pinctrl-mtk-common-v2.c:70:39: warning:
 variable e set but not used [-Wunused-but-set-variable]

Since commit 3de7deefce69 ("pinctrl: mediatek: Check gpio pin
number and use binary search in mtk_hw_pin_field_lookup()"),
it is not used any more, so remove it, also remove redundant
assignment to variable c, it will be assigned a new value later
before used.

Reported-by: Hulk Robot 
Signed-off-by: YueHaibing 
Reviewed-by: Matthias Brugger 
Link: https://lore.kernel.org/r/20200218023625.14324-1-yuehaibing@huawei.com
Signed-off-by: Linus Walleij 
Signed-off-by: Greg Kroah-Hartman

pinctrl: mediatek: Fix some off by one bugs

2024-05-17T09:43:55+00:00

commit 3385ab72d995fc0b876818a36203bf2429445686 upstream.

These comparisons should be >= instead of > to prevent accessing one
element beyond the end of the hw->soc->pins[] array.

Fixes: 3de7deefce69 ("pinctrl: mediatek: Check gpio pin number and use binary search in mtk_hw_pin_field_lookup()")
Fixes: 184d8e13f9b1 ("pinctrl: mediatek: Add support for pin configuration dump via debugfs.")
Signed-off-by: Dan Carpenter 
Link: https://lore.kernel.org/r/20200218055247.74s2xa7veqx2do34@kili.mountain
Reviewed-by: Matthias Brugger 
Signed-off-by: Linus Walleij 
Signed-off-by: Greg Kroah-Hartman

pinctrl: mediatek: Fix fallback behavior for bias_set_combo

2024-05-17T09:43:55+00:00

commit 798a315fc359aa6dbe48e09d802aa59b7e158ffc upstream.

Some pin doesn't support PUPD register, if it fails and fallbacks with
bias_set_combo case, it will call mtk_pinconf_bias_set_pupd_r1_r0() to
modify the PUPD pin again.

Since the general bias set are either PU/PD or PULLSEL/PULLEN, try
bias_set or bias_set_rev1 for the other fallback case. If the pin
doesn't support neither PU/PD nor PULLSEL/PULLEN, it will return
-ENOTSUPP.

Fixes: 81bd1579b43e ("pinctrl: mediatek: Fix fallback call path")
Signed-off-by: Hsin-Yi Wang 
Reviewed-by: Chen-Yu Tsai 
Reviewed-by: Zhiyong Tao 
Link: https://lore.kernel.org/r/20210701080955.2660294-1-hsinyi@chromium.org
Signed-off-by: Linus Walleij 
Signed-off-by: Greg Kroah-Hartman

pinctrl: mediatek: Fix fallback call path

2024-05-17T09:43:53+00:00

[ Upstream commit 81bd1579b43e0e285cba667399f1b063f1ce7672 ]

Some SoCs, eg. mt8183, are using a pinconfig operation bias_set_combo.
The fallback path in mtk_pinconf_adv_pull_set() should also try this
operation.

Fixes: cafe19db7751 ("pinctrl: mediatek: Backward compatible to previous Mediatek's bias-pull usage")
Signed-off-by: Hsin-Yi Wang 
Acked-by: Sean Wang 
Link: https://lore.kernel.org/r/20201228090425.2130569-1-hsinyi@chromium.org
Signed-off-by: Linus Walleij 
Signed-off-by: Sasha Levin

pinctrl: devicetree: fix refcount leak in pinctrl_dt_to_map()

2024-05-17T09:43:48+00:00

[ Upstream commit a0cedbcc8852d6c77b00634b81e41f17f29d9404 ]

If we fail to allocate propname buffer, we need to drop the reference
count we just took. Because the pinctrl_dt_free_maps() includes the
droping operation, here we call it directly.

Fixes: 91d5c5060ee2 ("pinctrl: devicetree: fix null pointer dereferencing in pinctrl_dt_to_map")
Suggested-by: Dan Carpenter 
Signed-off-by: Zeng Heng 
Reviewed-by: Dan Carpenter 
Message-ID: <20240415105328.3651441-1-zengheng4@huawei.com>
Signed-off-by: Linus Walleij 
Signed-off-by: Sasha Levin

pinctrl: mediatek: paris: Rework support for PIN_CONFIG_{INPUT,OUTPUT}_ENABLE

2024-05-17T09:43:48+00:00

[ Upstream commit c5d3b64c568a344e998830e0e94a7c04e372f89b ]

There is a misinterpretation of some of the PIN_CONFIG_* options in this
driver library. PIN_CONFIG_OUTPUT_ENABLE should refer to a buffer or
switch in the output direction of the electrical path. The MediaTek
hardware does not have such a thing. The driver incorrectly maps this
option to the GPIO function's direction.

Likewise, PIN_CONFIG_INPUT_ENABLE should refer to a buffer or switch in
the input direction. The hardware does have such a mechanism, and is
mapped to the IES bit. The driver however sets the direction in addition
to the IES bit, which is incorrect. On readback, the IES bit isn't even
considered.

Ironically, the driver does not support readback for PIN_CONFIG_OUTPUT,
while its readback of PIN_CONFIG_{INPUT,OUTPUT}_ENABLE is what it should
be doing for PIN_CONFIG_OUTPUT.

Rework support for these three options, so that PIN_CONFIG_OUTPUT_ENABLE
is completely removed, PIN_CONFIG_INPUT_ENABLE is only linked to the IES
bit, and PIN_CONFIG_OUTPUT is linked to the GPIO function's direction
and output level.

Fixes: 805250982bb5 ("pinctrl: mediatek: add pinctrl-paris that implements the vendor dt-bindings")
Signed-off-by: Chen-Yu Tsai 
Reviewed-by: AngeloGioacchino Del Regno 
Message-ID: <20240327091336.3434141-3-wenst@chromium.org>
Signed-off-by: Linus Walleij 
Signed-off-by: Sasha Levin

pinctrl: mediatek: paris: Rework mtk_pinconf_{get,set} switch/case logic

2024-05-17T09:43:47+00:00

[ Upstream commit 9b780fa1ff14663c2e0f07ad098b96b8337f27a4 ]

The current code deals with optional features by testing for the
function pointers and returning -ENOTSUPP if it is not valid. This is
done for multiple pin config settings and results in the code that
handles the supporting cases to get indented by one level. This is
aggrevated by the fact that some features require another level of
conditionals.

Instead of assigning the same error code in all unsupported optional
feature cases, simply have that error code as the default, and break
out of the switch/case block whenever a feature is unsupported, or an
error is returned. This reduces indentation by one level for the useful
code.

Also replace the goto statements with break statements. The result is
the same, as the gotos simply exit the switch/case block, which can
also be achieved with a break statement. With the latter the intent
is clear and easier to understand.

Signed-off-by: Chen-Yu Tsai 
Reviewed-by: AngeloGioacchino Del Regno 
Link: https://lore.kernel.org/r/20220308100956.2750295-8-wenst@chromium.org
Signed-off-by: Linus Walleij 
Stable-dep-of: c5d3b64c568a ("pinctrl: mediatek: paris: Rework support for PIN_CONFIG_{INPUT,OUTPUT}_ENABLE")
Signed-off-by: Sasha Levin

pinctrl: mediatek: paris: Fix PIN_CONFIG_BIAS_* readback

2024-05-17T09:43:47+00:00

[ Upstream commit 3e8c6bc608480010f360c4a59578d7841726137d ]

When reading back pin bias settings, if the pin is not in the
corresponding bias state, the function should return -EINVAL.

Fix this in the mediatek-paris pinctrl library so that the read back
state is not littered with bogus a "input bias disabled" combined with
"pull up" or "pull down" states.

Fixes: 805250982bb5 ("pinctrl: mediatek: add pinctrl-paris that implements the vendor dt-bindings")
Signed-off-by: Chen-Yu Tsai 
Reviewed-by: AngeloGioacchino Del Regno 
Link: https://lore.kernel.org/r/20220308100956.2750295-3-wenst@chromium.org
Signed-off-by: Linus Walleij 
Stable-dep-of: c5d3b64c568a ("pinctrl: mediatek: paris: Rework support for PIN_CONFIG_{INPUT,OUTPUT}_ENABLE")
Signed-off-by: Sasha Levin