The BIT() macro uses unsigned long, which is 32 bits on 32-bit
architectures. When iterating over GPIO pins with index >= 32,
the expression (*value & BIT(i)) causes undefined behavior due
to shifting by a value >= type width.

Since 'value' is a pointer to u64, use BIT_ULL() to ensure correct
64-bit mask on all architectures.

Found by Linux Verification Center (linuxtesting.org) with Svace.

Fixes: 2c4d00cb8fc5 ("gpiolib: acpi: Use BIT() macro to increase readability")
Signed-off-by: Denis Sergeev <denserg.edu@gmail.com>

---
The ACPI specification does not define a strict upper bound for the
number of GPIO pins in the Pin Table. The value is derived from
16-bit offsets inside the resource descriptor, which theoretically
allows far more than 64 pins.

However, the current Linux GPIO ACPI OpRegion handler represents the
pin state as a single u64 value, which inherently limits the number of
addressable pins to 64. Thus, even though the specification permits
larger tables, the existing implementation already assumes a <= 64 pin
mask.

This patch fixes undefined behavior in the valid range [32, 63] on
32-bit architectures. Extending support beyond 64 pins would require
a different representation (e.g. bitmap) and is outside the scope of
this fix.

 drivers/gpio/gpiolib-acpi-core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpio/gpiolib-acpi-core.c b/drivers/gpio/gpiolib-acpi-core.c
index 83dd227dbbec..d42f769eeb11 100644
--- a/drivers/gpio/gpiolib-acpi-core.c
+++ b/drivers/gpio/gpiolib-acpi-core.c
@@ -1159,7 +1159,7 @@ acpi_gpio_adr_space_handler(u32 function, acpi_physical_address address,
 		mutex_unlock(&achip->conn_lock);
 
 		if (function == ACPI_WRITE)
-			gpiod_set_raw_value_cansleep(desc, !!(*value & BIT(i)));
+			gpiod_set_raw_value_cansleep(desc, !!(*value & BIT_ULL(i)));
 		else
 			*value |= (u64)gpiod_get_raw_value_cansleep(desc) << i;
 	}
-- 
2.50.1

On Mon, 26 Jan 2026 06:59:14 +0300, Denis Sergeev wrote:
> The BIT() macro uses unsigned long, which is 32 bits on 32-bit
> architectures. When iterating over GPIO pins with index >= 32,
> the expression (*value & BIT(i)) causes undefined behavior due
> to shifting by a value >= type width.
> 
> Since 'value' is a pointer to u64, use BIT_ULL() to ensure correct
> 64-bit mask on all architectures.
> 
> [...]

Applied, thanks!

[1/1] gpiolib: acpi: use BIT_ULL() for u64 mask in address space handler
      commit: 59084c564c412b1d537f90bd70fa1d6bfb584e82

Best regards,
-- 
Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>

On Mon, Jan 26, 2026 at 06:59:14AM +0300, Denis Sergeev wrote:
> The BIT() macro uses unsigned long, which is 32 bits on 32-bit
> architectures. When iterating over GPIO pins with index >= 32,
> the expression (*value & BIT(i)) causes undefined behavior due
> to shifting by a value >= type width.
> 
> Since 'value' is a pointer to u64, use BIT_ULL() to ensure correct
> 64-bit mask on all architectures.
> 
> Found by Linux Verification Center (linuxtesting.org) with Svace.
> 
> Fixes: 2c4d00cb8fc5 ("gpiolib: acpi: Use BIT() macro to increase readability")
> Signed-off-by: Denis Sergeev <denserg.edu@gmail.com>

Reviewed-by: Mika Westerberg <westeri@kernel.org>

On Tue, Jan 27, 2026 at 9:08 AM Mika Westerberg
<mika.westerberg@linux.intel.com> wrote:
>
> On Mon, Jan 26, 2026 at 06:59:14AM +0300, Denis Sergeev wrote:
> > The BIT() macro uses unsigned long, which is 32 bits on 32-bit
> > architectures. When iterating over GPIO pins with index >= 32,
> > the expression (*value & BIT(i)) causes undefined behavior due
> > to shifting by a value >= type width.
> >
> > Since 'value' is a pointer to u64, use BIT_ULL() to ensure correct
> > 64-bit mask on all architectures.
> >
> > Found by Linux Verification Center (linuxtesting.org) with Svace.
> >
> > Fixes: 2c4d00cb8fc5 ("gpiolib: acpi: Use BIT() macro to increase readability")
> > Signed-off-by: Denis Sergeev <denserg.edu@gmail.com>
>
> Reviewed-by: Mika Westerberg <westeri@kernel.org>

I'll queue this for v6.19-rc8.

Bart

On Tue, Jan 27, 2026 at 10:06:33AM +0100, Bartosz Golaszewski wrote:
> On Tue, Jan 27, 2026 at 9:08 AM Mika Westerberg
> <mika.westerberg@linux.intel.com> wrote:
> >
> > On Mon, Jan 26, 2026 at 06:59:14AM +0300, Denis Sergeev wrote:
> > > The BIT() macro uses unsigned long, which is 32 bits on 32-bit
> > > architectures. When iterating over GPIO pins with index >= 32,
> > > the expression (*value & BIT(i)) causes undefined behavior due
> > > to shifting by a value >= type width.
> > >
> > > Since 'value' is a pointer to u64, use BIT_ULL() to ensure correct
> > > 64-bit mask on all architectures.
> > >
> > > Found by Linux Verification Center (linuxtesting.org) with Svace.
> > >
> > > Fixes: 2c4d00cb8fc5 ("gpiolib: acpi: Use BIT() macro to increase readability")
> > > Signed-off-by: Denis Sergeev <denserg.edu@gmail.com>
> >
> > Reviewed-by: Mika Westerberg <westeri@kernel.org>
> 
> I'll queue this for v6.19-rc8.

This is half-baked solution... But okay, let's shut up at least the stream of
this type of "fixes" in this file.

-- 
With Best Regards,
Andy Shevchenko