From: msizanoen <msizanoen@qtmlabs.xyz>

The AlpsPS/2 code previously relied on the assumption that `char` is a
signed type, which was true on x86 platforms (the only place where this
driver is used) before kernel 6.2. However, on 6.2 and later, this
assumption is broken due to the introduction of -funsigned-char as a new
global compiler flag.

Fix this by explicitly specifying the signedness of `char` when sign
extending the values received from the device.

v2:
	Add explicit signedness to more places

v3:
	Use `s8` instead of `signed char`

Fixes: f3f33c677699 ("Input: alps - Rushmore and v7 resolution support")
Cc: stable@vger.kernel.org
Signed-off-by: msizanoen <msizanoen@qtmlabs.xyz>
---
 drivers/input/mouse/alps.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/drivers/input/mouse/alps.c b/drivers/input/mouse/alps.c
index 989228b5a0a4..e2c11d9f3868 100644
--- a/drivers/input/mouse/alps.c
+++ b/drivers/input/mouse/alps.c
@@ -852,8 +852,8 @@ static void alps_process_packet_v6(struct psmouse *psmouse)
 			x = y = z = 0;
 
 		/* Divide 4 since trackpoint's speed is too fast */
-		input_report_rel(dev2, REL_X, (char)x / 4);
-		input_report_rel(dev2, REL_Y, -((char)y / 4));
+		input_report_rel(dev2, REL_X, (s8)x / 4);
+		input_report_rel(dev2, REL_Y, -((s8)y / 4));
 
 		psmouse_report_standard_buttons(dev2, packet[3]);
 
@@ -1104,8 +1104,8 @@ static void alps_process_trackstick_packet_v7(struct psmouse *psmouse)
 	    ((packet[3] & 0x20) << 1);
 	z = (packet[5] & 0x3f) | ((packet[3] & 0x80) >> 1);
 
-	input_report_rel(dev2, REL_X, (char)x);
-	input_report_rel(dev2, REL_Y, -((char)y));
+	input_report_rel(dev2, REL_X, (s8)x);
+	input_report_rel(dev2, REL_Y, -((s8)y));
 	input_report_abs(dev2, ABS_PRESSURE, z);
 
 	psmouse_report_standard_buttons(dev2, packet[1]);
@@ -2294,20 +2294,20 @@ static int alps_get_v3_v7_resolution(struct psmouse *psmouse, int reg_pitch)
 	if (reg < 0)
 		return reg;
 
-	x_pitch = (char)(reg << 4) >> 4; /* sign extend lower 4 bits */
+	x_pitch = (s8)(reg << 4) >> 4; /* sign extend lower 4 bits */
 	x_pitch = 50 + 2 * x_pitch; /* In 0.1 mm units */
 
-	y_pitch = (char)reg >> 4; /* sign extend upper 4 bits */
+	y_pitch = (s8)reg >> 4; /* sign extend upper 4 bits */
 	y_pitch = 36 + 2 * y_pitch; /* In 0.1 mm units */
 
 	reg = alps_command_mode_read_reg(psmouse, reg_pitch + 1);
 	if (reg < 0)
 		return reg;
 
-	x_electrode = (char)(reg << 4) >> 4; /* sign extend lower 4 bits */
+	x_electrode = (s8)(reg << 4) >> 4; /* sign extend lower 4 bits */
 	x_electrode = 17 + x_electrode;
 
-	y_electrode = (char)reg >> 4; /* sign extend upper 4 bits */
+	y_electrode = (s8)reg >> 4; /* sign extend upper 4 bits */
 	y_electrode = 13 + y_electrode;
 
 	x_phys = x_pitch * (x_electrode - 1); /* In 0.1 mm units */
-- 
2.39.2

On Mon, Mar 20, 2023 at 05:52:29AM +0100, msizanoen1 wrote:
> From: msizanoen <msizanoen@qtmlabs.xyz>
> 
> The AlpsPS/2 code previously relied on the assumption that `char` is a
> signed type, which was true on x86 platforms (the only place where this
> driver is used) before kernel 6.2. However, on 6.2 and later, this
> assumption is broken due to the introduction of -funsigned-char as a new
> global compiler flag.
> 
> Fix this by explicitly specifying the signedness of `char` when sign
> extending the values received from the device.
> 
> v2:
> 	Add explicit signedness to more places
> 
> v3:
> 	Use `s8` instead of `signed char`
> 
> Fixes: f3f33c677699 ("Input: alps - Rushmore and v7 resolution support")
> Cc: stable@vger.kernel.org
> Signed-off-by: msizanoen <msizanoen@qtmlabs.xyz>

Applied, thank you.

-- 
Dmitry