clocksource: mips-gic-timer: fix clocksource counter width

This code needs to use ffs instead of fls on the mask to determine the shift for reading the GIC_CONFIG_COUNTBITS field.

Fixes e07127a077c7 ("clocksource: mips-gic-timer: Use new GIC accessor functions")

Cc Paul Burton <[email protected]>

Cc [email protected]

Signed-off-by Felix Fietkau <[email protected]>

---
 drivers/clocksource/mips-gic-timer.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

drivers/clocksource/mips-gic-timer.c +1 -1

index a04808a21d4e..bf0eee78c6ef 100644
@@ -166,7 +166,7 @@ static int __init __gic_clocksource_init(void)
 
 	/* Set clocksource mask. */
 	count_width = read_gic_config() & GIC_CONFIG_COUNTBITS;
-	count_width >>= __fls(GIC_CONFIG_COUNTBITS);
+	count_width >>= __ffs(GIC_CONFIG_COUNTBITS);
 	count_width *= 4;
 	count_width += 32;
 	gic_clocksource.mask = CLOCKSOURCE_MASK(count_width);

Signature · 1 line

2.14.2

Quoted 2 lines · Felix Fietkau

This code needs to use ffs instead of fls on the mask to determine the shift for reading the GIC_CONFIG_COUNTBITS field.

Why?

drivers/clocksource/mips-gic-timer.c static int __init __gic_clocksource_init(void) · line 169

Quoted 3 lines

drivers/clocksource/mips-gic-timer.c +1 -1

 	count_width = read_gic_config() & GIC_CONFIG_COUNTBITS;
-	count_width >>= __fls(GIC_CONFIG_COUNTBITS);
+	count_width >>= __ffs(GIC_CONFIG_COUNTBITS);

Quoted 5 lines · Thomas Gleixner

Quoted 2 lines · Felix Fietkau

This code needs to use ffs instead of fls on the mask to determine the shift for reading the GIC_CONFIG_COUNTBITS field.

Quoted 4 lines

Why?

drivers/clocksource/mips-gic-timer.c static int __init __gic_clocksource_init(void) · line 169

Quoted 3 lines

drivers/clocksource/mips-gic-timer.c +1 -1

 	count_width = read_gic_config() & GIC_CONFIG_COUNTBITS;
-	count_width >>= __fls(GIC_CONFIG_COUNTBITS);
+	count_width >>= __ffs(GIC_CONFIG_COUNTBITS);

This code is trying to extract the GIC_CONFIG_COUNTBITS field from read_gic_config(), so it needs to shift count_width right by the index of the least significant bit (__ffs) instead of the most significant bit (__fls) of the field mask.