On Thu, Feb 20, 2025 at 11:32:10AM -0600, David Lechner wrote:

On 2/20/25 4:17 AM, Simon Horman wrote:

quoted

On Mon, Feb 10, 2025 at 04:33:31PM -0600, David Lechner wrote:

quoted

Use bitmap_get_value8() instead of accessing the bitmap directly.

Accessing the bitmap directly is not considered good practice. We now
have a helper function that can be used instead, so let's use it.

Suggested-by: Andy Shevchenko <redacted>

u> Signed-off-by: David Lechner [off-list ref]

quoted

---
 drivers/bus/ts-nbus.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/drivers/bus/ts-nbus.c b/drivers/bus/ts-nbus.c
index b4c9308caf0647a3261071d9527fffce77784af2..beac67f3b820377f8bb1fc4f4ee77e15ee240834 100644
--- a/drivers/bus/ts-nbus.c
+++ b/drivers/bus/ts-nbus.c

@@ -10,6 +10,7 @@
  * TS-4600 SoM.
  */
 
+#include <linux/bitmap.h>
 #include <linux/bitops.h>
 #include <linux/gpio/consumer.h>
 #include <linux/kernel.h>

@@ -107,7 +108,7 @@ static void ts_nbus_reset_bus(struct ts_nbus *ts_nbus)
 {
 	DECLARE_BITMAP(values, 8);
 
-	values[0] = 0;
+	bitmap_set_value8(values, byte, 0);

Hi David,

byte doesn't appear to exist in the scope of this function.

I tried this:

	bitmap_set_value8(values, 0, 8);

But when compiling with GCC 14.2.0 I see warnings that values
is used uninitialised - bitmap_set_value8() appears to rely on
it being so.

Ah yes, I see the problem (I don't think this driver compiles with
allmodconfig so the compiler didn't catch it for me).

Thanks, that would explain things.

FWIIW, I think you can exercise this with allmodconfig by simply running:

  make drivers/bus/ts-nbus.o

quoted

  CC      drivers/bus/ts-nbus.o
In file included from drivers/bus/ts-nbus.c:13:
In function ‘bitmap_write’,
    inlined from ‘ts_nbus_reset_bus’ at drivers/bus/ts-nbus.c:111:2:
./include/linux/bitmap.h:818:12: error: ‘values’ is used uninitialized [-Werror=uninitialized]
  818 |         map[index] &= (fit ? (~(mask << offset)) : ~BITMAP_FIRST_WORD_MASK(start));
      |         ~~~^~~~~~~
In file included from ./include/linux/kasan-checks.h:5,
                 from ./include/asm-generic/rwonce.h:26,
                 from ./arch/x86/include/generated/asm/rwonce.h:1,
                 from ./include/linux/compiler.h:344,
                 from ./include/linux/build_bug.h:5,
                 from ./include/linux/bits.h:22,
                 from ./include/linux/bitops.h:6,
                 from ./include/linux/bitmap.h:8:
drivers/bus/ts-nbus.c: In function ‘ts_nbus_reset_bus’:
drivers/bus/ts-nbus.c:109:24: note: ‘values’ declared here
  109 |         DECLARE_BITMAP(values, 8);
      |                        ^~~~~~
./include/linux/types.h:11:23: note: in definition of macro ‘DECLARE_BITMAP’
   11 |         unsigned long name[BITS_TO_LONGS(bits)]
      |                       ^~~~

quoted

 
 	gpiod_multi_set_value_cansleep(ts_nbus->data, values);
 	gpiod_set_value_cansleep(ts_nbus->csn, 0);

@@ -151,7 +152,7 @@ static void ts_nbus_write_byte(struct ts_nbus *ts_nbus, u8 byte)
 {
 	DECLARE_BITMAP(values, 8);

We can fix by zero-initialing the bitmap.

	DECLARE_BITMAP(values, 8) = { };

Thanks, I confirmed that adding that to ts_nbus_reset_bus()
makes the compiler happy. And it seems sensible to me.

I guess that theoretically it should also be added to ts_nbus_write_byte(),
although GCC has nothing to say about that either way.

Would you like me to send a new version of the patch?

It's not really my call. But I would expect that is a good next step.

quoted

quoted

-	values[0] = byte;
+	bitmap_set_value8(values, byte, 8);
 
 	gpiod_multi_set_value_cansleep(ts_nbus->data, values);
 }