It is a common code pattern to modify a bitfield by masking the field
and performing a bitwise OR with the respective FIELD_PREP. Wrap such a
task into a macro by introducing FIELD_MODIFY() which modifies the field
specified by a mask from a bitfield by putting a val in the field.

Signed-off-by: William Breathitt Gray <william.gray@linaro.org>
---
 include/linux/bitfield.h | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/include/linux/bitfield.h b/include/linux/bitfield.h
index ebfa12f69501..b06a98f0a87f 100644
--- a/include/linux/bitfield.h
+++ b/include/linux/bitfield.h
@@ -38,8 +38,7 @@
  *	  FIELD_PREP(REG_FIELD_D, 0x40);
  *
  * Modify:
- *  reg &= ~REG_FIELD_C;
- *  reg |= FIELD_PREP(REG_FIELD_C, c);
+ *  reg = FIELD_MODIFY(REG_FIELD_C, reg, c);
  */
 
 #define __bf_shf(x) (__builtin_ffsll(x) - 1)
@@ -155,6 +154,21 @@
 		(typeof(_mask))(((_reg) & (_mask)) >> __bf_shf(_mask));	\
 	})
 
+/**
+ * FIELD_MODIFY() - modify a bitfield element
+ * @_mask: shifted mask defining the field's length and position
+ * @_reg:  value of entire bitfield
+ * @_val:  value to put in the field
+ *
+ * FIELD_MODIFY() modifies the field specified by @_mask from the
+ * bitfield passed in as @_reg by putting @val in the field.
+ */
+#define FIELD_MODIFY(_mask, _reg, _val)						\
+	({									\
+		__BF_FIELD_CHECK(_mask, _reg, _val, "FIELD_MODIFY: ");		\
+		(typeof(_mask))(((_reg) & ~(_mask)) | FIELD_PREP(_mask, _val));	\
+	})
+
 extern void __compiletime_error("value doesn't fit into mask")
 __field_overflow(void);
 extern void __compiletime_error("bad bitfield mask")
-- 
2.39.2

On Sat, 2023-03-18 at 14:59 +0000, William Breathitt Gray wrote:
> It is a common code pattern to modify a bitfield by masking the field
> and performing a bitwise OR with the respective FIELD_PREP. Wrap such a
> task into a macro by introducing FIELD_MODIFY() which modifies the field
> specified by a mask from a bitfield by putting a val in the field.

So I have no objection to adding this and you using FIELD_* macros, but
just wanted to say that personally I've come to prefer the typed
versions declared later in the fiel, and there we have
<type>_replace_bits() already.

Hmm. And now that I mentioned that, maybe that means FIELD_REPLACE()
would be nicer as a name?

johannes

On Mon, Mar 20, 2023 at 09:50:35AM +0100, Johannes Berg wrote:
> On Sat, 2023-03-18 at 14:59 +0000, William Breathitt Gray wrote:
> > It is a common code pattern to modify a bitfield by masking the field
> > and performing a bitwise OR with the respective FIELD_PREP. Wrap such a
> > task into a macro by introducing FIELD_MODIFY() which modifies the field
> > specified by a mask from a bitfield by putting a val in the field.
> 
> So I have no objection to adding this and you using FIELD_* macros, but
> just wanted to say that personally I've come to prefer the typed
> versions declared later in the fiel, and there we have
> <type>_replace_bits() already.
> 
> Hmm. And now that I mentioned that, maybe that means FIELD_REPLACE()
> would be nicer as a name?
> 
> johannes

Perhaps I can convert all of these FIELD_GET(), FIELD_MODIFY(), and
FIELD_GET() to the equivalent of u8_get_bits(), u8p_replace_bits(), and
u8_encode_bits(). If that works, then I'll just drop the FIELD_MODIFY()
patch in the v2 patchset submission.

William Breathitt Gray

On Mon, Mar 20, 2023 at 09:50:35AM +0100, Johannes Berg wrote:
> On Sat, 2023-03-18 at 14:59 +0000, William Breathitt Gray wrote:
> > It is a common code pattern to modify a bitfield by masking the field
> > and performing a bitwise OR with the respective FIELD_PREP. Wrap such a
> > task into a macro by introducing FIELD_MODIFY() which modifies the field
> > specified by a mask from a bitfield by putting a val in the field.
> 
> So I have no objection to adding this and you using FIELD_* macros, but
> just wanted to say that personally I've come to prefer the typed
> versions declared later in the fiel, and there we have
> <type>_replace_bits() already.
> 
> Hmm. And now that I mentioned that, maybe that means FIELD_REPLACE()
> would be nicer as a name?

+1 here with the similar thoughts.

One thing I hate about macros like above mentioned is that Elixir or similar
code browsing tools can't find. In net there are specific #if 0 ... #endif
sections for mitigating that.

Shouldn't we add the similar into bitfield.h?

-- 
With Best Regards,
Andy Shevchenko