[PATCH v2 0/3] bits: Split asm and non-asm GENMASK*() and unify definitions

[Yury Norov]

On Mon, Jun 09, 2025 at 11:45:44AM +0900, Vincent Mailhol via B4 Relay wrote:
> This is a subset of below series:
> 
>   bits: Fixed-type GENMASK_U*() and BIT_U*()
>   Link: https://lore.kernel.org/r/20250308-fixed-type-genmasks-v6-0-f59315e73c29@wanadoo.fr
> 
> Yury suggested to split the above series in two steps:
> 
>   #1 Introduce the new fixed type GENMASK_U*() (already merged upstream)
>   #2 Consolidate the existing GENMASK*()
> 
> This new series is the resulting step #2 following the split.
> 
> And thus, this series consolidate all the non-asm GENMASK*() so that
> they now all depend on GENMASK_TYPE() which was introduced in step #1.
> 
> To do so, I had to split the definition of the asm and non-asm
> GENMASK(). I think this is controversial. So I initially implemented a
> first draft in which both the asm and non-asm version would rely on
> the same helper macro, i.e. adding this:
> 
>   #define __GENMASK_TYPE(t, w, h, l)		\
>   	(((t)~_ULL(0) << (l)) &			\
>   	 ((t)~_ULL(0) >> (w - 1 - (h))))
> 
> to uapi/bits.h. And then, the different GENMASK()s would look like
> this:
> 
>   #define __GENMASK(h, l) __GENMASK_TYPE(unsigned long, __BITS_PER_LONG, h, l)
> 
> and so on.
> 
> I implemented it, and the final result looked quite ugly. Not only do
> we need to manually provide the width each time, the biggest concern
> is that adding this to the uapi is asking for trouble. Who knows how
> people are going to use this? And once it is in the uapi, there is
> virtually no way back.
> 
> Adding to this, that macro can not even be generalised to u128
> integers, whereas after the split, it can.
> 
> And so, after implementing both, the asm seems way cleaner than the
> non-asm split and is, I think, the best compromise.
> 
> Aside from the split, the asm's GENMASK() and GENMASK_ULL() are left
> untouched. While there are some strong incentives to also simplify
> these as pointed by David Laight in this thread:
> 
>   https://lore.kernel.org/all/20250309102312.4ff08576@pumpkin/
> 
> this series deliberately limit its scope to the non-asm variants.
> 
> Here are the bloat-o-meter stats:
> 
>   $ ./scripts/bloat-o-meter vmlinux_before.o vmlinux_after.o
>   add/remove: 0/0 grow/shrink: 4/2 up/down: 5/-9 (-4)
>   Function                                     old     new   delta
>   intel_psr_invalidate                         352     354      +2
>   mst_stream_compute_config                   1589    1590      +1
>   intel_psr_flush                              707     708      +1
>   intel_dp_compute_link_config                1338    1339      +1
>   intel_drrs_activate                          398     395      -3
>   cfg80211_inform_bss_data                    5137    5131      -6
>   Total: Before=23333846, After=23333842, chg -0.00%
> 
> (done with GCC 12.4.1 on an x86_64 defconfig)

So, I'm still concerned about that split for C and asm implementations.
But seemingly nobody else does, and after all it's a nice consolidation.

I've moved this in bitmap-for-next for testing. Thank you Vincent for
your work.

Thanks,
Yury