Re: [PATCH 2/3] linux/bits.h: Add fixed-width GENMASK and BIT macros

[Lucas De Marchi <lucas.demarchi@intel.com>]

On Wed, Jun 21, 2023 at 07:20:59PM -0700, Yury Norov wrote:
>Hi Lucas, all!
>
>(Thanks, Andy, for pointing to this thread.)
>
>On Mon, May 08, 2023 at 10:14:02PM -0700, Lucas De Marchi wrote:
>> Add GENMASK_U32(), GENMASK_U16() and GENMASK_U8()  macros to create
>> masks for fixed-width types and also the corresponding BIT_U32(),
>> BIT_U16() and BIT_U8().
>
>Can you split BIT() and GENMASK() material to separate patches?
>
>> All of those depend on a new "U" suffix added to the integer constant.
>> Due to naming clashes it's better to call the macro U32. Since C doesn't
>> have a proper suffix for short and char types, the U16 and U18 variants
>> just use U32 with one additional check in the BIT_* macros to make
>> sure the compiler gives an error when the those types overflow.
>
>I feel like I don't understand the sentence...

maybe it was a digression of the integer constants

>
>> The BIT_U16() and BIT_U8() need the help of GENMASK_INPUT_CHECK(),
>> as otherwise they would allow an invalid bit to be passed. Hence
>> implement them in include/linux/bits.h rather than together with
>> the other BIT* variants.
>
>I don't think it's a good way to go because BIT() belongs to a more basic
>level than GENMASK(). Not mentioning possible header dependency issues.
>If you need to test against tighter numeric region, I'd suggest to
>do the same trick as  GENMASK_INPUT_CHECK() does, but in uapi/linux/const.h
>directly. Something like:
>        #define _U8(x)		(CONST_GT(U8_MAX, x) + _AC(x, U))
>
>> The following test file is is used to test this:
>>
>> 	$ cat mask.c
>> 	#include <linux/types.h>
>> 	#include <linux/bits.h>
>>
>> 	static const u32 a = GENMASK_U32(31, 0);
>> 	static const u16 b = GENMASK_U16(15, 0);
>> 	static const u8 c = GENMASK_U8(7, 0);
>> 	static const u32 x = BIT_U32(31);
>> 	static const u16 y = BIT_U16(15);
>> 	static const u8 z = BIT_U8(7);
>>
>> 	#if FAIL
>> 	static const u32 a2 = GENMASK_U32(32, 0);
>> 	static const u16 b2 = GENMASK_U16(16, 0);
>> 	static const u8 c2 = GENMASK_U8(8, 0);
>> 	static const u32 x2 = BIT_U32(32);
>> 	static const u16 y2 = BIT_U16(16);
>> 	static const u8 z2 = BIT_U8(8);
>> 	#endif
>>
>> Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
>> ---
>>  include/linux/bits.h       | 22 ++++++++++++++++++++++
>>  include/uapi/linux/const.h |  2 ++
>>  include/vdso/const.h       |  1 +
>>  3 files changed, 25 insertions(+)
>>
>> diff --git a/include/linux/bits.h b/include/linux/bits.h
>> index 7c0cf5031abe..ff4786c99b8c 100644
>> --- a/include/linux/bits.h
>> +++ b/include/linux/bits.h
>> @@ -42,4 +42,26 @@
>>  #define GENMASK_ULL(h, l) \
>>  	(GENMASK_INPUT_CHECK(h, l) + __GENMASK_ULL(h, l))
>>
>> +#define __GENMASK_U32(h, l) \
>> +	(((~U32(0)) - (U32(1) << (l)) + 1) & \
>> +	 (~U32(0) >> (32 - 1 - (h))))
>> +#define GENMASK_U32(h, l) \
>> +	(GENMASK_INPUT_CHECK(h, l) + __GENMASK_U32(h, l))
>> +
>> +#define __GENMASK_U16(h, l) \
>> +	((U32(0xffff) - (U32(1) << (l)) + 1) & \
>> +	 (U32(0xffff) >> (16 - 1 - (h))))
>> +#define GENMASK_U16(h, l) \
>> +	(GENMASK_INPUT_CHECK(h, l) + __GENMASK_U16(h, l))
>> +
>> +#define __GENMASK_U8(h, l) \
>> +	(((U32(0xff)) - (U32(1) << (l)) + 1) & \
>> +	 (U32(0xff) >> (8 - 1 - (h))))
>> +#define GENMASK_U8(h, l) \
>> +	(GENMASK_INPUT_CHECK(h, l) + __GENMASK_U8(h, l))
>
>[...]
>
>I see nothing wrong with fixed-wight versions of GENMASK if it helps
>people to write safer code. Can you please in commit message mention
>the exact patch(es) that added a bug related to GENMASK() misuse? It
>would be easier to advocate the purpose of new API with that in mind.
>
>Regarding implementation - we should avoid copy-pasting in cases
>like this. Below is the patch that I boot-tested for x86_64 and
>compile-tested for arm64.
>
>It looks less opencoded, and maybe Andy will be less skeptical about
>this approach because of less maintenance burden. Please take it if
>you like for v2.
>
>Thanks,
>Yury
>
>From 39c5b35075df67e7d88644470ca78a3486367c02 Mon Sep 17 00:00:00 2001
>From: Yury Norov <yury.norov@gmail.com>
>Date: Wed, 21 Jun 2023 15:27:29 -0700
>Subject: [PATCH] bits: introduce fixed-type genmasks
>
>Generalize __GENMASK() to support different types, and implement
>fixed-types versions of GENMASK() based on it.
>
>Signed-off-by: Yury Norov <yury.norov@gmail.com>
>---
> include/linux/bitops.h |  1 -
> include/linux/bits.h   | 22 ++++++++++++----------
> 2 files changed, 12 insertions(+), 11 deletions(-)
>
>diff --git a/include/linux/bitops.h b/include/linux/bitops.h
>index 2ba557e067fe..1db50c69cfdb 100644
>--- a/include/linux/bitops.h
>+++ b/include/linux/bitops.h
>@@ -15,7 +15,6 @@
> #  define aligned_byte_mask(n) (~0xffUL << (BITS_PER_LONG - 8 - 8*(n)))
> #endif
>
>-#define BITS_PER_TYPE(type)	(sizeof(type) * BITS_PER_BYTE)
> #define BITS_TO_LONGS(nr)	__KERNEL_DIV_ROUND_UP(nr, BITS_PER_TYPE(long))
> #define BITS_TO_U64(nr)		__KERNEL_DIV_ROUND_UP(nr, BITS_PER_TYPE(u64))
> #define BITS_TO_U32(nr)		__KERNEL_DIV_ROUND_UP(nr, BITS_PER_TYPE(u32))
>diff --git a/include/linux/bits.h b/include/linux/bits.h
>index 7c0cf5031abe..cb94128171b2 100644
>--- a/include/linux/bits.h
>+++ b/include/linux/bits.h
>@@ -6,6 +6,8 @@
> #include <vdso/bits.h>
> #include <asm/bitsperlong.h>
>
>+#define BITS_PER_TYPE(type)	(sizeof(type) * BITS_PER_BYTE)
>+
> #define BIT_MASK(nr)		(UL(1) << ((nr) % BITS_PER_LONG))
> #define BIT_WORD(nr)		((nr) / BITS_PER_LONG)
> #define BIT_ULL_MASK(nr)	(ULL(1) << ((nr) % BITS_PER_LONG_LONG))
>@@ -30,16 +32,16 @@
> #define GENMASK_INPUT_CHECK(h, l) 0
> #endif
>
>-#define __GENMASK(h, l) \
>-	(((~UL(0)) - (UL(1) << (l)) + 1) & \
>-	 (~UL(0) >> (BITS_PER_LONG - 1 - (h))))
>-#define GENMASK(h, l) \
>-	(GENMASK_INPUT_CHECK(h, l) + __GENMASK(h, l))
>+#define __GENMASK(t, h, l) \
>+	(GENMASK_INPUT_CHECK(h, l) + \
>+	 (((t)~0ULL - ((t)(1) << (l)) + 1) & \
>+	 ((t)~0ULL >> (BITS_PER_TYPE(t) - 1 - (h)))))

yeah... forcing the use of ull and then casting to the type is simpler
and does the job. Checked that it does not break the build if h is
greater than the type and it works

../include/linux/bits.h:40:20: error: right shift count >= width of type [-Werror=shift-count-overflow]
    40 |          ((t)~0ULL >> (BITS_PER_TYPE(t) - 1 - (h)))))
       |                    ^~

However this new version does increase the size. Using i915 module
to test:

$ size build64/drivers/gpu/drm/i915/i915.ko*
    text    data     bss     dec     hex filename
4355676  213473    7048 4576197  45d3c5 build64/drivers/gpu/drm/i915/i915.ko
4361052  213505    7048 4581605  45e8e5 build64/drivers/gpu/drm/i915/i915.ko.new

Lucas De Marchi

>
>-#define __GENMASK_ULL(h, l) \
>-	(((~ULL(0)) - (ULL(1) << (l)) + 1) & \
>-	 (~ULL(0) >> (BITS_PER_LONG_LONG - 1 - (h))))
>-#define GENMASK_ULL(h, l) \
>-	(GENMASK_INPUT_CHECK(h, l) + __GENMASK_ULL(h, l))
>+#define GENMASK(h, l)		__GENMASK(unsigned long,  h, l)
>+#define GENMASK_ULL(h, l)	__GENMASK(unsigned long long, h, l)
>+#define GENMASK_U8(h, l)	__GENMASK(u8,  h, l)
>+#define GENMASK_U16(h, l)	__GENMASK(u16, h, l)
>+#define GENMASK_U32(h, l)	__GENMASK(u32, h, l)
>+#define GENMASK_U64(h, l)	__GENMASK(u64, h, l)
>
> #endif	/* __LINUX_BITS_H */
>-- 
>2.39.2
>