linux-kernel - Re: [PATCH next 12/14] bits: move the defitions of BIT() and BIT

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20260122082822-f25acbd0-6f7b-41a6-92a1-0873f967ca18@linutronix.de>
Date: Thu, 22 Jan 2026 08:39:42 +0100
From: Thomas Weißschuh <thomas.weissschuh@...utronix.de>
To: David Laight <david.laight.linux@...il.com>
Cc: Nathan Chancellor <nathan@...nel.org>, 
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>, Thomas Gleixner <tglx@...utronix.de>, 
	Peter Zijlstra <peterz@...radead.org>, Ingo Molnar <mingo@...nel.org>, 
	Mathieu Desnoyers <mathieu.desnoyers@...icios.com>, Arnd Bergmann <arnd@...db.de>, linux-arch@...r.kernel.org, 
	linux-kernel@...r.kernel.org, Yury Norov <yury.norov@...il.com>, 
	Lucas De Marchi <lucas.demarchi@...el.com>, Jani Nikula <jani.nikula@...el.com>, 
	Vincent Mailhol <mailhol.vincent@...adoo.fr>, Andy Shevchenko <andriy.shevchenko@...ux.intel.com>, 
	Kees Cook <keescook@...omium.org>, Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [PATCH next 12/14] bits: move the defitions of BIT() and
 BIT_ULL() back to linux/bits.h

On Wed, Jan 21, 2026 at 07:24:51PM +0000, David Laight wrote:
> On Wed, 21 Jan 2026 16:17:18 +0100
> Thomas Weißschuh <thomas.weissschuh@...utronix.de> wrote:
> 
> > On Wed, Jan 21, 2026 at 02:57:29PM +0000, david.laight.linux@...il.com wrote:
> > > From: David Laight <david.laight.linux@...il.com>
> > > 
> > > The definition of BIT() was moved from linux/bits.h to vdso/bits.h to
> > > isolate the vdso from 'normal' kernel headers.
> > > BIT_ULL() was then moved to be defined in the same place for consistency.
> > > 
> > > Since then linux/bits.h had gained BIT_Unn() and it really makes sense
> > > for BIT() and BIT_ULL() to be defined in the same place.
> > > 
> > > Move BIT_ULL() and make code that include both headers use the definition
> > > of BIT() from linux/bits.h
> > > Add BIT_U128() for completness.
> > > 
> > > This lets BIT() pick up the extra compile time checks for W=[1c] builds
> > > that detect errors like:
> > > 	long foo(void) { int x = 64; return BIT(x); }
> > > For which clang (silently) just generates a 'return' instruction.
> > > 
> > > Note that nothing the the x86-64 build relies on the definition in
> > > vdso/bits.h, linux/bits.h is always included.
> > > 
> > > Signed-off-by: David Laight <david.laight.linux@...il.com>
> > > ---
> > >  include/linux/bits.h | 7 ++++++-
> > >  include/vdso/bits.h  | 2 +-
> > >  2 files changed, 7 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/include/linux/bits.h b/include/linux/bits.h
> > > index 0f559038981d..3dd32b9eef35 100644
> > > --- a/include/linux/bits.h
> > > +++ b/include/linux/bits.h
> > > @@ -2,7 +2,6 @@
> > >  #ifndef __LINUX_BITS_H
> > >  #define __LINUX_BITS_H
> > >  
> > > -#include <vdso/bits.h>
> > >  #include <uapi/linux/bits.h>
> > >  
> > >  #define BIT_MASK(nr)		(UL(1) << ((nr) % BITS_PER_LONG))
> > > @@ -89,10 +88,16 @@ int BIT_INPUT_CHECK_FAIL(void) __compiletime_error("Bit number out of range");
> > >  	((unsigned int)BIT_INPUT_CHECK(+(nr), BITS_PER_TYPE(type)) + ((type)1 << (nr)))
> > >  #endif /* defined(__ASSEMBLY__) */
> > >  
> > > +/* Prefer this definition of BIT() to the one in vdso/bits.h */
> > > +#undef BIT
> > > +#define __VDSO_BITS_H  
> > 
> > This is ugly.
> 
> It works :-)
> I could have put an #ifndef BIT in vdso/bits.h instead.

The #ifndef in vdso/bits.h is not much better IMO.
Given that there a multiple nicer options why not use on of those?

> > Why can't the vDSO code make use of those checks, too?
> > Or use _BITUL() from the UAPI in the vDSO and remove vdso/bits.h.
> 
> I didn't actually find anything that just needed vdso/bits.h
> linux/bits.h would get included - eg (IIRC) because of warn_on_once().

That is a bug. The vdso code should only include the vdso/ namespace.
Right now the inclusions are all over the place. I'd like to clean that
up at some point.

> I'm not that sure why it got separated, it isn't as though it defines
> anything that is code version specific.

I don't understand "code version specific".

There are various issues with including normal kernel headers from the vDSO.
Symbols defined in terms of kconfig options, like BITS_PER_LONG using
CONFIG_64BIT are wrong in a compat vDSO.

> > > +#define BIT(nr)		BIT_TYPE(unsigned long, nr)
> > > +#define BIT_ULL(nr)	BIT_TYPE(unsigned long long, nr)
> > >  #define BIT_U8(nr)	BIT_TYPE(u8, nr)
> > >  #define BIT_U16(nr)	BIT_TYPE(u16, nr)
> > >  #define BIT_U32(nr)	BIT_TYPE(u32, nr)
> > >  #define BIT_U64(nr)	BIT_TYPE(u64, nr)
> > > +#define BIT_U128(nr)	BIT_TYPE(u128, nr)
> > >  
> > >  #if defined(__ASSEMBLY__)
> > >  
> > > diff --git a/include/vdso/bits.h b/include/vdso/bits.h
> > > index 388b212088ea..a6ac1e6b637c 100644
> > > --- a/include/vdso/bits.h
> > > +++ b/include/vdso/bits.h
> > > @@ -4,7 +4,7 @@  
> > >  >  #include <vdso/const.h>  
> > >  
> > > +/* Most code picks up BIT() from linux/bits.h */
> > >  #define BIT(nr)			(UL(1) << (nr))
> > > -#define BIT_ULL(nr)		(ULL(1) << (nr))
> > >  
> > >  #endif	/* __VDSO_BITS_H */
> > > -- 
> > > 2.39.5
> > >   
>