Re: [patch 3/6] futex: Get rid of the val2 conditional dance

From: "André Almeida" <andrealmeid@collabora.com>
To: Thomas Gleixner <tglx@linutronix.de>,
	LKML <linux-kernel@vger.kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>,
	Adhemerval Zanella <adhemerval.zanella@linaro.org>,
	Lukasz Majewski <lukma@denx.de>,
	Florian Weimer <fweimer@redhat.com>,
	Carlos O'Donell <carlos@redhat.com>,
	"Michael Kerrisk (man-pages)" <mtk.manpages@gmail.com>,
	Davidlohr Bueso <dave@stgolabs.net>,
	Ingo Molnar <mingo@kernel.org>,
	Darren Hart <dvhart@infradead.org>,
	Andrei Vagin <avagin@gmail.com>,
	Kurt Kanzenbach <kurt@linutronix.de>,
	kernel@collabora.com
Subject: Re: [patch 3/6] futex: Get rid of the val2 conditional dance
Date: Fri, 23 Apr 2021 20:21:58 -0300	[thread overview]
Message-ID: <5817ec9e-9e80-79ef-d09c-6717b0816963@collabora.com> (raw)
In-Reply-To: <87r1j0vdev.ffs@nanos.tec.linutronix.de>

Às 19:34 de 23/04/21, Thomas Gleixner escreveu:
> On Fri, Apr 23 2021 at 18:40, André Almeida wrote:
>>>    
>>> -	return do_futex(uaddr, op, val, tp, uaddr2, val2, val3);
>>> +	return do_futex(uaddr, op, val, tp, uaddr2, (unsigned long)utime, val3);
>>
>> Given do_futex()'s type signature, I think it makes more sense to cast
>> utime to u32.
> 
> It's a pointer which you better force cast to unsigned long first.
> 
> So the explicit thing would be '(u32)(unsigned long) utime' which is
> what the val2 dance stupidly did with 'int'
> 
> 		val2 = (int) (unsigned long) utime;
> 
> But with doing it at function call argument it's implicit, because the
> 
>    unsigned long  to u32 conversion is well defined
> 
> while
> 
>    (u32)ptr
> 
> is only well defined on 32bit.
> 

I see, thank you for the clarification!

> Thanks,
> 
>          tglx
> 