Re: [PATCH 1/5] ARM: findbit: document ARMv5 bit offset calculation

From: "Russell King (Oracle)" <linux@armlinux.org.uk>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Yury Norov <yury.norov@gmail.com>,
	Catalin Marinas <catalin.marinas@arm.com>,
	Mark Rutland <mark.rutland@arm.com>,
	Will Deacon <will@kernel.org>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH 1/5] ARM: findbit: document ARMv5 bit offset calculation
Date: Fri, 28 Oct 2022 20:46:48 +0100	[thread overview]
Message-ID: <Y1wxqPccRE+biAfw@shell.armlinux.org.uk> (raw)
In-Reply-To: <CAHk-=wh75aOWvaEhYsAcCJSRzhy8MO_SdvDzB6zmvVs+FeAr6w@mail.gmail.com>

On Fri, Oct 28, 2022 at 12:01:00PM -0700, Linus Torvalds wrote:
> Hmm. Interestingly, your _find_first_zero_bit_le() (which
> find_next_bit ends up using except for the first byte) ends up doing
> an optimization that is technically not valid.
> 
> In particular, the *generic* code does
> 
>                         sz = min(idx * BITS_PER_LONG + __ffs(MUNGE(val)), sz);
> 
> for the final result.
> 
> In contrast, the arm code doesn't do the "min()" at all, and if there
> are bits after the bitmap (in a partial byte), it will just return
> those bits.

You've missed how the min() is coded. Specifically, that's handled by:

  cc:   e1510000        cmp     r1, r0
  d0:   31a00001        movcc   r0, r1

which clamps the returned index to the size of the array (held in r1).
So everything is in fact fine - and I think your analysis is incorrect.

Please could you take another look and evaluate whether you think the
arm assembly is incorrect.

I kind'a stopped reading here on the assumption that the remainder of
your email was based on this misinterpretation of the code.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel