linux-kernel - RE: [PATCH v4 1/3] math.h: add DIV_ROUND_UP_NO

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: <2873863a22f441c9889ba405d8a71d97@AcuMS.aculab.com>
Date:   Thu, 26 Oct 2023 08:41:47 +0000
From:   David Laight <David.Laight@...LAB.COM>
To:     'Linus Torvalds' <torvalds@...ux-foundation.org>
CC:     Sebastian Reichel <sebastian.reichel@...labora.com>,
        Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
        Michael Turquette <mturquette@...libre.com>,
        Stephen Boyd <sboyd@...nel.org>,
        "linux-clk@...r.kernel.org" <linux-clk@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        Vasily Gorbik <gor@...ux.ibm.com>,
        Niklas Schnelle <schnelle@...ux.ibm.com>,
        "kernel@...labora.com" <kernel@...labora.com>
Subject: RE: [PATCH v4 1/3] math.h: add DIV_ROUND_UP_NO_OVERFLOW

From: Linus Torvalds 
> Sent: 25 October 2023 18:41
> 
> On Tue, 24 Oct 2023 at 22:38, David Laight <David.Laight@...lab.com> wrote:
> >
> > From: Linus Torvalds
> > > >  - we can't even use a statement expression and __auto_type, because
> > > > these things are used in type definitions etc and need to be constant
> > > > expressions
> >
> > Doesn't min() get around that by using is_constexpr() and
> > __builtin_choose_exptr() - the same could be done here.
> 
> Nope. I wanted to do it that way - it would have made things much
> simpler and avoid the whole _Generic() thing, but try it - you cannot
> use statement expressions in a non-function context even with
> __builtin_choose_expr().

_Generic() has exactly the same issues as __builtin_choose_expr().
All the code has to be valid for all types.
Makes testing for negative constants a PITA.
Something like this worked:
	(__buitin_choose_expr(__is_constexpr((x) && is_signed_type(__typeof(x)), x, 1) < 0)
Although our is_signed_type() isn't constant (enough) for pointer types.
So you need another check for (typeof(x))1 being constant.

> And no, min/max have never been usable in that context

I've clearly not tested it and assumed the comment implied
that was ok.

It'd probably work in C++ - which is perfectly willing to
add functions to initialise static data - even when you'd
rather have a compile time error.

	David

> 
>                  Linus

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)