linux-kernel - Re: [PATCH] mm: vmscan: Avoid signedness error for GCC 5.4

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: <20250507130629.303b01f8@pumpkin>
Date: Wed, 7 May 2025 13:06:29 +0100
From: David Laight <david.laight.linux@...il.com>
To: Matthew Wilcox <willy@...radead.org>
Cc: WangYuli <wangyuli@...ontech.com>, akpm@...ux-foundation.org,
 linux-mm@...ck.org, linux-kernel@...r.kernel.org, yuzhao@...gle.com,
 stevensd@...omium.org, kaleshsingh@...gle.com, zhanjun@...ontech.com,
 niecheng1@...ontech.com, guanwentao@...ontech.com
Subject: Re: [PATCH] mm: vmscan: Avoid signedness error for GCC 5.4

On Tue, 6 May 2025 17:22:51 +0100
Matthew Wilcox <willy@...radead.org> wrote:

> On Wed, May 07, 2025 at 12:02:38AM +0800, WangYuli wrote:
> > To the compiler, (MAX_NR_TIERS - 1) (i.e., (4U - 1)) is unsigned,
> > whereas tier is a signed integer.
> > 
> > GCC 5.4 does not permit the minimum operation on such
> > type-inconsistent operands.  
> 
> 1. This has nothing to do with the compiler version; the type-checking
> is built into min().
> 2. We have min_t for a reason

Mostly historical - to match the original inline function min().
min_t() is definitely overused, it should be the 'last resort'
for a type mismatch, not the first.

> 3. Why is a signed min the right answer instead of an unsigned min?
> 

I don't seem to have the patch itself, but I' guessing it is for:

for (i = tier % MAX_NR_TIERS; i <= min(tier, MAX_NR_TIERS - 1); i++) {

which seems to have been added for 6.14-rc1 - so why is it only an issue now.

Looks closer, I bet the function is usually inlined.

	David