linux-kernel - Re: [RFC] mm/page_alloc: Enumerate bad page reasons

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 for Android: free password hash cracker in your pocket

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

Message-ID: <689b594f-c18a-4131-8049-ac917345099b@arm.com>
Date:   Mon, 30 Mar 2020 17:55:14 +0530
From:   Anshuman Khandual <anshuman.khandual@....com>
To:     Michal Hocko <mhocko@...nel.org>
Cc:     linux-mm@...ck.org, Andrew Morton <akpm@...ux-foundation.org>,
        David Hildenbrand <david@...hat.com>,
        Dan Williams <dan.j.williams@...el.com>,
        Pavel Tatashin <pasha.tatashin@...een.com>,
        linux-kernel@...r.kernel.org
Subject: Re: [RFC] mm/page_alloc: Enumerate bad page reasons



On 03/30/2020 02:13 PM, Michal Hocko wrote:
> On Mon 30-03-20 12:21:37, Anshuman Khandual wrote:
>> Enumerate all existing bad page reasons which can be used in bad_page() for
>> reporting via __dump_page(). Unfortunately __dump_page() cannot be changed.
>> __dump_page() is called from dump_page() that accepts a raw string and is
>> also an exported symbol that is currently being used from various generic
>> memory functions and other drivers. This reduces code duplication while
>> reporting bad pages.
> 
> I dunno. It sounds like over engineering something that is an internal
> stuff. Besides that I consider string reasons kinda obvious and I am
> pretty sure I would have to check them for each numeric alias when want
> to read the code. Yeah, yeah, nothing really hard but still...

Right these are very much self explanatory. Would moving these aliases into
mm/page_alloc.c itself, make it any better for quicker access ?

> 
> So I am not really sure this is all worth the code churn. Besides

I understand but is not just repeating the same strings in similar functions
bit suboptimal as well.

> that I stongly suspect you wanted ...
> 
>> -static void bad_page(struct page *page, const char *reason,
>> +static void bad_page(struct page *page, int reason,
>>  		unsigned long bad_flags)
> 
> ... enum page_bad_reason reason here, right? What is the point of declaring
> an enum when you are not using it?

Sure, will replace here and other local reason variables which are 'int'.