linux-kernel - Re: [RESEND] Re: [PATCH] printk: Fix _DESCS

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] [day] [month] [year] [list]

Message-ID: <aYIvQlbgtO4LjUqU@pathway.suse.cz>
Date: Tue, 3 Feb 2026 18:24:18 +0100
From: Petr Mladek <pmladek@...e.com>
To: "feng.zhou" <realsummitzhou@...il.com>
Cc: john.ogness@...utronix.de, linux-kernel@...r.kernel.org,
	rostedt@...dmis.org, senozhatsky@...omium.org
Subject: Re: [RESEND] Re: [PATCH] printk: Fix _DESCS_COUNT type for 64-bit
 systems

On Mon 2026-02-02 23:07:50, feng.zhou wrote:
> Hi John and maintainers,
> 
> (Resending to the list - apologies for the initial off-list reply)
> 
> On Mon, Feb 2, 2026, John Ogness wrote:
> > This is important to us as well! And we really do want those
> > wrap-arounds to happen early.
> 
> > Your v1 is certainly OK as-is. I just want to do some more investigating
> > and testing. With your patch it will be the first time we experience
> > descriptor ID wraps.
> >
> > As to the other U->UL conversion, I really have no preference. Let's
> > just wait a bit in case Petr wants to comment.
>
> Thank you for the feedback! I understand v1 is acceptable and you're doing
> additional testing. I'll wait for Petr's comments and your investigation
> results before any further action.

JFYI, this patch is on my radar. But the review might take some time:

  1. I am a bit overloaded.

  2. This patch has low priority. It helps to test a rather
     non-realistic scenario.

  3. The review is far from easy because the macro _DESCS_COUNT()
     is used on many cricial parts of the lock-less code via
     DESC_COUNT() macro.

> For the record, here's a summary of our off-list discussion:
> 
> 1. Early wrap-around testing is important for the printk_ringbuffer design
>    itself, not just for matching documentation.
> 
> 2. John noted this is interesting because it will be the first time descriptor
>    ID wraps are actually exercised in practice.
> 
> 3. There's a similar issue with the text buffer array size definition using 1U,
>    but we agreed to wait for broader maintainer input on whether to address it
>    in the same patch or separately.

This is another complicated problem. The printk log buffer size is
limited to 2GB. I am not exactly sure why. Some clue can be found
at https://lore.kernel.org/all/20181008135916.gg4kkmoki5bgtco5@pathway.suse.cz/


> I apologize for not using "Reply All" initially - I'm still learning the
> kernel development process. I'll make sure all future communications stay
> on-list.

No problem. It happens. It is great that you are learning.

> Thanks again for your patience and guidance!

You are welcome. Please, have patience with us as well.

Best Regards,
Petr