linux-kernel - Re: pktcdvd

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: <1dea09ad-73b8-8f9f-c0d1-5637413349b5@kernel.dk>
Date:   Sun, 29 Jan 2023 15:34:04 -0700
From:   Jens Axboe <axboe@...nel.dk>
To:     Pali Rohár <pali@...nel.org>
Cc:     Linus Torvalds <torvalds@...ux-foundation.org>,
        Christoph Hellwig <hch@...radead.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: pktcdvd

On 1/29/23 3:21 PM, Pali Rohár wrote:
> On Sunday 29 January 2023 14:55:27 Jens Axboe wrote:
>> This does not look like a new issue (eg related to this series), and
>> this is exactly one of the reasons we wanted to get rid of this code -
>> basically nobody tests it, as nobody has the ability to. That means it's
>> very time consuming to debug issues with it.
> 
> I understand. I could try to look at it. The main issue for me is that I
> have not looked at this low level block device kernel area and I do not
> understand what is the code around doing...

I'll be happy to answer questions if you want to dive in... Looking at
the oops, it's clear that a bio arrived (or was split) that didn't end
up on a packet size alignment. We didn't error on the full size
alignment, so the total size of the write is a multiple of the packet
size. The bio_split() is pretty straightforward, so perhaps the starting
sector wasn't a multiple of the packet size to begin with?

>> What is the newest kernel you've run the pktcdvd driver on?
> 
> 4.19 from Debian 10. I'm not sure if I used newer version as most stuff
> is still on Debian 10 as I have not find a time to do upgrade of
> everything.

Ok, yeah that's pretty old...

-- 
Jens Axboe