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: <4a360dec-79ff-1444-6c1e-830f43b13c2f@huaweicloud.com>
Date: Tue, 2 Sep 2025 16:25:54 +0800
From: Yu Kuai <yukuai1@...weicloud.com>
To: John Garry <john.g.garry@...cle.com>,
 Christoph Hellwig <hch@...radead.org>, Yu Kuai <yukuai1@...weicloud.com>
Cc: anthony <antmbox@...ngman.org.uk>, colyli@...nel.org, hare@...e.de,
 tieren@...as.com, axboe@...nel.dk, tj@...nel.org, josef@...icpanda.com,
 song@...nel.org, akpm@...ux-foundation.org, neil@...wn.name,
 linux-block@...r.kernel.org, linux-kernel@...r.kernel.org,
 cgroups@...r.kernel.org, linux-raid@...r.kernel.org, yi.zhang@...wei.com,
 yangerkun@...wei.com, johnny.chenyi@...wei.com,
 "yukuai (C)" <yukuai3@...wei.com>
Subject: Re: [PATCH RFC 4/7] md/raid10: convert read/write to use
 bio_submit_split()

Hi,

在 2025/09/02 14:58, John Garry 写道:
> On 02/09/2025 07:30, Christoph Hellwig wrote:
>> On Tue, Sep 02, 2025 at 07:18:01AM +0100, John Garry wrote:
>>> BTW, do we realistically expect atomic writes HW support and bad 
>>> blocks ever
>>> to meet?
>>
>> That's the point I'm trying to make.  bad block tracking is stupid
>> with modern hardware.  Both SSDs and HDDs are overprovisioned on
>> physical "blocks", and once they run out fine grained bad block tracking
>> is not going to help.  І really do not understand why md even tries
>> to do this bad block tracking, 
> 
> Just because they can try to deal with bad blocks for some (mirroring) 
> personalities, I suppose.

I agree it's useless for enterprise storage, however, for personal
storage, there are lots of users using cost-effective (often aging)
disks, badblocks tracking can reduce the risk of data lost, and
make sure these devices will not become waste.

> 
>> but claiming to support atomic writes
>> while it does is actively harmful.
>>
> 
> There does not look to be some switch to turn off bad block support. 
> That's from briefly checking raid10.c anyway. Kuai, any thoughts on 
> whether we should allow this to be disabled?
> 

I remember that I used to suggest always enable failfast in this case,
and badblocks can be bypassed. Anyway, I think it's good to allow this
to be disabled, it will behave very similar to failfast.

Thanks,
Kuai

> Thanks,
> John
> 
> .
> 


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ