linux-kernel - Re: [PATCH v2] block: ublk: enable zoned storage support

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: <0d0e5c63-a155-d2ad-1829-c016c61acad6@opensource.wdc.com>
Date:   Thu, 2 Mar 2023 17:19:20 +0900
From:   Damien Le Moal <damien.lemoal@...nsource.wdc.com>
To:     Andreas Hindborg <nmi@...aspace.dk>, Ming Lei <ming.lei@...hat.com>
Cc:     linux-block@...r.kernel.org, Hans Holmberg <Hans.Holmberg@....com>,
        Matias Bjorling <Matias.Bjorling@....com>,
        Niklas Cassel <Niklas.Cassel@....com>,
        kernel test robot <lkp@...el.com>,
        Jens Axboe <axboe@...nel.dk>,
        open list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2] block: ublk: enable zoned storage support

On 3/2/23 16:31, Andreas Hindborg wrote:
>>> +struct ublk_param_zoned {
>>> +	__u64	max_open_zones;
>>> +	__u64	max_active_zones;
>>> +	__u64	max_append_size;
>>> +};
>>
>> Is the above zoned parameter enough for future extension?
>> Does ZNS need extra parameter? Or some zoned new(important) features?
> 
> @Damien, @Hans, @Matias, what do you think?

Yes, add some reserved fields. The struct is 24 B for now, you can make it 32 B.
But it is a little odd: why 64 bits for max open/active zones and max append ?
bio len is 32 bits and number of zones also 32 bits. You do not need 64 bits for
all these fields. Also, no zone model reported with this ? What about write
granularity (for SMR HDDs) too ?

So something like:

struct ublk_param_zoned {
	__u32	model;
	__u32	write_granularity;
	__u32	max_open_zones;
	__u32	max_active_zones;
	__u32	max_append_size;
	__u8	reserved[12];
};

looks better to me. Note sure about the need for model and write_granularity
here though. I did not follow zoned ublk patches.


-- 
Damien Le Moal
Western Digital Research