linux-kernel - Re: [PATCH v2 06/10] nvme/core: add mdev interfaces

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: <20190506125909.GB5288@lst.de>
Date:   Mon, 6 May 2019 14:59:09 +0200
From:   Christoph Hellwig <hch@....de>
To:     Maxim Levitsky <mlevitsk@...hat.com>
Cc:     Christoph Hellwig <hch@....de>, Max Gurtovoy <maxg@...lanox.com>,
        Fam Zheng <fam@...hon.net>, kvm@...r.kernel.org,
        Wolfram Sang <wsa@...-dreams.de>,
        linux-nvme@...ts.infradead.org, linux-kernel@...r.kernel.org,
        Keith Busch <keith.busch@...el.com>,
        Kirti Wankhede <kwankhede@...dia.com>,
        Mauro Carvalho Chehab <mchehab+samsung@...nel.org>,
        "Paul E . McKenney" <paulmck@...ux.ibm.com>,
        Sagi Grimberg <sagi@...mberg.me>,
        Christoph Hellwig <hch@...radead.org>,
        Liang Cunming <cunming.liang@...el.com>,
        Jens Axboe <axboe@...com>,
        Alex Williamson <alex.williamson@...hat.com>,
        John Ferlan <jferlan@...hat.com>,
        Liu Changpeng <changpeng.liu@...el.com>,
        Jens Axboe <axboe@...nel.dk>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Nicolas Ferre <nicolas.ferre@...rochip.com>,
        Paolo Bonzini <pbonzini@...hat.com>,
        Amnon Ilan <ailan@...hat.com>,
        "David S . Miller" <davem@...emloft.net>
Subject: Re: [PATCH v2 06/10] nvme/core: add mdev interfaces

On Mon, May 06, 2019 at 11:31:27AM +0300, Maxim Levitsky wrote:
> Why are you saying that? I actualy prefer to use a sepearate queue per software
> nvme controller, tat because of lower overhead (about half than going through
> the block layer) and it better at QoS as the separate queue (or even few queues
> if needed) will give the guest a mostly guaranteed slice of the bandwidth of the
> device.

The downside is that your create tons of crap code in the core nvme driver
for your specific use case that no one cares about.  Which is why it is
completely unacceptable.  If you want things to go fast make the block
layer go fast, don't add your very special bypass.