linux-kernel - Re: [PATCH v16 6/9] nvmet-passthru: Add passthru code to process commands

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: <d5c64c38-066c-013c-7a79-46230e439c5e@deltatee.com>
Date:   Fri, 24 Jul 2020 13:40:20 -0600
From:   Logan Gunthorpe <logang@...tatee.com>
To:     Keith Busch <kbusch@...nel.org>
Cc:     linux-kernel@...r.kernel.org, linux-nvme@...ts.infradead.org,
        Christoph Hellwig <hch@....de>,
        Sagi Grimberg <sagi@...mberg.me>, Jens Axboe <axboe@...com>,
        Chaitanya Kulkarni <Chaitanya.Kulkarni@....com>,
        Max Gurtovoy <maxg@...lanox.com>,
        Stephen Bates <sbates@...thlin.com>
Subject: Re: [PATCH v16 6/9] nvmet-passthru: Add passthru code to process
 commands



On 2020-07-24 1:33 p.m., Keith Busch wrote:
> On Fri, Jul 24, 2020 at 11:25:17AM -0600, Logan Gunthorpe wrote:
>> +	/*
>> +	 * The passthru NVMe driver may have a limit on the number of segments
>> +	 * which depends on the host's memory fragementation. To solve this,
>> +	 * ensure mdts is limitted to the pages equal to the number of
> 
>                           limited
> 
>> +	/* don't support fuse commands */
>> +	id->fuses = 0;
> 
> If a host were to set a fuse, the target should return an Invalid Field
> error. Just to future-proof, rejecting commands with any flags set
> (other than SGL, which you handled in patch 1/9) is probably what should
> happen, like:

> 
>> +u16 nvmet_parse_passthru_io_cmd(struct nvmet_req *req)
>> +{
> 
> 	if (req->cmd->common.flags & ~NVME_CMD_SGL_ALL)
> 		return NVME_SC_INVALID_FIELD;


Yes, this seems like a good idea. I can add it.

> Or maybe we could obviate the need for 1/9 with something like:
> 
> 	req->cmd->common.flags &= ~NVME_CMD_SGL_ALL;
> 	if (req->cmd->common.flags)
> 		return NVME_SC_INVALID_FIELD;

We used to clear the SGL flags in the target passthru code but Christoph
asked that it be done in the host code, hence patch 1/9.

Logan