linux-kernel - RE: [PATCH v2] scsi: ufs: Zero utp_upiu_req at the beginning of each command

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:
 <DM6PR04MB6575FD6BCCDC3327885E6DF6FC632@DM6PR04MB6575.namprd04.prod.outlook.com>
Date: Thu, 19 Sep 2024 06:59:35 +0000
From: Avri Altman <Avri.Altman@....com>
To: Bart Van Assche <bvanassche@....org>, "Martin K . Petersen"
	<martin.petersen@...cle.com>
CC: "linux-scsi@...r.kernel.org" <linux-scsi@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: RE: [PATCH v2] scsi: ufs: Zero utp_upiu_req at the beginning of each
 command

> On 9/17/24 11:03 PM, Avri Altman wrote:
> > My proposal is making 4 changes, attending the 5 upiu types:
> >   1) Zero query upiu and nop upiu in ufshcd_compose_devman_upiu
> >   2) zero command upiu in ufshcd_comp_scsi_upiu
> >   3) zero raw query upiu in ufshcd_issue_devman_upiu_cmd, and
> >   4) zero rpmb extended header (raw command upiu) in
> > ufshcd_advanced_rpmb_req_handler
> >
> > Your proposal is making 3 changes:
> >   - zero query upiu in ufshcd_prepare_utp_query_req_upiu
> >   - zero nop upiu in ufshcd_prepare_utp_nop_upiu
> >   - zero command upiu in ufshcd_prepare_utp_scsi_cmd_upiu And you
> > haven't zero the raw query upiu nor the rpmb extended header .
>   Hi Avri,
> 
> Would it be possible to combine our patches in such a way that all UPIU types are
> covered and such that the memset() calls for query, nop and command UPIUs
> occur in the functions that initialize *ucd_req_ptr->header?
I guess so.
One approach is to use the fact that all prep instances calls ufshcd_prepare_req_desc_hdr  which instantiate the UTRD header.
Although it reduces the memset() calls to a single call,
this option is less good IMO because the UTRD header shouldn't have anything to do with the upiu itself.

Alternatively, we can move it to ufshcd_setup_dev_cmd in which we'll end up with 2 calls:
One for command upiu in ufshcd_comp_scsi_upiu and one for all others in ufshcd_setup_dev_cmd,
Which makes more sense to me.

Thanks,
Avri
> 
> Thanks,
> 
> Bart.