From: Can Guo <cang@codeaurora.org>
To: Bean Huo <huobean@gmail.com>
Cc: Stanley Chu <stanley.chu@mediatek.com>,
alim.akhtar@samsung.com, avri.altman@wdc.com,
asutoshd@codeaurora.org, jejb@linux.ibm.com,
martin.petersen@oracle.com, beanhuo@micron.com,
bvanassche@acm.org, tomas.winkler@intel.com,
linux-scsi@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v5 1/7] scsi: ufs: Add "wb_on" sysfs node to control WB on/off
Date: Wed, 23 Dec 2020 09:31:49 +0800 [thread overview]
Message-ID: <862483add1462510b809aee6d3678435@codeaurora.org> (raw)
In-Reply-To: <28211d08700d1e4876a9aea342e8fcb79534cd2c.camel@gmail.com>
On 2020-12-23 06:11, Bean Huo wrote:
> On Tue, 2020-12-22 at 21:57 +0100, Bean Huo wrote:
>> > > May this operation race with UFS shutdown flow?
>> > >
>> > > To be more clear, ufshcd_wb_ctrl() here may be executed after
>> > > host
>> > > clock
>> > > is disabled by shutdown flow?
>> > >
>> > > If yes, we need to avoid it.
>> >
>> > I have the same doubt - can user still access sysfs nodes after
>> > system
>> > starts to run shutdown routines? If yes, then we need to remove all
>> > UFS
>> > sysfs nodes in ufshcd_shutdown().
>> >
>>
>> No, we shouldn't do in this way, user space complains this. I think
>> the nodes in the sysfs can be shileded write, but the nodes shouldn't
>> be flash of its presence frequently.
>>
>> Thanks,
>> Bean
>>
>>
>> > Thanks,
>> >
>> > Can Guo.
>
>
> Hi Can
> Got your point, you don't want user space to interrupt UFS by sysyfs if
> UFS is in power down mode. if this is true, insteading of removing all
> sysfs node in ufshcd_shutdown, maybe just add this checkup before
> accessing UFS device descriptors/flag/attributes/LU:
>
> for example, for the device descriptor:
>
>
> diff --git a/drivers/scsi/ufs/ufs-sysfs.c b/drivers/scsi/ufs/ufs-
> sysfs.c
> index b3bf7fca00e5..881fe1c24a9f 100644
> --- a/drivers/scsi/ufs/ufs-sysfs.c
> +++ b/drivers/scsi/ufs/ufs-sysfs.c
> @@ -262,6 +262,9 @@ static ssize_t ufs_sysfs_read_desc_param(struct
> ufs_hba *hba,
> u8 desc_buf[8] = {0};
> int ret;
>
> + if (!ufshcd_is_ufs_dev_active(hba) ||
> !ufshcd_is_link_active(hba))
> + return -EACCES;
> +
>
> Bean
First of all, this check is not helping at all, during
ufshcd_shutdown(),
both the link and dev can be active for a moment, so this check is not
helping the race condition. And assume you come up with a better check,
you want to add the check everywhere? You must have noticed the fix to
the func ufshcd_clk_gate_enable_store() from Jaegeuk Kim. So, don't
expect any luck from user space, so long the sysfs nodes are available,
users can grasp them and even stress them just for testing purposes.
I don't see why removing the sysfs nodes during ufshcd_shutdown() is a
concern to customer - we should do whatever is needed to protect LLD
contexts from user space intervene. What do you think?
Can Guo.
Thanks,
Can Guo.
next prev parent reply other threads:[~2020-12-23 1:32 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-12-15 23:05 [PATCH v5 0/7] Several changes for UFS WriteBooster Bean Huo
2020-12-15 23:05 ` [PATCH v5 1/7] scsi: ufs: Add "wb_on" sysfs node to control WB on/off Bean Huo
2020-12-22 6:08 ` Stanley Chu
2020-12-22 6:12 ` Can Guo
2020-12-22 20:57 ` Bean Huo
2020-12-22 22:11 ` Bean Huo
2020-12-23 1:31 ` Can Guo [this message]
2020-12-23 8:28 ` Bean Huo
2020-12-23 8:30 ` Bean Huo
2020-12-23 9:11 ` Bean Huo
2020-12-22 20:50 ` Bean Huo
2020-12-22 21:08 ` Bean Huo
2020-12-23 21:52 ` Bean Huo
2020-12-15 23:05 ` [PATCH v5 2/7] docs: ABI: Add wb_on documentation for UFS sysfs Bean Huo
2020-12-15 23:05 ` [PATCH v5 3/7] scsi: ufs: Changes comment in the function ufshcd_wb_probe() Bean Huo
2020-12-15 23:05 ` [PATCH v5 4/7] scsi: ufs: Remove two WB related fields from struct ufs_dev_info Bean Huo
2020-12-16 4:15 ` Stanley Chu
2020-12-15 23:05 ` [PATCH v5 5/7] scsi: ufs: Group UFS WB related flags to " Bean Huo
2020-12-16 4:16 ` Stanley Chu
2020-12-22 6:14 ` Can Guo
2020-12-15 23:05 ` [PATCH v5 6/7] scsi: ufs: Cleanup WB buffer flush toggle implementation Bean Huo
2020-12-16 4:17 ` Stanley Chu
2020-12-22 6:14 ` Can Guo
2020-12-15 23:05 ` [PATCH v5 7/7] scsi: ufs: Keep device active mode only fWriteBoosterBufferFlushDuringHibernate == 1 Bean Huo
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=862483add1462510b809aee6d3678435@codeaurora.org \
--to=cang@codeaurora.org \
--cc=alim.akhtar@samsung.com \
--cc=asutoshd@codeaurora.org \
--cc=avri.altman@wdc.com \
--cc=beanhuo@micron.com \
--cc=bvanassche@acm.org \
--cc=huobean@gmail.com \
--cc=jejb@linux.ibm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-scsi@vger.kernel.org \
--cc=martin.petersen@oracle.com \
--cc=stanley.chu@mediatek.com \
--cc=tomas.winkler@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).