NVMe: Fix partition detection issue during hotplug

[Shiro Itou 伊東]

Hi Kieth,

My mistake, sorry. I will use text mail.

Some our SCSI engineers and some customer suggest SCSI driver work the same way for the use case as it is with patch I sent.

Thank you, please advice right solution as you suggest.

Thank You,
Shiro

Date: Tue, 5 Aug 2014 23:47:04 -0600
From: keith.busch at intel.com
To: shiro.itou at outlook.com
Subject: Re: NVMe: Fix partition detection issue during hotplug
CC: keith.busch at intel.com; willy at linux.intel.com; linux-nvme at lists.infradead.org

Hi Shiro,

Your email will not show up on the mailing list if you use html instead
of plain text.

Regarding your use case, I think you are inviting data loss among other
problems by hot removing a mounted drive no matter what the driver is
doing with the device minors, but I'll see if there is a way we can keep
dynamic partitions and get the same functionality you're looking for.

On Tue, 5 Aug 2014, Shiro Itou 伊東 wrote:
> Hi All,
> 
> In out setup we see nighlty test fail when device mount
> and hot-remove and hot-plug sequence. read-write command
> fail to do on device after hotplug, because partition do
> not come visible.
>  
> The test sequence create,
>  a. multi-partition on device
>  b. and mount
>  c. stress test from mount folder
>  d. and hot-remove
>   e. hot-add
>   
>   now after 'step-e' read-write fail on device. Here device
>   partition do not come visible.
>   
>   This patch fix the issue by revert:469071a37afc8a627b6b2ddf29db0a097d864845
>   and minor change.
> 
> Thank You,
> Shiro
> 
>

_______________________________________________ Linux-nvme mailing list Linux-nvme at lists.infradead.org http://lists.infradead.org/mailman/listinfo/linux-nvme