Re: [PATCH] i40e: fix the wrong PTP frequency calculation

[Greg KH <gregkh@linuxfoundation.org>]

On Tue, Sep 26, 2023 at 09:54:29AM +0800, Yajun Deng wrote:
> 
> On 2023/9/26 07:59, Tony Nguyen wrote:
> > On 9/25/2023 12:55 AM, Yajun Deng wrote:
> > > 
> > > On 2023/6/28 04:20, Jacob Keller wrote:
> > > > 
> > > > On 6/26/2023 7:26 PM, Yajun Deng wrote:
> > > > > The new adjustment should be based on the base frequency, not the
> > > > > I40E_PTP_40GB_INCVAL in i40e_ptp_adjfine().
> > > > > 
> > > > > This issue was introduced in commit 3626a690b717 ("i40e: use
> > > > > mul_u64_u64_div_u64 for PTP frequency calculation"), and was fixed in
> > > > > commit 1060707e3809 ("ptp: introduce helpers to adjust by scaled
> > > > > parts per million"). However the latter is a new feature and
> > > > > hasn't been
> > > > > backported to the stable releases.
> > > > > 
> > > > > This issue affects both v6.0 and v6.1 versions, and the v6.1
> > > > > version is
> > > > > an LTS version.
> > > > > 
> > 
> > ...
> > 
> > > > 
> > > > Thanks for finding and fixing this mistake. I think its the
> > > > simplest fix
> > > > to get into the stable kernel that are broken, since taking the
> > > > adjust_by_scaled_ppm version would require additional patches.
> > > > 
> > > > Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
> > > > 
> > > Kindly ping...
> > 
> > As this patch looks to be for stable, you need to follow the process for
> > that. I believe your situation would fall into option 3:
> > https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html#option-3
> > 
> > 
> Yes, it needs an upstream commit ID. But this patch didn't need to apply to
> the upstream.
> 
> As the commit of the patch, the issue was fixed in
> commit 1060707e3809 ("ptp: introduce helpers to adjust by scaled
> parts per million"). However the commit is a new feature and hasn't been
> backported to the stable releases.
> 
> Therefore, the patch does not have an upstream commit ID, and only needs to
> be applied to stable.

That wasn't very obvious to most of us, perhaps resend it and explicitly
ask for acks/reviews so it can be only applied to the 6.1.y tree?

thanks,

greg k-h