Re: [PATCH 0/2] PCI: aardvark: Fix comphy with old ATF

From: Pali Rohár
Date: Fri Oct 02 2020 - 10:52:42 EST


On Friday 02 October 2020 15:38:51 Lorenzo Pieralisi wrote:
> On Fri, Oct 02, 2020 at 04:26:16PM +0200, Pali Rohár wrote:
> > On Friday 02 October 2020 14:37:13 Lorenzo Pieralisi wrote:
> > > On Wed, Sep 02, 2020 at 04:43:42PM +0200, Pali Rohár wrote:
> > > > This patch series fixes regression introduced in commit 366697018c9a
> > > > ("PCI: aardvark: Add PHY support") which caused aardvark driver
> > > > initialization failure on EspressoBin board with factory version of
> > > > Arm Trusted Firmware provided by Marvell.
> > > >
> > > > Second patch depends on the first patch, so please add appropriate
> > > > Fixes/Cc:stable@ tags to have both patches correctly backported to
> > > > stable kernels.
> > > >
> > > > I have tested both patches with Marvell ATF firmware ebin-17.10-uart.zip
> > > > and with upstream ATF+uboot and aardvark was initialized successfully.
> > > > Without this patch series on ebin-17.10-uart.zip aardvark initialization
> > > > failed.
> > > >
> > > > Pali Rohár (2):
> > > > phy: marvell: comphy: Convert internal SMCC firmware return codes to
> > > > errno
> > > > PCI: aardvark: Fix initialization with old Marvell's Arm Trusted
> > > > Firmware
> > > >
> > > > drivers/pci/controller/pci-aardvark.c | 4 +++-
> > > > drivers/phy/marvell/phy-mvebu-a3700-comphy.c | 14 +++++++++++---
> > > > drivers/phy/marvell/phy-mvebu-cp110-comphy.c | 14 +++++++++++---
> > > > 3 files changed, 25 insertions(+), 7 deletions(-)
> > >
> > > Applied to pci/aardvark (both patches), thanks.
> >
> > Ok! For second patch would be needed to put CC:stable line with
> > specifying prerequisite of first patch as written in document:
> >
> > https://www.kernel.org/doc/html/latest/process/stable-kernel-rules.html
> >
> > If I understood it correctly, second patch needs following line:
> >
> > Cc: <stable@xxxxxxxxxxxxxxx> # <commit_id_of_first_path>: comphy: errno return codes
> >
> > where <commit_id_of_first_path> is commit id of PATCH 1/2.
> >
> > (correct me if I'm wrong)
> >
> > Now when first commit has commit id, could you update second commit to
> > include this information about prerequisite?
>
> No I can't because they will be merged at the same time.

And it is a problem? Git commit id of first patch would not be changed,
so referencing to it is now possible from second commit (unless you do
rebasing).

> What we can do is mark the second patch for stable

This is done by adding "Cc: <stable@xxxxxxxxxxxxxxx>" line to second
patch which I suggested, right?

> and during the stable review
> ask to pull patch (1). Or better you shall send both patches to stable
> kernels when they hit Linus' tree.
>
> Lorenzo