Date:   Thu, 6 Jan 2022 12:00:26 -0600
From:   Bjorn Helgaas <helgaas@kernel.org>
To:     Pali =?iso-8859-1?Q?Roh=E1r?= <pali@kernel.org>
Cc:     Ray Jui <ray.jui@broadcom.com>,
        Roman Bacik <roman.bacik@broadcom.com>,
        Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>,
        Rob Herring <robh@kernel.org>,
        Krzysztof =?utf-8?Q?Wilczy=C5=84ski?= <kw@linux.com>,
        Bjorn Helgaas <bhelgaas@google.com>,
        bcm-kernel-feedback-list@broadcom.com, linux-pci@vger.kernel.org,
        linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] PCI: iproc: Set all 24 bits of PCI class code
Message-ID: <20220106180026.GA295674@bhelgaas>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <20220105181306.mkratasqg36tjf4e@pali>
Precedence: bulk

On Wed, Jan 05, 2022 at 07:13:06PM +0100, Pali Roh?r wrote:
> On Wednesday 05 January 2022 09:51:48 Ray Jui wrote:
> > On 1/5/2022 1:35 AM, Pali Roh?r wrote:

> > 2. I suppose 'PCI_CLASS_BRIDGE_PCI_NORMAL' is defined in some common PCI
> > header in a separate patch as described in the commit message. Then how
> > come these patches are not constructed with a patch series?
> 
> Yes, PCI_CLASS_BRIDGE_PCI_NORMAL is a new constant for common pci header
> file defined in patch linked in commit message.
> https://lore.kernel.org/linux-pci/20211220145140.31898-1-pali@kernel.org/
> 
> Originally I included this change in v1 of linked patch in December but
> I realized that it does not match standard PCI config space (different
> offset 0x43c vs 0x08 and also different shift 0x8 vs 0x0) and probably
> there is something either incorrect or really non-standard. So later in
> December I dropped iproc_pcie_check_link() change in v2 of the linked
> patch where is introduced PCI_CLASS_BRIDGE_PCI_NORMAL and now sent new
> change for iproc_pcie_check_link() separately.
> 
> Technically, linked patch in commit message is just extracting code into
> the common macros without any functional changed. But change in this
> iproc_pcie_check_link() has also functional change as now also lower 8
> bits of class code are changed. So in my opinion this patch should be
> really separate of linked patch.
> 
> I hope that Lorenzo and Bjorn take patches in correct order...

If patches are not sent together in a series, you can't assume
anything about the order they'll be applied in.  Adding a note about
"this patch depends patch X" helps a little but adds a fair amount of
friction to the process.

Bjorn