From mboxrd@z Thu Jan 1 00:00:00 1970 From: Simon Horman Date: Thu, 19 Dec 2013 09:04:20 +0000 Subject: Re: [PATCH] ARM: shmobile: koelsch: Conditionally select MICREL_PHY Message-Id: <20131219090420.GD12239@verge.net.au> List-Id: References: <1387153872-8387-1-git-send-email-horms+renesas@verge.net.au> <52AEFF56.5010602@cogentembedded.com> <52AF8D57.7010705@cogentembedded.com> <20131217003107.GF27171@verge.net.au> <20131218005326.GE18992@verge.net.au> In-Reply-To: <20131218005326.GE18992@verge.net.au> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: linux-arm-kernel@lists.infradead.org On Wed, Dec 18, 2013 at 09:53:26AM +0900, Simon Horman wrote: > On Tue, Dec 17, 2013 at 11:10:15AM +0100, Geert Uytterhoeven wrote: > > On Tue, Dec 17, 2013 at 1:31 AM, Simon Horman wrote: > > > On Tue, Dec 17, 2013 at 02:31:35AM +0300, Sergei Shtylyov wrote: > > >> On 12/16/2013 04:25 PM, I wrote: > > >> >>The koelsch board uses has an SH ethernet controller which uses a Micrel > > >> >>phy. Select MICREL_PHY for koelsch if SH_ETH is enabled to make use of the > > >> >>Micrel-specific phy driver rather than relying on the generic phy driver. > > >> > > >> > Frankly speaking, I'm seeing no advantages to doing that. > > >> > > >> Although wait, Geert has reported the absence of Rx FIFO overflow > > >> messages with this PHY driver, IIRC. > > > > I spoke too soon: apparently the statistical relevance of my reporting was > > too low. I still see them from time to time, but I think less than before (no > > numbers available). > > > > >> >>Signed-off-by: Simon Horman > > >> > > >> > Analogous changes are needed for BOCK-W and Lager boards since they all > > >> >use the same KSZ8041RNLI PHY chip. I was going to submit such changes with the > > >> >PHY IRQ patches but since no IRQ is generated still, this got delayed... > > >> > > >> Only I was going to do it via defconfigs... > > > > > > That was my initial approach too. However I have been convinced > > > that Kconfig is a better way forwards. > > > > Kconfig has the advantage that the minimal defconfigs will be smaller, so > > less churn there. > > However, it does mean you cannot disable the option anymore via your .config. > > > > In this case it's a bit difficult to decide, as sh_eth will work both > > with and without > > the specific PHY-driver. > > > > If no users want to disable it, I'd go for Kconfig. > > It is indeed a tricky choice. But I think we should try going for Kconfig. I have decided to queue this up. From mboxrd@z Thu Jan 1 00:00:00 1970 From: horms@verge.net.au (Simon Horman) Date: Thu, 19 Dec 2013 18:04:20 +0900 Subject: [PATCH] ARM: shmobile: koelsch: Conditionally select MICREL_PHY In-Reply-To: <20131218005326.GE18992@verge.net.au> References: <1387153872-8387-1-git-send-email-horms+renesas@verge.net.au> <52AEFF56.5010602@cogentembedded.com> <52AF8D57.7010705@cogentembedded.com> <20131217003107.GF27171@verge.net.au> <20131218005326.GE18992@verge.net.au> Message-ID: <20131219090420.GD12239@verge.net.au> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Wed, Dec 18, 2013 at 09:53:26AM +0900, Simon Horman wrote: > On Tue, Dec 17, 2013 at 11:10:15AM +0100, Geert Uytterhoeven wrote: > > On Tue, Dec 17, 2013 at 1:31 AM, Simon Horman wrote: > > > On Tue, Dec 17, 2013 at 02:31:35AM +0300, Sergei Shtylyov wrote: > > >> On 12/16/2013 04:25 PM, I wrote: > > >> >>The koelsch board uses has an SH ethernet controller which uses a Micrel > > >> >>phy. Select MICREL_PHY for koelsch if SH_ETH is enabled to make use of the > > >> >>Micrel-specific phy driver rather than relying on the generic phy driver. > > >> > > >> > Frankly speaking, I'm seeing no advantages to doing that. > > >> > > >> Although wait, Geert has reported the absence of Rx FIFO overflow > > >> messages with this PHY driver, IIRC. > > > > I spoke too soon: apparently the statistical relevance of my reporting was > > too low. I still see them from time to time, but I think less than before (no > > numbers available). > > > > >> >>Signed-off-by: Simon Horman > > >> > > >> > Analogous changes are needed for BOCK-W and Lager boards since they all > > >> >use the same KSZ8041RNLI PHY chip. I was going to submit such changes with the > > >> >PHY IRQ patches but since no IRQ is generated still, this got delayed... > > >> > > >> Only I was going to do it via defconfigs... > > > > > > That was my initial approach too. However I have been convinced > > > that Kconfig is a better way forwards. > > > > Kconfig has the advantage that the minimal defconfigs will be smaller, so > > less churn there. > > However, it does mean you cannot disable the option anymore via your .config. > > > > In this case it's a bit difficult to decide, as sh_eth will work both > > with and without > > the specific PHY-driver. > > > > If no users want to disable it, I'd go for Kconfig. > > It is indeed a tricky choice. But I think we should try going for Kconfig. I have decided to queue this up.