From mboxrd@z Thu Jan 1 00:00:00 1970 From: marek.vasut@gmail.com (Marek Vasut) Date: Wed, 25 Aug 2010 15:25:20 +0200 Subject: [PATCH 2/3] [ARM] pxa: MFU pin configuration to support Fast ethernet. In-Reply-To: References: <1282727921.6968.99.camel@pe-lt522.marvell.com> <201008251511.02665.marek.vasut@gmail.com> Message-ID: <201008251525.20575.marek.vasut@gmail.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Dne St 25. srpna 2010 15:20:15 Eric Miao napsal(a): > On Wed, Aug 25, 2010 at 9:11 PM, Marek Vasut wrote: > > Dne St 25. srpna 2010 11:18:41 Sachin Sanap napsal(a): > >> Signed-off-by: Sachin Sanap > >> --- > >> arch/arm/mach-mmp/aspenite.c | 18 ++++++++++++++++++ > >> arch/arm/mach-mmp/include/mach/mfp-pxa168.h | 19 +++++++++++++++++++ > >> 2 files changed, 37 insertions(+), 0 deletions(-) > >> > >> diff --git a/arch/arm/mach-mmp/aspenite.c b/arch/arm/mach-mmp/aspenite.c > >> index d19c26c..a235551 100644 > >> --- a/arch/arm/mach-mmp/aspenite.c > >> +++ b/arch/arm/mach-mmp/aspenite.c > >> @@ -74,6 +74,24 @@ static unsigned long common_pin_config[] __initdata = > >> { GPIO105_CI2C_SDA, > >> GPIO106_CI2C_SCL, > >> > >> + /* MFU */ > >> + GPIO86_TX_CLK, > >> + GPIO87_TX_EN, > >> + GPIO88_TX_DQ3, > >> + GPIO89_TX_DQ2, > >> + GPIO90_TX_DQ1, > >> + GPIO91_TX_DQ0, > >> + GPIO92_MII_CRS, > >> + GPIO93_MII_COL, > >> + GPIO94_RX_CLK, > >> + GPIO95_RX_ER, > >> + GPIO96_RX_DQ3, > >> + GPIO97_RX_DQ2, > >> + GPIO98_RX_DQ1, > >> + GPIO99_RX_DQ0, > >> + GPIO100_MII_MDC, > >> + GPIO101_MII_MDIO, > >> + GPIO103_RX_DV, > >> }; > > > > That's for ethernet, right ? Can you take aspenite apart (aka. remove the > > ethernet chip or whatnot) ? > > > > Maybe if you'd go the way I outlined in the previous mail commenting on > > your 1/3 patch and apply that approach on the whole aspenite, then > > there's a space for further improvement. That is, split the MFP config > > structure into smaller chunks and configure the pins only in case that > > particular device is enabled in kernel. See colibri_pxa320 for reference > > again. > > Well, if it's not separable, and that those pins are not able to be used > as other functions, I'd prefer this being in one consistent array actually. Hm, there is no public documentation for PXA168 (as always in marvell case), right ? It's hard to know if those pins can be used for something else or review the code at all :-(