Hi, On Mon, Jan 14, 2013 at 04:17:35PM +0800, Peter Chen wrote: > On Mon, Jan 14, 2013 at 09:48:58AM +0200, Felipe Balbi wrote: > > Hi, > > > > On Mon, Jan 14, 2013 at 03:18:17PM +0800, Peter Chen wrote: > > > It changes the driver to use platform_device_id rather than cpu_is_xxx > > > to determine the SoC type, and updates the platform code accordingly. > > > > > > Compile ok at imx_v6_v7_defconfig with CONFIG_USB_FSL_USB2 enable. > > > Tested at mx51 bbg board, it works ok after enable phy clock > > > (Need another patch to fix this problem) > > > > > > Signed-off-by: Peter Chen <peter.chen@xxxxxxxxxxxxx> > > > > not good for -rc. You have to break this down as you're solving at least > > three different problems with this patch. > > Felipe, all my changes are for this problem, these are fix build error and > let it work. > > arch/arm/mach-imx/clk-imx25.c | 6 +- > arch/arm/mach-imx/clk-imx27.c | 6 +- > arch/arm/mach-imx/clk-imx31.c | 6 +- > arch/arm/mach-imx/clk-imx35.c | 6 +- > arch/arm/mach-imx/clk-imx51-imx53.c | 6 +- > > As we change the connection-id, we need to change clock file or > the devm_clk_get will be failed. right right, that's ok. > arch/arm/mach-imx/devices/devices-common.h | 1 + > arch/arm/mach-imx/devices/platform-fsl-usb2-udc.c | 15 +++--- > > We need to differentiate SoCs, so I use platform_device_id to > instead of cpu_ix_xxx(), this is for machine file change fair enough. > drivers/usb/gadget/fsl_mxc_udc.c | 23 +++++---- > drivers/usb/gadget/fsl_udc_core.c | 52 +++++++++++++------- > drivers/usb/gadget/fsl_usb2_udc.h | 13 ++++-- > include/linux/fsl_devices.h | 8 +++ > > Need to get platform_device_id at driver, and replace the cpu_is_xxx to > platform_device_id. Meanwhile, needs a solution for replace > MX35_IO_ADDRESS. ok, here we go: You just listed to me three different fixes and each fix should be on a separate patch. Meaning that you should have one single patch to convert MX35_IO_ADDRESS() into ioremap(), another patch should be removing cpu_is_xxx() and the third one fixing connection-id. As you might remember, we want patches to be self-contained logical units and your patch contains 3 of such logical units. > > Besides there are parts which are not related to the build break. > > I can split this patch to two patches, one is driver part, the other is > machine part. It can also fix build break, but it needs two > patches to let the udc work. you will need as many patches as necessary, but they need to the broken correclty, the way you suggested above is wrong. -- balbi
Attachment:
signature.asc
Description: Digital signature