Dear Marek Vasut,
Dear Benoît Thébaudeau,
Use clock gate definitions having names showing clearly the gated clock instead of names giving only a register field index.
This change reveals that the USB PHY clock functions were broken on i.MX51, so this patch fixes those too.
Expanding CC /wrt this mx51 usb biz.
Signed-off-by: Benoît Thébaudeau benoit.thebaudeau@advansee.com Cc: Stefano Babic sbabic@denx.de Cc: Marek Vasut marex@denx.de
.../arch/arm/cpu/armv7/mx5/clock.c | 39 ++++- .../arch/arm/include/asm/arch-mx5/clock.h | 7 + .../arch/arm/include/asm/arch-mx5/crm_regs.h | 158 +++++++++++++++++++- .../drivers/usb/host/ehci-mx5.c | 5 + .../drivers/video/ipu_common.c | 2 +- 5 files changed, 200 insertions(+), 11 deletions(-)
diff --git u-boot-4d3c95f.orig/arch/arm/cpu/armv7/mx5/clock.c u-boot-4d3c95f/arch/arm/cpu/armv7/mx5/clock.c index c67c3cf..9b083c0 100644 --- u-boot-4d3c95f.orig/arch/arm/cpu/armv7/mx5/clock.c +++ u-boot-4d3c95f/arch/arm/cpu/armv7/mx5/clock.c @@ -110,10 +110,11 @@ void enable_usboh3_clk(unsigned char enable) unsigned int reg;
reg = readl(&mxc_ccm->CCGR2);
- reg &= ~(MXC_CCM_CCGR_CG_MASK <<
MXC_CCM_CCGR2_USBOH3_60M_OFFSET);
What's this addition?
It's to clear this 2-bit bit-field before setting it.
if (enable)
reg |= 1 << MXC_CCM_CCGR2_CG14_OFFSET;
reg |= MXC_CCM_CCGR_CG_ON << MXC_CCM_CCGR2_USBOH3_60M_OFFSET;Can't you make it like ... MXC_CCM_CCGR2_USBOH3_60M(<on/off>) ... which would emit the correct bit? So you'd wrap the bitshift into it as well ... also, using clrsetbits_le32() won't hurt here.
It's only to be consistent with the existing code. A cosmetic patch could be applied after that for the whole file.
else
reg &= ~(1 << MXC_CCM_CCGR2_CG14_OFFSET);
reg |= MXC_CCM_CCGR_CG_OFF << MXC_CCM_CCGR2_USBOH3_60M_OFFSET;}
@@ -137,6 +138,29 @@ int enable_i2c_clk(unsigned char enable, unsigned i2c_num) } #endif
+#if defined(CONFIG_MX51) +void set_usb_phy_clk(void) +{
- unsigned int reg;
- reg = readl(&mxc_ccm->cscmr1);
- reg &= ~MXC_CCM_CSCMR1_USB_PHY_CLK_SEL;
- writel(reg, &mxc_ccm->cscmr1);
clrbits_le32() etc., please fix globally.
Ditto.
This doesn't even fit into this patch, so please split away. [...]
These cosmetic changes are off topic for my series. I don't have much time to work on that. Perhaps this can be done by someone else later.
Best regards, Benoît