public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Daniel Mack <daniel@caiaq.de>
To: Dinh.Nguyen@freescale.com
Cc: linux-kernel@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org, linux@arm.linux.org.uk,
	s.hauer@pengutronix.de, valentin.longchamp@epfl.ch,
	grant.likely@secretlab.ca, bryan.wu@canonical.com,
	amit.kucheria@canonical.com, r.herring@freescale.com,
	Jun.Li@freescale.com, xiao-lizhang@freescale.com
Subject: Re: [PATCHv4 2.6.34-rc4 5/7] mxc: Add generic USB HW initialization for MX51
Date: Sat, 17 Apr 2010 19:07:07 +0200	[thread overview]
Message-ID: <20100417170706.GY30801@buzzloop.caiaq.de> (raw)
In-Reply-To: <1271445371-18501-5-git-send-email-Dinh.Nguyen@freescale.com>

On Fri, Apr 16, 2010 at 02:16:09PM -0500, Dinh.Nguyen@freescale.com wrote:
> This patch adds USB HW initializiation code to /plat-mxc/ehci.c.
> 	-Stops and resets USB HW
> 	-Sets some specific PHY settings
> 	-Stop and restart the USB HW.
> Renames mxc_set_usbcontrol to mxc_initialize_usb_hw.
> 
> This patch applies to 2.6.34-rc4.

Yes, the whole series looks _much_ better now, thanks!
Some minor things below.


> Signed-off-by: Dinh Nguyen <Dinh.Nguyen@freescale.com>
> ---
>  arch/arm/plat-mxc/ehci.c                  |  128 ++++++++++++++++++++++++++++-
>  arch/arm/plat-mxc/include/mach/mxc_ehci.h |    2 +-
>  2 files changed, 127 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/arm/plat-mxc/ehci.c b/arch/arm/plat-mxc/ehci.c
> index cb0b638..35ff1c1 100644
> --- a/arch/arm/plat-mxc/ehci.c
> +++ b/arch/arm/plat-mxc/ehci.c
> @@ -1,5 +1,6 @@
>  /*
>   * Copyright (c) 2009 Daniel Mack <daniel@caiaq.de>
> + * Copyright (C) 2010 Freescale Semiconductor, Inc.
>   *
>   * This program is free software; you can redistribute it and/or modify it
>   * under the terms of the GNU General Public License as published by the
> @@ -18,6 +19,7 @@
>  
>  #include <linux/platform_device.h>
>  #include <linux/io.h>
> +#include <linux/delay.h>
>  
>  #include <mach/hardware.h>
>  #include <mach/mxc_ehci.h>
> @@ -50,9 +52,14 @@
>  #define MX35_H1_TLL_BIT		(1 << 5)
>  #define MX35_H1_USBTE_BIT	(1 << 4)
>  
> -int mxc_set_usbcontrol(int port, unsigned int flags)
> +int mxc_intialize_usb_hw(int port, unsigned int flags)

This patch and the next one should be merged into one. The reason is
that you rename a function here and the next patch follows that change.
Especially when bisecting, this can lead to uncompilable checkouts of
the tree when stopping in the middle.

>  {
>  	unsigned int v;
> +	void __iomem *usb_base;
> +	u32 usbotg_base;
> +	u32 usbother_base;
> +	int timeout;
> +	int ret = 0;

As you only use these variable in the MX51 case, ...

>  #ifdef CONFIG_ARCH_MX3
>  	if (cpu_is_mx31()) {
>  		v = readl(MX31_IO_ADDRESS(MX31_OTG_BASE_ADDR +
> @@ -186,9 +193,126 @@ int mxc_set_usbcontrol(int port, unsigned int flags)
>  		return 0;
>  	}
>  #endif /* CONFIG_MACH_MX27 */
> +#ifdef CONFIG_ARCH_MX51
> +	if (cpu_is_mx51()) {

... they can be local -> here.

> +		usb_base = ioremap(MX51_OTG_BASE_ADDR, SZ_4K);
> +
> +		switch (port) {
> +		case 0:	/* OTG port */
> +			usbotg_base = (u32)usb_base + USBOTG_OFFSET;
> +			break;
> +		case 1:	/* Host 1 port */
> +			usbotg_base = (u32)usb_base + USBH1_OFFSET;
> +			break;
> +		default:
> +			printk(KERN_ERR"%s no such port %d\n", __func__, port);
> +			ret = -ENOENT;
> +			goto error;
> +		}
> +		usbother_base = (u32)usb_base + USBOTHER_REGS_OFFSET;

That cast shouldn't be needed.

> +		/* Stop then Reset */
> +		v = __raw_readl(usbotg_base + USBCMD_OFFSET);
> +		v &= ~UCMD_RUN_STOP;
> +		__raw_writel(v, usbotg_base + USBCMD_OFFSET);
> +		timeout = 0x100000;
> +		while (--timeout && __raw_readl(usbotg_base + USBCMD_OFFSET) & UCMD_RUN_STOP)
> +			cpu_relax();

I'm thinking whether this __raw_readl/timeout procedure could be
factored out to some static inline function?

Also, the timeout seem _huge_.

> +		if (!timeout) {
> +			printk(KERN_ERR "%s could not stop usb hardware\n", __func__);
> +			ret = -ETIMEDOUT;
> +			goto error;
> +		}
> +
> +		v = __raw_readl(usbotg_base + USBCMD_OFFSET);
> +		v |= UCMD_RESET;
> +		__raw_writel(v, usbotg_base + USBCMD_OFFSET);
> +		timeout = 0x100000;
> +		while (--timeout && __raw_readl(usbotg_base + USBCMD_OFFSET) & UCMD_RESET)
> +			cpu_relax();
> +		if (!timeout) {
> +			printk(KERN_ERR "%s could not reset usb hardware\n", __func__);
> +			ret = -ETIMEDOUT;
> +			goto error;
> +		}
> +
> +		switch (port) {
> +		case 0:	/*OTG port */
> +			v = __raw_readl(usbother_base + USB_PHY_CTR_FUNC_OFFSET);
> +			v |= USB_UTMI_PHYCTRL_OC_DIS; /* OC is not used */
> +			__raw_writel(v, usbother_base + USB_PHY_CTR_FUNC_OFFSET);
> +
> +			v = __raw_readl(usbother_base + USBCTRL_OFFSET);
> +			v &= ~(UCTRL_OPM | UCTRL_OWIE);/* OTG wakeup/power mask disable */
> +			__raw_writel(v, usbother_base + USBCTRL_OFFSET);
> +
> +			/* Set the PHY clock to 19.2MHz */
> +			v = __raw_readl(usbother_base + USB_PHY_CTR_FUNC2_OFFSET);
> +			v &= ~USB_UTMI_PHYCTRL2_PLLDIV_MASK;
> +			v |= 0x01;
> +			__raw_writel(v, usbother_base + USB_PHY_CTR_FUNC2_OFFSET);
> +			break;
> +		case 1:	/* Host 1 */
> +			/*Host ULPI */
> +			v = __raw_readl(usbother_base + USB_CTRL_1_OFFSET);
> +			__raw_writel(v | USB_CTRL_UH1_EXT_CLK_EN, usbother_base + USB_CTRL_1_OFFSET);
> +
> +			v = __raw_readl(usbother_base + USBCTRL_OFFSET);
> +			v &= ~(UCTRL_H1WIE | UCTRL_H1UIE);/* HOST1 wakeup/ULPI intr disable */
> +			v |= UCTRL_H1PM; /* HOST1 power mask */
> +			__raw_writel(v, usbother_base + USBCTRL_OFFSET);
> +
> +			v = __raw_readl(usbother_base + USB_PHY_CTR_FUNC_OFFSET);
> +			v |= USB_UH1_OC_DIS; /* OC is not used */
> +			__raw_writel(v, usbother_base + USB_PHY_CTR_FUNC_OFFSET);
> +
> +			v = __raw_readl(usbotg_base + USBCMD_OFFSET);
> +			/* Interrupt Threshold Control:Immediate (no threshold) */
> +			v &= UCMD_ITC_NO_THRESHOLD;
> +			__raw_writel(v, usbotg_base + USBCMD_OFFSET);
> +			break;
> +		}
> +
> +		/* need to reset the controller here so that the ID pin
> +		 * is correctly detected.
> +		 */
> +		/* Stop then Reset */
> +		v = __raw_readl(usbotg_base + USBCMD_OFFSET);
> +		v &= ~UCMD_RUN_STOP;
> +		__raw_writel(v, usbotg_base + USBCMD_OFFSET);
> +		timeout = 0x100000;
> +		while (--timeout && __raw_readl(usbotg_base + USBCMD_OFFSET) & UCMD_RUN_STOP)
> +			cpu_relax();
> +		if (!timeout) {
> +			printk(KERN_ERR "%s could not stop usb hardware\n", __func__);
> +			ret = -ETIMEDOUT;
> +			goto error;
> +		}
> +
> +		v = __raw_readl(usbotg_base + USBCMD_OFFSET);
> +		v |= UCMD_RESET;
> +		__raw_writel(v, usbotg_base + USBCMD_OFFSET);
> +		timeout = 0x100000;
> +		while (--timeout && __raw_readl(usbotg_base + USBCMD_OFFSET) & UCMD_RESET)
> +			cpu_relax();
> +		if (!timeout) {
> +			printk(KERN_ERR "%s could not reset usb hardware\n", __func__);
> +			ret = -ETIMEDOUT;
> +			goto error;
> +		}
> +
> +		/* allow controller to reset, and leave time for
> +		 * the ULPI transceiver to reset too.
> +		 */
> +		msleep(100);
> +error:
> +		iounmap(usb_base);
> +		return ret;
> +	}

And especially because of the jump label, I think that function could be
split now into individual chunks for the different platforms now.
But that can be done later. I'll send in a patch once things are merged.

Thanks,
Daniel


> +#endif
>  	printk(KERN_WARNING
>  		"%s() unable to setup USBCONTROL for this CPU\n", __func__);
>  	return -EINVAL;
>  }
> -EXPORT_SYMBOL(mxc_set_usbcontrol);
> +EXPORT_SYMBOL(mxc_intialize_usb_hw);
>  
> diff --git a/arch/arm/plat-mxc/include/mach/mxc_ehci.h b/arch/arm/plat-mxc/include/mach/mxc_ehci.h
> index c5b1e7b..e61feac 100644
> --- a/arch/arm/plat-mxc/include/mach/mxc_ehci.h
> +++ b/arch/arm/plat-mxc/include/mach/mxc_ehci.h
> @@ -83,7 +83,7 @@ struct mxc_usbh_platform_data {
>  	struct otg_transceiver	*otg;
>  };
>  
> -int mxc_set_usbcontrol(int port, unsigned int flags);
> +int mxc_intialize_usb_hw(int port, unsigned int flags);
>  
>  #endif /* __INCLUDE_ASM_ARCH_MXC_EHCI_H */
>  
> -- 
> 1.6.0.4
> 

  parent reply	other threads:[~2010-04-17 17:07 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-04-16 19:16 [PATCHv4 2.6.34-rc4 1/7] mxc: Update GPIO for USB support on Freescale MX51 Babbage HW Dinh.Nguyen
2010-04-16 19:16 ` [PATCHv4 2.6.34-rc4 2/7] mx5: Add USB device definitions for " Dinh.Nguyen
2010-04-16 19:16   ` [PATCHv4 2.6.34-rc4 3/7] mxc: Add new defines for USB HW register bits on Freescale SoCs Dinh.Nguyen
2010-04-16 19:16     ` [PATCHv4 2.6.34-rc4 4/7] mx5: Enable board specific functions for enabling USB host on Babbage Dinh.Nguyen
2010-04-16 19:16       ` [PATCHv4 2.6.34-rc4 5/7] mxc: Add generic USB HW initialization for MX51 Dinh.Nguyen
2010-04-16 19:16         ` [PATCHv4 2.6.34-rc4 6/7] mxc-usb: Add call to initialize MXC USB HW registers Dinh.Nguyen
2010-04-16 19:16           ` [PATCHv4 2.6.34-rc4 7/7] mx5: Add USB to Freescale MX51 defconfig Dinh.Nguyen
2010-04-17 17:07         ` Daniel Mack [this message]
2010-04-19  1:03           ` [PATCHv4 2.6.34-rc4 5/7] mxc: Add generic USB HW initialization for MX51 Amit Kucheria
2010-04-19  7:38             ` Daniel Mack
2010-04-18  2:07         ` Sascha Hauer
2010-04-19 15:40           ` [PATCHv4 2.6.34-rc4 5/7] mxc: Add generic USB HWinitialization " Nguyen Dinh-R00091
2010-04-19 15:58             ` Sascha Hauer
2010-04-19 16:59               ` Daniel Mack
2010-04-19 17:04                 ` Daniel Mack
2010-04-19  0:32   ` [PATCHv4 2.6.34-rc4 2/7] mx5: Add USB device definitions for Freescale MX51 Babbage HW Amit Kucheria
2010-04-19 14:48     ` Nguyen Dinh-R00091

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20100417170706.GY30801@buzzloop.caiaq.de \
    --to=daniel@caiaq.de \
    --cc=Dinh.Nguyen@freescale.com \
    --cc=Jun.Li@freescale.com \
    --cc=amit.kucheria@canonical.com \
    --cc=bryan.wu@canonical.com \
    --cc=grant.likely@secretlab.ca \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux@arm.linux.org.uk \
    --cc=r.herring@freescale.com \
    --cc=s.hauer@pengutronix.de \
    --cc=valentin.longchamp@epfl.ch \
    --cc=xiao-lizhang@freescale.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox