Skip to contentSkip to conversation

USB: host: Misc patches to remove hard-coded octeon platform information

PATCH Superseded Andreas Herrmann 7 participants 18 messages Nov 13, 2014 – Nov 25, 2014

Subsystems (4)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover USB: host: Misc patches to remove hard-coded octeon platform information

Hi Alan,

With following patches I want to base octeon ehci/ohci device configuration on device tree information.

I picked up patches that were submitted in May. See
http://marc.info/?l=linux-usb&m=140135823325811&w=2
and http://marc.info/?l=linux-mips&m=140139694721623&w=2

Patch #1 is your "untested preliminary pass" to remove
[oe]hci-octeon drivers.
Patch #2 is the removal of hard-coded platform information (but now
rebased on your patch)
Patch #3 adapts dma_mask for ehci (as used in ehci-octeon)

Overall diffstat is

 arch/mips/cavium-octeon/octeon-platform.c |  380 +++++++++++++++++++++++------
 arch/mips/configs/cavium_octeon_defconfig |    3 +
 drivers/usb/host/Kconfig                  |   18 +-
 drivers/usb/host/Makefile                 |    1 -
 drivers/usb/host/ehci-hcd.c               |    5 -
 drivers/usb/host/ehci-octeon.c            |  188 --------------
 drivers/usb/host/ehci-platform.c          |    4 +-
 drivers/usb/host/octeon2-common.c         |  200 ---------------
 drivers/usb/host/ohci-hcd.c               |    5 -
 drivers/usb/host/ohci-octeon.c            |  202 ---------------
 drivers/usb/host/ohci-platform.c          |    1 +
 include/linux/usb/ehci_pdriver.h          |    1 +
 12 files changed, 330 insertions(+), 678 deletions(-)

Patches are based on v3.18-rc4-65-g2c54396

Comments welcome.

Thanks,

Andreas

#1 lore raw

PATCH 1/3 USB: host: Remove ehci-octeon and ohci-octeon drivers

From: Alan Stern <[email protected]>

From: Alan Stern <[email protected]>

Remove special-purpose octeon drivers and instead use ehci-platform and ohci-platform as suggested with
http://marc.info/?l=linux-mips&m=140139694721623&w=2

[andreas.herrmann:
fixed compile error]

---
 arch/mips/cavium-octeon/octeon-platform.c |  274 ++++++++++++++++++++++++++++-
 arch/mips/configs/cavium_octeon_defconfig |    3 +
 drivers/usb/host/Kconfig                  |   18 +-
 drivers/usb/host/Makefile                 |    1 -
 drivers/usb/host/ehci-hcd.c               |    5 -
 drivers/usb/host/ehci-octeon.c            |  188 --------------------
 drivers/usb/host/octeon2-common.c         |  200 ---------------------
 drivers/usb/host/ohci-hcd.c               |    5 -
 drivers/usb/host/ohci-octeon.c            |  202 ---------------------
 9 files changed, 285 insertions(+), 611 deletions(-)
 delete mode 100644 drivers/usb/host/ehci-octeon.c
 delete mode 100644 drivers/usb/host/octeon2-common.c
 delete mode 100644 drivers/usb/host/ohci-octeon.c
arch/mips/cavium-octeon/octeon-platform.c +272 -2
index 6df0f4d..b67ddf0 100644
@@ -7,22 +7,27 @@
  * Copyright (C) 2008 Wind River Systems
  */
 
+#include <linux/delay.h>
 #include <linux/init.h>
 #include <linux/irq.h>
 #include <linux/i2c.h>
 #include <linux/usb.h>
 #include <linux/dma-mapping.h>
 #include <linux/module.h>
+#include <linux/mutex.h>
 #include <linux/slab.h>
 #include <linux/platform_device.h>
 #include <linux/of_platform.h>
 #include <linux/of_fdt.h>
 #include <linux/libfdt.h>
+#include <linux/usb/ehci_pdriver.h>
+#include <linux/usb/ohci_pdriver.h>
 
 #include <asm/octeon/octeon.h>
 #include <asm/octeon/cvmx-rnm-defs.h>
 #include <asm/octeon/cvmx-helper.h>
 #include <asm/octeon/cvmx-helper-board.h>
+#include <asm/octeon/cvmx-uctlx-defs.h>
 
 /* Octeon Random Number Generator.  */
 static int __init octeon_rng_device_init(void)
@@ -68,6 +73,229 @@ device_initcall(octeon_rng_device_init);
 
 #ifdef CONFIG_USB
 
+static DEFINE_MUTEX(octeon2_usb_clocks_mutex);
+
+static int octeon2_usb_clock_start_cnt;
+
+static void octeon2_usb_clocks_start(void)
+{
+	u64 div;
+	union cvmx_uctlx_if_ena if_ena;
+	union cvmx_uctlx_clk_rst_ctl clk_rst_ctl;
+	union cvmx_uctlx_uphy_ctl_status uphy_ctl_status;
+	union cvmx_uctlx_uphy_portx_ctl_status port_ctl_status;
+	int i;
+	unsigned long io_clk_64_to_ns;
+
+
+	mutex_lock(&octeon2_usb_clocks_mutex);
+
+	octeon2_usb_clock_start_cnt++;
+	if (octeon2_usb_clock_start_cnt != 1)
+		goto exit;
+
+	io_clk_64_to_ns = 64000000000ull / octeon_get_io_clock_rate();
+
+	/*
+	 * Step 1: Wait for voltages stable.  That surely happened
+	 * before starting the kernel.
+	 *
+	 * Step 2: Enable  SCLK of UCTL by writing UCTL0_IF_ENA[EN] = 1
+	 */
+	if_ena.u64 = 0;
+	if_ena.s.en = 1;
+	cvmx_write_csr(CVMX_UCTLX_IF_ENA(0), if_ena.u64);
+
+	/* Step 3: Configure the reference clock, PHY, and HCLK */
+	clk_rst_ctl.u64 = cvmx_read_csr(CVMX_UCTLX_CLK_RST_CTL(0));
+
+	/*
+	 * If the UCTL looks like it has already been started, skip
+	 * the initialization, otherwise bus errors are obtained.
+	 */
+	if (clk_rst_ctl.s.hrst)
+		goto end_clock;
+	/* 3a */
+	clk_rst_ctl.s.p_por = 1;
+	clk_rst_ctl.s.hrst = 0;
+	clk_rst_ctl.s.p_prst = 0;
+	clk_rst_ctl.s.h_clkdiv_rst = 0;
+	clk_rst_ctl.s.o_clkdiv_rst = 0;
+	clk_rst_ctl.s.h_clkdiv_en = 0;
+	clk_rst_ctl.s.o_clkdiv_en = 0;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* 3b */
+	/* 12MHz crystal. */
+	clk_rst_ctl.s.p_refclk_sel = 0;
+	clk_rst_ctl.s.p_refclk_div = 0;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* 3c */
+	div = octeon_get_io_clock_rate() / 130000000ull;
+
+	switch (div) {
+	case 0:
+		div = 1;
+		break;
+	case 1:
+	case 2:
+	case 3:
+	case 4:
+		break;
+	case 5:
+		div = 4;
+		break;
+	case 6:
+	case 7:
+		div = 6;
+		break;
+	case 8:
+	case 9:
+	case 10:
+	case 11:
+		div = 8;
+		break;
+	default:
+		div = 12;
+		break;
+	}
+	clk_rst_ctl.s.h_div = div;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+	/* Read it back, */
+	clk_rst_ctl.u64 = cvmx_read_csr(CVMX_UCTLX_CLK_RST_CTL(0));
+	clk_rst_ctl.s.h_clkdiv_en = 1;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+	/* 3d */
+	clk_rst_ctl.s.h_clkdiv_rst = 1;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* 3e: delay 64 io clocks */
+	ndelay(io_clk_64_to_ns);
+
+	/*
+	 * Step 4: Program the power-on reset field in the UCTL
+	 * clock-reset-control register.
+	 */
+	clk_rst_ctl.s.p_por = 0;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* Step 5:    Wait 1 ms for the PHY clock to start. */
+	mdelay(1);
+
+	/*
+	 * Step 6: Program the reset input from automatic test
+	 * equipment field in the UPHY CSR
+	 */
+	uphy_ctl_status.u64 = cvmx_read_csr(CVMX_UCTLX_UPHY_CTL_STATUS(0));
+	uphy_ctl_status.s.ate_reset = 1;
+	cvmx_write_csr(CVMX_UCTLX_UPHY_CTL_STATUS(0), uphy_ctl_status.u64);
+
+	/* Step 7: Wait for at least 10ns. */
+	ndelay(10);
+
+	/* Step 8: Clear the ATE_RESET field in the UPHY CSR. */
+	uphy_ctl_status.s.ate_reset = 0;
+	cvmx_write_csr(CVMX_UCTLX_UPHY_CTL_STATUS(0), uphy_ctl_status.u64);
+
+	/*
+	 * Step 9: Wait for at least 20ns for UPHY to output PHY clock
+	 * signals and OHCI_CLK48
+	 */
+	ndelay(20);
+
+	/* Step 10: Configure the OHCI_CLK48 and OHCI_CLK12 clocks. */
+	/* 10a */
+	clk_rst_ctl.s.o_clkdiv_rst = 1;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* 10b */
+	clk_rst_ctl.s.o_clkdiv_en = 1;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* 10c */
+	ndelay(io_clk_64_to_ns);
+
+	/*
+	 * Step 11: Program the PHY reset field:
+	 * UCTL0_CLK_RST_CTL[P_PRST] = 1
+	 */
+	clk_rst_ctl.s.p_prst = 1;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* Step 12: Wait 1 uS. */
+	udelay(1);
+
+	/* Step 13: Program the HRESET_N field: UCTL0_CLK_RST_CTL[HRST] = 1 */
+	clk_rst_ctl.s.hrst = 1;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+end_clock:
+	/* Now we can set some other registers.  */
+
+	for (i = 0; i <= 1; i++) {
+		port_ctl_status.u64 =
+			cvmx_read_csr(CVMX_UCTLX_UPHY_PORTX_CTL_STATUS(i, 0));
+		/* Set txvreftune to 15 to obtain compliant 'eye' diagram. */
+		port_ctl_status.s.txvreftune = 15;
+		port_ctl_status.s.txrisetune = 1;
+		port_ctl_status.s.txpreemphasistune = 1;
+		cvmx_write_csr(CVMX_UCTLX_UPHY_PORTX_CTL_STATUS(i, 0),
+			       port_ctl_status.u64);
+	}
+
+	/* Set uSOF cycle period to 60,000 bits. */
+	cvmx_write_csr(CVMX_UCTLX_EHCI_FLA(0), 0x20ull);
+exit:
+	mutex_unlock(&octeon2_usb_clocks_mutex);
+}
+
+static void octeon2_usb_clocks_stop(void)
+{
+	mutex_lock(&octeon2_usb_clocks_mutex);
+	octeon2_usb_clock_start_cnt--;
+	mutex_unlock(&octeon2_usb_clocks_mutex);
+}
+
+static int octeon_ehci_power_on(struct platform_device *pdev)
+{
+	octeon2_usb_clocks_start();
+	return 0;
+}
+
+static void octeon_ehci_power_off(struct platform_device *pdev)
+{
+	octeon2_usb_clocks_stop();
+}
+
+static struct usb_ehci_pdata octeon_ehci_pdata = {
+	/* Octeon EHCI matches CPU endianness. */
+#ifdef __BIG_ENDIAN
+	.big_endian_mmio	= 1,
+#endif
+	.power_on	= octeon_ehci_power_on,
+	.power_off	= octeon_ehci_power_off,
+};
+
+static void __init octeon_ehci_hw_start(void)
+{
+	union cvmx_uctlx_ehci_ctl ehci_ctl;
+
+	octeon2_usb_clocks_start();
+
+	ehci_ctl.u64 = cvmx_read_csr(CVMX_UCTLX_EHCI_CTL(0));
+	/* Use 64-bit addressing. */
+	ehci_ctl.s.ehci_64b_addr_en = 1;
+	ehci_ctl.s.l2c_addr_msb = 0;
+	ehci_ctl.s.l2c_buff_emod = 1; /* Byte swapped. */
+	ehci_ctl.s.l2c_desc_emod = 1; /* Byte swapped. */
+	cvmx_write_csr(CVMX_UCTLX_EHCI_CTL(0), ehci_ctl.u64);
+
+	octeon2_usb_clocks_stop();
+}
+
+static u64 octeon_ehci_dma_mask = DMA_BIT_MASK(64);
+
 static int __init octeon_ehci_device_init(void)
 {
 	struct platform_device *pd;
@@ -88,7 +316,7 @@ static int __init octeon_ehci_device_init(void)
 	if (octeon_is_simulation() || usb_disabled())
 		return 0; /* No USB in the simulator. */
 
-	pd = platform_device_alloc("octeon-ehci", 0);
+	pd = platform_device_alloc("ehci-platform", 0);
 	if (!pd) {
 		ret = -ENOMEM;
 		goto out;
@@ -105,6 +333,10 @@ static int __init octeon_ehci_device_init(void)
 	if (ret)
 		goto fail;
 
+	pd->dev.dma_mask = &octeon_ehci_dma_mask;
+	pd->dev.platform_data = &octeon_ehci_pdata;
+	octeon_ehci_hw_start();
+
 	ret = platform_device_add(pd);
 	if (ret)
 		goto fail;
@@ -117,6 +349,41 @@ out:
 }
 device_initcall(octeon_ehci_device_init);
 
+static int octeon_ohci_power_on(struct platform_device *pdev)
+{
+	octeon2_usb_clocks_start();
+	return 0;
+}
+
+static void octeon_ohci_power_off(struct platform_device *pdev)
+{
+	octeon2_usb_clocks_stop();
+}
+
+static struct usb_ohci_pdata octeon_ohci_pdata = {
+	/* Octeon OHCI matches CPU endianness. */
+#ifdef __BIG_ENDIAN
+	.big_endian_mmio	= 1,
+#endif
+	.power_on	= octeon_ohci_power_on,
+	.power_off	= octeon_ohci_power_off,
+};
+
+static void __init octeon_ohci_hw_start(void)
+{
+	union cvmx_uctlx_ohci_ctl ohci_ctl;
+
+	octeon2_usb_clocks_start();
+
+	ohci_ctl.u64 = cvmx_read_csr(CVMX_UCTLX_OHCI_CTL(0));
+	ohci_ctl.s.l2c_addr_msb = 0;
+	ohci_ctl.s.l2c_buff_emod = 1; /* Byte swapped. */
+	ohci_ctl.s.l2c_desc_emod = 1; /* Byte swapped. */
+	cvmx_write_csr(CVMX_UCTLX_OHCI_CTL(0), ohci_ctl.u64);
+
+	octeon2_usb_clocks_stop();
+}
+
 static int __init octeon_ohci_device_init(void)
 {
 	struct platform_device *pd;
@@ -137,7 +404,7 @@ static int __init octeon_ohci_device_init(void)
 	if (octeon_is_simulation() || usb_disabled())
 		return 0; /* No USB in the simulator. */
 
-	pd = platform_device_alloc("octeon-ohci", 0);
+	pd = platform_device_alloc("ohci-platform", 0);
 	if (!pd) {
 		ret = -ENOMEM;
 		goto out;
@@ -154,6 +421,9 @@ static int __init octeon_ohci_device_init(void)
 	if (ret)
 		goto fail;
 
+	pd->dev.platform_data = &octeon_ohci_pdata;
+	octeon_ohci_hw_start();
+
 	ret = platform_device_add(pd);
 	if (ret)
 		goto fail;
arch/mips/configs/cavium_octeon_defconfig +3 -0
index b2476a1..e57058d 100644
@@ -120,6 +120,9 @@ CONFIG_SPI_OCTEON=y
 # CONFIG_HWMON is not set
 CONFIG_WATCHDOG=y
 # CONFIG_USB_SUPPORT is not set
+CONFIG_USB_EHCI_BIG_ENDIAN_MMIO=y
+CONFIG_USB_OHCI_BIG_ENDIAN_MMIO=y
+CONFIG_USB_OHCI_LITTLE_ENDIAN=y
 CONFIG_RTC_CLASS=y
 CONFIG_RTC_DRV_DS1307=y
 CONFIG_STAGING=y
drivers/usb/host/Kconfig +10 -8
index a3ca137..fafc628 100644
@@ -292,11 +292,15 @@ config USB_EHCI_HCD_PLATFORM
 	  If unsure, say N.
 
 config USB_OCTEON_EHCI
-	bool "Octeon on-chip EHCI support"
+	bool "Octeon on-chip EHCI support (DEPRECATED)"
 	depends on CAVIUM_OCTEON_SOC
 	default n
 	select USB_EHCI_BIG_ENDIAN_MMIO
+	select USB_EHCI_HCD_PLATFORM
 	help
+	  This option is deprecated now and the driver was removed, use
+	  USB_EHCI_HCD_PLATFORM instead.
+
 	  Enable support for the Octeon II SOC's on-chip EHCI
 	  controller.  It is needed for high-speed (480Mbit/sec)
 	  USB 2.0 device support.  All CN6XXX based chips with USB are
@@ -575,12 +579,16 @@ config USB_OHCI_HCD_PLATFORM
 	  If unsure, say N.
 
 config USB_OCTEON_OHCI
-	bool "Octeon on-chip OHCI support"
+	bool "Octeon on-chip OHCI support (DEPRECATED)"
 	depends on CAVIUM_OCTEON_SOC
 	default USB_OCTEON_EHCI
 	select USB_OHCI_BIG_ENDIAN_MMIO
 	select USB_OHCI_LITTLE_ENDIAN
+	select USB_OHCI_HCD_PLATFORM
 	help
+	  This option is deprecated now and the driver was removed, use
+	  USB_OHCI_HCD_PLATFORM instead.
+
 	  Enable support for the Octeon II SOC's on-chip OHCI
 	  controller.  It is needed for low-speed USB 1.0 device
 	  support.  All CN6XXX based chips with USB are supported.
@@ -754,12 +762,6 @@ config USB_IMX21_HCD
          To compile this driver as a module, choose M here: the
          module will be called "imx21-hcd".
 
-
-
-config USB_OCTEON2_COMMON
-	bool
-	default y if USB_OCTEON_EHCI || USB_OCTEON_OHCI
-
 config USB_HCD_BCMA
 	tristate "BCMA usb host driver"
 	depends on BCMA
drivers/usb/host/Makefile +0 -1
index 348c243..d6216a4 100644
@@ -73,7 +73,6 @@ obj-$(CONFIG_USB_ISP1760_HCD)	+= isp1760.o
 obj-$(CONFIG_USB_HWA_HCD)	+= hwa-hc.o
 obj-$(CONFIG_USB_IMX21_HCD)	+= imx21-hcd.o
 obj-$(CONFIG_USB_FSL_MPH_DR_OF)	+= fsl-mph-dr-of.o
-obj-$(CONFIG_USB_OCTEON2_COMMON) += octeon2-common.o
 obj-$(CONFIG_USB_HCD_BCMA)	+= bcma-hcd.o
 obj-$(CONFIG_USB_HCD_SSB)	+= ssb-hcd.o
 obj-$(CONFIG_USB_FUSBH200_HCD)	+= fusbh200-hcd.o
drivers/usb/host/ehci-hcd.c +0 -5
index 15feaf9..5ea27a0 100644
@@ -1268,11 +1268,6 @@ MODULE_LICENSE ("GPL");
 #define XILINX_OF_PLATFORM_DRIVER	ehci_hcd_xilinx_of_driver
 #endif
 
-#ifdef CONFIG_USB_OCTEON_EHCI
-#include "ehci-octeon.c"
-#define PLATFORM_DRIVER		ehci_octeon_driver
-#endif
-
 #ifdef CONFIG_TILE_USB
 #include "ehci-tilegx.c"
 #define	PLATFORM_DRIVER		ehci_hcd_tilegx_driver
drivers/usb/host/ehci-octeon.c deleted +0 -188

188 lines not on this page · raw message

drivers/usb/host/octeon2-common.c deleted +0 -200

200 lines not on this page · raw message

drivers/usb/host/ohci-hcd.c +0 -5
index d664eda..1dab9df 100644
@@ -1249,11 +1249,6 @@ MODULE_LICENSE ("GPL");
 #define PLATFORM_DRIVER	ohci_hcd_jz4740_driver
 #endif
 
-#ifdef CONFIG_USB_OCTEON_OHCI
-#include "ohci-octeon.c"
-#define PLATFORM_DRIVER		ohci_octeon_driver
-#endif
-
 #ifdef CONFIG_TILE_USB
 #include "ohci-tilegx.c"
 #define PLATFORM_DRIVER		ohci_hcd_tilegx_driver
drivers/usb/host/ohci-octeon.c deleted +0 -202

202 lines not on this page · raw message

Signature · 1 line
1.7.9.5
Acked-by Acked-by on v1 1/3
#2 lore raw

Re: [PATCH 1/3] USB: host: Remove ehci-octeon and ohci-octeon drivers

Quoted 3 lines · Andreas Herrmann

From: Alan Stern <[email protected]>

From: Alan Stern <[email protected]>

Is there an echo?

Is there an echo?

Quoted 12 lines

Remove special-purpose octeon drivers and instead use ehci-platform and ohci-platform as suggested with
http://marc.info/?l=linux-mips&m=140139694721623&w=2

[andreas.herrmann:
fixed compile error]

Quoted 14 lines
---
 arch/mips/cavium-octeon/octeon-platform.c |  274 ++++++++++++++++++++++++++++-
 arch/mips/configs/cavium_octeon_defconfig |    3 +
 drivers/usb/host/Kconfig                  |   18 +-
 drivers/usb/host/Makefile                 |    1 -
 drivers/usb/host/ehci-hcd.c               |    5 -
 drivers/usb/host/ehci-octeon.c            |  188 --------------------
 drivers/usb/host/octeon2-common.c         |  200 ---------------------
 drivers/usb/host/ohci-hcd.c               |    5 -
 drivers/usb/host/ohci-octeon.c            |  202 ---------------------
 9 files changed, 285 insertions(+), 611 deletions(-)
 delete mode 100644 drivers/usb/host/ehci-octeon.c
 delete mode 100644 drivers/usb/host/octeon2-common.c
 delete mode 100644 drivers/usb/host/ohci-octeon.c

For the MIPS bits:

For the MIPS bits:

Ralf

Ralf

#3 lore raw

Re: [PATCH 1/3] USB: host: Remove ehci-octeon and ohci-octeon drivers

Quoted 7 lines · Ralf Baechle
Quoted 3 lines · Andreas Herrmann

From: Alan Stern <[email protected]>

From: Alan Stern <[email protected]>

Is there an echo?

Oops.

Quoted 1 line

Is there an echo?

LOL.

Quoted 31 lines
Quoted 12 lines

Remove special-purpose octeon drivers and instead use ehci-platform and ohci-platform as suggested with
http://marc.info/?l=linux-mips&m=140139694721623&w=2

[andreas.herrmann:
fixed compile error]

Quoted 14 lines
---
 arch/mips/cavium-octeon/octeon-platform.c |  274 ++++++++++++++++++++++++++++-
 arch/mips/configs/cavium_octeon_defconfig |    3 +
 drivers/usb/host/Kconfig                  |   18 +-
 drivers/usb/host/Makefile                 |    1 -
 drivers/usb/host/ehci-hcd.c               |    5 -
 drivers/usb/host/ehci-octeon.c            |  188 --------------------
 drivers/usb/host/octeon2-common.c         |  200 ---------------------
 drivers/usb/host/ohci-hcd.c               |    5 -
 drivers/usb/host/ohci-octeon.c            |  202 ---------------------
 9 files changed, 285 insertions(+), 611 deletions(-)
 delete mode 100644 drivers/usb/host/ehci-octeon.c
 delete mode 100644 drivers/usb/host/octeon2-common.c
 delete mode 100644 drivers/usb/host/ohci-octeon.c

For the MIPS bits:

For the MIPS bits:

Quoted 7 lines

Ralf

Ralf

Double thanks,

Andreas

#4 lore raw

Re: [PATCH 1/3] USB: host: Remove ehci-octeon and ohci-octeon drivers

In reply to Andreas Herrmann
Quoted 15 lines · Andreas Herrmann

From: Alan Stern <[email protected]>

Remove special-purpose octeon drivers and instead use ehci-platform and ohci-platform as suggested with
http://marc.info/?l=linux-mips&m=140139694721623&w=2

[andreas.herrmann:
fixed compile error]

Quoted 14 lines
---
 arch/mips/cavium-octeon/octeon-platform.c |  274 ++++++++++++++++++++++++++++-
 arch/mips/configs/cavium_octeon_defconfig |    3 +
 drivers/usb/host/Kconfig                  |   18 +-
 drivers/usb/host/Makefile                 |    1 -
 drivers/usb/host/ehci-hcd.c               |    5 -
 drivers/usb/host/ehci-octeon.c            |  188 --------------------
 drivers/usb/host/octeon2-common.c         |  200 ---------------------
 drivers/usb/host/ohci-hcd.c               |    5 -
 drivers/usb/host/ohci-octeon.c            |  202 ---------------------
 9 files changed, 285 insertions(+), 611 deletions(-)
 delete mode 100644 drivers/usb/host/ehci-octeon.c
 delete mode 100644 drivers/usb/host/octeon2-common.c
 delete mode 100644 drivers/usb/host/ohci-octeon.c

This doesn't apply to my usb-next or usb-testing branch of usb.git on git.kernel.org, so I can't apply it :(

#5 lore raw

Re: [PATCH 1/3] USB: host: Remove ehci-octeon and ohci-octeon drivers

Quoted 31 lines · Greg KH
Quoted 15 lines · Andreas Herrmann

From: Alan Stern <[email protected]>

Remove special-purpose octeon drivers and instead use ehci-platform and ohci-platform as suggested with
http://marc.info/?l=linux-mips&m=140139694721623&w=2

[andreas.herrmann:
fixed compile error]

Quoted 14 lines
---
 arch/mips/cavium-octeon/octeon-platform.c |  274 ++++++++++++++++++++++++++++-
 arch/mips/configs/cavium_octeon_defconfig |    3 +
 drivers/usb/host/Kconfig                  |   18 +-
 drivers/usb/host/Makefile                 |    1 -
 drivers/usb/host/ehci-hcd.c               |    5 -
 drivers/usb/host/ehci-octeon.c            |  188 --------------------
 drivers/usb/host/octeon2-common.c         |  200 ---------------------
 drivers/usb/host/ohci-hcd.c               |    5 -
 drivers/usb/host/ohci-octeon.c            |  202 ---------------------
 9 files changed, 285 insertions(+), 611 deletions(-)
 delete mode 100644 drivers/usb/host/ehci-octeon.c
 delete mode 100644 drivers/usb/host/octeon2-common.c
 delete mode 100644 drivers/usb/host/ohci-octeon.c
Quoted 2 lines

This doesn't apply to my usb-next or usb-testing branch of usb.git on git.kernel.org, so I can't apply it :(

Sorry, I need to rebase it (on usb-next).

Andreas

#6 lore raw

PATCH v2 1/3 USB: host: Remove ehci-octeon and ohci-octeon drivers

From: Alan Stern <[email protected]>

Remove special-purpose octeon drivers and instead use ehci-platform and ohci-platform as suggested with
http://marc.info/?l=linux-mips&m=140139694721623&w=2

[andreas.herrmann:
fixed compile error]

---
 arch/mips/cavium-octeon/octeon-platform.c |  274 ++++++++++++++++++++++++++++-
 arch/mips/configs/cavium_octeon_defconfig |    3 +
 drivers/usb/host/Kconfig                  |   18 +-
 drivers/usb/host/Makefile                 |    1 -
 drivers/usb/host/ehci-hcd.c               |    5 -
 drivers/usb/host/ehci-octeon.c            |  182 -------------------
 drivers/usb/host/octeon2-common.c         |  200 ---------------------
 drivers/usb/host/ohci-hcd.c               |    5 -
 drivers/usb/host/ohci-octeon.c            |  196 ---------------------
 9 files changed, 285 insertions(+), 599 deletions(-)
 delete mode 100644 drivers/usb/host/ehci-octeon.c
 delete mode 100644 drivers/usb/host/octeon2-common.c
 delete mode 100644 drivers/usb/host/ohci-octeon.c

There was a conflict with commits
073153bf22764 (host: ehci-octeon: remove duplicate check on resource) c6d413cebd82c (host: ohci-octeon: remove duplicate check on resource)

I rebased the patch to your usb-next branch as of
v3.18-rc4-66-g69b7290.

Patch 2 and 3 of the series should apply w/o issues.

Thanks,

Andreas

arch/mips/cavium-octeon/octeon-platform.c +272 -2
index 6df0f4d..b67ddf0 100644
@@ -7,22 +7,27 @@
  * Copyright (C) 2008 Wind River Systems
  */
 
+#include <linux/delay.h>
 #include <linux/init.h>
 #include <linux/irq.h>
 #include <linux/i2c.h>
 #include <linux/usb.h>
 #include <linux/dma-mapping.h>
 #include <linux/module.h>
+#include <linux/mutex.h>
 #include <linux/slab.h>
 #include <linux/platform_device.h>
 #include <linux/of_platform.h>
 #include <linux/of_fdt.h>
 #include <linux/libfdt.h>
+#include <linux/usb/ehci_pdriver.h>
+#include <linux/usb/ohci_pdriver.h>
 
 #include <asm/octeon/octeon.h>
 #include <asm/octeon/cvmx-rnm-defs.h>
 #include <asm/octeon/cvmx-helper.h>
 #include <asm/octeon/cvmx-helper-board.h>
+#include <asm/octeon/cvmx-uctlx-defs.h>
 
 /* Octeon Random Number Generator.  */
 static int __init octeon_rng_device_init(void)
@@ -68,6 +73,229 @@ device_initcall(octeon_rng_device_init);
 
 #ifdef CONFIG_USB
 
+static DEFINE_MUTEX(octeon2_usb_clocks_mutex);
+
+static int octeon2_usb_clock_start_cnt;
+
+static void octeon2_usb_clocks_start(void)
+{
+	u64 div;
+	union cvmx_uctlx_if_ena if_ena;
+	union cvmx_uctlx_clk_rst_ctl clk_rst_ctl;
+	union cvmx_uctlx_uphy_ctl_status uphy_ctl_status;
+	union cvmx_uctlx_uphy_portx_ctl_status port_ctl_status;
+	int i;
+	unsigned long io_clk_64_to_ns;
+
+
+	mutex_lock(&octeon2_usb_clocks_mutex);
+
+	octeon2_usb_clock_start_cnt++;
+	if (octeon2_usb_clock_start_cnt != 1)
+		goto exit;
+
+	io_clk_64_to_ns = 64000000000ull / octeon_get_io_clock_rate();
+
+	/*
+	 * Step 1: Wait for voltages stable.  That surely happened
+	 * before starting the kernel.
+	 *
+	 * Step 2: Enable  SCLK of UCTL by writing UCTL0_IF_ENA[EN] = 1
+	 */
+	if_ena.u64 = 0;
+	if_ena.s.en = 1;
+	cvmx_write_csr(CVMX_UCTLX_IF_ENA(0), if_ena.u64);
+
+	/* Step 3: Configure the reference clock, PHY, and HCLK */
+	clk_rst_ctl.u64 = cvmx_read_csr(CVMX_UCTLX_CLK_RST_CTL(0));
+
+	/*
+	 * If the UCTL looks like it has already been started, skip
+	 * the initialization, otherwise bus errors are obtained.
+	 */
+	if (clk_rst_ctl.s.hrst)
+		goto end_clock;
+	/* 3a */
+	clk_rst_ctl.s.p_por = 1;
+	clk_rst_ctl.s.hrst = 0;
+	clk_rst_ctl.s.p_prst = 0;
+	clk_rst_ctl.s.h_clkdiv_rst = 0;
+	clk_rst_ctl.s.o_clkdiv_rst = 0;
+	clk_rst_ctl.s.h_clkdiv_en = 0;
+	clk_rst_ctl.s.o_clkdiv_en = 0;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* 3b */
+	/* 12MHz crystal. */
+	clk_rst_ctl.s.p_refclk_sel = 0;
+	clk_rst_ctl.s.p_refclk_div = 0;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* 3c */
+	div = octeon_get_io_clock_rate() / 130000000ull;
+
+	switch (div) {
+	case 0:
+		div = 1;
+		break;
+	case 1:
+	case 2:
+	case 3:
+	case 4:
+		break;
+	case 5:
+		div = 4;
+		break;
+	case 6:
+	case 7:
+		div = 6;
+		break;
+	case 8:
+	case 9:
+	case 10:
+	case 11:
+		div = 8;
+		break;
+	default:
+		div = 12;
+		break;
+	}
+	clk_rst_ctl.s.h_div = div;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+	/* Read it back, */
+	clk_rst_ctl.u64 = cvmx_read_csr(CVMX_UCTLX_CLK_RST_CTL(0));
+	clk_rst_ctl.s.h_clkdiv_en = 1;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+	/* 3d */
+	clk_rst_ctl.s.h_clkdiv_rst = 1;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* 3e: delay 64 io clocks */
+	ndelay(io_clk_64_to_ns);
+
+	/*
+	 * Step 4: Program the power-on reset field in the UCTL
+	 * clock-reset-control register.
+	 */
+	clk_rst_ctl.s.p_por = 0;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* Step 5:    Wait 1 ms for the PHY clock to start. */
+	mdelay(1);
+
+	/*
+	 * Step 6: Program the reset input from automatic test
+	 * equipment field in the UPHY CSR
+	 */
+	uphy_ctl_status.u64 = cvmx_read_csr(CVMX_UCTLX_UPHY_CTL_STATUS(0));
+	uphy_ctl_status.s.ate_reset = 1;
+	cvmx_write_csr(CVMX_UCTLX_UPHY_CTL_STATUS(0), uphy_ctl_status.u64);
+
+	/* Step 7: Wait for at least 10ns. */
+	ndelay(10);
+
+	/* Step 8: Clear the ATE_RESET field in the UPHY CSR. */
+	uphy_ctl_status.s.ate_reset = 0;
+	cvmx_write_csr(CVMX_UCTLX_UPHY_CTL_STATUS(0), uphy_ctl_status.u64);
+
+	/*
+	 * Step 9: Wait for at least 20ns for UPHY to output PHY clock
+	 * signals and OHCI_CLK48
+	 */
+	ndelay(20);
+
+	/* Step 10: Configure the OHCI_CLK48 and OHCI_CLK12 clocks. */
+	/* 10a */
+	clk_rst_ctl.s.o_clkdiv_rst = 1;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* 10b */
+	clk_rst_ctl.s.o_clkdiv_en = 1;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* 10c */
+	ndelay(io_clk_64_to_ns);
+
+	/*
+	 * Step 11: Program the PHY reset field:
+	 * UCTL0_CLK_RST_CTL[P_PRST] = 1
+	 */
+	clk_rst_ctl.s.p_prst = 1;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+	/* Step 12: Wait 1 uS. */
+	udelay(1);
+
+	/* Step 13: Program the HRESET_N field: UCTL0_CLK_RST_CTL[HRST] = 1 */
+	clk_rst_ctl.s.hrst = 1;
+	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
+
+end_clock:
+	/* Now we can set some other registers.  */
+
+	for (i = 0; i <= 1; i++) {
+		port_ctl_status.u64 =
+			cvmx_read_csr(CVMX_UCTLX_UPHY_PORTX_CTL_STATUS(i, 0));
+		/* Set txvreftune to 15 to obtain compliant 'eye' diagram. */
+		port_ctl_status.s.txvreftune = 15;
+		port_ctl_status.s.txrisetune = 1;
+		port_ctl_status.s.txpreemphasistune = 1;
+		cvmx_write_csr(CVMX_UCTLX_UPHY_PORTX_CTL_STATUS(i, 0),
+			       port_ctl_status.u64);
+	}
+
+	/* Set uSOF cycle period to 60,000 bits. */
+	cvmx_write_csr(CVMX_UCTLX_EHCI_FLA(0), 0x20ull);
+exit:
+	mutex_unlock(&octeon2_usb_clocks_mutex);
+}
+
+static void octeon2_usb_clocks_stop(void)
+{
+	mutex_lock(&octeon2_usb_clocks_mutex);
+	octeon2_usb_clock_start_cnt--;
+	mutex_unlock(&octeon2_usb_clocks_mutex);
+}
+
+static int octeon_ehci_power_on(struct platform_device *pdev)
+{
+	octeon2_usb_clocks_start();
+	return 0;
+}
+
+static void octeon_ehci_power_off(struct platform_device *pdev)
+{
+	octeon2_usb_clocks_stop();
+}
+
+static struct usb_ehci_pdata octeon_ehci_pdata = {
+	/* Octeon EHCI matches CPU endianness. */
+#ifdef __BIG_ENDIAN
+	.big_endian_mmio	= 1,
+#endif
+	.power_on	= octeon_ehci_power_on,
+	.power_off	= octeon_ehci_power_off,
+};
+
+static void __init octeon_ehci_hw_start(void)
+{
+	union cvmx_uctlx_ehci_ctl ehci_ctl;
+
+	octeon2_usb_clocks_start();
+
+	ehci_ctl.u64 = cvmx_read_csr(CVMX_UCTLX_EHCI_CTL(0));
+	/* Use 64-bit addressing. */
+	ehci_ctl.s.ehci_64b_addr_en = 1;
+	ehci_ctl.s.l2c_addr_msb = 0;
+	ehci_ctl.s.l2c_buff_emod = 1; /* Byte swapped. */
+	ehci_ctl.s.l2c_desc_emod = 1; /* Byte swapped. */
+	cvmx_write_csr(CVMX_UCTLX_EHCI_CTL(0), ehci_ctl.u64);
+
+	octeon2_usb_clocks_stop();
+}
+
+static u64 octeon_ehci_dma_mask = DMA_BIT_MASK(64);
+
 static int __init octeon_ehci_device_init(void)
 {
 	struct platform_device *pd;
@@ -88,7 +316,7 @@ static int __init octeon_ehci_device_init(void)
 	if (octeon_is_simulation() || usb_disabled())
 		return 0; /* No USB in the simulator. */
 
-	pd = platform_device_alloc("octeon-ehci", 0);
+	pd = platform_device_alloc("ehci-platform", 0);
 	if (!pd) {
 		ret = -ENOMEM;
 		goto out;
@@ -105,6 +333,10 @@ static int __init octeon_ehci_device_init(void)
 	if (ret)
 		goto fail;
 
+	pd->dev.dma_mask = &octeon_ehci_dma_mask;
+	pd->dev.platform_data = &octeon_ehci_pdata;
+	octeon_ehci_hw_start();
+
 	ret = platform_device_add(pd);
 	if (ret)
 		goto fail;
@@ -117,6 +349,41 @@ out:
 }
 device_initcall(octeon_ehci_device_init);
 
+static int octeon_ohci_power_on(struct platform_device *pdev)
+{
+	octeon2_usb_clocks_start();
+	return 0;
+}
+
+static void octeon_ohci_power_off(struct platform_device *pdev)
+{
+	octeon2_usb_clocks_stop();
+}
+
+static struct usb_ohci_pdata octeon_ohci_pdata = {
+	/* Octeon OHCI matches CPU endianness. */
+#ifdef __BIG_ENDIAN
+	.big_endian_mmio	= 1,
+#endif
+	.power_on	= octeon_ohci_power_on,
+	.power_off	= octeon_ohci_power_off,
+};
+
+static void __init octeon_ohci_hw_start(void)
+{
+	union cvmx_uctlx_ohci_ctl ohci_ctl;
+
+	octeon2_usb_clocks_start();
+
+	ohci_ctl.u64 = cvmx_read_csr(CVMX_UCTLX_OHCI_CTL(0));
+	ohci_ctl.s.l2c_addr_msb = 0;
+	ohci_ctl.s.l2c_buff_emod = 1; /* Byte swapped. */
+	ohci_ctl.s.l2c_desc_emod = 1; /* Byte swapped. */
+	cvmx_write_csr(CVMX_UCTLX_OHCI_CTL(0), ohci_ctl.u64);
+
+	octeon2_usb_clocks_stop();
+}
+
 static int __init octeon_ohci_device_init(void)
 {
 	struct platform_device *pd;
@@ -137,7 +404,7 @@ static int __init octeon_ohci_device_init(void)
 	if (octeon_is_simulation() || usb_disabled())
 		return 0; /* No USB in the simulator. */
 
-	pd = platform_device_alloc("octeon-ohci", 0);
+	pd = platform_device_alloc("ohci-platform", 0);
 	if (!pd) {
 		ret = -ENOMEM;
 		goto out;
@@ -154,6 +421,9 @@ static int __init octeon_ohci_device_init(void)
 	if (ret)
 		goto fail;
 
+	pd->dev.platform_data = &octeon_ohci_pdata;
+	octeon_ohci_hw_start();
+
 	ret = platform_device_add(pd);
 	if (ret)
 		goto fail;
arch/mips/configs/cavium_octeon_defconfig +3 -0
index b2476a1..e57058d 100644
@@ -120,6 +120,9 @@ CONFIG_SPI_OCTEON=y
 # CONFIG_HWMON is not set
 CONFIG_WATCHDOG=y
 # CONFIG_USB_SUPPORT is not set
+CONFIG_USB_EHCI_BIG_ENDIAN_MMIO=y
+CONFIG_USB_OHCI_BIG_ENDIAN_MMIO=y
+CONFIG_USB_OHCI_LITTLE_ENDIAN=y
 CONFIG_RTC_CLASS=y
 CONFIG_RTC_DRV_DS1307=y
 CONFIG_STAGING=y
drivers/usb/host/Kconfig +10 -8
index a3ca137..fafc628 100644
@@ -292,11 +292,15 @@ config USB_EHCI_HCD_PLATFORM
 	  If unsure, say N.
 
 config USB_OCTEON_EHCI
-	bool "Octeon on-chip EHCI support"
+	bool "Octeon on-chip EHCI support (DEPRECATED)"
 	depends on CAVIUM_OCTEON_SOC
 	default n
 	select USB_EHCI_BIG_ENDIAN_MMIO
+	select USB_EHCI_HCD_PLATFORM
 	help
+	  This option is deprecated now and the driver was removed, use
+	  USB_EHCI_HCD_PLATFORM instead.
+
 	  Enable support for the Octeon II SOC's on-chip EHCI
 	  controller.  It is needed for high-speed (480Mbit/sec)
 	  USB 2.0 device support.  All CN6XXX based chips with USB are
@@ -575,12 +579,16 @@ config USB_OHCI_HCD_PLATFORM
 	  If unsure, say N.
 
 config USB_OCTEON_OHCI
-	bool "Octeon on-chip OHCI support"
+	bool "Octeon on-chip OHCI support (DEPRECATED)"
 	depends on CAVIUM_OCTEON_SOC
 	default USB_OCTEON_EHCI
 	select USB_OHCI_BIG_ENDIAN_MMIO
 	select USB_OHCI_LITTLE_ENDIAN
+	select USB_OHCI_HCD_PLATFORM
 	help
+	  This option is deprecated now and the driver was removed, use
+	  USB_OHCI_HCD_PLATFORM instead.
+
 	  Enable support for the Octeon II SOC's on-chip OHCI
 	  controller.  It is needed for low-speed USB 1.0 device
 	  support.  All CN6XXX based chips with USB are supported.
@@ -754,12 +762,6 @@ config USB_IMX21_HCD
          To compile this driver as a module, choose M here: the
          module will be called "imx21-hcd".
 
-
-
-config USB_OCTEON2_COMMON
-	bool
-	default y if USB_OCTEON_EHCI || USB_OCTEON_OHCI
-
 config USB_HCD_BCMA
 	tristate "BCMA usb host driver"
 	depends on BCMA
drivers/usb/host/Makefile +0 -1
index 348c243..d6216a4 100644
@@ -73,7 +73,6 @@ obj-$(CONFIG_USB_ISP1760_HCD)	+= isp1760.o
 obj-$(CONFIG_USB_HWA_HCD)	+= hwa-hc.o
 obj-$(CONFIG_USB_IMX21_HCD)	+= imx21-hcd.o
 obj-$(CONFIG_USB_FSL_MPH_DR_OF)	+= fsl-mph-dr-of.o
-obj-$(CONFIG_USB_OCTEON2_COMMON) += octeon2-common.o
 obj-$(CONFIG_USB_HCD_BCMA)	+= bcma-hcd.o
 obj-$(CONFIG_USB_HCD_SSB)	+= ssb-hcd.o
 obj-$(CONFIG_USB_FUSBH200_HCD)	+= fusbh200-hcd.o
drivers/usb/host/ehci-hcd.c +0 -5
index df75b8e..38bfeed 100644
@@ -1275,11 +1275,6 @@ MODULE_LICENSE ("GPL");
 #define XILINX_OF_PLATFORM_DRIVER	ehci_hcd_xilinx_of_driver
 #endif
 
-#ifdef CONFIG_USB_OCTEON_EHCI
-#include "ehci-octeon.c"
-#define PLATFORM_DRIVER		ehci_octeon_driver
-#endif
-
 #ifdef CONFIG_TILE_USB
 #include "ehci-tilegx.c"
 #define	PLATFORM_DRIVER		ehci_hcd_tilegx_driver
drivers/usb/host/ehci-octeon.c deleted +0 -182

182 lines not on this page · raw message

drivers/usb/host/octeon2-common.c deleted +0 -200

200 lines not on this page · raw message

drivers/usb/host/ohci-hcd.c +0 -5
index d664eda..1dab9df 100644
@@ -1249,11 +1249,6 @@ MODULE_LICENSE ("GPL");
 #define PLATFORM_DRIVER	ohci_hcd_jz4740_driver
 #endif
 
-#ifdef CONFIG_USB_OCTEON_OHCI
-#include "ohci-octeon.c"
-#define PLATFORM_DRIVER		ohci_octeon_driver
-#endif
-
 #ifdef CONFIG_TILE_USB
 #include "ohci-tilegx.c"
 #define PLATFORM_DRIVER		ohci_hcd_tilegx_driver
drivers/usb/host/ohci-octeon.c deleted +0 -196

196 lines not on this page · raw message

Signature · 1 line
1.7.9.5
#7 lore raw

Re: [PATCH 1/3 v2] USB: host: Remove ehci-octeon and ohci-octeon drivers

Quoted 31 lines · Andreas Herrmann

From: Alan Stern <[email protected]>

Remove special-purpose octeon drivers and instead use ehci-platform and ohci-platform as suggested with
http://marc.info/?l=linux-mips&m=140139694721623&w=2

[andreas.herrmann:
fixed compile error]

---
 arch/mips/cavium-octeon/octeon-platform.c |  274 ++++++++++++++++++++++++++++-
 arch/mips/configs/cavium_octeon_defconfig |    3 +
 drivers/usb/host/Kconfig                  |   18 +-
 drivers/usb/host/Makefile                 |    1 -
 drivers/usb/host/ehci-hcd.c               |    5 -
 drivers/usb/host/ehci-octeon.c            |  182 -------------------
 drivers/usb/host/octeon2-common.c         |  200 ---------------------
 drivers/usb/host/ohci-hcd.c               |    5 -
 drivers/usb/host/ohci-octeon.c            |  196 ---------------------
 9 files changed, 285 insertions(+), 599 deletions(-)
 delete mode 100644 drivers/usb/host/ehci-octeon.c
 delete mode 100644 drivers/usb/host/octeon2-common.c
 delete mode 100644 drivers/usb/host/ohci-octeon.c
Quoted 8 lines

There was a conflict with commits
073153bf22764 (host: ehci-octeon: remove duplicate check on resource) c6d413cebd82c (host: ohci-octeon: remove duplicate check on resource)

I rebased the patch to your usb-next branch as of
v3.18-rc4-66-g69b7290.

Patch 2 and 3 of the series should apply w/o issues.

They are long-gone from my queue, please resend.

thanks,

greg k-h

#8 lore raw

PATCH 2/3 USB: host: Remove hard-coded octeon platform information for ehci/ohci

In reply to Andreas Herrmann

Instead rely on device tree information for ehci and ohci.

This was suggested with
http://www.linux-mips.org/archives/linux-mips/2014-05/msg00307.html

"The device tree will *always* have correct ehci/ohci clock configuration, so use it. This allows us to remove a big chunk of platform configuration code from octeon-platform.c."

More or less I rebased that patch on Alan's work to remove ehci-octeon and ohci-octeon drivers.

Cc David Daney <[email protected]>
Cc Alex Smith <[email protected]>
Cc Alan Stern <[email protected]>
---
 arch/mips/cavium-octeon/octeon-platform.c |  148 ++++++++++++-----------------
 drivers/usb/host/ehci-platform.c          |    1 +
 drivers/usb/host/ohci-platform.c          |    1 +
 3 files changed, 64 insertions(+), 86 deletions(-)
arch/mips/cavium-octeon/octeon-platform.c +62 -86
index b67ddf0..eea60b6 100644
@@ -77,7 +77,7 @@ static DEFINE_MUTEX(octeon2_usb_clocks_mutex);
 
 static int octeon2_usb_clock_start_cnt;
 
-static void octeon2_usb_clocks_start(void)
+static void octeon2_usb_clocks_start(struct device *dev)
 {
 	u64 div;
 	union cvmx_uctlx_if_ena if_ena;
@@ -86,6 +86,8 @@ static void octeon2_usb_clocks_start(void)
 	union cvmx_uctlx_uphy_portx_ctl_status port_ctl_status;
 	int i;
 	unsigned long io_clk_64_to_ns;
+	u32 clock_rate = 12000000;
+	bool is_crystal_clock = false;
 
 
 	mutex_lock(&octeon2_usb_clocks_mutex);
@@ -96,6 +98,28 @@ static void octeon2_usb_clocks_start(void)
 
 	io_clk_64_to_ns = 64000000000ull / octeon_get_io_clock_rate();
 
+	if (dev->of_node) {
+		struct device_node *uctl_node;
+		const char *clock_type;
+
+		uctl_node = of_get_parent(dev->of_node);
+		if (!uctl_node) {
+			dev_err(dev, "No UCTL device node\n");
+			goto exit;
+		}
+		i = of_property_read_u32(uctl_node,
+					 "refclk-frequency", &clock_rate);
+		if (i) {
+			dev_err(dev, "No UCTL \"refclk-frequency\"\n");
+			goto exit;
+		}
+		i = of_property_read_string(uctl_node,
+					    "refclk-type", &clock_type);
+
+		if (!i && strcmp("crystal", clock_type) == 0)
+			is_crystal_clock = true;
+	}
+
 	/*
 	 * Step 1: Wait for voltages stable.  That surely happened
 	 * before starting the kernel.
@@ -126,9 +150,22 @@ static void octeon2_usb_clocks_start(void)
 	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
 
 	/* 3b */
-	/* 12MHz crystal. */
-	clk_rst_ctl.s.p_refclk_sel = 0;
-	clk_rst_ctl.s.p_refclk_div = 0;
+	clk_rst_ctl.s.p_refclk_sel = is_crystal_clock ? 0 : 1;
+	switch (clock_rate) {
+	default:
+		pr_err("Invalid UCTL clock rate of %u, using 12000000 instead\n",
+			clock_rate);
+		/* Fall through */
+	case 12000000:
+		clk_rst_ctl.s.p_refclk_div = 0;
+		break;
+	case 24000000:
+		clk_rst_ctl.s.p_refclk_div = 1;
+		break;
+	case 48000000:
+		clk_rst_ctl.s.p_refclk_div = 2;
+		break;
+	}
 	cvmx_write_csr(CVMX_UCTLX_CLK_RST_CTL(0), clk_rst_ctl.u64);
 
 	/* 3c */
@@ -259,7 +296,7 @@ static void octeon2_usb_clocks_stop(void)
 
 static int octeon_ehci_power_on(struct platform_device *pdev)
 {
-	octeon2_usb_clocks_start();
+	octeon2_usb_clocks_start(&pdev->dev);
 	return 0;
 }
 
@@ -277,11 +314,11 @@ static struct usb_ehci_pdata octeon_ehci_pdata = {
 	.power_off	= octeon_ehci_power_off,
 };
 
-static void __init octeon_ehci_hw_start(void)
+static void __init octeon_ehci_hw_start(struct device *dev)
 {
 	union cvmx_uctlx_ehci_ctl ehci_ctl;
 
-	octeon2_usb_clocks_start();
+	octeon2_usb_clocks_start(dev);
 
 	ehci_ctl.u64 = cvmx_read_csr(CVMX_UCTLX_EHCI_CTL(0));
 	/* Use 64-bit addressing. */
@@ -299,59 +336,28 @@ static u64 octeon_ehci_dma_mask = DMA_BIT_MASK(64);
 static int __init octeon_ehci_device_init(void)
 {
 	struct platform_device *pd;
+	struct device_node *ehci_node;
 	int ret = 0;
 
-	struct resource usb_resources[] = {
-		{
-			.flags	= IORESOURCE_MEM,
-		}, {
-			.flags	= IORESOURCE_IRQ,
-		}
-	};
-
-	/* Only Octeon2 has ehci/ohci */
-	if (!OCTEON_IS_MODEL(OCTEON_CN63XX))
+	ehci_node = of_find_node_by_name(NULL, "ehci");
+	if (!ehci_node)
 		return 0;
 
-	if (octeon_is_simulation() || usb_disabled())
-		return 0; /* No USB in the simulator. */
-
-	pd = platform_device_alloc("ehci-platform", 0);
-	if (!pd) {
-		ret = -ENOMEM;
-		goto out;
-	}
-
-	usb_resources[0].start = 0x00016F0000000000ULL;
-	usb_resources[0].end = usb_resources[0].start + 0x100;
-
-	usb_resources[1].start = OCTEON_IRQ_USB0;
-	usb_resources[1].end = OCTEON_IRQ_USB0;
-
-	ret = platform_device_add_resources(pd, usb_resources,
-					    ARRAY_SIZE(usb_resources));
-	if (ret)
-		goto fail;
+	pd = of_find_device_by_node(ehci_node);
+	if (!pd)
+		return 0;
 
 	pd->dev.dma_mask = &octeon_ehci_dma_mask;
 	pd->dev.platform_data = &octeon_ehci_pdata;
-	octeon_ehci_hw_start();
-
-	ret = platform_device_add(pd);
-	if (ret)
-		goto fail;
+	octeon_ehci_hw_start(&pd->dev);
 
 	return ret;
-fail:
-	platform_device_put(pd);
-out:
-	return ret;
 }
 device_initcall(octeon_ehci_device_init);
 
 static int octeon_ohci_power_on(struct platform_device *pdev)
 {
-	octeon2_usb_clocks_start();
+	octeon2_usb_clocks_start(&pdev->dev);
 	return 0;
 }
 
@@ -369,11 +375,11 @@ static struct usb_ohci_pdata octeon_ohci_pdata = {
 	.power_off	= octeon_ohci_power_off,
 };
 
-static void __init octeon_ohci_hw_start(void)
+static void __init octeon_ohci_hw_start(struct device *dev)
 {
 	union cvmx_uctlx_ohci_ctl ohci_ctl;
 
-	octeon2_usb_clocks_start();
+	octeon2_usb_clocks_start(dev);
 
 	ohci_ctl.u64 = cvmx_read_csr(CVMX_UCTLX_OHCI_CTL(0));
 	ohci_ctl.s.l2c_addr_msb = 0;
@@ -387,57 +393,27 @@ static void __init octeon_ohci_hw_start(void)
 static int __init octeon_ohci_device_init(void)
 {
 	struct platform_device *pd;
+	struct device_node *ohci_node;
 	int ret = 0;
 
-	struct resource usb_resources[] = {
-		{
-			.flags	= IORESOURCE_MEM,
-		}, {
-			.flags	= IORESOURCE_IRQ,
-		}
-	};
-
-	/* Only Octeon2 has ehci/ohci */
-	if (!OCTEON_IS_MODEL(OCTEON_CN63XX))
+	ohci_node = of_find_node_by_name(NULL, "ohci");
+	if (!ohci_node)
 		return 0;
 
-	if (octeon_is_simulation() || usb_disabled())
-		return 0; /* No USB in the simulator. */
-
-	pd = platform_device_alloc("ohci-platform", 0);
-	if (!pd) {
-		ret = -ENOMEM;
-		goto out;
-	}
-
-	usb_resources[0].start = 0x00016F0000000400ULL;
-	usb_resources[0].end = usb_resources[0].start + 0x100;
-
-	usb_resources[1].start = OCTEON_IRQ_USB0;
-	usb_resources[1].end = OCTEON_IRQ_USB0;
-
-	ret = platform_device_add_resources(pd, usb_resources,
-					    ARRAY_SIZE(usb_resources));
-	if (ret)
-		goto fail;
+	pd = of_find_device_by_node(ohci_node);
+	if (!pd)
+		return 0;
 
 	pd->dev.platform_data = &octeon_ohci_pdata;
-	octeon_ohci_hw_start();
-
-	ret = platform_device_add(pd);
-	if (ret)
-		goto fail;
+	octeon_ohci_hw_start(&pd->dev);
 
 	return ret;
-fail:
-	platform_device_put(pd);
-out:
-	return ret;
 }
 device_initcall(octeon_ohci_device_init);
 
 #endif /* CONFIG_USB */
 
+
 static struct of_device_id __initdata octeon_ids[] = {
 	{ .compatible = "simple-bus", },
 	{ .compatible = "cavium,octeon-6335-uctl", },
drivers/usb/host/ehci-platform.c +1 -0
index 2f5b9ce..2da18ea 100644
@@ -358,6 +358,7 @@ static const struct of_device_id vt8500_ehci_ids[] = {
 	{ .compatible = "via,vt8500-ehci", },
 	{ .compatible = "wm,prizm-ehci", },
 	{ .compatible = "generic-ehci", },
+	{ .compatible = "cavium,octeon-6335-ehci", },
 	{}
 };
 MODULE_DEVICE_TABLE(of, vt8500_ehci_ids);
drivers/usb/host/ohci-platform.c +1 -0
index 4369299..cb29865 100644
@@ -343,6 +343,7 @@ static int ohci_platform_resume(struct device *dev)
 
 static const struct of_device_id ohci_platform_ids[] = {
 	{ .compatible = "generic-ohci", },
+	{ .compatible = "cavium,octeon-6335-ohci", },
 	{ }
 };
 MODULE_DEVICE_TABLE(of, ohci_platform_ids);
Signature · 1 line
1.7.9.5
Acked-by on v1 2/3
#9 lore raw

Re: [PATCH 2/3] USB: host: Remove hard-coded octeon platform information for ehci/ohci

Quoted 9 lines · Andreas Herrmann

Instead rely on device tree information for ehci and ohci.

This was suggested with
http://www.linux-mips.org/archives/linux-mips/2014-05/msg00307.html

"The device tree will *always* have correct ehci/ohci clock configuration, so use it. This allows us to remove a big chunk of platform configuration code from octeon-platform.c."

Quoted 7 lines

More or less I rebased that patch on Alan's work to remove ehci-octeon and ohci-octeon drivers.

Cc David Daney <[email protected]>
Cc Alex Smith <[email protected]>
Cc Alan Stern <[email protected]>

For the MIPS bits:

Ralf

#10 lore raw

PATCH 3/3 USB: host: Introduce flag to enable use of 64-bit dma_mask for ehci-platform

In reply to Andreas Herrmann

ehci-octeon driver used a 64-bit dma_mask. With removal of ehci-octeon and usage of ehci-platform ehci dma_mask is now limited to 32 bits (coerced in ehci_platform_probe).

Provide a flag in ehci platform data to allow use of 64 bits for dma_mask.

Cc David Daney <[email protected]>
Cc Alex Smith <[email protected]>
Cc Alan Stern <[email protected]>
---
 arch/mips/cavium-octeon/octeon-platform.c |    4 +---
 drivers/usb/host/ehci-platform.c          |    3 ++-
 include/linux/usb/ehci_pdriver.h          |    1 +
 3 files changed, 4 insertions(+), 4 deletions(-)
arch/mips/cavium-octeon/octeon-platform.c +1 -3
index eea60b6..12410a2 100644
@@ -310,6 +310,7 @@ static struct usb_ehci_pdata octeon_ehci_pdata = {
 #ifdef __BIG_ENDIAN
 	.big_endian_mmio	= 1,
 #endif
+	.dma_mask_64	= 1,
 	.power_on	= octeon_ehci_power_on,
 	.power_off	= octeon_ehci_power_off,
 };
@@ -331,8 +332,6 @@ static void __init octeon_ehci_hw_start(struct device *dev)
 	octeon2_usb_clocks_stop();
 }
 
-static u64 octeon_ehci_dma_mask = DMA_BIT_MASK(64);
-
 static int __init octeon_ehci_device_init(void)
 {
 	struct platform_device *pd;
@@ -347,7 +346,6 @@ static int __init octeon_ehci_device_init(void)
 	if (!pd)
 		return 0;
 
-	pd->dev.dma_mask = &octeon_ehci_dma_mask;
 	pd->dev.platform_data = &octeon_ehci_pdata;
 	octeon_ehci_hw_start(&pd->dev);
 
drivers/usb/host/ehci-platform.c +2 -1
index 2da18ea..6df808b 100644
@@ -155,7 +155,8 @@ static int ehci_platform_probe(struct platform_device *dev)
 	if (!pdata)
 		pdata = &ehci_platform_defaults;
 
-	err = dma_coerce_mask_and_coherent(&dev->dev, DMA_BIT_MASK(32));
+	err = dma_coerce_mask_and_coherent(&dev->dev,
+		pdata->dma_mask_64 ? DMA_BIT_MASK(64) : DMA_BIT_MASK(32));
 	if (err)
 		return err;
 
include/linux/usb/ehci_pdriver.h +1 -0
index 7eb4dcd..f69529e 100644
@@ -45,6 +45,7 @@ struct usb_ehci_pdata {
 	unsigned	big_endian_desc:1;
 	unsigned	big_endian_mmio:1;
 	unsigned	no_io_watchdog:1;
+	unsigned	dma_mask_64:1;
 
 	/* Turn on all power and clocks */
 	int (*power_on)(struct platform_device *pdev);
Signature · 1 line
1.7.9.5
#11 lore raw

Re: [PATCH 3/3] USB: host: Introduce flag to enable use of 64-bit dma_mask for ehci-platform

2014-11-13 13:36 GMT-08:00 Andreas Herrmann
<[email protected]>:

Quoted 6 lines

ehci-octeon driver used a 64-bit dma_mask. With removal of ehci-octeon and usage of ehci-platform ehci dma_mask is now limited to 32 bits (coerced in ehci_platform_probe).

Provide a flag in ehci platform data to allow use of 64 bits for dma_mask.

Why not just allow enforcing an arbitrary DMA mask?

Quoted 70 lines
Cc David Daney <[email protected]>
Cc Alex Smith <[email protected]>
Cc Alan Stern <[email protected]>
---
 arch/mips/cavium-octeon/octeon-platform.c |    4 +---
 drivers/usb/host/ehci-platform.c          |    3 ++-
 include/linux/usb/ehci_pdriver.h          |    1 +
 3 files changed, 4 insertions(+), 4 deletions(-)
arch/mips/cavium-octeon/octeon-platform.c +1 -3
index eea60b6..12410a2 100644
@@ -310,6 +310,7 @@ static struct usb_ehci_pdata octeon_ehci_pdata = {
 #ifdef __BIG_ENDIAN
        .big_endian_mmio        = 1,
 #endif
+       .dma_mask_64    = 1,
        .power_on       = octeon_ehci_power_on,
        .power_off      = octeon_ehci_power_off,
 };
@@ -331,8 +332,6 @@ static void __init octeon_ehci_hw_start(struct device *dev)
        octeon2_usb_clocks_stop();
 }
 
-static u64 octeon_ehci_dma_mask = DMA_BIT_MASK(64);
-
 static int __init octeon_ehci_device_init(void)
 {
        struct platform_device *pd;
@@ -347,7 +346,6 @@ static int __init octeon_ehci_device_init(void)
        if (!pd)
                return 0;
 
-       pd->dev.dma_mask = &octeon_ehci_dma_mask;
        pd->dev.platform_data = &octeon_ehci_pdata;
        octeon_ehci_hw_start(&pd->dev);
 
drivers/usb/host/ehci-platform.c +2 -1
index 2da18ea..6df808b 100644
@@ -155,7 +155,8 @@ static int ehci_platform_probe(struct platform_device *dev)
        if (!pdata)
                pdata = &ehci_platform_defaults;
 
-       err = dma_coerce_mask_and_coherent(&dev->dev, DMA_BIT_MASK(32));
+       err = dma_coerce_mask_and_coherent(&dev->dev,
+               pdata->dma_mask_64 ? DMA_BIT_MASK(64) : DMA_BIT_MASK(32));
        if (err)
                return err;
 
include/linux/usb/ehci_pdriver.h +1 -1
index 7eb4dcd..f69529e 100644
@@ -45,6 +45,7 @@ struct usb_ehci_pdata {
        unsigned        big_endian_desc:1;
        unsigned        big_endian_mmio:1;
        unsigned        no_io_watchdog:1;
+       unsigned        dma_mask_64:1;
 
        /* Turn on all power and clocks */
        int (*power_on)(struct platform_device *pdev);
--

1.7.9.5

Signature · 1 line
Florian
#12 lore raw

Re: [PATCH 3/3] USB: host: Introduce flag to enable use of 64-bit dma_mask for ehci-platform

Quoted 10 lines · Florian Fainelli

2014-11-13 13:36 GMT-08:00 Andreas Herrmann
<[email protected]>:

Quoted 6 lines

ehci-octeon driver used a 64-bit dma_mask. With removal of ehci-octeon and usage of ehci-platform ehci dma_mask is now limited to 32 bits (coerced in ehci_platform_probe).

Provide a flag in ehci platform data to allow use of 64 bits for dma_mask.

Why not just allow enforcing an arbitrary DMA mask?

I thought about that but as it's currently just 32 or 64 bits a flag is sufficient. (At the moment I am not aware that other ehci-platform devices would require something else.)

I'll change the flag to a mask if desired.
Alan, what's your opinion about this?

Andreas

Quoted 74 lines
Quoted 69 lines
Cc David Daney <[email protected]>
Cc Alex Smith <[email protected]>
Cc Alan Stern <[email protected]>
---
 arch/mips/cavium-octeon/octeon-platform.c |    4 +---
 drivers/usb/host/ehci-platform.c          |    3 ++-
 include/linux/usb/ehci_pdriver.h          |    1 +
 3 files changed, 4 insertions(+), 4 deletions(-)
arch/mips/cavium-octeon/octeon-platform.c +1 -3
index eea60b6..12410a2 100644
@@ -310,6 +310,7 @@ static struct usb_ehci_pdata octeon_ehci_pdata = {
 #ifdef __BIG_ENDIAN
        .big_endian_mmio        = 1,
 #endif
+       .dma_mask_64    = 1,
        .power_on       = octeon_ehci_power_on,
        .power_off      = octeon_ehci_power_off,
 };
@@ -331,8 +332,6 @@ static void __init octeon_ehci_hw_start(struct device *dev)
        octeon2_usb_clocks_stop();
 }
 
-static u64 octeon_ehci_dma_mask = DMA_BIT_MASK(64);
-
 static int __init octeon_ehci_device_init(void)
 {
        struct platform_device *pd;
@@ -347,7 +346,6 @@ static int __init octeon_ehci_device_init(void)
        if (!pd)
                return 0;
 
-       pd->dev.dma_mask = &octeon_ehci_dma_mask;
        pd->dev.platform_data = &octeon_ehci_pdata;
        octeon_ehci_hw_start(&pd->dev);
 
drivers/usb/host/ehci-platform.c +2 -1
index 2da18ea..6df808b 100644
@@ -155,7 +155,8 @@ static int ehci_platform_probe(struct platform_device *dev)
        if (!pdata)
                pdata = &ehci_platform_defaults;
 
-       err = dma_coerce_mask_and_coherent(&dev->dev, DMA_BIT_MASK(32));
+       err = dma_coerce_mask_and_coherent(&dev->dev,
+               pdata->dma_mask_64 ? DMA_BIT_MASK(64) : DMA_BIT_MASK(32));
        if (err)
                return err;
 
include/linux/usb/ehci_pdriver.h +1 -1
index 7eb4dcd..f69529e 100644
@@ -45,6 +45,7 @@ struct usb_ehci_pdata {
        unsigned        big_endian_desc:1;
        unsigned        big_endian_mmio:1;
        unsigned        no_io_watchdog:1;
+       unsigned        dma_mask_64:1;
 
        /* Turn on all power and clocks */
        int (*power_on)(struct platform_device *pdev);
--

1.7.9.5

Signature · 1 line
Florian
maintainer
#13 lore raw

Re: [PATCH 3/3] USB: host: Introduce flag to enable use of 64-bit dma_mask for ehci-platform

Quoted 10 lines · Andreas Herrmann
Quoted 9 lines · Florian Fainelli

2014-11-13 13:36 GMT-08:00 Andreas Herrmann
<[email protected]>:

Quoted 6 lines

ehci-octeon driver used a 64-bit dma_mask. With removal of ehci-octeon and usage of ehci-platform ehci dma_mask is now limited to 32 bits (coerced in ehci_platform_probe).

Provide a flag in ehci platform data to allow use of 64 bits for dma_mask.

Quoted 8 lines
Quoted 1 line

Why not just allow enforcing an arbitrary DMA mask?

I thought about that but as it's currently just 32 or 64 bits a flag is sufficient. (At the moment I am not aware that other ehci-platform devices would require something else.)

I'll change the flag to a mask if desired.
Alan, what's your opinion about this?

I'm not aware of any devices that need a different DMA mask either.

Florian, do you have any reason for thinking such a thing might come along? Like Andreas, I don't mind making it more general if there's a good reason to do so.

Alan Stern

#14 lore raw

Re: [PATCH 3/3] USB: host: Introduce flag to enable use of 64-bit dma_mask for ehci-platform

Quoted 18 lines · Alan Stern
Quoted 16 lines · Andreas Herrmann
Quoted 10 lines · Florian Fainelli

2014-11-13 13:36 GMT-08:00 Andreas Herrmann
<[email protected]>:

Quoted 6 lines

ehci-octeon driver used a 64-bit dma_mask. With removal of ehci-octeon and usage of ehci-platform ehci dma_mask is now limited to 32 bits (coerced in ehci_platform_probe).

Provide a flag in ehci platform data to allow use of 64 bits for dma_mask.

Why not just allow enforcing an arbitrary DMA mask?

I thought about that but as it's currently just 32 or 64 bits a flag is sufficient. (At the moment I am not aware that other ehci-platform devices would require something else.)

Quoted 8 lines
Quoted 2 lines

I'll change the flag to a mask if desired.
Alan, what's your opinion about this?

I'm not aware of any devices that need a different DMA mask either.

Florian, do you have any reason for thinking such a thing might come along? Like Andreas, I don't mind making it more general if there's a good reason to do so.

I don't have a specific platform I am thinking about, just that while we are there allowing a dma_mask to be specified, I would rather pass an u64 that covers all possible cases.

Thanks!
--
Florian

maintainer
#15 lore raw
In reply to Andreas Herrmann
Quoted 31 lines · Andreas Herrmann

Hi Alan,

With following patches I want to base octeon ehci/ohci device configuration on device tree information.

I picked up patches that were submitted in May. See
http://marc.info/?l=linux-usb&m=140135823325811&w=2
and http://marc.info/?l=linux-mips&m=140139694721623&w=2

Patch #1 is your "untested preliminary pass" to remove
[oe]hci-octeon drivers.
Patch #2 is the removal of hard-coded platform information (but now
rebased on your patch)
Patch #3 adapts dma_mask for ehci (as used in ehci-octeon)

Overall diffstat is

 arch/mips/cavium-octeon/octeon-platform.c |  380 +++++++++++++++++++++++------
 arch/mips/configs/cavium_octeon_defconfig |    3 +
 drivers/usb/host/Kconfig                  |   18 +-
 drivers/usb/host/Makefile                 |    1 -
 drivers/usb/host/ehci-hcd.c               |    5 -
 drivers/usb/host/ehci-octeon.c            |  188 --------------
 drivers/usb/host/ehci-platform.c          |    4 +-
 drivers/usb/host/octeon2-common.c         |  200 ---------------
 drivers/usb/host/ohci-hcd.c               |    5 -
 drivers/usb/host/ohci-octeon.c            |  202 ---------------
 drivers/usb/host/ohci-platform.c          |    1 +
 include/linux/usb/ehci_pdriver.h          |    1 +
 12 files changed, 330 insertions(+), 678 deletions(-)
Quoted 3 lines

Patches are based on v3.18-rc4-65-g2c54396

Comments welcome.

At a very quick first glance, it looks great. Have you tested it thoroughly?

Alan Stern

#16 lore raw
Quoted 33 lines · Alan Stern
Quoted 31 lines · Andreas Herrmann

Hi Alan,

With following patches I want to base octeon ehci/ohci device configuration on device tree information.

I picked up patches that were submitted in May. See
http://marc.info/?l=linux-usb&m=140135823325811&w=2
and http://marc.info/?l=linux-mips&m=140139694721623&w=2

Patch #1 is your "untested preliminary pass" to remove
[oe]hci-octeon drivers.
Patch #2 is the removal of hard-coded platform information (but now
rebased on your patch)
Patch #3 adapts dma_mask for ehci (as used in ehci-octeon)

Overall diffstat is

 arch/mips/cavium-octeon/octeon-platform.c |  380 +++++++++++++++++++++++------
 arch/mips/configs/cavium_octeon_defconfig |    3 +
 drivers/usb/host/Kconfig                  |   18 +-
 drivers/usb/host/Makefile                 |    1 -
 drivers/usb/host/ehci-hcd.c               |    5 -
 drivers/usb/host/ehci-octeon.c            |  188 --------------
 drivers/usb/host/ehci-platform.c          |    4 +-
 drivers/usb/host/octeon2-common.c         |  200 ---------------
 drivers/usb/host/ohci-hcd.c               |    5 -
 drivers/usb/host/ohci-octeon.c            |  202 ---------------
 drivers/usb/host/ohci-platform.c          |    1 +
 include/linux/usb/ehci_pdriver.h          |    1 +
 12 files changed, 330 insertions(+), 678 deletions(-)
Quoted 6 lines
Quoted 3 lines

Patches are based on v3.18-rc4-65-g2c54396

Comments welcome.

At a very quick first glance, it looks great. Have you tested it thoroughly?

[sorry have to use another mail account, so far your mail didn't show
up at my caviumnetworks account]

I've tested it only on an EdgeRouterPro (Octeon II system, which I have on-site).

octeon_ehci_device_init and octeon_ohci_device_init run way before ehci-platform and ohci-platform probe for devices. So everything should be initialized orderly.

With current mainline (w/o these patches) USB doesn't work on my EdgeRouterPro due to an (inappropriate) OCTEON_IS_MODEL check.

I'd say having the patches in linux-next for awhile wouldn't hurt.

Andreas

Tested-by
#17 lore raw
Quoted 9 lines · Andreas Herrmann
Quoted 2 lines · Alan Stern

At a very quick first glance, it looks great. Have you tested it thoroughly?

[sorry have to use another mail account, so far your mail didn't show
up at my caviumnetworks account]

I've tested it only on an EdgeRouterPro (Octeon II system, which I have on-site).

I have also tested them on EdgeRouter Pro. Thanks, this is great improvement.

A.