[U-Boot] Marvell Armada XP SATA support
Anton Schubert
anton.schubert at gmx.de
Wed Jul 1 12:20:06 CEST 2015
Hi,
I figured out my problem was just ATA security. After unlocking the code
above works fine for MV78260.
Thanks,
Anton
2015-06-30 11:28 GMT+02:00 Anton Schubert <anton.schubert at gmx.de>:
> That should be it. Although it's probably a good idea to do this in cpu.c
> as you did with your a38x ahci support.
> Thanks for your help.
>
> Anton
>
> ---
> arch/arm/mach-mvebu/include/mach/soc.h | 1 +
> drivers/block/mvsata_ide.c | 46
> ++++++++++++++++++++++++++++++++++
> include/configs/db-mv784mp-gp.h | 29 +++++++++++++++++++++
> 3 files changed, 76 insertions(+)
>
> diff --git a/arch/arm/mach-mvebu/include/mach/soc.h
> b/arch/arm/mach-mvebu/include/mach/soc.h
> index 0a9307c..2a8234d 100644
> --- a/arch/arm/mach-mvebu/include/mach/soc.h
> +++ b/arch/arm/mach-mvebu/include/mach/soc.h
> @@ -51,6 +51,7 @@
> #define MVEBU_REG_PCIE_BASE (MVEBU_REGISTER(0x40000))
> #define MVEBU_EGIGA0_BASE (MVEBU_REGISTER(0x70000))
> #define MVEBU_EGIGA1_BASE (MVEBU_REGISTER(0x74000))
> +#define MVEBU_SATA_BASE (MVEBU_REGISTER(0xA0000))
>
> #define SDRAM_MAX_CS 4
> #define SDRAM_ADDR_MASK 0xFF000000
> diff --git a/drivers/block/mvsata_ide.c b/drivers/block/mvsata_ide.c
> index e54d564..ace061a 100644
> --- a/drivers/block/mvsata_ide.c
> +++ b/drivers/block/mvsata_ide.c
> @@ -13,6 +13,8 @@
> #include <asm/arch/orion5x.h>
> #elif defined(CONFIG_KIRKWOOD)
> #include <asm/arch/soc.h>
> +#elif defined(CONFIG_ARMADA_XP)
> +#include <linux/mbus.h>
> #endif
>
> /* SATA port registers */
> @@ -90,6 +92,46 @@ struct mvsata_port_registers {
> #define MVSATA_STATUS_TIMEOUT -1
>
> /*
> + * Registers for SATA MBUS memory windows
> + */
> +
> +#define MVSATA_WIN_CONTROL(w) (MVEBU_SATA_BASE + 0x30 + ((w) << 4))
> +#define MVSATA_WIN_BASE(w) (MVEBU_SATA_BASE + 0x34 + ((w) << 4))
> +
> +/*
> + * Initialize SATA memory windows for Armada XP
> + */
> +
> +#ifdef CONFIG_ARMADA_XP
> +static void mvsata_ide_conf_mbus_windows(void)
> +{
> + const struct mbus_dram_target_info *dram;
> + int i;
> +
> + dram = mvebu_mbus_dram_info();
> +
> + /* Disable windows, Set Size/Base to 0 */
> + for (i = 0; i < 4; i++) {
> + writel(0, MVSATA_WIN_CONTROL(i));
> + writel(0, MVSATA_WIN_BASE(i));
> + }
> +
> + for (i = 0; i < dram->num_cs; i++) {
> + const struct mbus_dram_window *cs = dram->cs + i;
> +
> + printf("%s: sata window %d\n", __func__, i);
> + printf("%s: size %04x, attr %02x, target %01x\n", __func__,
> cs->size, cs->mbus_attr, dram->mbus_dram_target_id);
> + printf("%s: base %04x\n\n", __func__, cs->base);
> +
> + writel(((cs->size - 1) & 0xffff0000) | (cs->mbus_attr << 8) |
> + (dram->mbus_dram_target_id << 4) | 1,
> + MVSATA_WIN_CONTROL(i));
> + writel(cs->base & 0xffff0000, MVSATA_WIN_BASE(i));
> + }
> +}
> +#endif
> +
> +/*
> * Initialize one MVSATAHC port: set SControl's IPM to "always active"
> * and DET to "reset", then wait for SStatus's DET to become "device and
> * comm ok" (or time out after 50 us if no device), then set SControl's
> @@ -137,6 +179,10 @@ int ide_preinit(void)
> int ret = MVSATA_STATUS_TIMEOUT;
> int status;
>
> +#ifdef CONFIG_ARMADA_XP
> + mvsata_ide_conf_mbus_windows();
> +#endif
> +
> /* Enable ATA port 0 (could be SATA port 0 or 1) if declared */
> #if defined(CONFIG_SYS_ATA_IDE0_OFFSET)
> status = mvsata_ide_initialize_port(
> diff --git a/include/configs/db-mv784mp-gp.h
> b/include/configs/db-mv784mp-gp.h
> index c33a588..299bc82 100644
> --- a/include/configs/db-mv784mp-gp.h
> +++ b/include/configs/db-mv784mp-gp.h
> @@ -32,6 +32,8 @@
> #define CONFIG_CMD_SPI
> #define CONFIG_CMD_TFTPPUT
> #define CONFIG_CMD_TIME
> +#define CONFIG_CMD_BOOTZ
> +#define CONFIG_CMD_IDE
>
> /* I2C */
> #define CONFIG_SYS_I2C
> @@ -60,6 +62,33 @@
> #define CONFIG_SYS_CONSOLE_INFO_QUIET /* don't print console @ startup
> */
> #define CONFIG_SYS_ALT_MEMTEST
>
> +/* IDE/ATA/SATA */
> +#ifdef CONFIG_CMD_IDE
> +#define __io
> +#define CONFIG_IDE_PREINIT
> +#define CONFIG_MVSATA_IDE
> +
> +/* Needs byte-swapping for ATA data register */
> +#define CONFIG_IDE_SWAP_IO
> +
> +#define CONFIG_SYS_ATA_REG_OFFSET 0x0100 /* Offset for normal
> register accesses*/
> +#define CONFIG_SYS_ATA_DATA_OFFSET 0x0100 /* Offset for data I/O
> */
> +#define CONFIG_SYS_ATA_ALT_OFFSET 0x0100 /* Offset for data I/O */
> +
> +/* Each 8-bit ATA register is aligned to a 4-bytes address */
> +#define CONFIG_SYS_ATA_STRIDE 4
> +
> +/* CONFIG_CMD_IDE requires some #defines for ATA registers */
> +#define CONFIG_SYS_IDE_MAXBUS 2
> +#define CONFIG_SYS_IDE_MAXDEVICE 2
> +
> +/* ATA registers base is at SATA controller base */
> +#define CONFIG_SYS_ATA_BASE_ADDR MVEBU_SATA_BASE
> +#define CONFIG_SYS_ATA_IDE0_OFFSET 0x2000
> +
> +#define CONFIG_DOS_PARTITION
> +#endif /* CONFIG_CMD_IDE */
> +
> /*
> * mv-common.h should be defined after CMD configs since it used them
> * to enable certain macros
> --
> 2.3.6
>
>
> 2015-06-30 8:57 GMT+02:00 Stefan Roese <sr at denx.de>:
>
>> Hi Anton,
>>
>> On 29.06.2015 19:01, Anton Schubert wrote:
>>
>>> 2015-06-23 14:38 GMT+02:00 Stefan Roese <sr at denx.de <mailto:sr at denx.de
>>> >>:
>>>
>>> Hi Anton,
>>>
>>> On 23.06.2015 13:23, Bin Meng wrote:
>>>
>>> On Tue, Jun 23, 2015 at 5:35 PM, Anton Schubert
>>> <anton.schubert at gmx.de <mailto:anton.schubert at gmx.de>> wrote
>>>
>>>
>>>
>>> but noticed that there doesn't seem to be a driver for the
>>> SATA
>>> controller.
>>> Are there any plans to support SATA on Armada XP or is our
>>> only option to
>>> migrate the old driver?
>>>
>>>
>>> Is the Marvell SATA controller AHCI compatible? If yes, U-Boot
>>> has the
>>> AHCI driver which should work.
>>>
>>>
>>> No, the AXP SATA controller is unfortunately not AHCI compatible
>>> (the A38x seems to be though). The mvsata driver
>>> (drivers/block/mvsata_ide.c) might be a pretty good start for the
>>> AXP though (I didn't check this closely).
>>>
>>>
>>> Hi,
>>> many thanks for your hints.
>>>
>>> I managed to get ide_ident working with mvsata_ide.c by setting the Sata
>>> address windows (using mbus.c analogous to the mvneta driver). But I'm
>>> kind of stuck now as ATA reads will just result in timeouts:
>>>
>>> ide_read dev 0 start 0, blocks 1 buffer at 7FBA9C60
>>> ide_outb (dev= 0, port= 0x118, val= 0xe0) : @ 0xf10a2118
>>> ide_inb (dev= 0, port= 0x11c) : @ 0xf10a211c -> 0x50
>>> ide_outb (dev= 0, port= 0x11c, val= 0xe5) : @ 0xf10a211c
>>> ide_inb (dev= 0, port= 0x11c) : @ 0xf10a211c -> 0x50
>>> ide_inb (dev= 0, port= 0x108) : @ 0xf10a2108 -> 0xff
>>> Powersaving FF
>>> ide_inb (dev= 0, port= 0x11c) : @ 0xf10a211c -> 0x50
>>> ide_outb (dev= 0, port= 0x108, val= 0x01) : @ 0xf10a2108
>>> ide_outb (dev= 0, port= 0x10c, val= 0x00) : @ 0xf10a210c
>>> ide_outb (dev= 0, port= 0x110, val= 0x00) : @ 0xf10a2110
>>> ide_outb (dev= 0, port= 0x114, val= 0x00) : @ 0xf10a2114
>>> ide_outb (dev= 0, port= 0x118, val= 0xe0) : @ 0xf10a2118
>>> ide_outb (dev= 0, port= 0x11c, val= 0x20) : @ 0xf10a211c
>>> ide_inb (dev= 0, port= 0x11c) : @ 0xf10a211c -> 0xd0
>>> ide_inb (dev= 0, port= 0x11c) : @ 0xf10a211c -> 0x51
>>> Error (no IRQ) dev 0 blk 0: status 0x51
>>>
>>>
>>> This is my current ide config:
>>> #define __io
>>> #define CONFIG_IDE_PREINIT
>>> #define CONFIG_MVSATA_IDE
>>>
>>> /* Needs byte-swapping for ATA data register */
>>> #define CONFIG_IDE_SWAP_IO
>>>
>>> #define CONFIG_SYS_ATA_REG_OFFSET 0x0100 /* Offset for
>>> normal register accesses*/
>>> #define CONFIG_SYS_ATA_DATA_OFFSET 0x0100 /* Offset for data I/O
>>> */
>>> #define CONFIG_SYS_ATA_ALT_OFFSET 0x0100 /* Offset for data
>>> I/O */
>>>
>>> /* Each 8-bit ATA register is aligned to a 4-bytes address */
>>> #define CONFIG_SYS_ATA_STRIDE 4
>>>
>>> /* CONFIG_CMD_IDE requires some #defines for ATA registers */
>>> #define CONFIG_SYS_IDE_MAXBUS 2
>>> #define CONFIG_SYS_IDE_MAXDEVICE 2
>>>
>>> /* ATA registers base is at SATA controller base */
>>> #define CONFIG_SYS_ATA_BASE_ADDR MVEBU_SATA_BASE /* 0xA0000 */
>>> #define CONFIG_SYS_ATA_IDE0_OFFSET 0x2000
>>>
>>> #define CONFIG_DOS_PARTITION
>>>
>>> I would be grateful for any pointers.
>>>
>>
>> Just send me the missing code (mbus address window setup etc) and I'll
>> give it a try on my DB-MV784MP-GP board.
>>
>> Thanks,
>> Stefan
>>
>>
>
More information about the U-Boot
mailing list