[PATCH 1/7] ARM: zynqmp: set reset source

[Ahmad Fatoum]

On 24.06.21 15:20, Michael Tretter wrote:
> On Thu, 24 Jun 2021 12:48:44 +0200, Ahmad Fatoum wrote:
>> On 24.06.21 12:23, Michael Tretter wrote:
>>> The reset reason is available in the APB register set on the ZynqMP.
>>> Read the reset reason and set the reset source accordingly.
>>>
>>> There might be multiple bits set in the APB register. Use the MSB for
>>> determining the actual reset source.
>>
>> APB is usually the AMBA Advanced Peripheral Bus. Perhaps CRL is the
>> actual name of the register and APB just tells that it's mapped there?
> 
> Ack. The data sheet calls it "Clock and Reset control registers for LPD." I
> will fix it in v2.
> 
>>
>>>
>>> Signed-off-by: Michael Tretter <m.tretter at pengutronix.de>
>>> ---
>>>  arch/arm/mach-zynqmp/Makefile              |  1 +
>>>  arch/arm/mach-zynqmp/include/mach/zynqmp.h |  6 ++
>>>  arch/arm/mach-zynqmp/zynqmp.c              | 74 ++++++++++++++++++++++
>>>  3 files changed, 81 insertions(+)
>>>  create mode 100644 arch/arm/mach-zynqmp/include/mach/zynqmp.h
>>>  create mode 100644 arch/arm/mach-zynqmp/zynqmp.c
>>>
>>> diff --git a/arch/arm/mach-zynqmp/Makefile b/arch/arm/mach-zynqmp/Makefile
>>> index 021efc94afaf..14b8a4e46b87 100644
>>> --- a/arch/arm/mach-zynqmp/Makefile
>>> +++ b/arch/arm/mach-zynqmp/Makefile
>>> @@ -1,2 +1,3 @@
>>>  # SPDX-License-Identifier: GPL-2.0-or-later
>>>  obj-y += firmware-zynqmp.o
>>> +obj-y += zynqmp.o
>>> diff --git a/arch/arm/mach-zynqmp/include/mach/zynqmp.h b/arch/arm/mach-zynqmp/include/mach/zynqmp.h
>>> new file mode 100644
>>> index 000000000000..f6c05f35a470
>>> --- /dev/null
>>> +++ b/arch/arm/mach-zynqmp/include/mach/zynqmp.h
>>
>> Nitpick: revision.h sounds more self-describing.
> 
> Ok. I thought there might be some more soc specific function instead of
> revision specific functions here, but revision.h is fine for me as well.
> 
>>
>>> @@ -0,0 +1,6 @@
>>> +#ifndef __MACH_ZYNQMP_H
>>> +#define __MACH_ZYNQMP_H
>>> +
>>> +int zynqmp_soc_revision(void);
>>> +
>>> +#endif /* __MACH_ZYNQMP_H */
>>> diff --git a/arch/arm/mach-zynqmp/zynqmp.c b/arch/arm/mach-zynqmp/zynqmp.c
>>> new file mode 100644
>>> index 000000000000..2b3bd8406ce9
>>> --- /dev/null
>>> +++ b/arch/arm/mach-zynqmp/zynqmp.c
>>> @@ -0,0 +1,74 @@
>>> +// SPDX-License-Identifier: GPL-2.0-only
>>> +/*
>>> + * Copyright (C) 2020 Michael Tretter <m.tretter at pengutronix.de>
>>> + */
>>> +
>>> +#include <common.h>
>>> +#include <init.h>
>>> +#include <linux/types.h>
>>> +#include <reset_source.h>
>>> +
>>> +#include <mach/zynqmp.h>
>>> +
>>> +#define ZYNQMP_CRL_APB_BASE		0xff5e0000
>>> +#define ZYNQMP_CRL_APB_RESET_REASON	(ZYNQMP_CRL_APB_BASE + 0x220)
>>> +
>>> +/* External POR: The PS_POR_B reset signal pin was asserted. */
>>> +#define ZYNQMP_CRL_APB_RESET_REASON_EXTERNAL   BIT(0)
>>> +/* Internal POR: A system error triggered a POR reset. */
>>> +#define ZYNQMP_CRL_APB_RESET_REASON_INTERNAL   BIT(1)
>>> +/* Internal system reset; A system error triggered a system reset. */
>>> +#define ZYNQMP_CRL_APB_RESET_REASON_PMU        BIT(2)
>>> +/* PS-only reset: Write to PMU_GLOBAL.GLOBAL_RESET [PS_ONLY_RST]. */
>>> +#define ZYNQMP_CRL_APB_RESET_REASON_PSONLY     BIT(3)
>>> +/* External system reset: The PS_SRST_B reset signal pin was asserted. */
>>> +#define ZYNQMP_CRL_APB_RESET_REASON_SRST       BIT(4)
>>> +/* Software system reset: Write to RESET_CTRL [soft_reset]. */
>>> +#define ZYNQMP_CRL_APB_RESET_REASON_SOFT       BIT(5)
>>> +/* Software debugger reset: Write to BLOCKONLY_RST [debug_only]. */
>>> +#define ZYNQMP_CRL_APB_RESET_REASON_DEBUG_SYS  BIT(6)
>>> +
>>> +struct zynqmp_reset_reason {
>>> +	u32 mask;
>>> +	enum reset_src_type type;
>>> +};
>>> +
>>> +static const struct zynqmp_reset_reason reset_reasons[] = {
>>> +	{ ZYNQMP_CRL_APB_RESET_REASON_DEBUG_SYS,	RESET_UKWN },
>>
>> RESET_JTAG?
> 
> I am not sure, if I understand RESET_JTAG correctly. The reference manual says
> that it is like a soft reset while preserving the debug logic and originates
> in the DAP controller. Is this RESET_JTAG?

Yes. DAP may be used with SWD as well, but as far as barebox reset reasons are
concerned, every debugger related reset should be RESET_JTAG.

> 
>>
>>> +	{ ZYNQMP_CRL_APB_RESET_REASON_SOFT,		RESET_RST },
>>> +	{ ZYNQMP_CRL_APB_RESET_REASON_SRST,		RESET_POR },
>>> +	{ ZYNQMP_CRL_APB_RESET_REASON_PSONLY,		RESET_POR },
>>> +	{ ZYNQMP_CRL_APB_RESET_REASON_PMU,		RESET_POR },
>>> +	{ ZYNQMP_CRL_APB_RESET_REASON_INTERNAL,		RESET_POR },
>>> +	{ ZYNQMP_CRL_APB_RESET_REASON_EXTERNAL,		RESET_POR },
>>
>> RESET_EXT?
> 
> This bit is set for a normal POR, too.
> 
>>
>>> +	{ /* sentinel */ }
>>> +};
>>> +
>>> +static enum reset_src_type zynqmp_get_reset_src(void)
>>> +{
>>> +	enum reset_src_type type = RESET_UKWN;
>>> +	unsigned int i;
>>> +	u32 val;
>>> +
>>> +	val = readl(ZYNQMP_CRL_APB_RESET_REASON);
>>> +
>>> +	for (i = 0; i < ARRAY_SIZE(reset_reasons); i++) {
>>> +		if (val & reset_reasons[i].mask) {
>>> +			type = reset_reasons[i].type;
>>> +			break;
>>> +		}
>>> +	}
>>> +
>>> +	pr_info("ZynqMP reset reason %s (ZYNQMP_CRL_APB_RESET_REASON: 0x%08x)\n",
>>> +		reset_source_to_string(type), val);
>>> +
>>> +	return type;
>>> +}
>>> +
>>> +static int zynqmp_init(void)
>>> +{
>>> +	reset_source_set(zynqmp_get_reset_src());
>>> +
>>> +	return 0;
>>> +}
>>> +postcore_initcall(zynqmp_init);
>>>
> 

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |