RE: [PATCH v6 1/1] watchdog: aspeed: Update bootstatus handling

Hi Andrew,

-----Original Message-----
From: Andrew Jeffery <andrew@codeconstruct.com.au>
Sent: Monday, January 13, 2025 11:29 AM
Subject: Re: [PATCH v6 1/1] watchdog: aspeed: Update bootstatus handling

On Sun, 2025-01-12 at 16:12 +0800, Chin-Ting Kuo wrote:

quoted

The boot status in the watchdog device struct is updated during
controller probe stage. Application layer can get the boot status
through the command, cat /sys/class/watchdog/watchdogX/bootstatus.
The bootstatus can be,
WDIOF_CARDRESET => the system is reset by WDT SoC reset.
Others          => other reset events, e.g., power on reset.

On ASPEED platform, the boot status is recorded in the SCU registers.
- AST2400: Only a bit represents for any WDT reset.
- AST2500/AST2600: The reset triggered by different WDT controllers
                   can be distinguished by different SCU bits.

Besides, on AST2400 and AST2500, since alternating boot event is
triggered by WDT SoC reset, it is classified as WDIOF_CARDRESET.

Signed-off-by: Chin-Ting Kuo <chin-ting_kuo@aspeedtech.com>
---
 drivers/watchdog/aspeed_wdt.c | 81
++++++++++++++++++++++++++++++++++-
 1 file changed, 79 insertions(+), 2 deletions(-)

diff --git a/drivers/watchdog/aspeed_wdt.c

b/drivers/watchdog/aspeed_wdt.c index b4773a6aaf8c..369635b38ca0
100644

--- a/drivers/watchdog/aspeed_wdt.c
+++ b/drivers/watchdog/aspeed_wdt.c

@@ -11,21 +11,30 @@

 #include <linux/io.h>
 #include <linux/kernel.h>
 #include <linux/kstrtox.h>
+#include <linux/mfd/syscon.h>
 #include <linux/module.h>
 #include <linux/of.h>
 #include <linux/of_irq.h>
 #include <linux/platform_device.h>
+#include <linux/regmap.h>
 #include <linux/watchdog.h>

 static bool nowayout = WATCHDOG_NOWAYOUT;
 module_param(nowayout, bool, 0);
 MODULE_PARM_DESC(nowayout, "Watchdog cannot be stopped once

started

quoted

(default="
                                __MODULE_STRING(WATC

HDOG_NOWAYOUT)

quoted

")");
+struct aspeed_wdt_scu {
+       const char *compatible;
+       u32 reset_status_reg;
+       u32 wdt_reset_mask;
+       u32 wdt_reset_mask_shift;
+};

 struct aspeed_wdt_config {
        u32 ext_pulse_width_mask;
        u32 irq_shift;
        u32 irq_mask;
+       struct aspeed_wdt_scu scu;
 };

 struct aspeed_wdt {

@@ -39,18 +48,36 @@ static const struct aspeed_wdt_config

ast2400_config = {
        .ext_pulse_width_mask = 0xff,
        .irq_shift = 0,
        .irq_mask = 0,
+       .scu = {
+               .compatible = "aspeed,ast2400-scu",
+               .reset_status_reg = 0x3c,
+               .wdt_reset_mask = 0x1,
+               .wdt_reset_mask_shift = 1,
+       },
 };

 static const struct aspeed_wdt_config ast2500_config = {
        .ext_pulse_width_mask = 0xfffff,
        .irq_shift = 12,
        .irq_mask = GENMASK(31, 12),
+       .scu = {
+               .compatible = "aspeed,ast2500-scu",
+               .reset_status_reg = 0x3c,
+               .wdt_reset_mask = 0x1,
+               .wdt_reset_mask_shift = 2,
+       },
 };

 static const struct aspeed_wdt_config ast2600_config = {
        .ext_pulse_width_mask = 0xfffff,
        .irq_shift = 0,
        .irq_mask = GENMASK(31, 10),
+       .scu = {
+               .compatible = "aspeed,ast2600-scu",
+               .reset_status_reg = 0x74,
+               .wdt_reset_mask = 0xf,
+               .wdt_reset_mask_shift = 16,
+       },
 };

 static const struct of_device_id aspeed_wdt_of_table[] = { @@ -213,6
+240,56 @@ static int aspeed_wdt_restart(struct watchdog_device *wdd,
        return 0;
 }

+static void aspeed_wdt_update_bootstatus(struct platform_device
*pdev,
+                                        struct

aspeed_wdt *wdt) {

quoted

+       const struct resource *res;
+       struct aspeed_wdt_scu scu = wdt->cfg->scu;
+       struct regmap *scu_base;
+       u32 reset_mask_width;
+       u32 reset_mask_shift;
+       u32 idx = 0;
+       u32 status;
+       int ret;
+
+       if (!of_device_is_compatible(pdev->dev.of_node,
"aspeed,ast2400-wdt")) {
+               res = platform_get_resource(pdev,

IORESOURCE_MEM, 0);

quoted

+               idx = ((intptr_t)wdt->base & 0x00000fff) /
resource_size(res);
+       }
+
+       scu_base =
syscon_regmap_lookup_by_compatible(scu.compatible);
+       if (IS_ERR(scu_base)) {
+               wdt->wdd.bootstatus = WDIOS_UNKNOWN;
+               return;
+       }
+
+       ret = regmap_read(scu_base, scu.reset_status_reg, &status);
+       if (ret) {
+               wdt->wdd.bootstatus = WDIOS_UNKNOWN;
+               return;
+       }
+
+       reset_mask_width = hweight32(scu.wdt_reset_mask);
+       reset_mask_shift = scu.wdt_reset_mask_shift +
+                          reset_mask_width * idx;
+
+       if (status & (scu.wdt_reset_mask << reset_mask_shift))
+               wdt->wdd.bootstatus = WDIOF_CARDRESET;

How precise is the wording in your commit message? The AST2600, for
instance, has 4 reset types:

1. "SoC"
2. "Full"
3. "ARM"
4. "Software"

In the commit message, you said:

quoted

WDIOF_CARDRESET => the system is reset by WDT SoC reset.

The commit message should be changed for more precisely.
"WDIOF_CARDRESET => the system is reset due to WDT timeout occurs."
This change can be updated with a new patch version.

However the logic here (with the way you've initialised the config struct for the
AST2600) will signal WDIOF_CARDRESET even if what occurred was e.g. a
(graceful?) software reset.

As the discussion in the previous patch series, there is no consensus for graceful
reset flag in WDT framework. Thus, this patch only distinguishes WDT reset from
other reset reasons.

The only thing WDIOF_CARDRESET differentiates is a power-on reset from any
other kind of watchdog-driven reset.

Yes.

Is that what's intended? I'm just wary of confusion for what appears to be a
generic use of "SoC" in the commit message vs the specific "SoC"
mode of the watchdog controller (should some documentation be
written/updated?)

The commit message can be updated.

Chin-Ting