| Message ID | 20250323113544.7933-14-andre.przywara@arm.com |
|---|---|
| State | New |
| Delegated to: | Andre Przywara |
| Headers | show |
| Series | sunxi: clock refactoring and Allwinner A523 support | expand |
Dne nedelja, 23. marec 2025 ob 12:35:23 Srednjeevropski standardni čas je Andre Przywara napisal(a): > From: Jernej Skrabec <jernej.skrabec@gmail.com> > > The watchdog in the Allwinner A523 SoC differs a bit from the one in the > previous SoCs: it lives in a separate register frame, so no longer > inside some timer device, and it manages to shuffle around some > registers a bit. > > Provide a new struct describing the register layout, and adjust the > address calculation in the SPL code accoringly. This is guarded by the > MACH_SUN55I_A523 Kconfig variable. This one is missing SoB, apparently by me :) > --- > arch/arm/include/asm/arch-sunxi/watchdog.h | 12 ++++++++++++ > arch/arm/mach-sunxi/board.c | 6 ++++++ > 2 files changed, 18 insertions(+) > > diff --git a/arch/arm/include/asm/arch-sunxi/watchdog.h b/arch/arm/include/asm/arch-sunxi/watchdog.h > index 38e2ef2aca3..96d5725141e 100644 > --- a/arch/arm/include/asm/arch-sunxi/watchdog.h > +++ b/arch/arm/include/asm/arch-sunxi/watchdog.h > @@ -26,6 +26,18 @@ struct sunxi_wdog { > u32 res[2]; > }; > > +#elif defined(CONFIG_MACH_SUN55I_A523) > + > +struct sunxi_wdog { > + u32 irq_en; /* 0x00 */ > + u32 irq_sta; /* 0x04 */ > + u32 srst; /* 0x08 */ > + u32 ctl; /* 0x0c */ > + u32 cfg; /* 0x10 */ > + u32 mode; /* 0x14 */ > + u32 ocfg; /* 0x18 */ > +}; Anyway, since you announced battle against C structures for register layout, it seems contraproductive to add it here. What do you think? Best regards, Jernej > + > #else > > #define WDT_CFG_RESET (0x1) > diff --git a/arch/arm/mach-sunxi/board.c b/arch/arm/mach-sunxi/board.c > index 701899ee4b2..89aea61e8e8 100644 > --- a/arch/arm/mach-sunxi/board.c > +++ b/arch/arm/mach-sunxi/board.c > @@ -495,6 +495,12 @@ void reset_cpu(void) > /* sun5i sometimes gets stuck without this */ > writel(WDT_MODE_RESET_EN | WDT_MODE_EN, &wdog->mode); > } > +#elif defined(CONFIG_MACH_SUN55I_A523) > + static const struct sunxi_wdog *wdog = > + (struct sunxi_wdog *)SUNXI_TIMER_BASE; > + > + writel(WDT_CTRL_KEY | WDT_CTRL_RESTART, &wdog->srst); > + while (1) { } > #elif defined(CONFIG_SUNXI_GEN_SUN6I) || defined(CONFIG_SUN50I_GEN_H6) || defined(CONFIG_SUNXI_GEN_NCAT2) > #if defined(CONFIG_MACH_SUN50I_H6) > /* WDOG is broken for some H6 rev. use the R_WDOG instead */ >
On Sun, 23 Mar 2025 13:15:46 +0100 Jernej Škrabec <jernej.skrabec@gmail.com> wrote: Hi, > Dne nedelja, 23. marec 2025 ob 12:35:23 Srednjeevropski standardni čas je Andre Przywara napisal(a): > > From: Jernej Skrabec <jernej.skrabec@gmail.com> > > > > The watchdog in the Allwinner A523 SoC differs a bit from the one in the > > previous SoCs: it lives in a separate register frame, so no longer > > inside some timer device, and it manages to shuffle around some > > registers a bit. > > > > Provide a new struct describing the register layout, and adjust the > > address calculation in the SPL code accoringly. This is guarded by the > > MACH_SUN55I_A523 Kconfig variable. > > This one is missing SoB, apparently by me :) Yes, I don't think it's appropriate to add your Signed-off-by, so if you could give it here. > > > --- > > arch/arm/include/asm/arch-sunxi/watchdog.h | 12 ++++++++++++ > > arch/arm/mach-sunxi/board.c | 6 ++++++ > > 2 files changed, 18 insertions(+) > > > > diff --git a/arch/arm/include/asm/arch-sunxi/watchdog.h b/arch/arm/include/asm/arch-sunxi/watchdog.h > > index 38e2ef2aca3..96d5725141e 100644 > > --- a/arch/arm/include/asm/arch-sunxi/watchdog.h > > +++ b/arch/arm/include/asm/arch-sunxi/watchdog.h > > @@ -26,6 +26,18 @@ struct sunxi_wdog { > > u32 res[2]; > > }; > > > > +#elif defined(CONFIG_MACH_SUN55I_A523) > > + > > +struct sunxi_wdog { > > + u32 irq_en; /* 0x00 */ > > + u32 irq_sta; /* 0x04 */ > > + u32 srst; /* 0x08 */ > > + u32 ctl; /* 0x0c */ > > + u32 cfg; /* 0x10 */ > > + u32 mode; /* 0x14 */ > > + u32 ocfg; /* 0x18 */ > > +}; > > Anyway, since you announced battle against C structures for register > layout, it seems contraproductive to add it here. What do you think? Well, I am all for it, but I guess that's another refactor patch first, so please be my guest. I didn't deem this important enough to address this now (it's not blocking as for the clocks), and this series is already long enough, so I was just happy to take this working patch. For development I really like having a working "reset" command ;-) Cheers, Andre > > Best regards, > Jernej > > > + > > #else > > > > #define WDT_CFG_RESET (0x1) > > diff --git a/arch/arm/mach-sunxi/board.c b/arch/arm/mach-sunxi/board.c > > index 701899ee4b2..89aea61e8e8 100644 > > --- a/arch/arm/mach-sunxi/board.c > > +++ b/arch/arm/mach-sunxi/board.c > > @@ -495,6 +495,12 @@ void reset_cpu(void) > > /* sun5i sometimes gets stuck without this */ > > writel(WDT_MODE_RESET_EN | WDT_MODE_EN, &wdog->mode); > > } > > +#elif defined(CONFIG_MACH_SUN55I_A523) > > + static const struct sunxi_wdog *wdog = > > + (struct sunxi_wdog *)SUNXI_TIMER_BASE; > > + > > + writel(WDT_CTRL_KEY | WDT_CTRL_RESTART, &wdog->srst); > > + while (1) { } > > #elif defined(CONFIG_SUNXI_GEN_SUN6I) || defined(CONFIG_SUN50I_GEN_H6) || defined(CONFIG_SUNXI_GEN_NCAT2) > > #if defined(CONFIG_MACH_SUN50I_H6) > > /* WDOG is broken for some H6 rev. use the R_WDOG instead */ > > > > > > >
diff --git a/arch/arm/include/asm/arch-sunxi/watchdog.h b/arch/arm/include/asm/arch-sunxi/watchdog.h index 38e2ef2aca3..96d5725141e 100644 --- a/arch/arm/include/asm/arch-sunxi/watchdog.h +++ b/arch/arm/include/asm/arch-sunxi/watchdog.h @@ -26,6 +26,18 @@ struct sunxi_wdog { u32 res[2]; }; +#elif defined(CONFIG_MACH_SUN55I_A523) + +struct sunxi_wdog { + u32 irq_en; /* 0x00 */ + u32 irq_sta; /* 0x04 */ + u32 srst; /* 0x08 */ + u32 ctl; /* 0x0c */ + u32 cfg; /* 0x10 */ + u32 mode; /* 0x14 */ + u32 ocfg; /* 0x18 */ +}; + #else #define WDT_CFG_RESET (0x1) diff --git a/arch/arm/mach-sunxi/board.c b/arch/arm/mach-sunxi/board.c index 701899ee4b2..89aea61e8e8 100644 --- a/arch/arm/mach-sunxi/board.c +++ b/arch/arm/mach-sunxi/board.c @@ -495,6 +495,12 @@ void reset_cpu(void) /* sun5i sometimes gets stuck without this */ writel(WDT_MODE_RESET_EN | WDT_MODE_EN, &wdog->mode); } +#elif defined(CONFIG_MACH_SUN55I_A523) + static const struct sunxi_wdog *wdog = + (struct sunxi_wdog *)SUNXI_TIMER_BASE; + + writel(WDT_CTRL_KEY | WDT_CTRL_RESTART, &wdog->srst); + while (1) { } #elif defined(CONFIG_SUNXI_GEN_SUN6I) || defined(CONFIG_SUN50I_GEN_H6) || defined(CONFIG_SUNXI_GEN_NCAT2) #if defined(CONFIG_MACH_SUN50I_H6) /* WDOG is broken for some H6 rev. use the R_WDOG instead */
From: Jernej Skrabec <jernej.skrabec@gmail.com> The watchdog in the Allwinner A523 SoC differs a bit from the one in the previous SoCs: it lives in a separate register frame, so no longer inside some timer device, and it manages to shuffle around some registers a bit. Provide a new struct describing the register layout, and adjust the address calculation in the SPL code accoringly. This is guarded by the MACH_SUN55I_A523 Kconfig variable. --- arch/arm/include/asm/arch-sunxi/watchdog.h | 12 ++++++++++++ arch/arm/mach-sunxi/board.c | 6 ++++++ 2 files changed, 18 insertions(+)