[U-Boot] [PATCH 1/4 v5] watchdog: Implement generic watchdog_reset() version

[Bin Meng]

Hi Stefan,

On Fri, Aug 16, 2019 at 1:11 PM Stefan Roese <sr at denx.de> wrote:
>
> Hi Bin,
>
> On 15.08.19 16:19, Bin Meng wrote:
> > Hi Stefan,
> >
> > On Thu, Aug 15, 2019 at 2:07 PM Stefan Roese <sr at denx.de> wrote:
> >>
> >> Hi Simon,
> >>
> >> On 14.08.19 21:35, Simon Glass wrote:
> >>> Hi,
> >>>
> >>> On Wed, 14 Aug 2019 at 00:22, Stefan Roese <sr at denx.de> wrote:
> >>>>
> >>>> Hi Simon,
> >>>>
> >>>> (added Simon Glass and Bin to Cc)
> >>>>
> >>>> On 13.08.19 22:16, Simon Goldschmidt wrote:
> >>>>> Am 25.04.2019 um 09:17 schrieb Stefan Roese:
> >>>>>> This patch tries to implement a generic watchdog_reset() function that
> >>>>>> can be used by all boards that want to service the watchdog device in
> >>>>>> U-Boot. This watchdog servicing is enabled via CONFIG_WATCHDOG.
> >>>>>>
> >>>>>> Without this approach, new boards or platforms needed to implement a
> >>>>>> board specific version of this functionality, mostly copy'ing the same
> >>>>>> code over and over again into their board or platforms code base.
> >>>>>>
> >>>>>> With this new generic function, the scattered other functions are now
> >>>>>> removed to be replaced by the generic one. The new version also enables
> >>>>>> the configuration of the watchdog timeout via the DT "timeout-sec"
> >>>>>> property (if enabled via CONFIG_OF_CONTROL).
> >>>>>>
> >>>>>> This patch also adds a new flag to the GD flags, to flag that the
> >>>>>> watchdog is ready to use and adds the pointer to the watchdog device
> >>>>>> to the GD. This enables us to remove the global "watchdog_dev"
> >>>>>> variable, which was prone to cause problems because of its potentially
> >>>>>> very early use in watchdog_reset(), even before the BSS is cleared.
> >>>>>>
> >>>>>> Signed-off-by: Stefan Roese <sr at denx.de>
> >>>>>
> >>>>> <snip>
> >>>>>
> >>>>>> --- a/include/asm-generic/global_data.h
> >>>>>> +++ b/include/asm-generic/global_data.h
> >>>>>> @@ -133,6 +133,9 @@ typedef struct global_data {
> >>>>>>        struct spl_handoff *spl_handoff;
> >>>>>>      # endif
> >>>>>>      #endif
> >>>>>> +#if defined(CONFIG_WDT)
> >>>>>> +    struct udevice *watchdog_dev;
> >>>>>> +#endif
> >>>>>>      } gd_t;
> >>>>>>      #endif
> >>>>>>
> >>>>>> @@ -161,5 +164,6 @@ typedef struct global_data {
> >>>>>>      #define GD_FLG_ENV_DEFAULT        0x02000 /* Default variable flag           */
> >>>>>>      #define GD_FLG_SPL_EARLY_INIT     0x04000 /* Early SPL init is done          */
> >>>>>>      #define GD_FLG_LOG_READY  0x08000 /* Log system is ready for use     */
> >>>>>> +#define GD_FLG_WDT_READY    0x10000 /* Watchdog is ready for use       */
> >>>>>
> >>>>> Sorry to warm up a thread that is more than 4 months old, but I just
> >>>>> stumbled accross this line when searching for space in 'gd':
> >>>>>
> >>>>> The comment some lines above in global_data.h clearly states that the
> >>>>> top 16 bits of flags are reserved for arch-specific flags, and your
> >>>>> patch here uses the lowest of these 16 arch-specific flags for generic code.
> >>>>
> >>>> I totally missed this comment.
> >>>>
> >>>>> Is this a problem? Does any arch code use the upper 16 bits? I would
> >>>>> have thought you'd at least need to adjust the comment to reflect your
> >>>>> new usage...
> >>>>
> >>>> As stated above, I did not check about any other (arch-specific)
> >>>> GD_FLG_ definitions outside of this file.
> >>>>
> >>>>> The reason I ask is that I'd need a place to put some (~5?)
> >>>>> 'is_initialized' bits for some code running in SPL in the 'board_init_f'
> >>>>> code where BSS shouldn't be used. gd->flags would be ideal for that, but
> >>>>> I'm hesistant to dive in further into the 'arch-specific' upper 16 bits...
> >>>>
> >>>> And you should be. A quick grep shows that we already have a problem with
> >>>> my patch touching the upper bits:
> >>>>
> >>>> $ git grep "define GD_FLG_"
> >>>> arch/x86/include/asm/global_data.h:#define GD_FLG_COLD_BOOT     0x10000 /* Cold Boot */
> >>>> arch/x86/include/asm/global_data.h:#define GD_FLG_WARM_BOOT     0x20000 /* Warm Boot */
> >>>>
> >>>> This should definitely be fixed. I see 3 options right now:
> >>>>
> >>>> a) Reserve only the upper 8 bits for arch-specific stuff
> >>>> b) Use a new variable (gd->flags_arch ?) for this arch
> >>>> c) Remove the arch-specific GD_FLG's completely
> >>>>
> >>>> I can't tell if c) is doable - Bin and / or Simon Glass might know,
> >>>> if the x86 GD_FLG_foo_BOOT are really needed in gd->flags. I see that
> >>>> both are assigned in the .S files, but only GD_FLG_COLD_BOOT is
> >>>> referenced later on:
> >>>
> >>> Probably we can drop warm boot.
> >>
> >> Bin, do you think so as well?
> >>
> >
> > I believe we can drop these 2 flags completely. Currently usage of
> > warm/cold boot flags is only limited to coreboot codes.
> >
> > arch/x86/cpu/coreboot/coreboot.c::last_stage_init()
> >
> >          if (gd->flags & GD_FLG_COLD_BOOT)
> >                  timestamp_add_to_bootstage();
> >
> > timestamp_add_to_bootstage() will never be called for coreboot.
>
> Why is this the case? Will GD_FLG_COLD_BOOT never be set for the
> coreboot target?

GD_FLG_COLD_BOOT is only set in the 16-bit start code while on
coreboot it boots directly from the 32-bit start code.

Regards,
Bin