Hi Prabhakar, > -----Original Message----- > From: Lad, Prabhakar <prabhakar.csengg@xxxxxxxxx> > Sent: 16 December 2024 13:14 > Subject: Re: [RFC PATCH] watchdog: rzv2h_wdt: Add support to retrieve the bootstatus information > > Hi Biju, > > On Sun, Dec 15, 2024 at 9:15 AM Biju Das <biju.das.jz@xxxxxxxxxxxxxx> wrote: > > > > Hi Prabhakar, > > > > > -----Original Message----- > > > From: Lad, Prabhakar <prabhakar.csengg@xxxxxxxxx> > > > Sent: 14 December 2024 21:36 > > > Subject: Re: [RFC PATCH] watchdog: rzv2h_wdt: Add support to > > > retrieve the bootstatus information > > > > > > Hi Biju, > > > > > > On Sat, Dec 14, 2024 at 11:32 AM Biju Das <biju.das.jz@xxxxxxxxxxxxxx> wrote: > > > > > > > > Hi Lad, Prabhakar, > > > > > > > <snip> > > > > > > That is a change in behavior. Up to now the syscon phandle did > > > > > > not have to exist for the driver to work. Is it guaranteed to > > > > > > not result in regressions on systems where it doesn't ? Also, > > > > > > is this documented ? I don't seem to be able to > > > > > find it. > > > > > > > > > > > Agreed. I will add a fallback mechanism to handle cases where > > > > > the syscon property is not present in the WDT node. This will > > > > > ensure no regressions occur, and the bootstatus will simply be > > > > > set to 0 in such scenarios. As mentioned in the patch comments, > > > > > I have not yet submitted the DT binding changes because I wanted > > > > > feedback on the syscon approach. The new RZ SoCs > > > have registers scattered across various locations, and I was > > > exploring if there might be a better way to handle this. > > > > > > > > See, syscon compatible not needed with [1] > > > > > > > > [1] > > > > https://lore.kernel.org/all/20241211-syscon-fixes-v1-3-b5ac8c219e9 > > > > 6@ke > > > > rnel.org/ > > > > > > > As per my understanding, `syscon` compatible is required in this > > > case because the CPG driver does not register a regmap. With the > > > patch [1] (linked above), this applies to drivers that register a syscon regmap, where the > corresponding DT node does not necessarily need a `syscon` compatible. > > > > I guess you can use "syscon_node_to_regmap" for that as > > of_syscon_register_regmap() is for externally created regmaps?? > > > No, it doesn't work either; the CPG node block needs to have `syscon` compatible when you are using > syscon_node_to_regmap/syscon_regmap_lookup_by_phandle. See patch [0] in `device_node_get_regmap()` the > `of_syscon_register()` is only called when it has a `syscon` compatible or else `-EINVAL` is returned. > > [0] https://lore.kernel.org/all/20241211-syscon-fixes-v1-3-b5ac8c219e96@xxxxxxxxxx/ Not sure the patch series, intention is to support only external regmap for those who does not have "syscon" compatible?? I thought the intention was, without "syscon" compatible, still it creates internal regmap, based on reg property of the system controller device node. Cheers, Biju
