On Wed, 2019-08-21 at 21:15 +0300, Tero Kristo wrote: > On 21.8.2019 18.45, Suman Anna wrote: > > On 8/21/19 10:10 AM, Philipp Zabel wrote: [...] > > > In general, assuming the device tree contains no errors, this should not > > > matter much, but I think it is nice if the reset driver, even with a > > > misconfigured device tree, can't write into arbitrary bit fields. > > > > Tero, > > Can you add a check for this if possible? > > Well, I can enforce the usage of reset bit mapping, which I have already > implemented for some SoCs like am33xx. If the specific ID is not found, > I can bail out. So, basically in this example requesting reset at index > 3 would succeed, but it would fail for any other ID; this would be > direct HW bit mapping. That should be fine. regards Philipp
