> On suspend and resume, pixcir_ts tries to configure registers > during noirq. At that point we already have PM runtime > rpm_check_suspend_allowed() return -EACCESS for PM runtime > which leads to the pixcir -ETIMEDOUT 110 error from the i2c > controller. > > And in general, I don't think we should attempt any i2c during > noirq :) I totally agree. But wouldn't the proper solution be to fix the pixcir_ts driver to not do transfers during noirq in this specific case? Recalling your original patch from last year, you mentioned RTC devices. So, another reason for this patch is to enforce suspending the adapter, even when there is a child device connected which is for good reason not going to be suspended? This is to make sure I understand all this correctly. > > I don't know if you followed this, but we failed with a generic handling > > within the core. We have helpers now which drivers can apply > > individually. > > Hmm I don't follow you you here. Which helpers are you > talking about here? > > I'm only aware of your pending patches from thread > "[PATCH v2 0/9] i2c: move handling of suspended adapters > to the core". Yes, but the v2 approach caused problems, see the discussion there. I merged v1 (with helpers) a few days ago: [PATCH 00/10] i2c: move handling of suspended adapters to the core
Attachment:
signature.asc
Description: PGP signature
