[PATCH] bcm53xx: add nand subtarget

[Rafał Miłecki]

niedz., 20 sie 2023 o 20:50 Arınç ÜNAL <arinc.unal at arinc9.com> napisał(a):
> On 20 August 2023 17:15:51 GMT+03:00, "Rafał Miłecki" <zajec5 at gmail.com> wrote:
> >sob., 19 sie 2023 o 17:12 Arınç ÜNAL <arinc.unal at arinc9.com> napisał(a):
> >> The NVMEM_BRCM_NVRAM driver won't work properly with NVRAM in NAND. It
> >> causes the devices with NVRAM in NAND to bootloop. Create a subtarget for
> >> NAND devices and disable the driver on it.
> >>
> >> Disable NVMEM too as the bgmac_bcma driver will hang trying to retrieve the
> >> MAC address without NVMEM_BRCM_NVRAM enabled.
> >
> >Adding a new subtarget just because one driver doesn't work correctly
> >is an overkill.
> >
> >I'll handle that in some simple solution after holidays.
>
> Rafał, you've been aware of this problem and reportedly working on at least diagnosing it for the last 6 months. You did not come up with anything yet. Forgive me if I find it hard to believe that you will come up with a solution in a short amount of time, if that's what you mean by after holidays as I don't know what holidays you're referring to.
>
> In the meantime, there is a good amount of users unable to use the newer OpenWrt versions on the affected devices because of a driver that provides a rather trivial feature. I would like these devices to work again as soon as possible as we've already lost a lot of time.
>
> To have these devices work again, I see these solutions:
>
> - First and the most obvious, fix the driver. I do not know the NVMEM subsystem well enough to do it and am currently not interested to spend time studying it.
>
> - Prevent the driver to read from the flash if NAND flash is detected. Once the underlying cause is fixed, this can then be reverted. Like I stated above, I lack the knowledge to do this.
>
> - This patch. It can be easily reverted in the future when or should the issue be fixed.
>
> - My other patch submitted here that disables the driver for the whole target. Currently, only four devices do not have NAND flash and two of these devices are not set to be compiled anyway. I'd rather prefer this than convoluting OpenWrt with another subtarget. I'd much rather prefer two devices not benefiting the future provided by this driver compared to a lot more devices being unable to boot.
>
> I feel like writing this mail has been a waste of time as lately I've never seen you reply or in any way acknowledge that you've read any of my responses, yet I've done it anyway, at least for the mailing list.

I'm not going to argue I handled it properly. It looked at it once
then forgot it. My maintenance time is very limited and I'm not
denying this.

I assumed holidays end for most people with the end of August. I'm on
a 1-week holiday family trip and I don't have time for development or
hardware for testing.

In my very personal opinion I prefer to leave support for those
devices broken for another week (given it has been 6 months now)
rather than push a (in my opinion) pretty hacky workaround.

-- 
Rafał