On 30/08/2023 05:59, Jiaxun Yang wrote: > > > 在 2023/8/25 20:56, Krzysztof Kozlowski 写道: > [...] >> How did you sneak this property? The version - v2 - which was reviewed >> by Rob: >> https://lore.kernel.org/all/20190905144316.12527-7-jiaxun.yang@xxxxxxxxxxx/ >> did not have it. >> >> Now v3 suddenly appears with Rob's review and this property: >> https://lore.kernel.org/all/20200112081416.722218-4-jiaxun.yang@xxxxxxxxxxx/ >> >> Please help me understand this property appeared there and how did you >> get it reviewed? > Hi all, > > It has been some years since this series was merged. > My vague memory tells me there was some off-list discussion made in IRC with > linux-arch folks and IRQ folks to come up with this binding design. We would not suggest you property which in the name has underscores and duplicates interrupt-map property. > > In this case I guess I forgot to drop Rob's R-b tag when updating this patch > between reversions. I apologize for any inconvenience this may have caused. > >> >>> <0xffffffff>, /* int1 */ >>> <0x00000000>, /* int2 */ >>> <0x00000000>; /* int3 */ >> So now you will keep bringing more hacks for a hacky property. No, this >> cannot go on. > > What's the best way, in your opinion, to overhaul this property? As we don't > really care backward compatibility of DTBs on those systems we can just > redesign it. Deprecate the property in the bindings, allow driver to work with or without it and finally drop it entirely from DTS. > > A little bit background about this property, LIOINTC can route a > interrupt to any of > 4 upstream core interrupt pins. Downstream interrupt devicies should not > care about > which pin the interrupt go but we want to leave a knob in devicetree for > performance > tuning. So we designed such property that use masks corresponding to > each upsteam > interrupt pins to tell where should a interrupt go. Best regards, Krzysztof
