Core-chat-meeting-2018-04-05

10:43 < geertu> Welcome to today's Core Group Meeting!
10:43 < geertu> Agenda:
10:43 < geertu> 1. Status Updates
10:43 < geertu> 2. Discussion Topics
10:43 < geertu> Topic 1. Status updates
10:44 < geertu> A) What have we done since last time:
10:44 < geertu> Marek worked on U-Boot TPL ("Third Program Loader") support for Porter,
10:44 < geertu> updated R-Car M2-W JTAG support to latest OpenOCD.
10:44 < geertu> Shimoda-san submitted M3-N PFC patches, and discussed a plan to improve
10:44 < geertu> the IPMMU driver with Magnus-san.
10:44 < geertu> Geert sent a second clk and pfc pull request for v4.17, fixed the dead
10:44 < geertu> serial console after serial wake-up, reviewed RZ/G1C patches, and
10:44 < geertu> enjoyed some Easter holidays.
10:45 < geertu> (BTW, Simon is enjoying Easter holidays, and thus excused)
10:45 < geertu> B) What we plan to do till next time:
10:45 < geertu> Marek will clean up and submit TPL, RPC, and QSPI patches for U-Boot,
10:45 < geertu> JTAG patches for OpenOCD.
10:45 < geertu> Morimoto-san will prepare his stomach and mental state for Ninja food.
10:45 < geertu> Shimoda-san will pave the way forward for IPMMU, and will try to port
10:45 < geertu> initial code for R-Car E3 Ebisu.
10:45 < geertu> Geert will enjoy more Easter holidays. After that he will revisit vfio
10:45 < geertu> and qemu patches, do more periupport analysis, and will flush his patch
10:45 < geertu> queue after v4.17-rc1 has been released.
10:47 < geertu> C) Problems we have currently:
10:47 < geertu> Shimoda-san would like to know how to add a node for usb2.0 ch3 on H3
10:47 < geertu> ES2.0.
10:47 < geertu> Anything I missed?
10:48 < neg> Not from my side :-)
10:49 < geertu> Topic 2. Discussion Topics
10:50 < geertu> Shimoda-san would like to know how to add a node for usb2.0 ch3 on H3 ES2.0.
10:50 < geertu> https://patchwork.kernel.org/patch/9980275/
10:50 < geertu> Enabling this requires flipping a switch, right?
10:50 < neg> kbingham[m]: ping :-)
10:51 < kbingham> neg: pong :D
10:51 < shimoda> geertu: yes. (according to the ML archive and board datasheet)
10:52 < kbingham> So we must chose between ADV748x and USB2 ch3 and boot time somehow.
10:53 < wsa_> a bit like the PCIE/SATA switch maybe?
10:53 < kbingham> We wouldn't be so lucky to be able to read this switch at runtime would we ?
10:55 < geertu> kbingham: I'm afraid not
10:56 < kbingham> So how should this be handled for the generic case - does it need to be some sort of overlay support ?
10:57 < geertu> In upstream DTS, we have to assume a fixed switch configuration.
10:57 < kbingham> And would you like me to do anything in particular in adv748x ?
10:58 < geertu> shimoda: Have you tried my suggestion, of using GP6_3[01] in salvator-common.dtsi, and overriding it in r8a7795-salvator-xs.dtsi?
10:59 < shimoda> geertu: not yet. so, I will try later
11:00 < shimoda> if success, is it acceptable for upstream?
11:00 < geertu> I think so.
11:00 < geertu> IIRC, my suggestion means that everything keeps working as before, and people who want to make use of the extra channel on H3 ES2.0 and Salvator-XS need to flip the switch.
11:01 < geertu> And everything (e.g. ADV7482) still keeps working after that.
11:01 < kbingham> What will be the effect of flipping that switch - and *not* changing the DT ? (i.e. the user doesn't chagne the DT either by overlay or otherwise)
11:01 < kbingham> Will it just be that the USB2 ch3 won't work ? or will it break ADV748x ?
11:01 < geertu> Correction: after the DT change, r8a7796-salvator-xs.dts
11:02 < geertu> \deleteline -)
11:02 < geertu> Correction: after the DT change, ADV7482 will stop working, until the user flips the switch
11:02 < kbingham> Or more specifically - only one of the ADV748x interrupts ?
11:02 < geertu> Flipping the switch and not changing DT will break the ADV7482 interrupt only, which is not yet supported anyway?
11:02 < Marex> geertu: can DTO help here ?
11:03 < Marex> geertu: U-Boot can apply DTOs
11:03 < geertu> So "stop working" really means "yet unsupported ADV7482 interrupt cannot work
11:03 < geertu> ", so it is harmless?
11:03 < kbingham> Ok - so if I make the (not yet implemented) ADV748x hotplug use only the unaffected interrupt - will we be able to support both ADV748x and USB2 ch3 ?
11:04 < geertu> kbingham: I think so. The switch allows to wire both ADV and USB using the new switch setting.
11:05 < geertu> With the old setting it keeps backwards compatibility with Salvator-X, which doesn't have the switch, and doesn't support the extra USB channel.
11:06 < kbingham> Ok - so I think we have a good all round solution there then. I will have to make sure that only the unaffected IRQ is passed to the ADV748x, and the ADV will have to configure that interrupt to handle whatever it needs. As long as that second interrupt is definitely routed in :)
11:07 < geertu> kbingham: I don't think you have to do anything, the interrupt is to be specified in DT.
11:07 < kbingham> geertu: Quite - but the (not yet implemented) adv748x interrupt configuration will have to map to the correct interrupt pin.
11:07 < kbingham> But that's fine - as it will be named.
11:08 < geertu> Naming or not doesn't matter.
11:08 < geertu> The common salvator-common.dtsi would describe the hardwired (Salvator-X) or default switch setting (Salvator-XS)
11:09 < geertu> r8a7795-salvator-xs.dtsi would override the interrupts property to accommodate the changed switch setting
11:09 < geertu> Does it all make sense?
11:10 < kbingham> Sorry - I meant naming for the driver to know which of the two interrupt sources is connected - as it has two which can both be configured for all interrupt reasons I believ.
11:10 < kbingham> It means the AFE(analog) and HDMI would share the interrupt in the ADV748x, which is why I assume they provide two interrupts
11:10 < kbingham> But I don't think its a problem.
11:11 < geertu> In both cases you can use both interrupts
11:12 < geertu> They're just routed to different pairs of pins on the SoC.
11:12 < kbingham> Ah yes - and I just re-read your message above... the changes specfic to -xs will then be handled there.
11:13 < geertu> (Yes, you can use other switch combinations not listed in the table near SW31 ;-)
11:13 < geertu> (those are considered invalid)
11:13 < geertu> Is there anything else to discuss?
11:14 < shimoda> not from me. thanks!
11:14 < kbingham> shimoda: I posted a USB2 crash to linux-renesas-soc ...
11:14 < kbingham> And I've referenced it this morning for you.
11:15 < kbingham> Is it something that you're aware of? Or did I likely make a silly mistake ?
11:15 < kbingham> (like it being related to this ch3 or such )
11:15 < geertu> kbingham: You inserted the USB plug in the correct direction? ;-)
11:15 < kbingham> geertu: I always rotate my plugs 3 times before they fit :D
11:16 < geertu> kbingham: Three is considered a good number (https://en.wikipedia.org/wiki/Chinese_numerology#Three)
11:17 < geertu> Anyway, USB is I/O
11:17 < geertu> So let's conclude
11:17 < geertu> Thanks for joining, and have a nice continued day!