path: root/wiki/Chat_log/20201001-core-chatlog

09:31 < shimoda> marex: i'd like to support u-boot for v3u at least. but, if so, we cannot support PSCI. So, perhaps, we also should support atf? Or, u-boot only is enough?
09:31 < wsa> it is "best effort" for now as I understood it
09:32 < marex> shimoda: we can do both
09:32 < shimoda> ah, i also got boards datasheets.
09:32 < geertu> Note that upstream won't accept arm64 SMP support that does not use PSCI
09:32 < marex> shimoda: isnt v3u booting via tfa/uboot anyway ?
09:32 < shimoda> moriperi: could we share boards datasheets to europeri members?
09:32 < wsa> no PSCI? we need a proper reboot, then
09:33 < marex> geertu: you mean ARM Linux people won't accept a system which does not run ARM BSD-licensed firmware
09:33 < marex> geertu: sounds to me like there is incentive on both sides to force that BSD-licensed firmware onto you
09:34 < marex> wsa: PSCI isnt only about reboot, but also about bringing up secondary CPU cores
09:34 < marex> wsa: and I would highly recommend keeping it at that
09:34 < wsa> marex: yes, but that is stuff for core ;)
09:34 < shimoda> marex: v3u will boot from "ICUMXA" cpu core. and the cpu kicks cortex-A cpu0
09:34 < marex> wsa: anything beyond that, like power domains ... uuuurgh
09:34 < geertu> marex: PSCI is just a spec, so you can provide your own inmplementation?
09:34 < wsa> why is there no PSCI on v3u?
09:34 < marex> geertu: and the "only up to date implementation" is by whom ?
09:35 < marex> geertu: U-Boot implements only PSCI 0.2 I think, and on Gen3 the TFA already installs the handlers
09:35 < marex> wsa: there likely is ?
09:35 < moriperi> shimoda: I think I already did
09:35 < marex> shimoda: is it using the CR7 loader ?
09:36 < marex> shimoda: was that why you asked me about it some time ago ?
09:36 < marex> shimoda: or is that something else entirely ?
09:37 < wsa> seems we already switched to core
09:37 < wsa> is there something left to discuss for IO
09:37 < wsa> ?
09:37 < geertu> wsa: Shall we make that official?
09:37 < geertu> Next meeting?
09:38 < wsa> okay, then, this time a on-the-fly-mic-takeover, geertu have fun!
09:38 < geertu> Oct 29? (conflicts with KVM Forum, anyone who cares?)
09:38 < wsa> Oct, 29th?
09:39 < kbingham> Not much else after ELCE I guess ;-)
09:39 < shimoda> marex: CR7 loader is used on V3H. But, it seems local code. So, i'm thinking we should not use it on V3U because it's difficult to maintain such local code by ourselve.
09:40 < shimoda> i will take a day off at oct 29...
09:40 < kbingham> (seems marex is in the promo background for elce : https://events.linuxfoundation.org/embedded-linux-conference-europe/)
09:40 < moriperi> shimoda: oops, not shared ?
09:41 < marex> shimoda: then I need to read up on ICUMXA :)
09:41 < geertu> v850?
09:41 < geertu> Welcome to today's Core Group Chat Meeting!
09:41 < wsa> Nov, 5th is also fine for me
09:42 < geertu> nov 5 is fine forme
09:42 < marex> kbingham: michal simek (xilinx) is on the left side, same row, black shirt with some "blah code blah" writing on it
09:42 < geertu> Note that EU wil have switched to Winter Time by then
09:42 < geertu> marex: And he fancies the camera person?
09:43 < marex> geertu: heh
09:43 < geertu> Agenda:
09:43 < geertu> 1. Status Updates
09:43 < geertu> 2. Discussion Topics
09:43 < geertu> Topic 1. Status updates
09:43 < geertu> A) What have we done since last time:
09:43 < geertu> Kieran got local patchwork bot instances to run.
09:43 < geertu> Marek fixed parsing multiple memory nodes in U-Boot and OpTee-OS,
09:43 < geertu> assisted Renesas UK with RZ/G2 U-Boot support, and investigated and
09:43 < geertu> fixed PCIe L1 link state recovery on R-Car Gen2.
09:43 < geertu> Morimoto-san tried to control ULCB power/reset by GPIO.
09:43 < geertu> Niklas posted VIN stf8 PFC patches.
09:43 < geertu> Shimoda-san submitted R-Car V3U initial support and got the
09:43 < geertu> corresponding SoC and board manuals.
09:43 < geertu> Geert reviewed RZ/G and R-Car V3U patches, consolidated renesas-pinctrl,
09:43 < geertu> and sent the second batch of pull requests for v5.10.
09:44 < geertu> B) What we plan to do till next time:
09:44 < geertu> Marek plans to continue helping with RZ/G2 U-Boot patch review, and
09:44 < geertu> submit v3 of the PCIe L1 link state recovery patch.
09:44 < geertu> Morimoto-san plans to try to control power/reset on the Salvator-XS and
09:44 < geertu> Kingfisher boards.
09:44 < geertu> Niklas plans to post v2 of the VIN stf8 PFC patchset.
09:44 < geertu> Shimoda-san plans to correct more information about R-Car Gen3e, and may
09:44 < geertu> work on more support for R-ar V3U.
09:44 < geertu> Geert plans to develop v10 of obtaining the start of physical memory
09:44 < geertu> from DTB, and attend and present at virtual ELC-E.
09:45 < geertu> C) Problems we have currently:
09:45 < geertu> Marek wonders if we can tell RZ/G2 apart from R-Car Gen3.
09:45 < geertu> ---EOT---
09:45 < geertu> Anything I missed?
09:46 < geertu> moriperi: Please have a look at
09:46 < geertu> https://elinux.org/R-Car/Boards/Salvator-XS#Remote_Control
09:46 < geertu> Note that you can change power switch behavior (Salvator-X(S) defaults
09:46 < geertu> to Level, ULCB to Pulse), by changing the resistors (R413/R424) that
09:46 < geertu> control the PMIC's RSTBMODE configuration.
09:46 < geertu> With KingFisher attached, it's different, as ULCB is powered from
09:46 < geertu> KingFisher, IIRC.
09:46 < geertu> Topic 2. Discussion Topics
09:46 < marex> geertu: well, Marek also wonders about the V3U boot process
09:46 < marex> geertu: with ULCB you can use cpld-control too
09:48 < moriperi> geertu: thanks
09:51 < geertu> So Topic 1. R-Car V3U already started
09:51 < geertu> No PSCI
09:51 < marex> geertu: thats just weird
09:51 < geertu> Booted from ICUMXA
09:51 < marex> well, what is running on the ICUMXA, some custom preloader or what ?
09:51 < marex> and then it starts -- what -- on the cortexA ?
09:52 < marex> geertu: have you seen the V3U booting ?
09:53 < shimoda> marex: yes, custom preloader is running on the ICUMXA. and, perhaps the preloader will not be in public
09:53 < shimoda> marex: and it starts u-boot on the cortexA
09:53 < geertu> According to Ch. 19 ("Boot"), it can boot from Cortex-R52 or ICUMXA
09:53 < marex> shimoda: please dont do that, you are gonna turn it into unpopular horribleness like the imx8q* and its SCFW
09:54 < marex> shimoda: that is also a huge blob without public sources and there isn't much joy about that
09:56 < shimoda> marex: hmm, renesas plans such boot sequence on R-Car Gen4 SoCs too...
09:56 < marex> shimoda: is there any specific problem in opening the ICUMXA code ?
09:57 < marex> shimoda: so anyway, it starts u-boot on the cortexA, and then U-Boot runs in EL3 , and U-Boot starts Linux ?
09:57 < marex> shimoda: so no TFA involved ?
09:58 < marex> shimoda: to bring up the secondary cores, U-Boot does implement enough of the PSCI API, so if we can avoid ATF here, good
09:58 < shimoda> marex: the boot sequence is correct. but, as you said, we need PSCI somewhere fore reboot, SMP and so on
09:58 < marex> shimoda: so why no ATF this time ?
09:59 < geertu> marex: So what's the problem with PSCI? Reboot only?
09:59 < marex> geertu: Linux on aarch64 uses PSCI to bring up secondary cores
09:59 < marex> geertu: reboot is kinda secondary
09:59 < marex> as in, reboot is "extra"
09:59 < shimoda> marex: i don't know why renesas will not open ICUMXA code in public.
09:59 < geertu> marex: marex said "to bring up the secondary cores, U-Boot does implement enough of the PSCI API"
10:00 < geertu> So?
10:00 < marex> geertu: yeah, so, you can have U-Boot behave as the PSCI firmware counterpart and bring the secondary cores up
10:00 < shimoda> marex: i think it's just renesas internal software resource problem. on Gen3, IPL team develops ATF. But, now IPL team develops ICUMXA loader.
10:01 < shimoda> Linux BSP team develops uboot on both Gen3 and V3U.
10:01 < marex> shimoda: yikes :)
10:02 -!- moriperi [~user@relprex1.renesas.com] has quit [Read error: Connection reset by peer]
10:02 < geertu> bummer, parental control has offlined moriperi?
10:02 < shimoda> marex: :)
10:02 < marex> shimoda: but I guess, if you look at it, the V3M/V3H CR7 loader was some sort of pre-ICUMXA loader
10:02 < marex> shimoda: so things were headed in that direction anyway
10:02 < marex> for a longer time no less
10:03 < marex> except V3M/V3H also have ATF port
10:03 -!- moriperi [~user@relprex1.renesas.com] has joined #periperi
10:03 < marex> so either there will be ATF port similar to those V3H (which is a bit of a thin shim sandwiched between the CR7 loader and U-Boot), or its really only gonna be U-Boot
10:04 < marex> maybe thats what should be double-checked
10:04 < marex> I would prefer the later of course
10:04 < marex> then we can add the PSCI callbacks to U-Boot , drop EL and boot Linux , without ATF, that would be nice
10:04  * pinchartl wonders how many (pre-/early-)boot loader stages a platform really need
10:05 < kbingham> pinchartl, It's turtles all the way down ... no wait - that'ssomething else ;-)
10:05 < marex> pinchartl: more than we have, just look at intel
10:05 < marex> kbingham: nice
10:06 < wsa> pinchartl: I think this correlates exactly to the number which SoC generation it is
10:06  * geertu thinks Super-H and RISC-V early stages are missing
10:07 < marex> shimoda: so, maybe we still need to flesh the boot process out ? :)
10:07 -!- kbingham[m] [kbinghamma@gateway/shell/matrix.org/x-tqvvdwyhjxnfmmxl] has joined #periperi
10:09 < shimoda> marex: sorry i could not understand "flesh"
10:09 < geertu> sort out
10:09 < marex> shimoda: flesh == meat
10:09 < marex> shimoda: what you are made of
10:09 < marex> shimoda: flesh out == sort out
10:10 < wsa> I also didn't know this expression
10:10 < shimoda> marex: ah, i got it. yes, i think so :)
10:10 < marex> geertu: would that translate as 肉-out ? :-)
10:12 < geertu> Anything else to discuss?
10:12 < marex> shimoda: can you check about the PSCI and ATF internally please ?
10:12 < marex> shimoda: it would be good to confirm, otherwise, well, we can start adding the PSCI bits into U-Boot unless renesas did already
10:12 < marex> shimoda: or how else do you test the secondary core bringup ?
10:13 < shimoda> marex: for now, bsp team and i don't test the secondary core bringup.
10:14 < marex> shimoda: oh, ok
10:16 < geertu> I think it's time to pass the torch^Wmic to MM
10:16 < geertu> Thanks for joining, and have a nice continued day!
10:16 < geertu> pinchartl: mic
10:16 < marex> shimoda: I am now looking at the U-Boot port for V3U, there isn't much, that's real cool :)
10:16 < marex> ttyl, stay healthy
10:17 < geertu> marex: where is the U-Boot port?
10:17 < marex> geertu: https://github.com/renesas-rcar/u-boot.git