diff options
author | Geert Uytterhoeven <geert+renesas@glider.be> | 2020-10-01 10:29:30 +0200 |
---|---|---|
committer | Geert Uytterhoeven <geert+renesas@glider.be> | 2020-10-01 10:29:30 +0200 |
commit | f0c6360e9ed16e120c0e69b803640a558dbe011f (patch) | |
tree | 2e2bd24e26121ff8e2915b201fc1af6852e36c89 /wiki/Chat_log | |
parent | 0b7db2f78815bb2ad9776df2feaea60e010e4512 (diff) |
wiki: Add Core chatlog for 2020-10-01
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Diffstat (limited to 'wiki/Chat_log')
-rw-r--r-- | wiki/Chat_log/20201001-core-chatlog | 157 |
1 files changed, 157 insertions, 0 deletions
diff --git a/wiki/Chat_log/20201001-core-chatlog b/wiki/Chat_log/20201001-core-chatlog new file mode 100644 index 0000000..8cf40d0 --- /dev/null +++ b/wiki/Chat_log/20201001-core-chatlog @@ -0,0 +1,157 @@ +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 |