09:33 < geertu> Welcome to Today's Core Group Chat!
09:34 < geertu> Due to Wolfram's overbooking, and Laurent's misbooking, there are less people than usual
09:34 < geertu> Agenda:
09:34 < geertu> 1. Status Updates
09:34 < geertu> 2. Discussion Topics
09:34 < geertu> Topic 1. Status updates
09:34 < geertu> A) What have we done since last time:
09:34 < geertu> Marek worked on U-Boot (HS400 on R-Car Gen3, special SD card support)
09:34 < geertu> and Linux (Upported M3-N and E3 CAN/CANFD support).
09:34 < geertu> Geert resubmitted the move of SoC-specific ARCH_* symbols, and generic
09:34 < geertu> reset support for VFIO platform, and explored SYSC errata.
09:35 < geertu> B) What we plan to do till next time:
09:35 < geertu> Marek plans to work on PCA953x suspend/resume support for Salvator-XS
09:35 < geertu> SATA, and will continue researching HS400 performance on E3.
09:35 < geertu> Shimoda-san plans to continue to make an IPMMU driver's plan with
09:35 < geertu> Magnus-san.
09:35 < geertu> Geert will sent the first CLK and PFC pull requests for v4.21, and plans
09:35 < geertu> to continue exploring SYSC errata, and submit resulting patches.
09:36 < geertu> C) Problems we have currently:
09:36 < geertu> Kaneko-san reports that the R-Car E3 thermal driver and bindings have
09:36 < geertu> been reviewed, but not merged.
09:36 < geertu> Anything I missed?
09:37 < patersonc> Marek: What is the status of M3N/E3 CAN/CAN FD?
09:37 < Marex> patersonc: I submitted the patches, tested on E3, for M3N I need to cook some small HW with MCP2551 or so
09:37 < patersonc> Marex: Thanks
09:37 < Marex> patersonc: since there's no CAN transceiver on the M3N S-XS, I need to build one ...
09:38 < patersonc> Yep
09:38 < patersonc> https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?id=a414cb9dc73a746c71ce79d6422d086745311626
09:38 < patersonc> Looks like E3 is in
09:38 < Marex> patersonc: you have M3N with CAN(FD) transceiver ? :)
09:38 < Marex> maybe you can test my patch
09:38 < Marex> s/my/Kaneko-san's/
09:39 < horms> patersonc: yes, I applied E3 CAN DT patches earlier this week
09:40 < patersonc> Marex: As it happens we do have a small CAN board for Sal-XS. It was developed by ze Germans
09:41 < Marex> patersonc: Schön
09:41 < Marex> patersonc: maybe you can save me a trip to local parts store ?
09:42 < patersonc> Sure
09:42 < Marex> patersonc: ah, sweet :)
09:42 < patersonc> I'll sort it before the boss comes in ;)
09:42 < geertu> patersonc: thx!
09:44 < patersonc> Marex: Just ping your patches/branch over
09:44 < geertu> I guess the R-Car E3 thermal  patches will be applied, eventually, as usual?
09:44 < Marex> geertu: hrm, but he'd need a DT patch adding the pinmux etc. on M3N
09:44 < Marex> geertu: I might build me the CAN transceiver anyway
09:44 < horms> geertu: I guess so too. I'm holding off on the DT patches until the driver patches are accepted.
09:44 < horms> But perhaps I should just apply them
09:45 < geertu> Marex: That's a 5 line DT patch?
09:45 < Marex> patersonc: will do, sec
09:46 < Marex> geertu: once you know the mux ...
09:47 < Marex> geertu: although, there are not that many options
09:48 < geertu> Marex: If they have a CAN extensaion board for Salvator-X(S), they should have some DT description, no?
09:49 < geertu> can1_data or canfd1_data
09:49 < Marex> geertu: not necessarily
09:49 < geertu> and perhaps can_clk
09:49 < Marex> geertu: is can_clk actually needed ?
09:50 < geertu> Marex: depends on the hardware expansion, I guess
09:50 < geertu> Topic 2. Discussion Topics
09:50 < Marex> patersonc: https://patchwork.kernel.org/project/linux-renesas-soc/list/?series=44491 https://patchwork.kernel.org/patch/10691447/
09:50 < geertu> Anything else to discuss (CAN is actually I/O ;-)
09:51 < Marex> patersonc: you need these two series + the M3N extras enabling the can controller
09:51 < Marex> geertu: well, HS400 slowness on E3 , it's unclear if that's IO or core, since it might be QoS issue
09:51 < patersonc> Marex: Ta. I'll try and do it tomorrow. Please nag.
09:51 < Marex> patersonc: did you have a chance to try HS400 performance on E3 ?
09:51 < Marex> patersonc: will do, thanks
09:52 < Marex> patersonc: I'll also grab the transceiver today, just to be sure
09:52 < geertu> Marex: Well, that's indeed to be seen
09:53 < geertu> Marex: Tried a logic analyzer, to compare with a "fast" board?
09:53 < Marex> geertu: we're talking 200 MHz LVTTL here
09:54 < Marex> geertu: scope which can sample that reliably costs like a small car
09:54 < geertu> Marex: OK
09:55 < patersonc> Marex: I can tell you HS400 works on E3 :)
09:55 < geertu> Could it be the clock is running at half the frequency of what you'd expect?
09:55 < Marex> patersonc: it works for me too
09:55 < Marex> patersonc: it's just slower than I'd expect
09:55 < Marex> geertu: I checked the clock, pinmux, memory bandwidth, nope ...
09:56 < Marex> geertu: clock was the issue with HS400 I had before, so I checked that first
09:56 < Marex> patersonc: can you do a dd if=/dev/mmcblk0 bs=32M count=16 of=/dev/null and tell me how fast that goes ?
09:56 < geertu> Marex: I meant measured the clock?
09:57 < Marex> patersonc: on the card that's on Ebisu, it should be some 230-250 MB/s
09:57 < patersonc> Marex: I'll check
09:57 < Marex> geertu: don't have such a good equipment here, so no
09:58 < geertu> Marex: Add some high-speed flip-flops to reduce the clock rate to something you can measure?
09:59 < Marex> geertu: did you see _any_ such high speed stuff recently ?
10:00 < geertu> Marex: The above dd command gives me 128 MB/s on Ebisu
10:00 < Marex> geertu: you also forgot the antenna effect, capacitance of the extra stuff etc
10:00 < Marex> geertu: we're talking 200 MHz here, very different from 20 MHz
10:00 < Marex> geertu: yeah, I see 128 MB/s with Ebisu 4D using vendorkernel which reports HS400 too
10:00 < geertu> Marex: Are slower speeds also slower than expected?
10:00 < Marex> geertu: U-Boot gives me up to 170 MB/s, but that's still too slow
10:01 < Marex> geertu: U-Boot gives me up to 170 MB/s in HS200, which is what I'd expect
10:01 < Marex> geertu: U-Boot gives me up to 172 MB/s in HS400, which is too slow
10:01  * geertu has a kernel without HS400 support, so test result was for HS200
10:02 < geertu> I mean anything slower that you can measure? The clocks may be off in all modes
10:02 < Marex> geertu: with HS200, you should get 170 MB/s, that's what the datasheet says and what my measurements confirm on this board and other boards with the same eMMC
10:03 < geertu> We used to have off-clocks on Armadillo, which I only discovered by measuring
10:04 < geertu> And e.g. R-Car H3 ES1.0 lacked a by-2 divider after some PLLs
10:04 < geertu> Anyway, time to conclude
10:04 < geertu> Thanks for joining, and have a nice continued day!