path: root/wiki/Chat_log/20210617-io-chatlog

09:03 < wsa> okay, everyone, let's start the meeting
09:03 < wsa> I hope you are all well
09:04 < marex> wsa: oh hey btw, that reminds me, I found microSD-to-M2 (SDIO) adapter available, would it be useful for SDHI testing ?
09:04 < wsa> marex: wasn't m2 an old mobile phone memory card standard?
09:05 < wsa> marex: but microsd-to-sdio sure sounds interesting!
09:05 < marex> wsa: its the successor of miniPCIe
09:05 < wsa> so, we ran out of acronyms
09:05 < marex> wsa: the slot is smaller and more dense, and has SDIO on it in addition to other things
09:05 < wsa> and you have them lying around or found a good place to order?
09:05 < marex> wsa: https://eu.mouser.com/ProductDetail/Murata-Electronics/LBEE0ZZ1WE-TEMP?qs=%2Fha2pyFaduh0gSsb%252BQ0n4%2F%252BKgmdBFdpEEIgpCrs2TO4zUBuDL3QyXA%3D%3D
09:06 < geertu> marex: wsa: Na, need to add the missing dots: "M.2"?
09:06 < wsa> I will have a look. Looks "nice to have" right now, not really "we need that!!!"
09:06 < wsa> okay, welcome to the IO meeting!
09:06 < wsa> Here are the status updates:
09:06 < wsa> A - what have I done since last time
09:06 < wsa> ------------------------------------
09:06 < wsa> Geert
09:06 < wsa> : investigated IIC automatic transmission with a logic analyzer, fixed HS400 on R-Car M3-W+, DT binding validation related fixes, incl. conversion to json-schema for sms911x, ccree, pcf8563, st1232, ilitek_ts_i2c, ti,bq32k
09:06 < wsa> Niklas
09:06 < wsa> : fixed coefficient calculations for Thermal, discussed sh_cmt single channel fix on list
09:06 < wsa> Shimoda-san
09:06 < wsa> : fixed a race in USB, avoided unneeded irqs with USB gadget, and improved DMA termination with SCIF, continued to improve R-Car S4 Ethernet driver, inspected V4H spec, discussed SDHI tuning issues, submitted eLinux articles about XHCI firmware loading and performance penalties of debugging options
09:06 < wsa> Ulrich
09:06 < wsa> : found out what causes the "tuning execution failed" message when removing an SD card, patch review
09:06 < wsa> Wolfram
09:06 < wsa> : sent SDHI patch to abort tuning cmd when timeout was detected and improved MMC core a little in that regard, investigated that MMC core behaves correctly with stop after tuning, improved documentation of debugfs, reviewed Geert's I2C related JSON conversion patches, sketched design for SDHI clock refactoring, reviewed SDHI & I2C patches, sent small patches: GPIO keys on Alt, bugfix for
09:06 < wsa>  DA9063, V3U TMU resend
09:06 < wsa> B - what I want to do until next time
09:06 < wsa> -------------------------------------
09:06 < wsa> Geert
09:06 < wsa> : wants to make all IIC instances compatible with the generic version and update the DT bindings
09:06 < wsa> Niklas
09:06 < wsa> : wants to work on the next batch of thermal upporting
09:06 < wsa> Shimoda-san
09:06 < wsa> : wants to continue to investigate SDHI driver's issues, continue to improve R-Car S4 Ethernet driver, continue to make more eLinux articles
09:06 < wsa> Wolfram
09:06 < wsa> : wants to refactor SDHn to be a separate clock, review Uli's SDHI findings, investigate Geert's regression report about SDHI on Koelsch
09:06 < wsa> C - problems I currently have
09:06 < wsa> -----------------------------
09:06 < wsa> Wolfram
09:06 < wsa> : mentions that Geert's SDHI issue on Koelsch may be card dependant
09:06 < marex> wsa: you can plug in contemporary SDIO WiFI devices into it, that's a bag of breakage :(
09:07 < wsa> marex: "can" or "can't"
09:07 < wsa> neg: can you resend the thermal fix with the Fixes tag which shimoday suggested?
09:08 < wsa> neg: not that the maintainer waits for that
09:08 < wsa> uli_: sorry for not checking your SDHI investiagtions before this meeting. I have planned it for tomorrow
09:08 < uli_> np
09:08 < neg> Yes I'm planing to resend it with the tag, just waitinig a few more days in case some more tags arrive
09:09 < marex> wsa: can, I tested a few in other board
09:09 < wsa> ah, so it reveals a lot of problems, and because of that it is a "bag of breakage"?
09:09 < wsa> neg: super!
09:10 < wsa> neg: also, I am afraid we *need* to have a personal meeting somewhen soon again
09:10 < neg> wsa: Ohh no :-)
09:10 < wsa> neg: the weat^H^H^H^H problems demand it :D
09:10 < marex> wsa: SDIO WiFi devices and their drivers, that is just depressing, yes
09:11 < marex> wsa: sometimes it breaks on the SDIO host end, sometimes on the wifi end, usually both in one way or the other
09:12 < wsa> marex: yes, I am with you. That was no fun looking at them. I think it was the Marvell driver for me.
09:12 < marex> wsa: so far I think I found one wifi card which didn't trigger bugs in either host driver or its own driver, the brcmfmac, all the others I tried (mwifiex, ath10k, rs91x, and I think I forgot a few) just failed in different obscure ways
09:12 < marex> wsa: oh yes, the mwifiex (now NXP btw) is surprisingly broken, I didn;t expect that
09:13 < wsa> I don't have more comments to the status reports. If you do, fire away
09:13 < wsa> otherwise the first topic is...
09:13 < wsa> SCIF!
09:14 < wsa> I have a task open here that SysRq does not work with DMA; same with parity and break
09:14 < geertu> marex: broadcom rulez...
09:14 < neg> FWIW, I tested EVK-EMMY-W161-A with SDR104 on M2 a few weeks back, still breaks as it always have :-(
09:14 < marex> wsa: btw I did a bit of an SDIO wifi survey recently, hence my observations, so if you need SDIO devices to test ...
09:15 < wsa> do we *know* it does not work with DMA, or is the driver not up for that?
09:15 < wsa> I kind of expect the first, then I'd abandon this task
09:15 < wsa> marex: cool
09:15 < marex> wsa: yep, I keep tripping the sysrq regularly on next , meta-f b in minicom -> fail
09:16 < wsa> I only have two cards. One older one from Magnus, can't recall which right now. And this EMMY W161 from ublox (mwifiex)
09:17 < wsa> marex: the card from Magnus: Z-Com XG-182M
09:18 < wsa> geertu: uli_: any comments?
09:18 < uli_> no idea
09:18 < marex> wsa: I can give you some pointers if you want :)
09:18 < wsa> otherwise it might be a good next task for Uli to research that?
09:18 < wsa> marex: SDIO or SCIF?
09:18 < geertu> wsa: We disable interrupts through SCSCR_RIE when DMA is enabled
09:18 < marex> wsa: SDIO cards to test
09:19 < geertu> that not only disables receive interrupts, but also error and break interrupts
09:19 < wsa> marex: which card is the one that works?
09:19 < marex> wsa: redpine rs9116 is a fine generator of SDIO interrupts for example
09:20 < marex> wsa: laird lwb5+ sterling , that's bcm43455 wifi + uart bluetooth (you need extra wires for that)
09:20 < marex> wsa: there is some azurewave mwifiex (fubar'd, really)
09:21 < geertu> so the hardware seems to support it, at the expense of the software having to handle more interrupts (and ignore the receive interrupts, as DMA handles that)
09:22 < wsa> so, we disable interrupts with the big hammer when we could have this more fine grained?
09:23 < geertu> wsa: There doesn't seem to be more fine-grained bits, even on HSCIF
09:23 < wsa> ah, I get it know
09:23 < geertu> We might play a bit with the FIFO thresholds to reduce the number of unwanted interrupts, but that might impact DMA, too?
09:24 < wsa> ignoring the interrupts means code which ignores the interrupt
09:24 < wsa> not a bit which disables it
09:25 < wsa> on the other hand, detecting parity errors with DMA might be good, too?
09:25 < geertu> yes
09:26 < wsa> uli_: are you interested in researching that?
09:26 < uli_> i'm up
09:27 < wsa> uli_: like enabling this interrupt to have BREAK and parity support and see how it affects DMA speeds?
09:27 < uli_> yes
09:27 < geertu> SCIF on e.g. RZ/A does have multiple interrupts, so you can disable them selectively (at the GIC level)
09:27 < geertu> and RZ/G2L ;-)
09:27 < wsa> RZ/A is Genmai board?
09:28 < geertu> A.o.
09:28 < geertu> gr-peach, rskrza1, rza2mevb
09:29 < wsa> okay, cool, then uli's next task is decided
09:30 < wsa> that's all from my side, any other topics around?
09:30 < geertu> wsa: Are you happy with marking all IIC instances compatible with the generic version?
09:31 < wsa> Very much!
09:31 < wsa> This is my preferred way to go
09:32 < geertu> Ok, thx
09:32 < wsa> RZ/G2L is what gen3 equivalent?
09:32 < geertu> None
09:32 < wsa> what is it then?
09:32 < geertu> It's a blend of R-Car Gen3 and RZ/A2
09:32 < geertu> More the latter
09:32 < wsa> okay
09:33 < geertu> So the work on RZ/G2L may backfill missing support on RZ/A, like DMA
09:33 < wsa> ok, so Renesas Europe should also handle this for "their" SCIF variant
09:34 < wsa> we focus on what we agreed above
09:34 < wsa> cool
09:34 < wsa> geertu: ready to rock?
09:34 < geertu> RZ/G2L SCIF == RZ/A2 SCIF, AFAIK
09:34 < geertu> wsa: yep