summaryrefslogtreecommitdiff
path: root/wiki/Chat_log/20210617-io-chatlog
diff options
context:
space:
mode:
Diffstat (limited to 'wiki/Chat_log/20210617-io-chatlog')
-rw-r--r--wiki/Chat_log/20210617-io-chatlog121
1 files changed, 121 insertions, 0 deletions
diff --git a/wiki/Chat_log/20210617-io-chatlog b/wiki/Chat_log/20210617-io-chatlog
new file mode 100644
index 0000000..7bd08c1
--- /dev/null
+++ b/wiki/Chat_log/20210617-io-chatlog
@@ -0,0 +1,121 @@
+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