path: root/wiki/Chat_log/20150826-io-chatlog

--- Log opened Wed Aug 26 10:55:12 2015
10:55 -!- wsa_ [~wsa@p4FE25F24.dip0.t-ipconnect.de] has joined #periperi
10:55 -!- Irssi: #periperi: Total of 8 nicks [0 ops, 0 halfops, 0 voices, 8 normal]
10:55 -!- Irssi: Join to #periperi was synced in 0 secs
10:55 < geertu> Hi Wolfram
10:55 < wsa_> heyho!
10:55 < shimoda> Hi Wolfram-san!
10:55 < horms> Hi + Bye Wolfram
10:56 < wsa_> bye?
10:56 < wsa_> No IO Meeting for you Simon?
10:56  * horms goes home
10:56 < horms> oh we have a meeting?
10:57 < horms> sorry i missed that
10:57 < horms> i'll run home and log back in
10:57 < wsa_> ok
10:57 < wsa_> :9
10:57 < wsa_> :)
10:57 < wsa_> hi geert, shimoda-san!
10:59 < shimoda> hi!
11:01 < geertu> hi all
11:01 -!- horms [~horms@reginn.isobedori.kobe.vergenet.net] has quit Ping timeout: 244 seconds
11:01 < wsa_> so, simon will come back soon
11:01 < wsa_> is morimoto-san around?
11:02 < morimoto> Hi
11:02 < geertu> His bot is alive ;-)
11:02 < wsa_> :)
11:02 < shimoda> :)
11:03 < wsa_> a meeting bot
11:03 < wsa_> i guess that's a way to make $$$
11:03 < wsa_> ;)
11:04 < geertu> My former project manager was once convinced I was in a phone conf call and said things, while I wasn't there ;-)
11:05 < geertu> With IRC it's even easier...
11:05 -!- horms [~horms@121-80-213-238f1.hyg1.eonet.ne.jp] has joined #periperi
11:05 < geertu> welcome back
11:05 < horms> sorry about that. for some reason i hadn't added this meeting to my schedule
11:05 < wsa_> geertu: I hope you said good things while you weren't there
11:06 < horms> i said "I have a meeting" !
11:06 < horms> + hello to someone in the elevator
11:06 < wsa_> well, you are here nonetheless
11:06 < wsa_> so fine
11:07 < wsa_> and I will make sure I get confirmations for the meeting dates from all of you explicitly ;)
11:07 < wsa_> so, let's start I'd say
11:07 < wsa_> I see two topics
11:08 < wsa_> 1) Gen3 status update
11:08 < wsa_> 2) SCIF
11:08 < wsa_> and miscellaneous
11:08 < wsa_> any other topics from you?
11:09 < horms> not from me
11:10 < wsa_> then let's start with Gen3
11:10 < wsa_> we can add new topics as they come
11:11 < wsa_> is there an update to the timeline when which devices need to be enabled when?
11:11 < horms> i can give an etheravb update of sorts; when you are ready
11:12 < wsa_> i did not get any feedback to my I2C patches, I guess everyone was busy doing core bringup...
11:12 < shimoda> The current plan is, SPI needs by 1th Sep. but, i don't know it can be used on gen3 board
11:12 < morimoto> This is different topics, but now I could use GPIO and I2C on Gen3 board. it is not included on v7, but I can add it on v8
11:13 < geertu> That's MSIOF?
11:13 < shimoda> geertu: yes
11:13 < geertu> The only usable point so far is SoftSW
11:13 < geertu> Haven't checked pinctrl and EXIO yet
11:13 < geertu> BTW, the pintrl part in the datasheet is "difficult"
11:13 < wsa_> morimoto-san: you mean I2C native or I2C using GPIO?
11:14 < geertu> There are no tables like on Gen2 (I want to add HSCIF pinctrl)
11:14 < geertu> Or am I missing something>
11:14 < geertu> Still, SoftSW would require soldering
11:15 < geertu> Or the CP[1-4] test points.
11:15 < shimoda> usb2 host needs by 25th August, and renesas-drivers-2015-08-25-v4.2-rc8 works correctly
11:15 < morimoto> geertu: Gen3 datasheet was exchanged. yes it is difficult to read
11:16 < wsa_> so I2C is closed case now?
11:17 < wsa_> and GPIO? (which is requirement for EtherAVB)
11:17 < morimoto> wsa_: I'm using renesas,i2c-r8a7795 for i2c. i2c driver seems working. I could chance to use gpio-i2c too
11:17 < wsa_> well, i am fine with renesas,i2c-r8a7795. that's enough for me.
11:17 < wsa_> nice
11:18 < morimoto> wsa_: please keep i2c/gpio + 1week. I need to check it on new board. (I used old and broken board)
11:18 < wsa_> ok
11:19 < shimoda> HSCIF and USB func need by 22th Sep.
11:20 < geertu> Working on HSCIF, now I have a serial link to DEBUG SERIAL-1
11:20 < geertu> SCIF1 on those pins works
11:21 < shimoda> nice!
11:21 < geertu> shimoda: Or is MSIOF the slave-only DRIF?
11:21 < wsa_> so, the next work items would be: SPI (but we don't have proper HW setup currently) and testing I2C/GPIO with the new board (morimoto-san)
11:21 < wsa_> usb works
11:22 < morimoto> I didn't try ii2
11:22 < morimoto> but i2
11:22 < morimoto> i2c
11:22 < morimoto> please care ii2
11:23 < morimoto> s/ii2/iic/g
11:23 < shimoda> geertu: just MSIOF, not DRIF
11:24 < wsa_> morimoto: what do you mean by care?
11:25 < morimoto> Gen3 has I2C/IIC. I can try I2C, but I don't try IIC
11:25 < horms> care -> please look after it
11:25 < wsa_> by the way, does magnus have a "new board" or an "old broken board" in his lab?
11:26 < morimoto> His one is OK board
11:26 < morimoto> but he is using old uboot
11:26 < wsa_> good
11:26 < morimoto> Ah, but 1 note.
11:26 < wsa_> morimoto: do you have a patch for adding all i2c to the r8a7795 dtsi or shall I make one?
11:27 < morimoto> His one has 16.000Mhz as extal clock.
11:27 < morimoto> new board has 16.666HMhz
11:27 < morimoto> First plan was 33.333MHz, but...
11:28 < morimoto> wsa_: I have it, I can send it in v8
11:28 < wsa_> ok
11:28 < morimoto> wsa_: it is including PFC / MSTP too
11:28 < wsa_> great
11:28 < geertu> Before someone starts writing the SCIF/HSCIF dtsi patch, I have it
11:29 < morimoto> Mine has only SCIF2, can I update it to your one? or do you send it as incremental patch ?
11:29 < morimoto> in v8
11:29 < geertu> I will send an incremental patch
11:29 < wsa_> is there some schedule when to use magnus gen3 board?
11:29 < morimoto> OK
11:30 < wsa_> i assume core group needs it a lot?
11:30 < wsa_> or first come first serve? :)
11:30 < geertu> It's mutex is kept on #periperi
11:30 < geertu> Currently I have it ;-)
11:30 < wsa_> ok
11:31 < wsa_> anything else related to gen3?
11:32 < horms> yes
11:32 < horms> the bsp team have an etheravb patchset, which i now also have
11:33 < horms> it depends on gpio which ulrich is handling; there were patches for that too which he now has
11:33 < horms> i plan to test the patchset in the near future
11:33 < horms> a blocking item is that it has some non-trivial driver changes that we would like to check for regresssions on 32bit
11:33 < horms> which requires testing
11:33 < wsa_> i see
11:33 < horms> i believe shimoda-san will handle that when he has time
11:34 < wsa_> are the changes needed for basic support or for the extended feature set?
11:34 < horms> in the short term we may be able to get 100Mbit working and on a ML
11:34 < shimoda> horms: yes, i will try that on alt + expantion board
11:34 < horms> there is some special sauce required for futher support; a documentation update is a dependency there
11:34 < shimoda> s/expantion/expansion/
11:35 < horms> wsa_: for basic support, i believe
11:35 < horms> i'll double check that
11:35 < horms> basucically i think we are talking about crash-bugs on 64bit
11:35 < wsa_> i see
11:36 < horms> probably the best things is for me to test things on 64bit, polish things up and post an rfc
11:36 < horms> with no special sauce
11:36 < horms> so i nominate that as my near term goal
11:36 < wsa_> sounds reasonable
11:37 < horms> one question i have; i assume the main use of this to us is nfs root. but that also seems not very useful to the people in europe
11:37 < horms> which is a good portion of us
11:37 < horms> am I reading things right?
11:37 < horms> shimoda: thanks
11:38 < geertu> At least NFS would allow to import/export files in the initrd
11:38 < wsa_> I mainly use initramfs, but my userspace needs are not very high...
11:38 < horms> geertu: good point
11:38 < geertu> But from my experience, just mounting NFS and writing a small file takes a minute
11:38 < horms> anyway, i'll work on it
11:39 < geertu> Right now I'd like to copy my HSCIF1 overlay dtbo to my initrd ;-)
11:39 < geertu> I could modify the initrd
11:39 < geertu> But for exporting, that doesn't work
11:39 < horms> so you tftp things once you are in usespace; in an ideal situation
11:39 < geertu> Probably we can set up an NFS root on the management VM, too?
11:40 < horms> probably
11:40 < geertu> AH, TFTP is an alternative
11:40 < geertu> Probably faster
11:40 < horms> i believe that is magnus's preferred method for himself
11:40 < wsa_> Yup, TFTP does it for me. Also, less to setup
11:40 < horms> but i agree a more local nfs server would be useful
11:41 < geertu> And we already have tftp setup on the management vm
11:41 < horms> but i think its somewhat off topic for this meeting
11:41 < geertu> yep
11:41 < horms> i think thats all for etheravb from me
11:41 < wsa_> thanks, simon
11:42 < wsa_> so, let's move to SCIF
11:42 < geertu> I submitted a new patch series last week
11:42 < wsa_> geertu has sent an awesome new series (actually two) which make DMA work now
11:42 < wsa_> so \o/
11:43 < horms> yay indeed
11:43 < geertu> Major change is submitting the RX DMA request from the RX interrupt, which is what many other serial drivers do
11:43 < geertu> + lots of cleanups, and reordering of the patches to make more sense
11:44 < wsa_> did that give you any more insights about how all this affects non-Rcar SCIF and how/what could be backported to the BSP?
11:44 < geertu> I have to do more testing, though, as you may receive bogus data during overrunrs
11:45 < geertu> Non-R-Car SCIF* may be an issue
11:46 < geertu> We don't have DT-based DMA on any other ARM platform
11:46 < shimoda> wsa_: I think we don't need backport to the R-Car Gen2 BSP because it's maintenance mode
11:46 < geertu> kzm9g/armadillo/ape6evm use SCIFA for console, though
11:47 < geertu> in -legacy they had some DMA support, but it was not enabled for SCIFA (and didn't work when I tried to enable it on Armadillo)
11:47 < horms> do you have access to any of those boards?
11:47 < geertu> I have kzm9g/armadillo/ape6evm
11:47 < horms> a flush!
11:47 < geertu> So PIO SCIFA keeps on working
11:49 < geertu> I could try SCIFA DMA on e.g. Armadillo if I (1) revive -legacy and (2) try to fix the legacy shdma driver for modern residue handling (untested RFC patch posted)
11:49 < geertu> But in the long term, that's insignificant
11:49 < wsa_> i agree
11:49 < geertu> SCIF DMA was originally written for SH (sh7724?)
11:50 < geertu> hut never wired up in the sh platform code, so there are no users.
11:50 < geertu> s/hut/but/
11:50 < geertu> So for now R-Car Gen2 (M2/koelsch) is what we have to work with
11:51 < wsa_> I think we are fine as long as PIO keeps working
11:51 < geertu> And Gen3 soon
11:51 < wsa_> on the old platforms
11:51 < wsa_> yes
11:51 < wsa_> is there any gain if I test them on lager?
11:51 < geertu> I think Shimoda-san already did, at least with the previous version
11:51 < wsa_> except that more testing is always a gain ;)
11:52 < geertu> On Koelsch, I test SCIF/SCIFA/SCIFB/HSCIF on EXIO
11:52 < shimoda> geertu: yes, I tested v2 version on lager. but i don't test v3 yet.
11:53 < geertu> BTW, SCIFB receive doesn't seem to work on one of the EXIOs. It may be a bug in pinctrl (code/doc)
11:53 < wsa_> so, gen2 bsp is in maintenance mode and gen3 bsp could pick up geert's patches as is?
11:54 < wsa_> or do we still have a version gap?
11:54 < wsa_> we talked last time about it
11:54 < geertu> I think Magnus would like the BSP team to switch over once my series is "finished"
11:54 < wsa_> yes, we would all like that, I assume :)
11:55 < horms> that makes sense to me unless there is a pressing need for it in the gen3 bsp in the near term
11:56 < wsa_> sounds good, i don't see any obstacles in how to go forward...
11:56 < shimoda> wsa_: about gen3 bsp plan, it uses upstream version (renesas-driver.git) until end of next year
11:57 < wsa_> next year? wow!
11:58 < wsa_> and yay!
11:58 < wsa_> so, that's it for SCIF?
11:58 < shimoda> oops, gen3 bsp target is "LTSI-2017". so, this might mean mid of next year
11:59 < geertu> I have one more question
11:59 < shimoda> anyway, bsp team looks upstream in this year
11:59 < geertu> What's up with the IRDA MSTP clock and SCIF2?
11:59 < wsa_> mid-2016 is still good IMO
12:00 < morimoto> IrDA on Gen3 is secret feature.
12:00 < morimoto> It will be removed from datasheet
12:00 < morimoto> and, please don't ask it on ML :)
12:01 < geertu> So there's a hole in the SCIFx MSTP2 range, and SCIF2 uses a clock from MSTP3
12:01 < morimoto> About MSTP, IrDA and SCIF2 is using same one
12:02 < morimoto> Yes
12:06 < wsa_> ok
12:07 < wsa_> misc
12:07 < wsa_> i pulled in my i2c-slave updates
12:08 < wsa_> and I am currently scoping the rcar-i2c core to understand all the repeated start/NACK/auto retransmission issues we were seeing
12:09 < horms> shimoda san sent me a Kconfig fix for our ethernet drivers which has been merged
12:09 < wsa_> shimoda-san also got his PWM driver accepted \o/
12:10 < horms> https://git.kernel.org/cgit/linux/kernel/git/davem/net.git/commit/?id=83bc805bff89854e4c81c67633a52ce1015b1502
12:11 < wsa_> that was my misc part
12:11 < shimoda> wsa_: about PWM, yes, the PWM driver seems accepted but i cannot find the pwm maintainer git :)
12:12 < wsa_> yes, he said he will do the final changes but it is not clear if he actually did them
12:12 < geertu> git://git.kernel.org/pub/scm/linux/kernel/git/thierry.reding/linux-pwm.git
12:12 < geertu> it's part of renesas-drivers
12:14 < shimoda> oops, i meant that i could not find the PWM driver commit in his repo
12:16 < wsa_> so, final question would be when to meet next time
12:17 < wsa_> to be honest, I don't think we need the next meeting in two weeks. what about three weeks?
12:18 < wsa_> sorry, just found one thing
12:18 < wsa_> shimoda: what is the status about USB+IPMMU issues?
12:18 < horms> so far my evening schedule is open in both 2 and 3 weeks
12:18 < geertu> shimoda: It's indeed not included yet, probably because it's meant for v4.4
12:19 < geertu> My schedule is open in both 2 and 3 weeks, except on Wednesdays
12:20 < shimoda> wsa_: no update for now
12:20 < shimoda> geertu: thank you for the checking
12:21 < geertu> shimoda: we have to have a bit more patience, as not many patches will be accepted between now and v4.3-rc1
12:22 < geertu> (this also applies to the Gen3 patch stack!)
12:22 < morimoto> About next meeting, I don't want 16th Sep, other days are OK
12:22 < morimoto> 16th and 23th Sep
12:23 < wsa_> geertu: well, thierry could apply the new-driver-rule. but it is his choice, of course...
12:23 < horms> btw, 21, 22, 23 are holidays in the land of the rising sun
12:23 < wsa_> i suggest Sep 15th
12:23 < wsa_> same time
12:23 < morimoto> 15th Sep is OK for me
12:24 < horms> This time its in my schedule properly :)
12:24 < shimoda> geertu: I got it
12:24 < wsa_> \o/
12:24 < shimoda> wsa_: 15th is OK to me
12:26 < geertu> booked
12:26 < morimoto> same here
12:26 < wsa_> great
12:27 < wsa_> that's it I assume
12:27 < wsa_> thank you!
12:27 < horms> wsa_: thanks
12:27 < geertu> thx
12:27 < morimoto> Thanks
12:27 < morimoto> Bye x2
12:27 < wsa_> have a great evening/rest of day!
12:28 < shimoda> thank you, bye!
12:28 -!- shimoda [~shimoda@relprex2.renesas.com] has quit Quit: WeeChat 0.4.2
12:28 < geertu> bye
--- Log closed Wed Aug 26 12:30:04 2015