Core-chat-meeting-2016-01-06

10:03 < geertu> Today's Agenda:
10:03 < geertu> 1. Upcoming Gen3 development for the Core group,
10:03 < geertu> 2. Status check for tasks from last meeting - what is remaining?
10:03 < geertu> Topic 1. Upcoming Gen3 development for the Core group
10:04 < dammsan> i'm working on the IPMMU
10:04 < dammsan> i've updated the boot loader and 4GiB of RAM is supposed to be available
10:05 < geertu> "supposed to be"
10:05 < dammsan> need to test that, and try to use IPMMU with the DU/VSP
10:05 < dammsan> i'd like to sync the boot loader update with other people as well
10:05 < pinchartl> dammsan: I was going to mention that
10:06 < dammsan> morimoto: shimoda: how can we share boot loader binaries with all the members?
10:06 < pinchartl> should we make sure we all use the same bootloader ?
10:06 < dammsan> i think so
10:06 < pinchartl> or should we make sure we test our kernel on a variety of bootloaders ?
10:06 < geertu> Both are valuable options
10:06 < dammsan> there are too many combos
10:06 < morimoto> dammsan: do you mean export problem ?
10:06 < dammsan> if you mix and match
10:07 < dammsan> morimoto: that could be some issue. i was more thinking of how to provide the information
10:07 < geertu> We mix toolchains, kernel configs, userland, ... too
10:07 < dammsan> morimoto: is it under NDA?
10:07 < dammsan> geertu: that is true
10:08 < pinchartl> I believe it depends on whether we can expect customers to use a recent boot loader
10:08 < morimoto> dammsan: it is difficult question...
10:08 < geertu> The bad thing (compared to toolchains, kernel configs, userland) is that the firmware is a black box
10:08 < pinchartl> there's no point in testing a boot loader that is not used
10:08 < dammsan> table.txt describes six different components at this point
10:08 < dammsan> just to get to u-boot
10:09 < dammsan> i recommend using complete releases, instead of mixing and matching
10:09 < dammsan> morimoto: can you ask ito-san about how to share boot loader stuff?
10:09 < morimoto> dammsan: OK but tomorrow
10:09 < dammsan> morimoto: i think other guys like baylibre also need new boot loader
10:09 < dammsan> sure
10:10 < dammsan> maybe we can have boot loader upgrade fest at peripericon
10:10 < dammsan> anyway, i'm poking around with those things
10:10 < dammsan> other people?
10:11 < neg> working on SYS-DMAC, making slow progress :)
10:11 < dammsan> neg: i can share my older gen2 bits with you
10:11 < dammsan> not sure if i made them public
10:12 < dammsan> i probably did
10:12 < neg> '[PATCH] ARM: shmobile: IPMMU and DMAC prototype' ?
10:12 < pinchartl> dammsan: table.txt ? do you mean table.html ?
10:13 < geertu> dammsan: How many boards will survive the boot loader upgrade and capacitor soldering fest? ;-)
10:13 < dammsan> neg: need to look into detail, lets chat more later after this meeting
10:13 < neg> geertu: did you evere here back from Vinod Koul for on '[PATCH] dmaengine: use phys_addr_t for slave configuration' ?
10:13 < neg> dammsan: ok
10:13 < geertu> neg: 2015-12-15: "I should be able to do this in next few days...
10:14 < geertu> "
10:14 < dammsan> pinchartl: table.txt is from SCIF D/L mode boot loader update script sent yesterday
10:14 < dammsan> geertu: prediction - zero =)
10:14 < pinchartl> dammsan: ah ok
10:15 < shimoda> about boot loader, yocto team releases it at github, but it is a few old
10:15 < shimoda> https://github.com/renesas-rcar/meta-renesas/tree/fido/meta-rcar-gen3/recipes-bsp
10:15 < geertu> How many of these 6 parts contain GPLed code?
10:18 < pinchartl> I see 3 boot loaders, one (un)trusted OS, a certificate and boot parameters
10:19 < pinchartl> do we know how the pieces fit together exactly ?
10:20 < pinchartl> it looks like source code is available for everything but the boot parameters and the certificate
10:20 -!- shimoda [~shimoda@relprex1.renesas.com] has quit [Ping timeout: 240 seconds]
10:20 -!- morimoto [~user@relprex1.renesas.com] has quit [Ping timeout: 260 seconds]
10:21 < pinchartl> will dammsan be next ?
10:21 < dammsan> shimoda: indeed looks a bit old
10:21 < dammsan> geertu: fortunately i don't have to look inside the box
10:22 < dammsan> if you find something fishy let us know and we will try to improve the situation
10:22 < dammsan> neg: i sent you my more recent gen2 experiment via email
10:23 < dammsan> pinchartl: you only have to care about running the script that i sent you
10:23 < neg> dammsan: thanks
10:23 < dammsan> good thing is that the SoC has Mask ROM SCIF D/L support so it is difficult to brick
10:23 < pinchartl> dammsan: with the binaries
10:24 < pinchartl> difficult or impossible ? :-)
10:24 < dammsan> pinchartl: if you have time then i can think of many more important things to do =)
10:24 < pinchartl> :-)
10:25 < dammsan> pinchartl: i build my own u-boot and omitted some secure code when i received the board during summer vacation
10:25 < dammsan> but there is enough bugs as-is without trying to poke around
10:25 < dammsan> for instance, to get IPMMU working update is needed
10:25 < dammsan> i am also pretty sure that SMP will need more work
10:26 < pinchartl> do we need to update before february ?
10:26 < dammsan> anyway, i think we should consider upgrading rather soon
10:27 < dammsan> so we can enable more memory upstream
10:27 < pinchartl> ok
10:27 < dammsan> lets see what morimoto-san and shimoda-san says
10:27 < dammsan> what are other people hacking on?
10:28 < geertu> I want to look into SYSC PM Domains and integrate it with your DT CPU bringup series
10:28 < dammsan> cool
10:29 < dammsan> when i hear PM i am also thinking of making sure Suspend-to-RAM works
10:29 < dammsan> ideally i wanted to get that going before enabling SMP
10:29 < geertu> SMP is already enabled ;-)
10:30 < geertu> And I want to fix suspend with CONFIG_DEBUG_RODATA=y
10:30 < dammsan> yeah, so now we need to consider SMP as well
10:30 < dammsan> which drags in the boot loader dependency
10:30 < dammsan> geertu: we were talking about SCIF hardening with DMAC
10:31 < dammsan> part of DMAC enablement i guess
10:31 < geertu> "ARM: mm: flip priority of CONFIG_DEBUG_RODATA" revealed a big in the shmobile asm suspend code. The patch was reverted, but will come back in next merge window
10:31 < geertu> Yeah, another thing I plan to do
10:32 < geertu> s/big/bug/
10:32 < dammsan> geertu: may i ask about ordering
10:32 < dammsan> with SYS-DMAC enablement and also PM
10:32 < geertu> Sure
10:32 < geertu> What's your preferred order?
10:32 < dammsan> does it make sense to enable some initial support before updating the boot loader?
10:33 < dammsan> geertu: did SMP work for you or not - i don't remember
10:33 < geertu> Yes, SMP works (all 8 cores)
10:33 < geertu> If it doesn't work, try SW10
10:33 < dammsan> good!
10:33 < dammsan> i've spent quite some time with SW10 updating boot loaders thanks =)
10:34 < dammsan> so i guess enabling DMAC is a rather good activity
10:34 < dammsan> doing that this month, does that make sense?
10:35 < geertu> Sure
10:35 < dammsan> and later on roll in IPMMU
10:35 < dammsan> how about the PM bits
10:35 < dammsan> doing Suspend-to-RAM before or after or parallel?
10:36 < dammsan> i think the timer needs some work too
10:36 < geertu> after / parallel (if the other becomes blocked)?
10:37 < dammsan> sounds good
10:37 < dammsan> it would be neat to get kexec working too
10:37 < geertu> Is it working now on arm64 in general?
10:37 < dammsan> i would assume no
10:38 < dammsan> so geertu: for SCIF hardening and DMAC enablement, what is needed apart from DTS?
10:39 < dammsan> i saw your recent BRG patches, shall Mbps setting over loopback work?
10:39 < geertu> It should
10:40 < dammsan> great!!
10:40 < geertu> dammsan: Fixing the hardware (perhaps)?
10:40 < dammsan> obviously with b0rk timer nothing works well
10:41 < dammsan> geertu: what pending issue are you thinking of?
10:41 < geertu> last time I tried, RX SCIF DMA didn't work at all
10:41 < geertu> (Yes, I have to retry)
10:41 < geertu> on r8a7795
10:42 < dammsan> geertu: maybe we can revisit after data sheet update and boot loader update
10:42 < dammsan> how about other people?
10:43 < dammsan> pinchartl: you were proposing to conver the IOMMU driver to be more modern for ARM32? =)
10:43 < geertu> dammsan: Speaking about datasheet update, I have to look into the recent errata for CPG
10:43 < pinchartl> dammsan: yes
10:44 < dammsan> geertu: yeah, errata.. also we were talking about keeping the GPIO MSTP bit enabled
10:44 < pinchartl> if we want to use the same IOMMU driver on ARM32 and ARM64 we'll need that
10:45 < dammsan> pinchartl: i was hoping that you could work with neg for the framework bit for IPMMU + DMAC
10:45 < pinchartl> do you mean guiding Niklas or writing the code myself ? :-)
10:45 < dammsan> pinchartl: i was never opposed to that change, but i did not want to block r8a7795 prototype support on that
10:46 < dammsan> pinchartl: how about you work together and figure out the details by yourself? =)
10:46 < pinchartl> I won't have time to do the work in January, but I can guide Niklas
10:46 < dammsan> geertu: what do we have on the todo list that need attention from pinchartl?
10:47 < dammsan> i want renesas-drivers to have working DU VGA-out on Salvator-X
10:47 < dammsan> so i can poke around with IPMMU on top of that
10:48 < pinchartl> that is on my to-do list
10:48 < geertu> dammsan: pinctrl documentation acked-by ;-)
10:48 < dammsan> pinchartl: guiding neg sounds good
10:48 < pinchartl> speaking of DU VGA out
10:48 < pinchartl> that involves the VSP1
10:48 < pinchartl> and thus V4L2
10:48 < geertu> dammsan: a patch fixing the vsp1 BUG() with media-next?
10:48 < pinchartl> I'm not in best terms with the V4L2 maintainer at the moment
10:48 < pinchartl> which could introduce delays in getting patches upstream
10:49 < pinchartl> geertu: that's an open issue
10:49 < pinchartl> media-next is broken
10:49 < dammsan> pinchartl: thats fine
10:49 < pinchartl> and if it comes to the worst I'll nack the pull request to Linus
10:49 -!- shimoda [~shimoda@relprex1.renesas.com] has joined #periperi
10:49 < geertu> shimoda-san: Welcome back!
10:50 < dammsan> pinchartl: we need to have something working for use by ourselves and the BSP team
10:50 < pinchartl> dammsan: internally or upstream ?
10:50 < dammsan> in renesas-drivers to begin with
10:50 < dammsan> upstream too, but i understand it may be difficult to control that timing
10:50 < pinchartl> it's very difficult at the moment
10:50 < geertu> Does vsp1 work with current renesas-drivers (i.e. excl. current media-next)?
10:51 < pinchartl> it should but I haven't updated renesas-drivers for about a month
10:51 < pinchartl> I'll test it
10:51 < dammsan> gpinchartl: please do =)
10:51 < pinchartl> for upstream I'm not sure what I can do
10:52 < pinchartl> neither now nor in the future
10:52 < dammsan> please guide morimoto with HDMI audio
10:52 < pinchartl> HDMI audio on Gen3 ?
10:52 < shimoda> geertu: hi :)
10:52 < dammsan> gen2 to begin with
10:52 < pinchartl> that should be fine
10:53 < pinchartl> but isn't this a core meeting, not a multimedia meeting ? :-)
10:53 < dammsan> aright
10:53 < pinchartl> I'm fine discussing multimedia
10:53 < pinchartl> but we have a multimedia meeting tomorrow
10:53 < pinchartl> I don't want to hijack today's meeting
10:53 < dammsan> yes lets do that then
10:53 < dammsan> but i need to make sure we can consume your stuff
10:53 < dammsan> and it goes through geerts tree
10:54 < dammsan> DU+VSP is starting to become a bottleneck
10:54 < dammsan> thanks for looking into that
10:54 < pinchartl> I'll make sure it works with renesas-drivers excluding media-next
10:54 < geertu> git rerere is my friend
10:54 < dammsan> please work with geert to sort it out
10:54 < pinchartl> but for upstream I have no idea how things will turn out
10:54 < pinchartl> if I nack the media-next pull request my ability to get anything merged in media will be diminished
10:54 < dammsan> pinchartl: please talk with hayama-san about that too
10:55 < geertu> For now I can keep excluding media-next, but as soon as Linus has pulled it, we're in trouble
10:55 < pinchartl> if I don't nack it, all kind of crap will end up in mainline
10:55 < geertu> Good, Linus doesn't like crap
10:56 < pinchartl> Linus knows about this already, I know that Mauro has personally contacted him to complain about me
10:56 < pinchartl> I don't know what the outcome of their discussion was though
10:56 < geertu> oops
10:58 < dammsan> regarding upstream merge, i guess we're waiting for the CPG and integration code to hit mainline?
10:58 < pinchartl> upstream merge of what ?
10:58 < geertu> They're in next
10:58 < dammsan> yeah
10:58 < dammsan> a lot of integration code is built but not yet merged upstream
10:59 < dammsan> not sure if any V4L2 explosion will affect ARM64 merge?
10:59 < pinchartl> dammsan: regarding the upstream merge of what ?
10:59 < dammsan> pinchartl: basic initial soc support
11:00 < pinchartl> ok
11:00 < dammsan> i keep my fingers crossed
11:01 < dammsan> neg: do you have any short term "easy stuff" on your todo?
11:01 < dammsan> you may be busy with what you have, but if you need something simple to switch to then please check with geert
11:01 < neg> will do thanks
11:02 < pinchartl> by the way, this is off-topic fore core, but if anyone has advices regarding nacking the media-next pull request vs. not doing it and resigning from V4L2 development, I'll be happy to hear them. we can discuss that later
11:02 < dammsan> i suggest expressing your disagrement lightly in a public space and then take some vacation
11:03 < dammsan> vacation is underrated
11:04 < pinchartl> I wish I could take vacations now :-)
11:05 < dammsan> speaking of which, we need to find some sight seeing spot for peripericon and the nice weather
11:06 < dammsan> pinchartl: let me know how i can help you
11:06 < geertu> FOSDEM is the coldest period of the year in Belgium
11:06 < geertu> Still hot for the Finnish
11:06 < pinchartl> dammsan: if I know how you could, I'd have asked you already :-)
11:09 < dammsan> geertu: we can perhaps browse the local town
11:10 < dammsan> while wearing snow board gear
11:10 < geertu> http://www.oudemarktleuven.be/
11:10 < geertu> Longest bar in the world ;-)
11:11 < geertu> Any other Upcoming Gen3 development for the Core group?
11:12 < dammsan> geertu: you cant go wrong with that
11:13 < dammsan> geertu: maybe not development, but did anyone plan on building loopback adapters following my plan (or something else?)
11:13 < dammsan> i wanted to do MSIOF loopback but that did not work, need to redo on Gen2 for testing purpose
11:13 < geertu> MSIOF is broken on Gen3
11:14 < geertu> All my (3) adapters are in-use on Koelsch
11:14 < geertu> Will try your setup for SCIF DMA on Salvator-X
11:15 < dammsan> neg: my plan is to give you a break out adapter at peripericon
11:16 < dammsan> geertu: sounds good
11:16 < dammsan> pinchartl: just one final note about IPMMU on Gen3 - some errata exists so i need to keep on doing prototype code with workaround
11:17 < neg> nice, break out adapter for what interface?
11:17 < dammsan> a general purpose thing located by geert, you can use it for many different things
11:17 < dammsan> for instance serial port loopback
11:17 < pinchartl> dammsan: other than the errata I've already received ?
11:17 < neg> ahh ok
11:17 < dammsan> geertu: can you look into using the DT overlays then?
11:18 < dammsan> pinchartl: there is no public information available yet
11:18 < geertu> dammsan: Sure, using overlays all over the place
11:18 < dammsan> or even closed one
11:18 < geertu> dammsan: topic/overlays and topic/renesas/overlays
11:18 < dammsan> geertu: great
11:18 < dammsan> so expect slow IPMMU progress
11:19 < dammsan> thats all from my side
11:19 < geertu> neg: http://www.zebax.com/doc/ZX1/ZX145QTE-020.pdf
11:19 < pinchartl> dammsan: I have received
11:19 < pinchartl> R-CarGen3_HardWare_Manual_Errata_for_Rev050_Oct_31.xlsx
11:19 < pinchartl> R-CarGen3_HardWare_Manual_Errata_for_Rev050_Nov_30.xlsx
11:19 < pinchartl> R-CarGen3_HardWare_Manual_Errata_for_Rev050_Dec_31.xlsx
11:19  * geertu too
11:21 < shimoda> oops, I also have Aug and Sep versions, so I will send them to you later via jinso
11:21 < pinchartl> shimoda: thank you
11:21 < geertu> Would be great, thanks!
11:25 < geertu> Topic 2. Status check for tasks from last meeting - what is remaining?
11:25 < geertu> https://osdr.renesas.com/projects/linux-kernel-development/wiki/Core-todo-list
11:25 < geertu> Is Simon still in the US?
11:25 < dammsan> AU
11:25 < dammsan> i think he arrived in japan on friday
11:25 < dammsan> arrives
11:26 < dammsan> so yes
11:26 < geertu> But still living in a US timezone ;-)
11:26 < dammsan> somehow
11:26 < geertu> I think I have to s/v4.4/v4.6/g and s/v4.5/v4.6/g?
11:27 < geertu> Except for CPG/MSSR
11:27 < dammsan> most likely
11:29 < geertu> dammsan: What happened to adding renesas,intc-ex-r8a7795 to the driver?
11:30 < dammsan> geertu: nothing yet, IRQC and timers are still on my todo. currently overwhelmed by IPMMU workaround stuff
11:31 < dammsan> geertu: i got as far as cooking up a loop back adapter for IRQ pins, but not yet tested
11:34 < geertu> Anything else?
11:35 < dammsan> not from me
11:35 < geertu> For next meeting, I propose Tuesday, January 19, 18h00-19h30 JST / 10h00-11h30 CET
11:36 < pinchartl> geertu: I'll be away that week
11:36 < pinchartl> (on vacation actually, but with e-mail access)
11:36 < shimoda> I'm OK at 19th
11:37 < geertu> pinchartl: Probably we can do without you for once?
11:37 < pinchartl> geertu: you've managed so far, so I think you'll be fine :-)
11:37 < geertu> I'd like not to postpone by one week, as two weeks later we have PeriPeriCon anyway
11:38 < dammsan> 19th sounds good
11:38 < uli___> same here
11:38 < neg> works for me
11:39 < pinchartl> geertu: no worries at all
11:39 < pinchartl> I wasn't asking you to postpone :-)
11:40 < geertu> I just wanted to make you feel more important ;-)
11:40 < pinchartl> :-)
11:40 < dammsan> i think the renesas network is down, but we can assume shimoda-san and morimoto-san are ok
11:41 < geertu> I do see Shimoda-san talking...
11:41 < shimoda> I'm here :)
11:42 < shimoda> however, Morimoto-san already went to home
11:42 < dammsan> geertu: good - that makes one of us
11:42 < dammsan> or two when i look a bit closer
11:43 < geertu> OK, Time to go home ;-)
11:44 < geertu> Thanks for joining