diff options
Diffstat (limited to 'wiki/Chat_log/20180607-io-chatlog')
| -rw-r--r-- | wiki/Chat_log/20180607-io-chatlog | 620 |
1 files changed, 620 insertions, 0 deletions
diff --git a/wiki/Chat_log/20180607-io-chatlog b/wiki/Chat_log/20180607-io-chatlog new file mode 100644 index 0000000..3cae41d --- /dev/null +++ b/wiki/Chat_log/20180607-io-chatlog @@ -0,0 +1,620 @@ +09:03 < wsa_> Welcome to this IO meeting +09:03 < wsa_> first the status updates with a focus on the C) parts +09:03 < pinchartl> hello +09:04 < wsa_> A - what have I done since last time +09:04 < wsa_> ------------------------------------ +09:04 < wsa_> Geert +09:04 < wsa_> : upported R-Car M3-N HSCIF support (incl. PFC), discovered SCIF work_tx race condition, +09:04 < wsa_> discussed MSIOF related BSP patch about DMA +09:04 < wsa_> Kaneko-san +09:04 < wsa_> : posted M3-W PCIE support +09:04 < wsa_> Marek +09:04 < wsa_> : has posted multiple iterations of PCIE fixes, with success +09:04 < wsa_> Niklas +09:04 < wsa_> : sent out two SDHI related patch series (tuning behaviour and reset procedure), +09:04 < wsa_> retested SDIO on Koelsch (still fails, but a step later now) +09:04 < wsa_> Shimoda-san +09:04 < wsa_> : further investigated USB susped/resume issue, sent out new versions for USB role switch feature, +09:04 < wsa_> upstreamed USB for E3, and communicated various issue from the BSP team to us +09:04 < wsa_> Simon +09:04 < wsa_> : began reworking HS400 patches as per Ulf's feedback +09:04 < wsa_> Wolfram +09:04 < wsa_> : discussed various items and implemented proposed solutions (SDHI WP regression, I2C DMA RX handling, +09:04 < wsa_> I2C recovery flaw), upported SDHI DATA_STROBE and faster I2C speed patches, discussed periject tool, +09:04 < wsa_> added new fault injector to i2c-gpio, proposed new QEMU I2C passthrough sketch to Magnus, +09:04 < wsa_> various patch review and some backporting assistance, prepared travel to Japan +09:04 < wsa_> B - what I want to do until next time +09:04 < wsa_> ------------------------------------- +09:04 < wsa_> Geert +09:04 < wsa_> : wants to fix the SCIF work_tx issue +09:04 < wsa_> Kaneko-san +09:04 < wsa_> : wants to continue upporting M3-W and M3-N PCIE support +09:04 < wsa_> Niklas +09:04 < wsa_> : wants to keep upporting SDHI and keep at SDIO testing if time allows +09:04 < wsa_> Shimoda-san +09:04 < wsa_> : wants to keep at the USB role switch patches, continue to plan for GPIO PHY support for E3, +09:04 < wsa_> learn about KVM and USB virtualization +09:04 < wsa_> Simon +09:04 < wsa_> : wants to continue HS400 upstreaming activity +09:04 < wsa_> Wolfram +09:04 < wsa_> : wants to continue the above discussions, prepare the talk for ALS18, continue +09:04 < wsa_> SPDX conversion for Renesas drivers, start implementing QEMU solution (f Magnus +09:04 < wsa_> agrees) +09:04 < wsa_> C - problems I currently have +09:04 < wsa_> ----------------------------- +09:04 < wsa_> Niklas +09:04 < wsa_> : is still waiting for Gen3 thermal fuses +09:04 < wsa_> Shimoda-san +09:04 < wsa_> : has to struggle with microB connector on E3 Ebisu instead of microAB +09:04 < wsa_> Simon +09:04 < wsa_> : wants to discuss how to handle different HS400 versions on Gen3 +09:04 < wsa_> Wolfram +09:04 < wsa_> : is travelling until after next meeting, so has no direct HW access +09:05 < wsa_> neg: yes, well, I have it in the IO todo file, so it won't be forgotten. I guess no need to wait bi-weekly for it :/ +09:06 < wsa_> shimoda: well, not much we can do with the "wrong" connector, or? :) +09:06 < wsa_> horms: so far, I'd think the 0,4,8-tap decision is per-SoC only +09:06 < horms> wsa_: agreed +09:06 < neg> wsa_: check, will drop it from future status updates +09:07 < wsa_> I wonder if such a thing will change from one ES to another? +09:07 < wsa_> does JapaPERI have an idea? +09:07 < horms> My question is like this +09:07 < horms> Currently its between SoCs +09:07 < horms> and traditionally we handled such differences via the compat string +09:07 < horms> which may well be the cleanest option here +09:07 < horms> but it may be also possible to use soc_match +09:08 < neg> horms: in BSP there is a define HS400_USE_4TAP and bsp commit ce12154b826b955b6439546014870d038eff19ce seems to indicate this might be per SoC +09:08 < horms> Keep in mind that the per-SOC compat strings are more or less unimplemented for Gen3 SoCs in the SDHI driver at this time +09:08 < geertu> If there's no need to check for SoC revision, IMHO using the compat string is the easiest. +09:08 < wsa_> I'd agree with geertu +09:08 < horms> ok, let me do that then +09:09 < horms> no objection from my side +09:09 < wsa_> we can switch to soc_device_match when needed` +09:09 < wsa_> ? +09:09 < geertu> If you store (a pointer to) the features in soc_device_id.data, it's easy to retrieve. +09:09 < horms> yes, if its an implementation detail that should be possible +09:09 < horms> geertu: ok, thanks +09:10 < shimoda> wsa_: correct. i cannot development with the wrong connector. So, i intend to use DIP-SW gpio instead of actual ID signal to develop :) +09:10 < geertu> If a SoC revision check is needed later, it can be easily handled by taking the (pointer to) the features) from soc_device_match()->data +09:10 < wsa_> shimoda: role-DIP-switch, nice :D +09:11 < Marex> shimoda: can't you use hotair and replace the connector ? :) +09:11 < geertu> horms: drivers/media/platform/rcar-vin/rcar-core.c is a nice example: +09:11 < geertu> vin->info = of_device_get_match_data(&pdev->dev); +09:11 < geertu> attr = soc_device_match(r8a7795es1); +09:11 < geertu> if (attr) +09:11 < geertu> vin->info = attr->data; +09:12 < wsa_> so, with me being on the road already... +09:12 < shimoda> wsa_: :D +09:12 < wsa_> dammsan: can you give me remote access to H3 ES2.0 Salvator-XS? So, I can check stuff remote at least +09:13 < shimoda> Marex: it's difficult to me (my skill and Renesas office rule) +09:13 < horms> geertu: I am confused now. I see that is an exmple of using soc_device_match. And indeed there are already examples in the SDHI driver. But I thought that we decided to use compat strings for the case in question? +09:13 < geertu> horms: You can start with of_device_get_match_data() +09:14 < Marex> shimoda: ah :) +09:14 < geertu> If needed later, you add the next 3 lines enabling soc_device_match(), but needing no further driver logic change +09:14 < horms> Oh, I see. Very nice. +09:15 < wsa_> So, anything more from your side about the status updates? +09:16 < Marex> wsa_: PCIe32 got one reply ! +09:16 < wsa_> \o/ +09:16 < Marex> wsa_: and I keep resubmitting DA9063L patchset +09:17 < wsa_> Maybe some of the ARM core people are in Tokyo? It seems fastest to me to poke them personally, or get the right group together there... +09:17 < Marex> wsa_: let's see what happens at OSSJ ? +09:18 < Marex> wsa_: although, I doubt that +09:18 < wsa_> I share that feeling, but let's hope a bit... +09:18 < wsa_> okay, for the topic, there is only "add. tasks" from my side +09:18 < wsa_> although there is not much change for the IO group +09:19 < wsa_> pinchartl: notification for you :) +09:19 < wsa_> the focus is still on upporting and we should have increased base time for that +09:19 < wsa_> I'll contact you about upporting individually +09:19 < pinchartl> wsa_: I'm here :-) +09:20 < pinchartl> dammsan: ping +09:20 < pinchartl> we're missing Morimoto-san +09:20 < pinchartl> kbingham: ping +09:20 < wsa_> Add. tasks are possible on request which then are discussed by Renesas +09:20 < dammsan> wsa_: port 9006 is yours +09:20 < wsa_> but no request was made to me, and I don't see one urgent in my list +09:20 < wsa_> dammsan: thanks! +09:21 < kbingham> hola +09:21 < dammsan> pinchartl: whatsup? +09:21 < wsa_> so, doesn't look like add. tasks for IO +09:22 < shimoda> pinchartl: Morimoto-san has a trouble about his PC network setting on Ubuntu 18 :) +09:22 < wsa_> dammsan: unless there is something new I don't know up to now? :) +09:22 < pinchartl> dammsan: Wolfram mentioned additional tasks +09:22 < pinchartl> I requested that to be added to the agenda +09:22 < pinchartl> as this topic concerns all groups, should we discuss it now ? +09:23 < wsa_> Frankly, I am more interested in the base time distribution again +09:23 < dammsan> sure what do you want to know? +09:24 < pinchartl> dammsan: I want to know how we can get additional tasks working +09:24 < pinchartl> as in scheduled in time +09:24 < wsa_> horms: I assume you are so busy backporting LTSI that ravb upporting had to stand back? +09:24 < pinchartl> and not repeat the Q2.2 fiasco +09:24 < dammsan> well i think we need to do two things: +09:24 < pinchartl> we've had a lengthy e-mail discussion but that wasn't public, so I'd like to broaden the audience as I don't think I'm the only one to be bothered by this +09:24 < dammsan> 1) agree on quarterly schedule for meeting dates and additional due dates first +09:25 < horms> wsa_: correct +09:25 < shimoda> wsa_: after discuss about add. task, I'd like to ask you about i2c recovery. Did you see any issues (e.g. write data to a slave wronglty) on your R-Car environment? +09:25 < dammsan> 2) do task break out ahead of time +09:25 < horms> wsa_: also, I think that the ravb backporting is a very low yeild exercise +09:25 < dammsan> pinchartl: who else is bothered? +09:25 < pinchartl> kbingham: the mic is yours :-) +09:26 * kbingham has just about caught up wit hthe mornings mails ... +09:26 < pinchartl> who thinks additional tasks work just fine, and who tihnk they don't ? +09:26 < pinchartl> s/tihnk/thinks/ +09:26 < wsa_> shimoda: sure! +09:26 < kbingham> dammsan, I agree where you said this needs an F2F ... and fortunately one is coming up :) +09:26 < kbingham> Lets make sure we have time :) +09:26 < wsa_> horms: "low yield" means most patches are not suitable for upstream? +09:26 < dammsan> unfortunately not all members are present +09:27 < geertu> Personally, I've never been a big fan of additional tasks. +09:27 < horms> wsa_: yes, I think so +09:27 < horms> But I'll go over them again +09:27 < kbingham> Essentially - for me -the *big* problem is the 'rigid square wave workload' which it causes me. +09:27 < pinchartl> I've never been a big fan of additional tasks either. they were a minor nuisance when they were introduced, but they got more and more annoying over time as they got split in two batches per quarter, and we always got late negotiating them +09:28 < wsa_> "square wave workload" :D +09:28 < kbingham> But in particular - where the due dates are fixed, but tasks are always started late. +09:28 < kbingham> I realise that this last AT is a bit of a special case - but my concerns with planning have started long before. +09:28 < uli___> if i may weigh in: i like that additional tasks tell me what to do; i'm not a self-starter when it comes to work... +09:29 < uli___> the contractual framework is secondary to me +09:29 < kbingham> And it's just this one has been a cliff-edge drop for me :( +09:29 < geertu> kbingham: One very important thing to consider is that this is a prototype +09:29 < dammsan> so historically they were created to allow lower latency operation for our group +09:29 < geertu> which means many more unknowns than "normal" additional tasks +09:30 < dammsan> as opposed to have all the fixed tasks agreed on 3 months ahead +09:30 < pinchartl> dammsan: they have created a much higher latency +09:30 < geertu> IMHO they increase latency. +09:30 < dammsan> sure +09:30 < dammsan> but how can we keep the quality and perform development (not single patch wanking) and reduce paper work? +09:30 < geertu> If something is urgent, there may not be any base time available at that moment, as ATs have a fixed deadline, and thus preempt any base work +09:31 < kbingham> uli___, I also like that an AT specifies what to do - but in the latest case - that didn't happen - until it was too late for me to realistically be able to deliver that. +09:31 < pinchartl> in my opinion, they only advantage in additional tasks, and the reason why they have been introduced, is to give a stick to Renesas to hit developers who don't deliver. with fixed deliverables and fixed due dates the work has to be completed, no excuse +09:31 < pinchartl> apart from that I see no use for additional tasks +09:32 -!- morimoto [~user@relprex2.renesas.com] has joined #periperi +09:32 < geertu> An AT deadline around the closing of a maintainer's merge window causes patches to be delayed by one kernel release. +09:32 < geertu> pinchartl: +1 +09:32 < wsa_> horms: if you could add what you currently know to the periupport list? so, we at least know which patches to "drop" +09:32 < dammsan> so we have been discussing dropping AT before +09:33 < dammsan> but at that point i requested some way to get plans from the group leaders +09:33 < horms> wsa_: sure, sorry for not having already done that +09:33 < pinchartl> dammsan: we can reduce paperwork and increase quality by dropping additional tasks completely and measuring performances to make sure we don't slow down (or even sleep) +09:33 < dammsan> this was last year i recall +09:33 < pinchartl> alternatively, we can keep fixed-price, fixed-term tasks, but with a higher latency. they *have* to be negotiated in advance +09:33 < dammsan> it is not a matter of increasing speed +09:34 < pinchartl> that is a contract has to be signed at least 2 months before the due date +09:34 < dammsan> it is a matter of improving visibility and communication +09:34 < jmondi> toothless jmondi's here +09:35 < dammsan> i will ask renesas if they can start considering 2 month delay +09:35 < dammsan> it is perfect for the last quarter of this year +09:35 < pinchartl> if the goal is to improve visibility and communication, then let's focus on that +09:35 < dammsan> until that time perhaps it is best that we drop additional tasks all together? =) +09:36 < dammsan> please note that some pipe lining is required +09:36 < pinchartl> dammsan: and I will also say that I agree we need to measure performances +09:36 < dammsan> one thing is the performance +09:36 < dammsan> another thing is to agree on what to do +09:37 < dammsan> i would like to have both +09:37 < kbingham> I'd like to not have high pressure workloads with short timescales :D +09:37 < dammsan> sure +09:37 < kbingham> I'd go join a tiger team if I wanted that ;D +09:38 < dammsan> but tell me how your 5 weeks can be split onto the existing slots +09:38 < neg> That is what I like about additional contracts, it provieds me with a clear channel from Renesas on what new task/feature is moste valuable for them and therefor I should focus on it. +09:38 < pinchartl> neg: I can only agree partially to that +09:39 < pinchartl> we are tasked with proposing additional tasks +09:39 < Marex> I have to admit, I like my kinda-free reign over U-Boot, since I can focus on doing what needs to be done when it needs to be done +09:39 < pinchartl> without being told what Renesas needs +09:39 < Marex> but maybe Im wrong :) +09:39 < pinchartl> so we try to guess +09:39 < kbingham> Marex, Do you not have additional tasks ? +09:39 < pinchartl> and then we are either told that the guess was right +09:39 < dammsan> he does but he prefers not to consume them +09:39 < pinchartl> or that it was wrong, in which case we are asked to guess again +09:39 < Marex> kbingham: I do, but I can live with that +09:40 < dammsan> what a nice fellow =) +09:40 < pinchartl> neg: I don't count that as being told by Renesas what is valuable for them :-) +09:40 < dammsan> pincharl: you seem to think the content of the bsp up porting list is enough +09:41 < neg> pinchartl: agreed, sometime the planing and late scheduling/decision point for a add. task is messy and creates higher workload. All I'm saying is that for me the task once scheduled provieds value +09:41 < pinchartl> Marex: I think most of us would like to focus on what needs to be done when it needs to be done, with additional inputs from Renesas when they have specific urgent needs +09:42 < dammsan> so may i ask if fixing the quarterly dates ahead of time has any drawback? +09:42 < kbingham> dammsan, What benefit are you addding with fixed dates... +09:42 < dammsan> i am not saying it solves all the problems +09:42 < pinchartl> dammsan: do you mean the Q2.1 and Q2.2 due dates, as in Q+1.5 and Q+2.5 ? +09:42 < kbingham> the 'due dates' are already predictable ... +09:42 < dammsan> this means we are clear about the dates +09:43 < dammsan> yes and meeting dates for chat meetings +09:43 < dammsan> if we agree on these ahead of time +09:43 < pinchartl> kbingham: the due dates are not just predictable, they are fixed :-) +09:43 < dammsan> and yet we fail to schedule tasks, then we won't schedule any tasks +09:43 < kbingham> :) +09:43 < pinchartl> dammsan: but we have already agreed, haven't we ? +09:44 < dammsan> i thought we had agreed to the AT due dates somehow +09:44 < dammsan> but yet it seems that the discussion of no time available and conference timing seems to happen again and again =) +09:44 < geertu> dammsan: Agreed as in: yes, let's use the dates Jinso needs the deliverables? +09:44 < dammsan> so if we know ahead of time.. +09:44 < pinchartl> dammsan: no, the problem isn't conference timing, the problem is late scheduling +09:45 < kbingham> If I'm going to 'lose' 3 weeks of work - then I need to know in advance - so I can find another three weeks of work to fill that time. Or Hugo losses his college fund! +09:45 < pinchartl> conferences and other events happen from time to time +09:45 < pinchartl> I can handle them by myself +09:45 < dammsan> kbingham: sure understood +09:45 < pinchartl> if you give me 3 weeks of work 6 weeks ahead of the due date +09:45 < pinchartl> I'll work around the conference that happens during those 6 weeks +09:45 < dammsan> makes sense +09:45 < pinchartl> if you give me 3 weeks of work 3 weeks ahead of the due date, that won't be possible +09:45 < geertu> pinchartl: Now, if you have 2x3 weeks of AT... +09:46 < pinchartl> and that's what is happening all the time +09:46 < pinchartl> in the best case we get the AT for 3 weeks of work 3 weeks ahead of the due date... +09:46 < dammsan> that seems to be because we don't do bre +09:46 < pinchartl> sometimes it's even later +09:46 < wsa_> kbingham: Hugo should be a rich kernel developer at college time, or? ;) +09:46 < dammsan> akout early enough +09:47 < kbingham> wsa_, I'm hoping he'll be a kernel developer at primary school :D +09:47 < pinchartl> dammsan: what do you mean by not breaking out early enough ? +09:47 < dammsan> well +09:47 < dammsan> for instance this virtualization topic +09:47 < dammsan> i wish we could meet and discuss how to proceed with that +09:47 < dammsan> now we have not met so much recently +09:47 < dammsan> and when we met in san sebastian the purpose seemed more to be for hacking than planning +09:48 < dammsan> i would like to meet with group leaders (+ others maybe) to do some planning +09:48 < pinchartl> I don't think we could have planned work for Q2 2018 in detail in San Sebastien +09:48 < pinchartl> there are various ways to meet, face to face or online +09:48 < kbingham> dammsan, I get that it was a communications issue - but in the Virt task - I was only told that *I* was supposed to write the task descriptoin - well after the point at which I would have expected to start the task - - hence for me - it's just impossible to fit in to a schedule in this current instance. +09:49 < dammsan> ok but i got the impression that you wanted to do the task? +09:49 < dammsan> i was pleasantly surprised that you signed up =) +09:49 < dammsan> obviously it was a too big jump +09:49 < pinchartl> dammsan: in general I think most of us are willing to help with work that is considered to be important by Renesas or you +09:49 < pinchartl> that shouldn't be a surprise :-) +09:49 < geertu> Exactly. +09:50 < dammsan> sure +09:50 < dammsan> i feel that +09:50 < dammsan> thanks for your help +09:50 < geertu> kbingham: You should have had some idea of what the task as about when volunteering, right? +09:50 < dammsan> but how to plan? +09:50 < dammsan> (and break out) +09:50 < kbingham> I knew very little about the task. +09:50 < dammsan> (that makes you the brave man in the room) =) +09:51 < kbingham> We are trying to hunt in the dark for tasks to fit into slots. +09:51 < kbingham> From my perspective - laurent said that dammsan wants us to look into virtualisatoin topics ... +09:51 < dammsan> sorry for being unclear +09:51 < kbingham> And thus I was expeting a virt topic to fill my available slots +09:51 < kbingham> However - I was expecting it to have been planned early - and be clear so that I could ramp up +09:52 < dammsan> but we had not done enough break out +09:52 < dammsan> also i feel we are moving slowly +09:52 < dammsan> probably due to lack of resources +09:52 < kbingham> I don't even mind if it would take me longer than the available time slots to do the investigation / self learning / and ramp up - but for that to happen - +09:52 < kbingham> the task *must start weeeeeellll in advance* +09:52 < dammsan> i would love to ask geertu to focus on virt activities +09:53 < dammsan> but without AT i feel i cannot request stuff +09:53 < kbingham> I see AT time slots as time slots that I reserve in my planning for to allow Renesas to fill with tasks. +09:53 < kbingham> But they can't be booked or not booked with zero notice. +09:53 < dammsan> kbingham: that view makes sense +09:54 < dammsan> but the default IMO is unbooked unless otherwise agreed +09:54 < horms> dammsan: when you say break out, do you mean you think it would help if there were more face-to-face meetings between say yourself and group leaders? +09:54 < dammsan> horms: yes +09:54 < dammsan> thank you +09:54 < horms> ok, I see a farily obvious way to see if that helps +09:54 < pinchartl> dammsan: unbooked unless otherwise agreed isn't something I can agree on as long as we book at the last minute +09:55 < pinchartl> if you want that, then I want to be booked 3 months in advance +09:55 < kbingham> So I should assume that 6/12 weeks of my time are not scheduled - and find another customer to fill that time ? +09:55 < pinchartl> as Kieran said in an e-mail, you don't expect to call a builder for your house and have him show up the next morning +09:55 < dammsan> sure, that fits perfectly fine with current budget situation =) +09:56 < kbingham> Ok - if we have budget restrictions - can we be clear about that!? +09:56 < pinchartl> there were no budget restrictions for Q2, were there ? +09:56 < dammsan> sorry if that has not communicated enough +09:56 < dammsan> but ATs are not guaranteed to be filled +09:56 < pinchartl> dammsan: it has been communicated badly, as in "beware of the big bad wolf" +09:57 < dammsan> never has been, and they are getting closer to be more sparse if budget is going bad +09:57 < kbingham> (beware, followed by - but don't go away) +09:57 < dammsan> well +09:57 < dammsan> it was actually my proposal to renesas side to tell you in advance +09:57 < dammsan> before you signed your base contracts +09:57 < dammsan> that budget is looking bad +09:57 < dammsan> so you should look into other options if you want to sustain +09:58 < neg> I think the budget situation have been communicated and also that there are possibilities that not all additional task slots will be used. Is this not why the base time was adjusted? +09:58 < dammsan> of course i want to work to with you +09:58 < dammsan> neg: yes +09:58 < dammsan> but i can't control our budget +09:58 < kbingham> Ok - but if AT's are not to be filled - they need to be filled with something else +09:58 < dammsan> especially if i can't make plans to show to renesas side +09:58 < kbingham> and tha'ts not possible - if the AT is filled at the last minute +09:59 < pinchartl> dammsan: why have additional tasks for Q2.2 not been scheduled in time, as there was no budget restriction for that period ? +09:59 < dammsan> kbingham: we seem to have different ideas how the contracts work +10:00 < kbingham> dammsan, I agree :) +10:00 < dammsan> what says that there was no budget restriction that period? +10:00 < dammsan> i was asked to reduce cost +10:00 < dammsan> i asked to reduce travel +10:00 < dammsan> did not to to EU +10:00 < dammsan> now if other members ask renesas directly +10:01 < dammsan> and they change their mind over a week +10:01 < dammsan> then it is not my problem +10:01 < kbingham> It's a shame you could not come to EU brussels - as it would have been good to do F2F then... :( +10:01 < dammsan> i agree +10:01 < pinchartl> dammsan: when we discussed additional tasks for Q2.2 there was no budget issue that was brought up, and tasks were agreed on to fill slots (but too late to be completed on time) +10:02 < dammsan> i have no intention to reduce the slots just for the sake of it +10:02 < dammsan> but we need to know what to do before doing it +10:02 < dammsan> and like my old fried paul said +10:02 < dammsan> you don't pay contractors to learn +10:02 < dammsan> you pay contractors to do what they are good at +10:03 < dammsan> of course some ramp up time is needed +10:03 < dammsan> and acceptable +10:03 < dammsan> but either you can do what you say +10:03 < dammsan> or you cant +10:03 < kbingham> dammsan, I actually agree on that - +10:03 < dammsan> and if we cant agree before the due date +10:03 < dammsan> too bad +10:03 < kbingham> That's why I expected the virt work to be planned well in advance :) +10:03 < dammsan> sure +10:04 < dammsan> so we had a some kerfuffle there +10:04 < dammsan> and in general when it comes to task break out and planning that may happen +10:04 < dammsan> probably my fault as usual +10:04 < dammsan> =( +10:04 < dammsan> but i stand by that we need to know before we do it +10:04 < pinchartl> it's not that it may happen, it happens all the time :-( that's for me a clear indication that the process is wrong +10:05 < dammsan> sure +10:05 < dammsan> but ive asked for plans forever +10:05 < dammsan> i would like to improve the situation +10:05 < pinchartl> I can't make a plan without being able to allocate resources +10:05 < dammsan> well +10:05 < pinchartl> I can't say what we'll do and when we'll do it if I'm not given a predictable amount of development time +10:05 < kbingham> I think we're well in to the territory of best discussed F2F :) +10:06 < kbingham> which is conveniently in 2 weeks or less. +10:06 < dammsan> i would like to separate assignment from planning +10:06 < dammsan> also time required from planning +10:06 < dammsan> only focus on break out and dependencies +10:06 < dammsan> i would like to discuss this with group leaders +10:06 < dammsan> ahead of time +10:07 < dammsan> then from the result of that activity feed AT and base work +10:07 < dammsan> if we manage to move forward with base only then we can skip AT +10:07 < dammsan> but dropping AT without any plans seems insane to me +10:08 < dammsan> kbingham: sorry that it became a mess +10:08 < pinchartl> so how do we move forward ? +10:08 < dammsan> how can we proceed? +10:09 < dammsan> i proposed my two ideas to improve the planning +10:09 < pinchartl> my requirement is that anything with a fixed due date has to be scheduled (as in contracts signed) at due date - max(2 * duration, duration + 2 weeks) +10:09 < dammsan> dates ahead of time + separate planning and breakout +10:09 < pinchartl> (the numbers can be fine-tuned +10:09 < dammsan> pinchartl: i want to agree as early as you +10:10 < dammsan> but we need to work together to break out stuff +10:10 < dammsan> and say that we are late +10:10 < dammsan> what happens then? +10:10 < dammsan> i don't want to be late +10:10 < dammsan> but if we end up there anyway? +10:10 < pinchartl> if you remember, I initially proposed a long list of tasks, and most got rejected with "this should be base time" or "this isn't a priority". I was then suppose to magically come up with new ides +10:11 < dammsan> also, crossi +10:11 < dammsan> ng quarterly boundaries is difficult +10:11 < dammsan> pinchartl: sorry if i made you feel rejected +10:11 < pinchartl> if we always end up being late we'll reach a point where stress and frustration will be so high that we'll lose developers completely. it won't just be a matter of some slots not being filled, people will go away +10:12 < dammsan> i understand. that's no good +10:12 < dammsan> but lets agree on a view with the contracts +10:12 < dammsan> no contract means no agreement +10:12 < dammsan> when it comes to assigned work +10:12 < pinchartl> if Renesas prefer stopping upstream development and only release BSPs, I won't stop them +10:13 < dammsan> well +10:13 < dammsan> if renesas can chose by themselves then we all have been reduced to up-porting machines +10:13 < dammsan> so no development on the horizon there no +10:13 < pinchartl> I'm sure they can find plenty of cheap developers who can write bad code +10:13 < dammsan> yep +10:13 < dammsan> we agree on that +10:13 < dammsan> so question is how to remedy that situation +10:14 < dammsan> without leaving people in a bad situation +10:14 < dammsan> also +10:14 < pinchartl> and if they prefer going that way, I have no issue making sure those bad patches won't make it to mainline +10:14 < dammsan> i don't want people to think they should be spoon fed with work +10:14 < dammsan> i want to share the burden of breaking out stuff with you +10:15 < dammsan> so we can collaborate to learn what next steps we need to take +10:15 * kbingham doesn't full understand what "breaking out stuff" means +10:15 < kbingham> Is that investigation and planning ? +10:15 < dammsan> kbingham: both perhaps +10:15 < pinchartl> by trying to move away from spoon-feeding and asking people to come up with task ideas for themselves, with such a fine-grained control, we have reached a situation that reminds me of kids in school who have to ask permission to go to the toilet +10:15 < dammsan> i asked for the dma virtualization topic +10:16 < dammsan> but in reality pinchartl found it to be 4 subtasks +10:16 < dammsan> unfortunately the ordering was odd +10:16 < pinchartl> dammsan: and it shouldn't have been my job to breakout a core task... +10:16 < dammsan> so we did the paper work before the so it happened after the contract was finalized =) +10:17 < dammsan> pinchartl: it seems that your role as coworked of kbingham might be conflicting with group leader +10:17 < dammsan> so i think we need to be more clear +10:18 < dammsan> kbingham: do you understand what i mean by breakout now? +10:18 < pinchartl> dammsan: I don't see a conflict. if you expect developers to work with group leaders to break out tasks I'm fine with that. what I disliked here is that Kieran and I were supposed to do the work all by ourselves +10:18 < kbingham> 'breakout = create additional task details' +10:18 < dammsan> kbingham: and split into subtasks +10:18 < kbingham> Yes - I was about to add that ;) +10:19 < dammsan> if we would have broken out ahead of time +10:19 < dammsan> then we could have assigned you something smaller instead +10:19 < dammsan> pinchartl: it was unclear about responsibility for sure +10:20 < dammsan> so how to proceed in the short term? +10:20 < dammsan> i proposed serial pass through only or also include clocks like you mentioned before +10:21 < pinchartl> today is June the 7th. the due date is the 15th. I'll let Kieran decide as this is his task, but personally I think it's too late to do anything +10:22 < dammsan> i have been told we may push back the due date until later this month +10:22 < dammsan> but you have your travels too +10:22 < kbingham> for available time remaining, I think perhaps dropping that to a 5 day task to cover the investigation already completed, and we can consider revisiting the task for next quarter. +10:23 < dammsan> kbingham: so serial pass through only? +10:23 < dammsan> which due date would you like? +10:23 < dammsan> is 15th ok or is later better? +10:23 < kbingham> I have got serial pass through working - so I don't mind the usual due date for that. +10:23 < dammsan> good! +10:24 < dammsan> can you email me your proposal how to change the text? +10:24 < kbingham> Sure I'll do that after the meeting. +10:24 < dammsan> perfect - thanks +10:24 < dammsan> i'll deal with this later today +10:25 < dammsan> so as final topic, how can we agree on the same view for the contracts +10:25 < dammsan> ? +10:25 < kbingham> dammsan, By sitting down with a beer. +10:25 < dammsan> alrighty =) +10:26 < dammsan> geertu: what do you think about agreeing on dates one quarter ahead of time? +10:26 < kbingham> :D +10:26 < geertu> dammsan: due dates and meeting dates? +10:26 < dammsan> wsa_: and what do you think about agreeing on dates one quarter ahead? +10:26 < dammsan> yeah +10:27 < geertu> What does that solve? We know the due dates, and can guess meeting dates (about every two weeks)? +10:27 < wsa_> for add. tasks? +10:27 < pinchartl> dammsan: haven't we agreed on due dates until the end of times already ? they're at Q+1.5 and Q+2.5. what else is there to agree on ? +10:27 < dammsan> chat meeting dates +10:27 < dammsan> so we know when we have to fix the AT titles + descriptions +10:27 < geertu> Do we have an issue with not knowing all future chat meeting dates? +10:28 < dammsan> we have an issue with AT creation which operates on chat meeting tick +10:28 < horms> dammsan: the chat meeting dates seem pretty predictable from my side +10:28 < horms> is the problem that sometimes we skip a week? +10:28 < pinchartl> dammsan: why don't we then decouple AT creation and chat meetings ? +10:28 < pinchartl> we can have ad-hoc meetings to create ATs +10:29 < dammsan> horms: sspinchartl: i agree if we can schedule break out meetings first =) +10:29 < dammsan> horms: no problem i think +10:29 < horms> sorry, the breakout term is very confusing to me +10:30 < horms> do you mean meetings to break out AT tasks? +10:30 < dammsan> tasks in general +10:30 < dammsan> that may or may not turn into AT +10:30 < horms> Ok, so task discussion meetings, separate from our regular meetings? +10:30 < geertu> Don't we usually handle AT titles over email? Decoupled from chat meeting? +10:30 < dammsan> if we have a bunch of tasks already broken out +10:30 < kbingham> https://paste.ubuntu.com/p/JcT9DrMQkT/ :D +10:31 < dammsan> then those should be easier to assign +10:31 < dammsan> horms: these discussions take time +10:31 < dammsan> and usually they get lower priority than actual hacking +10:32 < dammsan> and with travels and conferences and whatnot +10:32 < dammsan> it seems that we so far have not been able to come up with any plan (visible to me) +10:32 < pinchartl> dammsan: what kind of plan to you expect ? +10:32 < dammsan> a good example is the dma virtualization topic +10:33 < dammsan> you broke it out into 4 subtasks +10:33 < pinchartl> as I said, I can't create a plan with dates as I don't have development time I can assign to tasks +10:33 < dammsan> no need for dates +10:33 < dammsan> or time estimates +10:33 < dammsan> only titles and dependencies +10:33 < pinchartl> so if there's no date or time estimate, what kind of plan do you expect ? +10:33 < dammsan> maybe some details if possible +10:33 < pinchartl> a list of tasks ? +10:33 < dammsan> yes +10:33 < dammsan> with dependencies +10:33 < dammsan> for instance, clocks are missing from QEMU +10:33 < dammsan> so lets include a prototype task for that +10:34 < dammsan> before we do real development +10:34 < pinchartl> to find that out, nearly 5 days of work were needed. how could we have found out in a better way ? +10:34 < dammsan> we could have done it at aa different time IMO +10:34 < dammsan> maybe we should have aimed at serial pass through instead of going directly to DMA virt +10:35 < pinchartl> yes, but how ? what should have triggered that investigaion ? +10:35 < dammsan> serial pass through prototype? +10:35 < pinchartl> ok +10:35 < pinchartl> is it acceptable, when we schedule a prototype AT, to not deliver the expected prototype ? +10:35 < dammsan> or us sitting down and looking of +10:35 < dammsan> at what is needed? +10:35 < pinchartl> when features are missing for instance +10:35 < pinchartl> sitting down for 5 days ? :-) +10:35 < dammsan> we need deliverables +10:35 < pinchartl> it really required work +10:36 < dammsan> sure no doubt about that +10:36 < dammsan> (that's why i don't want to do it myself) =) +10:36 < dammsan> anyway +10:36 < dammsan> we can set aside base time +10:36 < pinchartl> in my opinion, there's no guarantee that a prototype can be delivered, ever +10:36 < dammsan> and meet and discuss these things over some time +10:36 < dammsan> then we are aiming too high +10:36 < pinchartl> by definition we create prototypes when the risk is high +10:37 < dammsan> so far all prototypes have been delivered =) +10:37 < pinchartl> and so we can run into blocking issues that can't be solved within the scheduled time frame +10:37 < dammsan> well, you take shortcuts +10:37 < pinchartl> so far we're had a combination of luck and dipping in base time. I don't think that's something we want to rely on as a rule +10:37 < dammsan> and if you cant then you have aimed too high +10:38 < dammsan> i'm open to suggestions though +10:38 < pinchartl> you can't know for sure how to aim, that's the whole point of prototypes +10:38 < dammsan> if you have any other ideas how to figure it out +10:38 < pinchartl> and taking shortcut can be acceptable to some extent +10:38 < pinchartl> but if we end up rushing to deliver something by taking too many shortcuts, then we delivere useless crap +10:38 < pinchartl> and waste time and money +10:38 < dammsan> there is a balance for sure +10:38 < pinchartl> in this specific case +10:39 < pinchartl> you could rush to create lots of hacks that will allow serial pass-through +10:39 < pinchartl> or start working on the missing features and not deliver the pass-through in time +10:39 < pinchartl> in the first came the patches will be garbage, nobody will ever use time +10:39 < kbingham> If a task 'suggestion' has begun - Investigtiaons and breakout - can simply occur in base time - in the quarter before>? +10:39 < dammsan> yeah? +10:39 < pinchartl> while in the second case the time is invested in a more productive and useful way +10:40 < dammsan> pinchartl: except the second does not consider the output of broken out tasks as useful +10:40 < pinchartl> dammsan: sorry, I didn't get that +10:40 < dammsan> kbingham: i agree +10:40 < dammsan> no sane person will take a prototype and put it into production +10:41 < kbingham> dammsan, Great - so we just need to get task suggestions started early. +10:41 < dammsan> but we can learn various things from it +10:41 < pinchartl> it's not about putting things into production +10:41 < dammsan> of course the ultimate goal is to solve the problem +10:41 < dammsan> but we need to learn what the problems are too +10:41 < dammsan> to know what we need to do +10:42 < dammsan> is it any more clear? +10:42 < pinchartl> if you spend 5 days learning about the issues and finding out what has to be solved, and have 5 days of budget left, consuming the remaining 5 days to write throw-away hacks just to tick the "done" box in the report is useless +10:42 < pinchartl> prototypes are valuable to find out about issues +10:43 < pinchartl> but once the issues are known it's pointless to create the dirtiest hacks just to check a box +10:43 < dammsan> if the prototypes happened by themselves then i would be the happiest man on the planet =) +10:43 < pinchartl> some hacks can be useful when they allow moving to the next step and finding about the next problem +10:43 < dammsan> the question is how the issues become known =) +10:43 < geertu> The hacks are useful to unblock you, and continue getting it to work, perhaps finding more issues to solve later. +10:44 < dammsan> geertu: yes +10:44 < pinchartl> or when they make a dependency for another task available faster than the proper solution +10:44 < geertu> Ineed, you won't know about the issues if you don't try. +10:44 < pinchartl> but if nothing else depends on your prototype, more hacks after all problems are known become throw-away code +10:44 < dammsan> i believe prototypes should be quick hacks +10:45 < dammsan> and i think my request to you was over reaching =) +10:45 < pinchartl> I think it's time to wrap this discussion up +10:45 < pinchartl> can we schedule a face to face discussion in Tokyo ? +10:45 < dammsan> sure, but how to make non-present members not feeling omitted? +10:46 < pinchartl> geertu: I think that question is for you +10:46 * geertu is wondering about this recent invention called "tubes", or "the Internet" +10:46 < dammsan> this travel planning is a cluster flower +10:47 < dammsan> geertu: i need to read the friendly manual =) +10:47 < pinchartl> dammsan: we have been told that there's no travel budget, and now we find out a face to face meeting is needed. there's an unavoidable conflict between those two :-) +10:47 < dammsan> i know - welcome to my life +10:47 < dammsan> i don't recommend it +10:48 < pinchartl> dammsan: I have never doubted that I'm not the only person to be frustrated by this :-) +10:48 < geertu> pinchartl: fall-out from the lack of travel budget around FOSDEM? +10:48 < pinchartl> so how can we schedule a meeting ? +10:48 < horms> dammsan: I recommend you have a f2f meeting. If its possible let people know so they can post questions via chat or etherpad. But likely it won't be due to times zones, being in a bar, ... So pleasae provide some sort of summary so a follow-up discussion can occur via chat or email +10:49 < horms> s/please/I recommend/ +10:49 < dammsan> horms: sounds good +10:49 < dammsan> thanks +10:49 < dammsan> pinchartl: when are you guys available? +10:50 < pinchartl> for me, Wednesday, Thursday, Saturday, Sunday, +10:50 < kbingham> My talk is on the Thursday. +10:51 < kbingham> but it might be scheduled early - so after would be fine. +10:51 < pinchartl> my talk is on Friday at 15:15, so starting at 16:00 Friday is fine too +10:51 < dammsan> i have some plans sat + sun +10:51 < wsa_> not the weekend, please +10:52 < pinchartl> how about Wednesday then ? +10:52 < wsa_> how about we get some weapons at the NinjaRestaurant and fight it out? ;) +10:52 < neg> All OSS-J days and the weekend works for me +10:52 < dammsan> sure +10:52 < neg> wsa_: I like your style +10:52 < dammsan> 20th, right? +10:53 < wsa_> wednesday morning? +10:53 < wsa_> or did someone here want to go to the keynotes? ;) +10:53 < dammsan> in a ditch in east shinjuku =) +10:53 < pinchartl> wednesday morning sounds good, there are only keynots +10:53 < kbingham> Wednesday morning works :) +10:53 < neg> +1 +10:53 < pinchartl> 9:00 on Monday ? +10:54 < pinchartl> we might be able to host the meeting at our apartment +10:54 < pinchartl> who wants to join ? +10:54 < wsa_> monday? +10:54 < pinchartl> sorry +10:54 < pinchartl> Wednesday +10:54 < geertu> If you want people in EU to join, morning meetings may not be the right time... +10:55 < wsa_> wed 09:00 is fine with me +10:55 < wsa_> and if your apartment turns out to be not big enough, we will surely find a place at the conference +10:55 < wsa_> with only keynotes going on +10:56 < pinchartl> geertu: attending remotely would require a properly A/V equipped room. I don't know if we could get that +10:56 -!- shimoda [~shimoda@relprex2.renesas.com] has quit Read error: Connection reset by peer +10:56 < dammsan> morimoto-san seems to suggest 18 or 19 =) +10:57 < dammsan> but from my side +10:57 < dammsan> if we have this kind of business meeting then reneeesas should pay for airfares IMO +10:57 < pinchartl> dammsan: I'll fly in on the 18th, and on the 19th there's a whole day V4L2 meeting that I need to attend to make sure the right design decisions will be taken +10:57 < dammsan> pinchart: gotcha +10:58 < pinchartl> otherwise I would have proposed Tuesday +10:58 -!- shimoda [~shimoda@relprex2.renesas.com] has joined #periperi +10:58 < pinchartl> we might be able to to Tuesday evening, but it's probably better to start fresh +10:58 < wsa_> yes +10:59 < morimoto> all: I'm OK for Web +10:59 < pinchartl> I agree with Simon, I think we should start proposing ideas over e-mail first, so that everybody can express themselves +10:59 < dammsan> for anyone in town the weekend before is also possible from my side +10:59 < pinchartl> send the minutes of the discussions after the meeting +10:59 < pinchartl> and then wrap up with an online discussion for the final agreement so that nobody is left out +10:59 < wsa_> then again, being tired in a bar might also help to keep it short :D +11:00 < dammsan> i think we need several opportunities to vent =) +11:00 < pinchartl> wsa_: I want short and productive, but I'd rather have long and useful than short and useless :-) +11:00 < dammsan> so the more the better +11:00 < dammsan> pinchartl: is this on of those "size matters" discussions? +11:01 < pinchartl> so can we wrap up now ? +11:02 < dammsan> wednesday it is then? +11:02 < pinchartl> yes +11:02 < wsa_> booked +11:03 < neg> 09:00 what location? +11:03 < pinchartl> neg: possibly our apartment, or if it's too small, somewhere else +11:04 < pinchartl> geertu: I think the mic is now yours for the core meeting |