2022-05-08.log

cinap_lenrekhttps://felloff.net/usr/cinap_lenrek/9reform-catclock.png00:55
- S0rin (QUIT: Ping timeout: 256 seconds) (~S0rin@user/s0rin)01:25
minutecinap_lenrek: wow, nice! did you get a display to work?01:39
cinap_lenreknope.01:40
minuteha01:40
cinap_lenrekthis is remote, i got the ethernet driver to work :)01:40
cinap_lenrekthe display will be the very last thing to work on01:40
minutethat's cool though!01:41
cinap_lenrekin fact, i'd rather have a screen turn on in uboot01:41
cinap_lenrekwhich already has the graphics driver afaik, just not the bits to initialize the mipidsi bridge01:42
cinap_lenrekthay'd also be more usefull in the field, when something's wrong and theres no second computer with serial at hand01:43
+ S0rin (~S0rin@user/s0rin)01:48
- chomwitt (QUIT: Ping timeout: 248 seconds) (~chomwitt@2a02:587:dc19:d600:88e2:8613:1b70:14d7)01:51
- mtm (QUIT: Ping timeout: 260 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)02:02
- Christoph_ (QUIT: Remote host closed the connection) (~Christoph@p4fe733a7.dip0.t-ipconnect.de)02:13
- vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:20)03:01
littlebobeepminute: Did you ever see my questions about the video output on the LS1028A ?03:21
littlebobeepcinap_lenrek: What OS is that?03:21
cinap_lenrekplan9 / 9front :)03:23
ex-parrotcinap_lenrek: amazing work03:43
ex-parrotvarious people have tried getting the display working in u-boot, if you sorted that it would unblock a bunch of other projects03:43
littlebobeepex-parrot: Which SOC is the display broken in u-boot?03:47
ex-parrotthe standard imx8 one03:47
ex-parrotI think if we could get u-boot display working it would make getting OpenBSD going a lot quicker too03:51
slalso appreciated03:52
cinap_lenrekyep.03:52
cinap_lenrekany os really.03:52
- Guest6130 (QUIT: Ping timeout: 276 seconds) (~nicolas@83-49-142-46.pool.kielnet.net)03:52
ex-parrotoh you're back03:52
ex-parrotI'll stop trying to mention you in two channels at once :)03:52
littlebobeepHmmm I am more interested in the LS1028A because it can have 16GB of RAM on the SOM and no proprietary blobs to boot.... I am just wanting to know about the blobs required for video out03:53
+ nsc (~nicolas@85-49-142-46.pool.kielnet.net)03:53
ex-parrotit will be nice to have a beefier option03:54
ex-parrotfor me the imx8 is quite sufficient tho03:54
ex-parrotesp with 9front it should be nice03:54
littlebobeepAlso I am curious about the CM4-compatible options how do these cards fit in the motherboard when it normally uses SODIMM slot for the SOM cards?03:54
* nsc -> Guest693403:54
ex-parrotthere's an adapter board03:55
cinap_lenrekfuck raspbarrypis03:55
ex-parrotI've got a RISC-V based CM4 compatible module on order (supposedly)03:55
ex-parrotwould be fun to have a very slow RISC-V Reform03:55
cinap_lenrekat least nxp has some decent documentation03:55
littlebobeepex-parrot: Can you share a link or name of the module so I can research?03:59
ex-parrotsure, I'm not sure how much official info there is but there have been some pictures released03:59
ex-parrothttps://www.mntmn.com/media/reform_md/2022-03-03-reform-march-update.html <-- was mentioned here04:00
littlebobeepex-parrot: also the blog post on mntre.com has no links to the CM4 adapter board where can i look at this item?04:00
ex-parrothttps://www.mntmn.com/media/reform_v2_images/20220303-mnt-rcm4-render-with-pi.png04:00
littlebobeepex-parrot: that blog post only talks aboutFPGA04:00
ex-parrot"The most immediate, least expensive solution will be in the form of the RCM4 Raspberry Pi CM4 adapter."04:01
littlebobeepex-parrot: You ordered an FPGA or an actual RISC-V SOC ASIC?04:01
ex-parrotthe latter, from ClockworkPi04:01
littlebobeepex-parrot: great, what is the model?04:02
ex-parrotI'll find a link hang on04:02
littlebobeepthnx04:02
littlebobeepcinap_lenrek: Yeah I don't like RPis either but can't you just have a CM4 module that has nothing to do with RPis?04:03
ex-parrothttps://mastodon.social/@mntmn/107837330047564462 <-- slightly more info on the CM4 04:03
ex-parrotok sorry I was incorrect, the RISC-V module I ordered is a Pi CM3 form factor not CM404:04
ex-parrotbut I imagine an adapter could still be made for the Reform04:04
ex-parrothttps://www.clockworkpi.com/product-page/copy-of-clockworkpi-core-r-01 <-- this is the module I have on order04:04
littlebobeepex-parrot: thanks for mastodon link, I needed that pic on the left to understand... so on the left is the SODIMM board that just has the CM4 hat on top connecting with those two long plastic black contacts?04:06
ex-parrotyep thats the idea04:06
littlebobeepwhat are the two tan connectors on the far right edge of the board and the small metal connector on the bottom?04:06
ex-parrotI'd guess the bottom-most one is USB, not sure about the other two04:06
ex-parrotthe imx8 board has the wider flat flex cable as well04:07
ex-parrotI forget what is carried over it04:07
littlebobeepex-parrot: that clockworkpi is not SM4 or SM3, it does SODIMM.... does this plug into Reform motherboard and magically work?04:07
+ mtm (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)04:08
ex-parrotno, I thought it was a Pi3 Compute Module SIMM04:08
ex-parrotI could be wrong, I just kind of bought it on a whim04:08
ex-parrotthe Pi compute module changed from the SIMM form factor to the board to board interconnects between 3 and 404:08
littlebobeepAhhh well what can you plug it into when it arrives?  The only thing I ever used SODIMM for was DDR RAM modules :P04:09
ex-parrotthe Reform's motherboard slot is VERY DEFINITELY not CM3 compatible04:09
ex-parrotso no, I won't be plugging it in there :P04:09
ex-parrotthe interface is specific to the reform04:09
ex-parrotso to use the RISC-V board with the reform you'd need a similar adapter to the CM4 one04:09
littlebobeepOhhhh so you are saying CM3 is basically physically compatible to SODIMM (but maybe not logically)?04:09
ex-parrotright it's the same connector but totally different signals / pinout04:10
ex-parrotwould likely damage the reform and the module if plugged in to the reform without an adapter board04:10
littlebobeepWeird why choose SODIMM connector then that is confusing04:11
ex-parrotwell you could ask why did the Pi Foundation use the SODIMM connector for the CM3 when it was already used for RAM...04:12
ex-parrotit's just a suitable connector that is available, has enough pins and can carry the appropriate signals04:13
littlebobeepHmmm someone on the forum says the MNT Reform only lasts ~3 hours on full battery charge?  Is this true?05:54
ex-parrotthat sounds wrong06:28
ex-parrotdepends what you're using it for but I would say I get more like 5-6 hours06:29
ex-parrotnot amazing but not terrible06:29
littlebobeepAhhh okay just random forum user06:29
ex-parrotyou could probably kill it in 3 if you were compiling constanrtly and had the screen on high brightness etc06:29
ex-parrotthere are higher capacity batteries available now too, not sure how much that affects the runtime06:30
littlebobeepOhhh how many mAh are the cells compared to the original/default ones?06:32
josch_littlebobeep: 2000 mAh https://www.eremit.de/p/18650-3-2v-2-000mah-lifepo406:34
* josch_ -> josch06:34
ex-parrot^ those were the ones06:34
ex-parrotthere was talk of changing the batteries included with theu nit but I don't know if that has happened06:34
joschBoostisbetter said that those last about an hour longer than the default ones.06:41
littlebobeepjosch: thanks!06:45
ex-parroti switched from the included cells that I busted to some generic-ish ones from aliexpress06:46
ex-parrotand the battery life seems about the same06:46
ex-parrotwhich is good06:46
littlebobeepex-parrot: Were thye aliexpress ones different mAh?06:47
ex-parrotno, advertised as 1800maH06:47
ex-parrotmAh06:47
littlebobeepex-parrot: What is the motivation for buying a whole new set of batteries in this case?06:47
ex-parrotI overdischarged my original ones, and while they kept working for a while one cell suddenly died06:48
ex-parrotso I decided it was safer to replace them all with a new set06:48
ex-parrotthe overdischarge problem is improved with newer LPC firmware, and should be entirely fixed with the next revision battery boards06:48
littlebobeepLPC is a family of 32-bit microcontroller integrated circuits by NXP Semiconductors (formerly Philips Semiconductors06:54
ex-parrotjosch: am I right in thinking the Debian experimental 5.17 firmware is almost ready to run on the Reform minus the LCD?06:54
littlebobeepHad to look that up.... is a LPC armhf chip used for PMU?06:54
joschex-parrot: you mean the Debian 5.17 kernel?06:55
ex-parrotlittlebobeep: the "platform controller" if you like in the Reform is an NXP LPC chip06:55
ex-parrotjosch: err yeah, why did I write firmware06:55
littlebobeepex-parrot: What Debian firmware?06:55
ex-parrots/firmware/kernel/06:55
littlebobeepAh okay sorry I can look this up, I guess the platform controller is used for PMU06:55
ex-parrotyes06:55
joschex-parrot: well, it will run and give you a serial console (according to vagrantc) but it will not enable the internal display yet because those patches haven't been accepted upstream06:56
ex-parrotright. that's what I was thinking, thanks for confirming06:56
joschwe can carry unreleased patches in Debian but for that those patches should at least be accepted in some upstream branch06:56
ex-parrotyeah makes sense06:56
ex-parrotI imagine it's only a matter of time before they get accepted upstream anyway06:56
ex-parrotI thnk we never got there with the Novena but that platform was kind of a dead end anyway :(06:57
ex-parrotlittlebobeep: the keyboard and the LPC both are involved in the power management06:57
ex-parrotI think on the Reform specifically it's an lpc11u2406:58
littlebobeephttps://source.mnt.re/reform/reform/-/tree/master/reform2-lpc-fw       06:58
ex-parrotthat's the one06:58
littlebobeepam searching yesss06:58
ex-parrotcurrently the higher level power management is done through the keyboard though06:58
littlebobeepI do not understand what a keyboard has to do with a PMU06:59
ex-parrote.g. the keyboard tells the LPC to power on the machine07:00
joschex-parrot: well, currently lynxeye (lucas stach) is working on upstreaming those but but explained to me how their free time is limited right now, so upstreaming might still take a while07:00
ex-parrotbecause the power button is on the keyboard07:00
ex-parrotjosch: understandable07:00
ex-parrotand I believe currently the "power off" is done through a HID command sent to the keyboard, which then instructs the LPC to cut power07:01
ex-parrotbut I could be remembering that incorrectly. I still want to do a "proper" kernel driver for that when I have time / inclination07:01
littlebobeepjosch: if this is about display during bootloader is this patches bering merged into upstream u-boot you are talking about?07:01
joschlittlebobeep: no, i was talking about the kernel07:01
ex-parrotlittlebobeep: the u-boot display patches are a separate issue, we just straight up don't have that working yet07:01
littlebobeepex-parrot: really, just because the power button is there, the majority of the PMU logic is on the keybaord controller??07:01
joschwe also don't have upstream u-boot working07:01
ex-parrotlittlebobeep: I believe at the moment the i2c comms between the SoM and the LPC are not working07:02
ex-parrottho there is prob more subtlety to this than I am aware of07:02
ex-parrotthe keyboard has to be informed anyway so it can display the power state etc, so it's not entirely unreasonable to route power control requests through there at the moment07:03
littlebobeepex-parrot: what effects does broken i2c have?07:03
ex-parrotsorry it was SPI07:04
ex-parrotso the LPC is connected to the SoM over SPI, but it's not currently working07:04
ex-parrotwhich means that queries / commands to / from the LPC have to go through the keyboard07:04
ex-parrot(someone correct me if I'm wrong here, it was months and months ago we discussed this last)07:04
littlebobeepThe LPC is on the motherboard right?  So the SoM can speak SPI via the SODIMM connector to the LPC?07:04
ex-parrotright, except it's not currently working07:05
ex-parrotfor reasons unknown07:05
ex-parrotit would be nice to get it fixed at some point for tighter integration of power stuff07:05
littlebobeepHmmmm well the code is there, so....07:05
ex-parrotthe problem was that no SPI activity was observable across the bus07:05
ex-parrotso it's not clear if it's a software issue, configuration issue, electrical issue I think07:06
ex-parrotI'm not sure anyone has put a lot of time in to it yet07:06
littlebobeepI see07:06
littlebobeepWell I am not familiar with SoMs or SODIMM being used for this07:07
ex-parrotjust forget about the connector being a SODIMM, the connector isn't important :)07:07
littlebobeepahhh well07:07
littlebobeepmaybe07:07
ex-parrotSoM is the System-on-Module used, it's sort of a step up from an SoC in terms of what peripherals etc are included in the "package"07:08
littlebobeepex-parrot: what else besides LPDDR4 chips?07:38
- sl (QUIT: Remote host closed the connection) (~sl@beastie.sdf.org)07:48
- littlebobeep (QUIT: Ping timeout: 240 seconds) (~alMalsamo@gateway/tor-sasl/almalsamo)08:23
+ chomwitt (~chomwitt@2a02:587:dc19:d600:24ae:3716:1088:6d4d)08:29
- sigrid (QUIT: Ping timeout: 248 seconds) (~ftrvxmtrx@ftrv.se)09:41
+ sigrid (~ftrvxmtrx@ftrv.se)09:48
+ mtm- (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)11:54
- mtm (QUIT: Ping timeout: 276 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)11:56
- Ar|stote|is (QUIT: Quit: https://quassel-irc.org - Chat comfortably. Anywhere.) (~linx@149-210-16-210.mobile.nym.cosmote.net)12:17
+ Ar|stote|is (~linx@149-210-16-210.mobile.nym.cosmote.net)12:17
+ littlebobeep (~alMalsamo@gateway/tor-sasl/almalsamo)12:40
littlebobeep.12:40
+ MajorBiscuit (~MajorBisc@c-001-007-017.client.tudelft.eduvpn.nl)13:00
+ Christoph_ (~Christoph@p4fe73fde.dip0.t-ipconnect.de)13:31
- littlebobeep (QUIT: Ping timeout: 240 seconds) (~alMalsamo@gateway/tor-sasl/almalsamo)13:58
- mtm- (QUIT: Ping timeout: 260 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)14:02
- S0rin (QUIT: Ping timeout: 276 seconds) (~S0rin@user/s0rin)14:36
+ mtm (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)16:08
+ S0rin (~S0rin@user/s0rin)16:27
- S0rin (QUIT: Ping timeout: 248 seconds) (~S0rin@user/s0rin)16:51
+ sl (~sl@beastie.sdf.org)16:55
- Ar|stote|is (QUIT: Quit: https://quassel-irc.org - Chat comfortably. Anywhere.) (~linx@149-210-16-210.mobile.nym.cosmote.net)18:08
+ Ar|stote|is (~linx@149-210-16-210.mobile.nym.cosmote.net)18:15
- mtm (QUIT: Ping timeout: 248 seconds) (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)18:21
CliffyAin the system image v3 thread, bnys mentions "GPU glitches/freezes can occasionally happen (due to a mesa/new kernel issue)."18:27
CliffyAi was wondering if that's what's causing the flicker in Xwayland or if that's a different issue and i should create a report?18:28
CliffyAA quick repro description for anyone wondering: Start reform-windowmanager. Then start xterm. Type some text in then start moving the mouse around over the screen. The new text will flicker like there are two video buffers and only one was updated.18:32
CliffyAAnd I forgot to note that this does not happen with system image v2. I'll check back tomorrow thanks!18:42
cinap_lenrekhttps://git.9front.org/plan9front/plan9front/fff070f2cbb01b7c0879e9dcb13ee4e3ed2497f0/commit.html19:04
+ mtm (~mtm@c-73-27-62-116.hsd1.fl.comcast.net)20:27
+ S0rin (~S0rin@user/s0rin)20:48
- MajorBiscuit (QUIT: Quit: WeeChat 3.4) (~MajorBisc@c-001-007-017.client.tudelft.eduvpn.nl)22:08
cinap_lenrekaaaand we have multiprocessor up!22:27
cinap_lenrekhttps://felloff.net/usr/cinap_lenrek/9reform-multicore.png22:30
cinap_lenrekhttps://felloff.net/usr/cinap_lenrek/9reform-multicore2.png22:31
cinap_lenrektime for the beers :)22:33
minutecinap_lenrek: cheers!!22:33
+ littlebobeep (~alMalsamo@gateway/tor-sasl/almalsamo)22:42
minutecinap_lenrek: git.9front.org appears to be down now, though22:45
cinap_lenrekhaha22:45
cinap_lenrekok22:45
cinap_lenrekaand its back22:54
cinap_lenrekran out of processes :)22:55
minutenice22:59
- Christoph_ (QUIT: Remote host closed the connection) (~Christoph@p4fe73fde.dip0.t-ipconnect.de)23:59

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!