| - oomono (QUIT: Quit: Connection closed for inactivity) (uid328183@gateway/web/irccloud.com/x-xmuohfrqlsqtswow) | 00:09 | |
| mntmn | swivel: this is a misunderstanding. | 00:46 |
|---|---|---|
| mntmn | swivel: etnaviv is not a display device, only a renderer. | 00:46 |
| mntmn | swivel: so your modesetting display is dcss or mxsfb, and etnaviv renders to it. | 00:46 |
| mntmn | swivel: i have long given up on X (without wayland) so i cannot say how to make it work nowadays | 00:47 |
| swivel | mntmn: in your tech note spost, it explicitly mentioned etnaviv by name, but that's not actually an etnaviv on i.mx6 is it? putting "etnaviv" in the X config for Identifier is just a key for the config, not an actual driver anything right? | 00:49 |
| ex-parrot | mntmn: shirt arrived, is awesome \m/ | 00:55 |
| - Neelfyn (QUIT: Quit: Connection closed for inactivity) (uid180106@gateway/web/irccloud.com/x-ucptjhmsjhdanimw) | 05:23 | |
| - stepan (QUIT: Read error: Connection reset by peer) (sid147049@gateway/web/irccloud.com/x-ptpoibrwjerqziys) | 05:58 | |
| + stepan (sid147049@gateway/web/irccloud.com/x-esewxyteznyacmvu) | 05:58 | |
| + futarisIRCcloud (uid222239@gateway/web/irccloud.com/x-zuivisixaftfmnlx) | 06:19 | |
| - B[] (QUIT: Ping timeout: 240 seconds) (~Thunderbi@219-89-228-78-adsl.sparkbb.co.nz) | 07:22 | |
| - futarisIRCcloud (QUIT: Quit: Connection closed for inactivity) (uid222239@gateway/web/irccloud.com/x-zuivisixaftfmnlx) | 08:28 | |
| + oomono (uid328183@gateway/web/irccloud.com/x-lutihhkalrygojec) | 09:39 | |
| paulk-leonov | hi | 10:45 |
| paulk-leonov | were there any developments regarding blobs on the imx8? | 10:45 |
| paulk-leonov | especially DDR PHY and HDMI/DP | 10:45 |
| + rasmus-mk (~rasmus@c83-253-223-217.bredband.comhem.se) | 12:54 | |
| * debacle -> demoveu | 13:01 | |
| * demoveu -> debacle | 13:36 | |
| + Neelfyn (uid180106@gateway/web/irccloud.com/x-pvukjbjkbggrtkjz) | 14:40 | |
| - erlehmann (QUIT: Ping timeout: 240 seconds) (~erlehmann@dynamic-046-114-035-154.46.114.pool.telefonica.de) | 14:53 | |
| + erlehmann (~erlehmann@dynamic-046-114-035-154.46.114.pool.telefonica.de) | 14:55 | |
| + futarisIRCcloud (uid222239@gateway/web/irccloud.com/x-yxwfbuptjtfxzoed) | 17:13 | |
| - erlehmann (QUIT: Read error: Connection reset by peer) (~erlehmann@dynamic-046-114-035-154.46.114.pool.telefonica.de) | 17:20 | |
| + erlehmann (~erlehmann@dynamic-046-114-035-154.46.114.pool.telefonica.de) | 17:20 | |
| - DanC (QUIT: Quit: ZNC 1.6.6+deb1ubuntu0.2 - http://znc.in) (~DanC@2605:a601:ac60:1000:dc72:f722:2968:d479) | 17:36 | |
| + DanC (~DanC@2605:a601:ac60:1000:481f:e274:1264:569b) | 17:37 | |
| - oomono (QUIT: Quit: Connection closed for inactivity) (uid328183@gateway/web/irccloud.com/x-lutihhkalrygojec) | 17:39 | |
| jackhill | paulk-leonov: I'm just a random person, but https://www.devever.net/~hl/imx8 seems to suggest that there is no hope for HDMI | 17:48 |
| + oomono (uid328183@gateway/web/irccloud.com/x-gcbbfxxkyfwroemg) | 17:49 | |
| * debacle -> demoveu | 18:17 | |
| paulk-leonov | jackhill: thanks for the link! | 18:34 |
| paulk-leonov | looks like it would need a pci-e display controller | 18:34 |
| - oomono (QUIT: Read error: Connection reset by peer) (uid328183@gateway/web/irccloud.com/x-gcbbfxxkyfwroemg) | 19:01 | |
| + oomono (uid328183@gateway/web/irccloud.com/x-teagdqzowwisxlkg) | 19:02 | |
| * demoveu -> debacle | 19:27 | |
| + B[] (~Thunderbi@219-89-228-78-adsl.sparkbb.co.nz) | 19:59 | |
| * debacle -> demoveu | 21:21 | |
| mntmn | swivel: correct, this "etnaviv" is just a nickname there | 21:28 |
| mntmn | paulk-leonov, jackhill, the DDR4 could be REd in theory because it is not signed. the hdmi is signed unfortunately, but i am working on a USB3 display adapter dongle thing | 21:29 |
| mntmn | OTOH you probably disagree but i consider these blobs harmless, as they don't run on the ARM and have no bus access | 21:30 |
| mntmn | (and the DDR one only runs at boot and then never again) | 21:30 |
| mntmn | they are not nice of course, but not a big difference to closed-source silicon | 21:30 |
| mntmn | which is most of the silicon | 21:30 |
| mntmn | i would not tolerate a software blob running on the OS/in the ARM though | 21:31 |
| paulk-leonov | yeah it's not so much about security impact, rather about hiding out knowledge and exclusivity of modification ability | 21:31 |
| mntmn | yap. the DDR4 thing is def. hackable, i played around with it. you can tweak the DDRC registers also manually without the blob | 21:32 |
| mntmn | but you would have to write a new calibration routine | 21:32 |
| paulk-leonov | I see | 21:32 |
| mntmn | i don't have the experience+knowledge to do that though | 21:32 |
| mntmn | the blob is a piece of code that runs in the DDR to do a part of the calibration | 21:33 |
| paulk-leonov | also silicon is a different technological nature, it's not possible to modify it | 21:33 |
| mntmn | slash training | 21:33 |
| paulk-leonov | yep | 21:33 |
| paulk-leonov | isn't it possible to just run it once and then reuse whatever it produced? | 21:33 |
| mntmn | yes. to modify the HDMI thing you would need to crack the key | 21:33 |
| mntmn | paulk-leonov: yes, i had the same idea | 21:33 |
| mntmn | i couldn't get it to work but i think it should be doable | 21:34 |
| paulk-leonov | because if it's soldered DRAM chips I suspect the calibrated result won't change much | 21:34 |
| mntmn | given time + resources which i didn't have (i spent parts of a week on it) | 21:34 |
| paulk-leonov | except maybe with temperature | 21:34 |
| mntmn | correct | 21:34 |
| mntmn | not even with temp | 21:34 |
| paulk-leonov | thanks for trying! | 21:34 |
| mntmn | the temp stuff is auto-adjusted in silicon | 21:34 |
| paulk-leonov | neat | 21:34 |
| mntmn | if i understand it correctly the training is all about the PCB/trace situation | 21:35 |
| paulk-leonov | but well, if the design is that the firmware should run, there may also be unreasonable blockers built-in | 21:35 |
| mntmn | correct, for example there could be some register write that you cannot do from the outside | 21:35 |
| paulk-leonov | like waiting for the firmware to flip a bit saying it's done | 21:35 |
| mntmn | yep | 21:35 |
| paulk-leonov | exactly | 21:35 |
| mntmn | but we don't know that (yet) | 21:35 |
| paulk-leonov | indeed | 21:35 |
| paulk-leonov | so still worth a try | 21:35 |
| mntmn | so, it is still an open topic | 21:35 |
| mntmn | yep | 21:35 |
| paulk-leonov | do you know if the firmware's ISA is public? | 21:36 |
| mntmn | semi | 21:36 |
| mntmn | IDA can disassemble it | 21:36 |
| paulk-leonov | ah like, general-purpose with custom extensions? | 21:36 |
| mntmn | but IIRC there are vendor extensions in the ISA, not sure atm | 21:36 |
| mntmn | yep | 21:36 |
| Kooda | mntmn: isn’t the HDMI blob source open-source (maybe decompiled?) Being able to inspect it, and maybe reproduce the binary, would at least be *something* | 21:36 |
| paulk-leonov | I see | 21:36 |
| mntmn | it is ARCcompact | 21:36 |
| mntmn | Kooda: huh? where did you see it? | 21:37 |
| paulk-leonov | ofc, synopsys stuff ;) | 21:37 |
| mntmn | sorry ARCompact | 21:37 |
| mntmn | paulk-leonov: yeah | 21:37 |
| mntmn | it is synopsys ip | 21:37 |
| Kooda | mntmn: I can’t remember, maybe I dreamed it xD | 21:37 |
| mntmn | Kooda: the hdmi thing is Xtensa ISA. it is probably possible to disassemble but when i tried i had only garbage results | 21:37 |
| mntmn | but i'm also not an experienced RE | 21:38 |
| paulk-leonov | it's really a shame, the imx8m family looks pretty neat otherwise | 21:38 |
| mntmn | yeah i have made my piece with it... there are lots of little things to like | 21:38 |
| mntmn | for example, you can just poke the registers at runtime with memtool | 21:38 |
| mntmn | and do all kinds of weird stuff | 21:38 |
| paulk-leonov | hehe | 21:38 |
| mntmn | that has a retro feel to it | 21:39 |
| paulk-leonov | are there platforms that don't allow that somehow? | 21:39 |
| mntmn | i guess non SoC platforms | 21:40 |
| paulk-leonov | MMIO vs PIO I guess | 21:41 |
| mntmn | yeah | 21:41 |
| mntmn | also i really like that we have etnaviv in the first place | 21:41 |
| mntmn | even if it is GLES2.0 you can do a lot of fun things with that | 21:42 |
| Kooda | The case about the HDMI blob reminds me of video game console cracking, as well. Some people can figure out keys and/or bugs in the security that allow you to use unsigned code. It would be neat to find that for that SoC. :Þ | 21:42 |
| mntmn | yes | 21:42 |
| mntmn | Kooda: i would not be surprised | 21:42 |
| Kooda | mntmn: yes that’s (open-source GL) one of the main thing I was looking for in an open-source machine. | 21:42 |
| mntmn | i hope to see more fun stuff once all those Reforms are in peoples' hands and also those librem phones too | 21:43 |
| Kooda | Yep :) | 21:43 |
| Kooda | I will definitely tinker with mine :) | 21:43 |
| paulk-leonov | free software graphics support is in quite good shape on embedded these days | 21:43 |
| paulk-leonov | powervr is slowly dying out | 21:44 |
| mntmn | except for the stuff with powervr yeah you named it | 21:44 |
| paulk-leonov | others are doing better and better | 21:44 |
| mntmn | yep | 21:44 |
| mntmn | it surprises me that arm didn't just open their gpu stuff in the first place, it seems dumb | 21:44 |
| paulk-leonov | yeah, pattent stuff and worried lawyers I guess | 21:45 |
| Kooda | I wonder what will happen now that nvidia own them | 21:45 |
| Kooda | They aren’t really keen on openning anything | 21:45 |
| mntmn | the nvidia deal is still not 100% through, right? | 21:45 |
| mntmn | there could still be some antitrust thing | 21:45 |
| Kooda | I admit I don’t know the details | 21:45 |
| paulk-leonov | yeah let's see | 21:46 |
| mntmn | i just read that a bunch of other big corps (incl. MS) wanted to protest | 21:46 |
| paulk-leonov | but indeed nvidia has a pretty bad track record | 21:46 |
| mntmn | i'm also for sure looking forward to RISC-V maturity. my strategy is to support both ARM+RISC-V until RISC-V is in really good shape. | 21:47 |
| mntmn | but we still don't have a GPU story in that space | 21:47 |
| mntmn | there might be some RISC-V chip with vivante coming but still vivante is not open, just REd | 21:47 |
| Kooda | It will take a while yeah | 21:47 |
| mntmn | yeah | 21:47 |
| Kooda | But it’s really exciting :) | 21:47 |
| mntmn | but in the meantime we can learn a lot about building devices around chips | 21:47 |
| mntmn | open devices that is | 21:48 |
| mntmn | and we will have an FPGA module to experiment with open cores | 21:48 |
| Kooda | Yep, that’s awesome. Thanks for looking into that :) | 21:50 |
| * demoveu -> debacle | 21:52 | |
| * debacle -> demoveu | 21:52 | |
| - Neelfyn (QUIT: Quit: Connection closed for inactivity) (uid180106@gateway/web/irccloud.com/x-pvukjbjkbggrtkjz) | 22:11 | |
| * demoveu -> debacle | 22:19 | |
| - rasmus-mk (QUIT: Quit: Konversation terminated!) (~rasmus@c83-253-223-217.bredband.comhem.se) | 22:40 | |
| - erlehmann (QUIT: Read error: Connection reset by peer) (~erlehmann@dynamic-046-114-035-154.46.114.pool.telefonica.de) | 23:10 | |
| + erlehmann (~erlehmann@dynamic-046-114-037-211.46.114.pool.telefonica.de) | 23:25 | |
| - oomono (QUIT: Quit: Connection closed for inactivity) (uid328183@gateway/web/irccloud.com/x-teagdqzowwisxlkg) | 23:28 | |
| * debacle -> demoveu | 23:47 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!