| gnarface | freaxeh: you just need to install firmware-amd-graphics from non-free-firmware (moved to non-free-firmware from non-free as of daedalus) | 02:53 |
|---|---|---|
| gnarface | sorry if you already figured that out | 02:53 |
| gnarface | just catching up | 02:53 |
| freemangordon | rrq: af4c84ce8 is shown to be there, but the changes it makes are not. | 06:44 |
| freemangordon | so busid is not set in lnx_platform.c | 06:45 |
| freemangordon | but in udev.c instead and *only* for PCI devices, see https://git.devuan.org/devuan/xorg-server/src/branch/suites/daedalus-security/config/udev.c#L546 | 06:47 |
| freemangordon | so, for platform devices (like omapdrm), busid is not set | 06:47 |
| freemangordon | dunno, maybe that's on purpose and I cannot get it. | 06:48 |
| freemangordon | but at least modesetting does this https://git.devuan.org/devuan/xorg-server/src/branch/suites/daedalus-security/hw/xfree86/drivers/modesetting/driver.c#L1019 | 06:53 |
| freemangordon | which I cannot do in DDX to get mediated fd, because pEnt->location.type in DDX is set to BUS_NONE and actually there is no data | 06:55 |
| rrq | I need to do "dpkg-source --before-build ." to apply the patches before looking at the code | 07:05 |
| rrq | then that commit edits show up | 07:05 |
| freemangordon | patches do not touch that code, but yeah, it is better someone else to confirm | 07:06 |
| rrq | mmm I used suites/daedalus-proposed-updates which I think is the current | 07:08 |
| freemangordon | well, I used security which is the latest relesed | 07:09 |
| freemangordon | but should not matter as those 2 files are not changed by devuan | 07:09 |
| rrq | yes it is | 07:09 |
| rrq | I've changed to daedalus-security.. (my "yes" was for the branch | 07:11 |
| rrq | I dont know the code terribly well... | 07:14 |
| rrq | what's the core problem? | 07:15 |
| freemangordon | "for platform devices (like omapdrm), busid is not set" | 07:16 |
| rrq | and busid is needed? | 07:17 |
| freemangordon | how to get mediated fd otherwise? | 07:17 |
| freemangordon | https://git.devuan.org/devuan/xorg-server/src/branch/suites/daedalus-security/hw/xfree86/drivers/modesetting/driver.c#L1019 | 07:17 |
| freemangordon | hmm, how to rebuild xorg packages without recompiling everything when only a single file has changed? -nc seems to do nothing :( | 07:21 |
| rrq | run "debian/rules binary" ... following the "dpkg-source --before-build . | 07:22 |
| rrq | well maybe that builds everything (?) | 07:23 |
| freemangordon | lemme try | 07:24 |
| freemangordon | nope | 07:24 |
| freemangordon | anyway, will rebuild everuything | 07:24 |
| freemangordon | I think I have a patch for that, but if every test will require to rebuild the whole tree, I may retire before I'm done with that | 07:30 |
| rrq | is xf86_find_platform_device_by_devnum useful? | 07:31 |
| freemangordon | didn't try it, but wouldn't that be a hack? | 07:32 |
| freemangordon | omapdrm *is* on PLATFORM_BUS | 07:33 |
| rrq | there's the global struct xf86_platform_device *xf86_platform_devices; | 07:33 |
| freemangordon | or BUS_PLATFORM or whatever it is | 07:33 |
| freemangordon | ok, lemme try the patch first, will try is xf86_find_platform_device_by_devnum later on | 07:34 |
| rrq | and global int xf86_num_platform_devices; | 07:34 |
| rrq | hw/xfree86/common/xf86platformBus.c | 07:35 |
| rrq | when you find the right xf86_platform_device struct it'll be attribs->fd | 07:37 |
| freemangordon | yeah | 07:37 |
| rrq | (I think) | 07:37 |
| freemangordon | I think too | 07:37 |
| freemangordon | I have the feeling modesetting will not run too, lemme try | 07:38 |
| freemangordon | mhm, modesetting fails too | 07:51 |
| freemangordon | rrq: https://pastebin.com/zX6Kvp1k | 07:53 |
| rrq | is that where /dev/vga_arbiter is of use? does the user have permissions? | 07:59 |
| rrq | (or I misunderstand the issue) | 07:59 |
| freemangordon | (II) modeset(G0): using drv /dev/dri/card1 | 08:00 |
| freemangordon | Cannot run in framebuffer mode. Please specify busIDs for all framebuffer devices | 08:00 |
| freemangordon | that's the same issue to me | 08:01 |
| freemangordon | omapdrm does not have busid | 08:01 |
| freemangordon | user@devuan-droid4:~$ ls -al /dev/dri/by-path/platform-omapdrm.0-card | 08:01 |
| freemangordon | lrwxrwxrwx 1 root root 8 Jan 10 06:48 /dev/dri/by-path/platform-omapdrm.0-card -> ../card1 | 08:01 |
| rrq | I was also reading "VGA arbiter: cannot open kernel arbiter, no multi-card support" .. but maybe that's a herring | 08:03 |
| freemangordon | it is | 08:03 |
| rrq | ok | 08:03 |
| freemangordon | it cannot open it, but it opens the correct device | 08:03 |
| rrq | is card1 opened with mdeiation? | 08:03 |
| freemangordon | should be | 08:04 |
| freemangordon | seatd_libseat try open graphics /dev/dri/card1 | 08:04 |
| freemangordon | [libseat/backend/logind.c:161] DRM device opened, current total: 2 | 08:04 |
| freemangordon | it is | 08:04 |
| freemangordon | will try with this patch https://pastebin.com/dTUu45eH on top | 08:06 |
| rrq | seems good | 08:08 |
| freemangordon | lets hope it will fix the issue | 08:10 |
| rrq | btw you could insert "/raw" in the url (https://pastebin.com/raw/dTUu45eH) to spare people from watching ads :) | 08:15 |
| freemangordon | oh, sorry, I am using adblocker and have no clue there are ads | 08:16 |
| freemangordon | but yeah, thanks, will do | 08:17 |
| * rrq off on an outing. biab. | 08:18 | |
| rrq | please send me the patch (when it works) so I can tape it onto the devuan source | 09:40 |
| rrq | freemangordon: ^ | 09:41 |
| rrq | (if it needs more than recognising "platform-") | 09:52 |
| freemangordon | rrq: sure, thanks! | 10:28 |
| freaxeh2 | <freaxeh2> does devuan work ok with say an rtx 4070? | 12:06 |
| freaxeh2 | <freaxeh2> or is it best to stick to amd cards | 12:06 |
| freaxeh2 | <freaxeh2> was just wondering what the current state of affairs is | 12:06 |
| freaxeh2 | before i blow a heap of cash in a few months time | 12:06 |
| kennylevinsen | some entirely untested possible improvements to the xserver libseat patch: https://gitlab.freedesktop.org/kennylevinsen/xserver/-/tree/devuan-libseat-fix?ref_type=heads | 13:13 |
| freemangordon | kennylevinsen: what to use as libsystemd-dev replacement? | 13:36 |
| freemangordon | if I want to test without seatd | 13:36 |
| kennylevinsen | well you can use libseat with seatd *or* logind, but if you want to use xorg's own logind impl, libelogind is a libsystemd substitute purely for the logind side of things. Not sure if xserver is wired up for it upstream though. | 13:38 |
| freaxeh | ty gnarface | 14:49 |
| djph | man, what a pain to get CUPS to finally behave with the 'lp' command in Daedalus ... no idea why "AuthInfoRequired none" was ultimately invalid ... *sigh* | 15:24 |
| djph | turned out to need "AuthInfoRequired negotiate" :( | 16:43 |
| gnarface | freaxeh_: you should be able to figure it out by looking at the package versions in the repo then comparing it to the supported cards list on nvidia's own download site for that particular version, but personally i would recommend you stick with AMD cards based on my experience that even to this day all the king's horses and all the king's men can't make nvidia's drivers stop crashing | 21:19 |
| gnarface | btw did you get that firmware-amd-graphics package installed and did it give you all your video card features back? (if you got resolution control back but are still missing other features you're probably just still missing some "optional" mesa packages) | 21:20 |
| gnarface | freaxeh_: freaxeh and freaxeh2 are both you as well, right? just confirming to be sure... | 21:21 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!