You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am transferring a bunch of (late night your time , early morning my time ) posts from the Discord to here, try to de-clutter the Discord channels (if I can).
@paulwratt — 2023/10/31 at 10:27 AM
I have not been able to get X running, so I cant use FireFox on that machine yet. X did a core-dump (and hang) the 1st time I ran it (as root), but it just quits if I startx as a user "unknown display: t2-5700G:0", when it doesn't core-dump and (looks like a) hang
paulwratt — 2023/10/31 at 10:35 AM
xauth: (argv):1: bad display name "t2-5700G:0" in "remove" command
VT switchinit: connection to X server lost
am I having problems with the default Xorg configuration because I have a - symbol in the hostname?
@r2t2 — 2023/10/31 at 10:51 AM
I have seen this sometimes, should probably have debugged where this regression is coming from 😕
which gpu is this? or VM?
@paulwratt — 2023/10/31 at 11:10 AM amdgpu - but I also noticed in ... /xor.conf.d/ that there is both 10-amdgpu.conf and 10-radeon.conf although the screen and card sections reference driver=amdgpu .
its the internal GPU for the 5700G on an Asus Tuff Gaming B550M Plus (WiFi)
@paulwratt — 2023/10/31 at 11:22 AM
I too like the idea of a PID. dont have to grep a special PS then. I think that last entry in rocknet_dhcp is missing another line:
addcode down 5 6 "sleep 1; ip link set $if down"
@paulwratt — 2023/10/31 at 12:45 PM
Hmm .. I just ran /etc/init.d/xcfgt2 start and it says X Driver: radeon
and there is no SystemID either
OK that "no SystemID" is because dmidecode (and ddcprobe) is not install on the base ISO
( I am using t2-23.10-x86-64-minimal-firefox-gcc-musl-nocona.iso )
@paulwratt — 2023/10/31 at 1:00 PM xcfgt2 has 2 ways of picking up the graphics card, the one it is showing me is the one from lspci which looks for a VGA, but if that fails, the fallback is /sys/class/graphics/fb0/name and on my setup that is amdgpudrmfb
@paulwratt — 2023/10/31 at 1:08 PM
note that the fallback if either of those fail (older graphics card) is to look in uname -m for i*86|x86*64 and use vesa otherwise use fbdev, but this kernel outputs x86-64 and we already know from above that it should choose fbdev on this machine - it never gets to keep the output of this section tho, because in ..
the next section, the 1st match is for radeon so ..
there should be a selection before that that reads:
*amdgpu*) xdrv=amdgpu
and probably have 2x checks for the card so you can print the lspci entry, but if there is anything in ../fb0/name then it should use that to match for the correct driver
@paulwratt — Yesterday at 1:21 PM
BTW I have an old P4 with an HD5470 PCI, so I could (potentially) test t2 for a valid Radeon card
RE: 2x gfx card checks: or add a clause after that CASE section, like the check for xdrv=nv
@paulwratt — 2023/10/31 at 1:32 PM
Hmm .. I am not 110% sure but I think there is an X Windows DRM graphics driver specifically called amdgpudrmfb. EDIT: I might have confused that with DRI ..
(I am going of some vague items I saw on the RPi and using DRM in an xorg.conf)
@paulwratt — 2023/10/31 at 5:35 PM
ok, did some changes, and the result is ... exactly the same as the 1st time I ran it as root, VT7 has a blinking cursor, But VT5 (where I run it from) is blank, and when I (gracefully) kill xinit there is a bunch of XFCE errors (WARNINGs actually, all with the wrong EOL characters), and :
No window manager registered on screen 0. To start the xfdesktop without this check, run with --disable-um-check
and that same xauth: error posted above.
@paulwratt — 2023/10/31 at 5:44 PM
Note that is the same end result as when Xorg is run with the radeon driver too
I would need console soft-scrollback to be able to debug this any further
The following is a list that contains general information about GPUs and video cards by AMD, including those by ATI Technologies before 2006, based on official specifications in table-form.
amdgpu is the correct driver for a VEGA class GPU (like the one in the 5700G)
@r2t2 — 2023/10/31 at 8:21 PM
Best run config less. Xcfgt2 is an old thing from when a config was needed
I mostly left it for illustration purposes ONLY
I missed this post for some reason ???
@paulwratt — 2023/11/01 at 12:09 AM
Just spent a couple of (11.5?) hours looking at the output of :
X -verbose 6 > ~/xlog.txt 2>&1
and it appears that after much copy and pasting, editing and reviewing, that the failure might be due to the disconnected DisplayPort using Monitor1 and the connected HDMINOT having a Monitor section (oh joy)
@paulwratt — 2023/11/01 at 12:17 AM
Well at least I got Modelines out of it, and an EDID text binary blob 🙂
The text was updated successfully, but these errors were encountered:
After I have pushed some of these post install fixes, I can do another install processing.
I don't mind updating xcfg2 if it becomes useful again.
I'd like to get a working no-config boot straight into X after 1st or 2nd reboot, even if that requires fixing some stuff.
If there are some other partitioning related stuff, I'd like to burn thru that as well, because altho I can probably do infinite test installs on this machine, I do need the partitions to be fixed (unchanging) at some point so I can start work on some other projects.
To that end one of the 1st things I want to get set in STONE ( sic ) is a working hybrid partition scheme. I have a P4 that I can test BIOS boot on, and it has a PCI HD5470 which i would like to get " booting to X after 1st or 2nd reboot " too.
I am transferring a bunch of (late night your time , early morning my time ) posts from the Discord to here, try to de-clutter the Discord channels (if I can).
@paulwratt — 2023/10/31 at 10:27 AM
I have not been able to get X running, so I cant use FireFox on that machine yet. X did a core-dump (and hang) the 1st time I ran it (as root), but it just quits if I
startx
as a user "unknown display:t2-5700G:0
", when it doesn'tcore-dump
and (looks like a) hangpaulwratt — 2023/10/31 at 10:35 AM
am I having problems with the default Xorg configuration because I have a
-
symbol in thehostname
?@r2t2 — 2023/10/31 at 10:51 AM
I have seen this sometimes, should probably have debugged where this regression is coming from 😕
which gpu is this? or VM?
@paulwratt — 2023/10/31 at 11:10 AM
amdgpu
- but I also noticed in... /xor.conf.d/
that there is both10-amdgpu.conf
and10-radeon.conf
although the screen and card sections referencedriver=amdgpu
.its the internal GPU for the 5700G on an Asus Tuff Gaming B550M Plus (WiFi)
@paulwratt — 2023/10/31 at 11:22 AM
I too like the idea of a PID. dont have to grep a special PS then. I think that last entry in
rocknet_dhcp
is missing another line:@paulwratt — 2023/10/31 at 12:45 PM
Hmm .. I just ran
/etc/init.d/xcfgt2 start
and it saysX Driver: radeon
and there is no SystemID either
OK that "no SystemID" is because
dmidecode
(andddcprobe
) is not install on the base ISO( I am using
t2-23.10-x86-64-minimal-firefox-gcc-musl-nocona.iso
)@paulwratt — 2023/10/31 at 1:00 PM
xcfgt2
has 2 ways of picking up the graphics card, the one it is showing me is the one fromlspci
which looks for aVGA
, but if that fails, the fallback is/sys/class/graphics/fb0/name
and on my setup that isamdgpudrmfb
@paulwratt — 2023/10/31 at 1:08 PM
note that the fallback if either of those fail (older graphics card) is to look in
uname -m
fori*86|x86*64
and usevesa
otherwise use fbdev, but this kernel outputsx86-64
and we already know from above that it should choosefbdev
on this machine - it never gets to keep the output of this section tho, because in ..the next section, the 1st match is for radeon so ..
there should be a selection before that that reads:
and probably have 2x checks for the card so you can print the
lspci
entry, but if there is anything in../fb0/name
then it should use that to match for the correct driver@paulwratt — Yesterday at 1:21 PM
BTW I have an old P4 with an HD5470 PCI, so I could (potentially) test t2 for a valid Radeon card
RE: 2x gfx card checks: or add a clause after that CASE section, like the check for
xdrv=nv
@paulwratt — 2023/10/31 at 1:32 PM
Hmm .. I am not 110% sure but I think there is an X Windows DRM graphics driver specifically called
amdgpudrmfb
. EDIT: I might have confused that with DRI ..(I am going of some vague items I saw on the RPi and using DRM in an xorg.conf)
@paulwratt — 2023/10/31 at 5:35 PM
ok, did some changes, and the result is ... exactly the same as the 1st time I ran it as root, VT7 has a blinking cursor, But VT5 (where I run it from) is blank, and when I (gracefully) kill
xinit
there is a bunch of XFCE errors (WARNINGs actually, all with the wrong EOL characters), and :and that same
xauth:
error posted above.@paulwratt — 2023/10/31 at 5:44 PM
Note that is the same end result as when Xorg is run with the radeon driver too
I would need console soft-scrollback to be able to debug this any further
I looked at the Arch Linux wiki:
@paulwratt — 2023/10/31 at 5:51 PM
https://wiki.archlinux.org/title/Xorg#AMD
Xorg
and I checked their reference with WikiPedia:
https://en.wikipedia.org/wiki/List_of_AMD_graphics_processing_units#Features_overview
List of AMD graphics processing units
amdgpu
is the correct driver for a VEGA class GPU (like the one in the 5700G)@r2t2 — 2023/10/31 at 8:21 PM
Best run config less. Xcfgt2 is an old thing from when a config was needed
I mostly left it for illustration purposes ONLY
@paulwratt — 2023/11/01 at 12:09 AM
Just spent a couple of (11.5?) hours looking at the output of :
and it appears that after much copy and pasting, editing and reviewing, that the failure might be due to the disconnected DisplayPort using
Monitor1
and the connected HDMI NOT having aMonitor
section (oh joy)@paulwratt — 2023/11/01 at 12:17 AM
Well at least I got
Modelines
out of it, and an EDID text binary blob 🙂The text was updated successfully, but these errors were encountered: