Bug#861015: stretch installer RC2
Package: installation-reports
Boot method: dvd
Image version:
http://cdimage.debian.org/cdimage/stretch_di_rc2/amd64/iso-cd/debian-stretch-DI-rc2-amd64-netinst.iso
Date: April 2, 2017
Machine: hp pavilion a6110n
Processor: amd 64 bit, dual core, 2.3Ghz
Memory: 2G
Output of lspci -knn:
00:00.0 RAM memory [0500]: nVidia Corporation MCP61 Memory Controller
[10de:03ea] (rev a1)
Subsystem: Hewlett-Packard Company Device [103c:2a58]
00:01.0 ISA bridge [0601]: nVidia Corporation MCP61 LPC Bridge
[10de:03e0] (rev a2)
Subsystem: Hewlett-Packard Company Device [103c:2a58]
00:01.1 SMBus [0c05]: nVidia Corporation MCP61 SMBus [10de:03eb] (rev a2)
Subsystem: Hewlett-Packard Company Device [103c:2a58]
Kernel driver in use: nForce2_smbus
Kernel modules: i2c-nforce2
00:01.2 RAM memory [0500]: nVidia Corporation MCP61 Memory Controller
[10de:03f5] (rev a2)
Subsystem: Hewlett-Packard Company Device [103c:2a58]
00:02.0 USB Controller [0c03]: nVidia Corporation MCP61 USB Controller
[10de:03f1] (rev a3)
Subsystem: Hewlett-Packard Company Device [103c:2a58]
Kernel driver in use: ohci_hcd
00:02.1 USB Controller [0c03]: nVidia Corporation MCP61 USB Controller
[10de:03f2] (rev a3)
Subsystem: Hewlett-Packard Company Device [103c:2a58]
Kernel driver in use: ehci_hcd
00:04.0 PCI bridge [0604]: nVidia Corporation MCP61 PCI bridge
[10de:03f3] (rev a1)
00:05.0 Audio device [0403]: nVidia Corporation MCP61 High Definition
Audio [10de:03f0] (rev a2)
Subsystem: Hewlett-Packard Company Device [103c:2a58]
Kernel driver in use: HDA Intel
Kernel modules: snd-hda-intel
00:06.0 IDE interface [0101]: nVidia Corporation MCP61 IDE [10de:03ec]
(rev a2)
Subsystem: Hewlett-Packard Company Device [103c:2a58]
Kernel driver in use: pata_amd
Kernel modules: ata_generic, pata_acpi, pata_amd
00:07.0 Bridge [0680]: nVidia Corporation MCP61 Ethernet [10de:03ef]
(rev a2)
Subsystem: Hewlett-Packard Company Device [103c:2a58]
Kernel driver in use: forcedeth
Kernel modules: forcedeth
00:08.0 IDE interface [0101]: nVidia Corporation MCP61 SATA Controller
[10de:03f6] (rev a2)
Subsystem: Hewlett-Packard Company Device [103c:2a58]
Kernel driver in use: sata_nv
Kernel modules: ata_generic, pata_acpi, sata_nv
00:08.1 IDE interface [0101]: nVidia Corporation MCP61 SATA Controller
[10de:03f6] (rev a2)
Subsystem: Hewlett-Packard Company Device [103c:2a58]
Kernel driver in use: sata_nv
Kernel modules: ata_generic, pata_acpi, sata_nv
00:09.0 PCI bridge [0604]: nVidia Corporation MCP61 PCI Express bridge
[10de:03e8] (rev a2)
Kernel driver in use: pcieport
Kernel modules: shpchp
00:0b.0 PCI bridge [0604]: nVidia Corporation MCP61 PCI Express bridge
[10de:03e9] (rev a2)
Kernel driver in use: pcieport
Kernel modules: shpchp
00:0d.0 VGA compatible controller [0300]: nVidia Corporation C61
[GeForce 6150SE nForce 430] [10de:03d0] (rev a2)
Subsystem: Hewlett-Packard Company Device [103c:2a58]
Kernel driver in use: nvidia
Kernel modules: nvidia, nouveau, nvidiafb
00:18.0 Host bridge [0600]: Advanced Micro Devices [AMD] K8
[Athlon64/Opteron] HyperTransport Technology Configuration [1022:1100]
00:18.1 Host bridge [0600]: Advanced Micro Devices [AMD] K8
[Athlon64/Opteron] Address Map [1022:1101]
00:18.2 Host bridge [0600]: Advanced Micro Devices [AMD] K8
[Athlon64/Opteron] DRAM Controller [1022:1102]
Kernel modules: amd64_edac_mod
00:18.3 Host bridge [0600]: Advanced Micro Devices [AMD] K8
[Athlon64/Opteron] Miscellaneous Control [1022:1103]
Kernel driver in use: k8temp
Kernel modules: k8temp
01:05.0 FireWire (IEEE 1394) [0c00]: Agere Systems FW322/323 [11c1:5811]
(rev 70)
Subsystem: Hewlett-Packard Company Device [103c:2a58]
Kernel driver in use: firewire_ohci
Kernel modules: firewire-ohci
01:09.0 Multimedia video controller [0400]: Conexant Systems, Inc.
CX23880/1/2/3 PCI Video and Audio Decoder [14f1:8800] (rev 05)
Subsystem: ATI Technologies Inc HDTV Wonder [1002:a101]
Kernel driver in use: cx8800
Kernel modules: cx8800
01:09.1 Multimedia controller [0480]: Conexant Systems, Inc.
CX23880/1/2/3 PCI Video and Audio Decoder [Audio Port] [14f1:8801] (rev 05)
Subsystem: ATI Technologies Inc Device [1002:a101]
Kernel driver in use: cx88_audio
Kernel modules: cx88-alsa
01:09.2 Multimedia controller [0480]: Conexant Systems, Inc.
CX23880/1/2/3 PCI Video and Audio Decoder [MPEG Port] [14f1:8802] (rev 05)
Subsystem: ATI Technologies Inc Device [1002:a101]
Kernel driver in use: cx88-mpeg driver manager
Kernel modules: cx8802
01:0a.0 Network controller [0280]: Broadcom Corporation BCM4306
802.11b/g Wireless LAN Controller [14e4:4320] (rev 02)
Subsystem: Linksys Device [1737:0013]
Kernel driver in use: b43-pci-bridge
Kernel modules: ssb
Base System Installation Checklist:
[O] = OK, [E] = Error (please elaborate below), [ ] = didn't try it
Initial boot: [O ]
Detect network card: [O ]
Configure network: [O ]
Detect CD: [O ]
Load installer modules: [O ]
Detect hard drives: [O ]
Partition hard drives: [O ]
Install base system: [O ]
Clock/timezone setup: [O ]
User/password setup: [O ]
Install tasks: [O ]
Install boot loader: [E ]
Overall install: [E ]
Comments/Problems:
There are two things unusual about this hardware.
First, there are three hard drives, plus four memory card slots which are
assigned drive letters even when there is no card in the slot, plus a usb
drive I am using to take notes on the installation, and to transfer data
between operating systems. So there are a lot of drive letters in use. One
of the hard drives has a working installation of fedora 13 and old grub. I
have been using fedora 13 for years, and I do not want the stretch installer
to make any changes to it. I am trying to install stretch to a hard drive
which has nothing on it.
Second, the display has an invalid edid checksum. The display works fine
with any software which ignores the edid checksum. In fedora 13, I have
been using the xorg nvidia driver which accepts option IgnoreEDIDChecksum in
xorg.conf, and the display works perfectly.
While I am running the installer, if I switch to a different virtual
terminal, the first two columns are missing. If I enter the command 'echo
12345', it displays '345' in the first three columns on the next line, and
the '12' is not visible anywhere on the display. This problem also occurs
in rescue mode. This problem does not occur if I use a different display.
The problem is not changed if I use kernel boot parameters nomodeset
nouveau.modeset=0.
The installer prompted to install grub onto the first hard drive. I did not
know which hard drive the installer meant by first. I was afraid the
installer would put grub on the fedora 13 drive instead of the stretch
drive, so I clicked no. The next screen of the installer offered a choice
of locations to install grub. I selected the correct hard drive and
installed grub.
After the installer finished, I tried to reboot but the computer would not
boot from the stretch hard drive. The computer went through the bios ok,
then nothing, no messages at all, just a cursor in the upper left corner;
which usually means no boot loader. It is possible that in my confusion
about which hard drive was which, I pressed the wrong key at some point.
The computer would still boot normally from the fedora 13 hard drive, same
as before I installed stretch. I used the install disk in rescue mode to
reinstall grub, and then the computer would boot from the stretch hard
drive.
I think that when the installer prompts to install grub, it should display
more information about the hard drives to make it easier for the user to
select the correct hard drive. The 'no' option should not be called 'no'
because that seems like not installing grub at all. Instead it should be
called install grub to alternate location or install grub manually or expert
mode.
When stretch booted, it displayed some messages saying the root partition
was ok, then some messages saying the EDID checksum was bad, then there was
a black screen with an arrow shaped cursor in the center. The computer did
not respond to the keyboard or mouse. I could not change to a different
virtual terminal. No login prompt was displayed, so the crash probably
occurred before the login prompt. Control-alternate-delete and
control-alternate-backspace had no effect.
I used the install disk in rescue mode to check the logs. The last
messages in
/var/log/messages were:
Apr 16 14:32:59 hpx2 gnome-shell[687]: Failed to apply DRM plane
transform 0: Invalid argument
Apr 16 14:32:59 hpx2 org.gnome.Shell.desktop[687]: glamor: EGL version
1.4 (DRI2):
Apr 16 14:33:00 hpx2 org.a11y.atspi.Registry[726]: SpiRegistry daemon is
running with well-known name - org.a11y.atspi.Registry
Apr 16 14:33:01 hpx2 pulseaudio[734]: [pulseaudio] source.c: Default and
alternate sample rates are the same.
Apr 16 14:33:02 hpx2 pulseaudio[734]: [pulseaudio] authkey.c: Failed to
open cookie file '/var/lib/gdm3/.config/pulse/cookie': No such file or
directory
Apr 16 14:33:02 hpx2 pulseaudio[734]: [pulseaudio] authkey.c: Failed to
load authentication key '/var/lib/gdm3/.config/pulse/cookie': No such
file or directory
Apr 16 14:33:02 hpx2 pulseaudio[734]: [pulseaudio] authkey.c: Failed to
open cookie file '/var/lib/gdm3/.pulse-cookie': No such file or directory
Apr 16 14:33:02 hpx2 pulseaudio[734]: [pulseaudio] authkey.c: Failed to
load authentication key '/var/lib/gdm3/.pulse-cookie': No such file or
directory
Apr 16 14:33:07 hpx2 gnome-shell[687]: JS LOG: Failed to launch
ibus-daemon: Failed to execute child process "ibus-daemon" (No such file
or directory)
Apr 16 14:33:10 hpx2 gnome-session-binary[679]: Entering running state
Apr 16 14:33:10 hpx2 colord[775]: failed to get edid data: EDID length
is too small
Apr 16 14:33:10 hpx2 xbrlapi.desktop[773]: openConnection: connect: No
such file or directory
Apr 16 14:33:10 hpx2 xbrlapi.desktop[773]: cannot connect to braille
devices daemon brltty at :0
Apr 16 14:33:11 hpx2 gnome-settings-[750]: g_task_return_error:
assertion 'error != NULL' failed
Apr 16 14:33:11 hpx2 gnome-settings-[750]: failed to get edid: unable to
get EDID for output
Apr 16 14:33:11 hpx2 gnome-settings-[750]: unable to get EDID for
xrandr-VGA-1: unable to get EDID for output
Using the install disk in rescue mode, with the stretch partition selected
as the root partition: If I select 'execute a shell in the installer
environment', then ping google.com works. If I select 'execute a shell in
/dev/sdg1' (the stretch partition), ping google.com says name or service
unknown. Maybe DNS does not work because the nameservers are lost when it
does chroot? Lack of networking makes it hard to install packages from
repositories.
I tried a different display. Stretch booted ok with a different display.
The grub boot menu is only visible for 3 seconds. This is not enough time
for me to choose recovery mode or any other boot options. I used the
install disk in rescue mode to do
echo 'set timeout=-1' >> /boot/grub/custom.cfg
Stretch crashes in the same manner in recovery mode. Recovery mode should
not require xwindows, or else there should be a alternate non xwindows
recovery mode. The grub menu offers three different ways to boot
stretch, and all three are the same, or at least result in the same crash. I
made a temporary rescue mode by editing the kernel boot parameters and
adding init=/bin/sh.
I copied the menu items from /boot/grub/grub.cfg to
/boot/grub/custom.cfg and
edited the kernel boot parameters.
Adding kernel boot parameters nomodeset nouveau.modeset=0 caused stretch to
crash earlier in the boot process. The last message displayed on the screen
was KVM: disabled by bios.
Adding kernel boot parameters edid_strict=0 drm_edid_strict=0
drm.edid_strict=0
had no effect. This option is for a proposed kernel patch at
https://lists.freedesktop.org/archives/dri-devel/2011-January/006778.html.
I do not know if that proposed kernel patch was ever accepted into the
kernel, and do not know the exact syntax, but I figured it would not hurt
to try.
Adding kernel boot parameter drm_kms_helper.edid_firmware=edid/1024x768.bin
had no effect. I think this kernel option is automatically activated by
edid problems, which is why giving it had no effect for me. If I use this
kernel boot parameter with a different display, stretch does not crash.
Adding kernel boot parameter drm_kms_helper.edid_firmware=edid/1680x1050.bin
caused the text in the displayed error messags to become smaller, but
otherwise stretch crashed the same.
I tried kernel boot parameter drm_kms_helper.edid_firmware=edid.bin.
/lib/firmware/edid.bin was a file created by the write edid to file function
of the nvidia utilities on fedora 13. This resulted in an error message
saying base block of edid firmware is invalid. I do not know if the edid
file generated by nvidia utilities is the correct format for the kernel. I
tried generating another edid file with get-edid, but get-edid failed.
I tried installing the nvidia driver with:
dpkg --add-architecture i386
apt update
apt install nvidia-legacy-304xx-driver
cat << endcat > /etc/X11/xorg.conf
Section "Device"
Identifier "Device0"
Driver "nvidia"
Option "IgnoreEDIDChecksum" "CRT-0"
EndSection
endcat
Xwindows ran very slowly and crashed frequently. The same thing happened if
I used a different display.
I tried deleting the nvidia driver and switching to the vesa driver:
apt autoremove nvidia-legacy-304xx-driver
cat << endcat > /etc/X11/xorg.conf
Section "Device"
Identifier "Device0"
Driver "vesa"
EndSection
endcat
Messages displayed during apt autoremove said it was remaking the initramfs
and the grub configuration. I vaguely recall that apt install said the
nvidia driver would use about 300M of disk space, and apt autoremove said
40M of disk space would be freed up.
I rebooted using kernel boot parameter systemd.unit=multi-user.target.
Stretch ran ok in multi-user mode, but in 80 columns x 25 rows. I did
startx. Xwindows failed to start. One of the error messages said vesa:
Ignoring device with a bound kernel driver. Maybe the nvidia driver did not
uninstall cleanly. Maybe I should have tried the vesa driver before trying
the nvidia driver.
So stretch works in multi-user mode, but I cannot use xwindows.
I think the default /etc/apt/sources should include stretch-updates.
Reply to: