Re: Installing Buster on a Fit-PC: Can't find the network.
On Thu, 28 May 2020 22:49:58 -0500
David Wright <deblis@lionunicorn.co.uk> wrote:
> On Thu 28 May 2020 at 14:23:31 (-0600), Charles Curley wrote:
> > On Fri, 15 May 2020 19:47:02 -0500 David Wright
> > <deblis@lionunicorn.co.uk> wrote:
> > > On Fri 15 May 2020 at 16:03:12 (-0600), Charles Curley wrote:
>
> It shouldn't be. I only mentioned it because the installer mentions
> it. I'm sure the d-i itself boots up a non-pae kernel.
It does.
>
> In may depend on how you run the d-i. In expert mode, you get asked,
> as you might want to use the installation with less well endowed
> hardware. For stretch, the dialogue is
<mangled text capture deleted>
Right, I've seen a shorter version of that as well. Probably no pae
options.
> >
> > However, are these the kernel that is installed, or the one that
> > runs for the installer? Or both?
>
> These are packages for installation according to the choice you make
> as above (or is made for you). The installer's kernel is the vmlinuz
> in installer.386 (one of the three).
>
Which is not the same as what you showed earlier. According to the
hardware summary file (from yesterday's run of the installer):
lspci -knn: 00:0d.0 Ethernet controller [0200]: Realtek Semiconductor Co., Ltd. RTL-8100/8101L/8139 PCI Fast Ethernet Adapter [10ec:8139] (rev 10)
lspci -knn: Subsystem: Realtek Semiconductor Co., Ltd. RTL-8100/8101L/8139 PCI Fast Ethernet Adapter [10ec:8139]
syslog reports as follows:
May 28 18:38:07 kernel: [ 0.412728] pci 0000:00:0d.0: [10ec:8139] type 00 class 0x020000
May 28 18:38:07 kernel: [ 0.412771] pci 0000:00:0d.0: reg 0x10: [io 0xf800-0xf8ff]
May 28 18:38:07 kernel: [ 0.412797] pci 0000:00:0d.0: reg 0x14: [mem 0xe1014000-0xe10140ff]
May 28 18:38:07 kernel: [ 0.412913] pci 0000:00:0d.0: supports D1 D2
May 28 18:38:07 kernel: [ 0.412926] pci 0000:00:0d.0: PME# supported from D1 D2 D3hot
May 28 18:38:07 kernel: [ 0.413358] pci 0000:00:0e.0: [10ec:8139] type 00 class 0x020000
May 28 18:38:07 kernel: [ 0.413399] pci 0000:00:0e.0: reg 0x10: [io 0xf400-0xf4ff]
May 28 18:38:07 kernel: [ 0.413425] pci 0000:00:0e.0: reg 0x14: [mem 0xe1014100-0xe10141ff]
May 28 18:38:07 kernel: [ 0.413538] pci 0000:00:0e.0: supports D1 D2
May 28 18:38:07 kernel: [ 0.413551] pci 0000:00:0e.0: PME# supported from D1 D2 D3hot
But I do not see any of the three drivers in the lsmod listing in the
hardware summary file. So lspci is finding the hardware but the kernel
is neglecting to load the driver.
I looked at syslog for a running instance of buster. It has a lot more
references to the 8139, including:
May 28 15:00:32 chaffee kernel: [ 0.400872] pci 0000:00:0d.0: [10ec:8139] type 00 class 0x020000
May 28 15:00:32 chaffee kernel: [ 0.400914] pci 0000:00:0d.0: reg 0x10: [io 0xf800-0xf8ff]
May 28 15:00:32 chaffee kernel: [ 0.400939] pci 0000:00:0d.0: reg 0x14: [mem 0xe1014000-0xe10140ff]
May 28 15:00:32 chaffee kernel: [ 0.401049] pci 0000:00:0d.0: supports D1 D2
May 28 15:00:32 chaffee kernel: [ 0.401063] pci 0000:00:0d.0: PME# supported from D1 D2 D3hot
May 28 15:00:32 chaffee kernel: [ 0.401489] pci 0000:00:0e.0: [10ec:8139] type 00 class 0x020000
May 28 15:00:32 chaffee kernel: [ 0.401529] pci 0000:00:0e.0: reg 0x10: [io 0xf400-0xf4ff]
May 28 15:00:32 chaffee kernel: [ 0.401555] pci 0000:00:0e.0: reg 0x14: [mem 0xe1014100-0xe10141ff]
May 28 15:00:32 chaffee kernel: [ 0.401664] pci 0000:00:0e.0: supports D1 D2
May 28 15:00:32 chaffee kernel: [ 0.401677] pci 0000:00:0e.0: PME# supported from D1 D2 D3hot
Which, other than the host name, looks identical to the above (and
should).
Further down, there are several lines of output from the driver as it
initializes itself and the hardware. Those are not present in the
syslog from the installer.
>
> I checked out the /var/log/installer/syslog and here's the section
> that involves the NIC. The first part shows where the modules
> originate, the second shows their being used. You can see that the
> pae modules are skipped as the installer kernel is non-pae.
> (I've removed the timestamps after the first ones):
>
> Mar 22 20:05:27 anna[1939]: DEBUG: retrieving mtd-core-modules-4.19.0-5-686-di 4.19.37-5
> anna[1939]: DEBUG: retrieving nic-modules-4.19.0-5-686-di 4.19.37-5
> anna[1939]: DEBUG: retrieving nic-pcmcia-modules-4.19.0-5-686-di 4.19.37-5
> anna[1939]: DEBUG: retrieving nic-shared-modules-4.19.0-5-686-di 4.19.37-5
> anna[1939]: DEBUG: retrieving nic-usb-modules-4.19.0-5-686-di 4.19.37-5
> anna[1939]: DEBUG: retrieving nic-wireless-modules-4.19.0-5-686-di 4.19.37-5
> anna[1939]: DEBUG: retrieving xfs-modules-4.19.0-5-686-di 4.19.37-5
Nope, not in my installer syslog.
> [… …]
> Mar 22 20:06:27 anna[1939]: DEBUG: retrieving xfsprogs-udeb 4.20.0-1
> main-menu[220]: INFO: Falling back to the package description for brltty-udeb
> main-menu[220]: INFO: Falling back to the package description for brltty-udeb
> main-menu[220]: INFO: Menu item 'ethdetect' selected
> kernel: [ 301.076362] 8139cp: 8139cp: 10/100 PCI Ethernet driver v1.3 (Mar 22, 2004)
> kernel: [ 301.076383] 8139cp 0000:00:0e.0: This (id 10ec:8139 rev 10) is not an 8139C+ compatible chip, use 8139too
> kernel: [ 301.177580] 8139too: 8139too Fast Ethernet driver 0.9.28
> kernel: [ 301.217875] PCI Interrupt Link [LNKB] enabled at IRQ 5
> kernel: [ 301.218014] PCI: setting IRQ 5 as level-triggered
> kernel: [ 301.234695] 8139too 0000:00:0e.0 eth0: RealTek RTL8139 at 0x35988a16, 54:e6:fc:82:3f:be, IRQ 5
> net/hw-detect.hotplug: Detected hotpluggable network interface lo
> kernel: [ 301.811162] 8139too 0000:00:0e.0 enp0s14: renamed from eth0
Also not in my installer syslog, but in chaffee's.
> net/hw-detect.hotplug: Detected hotpluggable network interface enp0s14
> check-missing-firmware: looking at dmesg again, restarting from \[ 71.504225\]
> check-missing-firmware: timestamp found, truncating dmesg accordingly
> check-missing-firmware: saving timestamp for a later use: [ 301.811162]
> check-missing-firmware: /dev/.udev/firmware-missing does not exist, skipping
> check-missing-firmware: /run/udev/firmware-missing does not exist, skipping
> check-missing-firmware: no missing firmware in loaded kernel modules
> check-missing-firmware: taking network interface enp0s14 up/down
> kernel: [ 304.676284] 8139too 0000:00:0e.0 enp0s14: link up, 100Mbps, full-duplex, lpa 0xC5E1
> check-missing-firmware: looking at dmesg again, restarting from \[ 301.811162\]
> check-missing-firmware: timestamp found, truncating dmesg accordingly
> check-missing-firmware: saving timestamp for a later use: [ 304.676284]
> check-missing-firmware: /dev/.udev/firmware-missing does not exist, skipping
> check-missing-firmware: /run/udev/firmware-missing does not exist, skipping
> check-missing-firmware: no missing firmware in loaded kernel modules
Interesting. I have some similar lines in the installer syslog, but not
in chaffee's syslog.
> main-menu[220]: INFO: Falling back to the package description for brltty-udeb
> main-menu[220]: INFO: Falling back to the package description for brltty-udeb
> main-menu[220]: INFO: Menu item 'netcfg' selected
> netcfg[3865]: INFO: Starting netcfg v.1.160
> netcfg[3865]: WARNING **: Couldn't read Wpasupplicant pid file, not trying to kill.
> netcfg[3865]: INFO: Could not find valid BOOTIF= entry in /proc/cmdline
> netcfg[3865]: INFO: Taking down interface enp0s14
> netcfg[3865]: INFO: Taking down interface lo
> netcfg[3865]: INFO: Activating interface enp0s14
> netcfg[3865]: DEBUG: State is now 0
> netcfg[3865]: DEBUG: Want link on enp0s14
> kernel: [ 379.008021] 8139too 0000:00:0e.0 enp0s14: link up, 100Mbps, full-duplex, lpa 0xC5E1
> netcfg[3865]: INFO: Waiting time set to 3
> netcfg[3865]: INFO: ethtool-lite: enp0s14: carrier up
> netcfg[3865]: INFO: Found link on enp0s14
> netcfg[3865]: DEBUG: Commencing network autoconfiguration on enp0s14
> netcfg[3865]: DEBUG: rdnssd started; PID: 3875
> netcfg[3865]: DEBUG: nc_v6_interface_configured(enp0s14, scope local)
> netcfg[3865]: DEBUG: Running ip addr show enp0s14 to look for address
> netcfg[3865]: DEBUG: ip line: 2: enp0s14: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast qlen 1000
> netcfg[3865]: DEBUG: ip line: link/ether 54:e6:fc:82:3f:be brd ff:ff:ff:ff:ff:ff
> netcfg[3865]: DEBUG: ip line: inet6 fe80::56e6:fcff:fe82:3fbe/64 scope link
> netcfg[3865]: DEBUG: Configured address found
> netcfg[3865]: DEBUG: ip line: valid_lft forever preferred_d_lft forever
I don't have any netcfg lines in either syslog.
I also booted the target computer to a Debian 10 live CD. It found the
network and networking worked just find.
Conclusion: between our two known good systems and the one I'm trying
to install on, the installer detects the hardware with lspci. At run
time, the kernel detects it but does not know what to do with it.
Installer kernel bug?
--
Does anybody read signatures any more?
https://charlescurley.com
https://charlescurley.com/blog/
Reply to: