Bug#1028547: Firmware is not udpkg -i before hardware detection

[Jamie Wilkinson]

I think I've figured out why we see these errors.

I restarted an Expert mode installation with the latest nightly DI
https://cdimage.debian.org/cdimage/daily-builds/daily/arch-latest/amd64/iso-cd/
dated
2023-03-07 04:13 size 758M

I ran through the installation to "load installer components from
installation media", then opened the terminal on tty2.

I changed into /cdrom/firmware and used `udpkg -i *.deb` to install
everything. There were many warnings about missing tools which are
now lost off the console.

I manually ran `modprobe mt7921e` and returned without error. `dmesg`
has the following in the tail:

mt7921e: firmware: direct-loading firmware
mediatek/WIFI_MT7961_patch_mcu_1_2_hdr.bin
mt7921e: HW/SW Version: 0x8a108a10, Build Time: 20230117170855a
mt7921e: firmware: direct-loading firmware mediatek/WIFI_RAM_CODE_MT7961_1.bin
mt7921e: HW/SW Version: ____010000, Build Time: 20230117170942
mt7921e: firmware: direct-loading firmware mediatek/WIFI_RAM_CODE_MT7961_1.bin
mt7921e: wlp3s0: renamed from wlan0

I then returned to the menu on tty1 and detected network hardware, and
the installer now sees the WiFi interface.

So I think the problem here is that the firmware udebs are not
unpacked before hardware detection starts.

[Cyril Brulebois]

Jamie Wilkinson <j...@debian.org> (2023-03-07):

> I think I've figured out why we see these errors.

I don't think you did.

> So I think the problem here is that the firmware udebs are not
> unpacked before hardware detection starts.

Those are firmware debs, not firmware udebs. And no, they shouldn't be

unpacked before hardware detection starts.

See my previous answer for the details.

Attaching a full syslog would help figure out what happened *if*
there's indeed something unexpected there.


Cheers,
--
Cyril Brulebois (ki...@debian.org) <https://debamax.com/>
D-I release manager -- Release team member -- Freelance Consultant

[Jamie Wilkinson]

With respect, you didn't read the first entry in the bug.  There is a syslog attached, and my syslog is not adding new information.

The mt7912e driver panics on rmmod and the firmware is never loaded.  The first part of the bug describes this.  I attempted this workaround in order to get the firmware available to the module before it performed hardware detection, because otherwise the installer hangs.

Perhaps that information is useful to you to understand the problem?  Perhaps it is useful to other users who encounter this bug.

I'm not trying to tell you what the fix is, I'm trying to give you additional information to assist in understanding the bug that (so far) two users with a Mediatek MT9721e have encountered.  I am not a d-i expert, I only know what I know, and this is the best I could do to work around the driver crash and thus installer hang.

I would appreciate you not replying with snide remarks in response to every paragraph in the bug report.   You don't need to repeat yourself either; I saw your previous update.  This is not a competition, I am only trying to offer you assistance.