Bug#1030636: Debian Installer complains about missing firmware in ath10k, even when using image with firmware included

[Alexander Dalm]

Hi Cyril,

> That's not what it does. It is there to determine whether it's a bad
> idea to reload a module. That's quite a different story… And no, between
> two mainloop iterations, that part wouldn't change anyway.

indeed.
I thought the device would be configured after upnics at the beginning or
reloading the module + upnics.
sorry for the inconvenience :)

> If we maintain a module/file mapping, we could:
>  - decide which modules need to be reloaded, instead of iterating on all
>of them (that's part of the reason why I had this idea in mind in the
>first place, looking around how to “reload dance” was implemented:
>walking through all modules unconditionally);
>  - decide that a module is “good to go” as soon as it's been reloaded
>once, i.e. some of the files it requested have been found.

The needed files for my WiFi module BCM43340 / CYW43340 / AP6234 are
basically a worst case scenario:
brcmfmac43340-sdio.bin -> required firmware file
(reload)
brcmfmac43340-sdio.txt -> required NVRAM file
(reload)
brcmfmac43340-sdio.clm_blob -> optional and not available for 43340

Depending on the Kernel version the brcmfmac module requests each file
first as
brcmfmac43340-sdio.[ManufacturerName]-[ProductName].bin / .txt / .clm_blob
and then as fallback:
brcmfmac43340-sdio.bin / .txt / .clm_blob
-> Debian 11.6: only .txt in [ManufacturerName]-[ProductName] format
-> Debian 12: all three files in [ManufacturerName]-[ProductName] format

My manufacturer was kinda lazy and left the ManufacturerName in BIOS as "To
be filled by O.E.M"
-> brcmfmac43340-sdio.To be filled by O.E.M.-Z83.bin
(this is where the spaces came from)

There will be some more prominent devices like the Raspberry Pi 3A / 3B /
4B / M2 / M3 affected with the new Debian version,
here's the file list of the firmware-brcm80211 package:
https://packages.debian.org/bullseye/all/firmware-brcm80211/filelist

I guess it's best to keep the reload dance.


Am Mo., 6. Feb. 2023 um 10:38 Uhr schrieb Cyril Brulebois :

> Alexander Dalm  (2023-02-06):
> > But there is actually a function nic_is_configured() that is used to
> > determine if the module has to be removed and reloaded after files are
> > installed.
>
> That's not what it does. It is there to determine whether it's a bad
> idea to reload a module. That's quite a different story… And no, between
> two mainloop iterations, that part wouldn't change anyway.
>
> > This function technically could be used before asking, so either:
> > a) optional files wouln't be displayed at all
> > b) optional files would be marked as optional
> > c) all files should be displayed regardless
> >
> > I think Cyril will step in to answer this question ;)
>
> See other reply.
>
>
> Cheers,
> --
> Cyril Brulebois (k...@debian.org)
> D-I release manager -- Release team member -- Freelance Consultant
>

Bug#1030636: Debian Installer complains about missing firmware in ath10k, even when using image with firmware included

[Brandon Werner]

On Mon, Feb 6, 2023, at 7:34 AM, Cyril Brulebois wrote:
>
> That's not to say I won't try and get a tentative patch before then…
> Would you be happy to test a custom netinst that I would build locally
> (i.e. outside cdimage.debian.org and the official infrastructure)?

I'd be happy to test an image you build locally on your machine to streamline 
the testing process. 

Bug#1030636: Debian Installer complains about missing firmware in ath10k, even when using image with firmware included

[Cyril Brulebois]

Brandon Werner  (2023-02-06):
> This makes sense, and both solutions seem like they would work. It
> seems like the second solution of keeping the unconditional reload and
> testing if the list of files was smaller after the reload would be
> easier to implement for Bookworm, but I think you and the rest of the
> installer team are better informed to make that decision. :)

Heh. :)

> I think assuming the module is working if the list of requested files
> is smaller after a reload is a fairly safe bet for network hardware,
> but If the installer team implemented either of these solutions, I
> could test on a bunch of old and new machines that are available at a
> computer club I attend. I unfortunately don't feel like I personally
> understand the installer well enough to fix this properly, and any
> merge request I would create would be a sad hack. Hopefully many folks
> will be testing Alpha 2 of Bookworm as well, to find any problems that
> would result from a change like this.

To be honest, there were a lot of changes already, and I'm not sure I'll
be able to squeeze that in before the next release (to be confirmed but
we were aiming at a release next week-end, and we still have to upload
d-i and see whether it becomes eligible for testing…). The safest bet at
this point would be to release with what's in the archive (basically
what you tested), warn against this specific issue (that wouldn't be a
regression from previous releases anyway), and mention it should go away
in the following release.

That's not to say I won't try and get a tentative patch before then…
Would you be happy to test a custom netinst that I would build locally
(i.e. outside cdimage.debian.org and the official infrastructure)?


Cheers,
-- 
Cyril Brulebois (k...@debian.org)
D-I release manager -- Release team member -- Freelance Consultant


signature.asc
Description: PGP signature

Bug#1030636: Debian Installer complains about missing firmware in ath10k, even when using image with firmware included

[Brandon Werner]

On Mon, Feb 6, 2023, at 5:33 AM, Cyril Brulebois wrote:
> Hi Brandon,
>
> Brandon Werner  (2023-02-06):
>> Thanks for your response. I have included the syslog lines from the
>> installer log you requested.
>
> OK, that's basically what I thought was happening, but it's always a
> good idea to check a hypothesis before deciding what to do about it. :)
>
> I think I might have mentioned the following in some discussion, or
> in some commit, but basically:
>  - we have a list of requested files;
>  - we have a list of requesting modules;
>  - some modules get reloaded.
>
> If we maintain a module/file mapping, we could:
>  - decide which modules need to be reloaded, instead of iterating on all
>of them (that's part of the reason why I had this idea in mind in the
>first place, looking around how to “reload dance” was implemented:
>walking through all modules unconditionally);
>  - decide that a module is “good to go” as soon as it's been reloaded
>once, i.e. some of the files it requested have been found.
>
> The second part would make the point about “required” vs. “optional”
> firmware moot, and would prevent extra dialogs. One could argue that
> maybe some other *.deb somewhere could be more recent or could have
> extra files, but then we don't implement anything when it comes to
> multiplicity anyway, so that wouldn't be a regression.
>
> Alternatively, we could keep the unconditionally reload dance, while
> still keeping track of files requested by each module over time.
> When the list gets smaller, its files start getting ignored.
>
>
> Does that make sense to you?
This makes sense, and both solutions seem like they would work. It seems like 
the second solution of keeping the unconditional reload and testing if the list 
of files was smaller after the reload would be easier to implement for 
Bookworm, but I think you and the rest of the installer team are better 
informed to make that decision. :)
I think assuming the module is working if the list of requested files is 
smaller after a reload is a fairly safe bet for network hardware, but If the 
installer team implemented either of these solutions, I could test on a bunch 
of old and new machines that are available at a computer club I attend. I 
unfortunately don't feel like I personally understand the installer well enough 
to fix this properly, and any merge request I would create would be a sad hack. 
Hopefully many folks will be testing Alpha 2 of Bookworm as well, to find any 
problems that would result from a change like this.

Bug#1030636: Debian Installer complains about missing firmware in ath10k, even when using image with firmware included

[Cyril Brulebois]

Hi Brandon,

Brandon Werner  (2023-02-06):
> Thanks for your response. I have included the syslog lines from the
> installer log you requested.

OK, that's basically what I thought was happening, but it's always a
good idea to check a hypothesis before deciding what to do about it. :)

I think I might have mentioned the following in some discussion, or
in some commit, but basically:
 - we have a list of requested files;
 - we have a list of requesting modules;
 - some modules get reloaded.

If we maintain a module/file mapping, we could:
 - decide which modules need to be reloaded, instead of iterating on all
   of them (that's part of the reason why I had this idea in mind in the
   first place, looking around how to “reload dance” was implemented:
   walking through all modules unconditionally);
 - decide that a module is “good to go” as soon as it's been reloaded
   once, i.e. some of the files it requested have been found.

The second part would make the point about “required” vs. “optional”
firmware moot, and would prevent extra dialogs. One could argue that
maybe some other *.deb somewhere could be more recent or could have
extra files, but then we don't implement anything when it comes to
multiplicity anyway, so that wouldn't be a regression.

Alternatively, we could keep the unconditionally reload dance, while
still keeping track of files requested by each module over time.
When the list gets smaller, its files start getting ignored.


Does that make sense to you?


Cheers,
-- 
Cyril Brulebois (k...@debian.org)
D-I release manager -- Release team member -- Freelance Consultant


signature.asc
Description: PGP signature

Bug#1030636: Debian Installer complains about missing firmware in ath10k, even when using image with firmware included

[Brandon Werner]

On Mon, Feb 6, 2023, at 4:36 AM, Cyril Brulebois wrote:
> Hi Brandon,
>
> Brandon Werner  (2023-02-05):
>> I saw the recent work to the installer surrounding firmware handling
>> and thought I would test on my machines to see how this all was
>> working. I used one of the daily sid_d-i netinst cds including
>> firmware. I noticed some problems around the installer asking for
>> firmware that was not neded for ath10k. I first tried on a laptop that
>> had QCA-6174 hw2.1. I noticed the prompt telling me about missing
>> firmware and asking if I wanted to load it from additional media,
>> which was puzzling for the firmware image. If I select no, the
>> installer continues, however, I thought this  could confuse users, so
>> I dug into it.
>
> Thanks for the tests and the report.
>
>> Before firmware atheros was loaded, the kernel tried to load versions
>> 6 through 2 of the firmware files as well as calibration firmware
>> files. After firmware atheros was installed, the card was brought up,
>> and this time, only three files were missing. The cal and pre-cal
>> files appear to be optional according to the driver source, and do not
>> exist in linux-firmware upstream, so I think them missing is no
>> problem. Firmware ver 6 doesn't exist yet in the upstream Linux repo
>> so maybe this is in the driver for future use? I guess the installer
>> still thinks there is missing firmware because of the kernel failing
>> to load these 3 unnecessary files. After version 5 of the firmware was
>> found, the kernel stopped trying to load versions 4 3 2, so there was
>> many fewer missing files on the second run of check-missing-firmware.
>
> We would need to see more of your log file. It starts with mainloop
> iteration #1, while the first check_missing call has happened already.

Thanks for your response. I have included the  syslog lines from the installer 
log you requested.

Feb  5 10:35:26 check-missing-firmware: looking at dmesg for the first time
Feb  5 10:35:26 check-missing-firmware: saving timestamp for a later use: [   
57.345819]
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/pre-cal-pci-:01:00.0.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/pre-cal-pci-:01:00.0.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/cal-pci-:01:00.0.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/cal-pci-:01:00.0.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/QCA6174/hw2.1/firmware-6.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/QCA6174/hw2.1/firmware-6.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/QCA6174/hw2.1/firmware-5.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/QCA6174/hw2.1/firmware-5.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/QCA6174/hw2.1/firmware-4.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/QCA6174/hw2.1/firmware-4.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/QCA6174/hw2.1/firmware-3.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/QCA6174/hw2.1/firmware-3.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/QCA6174/hw2.1/firmware-2.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: looking for firmware file 
ath10k/QCA6174/hw2.1/firmware-2.bin requested by ath10k_pci
Feb  5 10:35:26 check-missing-firmware: missing firmware files 
(ath10k/pre-cal-pci-:01:00.0.bin ath10k/pre-cal-pci-:01:00.0.bin 
ath10k/cal-pci-:01:00.0.bin ath10k/cal-pci-:01:00.0.bin 
ath10k/QCA6174/hw2.1/firmware-6.bin ath10k/QCA6174/hw2.1/firmware-6.bin 
ath10k/QCA6174/hw2.1/firmware-5.bin ath10k/QCA6174/hw2.1/firmware-5.bin 
ath10k/QCA6174/hw2.1/firmware-4.bin ath10k/QCA6174/hw2.1/firmware-4.bin 
ath10k/QCA6174/hw2.1/firmware-3.bin ath10k/QCA6174/hw2.1/firmware-3.bin 
ath10k/QCA6174/hw2.1/firmware-2.bin ath10k/QCA6174/hw2.1/firmware-2.bin) for 
ath10k_pci
Feb  5 10:35:26 check-missing-firmware: mainloop iteration #1
Feb  5 10:35:26 check-missing-firmware: lookup with 
/cdrom/firmware/Contents-firmware
Feb  5 10:35:26 check-missing-firmware: installing firmware package 
/cdrom/firmware/firmware-atheros_20221214-5_all.deb (non-free-firmware)
Feb  5 10:35:28 check-missing-firmware: removing and loading kernel module 
ath10k_pci
Feb  5 10:35:28 kernel: [   60.711599] DMAR: DRHD: handling fault status reg 2
Feb  5 10:35:28 kernel: [   60.711605] DMAR: [DMA Write NO_PASID] Request 
device [01:00.0] fault addr 0x7ee0 [fault 

Bug#1030636: Debian Installer complains about missing firmware in ath10k, even when using image with firmware included

[Cyril Brulebois]

Alexander Dalm  (2023-02-06):
> But there is actually a function nic_is_configured() that is used to
> determine if the module has to be removed and reloaded after files are
> installed.

That's not what it does. It is there to determine whether it's a bad
idea to reload a module. That's quite a different story… And no, between
two mainloop iterations, that part wouldn't change anyway.

> This function technically could be used before asking, so either:
> a) optional files wouln't be displayed at all
> b) optional files would be marked as optional
> c) all files should be displayed regardless
> 
> I think Cyril will step in to answer this question ;)

See other reply.


Cheers,
-- 
Cyril Brulebois (k...@debian.org)
D-I release manager -- Release team member -- Freelance Consultant


signature.asc
Description: PGP signature

Bug#1030636: Debian Installer complains about missing firmware in ath10k, even when using image with firmware included

[Cyril Brulebois]

Hi Brandon,

Brandon Werner  (2023-02-05):
> I saw the recent work to the installer surrounding firmware handling
> and thought I would test on my machines to see how this all was
> working. I used one of the daily sid_d-i netinst cds including
> firmware. I noticed some problems around the installer asking for
> firmware that was not neded for ath10k. I first tried on a laptop that
> had QCA-6174 hw2.1. I noticed the prompt telling me about missing
> firmware and asking if I wanted to load it from additional media,
> which was puzzling for the firmware image. If I select no, the
> installer continues, however, I thought this  could confuse users, so
> I dug into it.

Thanks for the tests and the report.

> Before firmware atheros was loaded, the kernel tried to load versions
> 6 through 2 of the firmware files as well as calibration firmware
> files. After firmware atheros was installed, the card was brought up,
> and this time, only three files were missing. The cal and pre-cal
> files appear to be optional according to the driver source, and do not
> exist in linux-firmware upstream, so I think them missing is no
> problem. Firmware ver 6 doesn't exist yet in the upstream Linux repo
> so maybe this is in the driver for future use? I guess the installer
> still thinks there is missing firmware because of the kernel failing
> to load these 3 unnecessary files. After version 5 of the firmware was
> found, the kernel stopped trying to load versions 4 3 2, so there was
> many fewer missing files on the second run of check-missing-firmware.

We would need to see more of your log file. It starts with mainloop
iteration #1, while the first check_missing call has happened already.

> I have another laptop with hw3.2 of QCA-6174 and on that machine, only
> pre-cal and cal are missing after firmware-atheros is loaded by the
> installer. I looked at hw-detect, and noticed there was a section in
> check-missing-firmware.sh ignoreing intel wifi debugging firmware,

That one was obvious enough to ignore. But how are we supposed to tell
which one(s) of the *68* files iwlwifi requests is(are) needed?

> but I think trying to ignore all the correct files in that location
> might be a bit tricky, especially if other net drivers try to load
> optional firmware. It also seems possible that the PCI IDs searched by
> the driver could be different for cal and pre-cal for different ath10k
> hardware although I didn't dig into this. I hope the information I
> provided is enough for package maintainers to determine a correct
> solution. Thanks for all the great work on the installer recently to
> make  firmware handling work better.

I don't think there's a way to identify what's needed and what's
optional. We can't even determine whether firmware is actually needed!
(e.g. my rtw8192cu requests firmware but works fine without it.)

> Below, a bit of text from the installer log, to show the driver is
> loading, but the installer still thinking there is missing firmware.

Please include at least all lines between the first and the last
check-missing-firmware lines.


Cheers,
-- 
Cyril Brulebois (k...@debian.org)
D-I release manager -- Release team member -- Freelance Consultant


signature.asc
Description: PGP signature

Bug#1030636: Debian Installer complains about missing firmware in ath10k, even when using image with firmware included

[Alexander Dalm]

Hi Brandon,

good point.
currently hw-detect / check-missing-firmware doesn't checks if a specific
firmware file is optional or required, I even considered asking for the
optional ones as well as a feature.
But there is actually a function nic_is_configured() that is used to
determine if the module has to be removed and reloaded after files are
installed.
This function technically could be used before asking, so either:
a) optional files wouln't be displayed at all
b) optional files would be marked as optional
c) all files should be displayed regardless

I think Cyril will step in to answer this question ;)

I will do some testing to check if nic_is_configured() is reliable enough
for this task by feeding the files one by one.

Bug#1030636: Debian Installer complains about missing firmware in ath10k, even when using image with firmware included

[Brandon Werner]

package: hw-detect
version: 1.154

Hello,
I saw the recent work to the installer surrounding firmware handling and 
thought I would test on my machines to see how this all was working. I used one 
of the daily sid_d-i netinst cds including firmware. I noticed some problems 
around the installer asking for  firmware that was not neded for ath10k. I 
first tried on a laptop that had QCA-6174 hw2.1. I noticed the prompt telling 
me about missing firmware and asking if I wanted to load it from additional 
media, which was puzzling for the firmware image. If I select no, the installer 
continues, however, I thought this  could confuse users, so I dug into it.

Before firmware atheros was loaded, the kernel tried to load versions 6 through 
2 of the firmware files as well as calibration firmware files. After firmware 
atheros was installed, the card was brought up, and this time, only three files 
were missing. The cal and pre-cal files appear to be optional according to the 
driver source, and do not exist in linux-firmware upstream, so I think them 
missing is no problem. Firmware ver 6 doesn't exist yet in the upstream Linux 
repo so maybe this is in the driver for future use? I guess the installer still 
thinks there is missing firmware because of the kernel failing to load these 3 
unnecessary files. After version 5 of the firmware was found, the kernel 
stopped trying to load versions 4 3 2, so there was many fewer missing files on 
the second run of check-missing-firmware.

I have another laptop with hw3.2 of QCA-6174 and on that machine, only pre-cal 
and cal are missing after firmware-atheros is loaded by the installer. I looked 
at hw-detect, and noticed there was a section in check-missing-firmware.sh 
ignoreing intel wifi debugging firmware, but I think trying to ignore all the 
correct files in that location might be a bit tricky, especially if other net 
drivers try to load optional firmware. It also seems possible that the PCI IDs 
searched by the driver could be different for cal and pre-cal for different 
ath10k hardware although I didn't dig into this. I hope the information I 
provided is enough for package maintainers to determine a correct solution. 
Thanks for all the great work on the installer recently to make  firmware 
handling work better.

Below, a bit of text from the installer log, to show the driver is loading, but 
the installer still thinking there is missing firmware.

Feb  5 10:35:26 check-missing-firmware: mainloop iteration #1
Feb  5 10:35:26 check-missing-firmware: lookup with 
/cdrom/firmware/Contents-firmware
Feb  5 10:35:26 check-missing-firmware: installing firmware package 
/cdrom/firmware/firmware-atheros_20221214-5_all.deb (non-free-firmware)
Feb  5 10:35:28 check-missing-firmware: removing and loading kernel module 
ath10k_pci
Feb  5 10:35:28 kernel: [   60.711599] DMAR: DRHD: handling fault status reg 2
Feb  5 10:35:28 kernel: [   60.711605] DMAR: [DMA Write NO_PASID] Request 
device [01:00.0] fault addr 0x7ee0 [fault reason 0x05] PTE Write access is 
not set
Feb  5 10:35:28 kernel: [   60.711661] ath10k_pci :01:00.0: pci irq msi 
oper_irq_mode 2 irq_mode 0 reset_mode 0
Feb  5 10:35:28 kernel: [   60.977712] ath10k_pci :01:00.0: firmware: 
failed to load ath10k/pre-cal-pci-:01:00.0.bin (-2)
Feb  5 10:35:28 kernel: [   60.977724] ath10k_pci :01:00.0: firmware: 
failed to load ath10k/pre-cal-pci-:01:00.0.bin (-2)
Feb  5 10:35:28 kernel: [   60.977735] ath10k_pci :01:00.0: firmware: 
failed to load ath10k/cal-pci-:01:00.0.bin (-2)
Feb  5 10:35:28 kernel: [   60.977744] ath10k_pci :01:00.0: firmware: 
failed to load ath10k/cal-pci-:01:00.0.bin (-2)
Feb  5 10:35:28 kernel: [   60.977755] ath10k_pci :01:00.0: firmware: 
failed to load ath10k/QCA6174/hw2.1/firmware-6.bin (-2)
Feb  5 10:35:28 kernel: [   60.977762] ath10k_pci :01:00.0: firmware: 
failed to load ath10k/QCA6174/hw2.1/firmware-6.bin (-2)
Feb  5 10:35:28 kernel: [   60.977973] ath10k_pci :01:00.0: firmware: 
direct-loading firmware ath10k/QCA6174/hw2.1/firmware-5.bin
Feb  5 10:35:28 kernel: [   60.977980] ath10k_pci :01:00.0: qca6174 hw2.1 
target 0x0501 chip_id 0x003405ff sub 144d:4125
Feb  5 10:35:28 kernel: [   60.977986] ath10k_pci :01:00.0: kconfig debug 0 
debugfs 0 tracing 0 dfs 0 testmode 0
Feb  5 10:35:28 kernel: [   60.978697] ath10k_pci :01:00.0: firmware ver 
SW_RM.1.1.1-00157-QCARMSWPZ-1 api 5 features ignore-otp,no-4addr-pad crc32 
10bf8e08
Feb  5 10:35:28 kernel: [   61.040011] ath10k_pci :01:00.0: firmware: 
direct-loading firmware ath10k/QCA6174/hw2.1/board-2.bin
Feb  5 10:35:28 kernel: [   61.040387] ath10k_pci :01:00.0: board_file api 
2 bmi_id N/A crc32 ae2e275a
Feb  5 10:35:29 check-missing-firmware: looking at dmesg again, restarting from 
timestamp: [   57.345819]
Feb  5 10:35:29 check-missing-firmware: timestamp found, truncating dmesg 
accordingly
Feb  5 10:35:29 check-missing-firmware: saving timestamp for a