Remove EISA support from main kernel image and make it M (modular)

[A. Eibach]

(This is a sort of RfC. )

I have a PC that still has the good old PCI ports, but I am 100% sure that the active use of EISA dates back yet more 10 years.
In other words, boards that _only_ support EISA date back 25 years now, while those that provide *one* EISA slot date back roughly 15 years (one of the Elitegroup K7 series comes to mind, from about 2001/02).

So the EISA support *should* be provided, as we should even support ancient hardware (Linux principle :))
However, why the EISA support is *compiled in* is beyond me.

For instance, it creates lots of totally unnecessary noise in dmesg:

[ 0.080000] EISA bus registered
[ 1.598316] platform eisa.0: Probing EISA bus 0
[ 1.598345] platform eisa.0: Cannot allocate resource for EISA slot 1 (repeated 10 times, i=i+1)
[ 1.598368] platform eisa.0: EISA: Detected 0 cards

For a 1997 mainboard, this would make perfect sense.
But it is very probable that said mainboard would require linux-image-extra nonetheless to support some days-of-yore chipset.

So why not leave it as a *module*, while removing its support from main kernel image once and for all?

Your turn.

[Brad Figg (brad-figg)]

This bug is missing log files that will aid in diagnosing the problem. From a terminal window please run:

apport-collect 1543919

and then change the status of the bug to 'Confirmed'.

If, due to the nature of the issue you have encountered, you are unable to run this command, please add a comment stating that fact and change the bug status to 'Confirmed'.

This change has been made by an automated script, maintained by the Ubuntu Kernel Team.

[A. Eibach (andi3)]

Why "Incomplete"? There's nothing to supply. I simply need some maintainer's/kernel team's/iuser's feedback on this.
Funny enough that even if I muzzled that bot, "New" still gets auto-changed to "Incomplete", this time even without an explanation.
Brad, let your bots behave a little more nicely. Yes, _still_.

[Jack Cook (twosheds)]

Hi all,
Noticed this in my Ubuntu 19.04 dmesg:

[ 2.094782] platform eisa.0: Probing EISA bus 0
[ 2.094783] platform eisa.0: EISA: Cannot allocate resource for mainboard
[ 2.094784] platform eisa.0: Cannot allocate resource for EISA slot 1
[ 2.094785] platform eisa.0: Cannot allocate resource for EISA slot 2
[ 2.094786] platform eisa.0: Cannot allocate resource for EISA slot 3
[ 2.094787] platform eisa.0: Cannot allocate resource for EISA slot 4
[ 2.094788] platform eisa.0: Cannot allocate resource for EISA slot 5
[ 2.094789] platform eisa.0: Cannot allocate resource for EISA slot 6
[ 2.094790] platform eisa.0: Cannot allocate resource for EISA slot 7
[ 2.094791] platform eisa.0: Cannot allocate resource for EISA slot 8
[ 2.094792] platform eisa.0: EISA: Detected 0 cards

First time I've ever seen this on Ubuntu or any other distributions. My motherboard is an Asus Crosshair V Formula - not that new but not that old also, I think from 2011 or there abouts.

[eresquigal (eresquigal)]

Maybe there is a kernel option to avoid scanning for EISA buses ?

[Mikko Rantalainen (mira)]

Still warning messages emitted to kernel log each boot.

Ubuntu 20.04.1 LTS
5.4.0-42-generic #46-Ubuntu SMP Fri Jul 10 00:24:02 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux

I'm running over 10 year old systems and I don't need built-in support for EISA. Does anybody even know anybody running old enough hardware to require built-in support? Does anybody even know a use case where a module wouldn't be enough?

[Bram Stolk (b-stolk)]

Still an issue with latest kernel on 20.10

# uname -a
Linux Workstation 5.8.0-43-generic #49-Ubuntu SMP Fri Feb 5 03:01:28 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux

[B. C. Schmerker (bcschmerker)]

I've an emachines®/acer® EL1210-09 (Advance Micro Devices Inc. Athlon LE-1620 (P/N ADH1620IAA5DH); nVIDIA® nForce® 780a SLI with planar C77 a2 GPU; nVIDIA® GF119 a1 GPU on PCIe 2.0 x16 slot). The system DMesg predictably detects no EISA resources on probing platform eisa.0, so EISA.c and related headers can be compiled as a module with no effect on my system's kernel initialization. AFAIK, Intel PCI has replaced EISA on all x86 and x86_64 systems built after 2000. The following redact from /var/log/dmesg.0 on 17 July 2021 illustrates typical info:

[ 0.000000] kernel: Linux version 5.8.0-59-generic (buildd@lcy01-amd64-022) (gcc (Ubuntu 9.3.0-17ubuntu1~20.04) 9.3.0, GNU ld (GNU Binutils for Ubuntu) 2.34) #66~20.04.1-Ubuntu SMP Thu Jun 17 11:14:10 UTC 2021 (Ubuntu 5.8.0-59.66~20.04.1-generic 5.8.18)
[ 0.000000] kernel: Command line: BOOT_IMAGE=/vmlinuz-5.8.0-59-generic root=UUID=c4248af1-a4cf-49a2-ba18-4de2c65c815c ro acpi_force_32bit_fadt_addr acpi_rev_override check_enable_amd_mmconf pcie_bus_tune=off pcie_aspm=off pci_ports=native intremap=off agp=off nohugeiomap parport=0 hashdist=0
...
[ 0.212298] kernel: EISA bus registered
...
[ 1.006643] kernel: platform eisa.0: Probing EISA bus 0
[ 1.006715] kernel: platform eisa.0: EISA: Cannot allocate resource for mainboard
[ 1.006779] kernel: platform eisa.0: Cannot allocate resource for EISA slot 1
[ 1.006855] kernel: platform eisa.0: Cannot allocate resource for EISA slot 8
[ 1.006917] kernel: platform eisa.0: EISA: Detected 0 cards

I concur on the feasibility of offloading EISA from Kernel core and building a Module EISA.ko.