Booting RPi4 from USB MSD by 5.10

[Ryutaroh Matsumoto]

Dear Nicolas,

Thank you very much for your attention.

>> CONFIG_RESET_RASPBERRYPI=m is insufficient.
> Agree I'll prepare a fix.

Later I found that loading reset-raspberrypi.ko by initramfs
fixes this symptom with Debian kernel. On Debian,
I thought another package "raspi-firmware" has repsonsibility to
load reset-raspberrypi.ko by initramfs, and reassgined
Debian bug #977694 to the raspi-firmware package. A related merge request
is at https://salsa.debian.org/debian/raspi-firmware/-/merge_requests/21

>> and almost 100% of CPU usage is consumed by udev, dbus and udisks2
>> combined. I wonder if anyone other than me has experienced the same issue.
>> Bug report was filed to Debian as
>> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=980980
> 
> I'll investigate.

I DID OBSERVE this symptom by my compiled kernel, whose config and binary
are
http://153.240.174.134:64193/kernel-deb-5.9/config-arm64-5.10.10.txt
http://153.240.174.134:64193/kernel-deb-5.9/linux-image-5.10.10raspi4usb_5.10.10raspi4usb-1_arm64.deb

But, with Debian stock kernel package
linux-image-5.10.0-2-arm64 (5.10.9-1)
with reset-raspberrypi.ko and raspberrypi-cpufreq.ko loaded by initramfs
I DO NOT observe the symptom.  So I closed the corresponding Debian bug #980980.

I am happy to provide extra information when necessary.

Best regards, Ryutaroh

From: Nicolas Saenz Julienne <nsaenzjulienne at suse.de>
Subject: Re: Booting RPi4 from USB MSD by 5.10
Date: Thu, 28 Jan 2021 18:28:42 +0100

> Hi Ryutaroh, thanks for the reports!
> 
> On Wed, 2021-01-27 at 12:25 +0900, Ryutaroh Matsumoto wrote:
>> Hi all,
>>
>> Just FYI, CONFIG_RESET_RASPBERRYPI=y seems necessary
>> when RPi4 is booted from USB MSD by the upstream kernel 5.10.x
>> and CONFIG_USB_XHCI_PCI=m.
>> CONFIG_RESET_RASPBERRYPI=m is insufficient.
> 
> Agree I'll prepare a fix.
> 
>> By the way, when 5.10 kernel is booted from USB MSD and udisks2 is
>> installed, kernel emits flood of false media change messages to udev,
>> and almost 100% of CPU usage is consumed by udev, dbus and udisks2
>> combined. I wonder if anyone other than me has experienced the same issue.
>> Bug report was filed to Debian as
>> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=980980
> 
> I'll investigate.
> 
> Regards,
> Nicolas
>