On Thu, 4 Jan 2018 13:22:39 +0300
Andrei Borzenkov <arvidjaar(a)gmail.com> wrote:
04.01.2018 12:06, Paul Neuwirth пишет:
KERNEL[487838.926954]
add
/devices/pci0000:00/0000:00:0a.0/0000:01:06.0/host0/target0:0:3/0:0:3:0/block/sdc
(block) KERNEL[487838.926986]
change
/devices/pci0000:00/0000:00:0a.0/0000:01:06.0/host0/target0:0:3/0:0:3:0/block/sdc
(block) KERNEL[487838.927021]
change
/devices/pci0000:00/0000:00:0a.0/0000:01:06.0/host0/target0:0:3/0:0:3:0/block/sdc
(block) KERNEL[487838.927054]
change
/devices/pci0000:00/0000:00:0a.0/0000:01:06.0/host0/target0:0:3/0:0:3:0/block/sdc
(block) KERNEL[487939.384425]
change
/devices/pci0000:00/0000:00:0a.0/0000:01:06.0/host0/target0:0:3/0:0:3:0/block/sdc
(block) KERNEL[488039.884765]
change
/devices/pci0000:00/0000:00:0a.0/0000:01:06.0/host0/target0:0:3/0:0:3:0/block/sdc
(block) KERNEL[488039.895676]
change
/devices/pci0000:00/0000:00:0a.0/0000:01:06.0/host0/target0:0:3/0:0:3:0/block/sdc
(block)
OK, this confirms that at the end it is kernel issue. You should
really report it to kernel folks as true solution must be implemented
in kernel driver.
So there are two issues - (initial) spin up and permanent stream of
"change" events. To workaround the latter - try disabling periodic
device polling in kernel:
echo 0 > /sys/block/sdc/events_poll_msecs
and check with "udevadm monitor" whether events stop.
For the former you can pass quirk parameter to scsi_mod; create
/etc/modprobe.d/teac.conf (name is arbitrary) with:
options scsi_mod dev_flags=TEAC:FC-5:4096
Check /sys/block/sdc/device/{vendor,model} for correct values. This
flag suppresses initial spin up of device. You will need to reboot
(and do not forget you also need to recreate initrd as scsi_mod is
most likely loaded there).
Hello,
one little question: what is the concerned part of kernel?
I did a bug general bug report on the kernel mailing list. But no
response until now. Maybe I should mention the kernel
module/part/whatever in the topic, and write a copy to the maintainer.
But I do not have a clue which part it is.
Thank you again
Paul