Skip to contentSkip to conversation

udev cdrom_id rules prevent unmounted CD from spinning down

PATCH Applied Nick Bowler 8 participants 32 messages May 13, 2010 – May 25, 2010 libata

lore
Share & tools
lore
#0 lore raw

Please CC me as I am not subscribed to the list.

The cdrom_id udev rules seem to be preventing the cdrom drive on my laptop from spinning down. Upon inserting a disk, it will spin up to full speed, then the motor will turn off. Before the disk stops completely, however, it will spin right up again, then begin to stop, then spin up, then begin to stop, ad infinitum.

This only occurs whilst no filesystem is mounted off the device, and deleting /lib/udev/rules.d/60-cdrom_id.rules solves the problem, at least until the next time udev gets updated.

My laptop is a ThinkPad T500, which has a SATA cdrom drive attached to an Intel ICH9M/M-E SATA AHCI controller.

Signature · 1 line
Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)
#1 lore raw
Quoted 1 line · Nick Bowler

The cdrom_id udev rules seem to be preventing the cdrom drive on my

Which udev release are you using?

Signature · 2 lines
ciao,
Marco
#2 lore raw
In reply to Nick Bowler
Quoted 4 lines · Marco d'Itri
Quoted 1 line · Nick Bowler

The cdrom_id udev rules seem to be preventing the cdrom drive on my

Which udev release are you using?

I was using 151, but I just upgraded to 154 and the problem persists.

Signature · 1 line
Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)
#3 lore raw
In reply to Nick Bowler
Quoted 7 lines · Nick Bowler
Quoted 4 lines · Marco d'Itri
Quoted 1 line · Nick Bowler

The cdrom_id udev rules seem to be preventing the cdrom drive on my

Which udev release are you using?

I was using 151, but I just upgraded to 154 and the problem persists.

Does:
udevadm monitor
print something?

Does killing udevd changes the issue?

Kay

#4 lore raw
In reply to Nick Bowler
Quoted 4 lines · Kay Sievers
Quoted 3 lines · Nick Bowler
Quoted 2 lines · Marco d'Itri

On May 13, Nick Bowler <[email protected]> wrote:

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines
Quoted 1 line

The cdrom_id udev rules seem to be preventing the cdrom drive on my

Which udev release are you using?

I was using 151, but I just upgraded to 154 and the problem persists.

Does:
udevadm monitor
print something?

Yes (copied by hand):

KERNEL[1273786356.346839] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0 (scsi) UDEV [1273786356.352880] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0 (block) KERNEL[1273786356.357396] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0 (block) UDEV [1273786356.361771] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0 (scsi) UDEV [1273786356.595945] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0 (block) KERNEL[1273786361.162051] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0 (scsi) KERNEL[1273786361.166466] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0 (block)

and it spews out a bunch like this every time the disk spins back up.

Quoted 1 line

Does killing udevd changes the issue?

Yes, after killing udevd the disk stops spinning.

Signature · 1 line
Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)
#5 lore raw
In reply to Nick Bowler

Nick Bowler [2010-05-13 17:37 -0400]:

Quoted 2 lines

KERNEL[1273786356.346839] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0 (scsi) KERNEL[1273786361.166466] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0 (block)

Seems this is a 5 second interval. Both udisks and hal poll CD-ROMs every two seconds, so at first sight this looks slightly unrelated. But those are the two that actually generate polling events. Does the issue also go away after killing hald and udisks-daemon (but leaving udevd running)?

Martin

Signature · 2 lines
Martin Pitt                        | http://www.piware.de
Ubuntu Developer (www.ubuntu.com)  | Debian Developer  (www.debian.org)
#6 lore raw
In reply to Nick Bowler
Quoted 5 lines · Martin Pitt

Nick Bowler [2010-05-13 17:37 -0400]:

Quoted 2 lines

KERNEL[1273786356.346839] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0 (scsi) KERNEL[1273786361.166466] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0 (block)

Seems this is a 5 second interval.

I suspect it's tied to the normal spindown time for the drive (the drive does _start_ to spin down, after all, before it gets kicked back into full gear).

Quoted 5 lines

Both udisks and hal poll CD-ROMs
every two seconds, so at first sight this looks slightly unrelated. But those are the two that actually generate polling events. Does the issue also go away after killing hald and udisks-daemon (but leaving udevd running)?

Neither hald nor udisks-daemon are even installed on the machine, let alone running. This occurs whilst running _nothing_ other than udevd (as well as during normal operation):

boot with init=/bin/sh

mount -t proc proc proc
mount -t sysfs sysfs sys
mount -t tmpfs udev dev
modprobe sr-mod

(insert cd: after a few seconds it spins down normally)

  udevd --daemon
  udevadm trigger

(remove/insert cd: it never spins down; this is when I ran udevadm monitor)

killall udevd

(after a few seconds the cd spins down normally)

Signature · 1 line
Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)
#7 lore raw
In reply to Nick Bowler
Quoted 33 lines · Nick Bowler
Quoted 5 lines · Martin Pitt

Nick Bowler [2010-05-13 17:37 -0400]:

Quoted 2 lines

KERNEL[1273786356.346839] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0 (scsi) KERNEL[1273786361.166466] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0 (block)

Seems this is a 5 second interval.

I suspect it's tied to the normal spindown time for the drive (the drive does _start_ to spin down, after all, before it gets kicked back into full gear).

Quoted 5 lines

Both udisks and hal poll CD-ROMs
every two seconds, so at first sight this looks slightly unrelated. But those are the two that actually generate polling events. Does the issue also go away after killing hald and udisks-daemon (but leaving udevd running)?

Neither hald nor udisks-daemon are even installed on the machine, let alone running.  This occurs whilst running _nothing_ other than udevd (as well as during normal operation):

 boot with init=/bin/sh

 mount -t proc proc proc
 mount -t sysfs sysfs sys
 mount -t tmpfs udev dev
 modprobe sr-mod

 (insert cd: after a few seconds it spins down normally)

 udevd --daemon
 udevadm trigger

Quoted 6 lines

 (remove/insert cd: it never spins down; this is when I ran udevadm monitor)

 killall udevd

 (after a few seconds the cd spins down normally)

Sounds odd, that the kernel/the device seems to flip the media present bit, and sends events for that after every access.

Does cdrom_id work as expected otherwise? What does:
/lib/udev/cdrom_id --debug /dev/sr0
print?

Does running cdrom_id spin up the drive?

There is a call to /sbin/blkid in:
/lib/udev/rules.d/60-persistent-storage.rules
with:
KERNEL="sr*", IMPORT{program}="/sbin/blkid ...
If you comment that out, does the issue go away?

Kay

#8 lore raw
In reply to Nick Bowler
Quoted 3 lines · Kay Sievers

Does cdrom_id work as expected otherwise? What does:
/lib/udev/cdrom_id --debug /dev/sr0
print?

Output from this command is appended.

Quoted 1 line

Does running cdrom_id spin up the drive?

No, the above does not cause the drive to spin up.

Quoted 5 lines

There is a call to /sbin/blkid in:
/lib/udev/rules.d/60-persistent-storage.rules
with:
KERNEL="sr*", IMPORT{program}="/sbin/blkid ...
If you comment that out, does the issue go away?

I commented both of the following lines:

KERNEL="sr*", ENV{ID_CDROM_MEDIA}="?*", ENV{ID_CDROM_MEDIA_SESSION_LAST_OFFSET}="?*", IMPORT{program}="/sbin/blkid -o udev -p -u noraid -O $env{ID_CDROM_MEDIA_SESSION_LAST_OFFSET} $tempnode" KERNEL="sr*", ENV{ID_CDROM_MEDIA}="?*", ENV{ID_CDROM_MEDIA_SESSION_LAST_OFFSET}="", IMPORT{program}="/sbin/blkid -o udev -p -u noraid $tempnode"

and the issue indeed goes away.

# /lib/udev/cdrom_id --debug /dev/sr0
main: probing: '/dev/sr0'
cd_inquiry: INQUIRY: [Optiarc ][DVD RW AD-7910S ][1.D2]
cd_profiles: GET CONFIGURATION: size of features buffer 0x0150
cd_profiles: GET CONFIGURATION: feature 'profiles', with 14 entries
feature_profiles: profile 0x2b dvd_plus_r_dl
feature_profiles: profile 0x1b dvd_plus_r
feature_profiles: profile 0x1a dvd_plus_rw
feature_profiles: profile 0x16 <ignored>
feature_profiles: profile 0x15 <ignored>
feature_profiles: profile 0x14 dvd_rw
feature_profiles: profile 0x13 dvd_rw
feature_profiles: profile 0x12 dvd_ram
feature_profiles: profile 0x11 <ignored>
feature_profiles: profile 0x10 dvd_rom
feature_profiles: profile 0x0a cd_rw
feature_profiles: profile 0x09 cd_r
feature_profiles: profile 0x08 cd_rom
feature_profiles: profile 0x02 <ignored>
cd_profiles: GET CONFIGURATION: feature 0x0001 <ignored>, with 0x08 bytes
cd_profiles: GET CONFIGURATION: feature 0x0002 <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x0003 <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x0004 <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x0010 <ignored>, with 0x08 bytes
cd_profiles: GET CONFIGURATION: feature 0x001d <ignored>, with 0x00 bytes
cd_profiles: GET CONFIGURATION: feature 0x001e <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x001f <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x0020 <ignored>, with 0x0c bytes
cd_profiles: GET CONFIGURATION: feature 0x0021 <ignored>, with 0x08 bytes
cd_profiles: GET CONFIGURATION: feature 0x0023 <ignored>, with 0x08 bytes
cd_profiles: GET CONFIGURATION: feature 0x0024 <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x0026 <ignored>, with 0x00 bytes
cd_profiles: GET CONFIGURATION: feature 0x002a <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x002b <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x002c <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x002d <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x002e <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x002f <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x0033 <ignored>, with 0x08 bytes
cd_profiles: GET CONFIGURATION: feature 0x0037 <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x003b <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x0100 <ignored>, with 0x00 bytes
cd_profiles: GET CONFIGURATION: feature 0x0103 <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x0105 <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x0106 <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x0107 <ignored>, with 0x04 bytes
cd_profiles: GET CONFIGURATION: feature 0x0108 <ignored>, with 0x0c bytes
cd_profiles: GET CONFIGURATION: feature 0x010a <ignored>, with 0x0c bytes
cd_profiles: current profile 0x11
cd_profiles: profile 0x11 media_dvd_r
cd_media_info: disk type 00
cd_media_toc: READ TOC: len: 20
cd_media_toc: track=1 info=0x4(data) start_block=0
cd_media_toc: last track 1 starts at block 0
ID_CDROM=1
ID_CDROM_CD=1
ID_CDROM_CD_R=1
ID_CDROM_CD_RW=1
ID_CDROM_DVD=1
ID_CDROM_DVD_R=1
ID_CDROM_DVD_RW=1
ID_CDROM_DVD_RAM=1
ID_CDROM_DVD_PLUS_R=1
ID_CDROM_DVD_PLUS_RW=1
ID_CDROM_DVD_PLUS_R_DL=1
ID_CDROM_MRW=1
ID_CDROM_MRW_W=1
ID_CDROM_MEDIA=1
ID_CDROM_MEDIA_DVD_R=1
ID_CDROM_MEDIA_STATE=appendable
ID_CDROM_MEDIA_SESSION_NEXT=2
ID_CDROM_MEDIA_SESSION_COUNT=2
ID_CDROM_MEDIA_TRACK_COUNT=2
ID_CDROM_MEDIA_TRACK_COUNT_DATA=1
Signature · 1 line
Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)
#9 lore raw
In reply to Nick Bowler
Quoted 17 lines · Nick Bowler
Quoted 3 lines · Kay Sievers

Does cdrom_id work as expected otherwise? What does:
  /lib/udev/cdrom_id --debug /dev/sr0
print?

Output from this command is appended.

Quoted 1 line

Does running cdrom_id spin up the drive?

No, the above does not cause the drive to spin up.

Quoted 5 lines

There is a call to /sbin/blkid in:
  /lib/udev/rules.d/60-persistent-storage.rules
with:
  KERNEL="sr*", IMPORT{program}="/sbin/blkid ...
If you comment that out, does the issue go away?

Quoted 6 lines

I commented both of the following lines:

 KERNEL="sr*", ENV{ID_CDROM_MEDIA}="?*", ENV{ID_CDROM_MEDIA_SESSION_LAST_OFFSET}="?*", IMPORT{program}="/sbin/blkid -o udev -p -u noraid -O $env{ID_CDROM_MEDIA_SESSION_LAST_OFFSET} $tempnode" KERNEL="sr*", ENV{ID_CDROM_MEDIA}="?*", ENV{ID_CDROM_MEDIA_SESSION_LAST_OFFSET}="", IMPORT{program}="/sbin/blkid -o udev -p -u noraid $tempnode"

and the issue indeed goes away.

Hmm, weird behavior.

To make sure you don't have any other stuff accessing the device, can you please attach the output of:
udevadm test /class/block/sr0
so we can make sure it's nothing we don't think of.

To narrow down the blkid issue, if you comment out the rule, and let the drive spin down, then run:
/sbin/blkid -p -oudev /dev/sr0
manually, do you see while doing that any events generated in:
udevadm monitor
?

That would look like a bug in the device or the kernel. Because they are not supposed to flip the media-present bit when the media has not changed.

Thanks,
Kay

#10 lore raw
In reply to Nick Bowler
Quoted 4 lines · Kay Sievers

To make sure you don't have any other stuff accessing the device, can you please attach the output of:
udevadm test /class/block/sr0
so we can make sure it's nothing we don't think of.

Appended.

Quoted 6 lines

To narrow down the blkid issue, if you comment out the rule, and let the drive spin down, then run:
/sbin/blkid -p -oudev /dev/sr0
manually, do you see while doing that any events generated in:
udevadm monitor
?

The command causes the cd to spin up, but nothing is printed by udevadm monitor as a result.

# udevadm test /class/block/sr0
run_command: calling: test
udevadm_test: version 154
This program is for debugging only, it does not run any program,
specified by a RUN key. It may show incorrect results, because
some values may be different, or not available at a simulation run.

parse_file: reading '/dev/.udev/rules.d/10-root-link.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/30-kernel-compat.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/40-gentoo.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/40-usb_modeswitch.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/50-firmware.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/50-udev-default.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/60-cdrom_id.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/60-floppy.rules' as rules file
parse_file: reading '/etc/udev/rules.d/60-pcmcia.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/60-persistent-alsa.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/60-persistent-input.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/60-persistent-serial.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/60-persistent-storage-tape.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/60-persistent-storage.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/60-persistent-v4l.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/61-mobile-action.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/61-persistent-storage-edd.rules' as rules file
parse_file: reading '/etc/udev/rules.d/65-kvm.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/70-acl.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/70-hid2hci.rules' as rules file
parse_file: reading '/etc/udev/rules.d/70-libgphoto2.rules' as rules file
parse_file: reading '/etc/udev/rules.d/70-persistent-cd.rules' as rules file
parse_file: reading '/etc/udev/rules.d/70-persistent-net.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/75-cd-aliases-generator.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/75-net-description.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/75-persistent-net-generator.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/75-tty-description.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/78-sound-card.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/79-fstab_import.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/80-drivers.rules' as rules file
parse_file: reading '/etc/udev/rules.d/90-local.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/90-network.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/90-pulseaudio.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/95-keyboard-force-release.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/95-keymap.rules' as rules file
parse_file: reading '/lib64/udev/rules.d/95-udev-late.rules' as rules file
parse_file: reading '/etc/udev/rules.d/99-hsf.rules' as rules file
udev_rules_new: rules use 81948 bytes tokens (6829 * 12 bytes), 18007 bytes buffer
udev_rules_new: temporary index used 29900 bytes (1495 * 20 bytes)
udev_device_new_from_syspath: device 0x1790140 has devpath '/devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0' udev_device_new_from_syspath: device 0x17a7b30 has devpath '/devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0'
udev_device_read_db: device 0x17a7b30 filled with db file data
udev_rules_apply_to_event: LINK 'block/11:0' /lib64/udev/rules.d/50-udev-default.rules:3
udev_rules_apply_to_event: GROUP 6 /lib64/udev/rules.d/50-udev-default.rules:76
udev_rules_apply_to_event: GROUP 19 /lib64/udev/rules.d/50-udev-default.rules:82
udev_rules_apply_to_event: LINK 'scd0' /lib64/udev/rules.d/50-udev-default.rules:82
udev_rules_apply_to_event: IMPORT 'cdrom_id --export /dev/sr0' /lib64/udev/rules.d/60-cdrom_id.rules:9
util_run_program: 'cdrom_id --export /dev/sr0' started
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_CD=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_CD_R=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_CD_RW=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_DVD=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_DVD_R=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_DVD_RW=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_DVD_RAM=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_DVD_PLUS_R=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_DVD_PLUS_RW=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_DVD_PLUS_R_DL=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_MRW=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_MRW_W=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_MEDIA=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_MEDIA_CD=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_MEDIA_SESSION_COUNT=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_MEDIA_TRACK_COUNT=1'
util_run_program: '/lib64/udev/cdrom_id' (stdout) 'ID_CDROM_MEDIA_TRACK_COUNT_DATA=1'
util_run_program: 'cdrom_id --export /dev/sr0' returned with exitcode 0
udev_device_new_from_syspath: device 0x179d460 has devpath '/devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0' udev_device_new_from_syspath: device 0x179d7b0 has devpath '/devices/pci0000:00/0000:00:1f.2/host1/target1:0:0' udev_device_new_from_syspath: device 0x179dae0 has devpath '/devices/pci0000:00/0000:00:1f.2/host1'
udev_device_new_from_syspath: device 0x179de00 has devpath '/devices/pci0000:00/0000:00:1f.2'
udev_device_new_from_syspath: device 0x179e110 has devpath '/devices/pci0000:00'
udev_rules_apply_to_event: IMPORT 'scsi_id --export --whitelisted -d /dev/sr0' /lib64/udev/rules.d/60-persistent-storage.rules:28
util_run_program: 'scsi_id --export --whitelisted -d /dev/sr0' started
util_run_program: '/lib64/udev/scsi_id' (stdout) 'ID_SCSI=1'
util_run_program: '/lib64/udev/scsi_id' (stdout) 'ID_VENDOR=Optiarc'
util_run_program: '/lib64/udev/scsi_id' (stdout) 'ID_VENDOR_ENC=Optiarc\x20'
util_run_program: '/lib64/udev/scsi_id' (stdout) 'ID_MODEL=DVD_RW_AD-7910S'
util_run_program: '/lib64/udev/scsi_id' (stdout) 'ID_MODEL_ENC=DVD\x20RW\x20AD-7910S\x20'
util_run_program: '/lib64/udev/scsi_id' (stdout) 'ID_REVISION=1.D2'
util_run_program: '/lib64/udev/scsi_id' (stdout) 'ID_TYPEÍ'
util_run_program: 'scsi_id --export --whitelisted -d /dev/sr0' returned with exitcode 0
udev_rules_apply_to_event: IMPORT 'path_id /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0' /lib64/udev/rules.d/60-persistent-storage.rules:47
util_run_program: 'path_id /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0' started
util_run_program: '/lib64/udev/path_id' (stdout) 'ID_PATH=pci-0000:00:1f.2-scsi-1:0:0:0'
util_run_program: 'path_id /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0' returned with exitcode 0
udev_rules_apply_to_event: LINK 'disk/by-path/pci-0000:00:1f.2-scsi-1:0:0:0' /lib64/udev/rules.d/60-persistent-storage.rules:48
udev_rules_apply_to_event: LINK 'cdrom' /etc/udev/rules.d/70-persistent-cd.rules:8
udev_rules_apply_to_event: LINK 'cdrw' /etc/udev/rules.d/70-persistent-cd.rules:9
udev_rules_apply_to_event: LINK 'dvd' /etc/udev/rules.d/70-persistent-cd.rules:10
udev_rules_apply_to_event: LINK 'dvdrw' /etc/udev/rules.d/70-persistent-cd.rules:11
udev_event_execute_rules: no node name set, will use kernel supplied name 'sr0'
udev_device_update_db: created db file for '/devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0' in '/dev/.udev/db/block:sr0'
udev_node_add: creating device node '/dev/sr0', devnum:0, mode60, uid=0, gid
udev_node_mknod: preserve file '/dev/sr0', because it has correct dev_t
udev_node_mknod: preserve permissions /dev/sr0, 060660, uid=0, gid
node_symlink: preserve already existing symlink '/dev/block/11:0' to '../sr0'
link_find_prioritized: found 'b11:0' claiming '/dev/.udev/links/scd0'
link_update: creating link '/dev/scd0' to '/dev/sr0'
node_symlink: preserve already existing symlink '/dev/scd0' to 'sr0'
link_find_prioritized: found 'b11:0' claiming '/dev/.udev/links/disk\x2fby-path\x2fpci-0000:00:1f.2-scsi-1:0:0:0'
link_update: creating link '/dev/disk/by-path/pci-0000:00:1f.2-scsi-1:0:0:0' to '/dev/sr0'
node_symlink: preserve already existing symlink '/dev/disk/by-path/pci-0000:00:1f.2-scsi-1:0:0:0' to '../../sr0'
link_find_prioritized: found 'b11:0' claiming '/dev/.udev/links/cdrom'
link_update: creating link '/dev/cdrom' to '/dev/sr0'
node_symlink: preserve already existing symlink '/dev/cdrom' to 'sr0'
link_find_prioritized: found 'b11:0' claiming '/dev/.udev/links/cdrw'
link_update: creating link '/dev/cdrw' to '/dev/sr0'
node_symlink: preserve already existing symlink '/dev/cdrw' to 'sr0'
link_find_prioritized: found 'b11:0' claiming '/dev/.udev/links/dvd'
link_update: creating link '/dev/dvd' to '/dev/sr0'
node_symlink: preserve already existing symlink '/dev/dvd' to 'sr0'
link_find_prioritized: found 'b11:0' claiming '/dev/.udev/links/dvdrw'
link_update: creating link '/dev/dvdrw' to '/dev/sr0'
node_symlink: preserve already existing symlink '/dev/dvdrw' to 'sr0'
udevadm_test: UDEV_LOG=6
udevadm_test: DEVPATH=/devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0
udevadm_test: MAJOR
udevadm_test: MINOR=0
udevadm_test: DEVNAME=/dev/sr0
udevadm_test: DEVTYPE=disk
udevadm_test: ACTION­d
udevadm_test: SUBSYSTEM=block
udevadm_test: DEVLINKS=/dev/block/11:0 /dev/scd0 /dev/disk/by-path/pci-0000:00:1f.2-scsi-1:0:0:0 /dev/cdrom /dev/cdrw /dev/dvd /dev/dvdrw
udevadm_test: ID_CDROM=1
udevadm_test: ID_CDROM_CD=1
udevadm_test: ID_CDROM_CD_R=1
udevadm_test: ID_CDROM_CD_RW=1
udevadm_test: ID_CDROM_DVD=1
udevadm_test: ID_CDROM_DVD_R=1
udevadm_test: ID_CDROM_DVD_RW=1
udevadm_test: ID_CDROM_DVD_RAM=1
udevadm_test: ID_CDROM_DVD_PLUS_R=1
udevadm_test: ID_CDROM_DVD_PLUS_RW=1
udevadm_test: ID_CDROM_DVD_PLUS_R_DL=1
udevadm_test: ID_CDROM_MRW=1
udevadm_test: ID_CDROM_MRW_W=1
udevadm_test: ID_CDROM_MEDIA=1
udevadm_test: ID_CDROM_MEDIA_CD=1
udevadm_test: ID_CDROM_MEDIA_SESSION_COUNT=1
udevadm_test: ID_CDROM_MEDIA_TRACK_COUNT=1
udevadm_test: ID_CDROM_MEDIA_TRACK_COUNT_DATA=1
udevadm_test: ID_SCSI=1
udevadm_test: ID_VENDOR=Optiarc
udevadm_test: ID_VENDOR_ENC=Optiarc\x20
udevadm_test: ID_MODEL=DVD_RW_AD-7910S
udevadm_test: ID_MODEL_ENC=DVD\x20RW\x20AD-7910S\x20
udevadm_test: ID_REVISION=1.D2
udevadm_test: ID_TYPEÍ
udevadm_test: ID_BUS=scsi
udevadm_test: ID_PATH=pci-0000:00:1f.2-scsi-1:0:0:0
udevadm_test: TAGS=:udev-acl:
udevadm_test: GENERATED=1

Signature · 1 line
Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)
#11 lore raw
In reply to Nick Bowler
Quoted 7 lines · Nick Bowler
Quoted 4 lines · Kay Sievers

To make sure you don't have any other stuff accessing the device, can you please attach the output of:
  udevadm test /class/block/sr0
so we can make sure it's nothing we don't think of.

Appended.

Loos all fine and with the usual behavior so far.

Quoted 9 lines
Quoted 6 lines

To narrow down the blkid issue, if you comment out the rule, and let the drive spin down, then run:
  /sbin/blkid -p -oudev /dev/sr0
manually, do you see while doing that any events generated in:
  udevadm monitor
?

The command causes the cd to spin up, but nothing is printed by udevadm monitor as a result.

Ok, and if you enable the blkid rules again, then you see events in the monitor?

Thanks,
Kay

#12 lore raw
In reply to Nick Bowler
Quoted 19 lines · Kay Sievers
Quoted 7 lines · Nick Bowler
Quoted 4 lines · Kay Sievers

To make sure you don't have any other stuff accessing the device, can you please attach the output of:
udevadm test /class/block/sr0
so we can make sure it's nothing we don't think of.

Appended.

Loos all fine and with the usual behavior so far.

Quoted 7 lines
Quoted 6 lines

To narrow down the blkid issue, if you comment out the rule, and let the drive spin down, then run:
/sbin/blkid -p -oudev /dev/sr0
manually, do you see while doing that any events generated in:
udevadm monitor
?

Quoted 5 lines
Quoted 2 lines

The command causes the cd to spin up, but nothing is printed by udevadm monitor as a result.

Ok, and if you enable the blkid rules again, then you see events in the monitor?

Sorry, the above statement that udevadm monitor prints nothing was a lie! (I must have been insane and/or lacking in coffee first thing in the morning).

With the rules commented out, the command /sbin/blkid -p -oudev /dev/sr0 on the non-spinning drive causes three distinct bursts of events to occur:

(a) First, immediately after the disk spins up (and blkid prints stuff) KERNEL[1274136271.742645] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0 (scsi) KERNEL[1274136271.743204] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0 (block) UDEV [1274136271.743255] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0 (scsi) UDEV [1274136271.792800] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0 (block)

(b) Then, after a few seconds the disk spins down, but it's still humming:
KERNEL[1274136276.306739] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0 (scsi) KERNEL[1274136276.307182] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0 (block) UDEV [1274136276.307295] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0 (scsi) UDEV [1274136276.353864] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0 (block)

(c) Finally, after a few more seconds the drive totally shuts up:
KERNEL[1274136304.187396] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0 (scsi) KERNEL[1274136304.187954] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0 (block) UDEV [1274136304.188004] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0 (scsi) UDEV [1274136304.231548] change /devices/pci0000:00/0000:00:1f.2/host1/target1:0:0/1:0:0:0/block/sr0 (block)

#13 lore raw
In reply to Nick Bowler
Quoted 5 lines · Nicolas Thomas Bowler
Quoted 2 lines · Kay Sievers

Ok, and if you enable the blkid rules again, then you see events in the monitor?

Quoted 7 lines

Sorry, the above statement that udevadm monitor prints nothing was a lie! (I must have been insane and/or lacking in coffee first thing in the morning).

With the rules commented out, the command /sbin/blkid -p -oudev /dev/sr0 on the non-spinning drive causes three distinct bursts of events to occur:

Ok, the drive seems to send out new events with every access. Maybe it's a device which tries to send AN (async) events by itself.

What does:
cat /sys/class/block/sr0/../../evt_media_change
print for this device?

Kay

#14 lore raw
In reply to Nick Bowler
Quoted 7 lines · Kay Sievers

On Tue, May 18, 2010 at 00:54, Nicolas Thomas Bowler

Quoted 5 lines · Kay Sievers

With the rules commented out, the command /sbin/blkid -p -oudev /dev/sr0 on the non-spinning drive causes three distinct bursts of events to occur:

Quoted 6 lines

Ok, the drive seems to send out new events with every access. Maybe it's a device which tries to send AN (async) events by itself.

What does:
cat /sys/class/block/sr0/../../evt_media_change
print for this device?

Prints 1.

Signature · 1 line
Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)
#15 lore raw
In reply to Nick Bowler
Quoted 8 lines · Nick Bowler
Quoted 7 lines · Kay Sievers

On Tue, May 18, 2010 at 00:54, Nicolas Thomas Bowler

Quoted 5 lines · Kay Sievers

With the rules commented out, the command /sbin/blkid -p -oudev /dev/sr0 on the non-spinning drive causes three distinct bursts of events to occur:

Quoted 8 lines
Quoted 6 lines

Ok, the drive seems to send out new events with every access. Maybe it's a device which tries to send AN (async) events by itself.

What does:
  cat /sys/class/block/sr0/../../evt_media_change
print for this device?

Prints 1.

Oh, seems to be an AN capable device, which actually really sends events. :)

I don't know if that works, never tried, but if you chmod the evt_media_change file to 0664. and write 0 to it, does it still behave the same?

Kay

#16 lore raw
In reply to Nick Bowler
Quoted 6 lines · Kay Sievers
Quoted 5 lines · Nick Bowler
Quoted 3 lines · Kay Sievers

What does:
  cat /sys/class/block/sr0/../../evt_media_change
print for this device?

Quoted 7 lines
Quoted 1 line

Prints 1.

Oh, seems to be an AN capable device, which actually really sends events. :)

I don't know if that works, never tried, but if you chmod the evt_media_change file to 0664. and write 0 to it, does it still behave the same?

I can successfully chmod and write 0 to that file, and it reads back as 0, but the behaviour remains the same (the blkid rules are uncommented), and the disk never spins down.

Signature · 1 line
Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)
#17 lore raw
In reply to Nick Bowler
Quoted 11 lines · Nick Bowler
Quoted 10 lines · Kay Sievers
Quoted 6 lines · Nick Bowler
Quoted 3 lines · Kay Sievers

What does:
  cat /sys/class/block/sr0/../../evt_media_change
print for this device?

Prints 1.

Oh, seems to be an AN capable device, which actually really sends events. :)

Quoted 7 lines
Quoted 3 lines

I don't know if that works, never tried, but if you chmod the evt_media_change file to 0664. and write 0 to it, does it still behave the same?

I can successfully chmod and write 0 to that file, and it reads back as 0, but the behaviour remains the same (the blkid rules are uncommented), and the disk never spins down.

Ok, as said, it was just a try. Thanks for the test.

The kernel seems not to filter events based on that mask. So a fix for this would be in the kernel, to either provide information about the source of the event, or to be able to suppress these AN event, which have no obvious meaning.

Tejun,
an AN capable SATA optical drive seems to send AN events with every open(). For userspace these look the same as the media changed events, and cause a loop when udev checks if there is a new media. Any ideas if these AN events are expected, or useful for anything?

David,
this is stuff we might need to cover with udisks in the future to handle AN properly.

Thanks,
Kay

#18 lore raw
In reply to Nick Bowler

Hello,

Quoted 5 lines · Kay Sievers

Tejun,
an AN capable SATA optical drive seems to send AN events with every open(). For userspace these look the same as the media changed events, and cause a loop when udev checks if there is a new media. Any ideas if these AN events are expected, or useful for anything?

The open() path issues several commands and it could be that the drive is raising AN when certain conditions are met just in case. After all windows only uses single command for media presence polling, so there's nothing stopping vendors from using TUR, for example, as the level triggered check point to see whether media is available and raise AN if so. After all, that wouldn't lose any event and in case the internal state went out of sync, it would work as an easy resync point.

Long term, I think the only solution is to do what windows does for media presence polling; otherwise, we'll keep running into various obscure issues. For short term, I dunno. Don't use AN for the time being? :-)

Thanks.

Signature · 1 line
tejun
#19 lore raw
In reply to Nick Bowler
Quoted 16 lines · Tejun Heo
Quoted 5 lines · Kay Sievers

Tejun,
an AN capable SATA optical drive seems to send AN events with every open(). For userspace these look the same as the media changed events, and cause a loop when udev checks if there is a new media. Any ideas if these AN events are expected, or useful for anything?

The open() path issues several commands and it could be that the drive is raising AN when certain conditions are met just in case.  After all windows only uses single command for media presence polling, so there's nothing stopping vendors from using TUR, for example, as the level triggered check point to see whether media is available and raise AN if so.  After all, that wouldn't lose any event and in case the internal state went out of sync, it would work as an easy resync point.

Quoted 4 lines

Long term, I think the only solution is to do what windows does for media presence polling; otherwise, we'll keep running into various obscure issues.  For short term, I dunno.  Don't use AN for the time being?  :-)

Yeah, the problem is that we already have the problem with current software that causes the event loop. And we need a solution now. :)

We might need to switch the AN stuff off by default in the kernel, and only enable it when we know we are able to handle it?

Kay

#20 lore raw
In reply to Nick Bowler

Hello,

Quoted 5 lines · Kay Sievers

Yeah, the problem is that we already have the problem with current software that causes the event loop. And we need a solution now. :)

We might need to switch the AN stuff off by default in the kernel, and only enable it when we know we are able to handle it?

Yeah, somehow ignoring those events was what I suggested. I thought maybe it can be done in userland? Currently there is no switch on the driver side. If the feature is there and the host supports it, it's automatically enabled. Hmmm... When we have proper in kernel media presence polling, AN should be piped through there anyway so it makes sense to prevent it from reaching userland directly now. How does that sound?

Thanks.

Signature · 1 line
tejun
#21 lore raw
In reply to Nick Bowler
Quoted 9 lines · Tejun Heo
Quoted 5 lines · Kay Sievers

Yeah, the problem is that we already have the problem with current software that causes the event loop. And we need a solution now. :)

We might need to switch the AN stuff off by default in the kernel, and only enable it when we know we are able to handle it?

Yeah, somehow ignoring those events was what I suggested.  I thought maybe it can be done in userland?

It doesn't look like we can do that. But while thinking about that:
Nick, if you run:
udevadm monitor --property
and access the device that it generates an event. does it always contain the line:
SDEV_MEDIA_CHANGE=1
?
That would surely be incorrect to send from the kernel, if it isn't a media change. And we might check if we can suppress these.

Quoted 6 lines

Currently there is no switch on the
driver side.  If the feature is there and the host supports it, it's automatically enabled.  Hmmm... When we have proper in kernel media presence polling, AN should be piped through there anyway so it makes sense to prevent it from reaching userland directly now.  How does that sound?

Yeah, that sounds great. It would be just nice to have something now, so we can fix the current issues. The in-kernel polling might need some time still. :)

Kay

#22 lore raw
In reply to Nick Bowler
Quoted 8 lines · Kay Sievers

Nick, if you run:
udevadm monitor --property
and access the device that it generates an event. does it always contain the line:
SDEV_MEDIA_CHANGE=1
?
That would surely be incorrect to send from the kernel, if it isn't a media change. And we might check if we can suppress these.

I ran the earlier blkid test with the drive idle, and yes, SDEV_MEDIA_CHANGE=1 appears 6 times in the output.

Signature · 1 line
Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)
#23 lore raw
In reply to Nick Bowler
Quoted 10 lines · Nick Bowler
Quoted 8 lines · Kay Sievers

Nick, if you run:
  udevadm monitor --property
and access the device that it generates an event. does it always contain the line:
  SDEV_MEDIA_CHANGE=1
?
That would surely be incorrect to send from the kernel, if it isn't a media change. And we might check if we can suppress these.

Quoted 2 lines

I ran the earlier blkid test with the drive idle, and yes, SDEV_MEDIA_CHANGE=1 appears 6 times in the output.

Tejun, the stuff in:
drivers/ata/libata-eh.c::sata_async_notification()
seems to send events for this device, even when there is no changed media. Any ideas how to debug this? Or to check if these events are really events we should put out to userspace?

Thanks,
Kay

#24 lore raw
In reply to Nick Bowler

(cc'ing linux-ide and Jeff)
Hello,

For new comers: A cdrom is smart enough to actually raise AN but at the same time dumb enough to raise it again on each open leading to infinite loop involving the drive, kernel and udev.

Quoted 5 lines · Kay Sievers

Tejun, the stuff in:
drivers/ata/libata-eh.c::sata_async_notification()
seems to send events for this device, even when there is no changed media. Any ideas how to debug this? Or to check if these events are really events we should put out to userspace?

Hmmm... I've been looking through the standards and things look somewhat dodgy. ATA ACS continuously refers to SerialATA 2.6 for details on AN and SerialATA 2.6 standard in turn claims that it describes the AN mechanism but doesn't define all the events AN can be raised for and for those one should refer to command standards (in this case ACS), *but* both list media change event as a "yeah can be used like that" example. So, in the end, nobody clearly defines it. Sweet. :-)

That said, w/o issuing further commands, the driver can't tell what the AN exactly means. There's no further information at all. It's a simple notification from the device. For ATAPI devices, it would be safe to assume that checking media status is the right action to take although both ATA and SATA standards haven't been brave enough to actually define it precisely; however, exposing it directly to userland as a media change event seems quite naive.

I think it's best to plug AN till we have proper media presence detection inside kernel proper so that our behavior doesn't differ from windows apparently is the only thing many vendors are verifying their hardware against.

Thanks.

Signature · 1 line
tejun
#25 lore raw

PATCH 1/1 libata: disable ATAPI AN by default

There are ATAPI devices which raise AN when hit by commands issued by open(). This leads to infinite loop of AN -> MEDIA_CHANGE uevent -> udev open() to check media -> AN.

Both ACS and SerialATA standards don't define in which case ATAPI devices are supposed to raise or not raise AN. They both list media insertion event as a possible use case for ATAPI ANs but there is no clear description of what constitutes such events. As such, it seems a bit too naive to export ANs directly to userland as MEDIA_CHANGE events without further verification (which should behave similarly to windows as it apparently is the only thing that some hardware vendors are testing against).

This patch adds libata.atapi_an module parameter and disables ATAPI AN by default for now.

Cc Kay Sievers <[email protected]>
Cc Nick Bowler <[email protected]>
Cc David Zeuthen <[email protected]>
---
 drivers/ata/libata-core.c |    7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/ata/libata-core.c +6 -1
index 49cffb6..5abab5d 100644
@@ -160,6 +160,10 @@ int libata_allow_tpm = 0;
 module_param_named(allow_tpm, libata_allow_tpm, int, 0444);
 MODULE_PARM_DESC(allow_tpm, "Permit the use of TPM commands (0=off [default], 1=on)");
 
+static int atapi_an;
+module_param(atapi_an, int, 0444);
+MODULE_PARM_DESC(atapi_an, "Enable ATAPI AN media presence notification (0f [default], 1=on)");
+
 MODULE_AUTHOR("Jeff Garzik");
 MODULE_DESCRIPTION("Library module for ATA devices");
 MODULE_LICENSE("GPL");
@@ -2572,7 +2576,8 @@ int ata_dev_configure(struct ata_device *dev)
 		 * to enable ATAPI AN to discern between PHY status
 		 * changed notifications and ATAPI ANs.
 		 */
-		if ((ap->flags & ATA_FLAG_AN) && ata_id_has_atapi_AN(id) &&
+		if (atapi_an &&
+		    (ap->flags & ATA_FLAG_AN) && ata_id_has_atapi_AN(id) &&
 		    (!sata_pmp_attached(ap) ||
 		     sata_scr_read(&ap->link, SCR_NOTIFICATION, &sntf) = 0)) {
 			unsigned int err_mask;
#31 lore raw
In reply to Nick Bowler
Quoted 28 lines · Tejun Heo

For new comers: A cdrom is smart enough to actually raise AN but at the same time dumb enough to raise it again on each open leading to infinite loop involving the drive, kernel and udev.

Quoted 5 lines · Kay Sievers

Tejun, the stuff in:
  drivers/ata/libata-eh.c::sata_async_notification() seems to send events for this device, even when there is no changed media. Any ideas how to debug this? Or to check if these events are really events we should put out to userspace?

Hmmm... I've been looking through the standards and things look somewhat dodgy.  ATA ACS continuously refers to SerialATA 2.6 for details on AN and SerialATA 2.6 standard in turn claims that it describes the AN mechanism but doesn't define all the events AN can be raised for and for those one should refer to command standards (in this case ACS), *but* both list media change event as a "yeah can be used like that" example.  So, in the end, nobody clearly defines it. Sweet.  :-)

That said, w/o issuing further commands, the driver can't tell what the AN exactly means.  There's no further information at all.  It's a simple notification from the device.  For ATAPI devices, it would be safe to assume that checking media status is the right action to take although both ATA and SATA standards haven't been brave enough to actually define it precisely; however, exposing it directly to userland as a media change event seems quite naive.

Quoted 4 lines

I think it's best to plug AN till we have proper media presence detection inside kernel proper so that our behavior doesn't differ from windows apparently is the only thing many vendors are verifying their hardware against.

Yeah, that sounds good, and we can probably explicitly check with the in-kernel polling if we really have a media change, and only then send stuff out to userspace.

Sounds all good, and thanks a lot for the quick fix.

Thanks,
Kay

#26 lore raw

Re: [PATCH #upstream-fixes] libata: disable ATAPI AN by default

In reply to Tejun Heo
Quoted 4 lines · Tejun Heo

There are ATAPI devices which raise AN when hit by commands issued by open(). This leads to infinite loop of AN -> MEDIA_CHANGE uevent -> udev open() to check media -> AN.

Quoted 8 lines

Both ACS and SerialATA standards don't define in which case ATAPI devices are supposed to raise or not raise AN. They both list media insertion event as a possible use case for ATAPI ANs but there is no clear description of what constitutes such events. As such, it seems a bit too naive to export ANs directly to userland as MEDIA_CHANGE events without further verification (which should behave similarly to windows as it apparently is the only thing that some hardware vendors are testing against).

Windows completely disables AN?

Or do we simply need to be less naive about _delivery_ of AN's?

Jeff

#27 lore raw

Re: [PATCH #upstream-fixes] libata: disable ATAPI AN by default

Hello,

Quoted 3 lines · Jeff Garzik

Windows completely disables AN?

Or do we simply need to be less naive about _delivery_ of AN's?

I think we first need to verify what the event is about before delivering it to userland; IOW, it should kick a full polling op. The problem is the same with periodic polling. Windows does it with single command but we can't do that from userland as open() involves issuing more commands. That discrepancy is basically we're having all these problems with periodic polling and AN infinite looping.

Hardware vendors don't consider cases where their devices are hit with a series of commands periodically or after raising an AN event. We have quite a few drives which just die after being hit repeatedly with media presence polling commands and this one is causing infinite loop by re-raising AN.

So, until we can replicate the windows behavior (which actually is pretty reasonable - just use single GET_CONFIGURATION call for polling and status check), I think it's wiser to disable AN.

Thanks.

Signature · 1 line
tejun
author Tested-by on v1 1/1
#29 lore raw

Re: [PATCH #upstream-fixes] libata: disable ATAPI AN by default

In reply to Tejun Heo
Quoted 13 lines · Tejun Heo

There are ATAPI devices which raise AN when hit by commands issued by open(). This leads to infinite loop of AN -> MEDIA_CHANGE uevent -> udev open() to check media -> AN.

Both ACS and SerialATA standards don't define in which case ATAPI devices are supposed to raise or not raise AN. They both list media insertion event as a possible use case for ATAPI ANs but there is no clear description of what constitutes such events. As such, it seems a bit too naive to export ANs directly to userland as MEDIA_CHANGE events without further verification (which should behave similarly to windows as it apparently is the only thing that some hardware vendors are testing against).

Quoted 2 lines

This patch adds libata.atapi_an module parameter and disables ATAPI AN by default for now.

Works here, drive now spins down properly.

Signature · 1 line
Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)
#28 lore raw

Re: [PATCH #upstream-fixes] libata: disable ATAPI AN by default

In reply to Tejun Heo
Quoted 20 lines · Tejun Heo

Hello,

Quoted 3 lines · Jeff Garzik

Windows completely disables AN?

Or do we simply need to be less naive about _delivery_ of AN's?

I think we first need to verify what the event is about before delivering it to userland; IOW, it should kick a full polling op. The problem is the same with periodic polling. Windows does it with single command but we can't do that from userland as open() involves issuing more commands. That discrepancy is basically we're having all these problems with periodic polling and AN infinite looping.

Hardware vendors don't consider cases where their devices are hit with a series of commands periodically or after raising an AN event. We have quite a few drives which just die after being hit repeatedly with media presence polling commands and this one is causing infinite loop by re-raising AN.

Quoted 3 lines

So, until we can replicate the windows behavior (which actually is pretty reasonable - just use single GET_CONFIGURATION call for polling and status check), I think it's wiser to disable AN.

There are also some drives where AN is just broken - I have a Lite-ON DH-401S BD-ROM drive which raises AN on opening the tray, but not on closing with media inserted.

I don't know if Windows is using AN at all yet or not. Seems like a lot of drives aren't tested well with it though.

Applied
#30 lore raw

Re: [PATCH #upstream-fixes] libata: disable ATAPI AN by default

In reply to Tejun Heo
Quoted 16 lines · Tejun Heo

There are ATAPI devices which raise AN when hit by commands issued by open(). This leads to infinite loop of AN -> MEDIA_CHANGE uevent -> udev open() to check media -> AN.

Both ACS and SerialATA standards don't define in which case ATAPI devices are supposed to raise or not raise AN. They both list media insertion event as a possible use case for ATAPI ANs but there is no clear description of what constitutes such events. As such, it seems a bit too naive to export ANs directly to userland as MEDIA_CHANGE events without further verification (which should behave similarly to windows as it apparently is the only thing that some hardware vendors are testing against).

This patch adds libata.atapi_an module parameter and disables ATAPI AN by default for now.

Quoted 8 lines
---
  drivers/ata/libata-core.c |    7 ++++++-
  1 file changed, 6 insertions(+), 1 deletion(-)

applied