disk-shock-protection.rst

  1==========================
  2Hard disk shock protection
  3==========================
  4
  5Author: Elias Oltmanns <eo@nebensachen.de>
  6
  7Last modified: 2008-10-03
  8
  9
 10.. 0. Contents
 11
 12   1. Intro
 13   2. The interface
 14   3. References
 15   4. CREDITS
 16
 17
 181. Intro
 19--------
 20
 21ATA/ATAPI-7 specifies the IDLE IMMEDIATE command with unload feature.
 22Issuing this command should cause the drive to switch to idle mode and
 23unload disk heads. This feature is being used in modern laptops in
 24conjunction with accelerometers and appropriate software to implement
 25a shock protection facility. The idea is to stop all I/O operations on
 26the internal hard drive and park its heads on the ramp when critical
 27situations are anticipated. The desire to have such a feature
 28available on GNU/Linux systems has been the original motivation to
 29implement a generic disk head parking interface in the Linux kernel.
 30Please note, however, that other components have to be set up on your
 31system in order to get disk shock protection working (see
 32section 3. References below for pointers to more information about
 33that).
 34
 35
 362. The interface
 37----------------
 38
 39For each ATA device, the kernel exports the file
 40`block/*/device/unload_heads` in sysfs (here assumed to be mounted under
 41/sys). Access to `/sys/block/*/device/unload_heads` is denied with
 42-EOPNOTSUPP if the device does not support the unload feature.
 43Otherwise, writing an integer value to this file will take the heads
 44of the respective drive off the platter and block all I/O operations
 45for the specified number of milliseconds. When the timeout expires and
 46no further disk head park request has been issued in the meantime,
 47normal operation will be resumed. The maximal value accepted for a
 48timeout is 30000 milliseconds. Exceeding this limit will return
 49-EOVERFLOW, but heads will be parked anyway and the timeout will be
 50set to 30 seconds. However, you can always change a timeout to any
 51value between 0 and 30000 by issuing a subsequent head park request
 52before the timeout of the previous one has expired. In particular, the
 53total timeout can exceed 30 seconds and, more importantly, you can
 54cancel a previously set timeout and resume normal operation
 55immediately by specifying a timeout of 0. Values below -2 are rejected
 56with -EINVAL (see below for the special meaning of -1 and -2). If the
 57timeout specified for a recent head park request has not yet expired,
 58reading from `/sys/block/*/device/unload_heads` will report the number
 59of milliseconds remaining until normal operation will be resumed;
 60otherwise, reading the unload_heads attribute will return 0.
 61
 62For example, do the following in order to park the heads of drive
 63/dev/sda and stop all I/O operations for five seconds::
 64
 65	# echo 5000 > /sys/block/sda/device/unload_heads
 66
 67A simple::
 68
 69	# cat /sys/block/sda/device/unload_heads
 70
 71will show you how many milliseconds are left before normal operation
 72will be resumed.
 73
 74A word of caution: The fact that the interface operates on a basis of
 75milliseconds may raise expectations that cannot be satisfied in
 76reality. In fact, the ATA specs clearly state that the time for an
 77unload operation to complete is vendor specific. The hint in ATA-7
 78that this will typically be within 500 milliseconds apparently has
 79been dropped in ATA-8.
 80
 81There is a technical detail of this implementation that may cause some
 82confusion and should be discussed here. When a head park request has
 83been issued to a device successfully, all I/O operations on the
 84controller port this device is attached to will be deferred. That is
 85to say, any other device that may be connected to the same port will
 86be affected too. The only exception is that a subsequent head unload
 87request to that other device will be executed immediately. Further
 88operations on that port will be deferred until the timeout specified
 89for either device on the port has expired. As far as PATA (old style
 90IDE) configurations are concerned, there can only be two devices
 91attached to any single port. In SATA world we have port multipliers
 92which means that a user-issued head parking request to one device may
 93actually result in stopping I/O to a whole bunch of devices. However,
 94since this feature is supposed to be used on laptops and does not seem
 95to be very useful in any other environment, there will be mostly one
 96device per port. Even if the CD/DVD writer happens to be connected to
 97the same port as the hard drive, it generally *should* recover just
 98fine from the occasional buffer under-run incurred by a head park
 99request to the HD. Actually, when you are using an ide driver rather
100than its libata counterpart (i.e. your disk is called /dev/hda
101instead of /dev/sda), then parking the heads of one drive (drive X)
102will generally not affect the mode of operation of another drive
103(drive Y) on the same port as described above. It is only when a port
104reset is required to recover from an exception on drive Y that further
105I/O operations on that drive (and the reset itself) will be delayed
106until drive X is no longer in the parked state.
107
108Finally, there are some hard drives that only comply with an earlier
109version of the ATA standard than ATA-7, but do support the unload
110feature nonetheless. Unfortunately, there is no safe way Linux can
111detect these devices, so you won't be able to write to the
112unload_heads attribute. If you know that your device really does
113support the unload feature (for instance, because the vendor of your
114laptop or the hard drive itself told you so), then you can tell the
115kernel to enable the usage of this feature for that drive by writing
116the special value -1 to the unload_heads attribute::
117
118	# echo -1 > /sys/block/sda/device/unload_heads
119
120will enable the feature for /dev/sda, and giving -2 instead of -1 will
121disable it again.
122
123
1243. References
125-------------
126
127There are several laptops from different vendors featuring shock
128protection capabilities. As manufacturers have refused to support open
129source development of the required software components so far, Linux
130support for shock protection varies considerably between different
131hardware implementations. Ideally, this section should contain a list
132of pointers at different projects aiming at an implementation of shock
133protection on different systems. Unfortunately, I only know of a
134single project which, although still considered experimental, is fit
135for use. Please feel free to add projects that have been the victims
136of my ignorance.
137
138- https://www.thinkwiki.org/wiki/HDAPS
139
140  See this page for information about Linux support of the hard disk
141  active protection system as implemented in IBM/Lenovo Thinkpads.
142
143
1444. CREDITS
145----------
146
147This implementation of disk head parking has been inspired by a patch
148originally published by Jon Escombe <lists@dresco.co.uk>. My efforts
149to develop an implementation of this feature that is fit to be merged
150into mainline have been aided by various kernel developers, in
151particular by Tejun Heo and Bartlomiej Zolnierkiewicz.