1Hard disk shock protection
  2==========================
  3
  4Author: Elias Oltmanns <eo@nebensachen.de>
  5Last modified: 2008-10-03
  6
  7
  80. Contents
  9-----------
 10
 111. Intro
 122. The interface
 133. References
 144. CREDITS
 15
 16
 171. Intro
 18--------
 19
 20ATA/ATAPI-7 specifies the IDLE IMMEDIATE command with unload feature.
 21Issuing this command should cause the drive to switch to idle mode and
 22unload disk heads. This feature is being used in modern laptops in
 23conjunction with accelerometers and appropriate software to implement
 24a shock protection facility. The idea is to stop all I/O operations on
 25the internal hard drive and park its heads on the ramp when critical
 26situations are anticipated. The desire to have such a feature
 27available on GNU/Linux systems has been the original motivation to
 28implement a generic disk head parking interface in the Linux kernel.
 29Please note, however, that other components have to be set up on your
 30system in order to get disk shock protection working (see
 31section 3. References below for pointers to more information about
 32that).
 33
 34
 352. The interface
 36----------------
 37
 38For each ATA device, the kernel exports the file
 39block/*/device/unload_heads in sysfs (here assumed to be mounted under
 40/sys). Access to /sys/block/*/device/unload_heads is denied with
 41-EOPNOTSUPP if the device does not support the unload feature.
 42Otherwise, writing an integer value to this file will take the heads
 43of the respective drive off the platter and block all I/O operations
 44for the specified number of milliseconds. When the timeout expires and
 45no further disk head park request has been issued in the meantime,
 46normal operation will be resumed. The maximal value accepted for a
 47timeout is 30000 milliseconds. Exceeding this limit will return
 48-EOVERFLOW, but heads will be parked anyway and the timeout will be
 49set to 30 seconds. However, you can always change a timeout to any
 50value between 0 and 30000 by issuing a subsequent head park request
 51before the timeout of the previous one has expired. In particular, the
 52total timeout can exceed 30 seconds and, more importantly, you can
 53cancel a previously set timeout and resume normal operation
 54immediately by specifying a timeout of 0. Values below -2 are rejected
 55with -EINVAL (see below for the special meaning of -1 and -2). If the
 56timeout specified for a recent head park request has not yet expired,
 57reading from /sys/block/*/device/unload_heads will report the number
 58of milliseconds remaining until normal operation will be resumed;
 59otherwise, reading the unload_heads attribute will return 0.
 60
 61For example, do the following in order to park the heads of drive
 62/dev/sda and stop all I/O operations for five seconds:
 63
 64# echo 5000 > /sys/block/sda/device/unload_heads
 65
 66A simple
 67
 68# cat /sys/block/sda/device/unload_heads
 69
 70will show you how many milliseconds are left before normal operation
 71will be resumed.
 72
 73A word of caution: The fact that the interface operates on a basis of
 74milliseconds may raise expectations that cannot be satisfied in
 75reality. In fact, the ATA specs clearly state that the time for an
 76unload operation to complete is vendor specific. The hint in ATA-7
 77that this will typically be within 500 milliseconds apparently has
 78been dropped in ATA-8.
 79
 80There is a technical detail of this implementation that may cause some
 81confusion and should be discussed here. When a head park request has
 82been issued to a device successfully, all I/O operations on the
 83controller port this device is attached to will be deferred. That is
 84to say, any other device that may be connected to the same port will
 85be affected too. The only exception is that a subsequent head unload
 86request to that other device will be executed immediately. Further
 87operations on that port will be deferred until the timeout specified
 88for either device on the port has expired. As far as PATA (old style
 89IDE) configurations are concerned, there can only be two devices
 90attached to any single port. In SATA world we have port multipliers
 91which means that a user-issued head parking request to one device may
 92actually result in stopping I/O to a whole bunch of devices. However,
 93since this feature is supposed to be used on laptops and does not seem
 94to be very useful in any other environment, there will be mostly one
 95device per port. Even if the CD/DVD writer happens to be connected to
 96the same port as the hard drive, it generally *should* recover just
 97fine from the occasional buffer under-run incurred by a head park
 98request to the HD. Actually, when you are using an ide driver rather
 99than its libata counterpart (i.e. your disk is called /dev/hda
100instead of /dev/sda), then parking the heads of one drive (drive X)
101will generally not affect the mode of operation of another drive
102(drive Y) on the same port as described above. It is only when a port
103reset is required to recover from an exception on drive Y that further
104I/O operations on that drive (and the reset itself) will be delayed
105until drive X is no longer in the parked state.
106
107Finally, there are some hard drives that only comply with an earlier
108version of the ATA standard than ATA-7, but do support the unload
109feature nonetheless. Unfortunately, there is no safe way Linux can
110detect these devices, so you won't be able to write to the
111unload_heads attribute. If you know that your device really does
112support the unload feature (for instance, because the vendor of your
113laptop or the hard drive itself told you so), then you can tell the
114kernel to enable the usage of this feature for that drive by writing
115the special value -1 to the unload_heads attribute:
116
117# echo -1 > /sys/block/sda/device/unload_heads
118
119will enable the feature for /dev/sda, and giving -2 instead of -1 will
120disable it again.
121
122
1233. References
124-------------
125
126There are several laptops from different vendors featuring shock
127protection capabilities. As manufacturers have refused to support open
128source development of the required software components so far, Linux
129support for shock protection varies considerably between different
130hardware implementations. Ideally, this section should contain a list
131of pointers at different projects aiming at an implementation of shock
132protection on different systems. Unfortunately, I only know of a
133single project which, although still considered experimental, is fit
134for use. Please feel free to add projects that have been the victims
135of my ignorance.
136
137- http://www.thinkwiki.org/wiki/HDAPS
138  See this page for information about Linux support of the hard disk
139  active protection system as implemented in IBM/Lenovo Thinkpads.
140
141
1424. CREDITS
143----------
144
145This implementation of disk head parking has been inspired by a patch
146originally published by Jon Escombe <lists@dresco.co.uk>. My efforts
147to develop an implementation of this feature that is fit to be merged
148into mainline have been aided by various kernel developers, in
149particular by Tejun Heo and Bartlomiej Zolnierkiewicz.