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