blob: ba997577150369c5d7bc655c82704e01567e685d [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001<?xml version="1.0" encoding="UTF-8"?>
2<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.1.2//EN"
3 "http://www.oasis-open.org/docbook/xml/4.1.2/docbookx.dtd" []>
4
5<book id="libataDevGuide">
6 <bookinfo>
7 <title>libATA Developer's Guide</title>
8
9 <authorgroup>
10 <author>
11 <firstname>Jeff</firstname>
12 <surname>Garzik</surname>
13 </author>
14 </authorgroup>
15
16 <copyright>
Jeff Garzik4f931372006-09-29 05:07:25 -040017 <year>2003-2006</year>
Linus Torvalds1da177e2005-04-16 15:20:36 -070018 <holder>Jeff Garzik</holder>
19 </copyright>
20
21 <legalnotice>
22 <para>
23 The contents of this file are subject to the Open
24 Software License version 1.1 that can be found at
25 <ulink url="http://www.opensource.org/licenses/osl-1.1.txt">http://www.opensource.org/licenses/osl-1.1.txt</ulink> and is included herein
26 by reference.
27 </para>
28
29 <para>
30 Alternatively, the contents of this file may be used under the terms
31 of the GNU General Public License version 2 (the "GPL") as distributed
32 in the kernel source COPYING file, in which case the provisions of
33 the GPL are applicable instead of the above. If you wish to allow
34 the use of your version of this file only under the terms of the
35 GPL and not to allow others to use your version of this file under
36 the OSL, indicate your decision by deleting the provisions above and
37 replace them with the notice and other provisions required by the GPL.
38 If you do not delete the provisions above, a recipient may use your
39 version of this file under either the OSL or the GPL.
40 </para>
41
42 </legalnotice>
43 </bookinfo>
44
45<toc></toc>
46
Jeff Garzik07dd39b2005-05-30 13:15:52 -040047 <chapter id="libataIntroduction">
48 <title>Introduction</title>
49 <para>
50 libATA is a library used inside the Linux kernel to support ATA host
51 controllers and devices. libATA provides an ATA driver API, class
52 transports for ATA and ATAPI devices, and SCSI&lt;-&gt;ATA translation
53 for ATA devices according to the T10 SAT specification.
54 </para>
55 <para>
56 This Guide documents the libATA driver API, library functions, library
57 internals, and a couple sample ATA low-level drivers.
58 </para>
59 </chapter>
60
Linus Torvalds1da177e2005-04-16 15:20:36 -070061 <chapter id="libataDriverApi">
62 <title>libata Driver API</title>
Jeff Garzik92bab262005-05-31 20:43:57 -040063 <para>
64 struct ata_port_operations is defined for every low-level libata
65 hardware driver, and it controls how the low-level driver
66 interfaces with the ATA and SCSI layers.
67 </para>
68 <para>
69 FIS-based drivers will hook into the system with ->qc_prep() and
70 ->qc_issue() high-level hooks. Hardware which behaves in a manner
71 similar to PCI IDE hardware may utilize several generic helpers,
72 defining at a bare minimum the bus I/O addresses of the ATA shadow
73 register blocks.
74 </para>
Linus Torvalds1da177e2005-04-16 15:20:36 -070075 <sect1>
76 <title>struct ata_port_operations</title>
77
Jeff Garzik92bab262005-05-31 20:43:57 -040078 <sect2><title>Disable ATA port</title>
Linus Torvalds1da177e2005-04-16 15:20:36 -070079 <programlisting>
80void (*port_disable) (struct ata_port *);
81 </programlisting>
82
83 <para>
84 Called from ata_bus_probe() and ata_bus_reset() error paths,
85 as well as when unregistering from the SCSI module (rmmod, hot
86 unplug).
Edward Falk8b2af8f2005-06-15 14:26:39 -070087 This function should do whatever needs to be done to take the
88 port out of use. In most cases, ata_port_disable() can be used
89 as this hook.
90 </para>
91 <para>
92 Called from ata_bus_probe() on a failed probe.
93 Called from ata_bus_reset() on a failed bus reset.
94 Called from ata_scsi_release().
Linus Torvalds1da177e2005-04-16 15:20:36 -070095 </para>
96
Jeff Garzik92bab262005-05-31 20:43:57 -040097 </sect2>
98
99 <sect2><title>Post-IDENTIFY device configuration</title>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700100 <programlisting>
101void (*dev_config) (struct ata_port *, struct ata_device *);
102 </programlisting>
103
104 <para>
105 Called after IDENTIFY [PACKET] DEVICE is issued to each device
106 found. Typically used to apply device-specific fixups prior to
107 issue of SET FEATURES - XFER MODE, and prior to operation.
108 </para>
Edward Falk8b2af8f2005-06-15 14:26:39 -0700109 <para>
110 Called by ata_device_add() after ata_dev_identify() determines
111 a device is present.
112 </para>
113 <para>
114 This entry may be specified as NULL in ata_port_operations.
115 </para>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700116
Jeff Garzik92bab262005-05-31 20:43:57 -0400117 </sect2>
118
119 <sect2><title>Set PIO/DMA mode</title>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700120 <programlisting>
121void (*set_piomode) (struct ata_port *, struct ata_device *);
122void (*set_dmamode) (struct ata_port *, struct ata_device *);
Alan Cox5444a6f2006-03-27 18:58:20 +0100123void (*post_set_mode) (struct ata_port *);
124unsigned int (*mode_filter) (struct ata_port *, struct ata_device *, unsigned int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700125 </programlisting>
126
127 <para>
128 Hooks called prior to the issue of SET FEATURES - XFER MODE
Alan Cox5444a6f2006-03-27 18:58:20 +0100129 command. The optional ->mode_filter() hook is called when libata
130 has built a mask of the possible modes. This is passed to the
131 ->mode_filter() function which should return a mask of valid modes
132 after filtering those unsuitable due to hardware limits. It is not
133 valid to use this interface to add modes.
134 </para>
135 <para>
136 dev->pio_mode and dev->dma_mode are guaranteed to be valid when
137 ->set_piomode() and when ->set_dmamode() is called. The timings for
138 any other drive sharing the cable will also be valid at this point.
139 That is the library records the decisions for the modes of each
140 drive on a channel before it attempts to set any of them.
141 </para>
142 <para>
143 ->post_set_mode() is
Linus Torvalds1da177e2005-04-16 15:20:36 -0700144 called unconditionally, after the SET FEATURES - XFER MODE
145 command completes successfully.
146 </para>
147
148 <para>
149 ->set_piomode() is always called (if present), but
150 ->set_dma_mode() is only called if DMA is possible.
151 </para>
152
Jeff Garzik92bab262005-05-31 20:43:57 -0400153 </sect2>
154
155 <sect2><title>Taskfile read/write</title>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700156 <programlisting>
157void (*tf_load) (struct ata_port *ap, struct ata_taskfile *tf);
158void (*tf_read) (struct ata_port *ap, struct ata_taskfile *tf);
159 </programlisting>
160
161 <para>
162 ->tf_load() is called to load the given taskfile into hardware
163 registers / DMA buffers. ->tf_read() is called to read the
164 hardware registers / DMA buffers, to obtain the current set of
165 taskfile register values.
Edward Falk8b2af8f2005-06-15 14:26:39 -0700166 Most drivers for taskfile-based hardware (PIO or MMIO) use
167 ata_tf_load() and ata_tf_read() for these hooks.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700168 </para>
169
Jeff Garzik92bab262005-05-31 20:43:57 -0400170 </sect2>
171
Jeff Garzikbf717b112006-06-13 20:27:03 -0400172 <sect2><title>PIO data read/write</title>
173 <programlisting>
174void (*data_xfer) (struct ata_device *, unsigned char *, unsigned int, int);
175 </programlisting>
176
177 <para>
178All bmdma-style drivers must implement this hook. This is the low-level
179operation that actually copies the data bytes during a PIO data
180transfer.
181Typically the driver
182will choose one of ata_pio_data_xfer_noirq(), ata_pio_data_xfer(), or
183ata_mmio_data_xfer().
184 </para>
185
186 </sect2>
187
Jeff Garzik92bab262005-05-31 20:43:57 -0400188 <sect2><title>ATA command execute</title>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700189 <programlisting>
190void (*exec_command)(struct ata_port *ap, struct ata_taskfile *tf);
191 </programlisting>
192
193 <para>
194 causes an ATA command, previously loaded with
195 ->tf_load(), to be initiated in hardware.
Edward Falk8b2af8f2005-06-15 14:26:39 -0700196 Most drivers for taskfile-based hardware use ata_exec_command()
197 for this hook.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700198 </para>
199
Jeff Garzik92bab262005-05-31 20:43:57 -0400200 </sect2>
201
202 <sect2><title>Per-cmd ATAPI DMA capabilities filter</title>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700203 <programlisting>
Jeff Garzik780a87f2005-05-30 15:41:05 -0400204int (*check_atapi_dma) (struct ata_queued_cmd *qc);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700205 </programlisting>
206
207 <para>
Jeff Garzik780a87f2005-05-30 15:41:05 -0400208Allow low-level driver to filter ATA PACKET commands, returning a status
209indicating whether or not it is OK to use DMA for the supplied PACKET
210command.
211 </para>
Edward Falk8b2af8f2005-06-15 14:26:39 -0700212 <para>
213 This hook may be specified as NULL, in which case libata will
214 assume that atapi dma can be supported.
215 </para>
Jeff Garzik780a87f2005-05-30 15:41:05 -0400216
Jeff Garzik92bab262005-05-31 20:43:57 -0400217 </sect2>
218
219 <sect2><title>Read specific ATA shadow registers</title>
Jeff Garzik780a87f2005-05-30 15:41:05 -0400220 <programlisting>
221u8 (*check_status)(struct ata_port *ap);
222u8 (*check_altstatus)(struct ata_port *ap);
Jeff Garzik780a87f2005-05-30 15:41:05 -0400223 </programlisting>
224
225 <para>
Jeff Garzikbf717b112006-06-13 20:27:03 -0400226 Reads the Status/AltStatus ATA shadow register from
Jeff Garzik780a87f2005-05-30 15:41:05 -0400227 hardware. On some hardware, reading the Status register has
228 the side effect of clearing the interrupt condition.
Edward Falk8b2af8f2005-06-15 14:26:39 -0700229 Most drivers for taskfile-based hardware use
230 ata_check_status() for this hook.
231 </para>
232 <para>
233 Note that because this is called from ata_device_add(), at
234 least a dummy function that clears device interrupts must be
235 provided for all drivers, even if the controller doesn't
236 actually have a taskfile status register.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700237 </para>
238
Jeff Garzik92bab262005-05-31 20:43:57 -0400239 </sect2>
240
241 <sect2><title>Select ATA device on bus</title>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700242 <programlisting>
243void (*dev_select)(struct ata_port *ap, unsigned int device);
244 </programlisting>
245
246 <para>
247 Issues the low-level hardware command(s) that causes one of N
248 hardware devices to be considered 'selected' (active and
Jeff Garzik780a87f2005-05-30 15:41:05 -0400249 available for use) on the ATA bus. This generally has no
Edward Falk8b2af8f2005-06-15 14:26:39 -0700250 meaning on FIS-based devices.
251 </para>
252 <para>
253 Most drivers for taskfile-based hardware use
254 ata_std_dev_select() for this hook. Controllers which do not
255 support second drives on a port (such as SATA contollers) will
256 use ata_noop_dev_select().
Linus Torvalds1da177e2005-04-16 15:20:36 -0700257 </para>
258
Jeff Garzik92bab262005-05-31 20:43:57 -0400259 </sect2>
260
Alan Cox5444a6f2006-03-27 18:58:20 +0100261 <sect2><title>Private tuning method</title>
262 <programlisting>
263void (*set_mode) (struct ata_port *ap);
264 </programlisting>
265
266 <para>
267 By default libata performs drive and controller tuning in
268 accordance with the ATA timing rules and also applies blacklists
269 and cable limits. Some controllers need special handling and have
270 custom tuning rules, typically raid controllers that use ATA
271 commands but do not actually do drive timing.
272 </para>
273
274 <warning>
275 <para>
276 This hook should not be used to replace the standard controller
277 tuning logic when a controller has quirks. Replacing the default
278 tuning logic in that case would bypass handling for drive and
279 bridge quirks that may be important to data reliability. If a
280 controller needs to filter the mode selection it should use the
281 mode_filter hook instead.
282 </para>
283 </warning>
284
285 </sect2>
286
Jeff Garzik92bab262005-05-31 20:43:57 -0400287 <sect2><title>Control PCI IDE BMDMA engine</title>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700288 <programlisting>
289void (*bmdma_setup) (struct ata_queued_cmd *qc);
290void (*bmdma_start) (struct ata_queued_cmd *qc);
Jeff Garzik780a87f2005-05-30 15:41:05 -0400291void (*bmdma_stop) (struct ata_port *ap);
292u8 (*bmdma_status) (struct ata_port *ap);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700293 </programlisting>
294
295 <para>
Jeff Garzik780a87f2005-05-30 15:41:05 -0400296When setting up an IDE BMDMA transaction, these hooks arm
297(->bmdma_setup), fire (->bmdma_start), and halt (->bmdma_stop)
298the hardware's DMA engine. ->bmdma_status is used to read the standard
299PCI IDE DMA Status register.
300 </para>
301
302 <para>
303These hooks are typically either no-ops, or simply not implemented, in
304FIS-based drivers.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700305 </para>
Edward Falk8b2af8f2005-06-15 14:26:39 -0700306 <para>
307Most legacy IDE drivers use ata_bmdma_setup() for the bmdma_setup()
308hook. ata_bmdma_setup() will write the pointer to the PRD table to
309the IDE PRD Table Address register, enable DMA in the DMA Command
310register, and call exec_command() to begin the transfer.
311 </para>
312 <para>
313Most legacy IDE drivers use ata_bmdma_start() for the bmdma_start()
314hook. ata_bmdma_start() will write the ATA_DMA_START flag to the DMA
315Command register.
316 </para>
317 <para>
318Many legacy IDE drivers use ata_bmdma_stop() for the bmdma_stop()
319hook. ata_bmdma_stop() clears the ATA_DMA_START flag in the DMA
320command register.
321 </para>
322 <para>
323Many legacy IDE drivers use ata_bmdma_status() as the bmdma_status() hook.
324 </para>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700325
Jeff Garzik92bab262005-05-31 20:43:57 -0400326 </sect2>
327
328 <sect2><title>High-level taskfile hooks</title>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700329 <programlisting>
330void (*qc_prep) (struct ata_queued_cmd *qc);
331int (*qc_issue) (struct ata_queued_cmd *qc);
332 </programlisting>
333
334 <para>
335 Higher-level hooks, these two hooks can potentially supercede
336 several of the above taskfile/DMA engine hooks. ->qc_prep is
337 called after the buffers have been DMA-mapped, and is typically
338 used to populate the hardware's DMA scatter-gather table.
339 Most drivers use the standard ata_qc_prep() helper function, but
340 more advanced drivers roll their own.
341 </para>
342 <para>
343 ->qc_issue is used to make a command active, once the hardware
344 and S/G tables have been prepared. IDE BMDMA drivers use the
345 helper function ata_qc_issue_prot() for taskfile protocol-based
Jeff Garzik780a87f2005-05-30 15:41:05 -0400346 dispatch. More advanced drivers implement their own ->qc_issue.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700347 </para>
Edward Falk8b2af8f2005-06-15 14:26:39 -0700348 <para>
349 ata_qc_issue_prot() calls ->tf_load(), ->bmdma_setup(), and
350 ->bmdma_start() as necessary to initiate a transfer.
351 </para>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700352
Jeff Garzik92bab262005-05-31 20:43:57 -0400353 </sect2>
354
Jeff Garzikbf717b112006-06-13 20:27:03 -0400355 <sect2><title>Exception and probe handling (EH)</title>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700356 <programlisting>
357void (*eng_timeout) (struct ata_port *ap);
Jeff Garzikbf717b112006-06-13 20:27:03 -0400358void (*phy_reset) (struct ata_port *ap);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700359 </programlisting>
360
361 <para>
Jeff Garzikbf717b112006-06-13 20:27:03 -0400362Deprecated. Use ->error_handler() instead.
363 </para>
364
365 <programlisting>
366void (*freeze) (struct ata_port *ap);
367void (*thaw) (struct ata_port *ap);
368 </programlisting>
369
370 <para>
371ata_port_freeze() is called when HSM violations or some other
372condition disrupts normal operation of the port. A frozen port
373is not allowed to perform any operation until the port is
374thawed, which usually follows a successful reset.
375 </para>
376
377 <para>
378The optional ->freeze() callback can be used for freezing the port
379hardware-wise (e.g. mask interrupt and stop DMA engine). If a
380port cannot be frozen hardware-wise, the interrupt handler
381must ack and clear interrupts unconditionally while the port
382is frozen.
383 </para>
384 <para>
385The optional ->thaw() callback is called to perform the opposite of ->freeze():
386prepare the port for normal operation once again. Unmask interrupts,
387start DMA engine, etc.
388 </para>
389
390 <programlisting>
391void (*error_handler) (struct ata_port *ap);
392 </programlisting>
393
394 <para>
395->error_handler() is a driver's hook into probe, hotplug, and recovery
396and other exceptional conditions. The primary responsibility of an
397implementation is to call ata_do_eh() or ata_bmdma_drive_eh() with a set
398of EH hooks as arguments:
399 </para>
400
401 <para>
402'prereset' hook (may be NULL) is called during an EH reset, before any other actions
403are taken.
404 </para>
405
406 <para>
407'postreset' hook (may be NULL) is called after the EH reset is performed. Based on
408existing conditions, severity of the problem, and hardware capabilities,
409 </para>
410
411 <para>
412Either 'softreset' (may be NULL) or 'hardreset' (may be NULL) will be
413called to perform the low-level EH reset.
414 </para>
415
416 <programlisting>
417void (*post_internal_cmd) (struct ata_queued_cmd *qc);
418 </programlisting>
419
420 <para>
421Perform any hardware-specific actions necessary to finish processing
422after executing a probe-time or EH-time command via ata_exec_internal().
Linus Torvalds1da177e2005-04-16 15:20:36 -0700423 </para>
424
Jeff Garzik92bab262005-05-31 20:43:57 -0400425 </sect2>
426
427 <sect2><title>Hardware interrupt handling</title>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700428 <programlisting>
429irqreturn_t (*irq_handler)(int, void *, struct pt_regs *);
430void (*irq_clear) (struct ata_port *);
431 </programlisting>
432
433 <para>
434 ->irq_handler is the interrupt handling routine registered with
435 the system, by libata. ->irq_clear is called during probe just
436 before the interrupt handler is registered, to be sure hardware
437 is quiet.
438 </para>
Edward Falk8b2af8f2005-06-15 14:26:39 -0700439 <para>
440 The second argument, dev_instance, should be cast to a pointer
441 to struct ata_host_set.
442 </para>
443 <para>
444 Most legacy IDE drivers use ata_interrupt() for the
445 irq_handler hook, which scans all ports in the host_set,
446 determines which queued command was active (if any), and calls
447 ata_host_intr(ap,qc).
448 </para>
449 <para>
450 Most legacy IDE drivers use ata_bmdma_irq_clear() for the
451 irq_clear() hook, which simply clears the interrupt and error
452 flags in the DMA status register.
453 </para>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700454
Jeff Garzik92bab262005-05-31 20:43:57 -0400455 </sect2>
456
457 <sect2><title>SATA phy read/write</title>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700458 <programlisting>
James Bottomley110dd8f2007-07-20 13:11:44 -0500459int (*scr_read) (struct ata_port *ap, unsigned int sc_reg,
460 u32 *val);
461int (*scr_write) (struct ata_port *ap, unsigned int sc_reg,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700462 u32 val);
463 </programlisting>
464
465 <para>
466 Read and write standard SATA phy registers. Currently only used
467 if ->phy_reset hook called the sata_phy_reset() helper function.
Edward Falk8b2af8f2005-06-15 14:26:39 -0700468 sc_reg is one of SCR_STATUS, SCR_CONTROL, SCR_ERROR, or SCR_ACTIVE.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700469 </para>
470
Jeff Garzik92bab262005-05-31 20:43:57 -0400471 </sect2>
472
473 <sect2><title>Init and shutdown</title>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700474 <programlisting>
475int (*port_start) (struct ata_port *ap);
476void (*port_stop) (struct ata_port *ap);
477void (*host_stop) (struct ata_host_set *host_set);
478 </programlisting>
479
480 <para>
481 ->port_start() is called just after the data structures for each
482 port are initialized. Typically this is used to alloc per-port
483 DMA buffers / tables / rings, enable DMA engines, and similar
Edward Falk8b2af8f2005-06-15 14:26:39 -0700484 tasks. Some drivers also use this entry point as a chance to
485 allocate driver-private memory for ap->private_data.
486 </para>
487 <para>
488 Many drivers use ata_port_start() as this hook or call
489 it from their own port_start() hooks. ata_port_start()
490 allocates space for a legacy IDE PRD table and returns.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700491 </para>
492 <para>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700493 ->port_stop() is called after ->host_stop(). It's sole function
494 is to release DMA/memory resources, now that they are no longer
Edward Falk8b2af8f2005-06-15 14:26:39 -0700495 actively being used. Many drivers also free driver-private
496 data from port at this time.
497 </para>
498 <para>
499 Many drivers use ata_port_stop() as this hook, which frees the
500 PRD table.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700501 </para>
Jeff Garzik780a87f2005-05-30 15:41:05 -0400502 <para>
503 ->host_stop() is called after all ->port_stop() calls
504have completed. The hook must finalize hardware shutdown, release DMA
505and other resources, etc.
Edward Falk8b2af8f2005-06-15 14:26:39 -0700506 This hook may be specified as NULL, in which case it is not called.
Jeff Garzik780a87f2005-05-30 15:41:05 -0400507 </para>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700508
Jeff Garzik92bab262005-05-31 20:43:57 -0400509 </sect2>
510
Linus Torvalds1da177e2005-04-16 15:20:36 -0700511 </sect1>
Jeff Garzika1213492005-09-28 13:26:47 -0400512 </chapter>
513
514 <chapter id="libataEH">
Tejun Heobfd00722005-09-26 11:28:47 +0900515 <title>Error handling</title>
516
517 <para>
518 This chapter describes how errors are handled under libata.
519 Readers are advised to read SCSI EH
520 (Documentation/scsi/scsi_eh.txt) and ATA exceptions doc first.
521 </para>
522
Jeff Garzika1213492005-09-28 13:26:47 -0400523 <sect1><title>Origins of commands</title>
Tejun Heobfd00722005-09-26 11:28:47 +0900524 <para>
525 In libata, a command is represented with struct ata_queued_cmd
526 or qc. qc's are preallocated during port initialization and
527 repetitively used for command executions. Currently only one
528 qc is allocated per port but yet-to-be-merged NCQ branch
529 allocates one for each tag and maps each qc to NCQ tag 1-to-1.
530 </para>
531 <para>
532 libata commands can originate from two sources - libata itself
533 and SCSI midlayer. libata internal commands are used for
534 initialization and error handling. All normal blk requests
535 and commands for SCSI emulation are passed as SCSI commands
536 through queuecommand callback of SCSI host template.
537 </para>
Jeff Garzika1213492005-09-28 13:26:47 -0400538 </sect1>
Tejun Heobfd00722005-09-26 11:28:47 +0900539
Jeff Garzika1213492005-09-28 13:26:47 -0400540 <sect1><title>How commands are issued</title>
Tejun Heobfd00722005-09-26 11:28:47 +0900541
542 <variablelist>
543
544 <varlistentry><term>Internal commands</term>
545 <listitem>
546 <para>
547 First, qc is allocated and initialized using
548 ata_qc_new_init(). Although ata_qc_new_init() doesn't
549 implement any wait or retry mechanism when qc is not
550 available, internal commands are currently issued only during
551 initialization and error recovery, so no other command is
552 active and allocation is guaranteed to succeed.
553 </para>
554 <para>
555 Once allocated qc's taskfile is initialized for the command to
556 be executed. qc currently has two mechanisms to notify
557 completion. One is via qc->complete_fn() callback and the
558 other is completion qc->waiting. qc->complete_fn() callback
559 is the asynchronous path used by normal SCSI translated
560 commands and qc->waiting is the synchronous (issuer sleeps in
561 process context) path used by internal commands.
562 </para>
563 <para>
564 Once initialization is complete, host_set lock is acquired
565 and the qc is issued.
566 </para>
567 </listitem>
568 </varlistentry>
569
570 <varlistentry><term>SCSI commands</term>
571 <listitem>
572 <para>
573 All libata drivers use ata_scsi_queuecmd() as
574 hostt->queuecommand callback. scmds can either be simulated
575 or translated. No qc is involved in processing a simulated
576 scmd. The result is computed right away and the scmd is
577 completed.
578 </para>
579 <para>
580 For a translated scmd, ata_qc_new_init() is invoked to
581 allocate a qc and the scmd is translated into the qc. SCSI
582 midlayer's completion notification function pointer is stored
583 into qc->scsidone.
584 </para>
585 <para>
586 qc->complete_fn() callback is used for completion
587 notification. ATA commands use ata_scsi_qc_complete() while
588 ATAPI commands use atapi_qc_complete(). Both functions end up
589 calling qc->scsidone to notify upper layer when the qc is
590 finished. After translation is completed, the qc is issued
591 with ata_qc_issue().
592 </para>
593 <para>
594 Note that SCSI midlayer invokes hostt->queuecommand while
595 holding host_set lock, so all above occur while holding
596 host_set lock.
597 </para>
598 </listitem>
599 </varlistentry>
600
601 </variablelist>
Jeff Garzika1213492005-09-28 13:26:47 -0400602 </sect1>
Tejun Heobfd00722005-09-26 11:28:47 +0900603
Jeff Garzika1213492005-09-28 13:26:47 -0400604 <sect1><title>How commands are processed</title>
Tejun Heobfd00722005-09-26 11:28:47 +0900605 <para>
606 Depending on which protocol and which controller are used,
607 commands are processed differently. For the purpose of
608 discussion, a controller which uses taskfile interface and all
609 standard callbacks is assumed.
610 </para>
611 <para>
612 Currently 6 ATA command protocols are used. They can be
613 sorted into the following four categories according to how
614 they are processed.
615 </para>
616
617 <variablelist>
618 <varlistentry><term>ATA NO DATA or DMA</term>
619 <listitem>
620 <para>
621 ATA_PROT_NODATA and ATA_PROT_DMA fall into this category.
622 These types of commands don't require any software
623 intervention once issued. Device will raise interrupt on
624 completion.
625 </para>
626 </listitem>
627 </varlistentry>
628
629 <varlistentry><term>ATA PIO</term>
630 <listitem>
631 <para>
632 ATA_PROT_PIO is in this category. libata currently
633 implements PIO with polling. ATA_NIEN bit is set to turn
634 off interrupt and pio_task on ata_wq performs polling and
635 IO.
636 </para>
637 </listitem>
638 </varlistentry>
639
640 <varlistentry><term>ATAPI NODATA or DMA</term>
641 <listitem>
642 <para>
643 ATA_PROT_ATAPI_NODATA and ATA_PROT_ATAPI_DMA are in this
644 category. packet_task is used to poll BSY bit after
645 issuing PACKET command. Once BSY is turned off by the
646 device, packet_task transfers CDB and hands off processing
647 to interrupt handler.
648 </para>
649 </listitem>
650 </varlistentry>
651
652 <varlistentry><term>ATAPI PIO</term>
653 <listitem>
654 <para>
655 ATA_PROT_ATAPI is in this category. ATA_NIEN bit is set
656 and, as in ATAPI NODATA or DMA, packet_task submits cdb.
657 However, after submitting cdb, further processing (data
658 transfer) is handed off to pio_task.
659 </para>
660 </listitem>
661 </varlistentry>
662 </variablelist>
Jeff Garzika1213492005-09-28 13:26:47 -0400663 </sect1>
Tejun Heobfd00722005-09-26 11:28:47 +0900664
Jeff Garzika1213492005-09-28 13:26:47 -0400665 <sect1><title>How commands are completed</title>
Tejun Heobfd00722005-09-26 11:28:47 +0900666 <para>
667 Once issued, all qc's are either completed with
668 ata_qc_complete() or time out. For commands which are handled
669 by interrupts, ata_host_intr() invokes ata_qc_complete(), and,
670 for PIO tasks, pio_task invokes ata_qc_complete(). In error
671 cases, packet_task may also complete commands.
672 </para>
673 <para>
674 ata_qc_complete() does the following.
675 </para>
676
677 <orderedlist>
678
679 <listitem>
680 <para>
681 DMA memory is unmapped.
682 </para>
683 </listitem>
684
685 <listitem>
686 <para>
687 ATA_QCFLAG_ACTIVE is clared from qc->flags.
688 </para>
689 </listitem>
690
691 <listitem>
692 <para>
693 qc->complete_fn() callback is invoked. If the return value of
694 the callback is not zero. Completion is short circuited and
695 ata_qc_complete() returns.
696 </para>
697 </listitem>
698
699 <listitem>
700 <para>
701 __ata_qc_complete() is called, which does
702 <orderedlist>
703
704 <listitem>
705 <para>
706 qc->flags is cleared to zero.
707 </para>
708 </listitem>
709
710 <listitem>
711 <para>
712 ap->active_tag and qc->tag are poisoned.
713 </para>
714 </listitem>
715
716 <listitem>
717 <para>
718 qc->waiting is claread &amp; completed (in that order).
719 </para>
720 </listitem>
721
722 <listitem>
723 <para>
724 qc is deallocated by clearing appropriate bit in ap->qactive.
725 </para>
726 </listitem>
727
728 </orderedlist>
729 </para>
730 </listitem>
731
732 </orderedlist>
733
734 <para>
735 So, it basically notifies upper layer and deallocates qc. One
736 exception is short-circuit path in #3 which is used by
737 atapi_qc_complete().
738 </para>
739 <para>
740 For all non-ATAPI commands, whether it fails or not, almost
741 the same code path is taken and very little error handling
742 takes place. A qc is completed with success status if it
743 succeeded, with failed status otherwise.
744 </para>
745 <para>
746 However, failed ATAPI commands require more handling as
747 REQUEST SENSE is needed to acquire sense data. If an ATAPI
748 command fails, ata_qc_complete() is invoked with error status,
749 which in turn invokes atapi_qc_complete() via
750 qc->complete_fn() callback.
751 </para>
752 <para>
753 This makes atapi_qc_complete() set scmd->result to
754 SAM_STAT_CHECK_CONDITION, complete the scmd and return 1. As
755 the sense data is empty but scmd->result is CHECK CONDITION,
756 SCSI midlayer will invoke EH for the scmd, and returning 1
757 makes ata_qc_complete() to return without deallocating the qc.
758 This leads us to ata_scsi_error() with partially completed qc.
759 </para>
760
Jeff Garzika1213492005-09-28 13:26:47 -0400761 </sect1>
Tejun Heobfd00722005-09-26 11:28:47 +0900762
Jeff Garzika1213492005-09-28 13:26:47 -0400763 <sect1><title>ata_scsi_error()</title>
Tejun Heobfd00722005-09-26 11:28:47 +0900764 <para>
Christoph Hellwig9227c332006-04-01 19:21:04 +0200765 ata_scsi_error() is the current transportt->eh_strategy_handler()
Tejun Heobfd00722005-09-26 11:28:47 +0900766 for libata. As discussed above, this will be entered in two
767 cases - timeout and ATAPI error completion. This function
768 calls low level libata driver's eng_timeout() callback, the
769 standard callback for which is ata_eng_timeout(). It checks
770 if a qc is active and calls ata_qc_timeout() on the qc if so.
771 Actual error handling occurs in ata_qc_timeout().
772 </para>
773 <para>
774 If EH is invoked for timeout, ata_qc_timeout() stops BMDMA and
775 completes the qc. Note that as we're currently in EH, we
776 cannot call scsi_done. As described in SCSI EH doc, a
777 recovered scmd should be either retried with
778 scsi_queue_insert() or finished with scsi_finish_command().
779 Here, we override qc->scsidone with scsi_finish_command() and
780 calls ata_qc_complete().
781 </para>
782 <para>
783 If EH is invoked due to a failed ATAPI qc, the qc here is
784 completed but not deallocated. The purpose of this
785 half-completion is to use the qc as place holder to make EH
786 code reach this place. This is a bit hackish, but it works.
787 </para>
788 <para>
789 Once control reaches here, the qc is deallocated by invoking
790 __ata_qc_complete() explicitly. Then, internal qc for REQUEST
791 SENSE is issued. Once sense data is acquired, scmd is
792 finished by directly invoking scsi_finish_command() on the
793 scmd. Note that as we already have completed and deallocated
794 the qc which was associated with the scmd, we don't need
795 to/cannot call ata_qc_complete() again.
796 </para>
797
Jeff Garzika1213492005-09-28 13:26:47 -0400798 </sect1>
Tejun Heobfd00722005-09-26 11:28:47 +0900799
Jeff Garzika1213492005-09-28 13:26:47 -0400800 <sect1><title>Problems with the current EH</title>
Tejun Heobfd00722005-09-26 11:28:47 +0900801
802 <itemizedlist>
803
804 <listitem>
805 <para>
806 Error representation is too crude. Currently any and all
807 error conditions are represented with ATA STATUS and ERROR
808 registers. Errors which aren't ATA device errors are treated
809 as ATA device errors by setting ATA_ERR bit. Better error
810 descriptor which can properly represent ATA and other
811 errors/exceptions is needed.
812 </para>
813 </listitem>
814
815 <listitem>
816 <para>
817 When handling timeouts, no action is taken to make device
818 forget about the timed out command and ready for new commands.
819 </para>
820 </listitem>
821
822 <listitem>
823 <para>
824 EH handling via ata_scsi_error() is not properly protected
825 from usual command processing. On EH entrance, the device is
826 not in quiescent state. Timed out commands may succeed or
827 fail any time. pio_task and atapi_task may still be running.
828 </para>
829 </listitem>
830
831 <listitem>
832 <para>
833 Too weak error recovery. Devices / controllers causing HSM
834 mismatch errors and other errors quite often require reset to
835 return to known state. Also, advanced error handling is
836 necessary to support features like NCQ and hotplug.
837 </para>
838 </listitem>
839
840 <listitem>
841 <para>
842 ATA errors are directly handled in the interrupt handler and
843 PIO errors in pio_task. This is problematic for advanced
844 error handling for the following reasons.
845 </para>
846 <para>
847 First, advanced error handling often requires context and
848 internal qc execution.
849 </para>
850 <para>
851 Second, even a simple failure (say, CRC error) needs
852 information gathering and could trigger complex error handling
853 (say, resetting &amp; reconfiguring). Having multiple code
854 paths to gather information, enter EH and trigger actions
855 makes life painful.
856 </para>
857 <para>
858 Third, scattered EH code makes implementing low level drivers
859 difficult. Low level drivers override libata callbacks. If
860 EH is scattered over several places, each affected callbacks
861 should perform its part of error handling. This can be error
862 prone and painful.
863 </para>
864 </listitem>
865
866 </itemizedlist>
Jeff Garzika1213492005-09-28 13:26:47 -0400867 </sect1>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700868 </chapter>
869
870 <chapter id="libataExt">
871 <title>libata Library</title>
Henrik Kretzschmar58707cc2006-08-21 18:39:26 -0700872!Edrivers/ata/libata-core.c
Linus Torvalds1da177e2005-04-16 15:20:36 -0700873 </chapter>
874
875 <chapter id="libataInt">
876 <title>libata Core Internals</title>
Henrik Kretzschmar58707cc2006-08-21 18:39:26 -0700877!Idrivers/ata/libata-core.c
Linus Torvalds1da177e2005-04-16 15:20:36 -0700878 </chapter>
879
880 <chapter id="libataScsiInt">
881 <title>libata SCSI translation/emulation</title>
Henrik Kretzschmar58707cc2006-08-21 18:39:26 -0700882!Edrivers/ata/libata-scsi.c
883!Idrivers/ata/libata-scsi.c
Linus Torvalds1da177e2005-04-16 15:20:36 -0700884 </chapter>
885
Tejun Heofe998aa2005-10-02 11:54:29 +0900886 <chapter id="ataExceptions">
Robert P. J. Day7db51fb2007-01-08 02:22:03 -0500887 <title>ATA errors and exceptions</title>
Tejun Heofe998aa2005-10-02 11:54:29 +0900888
889 <para>
890 This chapter tries to identify what error/exception conditions exist
891 for ATA/ATAPI devices and describe how they should be handled in
892 implementation-neutral way.
893 </para>
894
895 <para>
896 The term 'error' is used to describe conditions where either an
897 explicit error condition is reported from device or a command has
898 timed out.
899 </para>
900
901 <para>
902 The term 'exception' is either used to describe exceptional
903 conditions which are not errors (say, power or hotplug events), or
904 to describe both errors and non-error exceptional conditions. Where
905 explicit distinction between error and exception is necessary, the
906 term 'non-error exception' is used.
907 </para>
908
909 <sect1 id="excat">
910 <title>Exception categories</title>
911 <para>
912 Exceptions are described primarily with respect to legacy
913 taskfile + bus master IDE interface. If a controller provides
914 other better mechanism for error reporting, mapping those into
915 categories described below shouldn't be difficult.
916 </para>
917
918 <para>
919 In the following sections, two recovery actions - reset and
920 reconfiguring transport - are mentioned. These are described
921 further in <xref linkend="exrec"/>.
922 </para>
923
924 <sect2 id="excatHSMviolation">
925 <title>HSM violation</title>
926 <para>
927 This error is indicated when STATUS value doesn't match HSM
928 requirement during issuing or excution any ATA/ATAPI command.
929 </para>
930
931 <itemizedlist>
932 <title>Examples</title>
933
934 <listitem>
935 <para>
936 ATA_STATUS doesn't contain !BSY &amp;&amp; DRDY &amp;&amp; !DRQ while trying
937 to issue a command.
938 </para>
939 </listitem>
940
941 <listitem>
942 <para>
943 !BSY &amp;&amp; !DRQ during PIO data transfer.
944 </para>
945 </listitem>
946
947 <listitem>
948 <para>
949 DRQ on command completion.
950 </para>
951 </listitem>
952
953 <listitem>
954 <para>
955 !BSY &amp;&amp; ERR after CDB tranfer starts but before the
956 last byte of CDB is transferred. ATA/ATAPI standard states
957 that &quot;The device shall not terminate the PACKET command
958 with an error before the last byte of the command packet has
959 been written&quot; in the error outputs description of PACKET
960 command and the state diagram doesn't include such
961 transitions.
962 </para>
963 </listitem>
964
965 </itemizedlist>
966
967 <para>
968 In these cases, HSM is violated and not much information
969 regarding the error can be acquired from STATUS or ERROR
970 register. IOW, this error can be anything - driver bug,
971 faulty device, controller and/or cable.
972 </para>
973
974 <para>
975 As HSM is violated, reset is necessary to restore known state.
976 Reconfiguring transport for lower speed might be helpful too
977 as transmission errors sometimes cause this kind of errors.
978 </para>
979 </sect2>
980
981 <sect2 id="excatDevErr">
982 <title>ATA/ATAPI device error (non-NCQ / non-CHECK CONDITION)</title>
983
984 <para>
985 These are errors detected and reported by ATA/ATAPI devices
986 indicating device problems. For this type of errors, STATUS
987 and ERROR register values are valid and describe error
988 condition. Note that some of ATA bus errors are detected by
989 ATA/ATAPI devices and reported using the same mechanism as
990 device errors. Those cases are described later in this
991 section.
992 </para>
993
994 <para>
995 For ATA commands, this type of errors are indicated by !BSY
996 &amp;&amp; ERR during command execution and on completion.
997 </para>
998
999 <para>For ATAPI commands,</para>
1000
1001 <itemizedlist>
1002
1003 <listitem>
1004 <para>
1005 !BSY &amp;&amp; ERR &amp;&amp; ABRT right after issuing PACKET
1006 indicates that PACKET command is not supported and falls in
1007 this category.
1008 </para>
1009 </listitem>
1010
1011 <listitem>
1012 <para>
1013 !BSY &amp;&amp; ERR(==CHK) &amp;&amp; !ABRT after the last
1014 byte of CDB is transferred indicates CHECK CONDITION and
1015 doesn't fall in this category.
1016 </para>
1017 </listitem>
1018
1019 <listitem>
1020 <para>
1021 !BSY &amp;&amp; ERR(==CHK) &amp;&amp; ABRT after the last byte
1022 of CDB is transferred *probably* indicates CHECK CONDITION and
1023 doesn't fall in this category.
1024 </para>
1025 </listitem>
1026
1027 </itemizedlist>
1028
1029 <para>
1030 Of errors detected as above, the followings are not ATA/ATAPI
1031 device errors but ATA bus errors and should be handled
1032 according to <xref linkend="excatATAbusErr"/>.
1033 </para>
1034
1035 <variablelist>
1036
1037 <varlistentry>
1038 <term>CRC error during data transfer</term>
1039 <listitem>
1040 <para>
1041 This is indicated by ICRC bit in the ERROR register and
1042 means that corruption occurred during data transfer. Upto
1043 ATA/ATAPI-7, the standard specifies that this bit is only
1044 applicable to UDMA transfers but ATA/ATAPI-8 draft revision
1045 1f says that the bit may be applicable to multiword DMA and
1046 PIO.
1047 </para>
1048 </listitem>
1049 </varlistentry>
1050
1051 <varlistentry>
1052 <term>ABRT error during data transfer or on completion</term>
1053 <listitem>
1054 <para>
1055 Upto ATA/ATAPI-7, the standard specifies that ABRT could be
1056 set on ICRC errors and on cases where a device is not able
1057 to complete a command. Combined with the fact that MWDMA
1058 and PIO transfer errors aren't allowed to use ICRC bit upto
1059 ATA/ATAPI-7, it seems to imply that ABRT bit alone could
1060 indicate tranfer errors.
1061 </para>
1062 <para>
1063 However, ATA/ATAPI-8 draft revision 1f removes the part
1064 that ICRC errors can turn on ABRT. So, this is kind of
1065 gray area. Some heuristics are needed here.
1066 </para>
1067 </listitem>
1068 </varlistentry>
1069
1070 </variablelist>
1071
1072 <para>
1073 ATA/ATAPI device errors can be further categorized as follows.
1074 </para>
1075
1076 <variablelist>
1077
1078 <varlistentry>
1079 <term>Media errors</term>
1080 <listitem>
1081 <para>
1082 This is indicated by UNC bit in the ERROR register. ATA
1083 devices reports UNC error only after certain number of
1084 retries cannot recover the data, so there's nothing much
1085 else to do other than notifying upper layer.
1086 </para>
1087 <para>
1088 READ and WRITE commands report CHS or LBA of the first
1089 failed sector but ATA/ATAPI standard specifies that the
1090 amount of transferred data on error completion is
1091 indeterminate, so we cannot assume that sectors preceding
1092 the failed sector have been transferred and thus cannot
1093 complete those sectors successfully as SCSI does.
1094 </para>
1095 </listitem>
1096 </varlistentry>
1097
1098 <varlistentry>
1099 <term>Media changed / media change requested error</term>
1100 <listitem>
1101 <para>
1102 &lt;&lt;TODO: fill here&gt;&gt;
1103 </para>
1104 </listitem>
1105 </varlistentry>
1106
1107 <varlistentry><term>Address error</term>
1108 <listitem>
1109 <para>
1110 This is indicated by IDNF bit in the ERROR register.
1111 Report to upper layer.
1112 </para>
1113 </listitem>
1114 </varlistentry>
1115
1116 <varlistentry><term>Other errors</term>
1117 <listitem>
1118 <para>
1119 This can be invalid command or parameter indicated by ABRT
1120 ERROR bit or some other error condition. Note that ABRT
1121 bit can indicate a lot of things including ICRC and Address
1122 errors. Heuristics needed.
1123 </para>
1124 </listitem>
1125 </varlistentry>
1126
1127 </variablelist>
1128
1129 <para>
1130 Depending on commands, not all STATUS/ERROR bits are
1131 applicable. These non-applicable bits are marked with
1132 &quot;na&quot; in the output descriptions but upto ATA/ATAPI-7
1133 no definition of &quot;na&quot; can be found. However,
1134 ATA/ATAPI-8 draft revision 1f describes &quot;N/A&quot; as
1135 follows.
1136 </para>
1137
1138 <blockquote>
1139 <variablelist>
1140 <varlistentry><term>3.2.3.3a N/A</term>
1141 <listitem>
1142 <para>
1143 A keyword the indicates a field has no defined value in
1144 this standard and should not be checked by the host or
1145 device. N/A fields should be cleared to zero.
1146 </para>
1147 </listitem>
1148 </varlistentry>
1149 </variablelist>
1150 </blockquote>
1151
1152 <para>
1153 So, it seems reasonable to assume that &quot;na&quot; bits are
1154 cleared to zero by devices and thus need no explicit masking.
1155 </para>
1156
1157 </sect2>
1158
1159 <sect2 id="excatATAPIcc">
1160 <title>ATAPI device CHECK CONDITION</title>
1161
1162 <para>
1163 ATAPI device CHECK CONDITION error is indicated by set CHK bit
1164 (ERR bit) in the STATUS register after the last byte of CDB is
1165 transferred for a PACKET command. For this kind of errors,
1166 sense data should be acquired to gather information regarding
1167 the errors. REQUEST SENSE packet command should be used to
1168 acquire sense data.
1169 </para>
1170
1171 <para>
1172 Once sense data is acquired, this type of errors can be
1173 handled similary to other SCSI errors. Note that sense data
1174 may indicate ATA bus error (e.g. Sense Key 04h HARDWARE ERROR
1175 &amp;&amp; ASC/ASCQ 47h/00h SCSI PARITY ERROR). In such
1176 cases, the error should be considered as an ATA bus error and
1177 handled according to <xref linkend="excatATAbusErr"/>.
1178 </para>
1179
1180 </sect2>
1181
1182 <sect2 id="excatNCQerr">
1183 <title>ATA device error (NCQ)</title>
1184
1185 <para>
1186 NCQ command error is indicated by cleared BSY and set ERR bit
1187 during NCQ command phase (one or more NCQ commands
1188 outstanding). Although STATUS and ERROR registers will
1189 contain valid values describing the error, READ LOG EXT is
1190 required to clear the error condition, determine which command
1191 has failed and acquire more information.
1192 </para>
1193
1194 <para>
1195 READ LOG EXT Log Page 10h reports which tag has failed and
1196 taskfile register values describing the error. With this
1197 information the failed command can be handled as a normal ATA
1198 command error as in <xref linkend="excatDevErr"/> and all
1199 other in-flight commands must be retried. Note that this
1200 retry should not be counted - it's likely that commands
1201 retried this way would have completed normally if it were not
1202 for the failed command.
1203 </para>
1204
1205 <para>
1206 Note that ATA bus errors can be reported as ATA device NCQ
1207 errors. This should be handled as described in <xref
1208 linkend="excatATAbusErr"/>.
1209 </para>
1210
1211 <para>
1212 If READ LOG EXT Log Page 10h fails or reports NQ, we're
1213 thoroughly screwed. This condition should be treated
1214 according to <xref linkend="excatHSMviolation"/>.
1215 </para>
1216
1217 </sect2>
1218
1219 <sect2 id="excatATAbusErr">
1220 <title>ATA bus error</title>
1221
1222 <para>
1223 ATA bus error means that data corruption occurred during
1224 transmission over ATA bus (SATA or PATA). This type of errors
1225 can be indicated by
1226 </para>
1227
1228 <itemizedlist>
1229
1230 <listitem>
1231 <para>
1232 ICRC or ABRT error as described in <xref linkend="excatDevErr"/>.
1233 </para>
1234 </listitem>
1235
1236 <listitem>
1237 <para>
1238 Controller-specific error completion with error information
1239 indicating transmission error.
1240 </para>
1241 </listitem>
1242
1243 <listitem>
1244 <para>
1245 On some controllers, command timeout. In this case, there may
1246 be a mechanism to determine that the timeout is due to
1247 transmission error.
1248 </para>
1249 </listitem>
1250
1251 <listitem>
1252 <para>
1253 Unknown/random errors, timeouts and all sorts of weirdities.
1254 </para>
1255 </listitem>
1256
1257 </itemizedlist>
1258
1259 <para>
1260 As described above, transmission errors can cause wide variety
1261 of symptoms ranging from device ICRC error to random device
1262 lockup, and, for many cases, there is no way to tell if an
1263 error condition is due to transmission error or not;
1264 therefore, it's necessary to employ some kind of heuristic
1265 when dealing with errors and timeouts. For example,
1266 encountering repetitive ABRT errors for known supported
1267 command is likely to indicate ATA bus error.
1268 </para>
1269
1270 <para>
1271 Once it's determined that ATA bus errors have possibly
1272 occurred, lowering ATA bus transmission speed is one of
1273 actions which may alleviate the problem. See <xref
1274 linkend="exrecReconf"/> for more information.
1275 </para>
1276
1277 </sect2>
1278
1279 <sect2 id="excatPCIbusErr">
1280 <title>PCI bus error</title>
1281
1282 <para>
1283 Data corruption or other failures during transmission over PCI
1284 (or other system bus). For standard BMDMA, this is indicated
1285 by Error bit in the BMDMA Status register. This type of
1286 errors must be logged as it indicates something is very wrong
1287 with the system. Resetting host controller is recommended.
1288 </para>
1289
1290 </sect2>
1291
1292 <sect2 id="excatLateCompletion">
1293 <title>Late completion</title>
1294
1295 <para>
1296 This occurs when timeout occurs and the timeout handler finds
1297 out that the timed out command has completed successfully or
1298 with error. This is usually caused by lost interrupts. This
1299 type of errors must be logged. Resetting host controller is
1300 recommended.
1301 </para>
1302
1303 </sect2>
1304
1305 <sect2 id="excatUnknown">
1306 <title>Unknown error (timeout)</title>
1307
1308 <para>
1309 This is when timeout occurs and the command is still
1310 processing or the host and device are in unknown state. When
1311 this occurs, HSM could be in any valid or invalid state. To
1312 bring the device to known state and make it forget about the
1313 timed out command, resetting is necessary. The timed out
1314 command may be retried.
1315 </para>
1316
1317 <para>
1318 Timeouts can also be caused by transmission errors. Refer to
1319 <xref linkend="excatATAbusErr"/> for more details.
1320 </para>
1321
1322 </sect2>
1323
1324 <sect2 id="excatHoplugPM">
1325 <title>Hotplug and power management exceptions</title>
1326
1327 <para>
1328 &lt;&lt;TODO: fill here&gt;&gt;
1329 </para>
1330
1331 </sect2>
1332
1333 </sect1>
1334
1335 <sect1 id="exrec">
1336 <title>EH recovery actions</title>
1337
1338 <para>
1339 This section discusses several important recovery actions.
1340 </para>
1341
1342 <sect2 id="exrecClr">
1343 <title>Clearing error condition</title>
1344
1345 <para>
1346 Many controllers require its error registers to be cleared by
1347 error handler. Different controllers may have different
1348 requirements.
1349 </para>
1350
1351 <para>
1352 For SATA, it's strongly recommended to clear at least SError
1353 register during error handling.
1354 </para>
1355 </sect2>
1356
1357 <sect2 id="exrecRst">
1358 <title>Reset</title>
1359
1360 <para>
1361 During EH, resetting is necessary in the following cases.
1362 </para>
1363
1364 <itemizedlist>
1365
1366 <listitem>
1367 <para>
1368 HSM is in unknown or invalid state
1369 </para>
1370 </listitem>
1371
1372 <listitem>
1373 <para>
1374 HBA is in unknown or invalid state
1375 </para>
1376 </listitem>
1377
1378 <listitem>
1379 <para>
1380 EH needs to make HBA/device forget about in-flight commands
1381 </para>
1382 </listitem>
1383
1384 <listitem>
1385 <para>
1386 HBA/device behaves weirdly
1387 </para>
1388 </listitem>
1389
1390 </itemizedlist>
1391
1392 <para>
1393 Resetting during EH might be a good idea regardless of error
1394 condition to improve EH robustness. Whether to reset both or
1395 either one of HBA and device depends on situation but the
1396 following scheme is recommended.
1397 </para>
1398
1399 <itemizedlist>
1400
1401 <listitem>
1402 <para>
1403 When it's known that HBA is in ready state but ATA/ATAPI
Paolo Ornati670e9f32006-10-03 22:57:56 +02001404 device is in unknown state, reset only device.
Tejun Heofe998aa2005-10-02 11:54:29 +09001405 </para>
1406 </listitem>
1407
1408 <listitem>
1409 <para>
1410 If HBA is in unknown state, reset both HBA and device.
1411 </para>
1412 </listitem>
1413
1414 </itemizedlist>
1415
1416 <para>
1417 HBA resetting is implementation specific. For a controller
1418 complying to taskfile/BMDMA PCI IDE, stopping active DMA
1419 transaction may be sufficient iff BMDMA state is the only HBA
1420 context. But even mostly taskfile/BMDMA PCI IDE complying
1421 controllers may have implementation specific requirements and
1422 mechanism to reset themselves. This must be addressed by
1423 specific drivers.
1424 </para>
1425
1426 <para>
1427 OTOH, ATA/ATAPI standard describes in detail ways to reset
1428 ATA/ATAPI devices.
1429 </para>
1430
1431 <variablelist>
1432
1433 <varlistentry><term>PATA hardware reset</term>
1434 <listitem>
1435 <para>
1436 This is hardware initiated device reset signalled with
1437 asserted PATA RESET- signal. There is no standard way to
1438 initiate hardware reset from software although some
1439 hardware provides registers that allow driver to directly
1440 tweak the RESET- signal.
1441 </para>
1442 </listitem>
1443 </varlistentry>
1444
1445 <varlistentry><term>Software reset</term>
1446 <listitem>
1447 <para>
1448 This is achieved by turning CONTROL SRST bit on for at
1449 least 5us. Both PATA and SATA support it but, in case of
1450 SATA, this may require controller-specific support as the
1451 second Register FIS to clear SRST should be transmitted
1452 while BSY bit is still set. Note that on PATA, this resets
1453 both master and slave devices on a channel.
1454 </para>
1455 </listitem>
1456 </varlistentry>
1457
1458 <varlistentry><term>EXECUTE DEVICE DIAGNOSTIC command</term>
1459 <listitem>
1460 <para>
1461 Although ATA/ATAPI standard doesn't describe exactly, EDD
1462 implies some level of resetting, possibly similar level
1463 with software reset. Host-side EDD protocol can be handled
1464 with normal command processing and most SATA controllers
1465 should be able to handle EDD's just like other commands.
1466 As in software reset, EDD affects both devices on a PATA
1467 bus.
1468 </para>
1469 <para>
1470 Although EDD does reset devices, this doesn't suit error
1471 handling as EDD cannot be issued while BSY is set and it's
1472 unclear how it will act when device is in unknown/weird
1473 state.
1474 </para>
1475 </listitem>
1476 </varlistentry>
1477
1478 <varlistentry><term>ATAPI DEVICE RESET command</term>
1479 <listitem>
1480 <para>
1481 This is very similar to software reset except that reset
1482 can be restricted to the selected device without affecting
1483 the other device sharing the cable.
1484 </para>
1485 </listitem>
1486 </varlistentry>
1487
1488 <varlistentry><term>SATA phy reset</term>
1489 <listitem>
1490 <para>
1491 This is the preferred way of resetting a SATA device. In
1492 effect, it's identical to PATA hardware reset. Note that
1493 this can be done with the standard SCR Control register.
1494 As such, it's usually easier to implement than software
1495 reset.
1496 </para>
1497 </listitem>
1498 </varlistentry>
1499
1500 </variablelist>
1501
1502 <para>
1503 One more thing to consider when resetting devices is that
1504 resetting clears certain configuration parameters and they
1505 need to be set to their previous or newly adjusted values
1506 after reset.
1507 </para>
1508
1509 <para>
1510 Parameters affected are.
1511 </para>
1512
1513 <itemizedlist>
1514
1515 <listitem>
1516 <para>
1517 CHS set up with INITIALIZE DEVICE PARAMETERS (seldomly used)
1518 </para>
1519 </listitem>
1520
1521 <listitem>
1522 <para>
1523 Parameters set with SET FEATURES including transfer mode setting
1524 </para>
1525 </listitem>
1526
1527 <listitem>
1528 <para>
1529 Block count set with SET MULTIPLE MODE
1530 </para>
1531 </listitem>
1532
1533 <listitem>
1534 <para>
1535 Other parameters (SET MAX, MEDIA LOCK...)
1536 </para>
1537 </listitem>
1538
1539 </itemizedlist>
1540
1541 <para>
1542 ATA/ATAPI standard specifies that some parameters must be
1543 maintained across hardware or software reset, but doesn't
1544 strictly specify all of them. Always reconfiguring needed
1545 parameters after reset is required for robustness. Note that
1546 this also applies when resuming from deep sleep (power-off).
1547 </para>
1548
1549 <para>
1550 Also, ATA/ATAPI standard requires that IDENTIFY DEVICE /
1551 IDENTIFY PACKET DEVICE is issued after any configuration
1552 parameter is updated or a hardware reset and the result used
1553 for further operation. OS driver is required to implement
1554 revalidation mechanism to support this.
1555 </para>
1556
1557 </sect2>
1558
1559 <sect2 id="exrecReconf">
1560 <title>Reconfigure transport</title>
1561
1562 <para>
1563 For both PATA and SATA, a lot of corners are cut for cheap
1564 connectors, cables or controllers and it's quite common to see
1565 high transmission error rate. This can be mitigated by
1566 lowering transmission speed.
1567 </para>
1568
1569 <para>
1570 The following is a possible scheme Jeff Garzik suggested.
1571 </para>
1572
1573 <blockquote>
1574 <para>
1575 If more than $N (3?) transmission errors happen in 15 minutes,
1576 </para>
1577 <itemizedlist>
1578 <listitem>
1579 <para>
1580 if SATA, decrease SATA PHY speed. if speed cannot be decreased,
1581 </para>
1582 </listitem>
1583 <listitem>
1584 <para>
1585 decrease UDMA xfer speed. if at UDMA0, switch to PIO4,
1586 </para>
1587 </listitem>
1588 <listitem>
1589 <para>
1590 decrease PIO xfer speed. if at PIO3, complain, but continue
1591 </para>
1592 </listitem>
1593 </itemizedlist>
1594 </blockquote>
1595
1596 </sect2>
1597
1598 </sect1>
1599
1600 </chapter>
1601
Linus Torvalds1da177e2005-04-16 15:20:36 -07001602 <chapter id="PiixInt">
1603 <title>ata_piix Internals</title>
Henrik Kretzschmar58707cc2006-08-21 18:39:26 -07001604!Idrivers/ata/ata_piix.c
Linus Torvalds1da177e2005-04-16 15:20:36 -07001605 </chapter>
1606
1607 <chapter id="SILInt">
1608 <title>sata_sil Internals</title>
Henrik Kretzschmar58707cc2006-08-21 18:39:26 -07001609!Idrivers/ata/sata_sil.c
Linus Torvalds1da177e2005-04-16 15:20:36 -07001610 </chapter>
1611
Jeff Garzik0cba6322005-05-30 19:49:12 -04001612 <chapter id="libataThanks">
1613 <title>Thanks</title>
1614 <para>
1615 The bulk of the ATA knowledge comes thanks to long conversations with
1616 Andre Hedrick (www.linux-ide.org), and long hours pondering the ATA
1617 and SCSI specifications.
1618 </para>
1619 <para>
1620 Thanks to Alan Cox for pointing out similarities
1621 between SATA and SCSI, and in general for motivation to hack on
1622 libata.
1623 </para>
1624 <para>
1625 libata's device detection
1626 method, ata_pio_devchk, and in general all the early probing was
1627 based on extensive study of Hale Landis's probe/reset code in his
1628 ATADRVR driver (www.ata-atapi.com).
1629 </para>
1630 </chapter>
1631
Linus Torvalds1da177e2005-04-16 15:20:36 -07001632</book>