| blob | 45d61ad8c6f7b3f7d0981216fc093bdc286c82fd |
2 -=< The IBM Microchannel SCSI-Subsystem >=-
4 for the IBM PS/2 series
6 Low Level Software-Driver for Linux
8 Copyright (c) 1995 Strom Systems, Inc. under the terms of the GNU
9 General Public License. Originally written by Martin Kolinek, December 1995.
10 Officially modified and maintained by Michael Lang since January 1999.
12 Version 4.0a
14 Last update: January 3, 2001
16 Before you Start
17 ----------------
18 This is the common README.ibmmca file for all driver releases of the
19 IBM MCA SCSI driver for Linux. Please note, that driver releases 4.0
20 or newer do not work with kernel versions older than 2.4.0, while driver
21 versions older than 4.0 do not work with kernels 2.4.0 or later! If you
22 try to compile your kernel with the wrong driver source, the
23 compilation is aborted and you get a corresponding error message. This is
24 no bug in the driver; it prevents you from using the wrong source code
25 with the wrong kernel version.
27 Authors of this Driver
28 ----------------------
29 - Chris Beauregard (improvement of the SCSI-device mapping by the driver)
30 - Martin Kolinek (origin, first release of this driver)
31 - Klaus Kudielka (multiple SCSI-host management/detection, adaption to
32 Linux Kernel 2.1.x, module support)
33 - Michael Lang (assigning original pun/lun mapping, dynamical ldn
34 assignment, rewritten adapter detection, this file,
35 patches, official driver maintenance and subsequent
36 debugging, related with the driver)
38 Table of Contents
39 -----------------
40 1 Abstract
41 2 Driver Description
42 2.1 IBM SCSI-Subsystem Detection
43 2.2 Physical Units, Logical Units, and Logical Devices
44 2.3 SCSI-Device Recognition and dynamical ldn Assignment
45 2.4 SCSI-Device Order
46 2.5 Regular SCSI-Command-Processing
47 2.6 Abort & Reset Commands
48 2.7 Disk Geometry
49 2.8 Kernel Boot Option
50 2.9 Driver Module Support
51 2.10 Multiple Hostadapter Support
52 2.11 /proc/scsi-Filesystem Information
53 2.12 /proc/mca-Filesystem Information
54 2.13 Supported IBM SCSI-Subsystems
55 2.14 Linux Kernel Versions
56 3 Code History
57 4 To do
58 5 Users' Manual
59 5.1 Commandline Parameters
60 5.2 Troubleshooting
61 5.3 Bug reports
62 5.4 Support WWW-page
63 6 References
64 7 Credits to
65 7.1 People
66 7.2 Sponsors & Supporters
67 8 Trademarks
68 9 Disclaimer
70 * * *
72 1 Abstract
73 ----------
74 This README-file describes the IBM SCSI-subsystem low level driver for
75 Linux. The descriptions which were formerly kept in the source code have
76 been taken out of this file to simplify the codes readability. The driver
77 description has been updated, as most of the former description was already
78 quite outdated. The history of the driver development is also kept inside
79 here. Multiple historical developments have been summarized to shorten the
80 text size a bit. At the end of this file you can find a small manual for
81 this driver and hints to get it running on your machine.
83 2 Driver Description
84 --------------------
85 2.1 IBM SCSI-Subsystem Detection
86 --------------------------------
87 This is done in the ibmmca_detect() function. It first checks, if the
88 Microchannel-bus support is enabled, as the IBM SCSI-subsystem needs the
89 Microchannel. In a next step, a free interrupt is chosen and the main
90 interrupt handler is connected to it to handle answers of the SCSI-
91 subsystem(s). If the F/W SCSI-adapter is forced by the BIOS to use IRQ11
92 instead of IRQ14, IRQ11 is used for the IBM SCSI-2 F/W adapter. In a
93 further step it is checked, if the adapter gets detected by force from
94 the kernel commandline, where the I/O port and the SCSI-subsystem id can
95 be specified. The next step checks if there is an integrated SCSI-subsystem
96 installed. This register area is fixed through all IBM PS/2 MCA-machines
97 and appears as something like a virtual slot 10 of the MCA-bus. On most
98 PS/2 machines, the POS registers of slot 10 are set to 0xff or 0x00 if not
99 integrated SCSI-controller is available. But on certain PS/2s, like model
100 9595, this slot 10 is used to store other information which at earlier
101 stage confused the driver and resulted in the detection of some ghost-SCSI.
102 If POS-register 2 and 3 are not 0x00 and not 0xff, but all other POS
103 registers are either 0xff or 0x00, there must be an integrated SCSI-
104 subsystem present and it will be registered as IBM Integrated SCSI-
105 Subsystem. The next step checks, if there is a slot-adapter installed on
106 the MCA-bus. To get this, the first two POS-registers, that represent the
107 adapter ID are checked. If they fit to one of the ids, stored in the
108 adapter list, a SCSI-subsystem is assumed to be found in a slot and will be
109 registered. This check is done through all possible MCA-bus slots to allow
110 more than one SCSI-adapter to be present in the PS/2-system and this is
111 already the first point of problems. Looking into the technical reference
112 manual for the IBM PS/2 common interfaces, the POS2 register must have
113 different interpretation of its single bits to avoid overlapping I/O
114 regions. While one can assume, that the integrated subsystem has a fix
115 I/O-address at 0x3540 - 0x3547, further installed IBM SCSI-adapters must
116 use a different I/O-address. This is expressed by bit 1 to 3 of POS2
117 (multiplied by 8 + 0x3540). Bits 2 and 3 are reserved for the integrated
118 subsystem, but not for the adapters! The following list shows, how the
119 bits of POS2 and POS3 should be interpreted.
121 The POS2-register of all PS/2 models' integrated SCSI-subsystems has the
122 following interpretation of bits:
123 Bit 7 - 4 : Chip Revision ID (Release)
124 Bit 3 - 2 : Reserved
125 Bit 1 : 8k NVRAM Disabled
126 Bit 0 : Chip Enable (EN-Signal)
127 The POS3-register is interpreted as follows (for most IBM SCSI-subsys.):
128 Bit 7 - 5 : SCSI ID
129 Bit 4 - 0 : Reserved = 0
130 The slot-adapters have different interpretation of these bits. The IBM SCSI
131 adapter (w/Cache) and the IBM SCSI-2 F/W adapter use the following
132 interpretation of the POS2 register:
133 Bit 7 - 4 : ROM Segment Address Select
134 Bit 3 - 1 : Adapter I/O Address Select (*8+0x3540)
135 Bit 0 : Adapter Enable (EN-Signal)
136 and for the POS3 register:
137 Bit 7 - 5 : SCSI ID
138 Bit 4 : Fairness Enable (SCSI ID3 f. F/W)
139 Bit 3 - 0 : Arbitration Level
140 The most modern product of the series is the IBM SCSI-2 F/W adapter, it
141 allows dual-bus SCSI and SCSI-wide addressing, which means, PUNs may be
142 between 0 and 15. Here, Bit 4 is the high-order bit of the 4-bit wide
143 adapter PUN expression. In short words, this means, that IBM PS/2 machines
144 can only support 1 single integrated subsystem by default. Additional
145 slot-adapters get ports assigned by the automatic configuration tool.
147 One day I found a patch in ibmmca_detect(), forcing the I/O-address to be
148 0x3540 for integrated SCSI-subsystems, there was a remark placed, that on
149 integrated IBM SCSI-subsystems of model 56, the POS2 register was showing 5.
150 This means, that really for these models, POS2 has to be interpreted
151 sticking to the technical reference guide. In this case, the bit 2 (4) is
152 a reserved bit and may not be interpreted. These differences between the
153 adapters and the integrated controllers are taken into account by the
154 detection routine of the driver on from version >3.0g.
156 Every time, a SCSI-subsystem is discovered, the ibmmca_register() function
157 is called. This function checks first, if the requested area for the I/O-
158 address of this SCSI-subsystem is still available and assigns this I/O-
159 area to the SCSI-subsystem. There are always 8 sequential I/O-addresses
160 taken for each individual SCSI-subsystem found, which are:
162 Offset Type Permissions
163 0 Command Interface Register 1 Read/Write
164 1 Command Interface Register 2 Read/Write
165 2 Command Interface Register 3 Read/Write
166 3 Command Interface Register 4 Read/Write
167 4 Attention Register Read/Write
168 5 Basic Control Register Read/Write
169 6 Interrupt Status Register Read
170 7 Basic Status Register Read
172 After the I/O-address range is assigned, the host-adapter is assigned
173 to a local structure which keeps all adapter information needed for the
174 driver itself and the mid- and higher-level SCSI-drivers. The SCSI pun/lun
175 and the adapters' ldn tables are initialized and get probed afterwards by
176 the check_devices() function. If no further adapters are found,
177 ibmmca_detect() quits.
179 2.2 Physical Units, Logical Units, and Logical Devices
180 ------------------------------------------------------
181 There can be up to 56 devices on the SCSI bus (besides the adapter):
182 there are up to 7 "physical units" (each identified by physical unit
183 number or pun, also called the scsi id, this is the number you select
184 with hardware jumpers), and each physical unit can have up to 8
185 "logical units" (each identified by logical unit number, or lun,
186 between 0 and 7). The IBM SCSI-2 F/W adapter offers this on up to two
187 busses and provides support for 30 logical devices at the same time, where
188 in wide-addressing mode you can have 16 puns with 32 luns on each device.
189 This section describes the handling of devices on non-F/W adapters.
190 Just imagine, that you can have 16 * 32 = 512 devices on a F/W adapter
191 which means a lot of possible devices for such a small machine.
193 Typically the adapter has pun=7, so puns of other physical units
194 are between 0 and 6(15). On a wide-adapter a pun higher than 7 is
195 possible, but is normally not used. Almost all physical units have only
196 one logical unit, with lun=0. A CD-ROM jukebox would be an example of a
197 physical unit with more than one logical unit.
199 The embedded microprocessor of the IBM SCSI-subsystem hides the complex
200 two-dimensional (pun,lun) organization from the operating system.
201 When the machine is powered-up (or rebooted), the embedded microprocessor
202 checks, on its own, all 56 possible (pun,lun) combinations, and the first
203 15 devices found are assigned into a one-dimensional array of so-called
204 "logical devices", identified by "logical device numbers" or ldn. The last
205 ldn=15 is reserved for the subsystem itself. Wide adapters may have
206 to check up to 15 * 8 = 120 pun/lun combinations.
208 2.3 SCSI-Device Recognition and Dynamical ldn Assignment
209 --------------------------------------------------------
210 One consequence of information hiding is that the real (pun,lun)
211 numbers are also hidden. The two possibilities to get around this problem
212 are to offer fake pun/lun combinations to the operating system or to
213 delete the whole mapping of the adapter and to reassign the ldns, using
214 the immediate assign command of the SCSI-subsystem for probing through
215 all possible pun/lun combinations. An ldn is a "logical device number"
216 which is used by IBM SCSI-subsystems to access some valid SCSI-device.
217 At the beginning of the development of this driver, the following approach
218 was used:
220 First, the driver checked the ldn's (0 to 6) to find out which ldn's
221 have devices assigned. This was done by the functions check_devices() and
222 device_exists(). The interrupt handler has a special paragraph of code
223 (see local_checking_phase_flag) to assist in the checking. Assume, for
224 example, that three logical devices were found assigned at ldn 0, 1, 2.
225 These are presented to the upper layer of Linux SCSI driver
226 as devices with bogus (pun, lun) equal to (0,0), (1,0), (2,0).
227 On the other hand, if the upper layer issues a command to device
228 say (4,0), this driver returns DID_NO_CONNECT error.
230 In a second step of the driver development, the following improvement has
231 been applied: The first approach limited the number of devices to 7, far
232 fewer than the 15 that it could use, then it just mapped ldn ->
233 (ldn/8,ldn%8) for pun,lun. We ended up with a real mishmash of puns
234 and luns, but it all seemed to work.
236 The latest development, which is implemented from the driver version 3.0
237 and later, realizes the device recognition in the following way:
238 The physical SCSI-devices on the SCSI-bus are probed via immediate_assign-
239 and device_inquiry-commands, that is all implemented in a completely new
240 made check_devices() subroutine. This delivers an exact map of the physical
241 SCSI-world that is now stored in the get_scsi[][]-array. This means,
242 that the once hidden pun,lun assignment is now known to this driver.
243 It no longer believes in default-settings of the subsystem and maps all
244 ldns to existing pun,lun "by foot". This assures full control of the ldn
245 mapping and allows dynamical remapping of ldns to different pun,lun, if
246 there are more SCSI-devices installed than ldns available (n>15). The
247 ldns from 0 to 6 get 'hardwired' by this driver to puns 0 to 7 at lun=0,
248 excluding the pun of the subsystem. This assures, that at least simple
249 SCSI-installations have optimum access-speed and are not touched by
250 dynamical remapping. The ldns 7 to 14 are put to existing devices with
251 lun>0 or to non-existing devices, in order to satisfy the subsystem, if
252 there are less than 15 SCSI-devices connected. In the case of more than 15
253 devices, the dynamical mapping goes active. If the get_scsi[][] reports a
254 device to be existent, but it has no ldn assigned, it gets an ldn out of 7
255 to 14. The numbers are assigned in cyclic order, therefore it takes 8
256 dynamical reassignments on the SCSI-devices until a certain device
257 loses its ldn again. This assures that dynamical remapping is avoided
258 during intense I/O between up to 15 SCSI-devices (means pun,lun
259 combinations). A further advantage of this method is that people who
260 build their kernel without probing on all luns will get what they expect,
261 because the driver just won't assign everything with lun>0 when
262 multiple lun probing is inactive.
264 2.4 SCSI-Device Order
265 ---------------------
266 Because of the now correct recognition of physical pun,lun, and
267 their report to mid-level- and higher-level-drivers, the new reported puns
268 can be different from the old, faked puns. Therefore, Linux will eventually
269 change /dev/sdXXX assignments and prompt you for corrupted superblock
270 repair on boottime. In this case DO NOT PANIC, YOUR DISKS ARE STILL OK!!!
271 You have to reboot (CTRL-D) with an old kernel and set the /etc/fstab-file
272 entries right. After that, the system should come up as errorfree as before.
273 If your boot-partition is not coming up, also edit the /etc/lilo.conf-file
274 in a Linux session booted on old kernel and run lilo before reboot. Check
275 lilo.conf anyway to get boot on other partitions with foreign OSes right
276 again. But there exists a feature of this driver that allows you to change
277 the assignment order of the SCSI-devices by flipping the PUN-assignment.
278 See the next paragraph for a description.
280 The problem for this is, that Linux does not assign the SCSI-devices in the
281 way as described in the ANSI-SCSI-standard. Linux assigns /dev/sda to
282 the device with at minimum id 0. But the first drive should be at id 6,
283 because for historical reasons, drive at id 6 has, by hardware, the highest
284 priority and a drive at id 0 the lowest. IBM was one of the rare producers,
285 where the BIOS assigns drives belonging to the ANSI-SCSI-standard. Most
286 other producers' BIOS does not (I think even Adaptec-BIOS). The
287 IBMMCA_SCSI_ORDER_STANDARD flag, which you set while configuring the
288 kernel enables to choose the preferred way of SCSI-device-assignment.
289 Defining this flag would result in Linux determining the devices in the
290 same order as DOS and OS/2 does on your MCA-machine. This is also standard
291 on most industrial computers and OSes, like e.g. OS-9. Leaving this flag
292 undefined will get your devices ordered in the default way of Linux. See
293 also the remarks of Chris Beauregard from Dec 15, 1997 and the followups
294 in section 3.
296 2.5 Regular SCSI-Command-Processing
297 -----------------------------------
298 Only three functions get involved: ibmmca_queuecommand(), issue_cmd(),
299 and interrupt_handler().
301 The upper layer issues a scsi command by calling function
302 ibmmca_queuecommand(). This function fills a "subsystem control block"
303 (scb) and calls a local function issue_cmd(), which writes a scb
304 command into subsystem I/O ports. Once the scb command is carried out,
305 the interrupt_handler() is invoked. If a device is determined to be
306 existant and it has not assigned any ldn, it gets one dynamically.
307 For this, the whole stuff is done in ibmmca_queuecommand().
309 2.6 Abort & Reset Commands
310 --------------------------
311 These are implemented with busy waiting for interrupt to arrive.
312 ibmmca_reset() and ibmmca_abort() do not work sufficiently well
313 up to now and need still a lot of development work. This seems
314 to be a problem with other low-level SCSI drivers too, however
315 this should be no excuse.
317 2.7 Disk Geometry
318 -----------------
319 The ibmmca_biosparams() function should return the same disk geometry
320 as the bios. This is needed for fdisk, etc. The returned geometry is
321 certainly correct for disks smaller than 1 gigabyte. In the meantime,
322 it has been proved, that this works fine even with disks larger than
323 1 gigabyte.
325 2.8 Kernel Boot Option
326 ----------------------
327 The function ibmmca_scsi_setup() is called if option ibmmcascsi=n
328 is passed to the kernel. See file linux/init/main.c for details.
330 2.9 Driver Module Support
331 -------------------------
332 Is implemented and tested by K. Kudielka. This could probably not work
333 on kernels <2.1.0.
335 2.10 Multiple Hostadapter Support
336 ---------------------------------
337 This driver supports up to eight interfaces of type IBM-SCSI-Subsystem.
338 Integrated-, and MCA-adapters are automatically recognized. Unrecognizable
339 IBM-SCSI-Subsystem interfaces can be specified as kernel-parameters.
341 2.11 /proc/scsi-Filesystem Information
342 --------------------------------------
343 Information about the driver condition is given in
344 /proc/scsi/ibmmca/<host_no>. ibmmca_proc_info() provides this information.
346 This table is quite informative for interested users. It shows the load
347 of commands on the subsystem and whether you are running the bypassed
348 (software) or integrated (hardware) SCSI-command set (see below). The
349 amount of accesses is shown. Read, write, modeselect is shown separately
350 in order to help debugging problems with CD-ROMs or tapedrives.
352 The following table shows the list of 15 logical device numbers, that are
353 used by the SCSI-subsystem. The load on each ldn is shown in the table,
354 again, read and write commands are split. The last column shows the amount
355 of reassignments, that have been applied to the ldns, if you have more than
356 15 pun/lun combinations available on the SCSI-bus.
358 The last two tables show the pun/lun map and the positions of the ldns
359 on this pun/lun map. This may change during operation, when a ldn is
360 reassigned to another pun/lun combination. If the necessity for dynamical
361 assignments is set to 'no', the ldn structure keeps static.
363 2.12 /proc/mca-Filesystem Information
364 -------------------------------------
365 The slot-file contains all default entries and in addition chip and I/O-
366 address information of the SCSI-subsystem. This information is provided
367 by ibmmca_getinfo().
369 2.13 Supported IBM SCSI-Subsystems
370 ----------------------------------
371 The following IBM SCSI-subsystems are supported by this driver:
373 - IBM Fast/Wide SCSI-2 Adapter
374 - IBM 7568 Industrial Computer SCSI Adapter w/Cache
375 - IBM Expansion Unit SCSI Controller
376 - IBM SCSI Adapter w/Cache
377 - IBM SCSI Adapter
378 - IBM Integrated SCSI Controller
379 - All clones, 100% compatible with the chipset and subsystem command
380 system of IBM SCSI-adapters (forced detection)
382 2.14 Linux Kernel Versions
383 --------------------------
384 The IBM SCSI-subsystem low level driver is prepared to be used with
385 all versions of Linux between 2.0.x and 2.4.x. The compatibility checks
386 are fully implemented up from version 3.1e of the driver. This means, that
387 you just need the latest ibmmca.h and ibmmca.c file and copy it in the
388 linux/drivers/scsi directory. The code is automatically adapted during
389 kernel compilation. This is different from kernel 2.4.0! Here version
390 4.0 or later of the driver must be used for kernel 2.4.0 or later. Version
391 4.0 or later does not work together with older kernels! Driver versions
392 older than 4.0 do not work together with kernel 2.4.0 or later. They work
393 on all older kernels.
395 3 Code History
396 --------------
397 Jan 15 1996: First public release.
398 - Martin Kolinek
400 Jan 23 1996: Scrapped code which reassigned scsi devices to logical
401 device numbers. Instead, the existing assignment (created
402 when the machine is powered-up or rebooted) is used.
403 A side effect is that the upper layer of Linux SCSI
404 device driver gets bogus scsi ids (this is benign),
405 and also the hard disks are ordered under Linux the
406 same way as they are under dos (i.e., C: disk is sda,
407 D: disk is sdb, etc.).
408 - Martin Kolinek
410 I think that the CD-ROM is now detected only if a CD is
411 inside CD_ROM while Linux boots. This can be fixed later,
412 once the driver works on all types of PS/2's.
413 - Martin Kolinek
415 Feb 7 1996: Modified biosparam function. Fixed the CD-ROM detection.
416 For now, devices other than harddisk and CD_ROM are
417 ignored. Temporarily modified abort() function
418 to behave like reset().
419 - Martin Kolinek
421 Mar 31 1996: The integrated scsi subsystem is correctly found
422 in PS/2 models 56,57, but not in model 76. Therefore
423 the ibmmca_scsi_setup() function has been added today.
424 This function allows the user to force detection of
425 scsi subsystem. The kernel option has format
426 ibmmcascsi=n
427 where n is the scsi_id (pun) of the subsystem. Most likely, n is 7.
428 - Martin Kolinek
430 Aug 21 1996: Modified the code which maps ldns to (pun,0). It was
431 insufficient for those of us with CD-ROM changers.
432 - Chris Beauregard
434 Dec 14 1996: More improvements to the ldn mapping. See check_devices
435 for details. Did more fiddling with the integrated SCSI detection,
436 but I think it's ultimately hopeless without actually testing the
437 model of the machine. The 56, 57, 76 and 95 (ultimedia) all have
438 different integrated SCSI register configurations. However, the 56
439 and 57 are the only ones that have problems with forced detection.
440 - Chris Beauregard
442 Mar 8-16 1997: Modified driver to run as a module and to support
443 multiple adapters. A structure, called ibmmca_hostdata, is now
444 present, containing all the variables, that were once only
445 available for one single adapter. The find_subsystem-routine has vanished.
446 The hardware recognition is now done in ibmmca_detect directly.
447 This routine checks for presence of MCA-bus, checks the interrupt
448 level and continues with checking the installed hardware.
449 Certain PS/2-models do not recognize a SCSI-subsystem automatically.
450 Hence, the setup defined by command-line-parameters is checked first.
451 Thereafter, the routine probes for an integrated SCSI-subsystem.
452 Finally, adapters are checked. This method has the advantage to cover all
453 possible combinations of multiple SCSI-subsystems on one MCA-board. Up to
454 eight SCSI-subsystems can be recognized and announced to the upper-level
455 drivers with this improvement. A set of defines made changes to other
456 routines as small as possible.
457 - Klaus Kudielka
459 May 30 1997: (v1.5b)
460 1) SCSI-command capability enlarged by the recognition of MODE_SELECT.
461 This needs the RD-Bit to be disabled on IM_OTHER_SCSI_CMD_CMD which
462 allows data to be written from the system to the device. It is a
463 necessary step to be allowed to set blocksize of SCSI-tape-drives and
464 the tape-speed, without confusing the SCSI-Subsystem.
465 2) The recognition of a tape is included in the check_devices routine.
466 This is done by checking for TYPE_TAPE, that is already defined in
467 the kernel-scsi-environment. The markup of a tape is done in the
468 global ldn_is_tape[] array. If the entry on index ldn
469 is 1, there is a tapedrive connected.
470 3) The ldn_is_tape[] array is necessary to distinguish between tape- and
471 other devices. Fixed blocklength devices should not cause a problem
472 with the SCB-command for read and write in the ibmmca_queuecommand
473 subroutine. Therefore, I only derivate the READ_XX, WRITE_XX for
474 the tape-devices, as recommended by IBM in this Technical Reference,
475 mentioned below. (IBM recommends to avoid using the read/write of the
476 subsystem, but the fact was, that read/write causes a command error from
477 the subsystem and this causes kernel-panic.)
478 4) In addition, I propose to use the ldn instead of a fix char for the
479 display of PS2_DISK_LED_ON(). On 95, one can distinguish between the
480 devices that are accessed. It shows activity and easyfies debugging.
481 The tape-support has been tested with a SONY SDT-5200 and a HP DDS-2
482 (I do not know yet the type). Optimization and CD-ROM audio-support,
483 I am working on ...
484 - Michael Lang
486 June 19 1997: (v1.6b)
487 1) Submitting the extra-array ldn_is_tape[] -> to the local ld[]
488 device-array.
489 2) CD-ROM Audio-Play seems to work now.
490 3) When using DDS-2 (120M) DAT-Tapes, mtst shows still density-code
491 0x13 for ordinary DDS (61000 BPM) instead 0x24 for DDS-2. This appears
492 also on Adaptec 2940 adaptor in a PCI-System. Therefore, I assume that
493 the problem is independent of the low-level-driver/bus-architecture.
494 4) Hexadecimal ldn on PS/2-95 LED-display.
495 5) Fixing of the PS/2-LED on/off that it works right with tapedrives and
496 does not confuse the disk_rw_in_progress counter.
497 - Michael Lang
499 June 21 1997: (v1.7b)
500 1) Adding of a proc_info routine to inform in /proc/scsi/ibmmca/<host> the
501 outer-world about operational load statistics on the different ldns,
502 seen by the driver. Everybody that has more than one IBM-SCSI should
503 test this, because I only have one and cannot see what happens with more
504 than one IBM-SCSI hosts.
505 2) Definition of a driver version-number to have a better recognition of
506 the source when there are existing too much releases that may confuse
507 the user, when reading about release-specific problems. Up to know,
508 I calculated the version-number to be 1.7. Because we are in BETA-test
509 yet, it is today 1.7b.
510 3) Sorry for the heavy bug I programmed on June 19 1997! After that, the
511 CD-ROM did not work any more! The C7-command was a fake impression
512 I got while programming. Now, the READ and WRITE commands for CD-ROM are
513 no longer running over the subsystem, but just over
514 IM_OTHER_SCSI_CMD_CMD. On my observations (PS/2-95), now CD-ROM mounts
515 much faster(!) and hopefully all fancy multimedia-functions, like direct
516 digital recording from audio-CDs also work. (I tried it with cdda2wav
517 from the cdwtools-package and it filled up the harddisk immediately :-).)
518 To easify boolean logics, a further local device-type in ld[], called
519 is_cdrom has been included.
520 4) If one uses a SCSI-device of unsupported type/commands, one
521 immediately runs into a kernel-panic caused by Command Error. To better
522 understand which SCSI-command caused the problem, I extended this
523 specific panic-message slightly.
524 - Michael Lang
526 June 25 1997: (v1.8b)
527 1) Some cosmetic changes for the handling of SCSI-device-types.
528 Now, also CD-Burners / WORMs and SCSI-scanners should work. For
529 MO-drives I have no experience, therefore not yet supported.
530 In logical_devices I changed from different type-variables to one
531 called 'device_type' where the values, corresponding to scsi.h,
532 of a SCSI-device are stored.
533 2) There existed a small bug, that maps a device, coming after a SCSI-tape
534 wrong. Therefore, e.g. a CD-ROM changer would have been mapped wrong
535 -> problem removed.
536 3) Extension of the logical_device structure. Now it contains also device,
537 vendor and revision-level of a SCSI-device for internal usage.
538 - Michael Lang
540 June 26-29 1997: (v2.0b)
541 1) The release number 2.0b is necessary because of the completely new done
542 recognition and handling of SCSI-devices with the adapter. As I got
543 from Chris the hint, that the subsystem can reassign ldns dynamically,
544 I remembered this immediate_assign-command, I found once in the handbook.
545 Now, the driver first kills all ldn assignments that are set by default
546 on the SCSI-subsystem. After that, it probes on all puns and luns for
547 devices by going through all combinations with immediate_assign and
548 probing for devices, using device_inquiry. The found physical(!) pun,lun
549 structure is stored in get_scsi[][] as device types. This is followed
550 by the assignment of all ldns to existing SCSI-devices. If more ldns
551 than devices are available, they are assigned to non existing pun,lun
552 combinations to satisfy the adapter. With this, the dynamical mapping
553 was possible to implement. (For further info see the text in the
554 source code and in the description below. Read the description
555 below BEFORE installing this driver on your system!)
556 2) Changed the name IBMMCA_DRIVER_VERSION to IBMMCA_SCSI_DRIVER_VERSION.
557 3) The LED-display shows on PS/2-95 no longer the ldn, but the SCSI-ID
558 (pun) of the accessed SCSI-device. This is now senseful, because the
559 pun known within the driver is exactly the pun of the physical device
560 and no longer a fake one.
561 4) The /proc/scsi/ibmmca/<host_no> consists now of the first part, where
562 hit-statistics of ldns is shown and a second part, where the maps of
563 physical and logical SCSI-devices are displayed. This could be very
564 interesting, when one is using more than 15 SCSI-devices in order to
565 follow the dynamical remapping of ldns.
566 - Michael Lang
568 June 26-29 1997: (v2.0b-1)
569 1) I forgot to switch the local_checking_phase_flag to 1 and back to 0
570 in the dynamical remapping part in ibmmca_queuecommand for the
571 device_exist routine. Sorry.
572 - Michael Lang
574 July 1-13 1997: (v3.0b,c)
575 1) Merging of the driver-developments of Klaus Kudielka and Michael Lang
576 in order to get a optimum and unified driver-release for the
577 IBM-SCSI-Subsystem-Adapter(s).
578 For people, using the Kernel-release >=2.1.0, module-support should
579 be no problem. For users, running under <2.1.0, module-support may not
580 work, because the methods have changed between 2.0.x and 2.1.x.
581 2) Added some more effective statistics for /proc-output.
582 3) Change typecasting at necessary points from (unsigned long) to
583 virt_to_bus().
584 4) Included #if... at special points to have specific adaption of the
585 driver to kernel 2.0.x and 2.1.x. It should therefore also run with
586 later releases.
587 5) Magneto-Optical drives and medium-changers are also recognized, now.
588 Therefore, we have a completely gapfree recognition of all SCSI-
589 device-types, that are known by Linux up to kernel 2.1.31.
590 6) The flag SCSI_IBMMCA_DEV_RESET has been inserted. If it is set within
591 the configuration, each connected SCSI-device will get a reset command
592 during boottime. This can be necessary for some special SCSI-devices.
593 This flag should be included in Config.in.
594 (See also the new Config.in file.)
595 Probable next improvement: bad disk handler.
596 - Michael Lang
598 Sept 14 1997: (v3.0c)
599 1) Some debugging and speed optimization applied.
600 - Michael Lang
602 Dec 15, 1997
603 - chrisb@truespectra.com
604 - made the front panel display thingy optional, specified from the
605 command-line via ibmmcascsi=display. Along the lines of the /LED
606 option for the OS/2 driver.
607 - fixed small bug in the LED display that would hang some machines.
608 - reversed ordering of the drives (using the
609 IBMMCA_SCSI_ORDER_STANDARD define). This is necessary for two main
610 reasons:
611 - users who've already installed Linux won't be screwed. Keep
612 in mind that not everyone is a kernel hacker.
613 - be consistent with the BIOS ordering of the drives. In the
614 BIOS, id 6 is C:, id 0 might be D:. With this scheme, they'd be
615 backwards. This confuses the crap out of those heathens who've
616 got a impure Linux installation (which, <wince>, I'm one of).
617 This whole problem arises because IBM is actually non-standard with
618 the id to BIOS mappings. You'll find, in fdomain.c, a similar
619 comment about a few FD BIOS revisions. The Linux (and apparently
620 industry) standard is that C: maps to scsi id (0,0). Let's stick
621 with that standard.
622 - Since this is technically a branch of my own, I changed the
623 version number to 3.0e-cpb.
625 Jan 17, 1998: (v3.0f)
626 1) Addition of some statistical info for /proc in proc_info.
627 2) Taking care of the SCSI-assignment problem, dealed by Chris at Dec 15
628 1997. In fact, IBM is right, concerning the assignment of SCSI-devices
629 to driveletters. It is conform to the ANSI-definition of the SCSI-
630 standard to assign drive C: to SCSI-id 6, because it is the highest
631 hardware priority after the hostadapter (that has still today by
632 default everywhere id 7). Also realtime-operating systems that I use,
633 like LynxOS and OS9, which are quite industrial systems use top-down
634 numbering of the harddisks, that is also starting at id 6. Now, one
635 sits a bit between two chairs. On one hand side, using the define
636 IBMMCA_SCSI_ORDER_STANDARD makes Linux assigning disks conform to
637 the IBM- and ANSI-SCSI-standard and keeps this driver downward
638 compatible to older releases, on the other hand side, people is quite
639 habituated in believing that C: is assigned to (0,0) and much other
640 SCSI-BIOS do so. Therefore, I moved the IBMMCA_SCSI_ORDER_STANDARD
641 define out of the driver and put it into Config.in as subitem of
642 'IBM SCSI support'. A help, added to Documentation/Configure.help
643 explains the differences between saying 'y' or 'n' to the user, when
644 IBMMCA_SCSI_ORDER_STANDARD prompts, so the ordinary user is enabled to
645 choose the way of assignment, depending on his own situation and gusto.
646 3) Adapted SCSI_IBMMCA_DEV_RESET to the local naming convention, so it is
647 now called IBMMCA_SCSI_DEV_RESET.
648 4) Optimization of proc_info and its subroutines.
649 5) Added more in-source-comments and extended the driver description by
650 some explanation about the SCSI-device-assignment problem.
651 - Michael Lang
653 Jan 18, 1998: (v3.0g)
654 1) Correcting names to be absolutely conform to the later 2.1.x releases.
655 This is necessary for
656 IBMMCA_SCSI_DEV_RESET -> CONFIG_IBMMCA_SCSI_DEV_RESET
657 IBMMCA_SCSI_ORDER_STANDARD -> CONFIG_IBMMCA_SCSI_ORDER_STANDARD
658 - Michael Lang
660 Jan 18, 1999: (v3.1 MCA-team internal)
661 1) The multiple hosts structure is accessed from every subroutine, so there
662 is no longer the address of the device structure passed from function
663 to function, but only the hostindex. A call by value, nothing more. This
664 should really be understood by the compiler and the subsystem should get
665 the right values and addresses.
666 2) The SCSI-subsystem detection was not complete and quite hugely buggy up
667 to now, compared to the technical manual. The interpretation of the pos2
668 register is not as assumed by people before, therefore, I dropped a note
669 in the ibmmca_detect function to show the registers' interpretation.
670 The pos-registers of integrated SCSI-subsystems do not contain any
671 information concerning the IO-port offset, really. Instead, they contain
672 some info about the adapter, the chip, the NVRAM .... The I/O-port is
673 fixed to 0x3540 - 0x3547. There can be more than one adapters in the
674 slots and they get an offset for the I/O area in order to get their own
675 I/O-address area. See chapter 2 for detailed description. At least, the
676 detection should now work right, even on models other than 95. The 95ers
677 came happily around the bug, as their pos2 register contains always 0
678 in the critical area. Reserved bits are not allowed to be interpreted,
679 therefore, IBM is allowed to set those bits as they like and they may
680 really vary between different PS/2 models. So, now, no interpretation
681 of reserved bits - hopefully no trouble here anymore.
682 3) The command error, which you may get on models 55, 56, 57, 70, 77 and
683 P70 may have been caused by the fact, that adapters of older design do
684 not like sending commands to non-existing SCSI-devices and will react
685 with a command error as a sign of protest. While this error is not
686 present on IBM SCSI Adapter w/cache, it appears on IBM Integrated SCSI
687 Adapters. Therefore, I implemented a workaround to forgive those
688 adapters their protests, but it is marked up in the statistics, so
689 after a successful boot, you can see in /proc/scsi/ibmmca/<host_number>
690 how often the command errors have been forgiven to the SCSI-subsystem.
691 If the number is bigger than 0, you have a SCSI subsystem of older
692 design, what should no longer matter.
693 4) ibmmca_getinfo() has been adapted very carefully, so it shows in the
694 slotn file really, what is senseful to be presented.
695 5) ibmmca_register() has been extended in its parameter list in order to
696 pass the right name of the SCSI-adapter to Linux.
697 - Michael Lang
699 Feb 6, 1999: (v3.1)
700 1) Finally, after some 3.1Beta-releases, the 3.1 release. Sorry, for
701 the delayed release, but it was not finished with the release of
702 Kernel 2.2.0.
703 - Michael Lang
705 Feb 10, 1999 (v3.1)
706 1) Added a new commandline parameter called 'bypass' in order to bypass
707 every integrated subsystem SCSI-command consequently in case of
708 troubles.
709 2) Concatenated read_capacity requests to the harddisks. It gave a lot
710 of troubles with some controllers and after I wanted to apply some
711 extensions, it jumped out in the same situation, on my w/cache, as like
712 on D. Weinehalls' Model 56, having integrated SCSI. This gave me the
713 decisive hint to move the code-part out and declare it global. Now
714 it seems to work far better and more stable. Let us see what
715 the world thinks of it...
716 3) By the way, only Sony DAT-drives seem to show density code 0x13. A
717 test with a HP drive gave right results, so the problem is vendor-
718 specific and not a problem of the OS or the driver.
719 - Michael Lang
721 Feb 18, 1999 (v3.1d)
722 1) The abort command and the reset function have been checked for
723 inconsistencies. From the logical point of thinking, they work
724 at their optimum, now, but as the subsystem does not answer with an
725 interrupt, abort never finishes, sigh...
726 2) Everything, that is accessed by a busmaster request from the adapter
727 is now declared as global variable, even the return-buffer in the
728 local checking phase. This assures, that no accesses to undefined memory
729 areas are performed.
730 3) In ibmmca.h, the line unchecked_isa_dma is added with 1 in order to
731 avoid memory-pointers for the areas higher than 16MByte in order to
732 be sure, it also works on 16-Bit Microchannel bus systems.
733 4) A lot of small things have been found, but nothing that endangered the
734 driver operations. Just it should be more stable, now.
735 - Michael Lang
737 Feb 20, 1999 (v3.1e)
738 1) I took the warning from the Linux Kernel Hackers Guide serious and
739 checked the cmd->result return value to the done-function very carefully.
740 It is obvious, that the IBM SCSI only delivers the tsb.dev_status, if
741 some error appeared, else it is undefined. Now, this is fixed. Before
742 any SCB command gets queued, the tsb.dev_status is set to 0, so the
743 cmd->result won't screw up Linux higher level drivers.
744 2) The reset-function has slightly improved. This is still planed for
745 abort. During the abort and the reset function, no interrupts are
746 allowed. This is however quite hard to cope with, so the INT-status
747 register is read. When the interrupt gets queued, one can find its
748 status immediately on that register and is enabled to continue in the
749 reset function. I had no chance to test this really, only in a bogus
750 situation, I got this function running, but the situation was too much
751 worse for Linux :-(, so tests will continue.
752 3) Buffers got now consistent. No open address mapping, as before and
753 therefore no further troubles with the unassigned memory segmentation
754 faults that scrambled probes on 95XX series and even on 85XX series,
755 when the kernel is done in a not so perfectly fitting way.
756 4) Spontaneous interrupts from the subsystem, appearing without any
757 command previously queued are answered with a DID_BAD_INTR result.
758 5) Taken into account ZP Gus' proposals to reverse the SCSI-device
759 scan order. As it does not work on Kernel 2.1.x or 2.2.x, as proposed
760 by him, I implemented it in a slightly derived way, which offers in
761 addition more flexibility.
762 - Michael Lang
764 Apr 23, 2000 (v3.2pre1)
765 1) During a very long time, I collected a huge amount of bug reports from
766 various people, trying really quite different things on their SCSI-
767 PS/2s. Today, all these bug reports are taken into account and should be
768 mostly solved. The major topics were:
769 - Driver crashes during boottime by no obvious reason.
770 - Driver panics while the midlevel-SCSI-driver is trying to inquire
771 the SCSI-device properties, even though hardware is in perfect state.
772 - Displayed info for the various slot-cards is interpreted wrong.
773 The main reasons for the crashes were two:
774 1) The commands to check for device information like INQUIRY,
775 TEST_UNIT_READY, REQUEST_SENSE and MODE_SENSE cause the devices
776 to deliver information of up to 255 bytes. Midlevel drivers offer
777 1024 bytes of space for the answer, but the IBM-SCSI-adapters do
778 not accept this, as they stick quite near to ANSI-SCSI and report
779 a COMMAND_ERROR message which causes the driver to panic. The main
780 problem was located around the INQUIRY command. Now, for all the
781 mentioned commands, the buffersize sent to the adapter is at
782 maximum 255 which seems to be a quite reasonable solution.
783 TEST_UNIT_READY gets a buffersize of 0 to make sure that no
784 data is transferred in order to avoid any possible command failure.
785 2) On unsuccessful TEST_UNIT_READY, the mid-level driver has to send
786 a REQUEST_SENSE in order to see where the problem is located. This
787 REQUEST_SENSE may have various length in its answer-buffer. IBM
788 SCSI-subsystems report a command failure if the returned buffersize
789 is different from the sent buffersize, but this can be suppressed by
790 a special bit, which is now done and problems seem to be solved.
791 2) Code adaption to all kernel-releases. Now, the 3.2 code compiles on
792 2.0.x, 2.1.x, 2.2.x and 2.3.x kernel releases without any code-changes.
793 3) Commandline-parameters are recognized again, even under Kernel 2.3.x or
794 higher.
795 - Michael Lang
797 April 27, 2000 (v3.2pre2)
798 1) Bypassed commands get read by the adapter by one cycle instead of two.
799 This increases SCSI-performance.
800 2) Synchronous datatransfer is provided for sure to be 5 MHz on older
801 SCSI and 10 MHz on internal F/W SCSI-adapter.
802 3) New commandline parameters allow to force the adapter to slow down while
803 in synchronous transfer. Could be helpful for very old devices.
804 - Michael Lang
806 June 2, 2000 (v3.2pre5)
807 1) Added Jim Shorney's contribution to make the activity indicator
808 flashing in addition to the LED-alphanumeric display-panel on
809 models 95A. To be enabled to choose this feature freely, a new
810 commandline parameter is added, called 'activity'.
811 2) Added the READ_CONTROL bit for test_unit_ready SCSI-command.
812 3) Added some suppress_exception bits to read_device_capacity and
813 all device_inquiry occurrences in the driver code.
814 4) Complaints about the various KERNEL_VERSION implementations are
815 taken into account. Every local_LinuxKernelVersion occurrence is
816 now replaced by KERNEL_VERSION, defined in linux/version.h.
817 Corresponding changes were applied to ibmmca.h, too. This was a
818 contribution to all kernel-parts by Philipp Hahn.
819 - Michael Lang
821 July 17, 2000 (v3.2pre8)
822 A long period of collecting bug reports from all corners of the world
823 now lead to the following corrections to the code:
824 1) SCSI-2 F/W support crashed with a COMMAND ERROR. The reason for this
825 was that it is possible to disable Fast-SCSI for the external bus.
826 The feature-control command, where this crash appeared regularly, tried
827 to set the maximum speed of 10MHz synchronous transfer speed and that
828 reports a COMMAND ERROR if external bus Fast-SCSI is disabled. Now,
829 the feature-command probes down from maximum speed until the adapter
830 stops to complain, which is at the same time the maximum possible
831 speed selected in the reference program. So, F/W external can run at
832 5 MHz (slow-) or 10 MHz (fast-SCSI). During feature probing, the
833 COMMAND ERROR message is used to detect if the adapter does not complain.
834 2) Up to now, only combined busmode is supported, if you use external
835 SCSI-devices, attached to the F/W-controller. If dual bus is selected,
836 only the internal SCSI-devices get accessed by Linux. For most
837 applications, this should do fine.
838 3) Wide-SCSI-addressing (16-Bit) is now possible for the internal F/W
839 bus on the F/W adapter. If F/W adapter is detected, the driver
840 automatically uses the extended PUN/LUN <-> LDN mapping tables, which
841 are now new from 3.2pre8. This allows PUNs between 0 and 15 and should
842 provide more fun with the F/W adapter.
843 4) Several machines use the SCSI: POS registers for internal/undocumented
844 storage of system relevant info. This confused the driver, mainly on
845 models 9595, as it expected no onboard SCSI only, if all POS in
846 the integrated SCSI-area are set to 0x00 or 0xff. Now, the mechanism
847 to check for integrated SCSI is much more restrictive and these problems
848 should be history.
849 - Michael Lang
851 July 18, 2000 (v3.2pre9)
852 This develop rather quickly at the moment. Two major things were still
853 missing in 3.2pre8:
854 1) The adapter PUN for F/W adapters has 4-bits, while all other adapters
855 have 3-bits. This is now taken into account for F/W.
856 2) When you select CONFIG_IBMMCA_SCSI_ORDER_STANDARD, you should
857 normally get the inverse probing order of your devices on the SCSI-bus.
858 The ANSI device order gets scrambled in version 3.2pre8!! Now, a new
859 and tested algorithm inverts the device-order on the SCSI-bus and
860 automatically avoids accidental access to whatever SCSI PUN the adapter
861 is set and works with SCSI- and Wide-SCSI-addressing.
862 - Michael Lang
864 July 23, 2000 (v3.2pre10 unpublished)
865 1) LED panel display supports wide-addressing in ibmmca=display mode.
866 2) Adapter-information and autoadaption to address-space is done.
867 3) Auto-probing for maximum synchronous SCSI transfer rate is working.
868 4) Optimization to some embedded function calls is applied.
869 5) Added some comment for the user to wait for SCSI-devices being probed.
870 6) Finished version 3.2 for Kernel 2.4.0. It least, I thought it is but...
871 - Michael Lang
873 July 26, 2000 (v3.2pre11)
874 1) I passed a horrible weekend getting mad with NMIs on kernel 2.2.14 and
875 a model 9595. Asking around in the community, nobody except of me has
876 seen such errors. Weird, but I am trying to recompile everything on
877 the model 9595. Maybe, as I use a specially modified gcc, that could
878 cause problems. But, it was not the reason. The true background was,
879 that the kernel was compiled for i386 and the 9595 has a 486DX-2.
880 Normally, no troubles should appear, but for this special machine,
881 only the right processor support is working fine!
882 2) Previous problems with synchronous speed, slowing down from one adapter
883 to the next during probing are corrected. Now, local variables store
884 the synchronous bitmask for every single adapter found on the MCA bus.
885 3) LED alphanumeric panel support for XX95 systems is now showing some
886 alive rotator during boottime. This makes sense, when no monitor is
887 connected to the system. You can get rid of all display activity, if
888 you do not use any parameter or just ibmmcascsi=activity, for the
889 harddrive activity LED, existent on all PS/2, except models 8595-XXX.
890 If no monitor is available, please use ibmmcascsi=display, which works
891 fine together with the linuxinfo utility for the LED-panel.
892 - Michael Lang
894 July 29, 2000 (v3.2)
895 1) Submission of this driver for kernel 2.4test-XX and 2.2.17.
896 - Michael Lang
898 December 28, 2000 (v3.2d / v4.0)
899 1) The interrupt handler had some wrong statement to wait for. This
900 was done due to experimental reasons during 3.2 development but it
901 has shown that this is not stable enough. Going back to wait for the
902 adapter to be not busy is best.
903 2) Inquiry requests can be shorter than 255 bytes of return buffer. Due
904 to a bug in the ibmmca_queuecommand routine, this buffer was forced
905 to 255 at minimum. If the memory address, this return buffer is pointing
906 to does not offer more space, invalid memory accesses destabilized the
907 kernel.
908 3) version 4.0 is only valid for kernel 2.4.0 or later. This is necessary
909 to remove old kernel version dependent waste from the driver. 3.2d is
910 only distributed with older kernels but keeps compatibility with older
911 kernel versions. 4.0 and higher versions cannot be used with older
912 kernels anymore!! You must have at least kernel 2.4.0!!
913 4) The commandline argument 'bypass' and all its functionality got removed
914 in version 4.0. This was never really necessary, as all troubles were
915 based on non-command related reasons up to now, so bypassing commands
916 did not help to avoid any bugs. It is kept in 3.2X for debugging reasons.
917 5) Dynamic reassignment of ldns was again verified and analyzed to be
918 completely inoperational. This is corrected and should work now.
919 6) All commands that get sent to the SCSI adapter were verified and
920 completed in such a way, that they are now completely conform to the
921 demands in the technical description of IBM. Main candidates were the
922 DEVICE_INQUIRY, REQUEST_SENSE and DEVICE_CAPACITY commands. They must
923 be transferred by bypassing the internal command buffer of the adapter
924 or else the response can be a random result. GET_POS_INFO would be more
925 safe in usage, if one could use the SUPRESS_EXCEPTION_SHORT, but this
926 is not allowed by the technical references of IBM. (Sorry, folks, the
927 model 80 problem is still a task to be solved in a different way.)
928 7) v3.2d is still hold back for some days for testing, while 4.0 is
929 released.
930 - Michael Lang
932 January 3, 2001 (v4.0a)
933 1) A lot of complains after the 2.4.0-prerelease kernel came in about
934 the impossibility to compile the driver as a module. This problem is
935 solved. In combination with that problem, some unprecise declaration
936 of the function option_setup() gave some warnings during compilation.
937 This is solved, too by a forward declaration in ibmmca.c.
938 2) #ifdef argument concerning CONFIG_SCSI_IBMMCA is no longer needed and
939 was entirely removed.
940 3) Some switch statements got optimized in code, as some minor variables
941 in internal SCSI-command handlers.
942 - Michael Lang
944 4 To do
945 -------
946 - IBM SCSI-2 F/W external SCSI bus support in separate mode!
947 - It seems that the handling of bad disks is really bad -
948 non-existent, in fact. However, a low-level driver cannot help
949 much, if such things happen.
951 5 Users' Manual
952 ---------------
953 5.1 Commandline Parameters
954 --------------------------
955 There exist several features for the IBM SCSI-subsystem driver.
956 The commandline parameter format is:
958 ibmmcascsi=<command1>,<command2>,<command3>,...
960 where commandN can be one of the following:
962 display Owners of a model 95 or other PS/2 systems with an
963 alphanumeric LED display may set this to have their
964 display showing the following output of the 8 digits:
966 ------DA
968 where '-' stays dark, 'D' shows the SCSI-device id
969 and 'A' shows the SCSI hostindex, being currently
970 accessed. During boottime, this will give the message
972 SCSIini*
974 on the LED-panel, where the * represents a rotator,
975 showing the activity during the probing phase of the
976 driver which can take up to two minutes per SCSI-adapter.
977 adisplay This works like display, but gives more optical overview
978 of the activities on the SCSI-bus. The display will have
979 the following output:
981 6543210A
983 where the numbers 0 to 6 light up at the shown position,
984 when the SCSI-device is accessed. 'A' shows again the SCSI
985 hostindex. If display nor adisplay is set, the internal
986 PS/2 harddisk LED is used for media-activities. So, if
987 you really do not have a system with a LED-display, you
988 should not set display or adisplay. Keep in mind, that
989 display and adisplay can only be used alternatively. It
990 is not recommended to use this option, if you have some
991 wide-addressed devices e.g. at the SCSI-2 F/W adapter in
992 your system. In addition, the usage of the display for
993 other tasks in parallel, like the linuxinfo-utility makes
994 no sense with this option.
995 activity This enables the PS/2 harddisk LED activity indicator.
996 Most PS/2 have no alphanumeric LED display, but some
997 indicator. So you should use this parameter to activate it.
998 If you own model 9595 (Server95), you can have both, the
999 LED panel and the activity indicator in parallel. However,
1000 some PS/2s, like the 8595 do not have any harddisk LED
1001 activity indicator, which means, that you must use the
1002 alphanumeric LED display if you want to monitor SCSI-
1003 activity.
1004 bypass This is obsolete from driver version 4.0, as the adapters
1005 got that far understood, that the selection between
1006 integrated and bypassed commands should now work completely
1007 correct! For historical reasons, the old description is
1008 kept here:
1009 This commandline parameter forces the driver never to use
1010 SCSI-subsystems' integrated SCSI-command set. Except of
1011 the immediate assign, which is of vital importance for
1012 every IBM SCSI-subsystem to set its ldns right. Instead,
1013 the ordinary ANSI-SCSI-commands are used and passed by the
1014 controller to the SCSI-devices, therefore 'bypass'. The
1015 effort, done by the subsystem is quite bogus and at a
1016 minimum and therefore it should work everywhere. This
1017 could maybe solve troubles with old or integrated SCSI-
1018 controllers and nasty harddisks. Keep in mind, that using
1019 this flag will slow-down SCSI-accesses slightly, as the
1020 software generated commands are always slower than the
1021 hardware. Non-harddisk devices always get read/write-
1022 commands in bypass mode. On the most recent releases of
1023 the Linux IBM-SCSI-driver, the bypass command should be
1024 no longer a necessary thing, if you are sure about your
1025 SCSI-hardware!
1026 normal This is the parameter, introduced on the 2.0.x development
1027 rail by ZP Gu. This parameter defines the SCSI-device
1028 scan order in the new industry standard. This means, that
1029 the first SCSI-device is the one with the lowest pun.
1030 E.g. harddisk at pun=0 is scanned before harddisk at
1031 pun=6, which means, that harddisk at pun=0 gets sda
1032 and the one at pun=6 gets sdb.
1033 ansi The ANSI-standard for the right scan order, as done by
1034 IBM, Microware and Microsoft, scans SCSI-devices starting
1035 at the highest pun, which means, that e.g. harddisk at
1036 pun=6 gets sda and a harddisk at pun=0 gets sdb. If you
1037 like to have the same SCSI-device order, as in DOS, OS-9
1038 or OS/2, just use this parameter.
1039 fast SCSI-I/O in synchronous mode is done at 5 MHz for IBM-
1040 SCSI-devices. SCSI-2 Fast/Wide Adapter/A external bus
1041 should then run at 10 MHz if Fast-SCSI is enabled,
1042 and at 5 MHz if Fast-SCSI is disabled on the external
1043 bus. This is the default setting when nothing is
1044 specified here.
1045 medium Synchronous rate is at 50% approximately, which means
1046 2.5 MHz for IBM SCSI-adapters and 5.0 MHz for F/W ext.
1047 SCSI-bus (when Fast-SCSI speed enabled on external bus).
1048 slow The slowest possible synchronous transfer rate is set.
1049 This means 1.82 MHz for IBM SCSI-adapters and 2.0 MHz
1050 for F/W external bus at Fast-SCSI speed on the external
1051 bus.
1053 A further option is that you can force the SCSI-driver to accept a SCSI-
1054 subsystem at a certain I/O-address with a predefined adapter PUN. This
1055 is done by entering
1057 commandN = I/O-base
1058 commandN+1 = adapter PUN
1060 e.g. ibmmcascsi=0x3540,7 will force the driver to detect a SCSI-subsystem
1061 at I/O-address 0x3540 with adapter PUN 7. Please only use this method, if
1062 the driver does really not recognize your SCSI-adapter! With driver version
1063 3.2, this recognition of various adapters was hugely improved and you
1064 should try first to remove your commandline arguments of such type with a
1065 newer driver. I bet, it will be recognized correctly. Even multiple and
1066 different types of IBM SCSI-adapters should be recognized correctly, too.
1067 Use the forced detection method only as last solution!
1069 Examples:
1071 ibmmcascsi=adisplay
1073 This will use the advanced display mode for the model 95 LED alphanumeric
1074 display.
1076 ibmmcascsi=display,0x3558,7
1078 This will activate the default display mode for the model 95 LED display
1079 and will force the driver to accept a SCSI-subsystem at I/O-base 0x3558
1080 with adapter PUN 7.
1082 5.2 Troubleshooting
1083 -------------------
1084 The following FAQs should help you to solve some major problems with this
1085 driver.
1087 Q: "Reset SCSI-devices at boottime" halts the system at boottime, why?
1088 A: This is only tested with the IBM SCSI Adapter w/cache. It is not
1089 yet proven to run on other adapters, however you may be lucky.
1090 In version 3.1d this has been hugely improved and should work better,
1091 now. Normally you really won't need to activate this flag in the
1092 kernel configuration, as all post 1989 SCSI-devices should accept
1093 the reset-signal, when the computer is switched on. The SCSI-
1094 subsystem generates this reset while being initialized. This flag
1095 is really reserved for users with very old, very strange or self-made
1096 SCSI-devices.
1097 Q: Why is the SCSI-order of my drives mirrored to the device-order
1098 seen from OS/2 or DOS ?
1099 A: It depends on the operating system, if it looks at the devices in
1100 ANSI-SCSI-standard (starting from pun 6 and going down to pun 0) or
1101 if it just starts at pun 0 and counts up. If you want to be conform
1102 with OS/2 and DOS, you have to activate this flag in the kernel
1103 configuration or you should set 'ansi' as parameter for the kernel.
1104 The parameter 'normal' sets the new industry standard, starting
1105 from pun 0, scanning up to pun 6. This allows you to change your
1106 opinion still after having already compiled the kernel.
1107 Q: Why can't I find IBM MCA SCSI support in the config menu?
1108 A: You have to activate MCA bus support, first.
1109 Q: Where can I find the latest info about this driver?
1110 A: See the file MAINTAINERS for the current WWW-address, which offers
1111 updates, info and Q/A lists. At this file's origin, the webaddress
1112 was: http://www.staff.uni-mainz.de/mlang/linux.html
1113 Q: My SCSI-adapter is not recognized by the driver, what can I do?
1114 A: Just force it to be recognized by kernel parameters. See section 5.1.
1115 If this really happens, do also send e-mail to the maintainer, as
1116 forced detection should be never necessary. Forced detection is in
1117 principal some flaw of the driver adapter detection and goes into
1118 bug reports.
1119 Q: The driver screws up, if it starts to probe SCSI-devices, is there
1120 some way out of it?
1121 A: Yes, that was some recognition problem of the correct SCSI-adapter
1122 and its I/O base addresses. Upgrade your driver to the latest release
1123 and it should be fine again.
1124 Q: I get a message: panic IBM MCA SCSI: command error .... , what can
1125 I do against this?
1126 A: Previously, I followed the way by ignoring command errors by using
1127 ibmmcascsi=forgiveall, but this command no longer exists and is
1128 obsolete. If such a problem appears, it is caused by some segmentation
1129 fault of the driver, which maps to some unallowed area. The latest
1130 version of the driver should be ok, as most bugs have been solved.
1131 Q: There are still kernel panics, even after having set
1132 ibmmcascsi=forgiveall. Are there other possibilities to prevent
1133 such panics?
1134 A: No, get just the latest release of the driver and it should work
1135 better and better with increasing version number. Forget about this
1136 ibmmcascsi=forgiveall, as also ignorecmd are obsolete.!
1137 Q: Linux panics or stops without any comment, but it is probable, that my
1138 harddisk(s) have bad blocks.
1139 A: Sorry, the bad-block handling is still a feeble point of this driver,
1140 but is on the schedule for development in the near future.
1141 Q: Linux panics while dynamically assigning SCSI-ids or ldns.
1142 A: If you disconnect a SCSI-device from the machine, while Linux is up
1143 and the driver uses dynamical reassignment of logical device numbers
1144 (ldn), it really gets "angry" if it won't find devices, that were still
1145 present at boottime and stops Linux.
1146 Q: The system does not recover after an abort-command has been generated.
1147 A: This is regrettably true, as it is not yet understood, why the
1148 SCSI-adapter does really NOT generate any interrupt at the end of
1149 the abort-command. As no interrupt is generated, the abort command
1150 cannot get finished and the system hangs, sorry, but checks are
1151 running to hunt down this problem. If there is a real pending command,
1152 the interrupt MUST get generated after abort. In this case, it
1153 should finish well.
1154 Q: The system gets in bad shape after a SCSI-reset, is this known?
1155 A: Yes, as there are a lot of prescriptions (see the Linux Hackers'
1156 Guide) what has to be done for reset, we still share the bad shape of
1157 the reset functions with all other low level SCSI-drivers.
1158 Astonishingly, reset works in most cases quite ok, but the harddisks
1159 won't run in synchronous mode anymore after a reset, until you reboot.
1160 Q: Why does my XXX w/Cache adapter not use read-prefetch?
1161 A: Ok, that is not completely possible. If a cache is present, the
1162 adapter tries to use it internally. Explicitly, one can use the cache
1163 with a read prefetch command, maybe in future, but this requires
1164 some major overhead of SCSI-commands that risks the performance to
1165 go down more than it gets improved. Tests with that are running.
1166 Q: I have a IBM SCSI-2 Fast/Wide adapter, it boots in some way and hangs.
1167 A: Yes, that is understood, as for sure, your SCSI-2 Fast/Wide adapter
1168 was in such a case recognized as integrated SCSI-adapter or something
1169 else, but not as the correct adapter. As the I/O-ports get assigned
1170 wrongly by that reason, the system should crash in most cases. You
1171 should upgrade to the latest release of the SCSI-driver. The
1172 recommended version is 3.2 or later. Here, the F/W support is in
1173 a stable and reliable condition. Wide-addressing is in addition
1174 supported.
1175 Q: I get an Oops message and something like "killing interrupt".
1176 A: The reason for this is that the IBM SCSI-subsystem only sends a
1177 termination status back, if some error appeared. In former releases
1178 of the driver, it was not checked, if the termination status block
1179 is NULL. From version 3.2, it is taken care of this.
1180 Q: I have a F/W adapter and the driver sees my internal SCSI-devices,
1181 but ignores the external ones.
1182 A: Select combined busmode in the IBM config-program and check for that
1183 no SCSI-id on the external devices appears on internal devices.
1184 Reboot afterwards. Dual busmode is supported, but works only for the
1185 internal bus, yet. External bus is still ignored. Take care for your
1186 SCSI-ids. If combined bus-mode is activated, on some adapters,
1187 the wide-addressing is not possible, so devices with ids between 8
1188 and 15 get ignored by the driver & adapter!
1189 Q: I have a 9595 and I get a NMI during heavy SCSI I/O e.g. during fsck.
1190 A COMMAND ERROR is reported and characters on the screen are missing.
1191 Warm reboot is not possible. Things look like quite weird.
1192 A: Check the processor type of your 9595. If you have an 80486 or 486DX-2
1193 processor complex on your mainboard and you compiled a kernel that
1194 supports 80386 processors, it is possible, that the kernel cannot
1195 keep track of the PS/2 interrupt handling and stops on an NMI. Just
1196 compile a kernel for the correct processor type of your PS/2 and
1197 everything should be fine. This is necessary even if one assumes,
1198 that some 80486 system should be downward compatible to 80386
1199 software.
1200 Q: Some commands hang and interrupts block the machine. After some
1201 timeout, the syslog reports that it tries to call abort, but the
1202 machine is frozen.
1203 A: This can be a busy wait bug in the interrupt handler of driver
1204 version 3.2. You should at least upgrade to 3.2c if you use
1205 kernel < 2.4.0 and driver version 4.0 if you use kernel 2.4.0 or
1206 later (including all test releases).
1207 Q: I have a PS/2 model 80 and more than 16 MBytes of RAM. The driver
1208 completely refuses to work, reports NMIs, COMMAND ERRORs or other
1209 ambiguous stuff. When reducing the RAM size down below 16 MB,
1210 everything is running smoothly.
1211 A: No real answer, yet. In any case, one should force the kernel to
1212 present SCBs only below the 16 MBytes barrier. Maybe this solves the
1213 problem. Not yet tried, but guessing that it could work. To get this,
1214 set unchecked_isa_dma argument of ibmmca.h from 0 to 1.
1216 5.3 Bug reports
1217 --------------
1218 If you really find bugs in the source code or the driver will successfully
1219 refuse to work on your machine, you should send a bug report to me. The
1220 best for this is to follow the instructions on the WWW-page for this
1221 driver. Fill out the bug-report form, placed on the WWW-page and ship it,
1222 so the bugs can be taken into account with maximum efforts. But, please
1223 do not send bug reports about this driver to Linus Torvalds or Leonard
1224 Zubkoff, as Linus is buried in E-Mail and Leonard is supervising all
1225 SCSI-drivers and won't have the time left to look inside every single
1226 driver to fix a bug and especially DO NOT send modified code to Linus
1227 Torvalds or Alan J. Cox which has not been checked here!!! They are both
1228 quite buried in E-mail (as me, sometimes, too) and one should first check
1229 for problems on my local teststand. Recently, I got a lot of
1230 bug reports for errors in the ibmmca.c code, which I could not imagine, but
1231 a look inside some Linux-distribution showed me quite often some modified
1232 code, which did no longer work on most other machines than the one of the
1233 modifier. Ok, so now that there is maintenance service available for this
1234 driver, please use this address first in order to keep the level of
1235 confusion low. Thank you!
1237 When you get a SCSI-error message that panics your system, a list of
1238 register-entries of the SCSI-subsystem is shown (from Version 3.1d). With
1239 this list, it is very easy for the maintainer to localize the problem in
1240 the driver or in the configuration of the user. Please write down all the
1241 values from this report and send them to the maintainer. This would really
1242 help a lot and makes life easier concerning misunderstandings.
1244 Use the bug-report form (see 5.4 for its address) to send all the bug-
1245 stuff to the maintainer or write e-mail with the values from the table.
1247 5.4 Support WWW-page
1248 --------------------
1249 The address of the IBM SCSI-subsystem supporting WWW-page is:
1251 http://www.staff.uni-mainz.de/mlang/linux.html
1253 Here you can find info about the background of this driver, patches,
1254 troubleshooting support, news and a bugreport form. Please check that
1255 WWW-page regularly for latest hints. If ever this URL changes, please
1256 refer to the MAINTAINERS file in order to get the latest address.
1258 For the bugreport, please fill out the formular on the corresponding
1259 WWW-page. Read the dedicated instructions and write as much as you
1260 know about your problem. If you do not like such formulars, please send
1261 some e-mail directly, but at least with the same information as required by
1262 the formular.
1264 If you have extensive bug reports, including Oops messages and
1265 screen-shots, please feel free to send it directly to the address
1266 of the maintainer, too. The current address of the maintainer is:
1268 Michael Lang <langa2@kph.uni-mainz.de>
1270 6 References
1271 ------------
1272 IBM Corp., "Update for the PS/2 Hardware Interface Technical Reference,
1273 Common Interfaces", Armonk, September 1991, PN 04G3281,
1274 (available in the U.S. for $21.75 at 1-800-IBM-PCTB or in Germany for
1275 around 40,-DM at "Hallo IBM").
1277 IBM Corp., "Personal System/2 Micro Channel SCSI
1278 Adapter with Cache Technical Reference", Armonk, March 1990, PN 68X2365.
1280 IBM Corp., "Personal System/2 Micro Channel SCSI
1281 Adapter Technical Reference", Armonk, March 1990, PN 68X2397.
1283 IBM Corp., "SCSI-2 Fast/Wide Adapter/A Technical Reference - Dual Bus",
1284 Armonk, March 1994, PN 83G7545.
1286 Friedhelm Schmidt, "SCSI-Bus und IDE-Schnittstelle - Moderne Peripherie-
1287 Schnittstellen: Hardware, Protokollbeschreibung und Anwendung", 2. Aufl.
1288 Addison Wesley, 1996.
1290 Michael K. Johnson, "The Linux Kernel Hackers' Guide", Version 0.6, Chapel
1291 Hill - North Carolina, 1995
1293 Andreas Kaiser, "SCSI TAPE BACKUP for OS/2 2.0", Version 2.12, Stuttgart
1294 1993
1296 Helmut Rompel, "IBM Computerwelt GUIDE", What is what bei IBM., Systeme *
1297 Programme * Begriffe, IWT-Verlag GmbH - Muenchen, 1988
1299 7 Credits to
1300 ------------
1301 7.1 People
1302 ----------
1303 Klaus Grimm
1304 who already a long time ago gave me the old code from the
1305 SCSI-driver in order to get it running for some old machine
1306 in our institute.
1307 Martin Kolinek
1308 who wrote the first release of the IBM SCSI-subsystem driver.
1309 Chris Beauregard
1310 who for a long time maintained MCA-Linux and the SCSI-driver
1311 in the beginning. Chris, wherever you are: Cheers to you!
1312 Klaus Kudielka
1313 with whom in the 2.1.x times, I had a quite fruitful
1314 cooperation to get the driver running as a module and to get
1315 it running with multiple SCSI-adapters.
1316 David Weinehall
1317 for his excellent maintenance of the MCA-stuff and the quite
1318 detailed bug reports and ideas for this driver (and his
1319 patience ;-)).
1320 Alan J. Cox
1321 for his bug reports and his bold activities in cross-checking
1322 the driver-code with his teststand.
1324 7.2 Sponsors & Supporters
1325 -------------------------
1326 "Hallo IBM",
1327 IBM-Deutschland GmbH
1328 the service of IBM-Deutschland for customers. Their E-Mail
1329 service is unbeatable. Whatever old stuff I asked for, I
1330 always got some helpful answers.
1331 Karl-Otto Reimers,
1332 IBM Klub - Sparte IBM Geschichte, Sindelfingen
1333 for sending me a copy of the w/Cache manual from the
1334 IBM-Deutschland archives.
1335 Harald Staiger
1336 for his extensive hardware donations which allows me today
1337 still to test the driver in various constellations.
1338 Erich Fritscher
1339 for his very kind sponsoring.
1340 Louis Ohland,
1341 Charles Lasitter
1342 for support by shipping me an IBM SCSI-2 Fast/Wide manual.
1343 In addition, the contribution of various hardware is quite
1344 decessive and will make it possible to add FWSR (RAID)
1345 adapter support to the driver in the near future! So,
1346 complaints about no RAID support won't remain forever.
1347 Yes, folks, that is no joke, RAID support is going to rise!
1348 Erik Weber
1349 for the great deal we made about a model 9595 and the nice
1350 surrounding equipment and the cool trip to Mannheim
1351 second-hand computer market. In addition, I would like
1352 to thank him for his exhaustive SCSI-driver testing on his
1353 95er PS/2 park.
1354 Anthony Hogbin
1355 for his direct shipment of a SCSI F/W adapter, which allowed
1356 me immediately on the first stage to try it on model 8557
1357 together with onboard SCSI adapter and some SCSI w/Cache.
1358 Andreas Hotz
1359 for his support by memory and an IBM SCSI-adapter. Collecting
1360 all this together now allows me to try really things with
1361 the driver at maximum load and variety on various models in
1362 a very quick and efficient way.
1363 Peter Jennewein
1364 for his model 30, which serves me as part of my teststand
1365 and his cool remark about how you make an ordinary diskette
1366 drive working and how to connect it to an IBM-diskette port.
1367 Johannes Gutenberg-Universitaet, Mainz &
1368 Institut fuer Kernphysik, Mainz Microtron (MAMI)
1369 for the offered space, the link, placed on the central
1370 homepage and the space to store and offer the driver and
1371 related material and the free working times, which allow
1372 me to answer all your e-mail.
1374 8 Trademarks
1375 ------------
1376 IBM, PS/2, OS/2, Microchannel are registered trademarks of International
1377 Business Machines Corporation
1379 MS-DOS is a registered trademark of Microsoft Corporation
1381 Microware, OS-9 are registered trademarks of Microware Systems
1383 9 Disclaimer
1384 ------------
1385 Beside the GNU General Public License and the dependent disclaimers and disclaimers
1386 concerning the Linux-kernel in special, this SCSI-driver comes without any
1387 warranty. Its functionality is tested as good as possible on certain
1388 machines and combinations of computer hardware, which does not exclude,
1389 that data loss or severe damage of hardware is possible while using this
1390 part of software on some arbitrary computer hardware or in combination
1391 with other software packages. It is highly recommended to make backup
1392 copies of your data before using this software. Furthermore, personal
1393 injuries by hardware defects, that could be caused by this SCSI-driver are
1394 not excluded and it is highly recommended to handle this driver with a
1395 maximum of carefulness.
1397 This driver supports hardware, produced by International Business Machines
1398 Corporation (IBM).
1400 ------
1401 Michael Lang
1402 (langa2@kph.uni-mainz.de)