| blob | d5fbbd59a2d72a2c34f56f3952416c1917d7dbf6 |
1 Thu Mar 11 23:00 1999 Gerard Roudier (groudier@club-internet.fr)
2 * revision 3.2 (8xx-896 driver bundle)
3 - Only define the host template in ncr53c8xx.h and include the
4 sym53c8xx_defs.h file.
5 - Declare static all symbols that donnot need to be visible from
6 outside the driver code.
7 - Add 'excl' boot command option that allows to pass to the driver
8 io address of devices not to attach.
9 - Add info() function called from the host template to print
10 driver/host information.
11 - Minor documentation additions.
13 Sat Mar 6 11:00 1999 Gerard Roudier (groudier@club-internet.fr)
14 * revision 3.1h
15 - Fix some oooold bug that hangs the bus if a device rejects a
16 negotiation. Btw, the corresponding stuff also needed some cleanup
17 and thus the change is a bit larger than it could have been.
18 - Still some typo that made compilation fail for 64 bit (trivial fix).
20 Sun Feb 14:00 1999 Gerard Roudier (groudier@club-internet.fr)
21 * revision 3.1g
22 - Deal correctly with 64 bit PCI address registers on Linux 2.2.
23 Pointed out by Leonard Zubkoff.
24 - Allow to tune request_irq() flags from the boot command line using
25 ncr53c8xx=irqm:??, as follows:
26 a) If bit 0x10 is set in irqm, SA_SHIRQ flag is not used.
27 b) If bit 0x20 is set in irqm, SA_INTERRUPT flag is not used.
28 By default the driver uses both SA_SHIRQ and SA_INTERRUPT.
29 Option 'ncr53c8xx=irqm:0x20' may be used when an IRQ is shared by
30 a 53C8XX adapter and a network board.
31 - Tiny mispelling fixed (ABORT instead of ABRT). Was fortunately
32 harmless.
33 - Negotiate SYNC data transfers with CCS devices.
35 Sat Jan 16 17:30 1999 Gerard Roudier (groudier@club-internet.fr)
36 * revision 3.1f
37 - Some PCI fix-ups not needed any more for PPC (from Cort).
38 - Cache line size set to 16 DWORDS for Sparc (from DSM).
39 - Waiting list look-up didn't work for the first command of the list.
40 - Remove 2 useless lines of code.
42 Sun Dec 13 18:00 1998 Gerard Roudier (groudier@club-internet.fr)
43 * revision 3.1e
44 - Same work-around as for the 53c876 rev <= 0x15 for 53c896 rev 1:
45 Disable overlapped arbitration. This will not make difference
46 since the chip has on-chip RAM.
48 Thu Nov 26 22:00 1998 Gerard Roudier (groudier@club-internet.fr)
49 * revision 3.1d
50 - The SISL RAID change requires now remap_pci_mem() stuff to be
51 compiled for __i386__ when normal IOs are used.
52 - Minor spelling fixes in doc files.
54 Sat Nov 21 18:00 1998 Gerard Roudier (groudier@club-internet.fr)
55 * revision 3.1c
56 - Ignore chips that are driven by SISL RAID (DAC 960).
57 Change sent by Leonard Zubkoff and slightly reworked.
58 - Still a buglet in the tags initial settings that needed to be fixed.
59 It was not possible to disable TGQ at system startup for devices
60 that claim TGQ support. The driver used at least 2 for the queue
61 depth but did'nt keep track of user settings for tags depth lower
62 than 2.
64 Wed Nov 11 10:00 1998 Gerard Roudier (groudier@club-internet.fr)
65 * revision 3.1b
66 - The driver was unhappy when configured with default_tags > MAX_TAGS
67 Hopefully doubly-fixed.
68 - Update the Configure.help driver section that speaks of TAGS.
70 Wed Oct 21 21:00 1998 Gerard Roudier (groudier@club-internet.fr)
71 * revision 3.1a
72 - Changes from Eddie Dost for Sparc and Alpha:
73 ioremap/iounmap support for Sparc.
74 pcivtophys changed to bus_dvma_to_phys.
75 - Add the 53c876 description to the chip table. This is only usefull
76 for printing the right name of the controller.
77 - DEL-441 Item 2 work-around for the 53c876 rev <= 5 (0x15).
78 - Add additionnal checking of INQUIRY data:
79 Check INQUIRY data received length is at least 7. Byte 7 of
80 inquiry data contains device features bits and the driver might
81 be confused by garbage. Also check peripheral qualifier.
82 - Cleanup of the SCSI tasks management:
83 Remove the special case for 32 tags. Now the driver only uses the
84 scheme that allows up to 64 tags per LUN.
85 Merge some code from the 896 driver.
86 Use a 1,3,5,...MAXTAGS*2+1 tag numbering. Previous driver could
87 use any tag number from 1 to 253 and some non conformant devices
88 might have problems with large tag numbers.
89 - 'no_sync' changed to 'no_disc' in the README file. This is an old
90 and trivial mistake that seems to demonstrate the README file is
91 not often read. :)
93 Sun Oct 4 14:00 1998 Gerard Roudier (groudier@club-internet.fr)
94 * revision 3.0i
95 - Cosmetic changes for sparc (but not for the driver) that needs
96 __irq_itoa() to be used for printed IRQ value to be understandable.
97 - Some problems with the driver that didn't occur using driver 2.5f
98 were due to a SCSI selection problem triggered by a clearly
99 documented feature that in fact seems not to work: (53C8XX chips
100 are claimed by the manuals to be able to execute SCSI scripts just
101 after abitration while the SCSI core is performing SCSI selection).
102 This optimization is broken and has been removed.
103 - Some broken scsi devices are confused when a negotiation is started
104 on a LUN that does not correspond to a real device. According to
105 SCSI specs, this is a device firmware bug. This has been worked
106 around by only starting negotiation if the LUN has previously be
107 used for at least 1 successful SCSI command.
108 - The 'last message sent' printed out on M_REJECT message reception
109 was read from the SFBR i/o register after the previous message had
110 been sent.
111 This was not correct and affects all previous driver versions and
112 the original FreeBSD one as well. The SCSI scripts has been fixed
113 so that it now provides the right information to the C code.
115 Sat Jul 18 13:00 1998 Gerard Roudier (groudier@club-internet.fr)
116 * revision 3.0g
117 - Preliminary fixes for Big Endian (sent by Eddie C. Dost).
118 Big Endian architectures should work again with the driver.
119 Eddie's patch has been partially applied since current 2.1.109
120 does not have all the Sparc changes of the vger tree.
121 - Use of BITS_PER_LONG instead of (~0UL == 0xffffffffUL) has fixed
122 the problem observed when the driver was compiled using EGCS or
123 PGCC.
125 Mon Jul 13 20:00 1998 Gerard Roudier (groudier@club-internet.fr)
126 * revision 3.0f
127 - Some spelling fixes.
128 - linux/config.h misplaced in ncr53c8xx.h
129 - MODULE_PARM stuff added for linux 2.1.
130 - check INQUIRY response data format is exactly 2.
131 - use BITS_PER_LONG if defined.
133 Sun Jun 28 12:00 1998 Gerard Roudier (groudier@club-internet.fr)
134 * revision 3.0e
135 - Some cleanup, spelling fixes, version checks, documentations
136 changes, etc ...
138 Sat Jun 20 20:00 1998 Gerard Roudier (groudier@club-internet.fr)
139 * revision 3.0c
140 - Add a boot setup option that allows to set up device queue depths
141 at boot-up. This option is very usefull since Linux does not
142 allow to change scsi device queue depth once the system has been
143 booted up.
145 Sun Jun 15 23:00 1998 Gerard Roudier (groudier@club-internet.fr)
146 * revision 3.0a
147 - Support for up to 64 TAGS per LUN.
148 - Rewrite the TARGET vs LUN capabilities management.
149 CmdQueue is now handled as a LUN capability as it shall be.
150 This also fixes a bug triggered when disabling tagged command
151 queuing for a device that had this feature enabled.
152 - Remove the ncr_opennings() stuff that was useless under Linux
153 and hard to understand to me.
154 - Add "setverbose" procfs driver command. It allows to tune
155 verbose level after boot-up. Setting this level to zero, for
156 example avoid flooding the syslog file.
157 - Add KERN_XXX to some printk's.
159 Tue Jun 10 23:00 1998 Gerard Roudier (groudier@club-internet.fr)
160 * revision 3.0
161 - Linux config changes for 2.0.34:
162 Remove NVRAM detection config option. This option is now enabled
163 by default but can be disabled by editing the driver header file.
164 Add a PROFILE config option.
165 - Update Configure.help
166 - Add calls to new function mdelay() for milli-seconds delay if
167 kernel version >= 2.1.105.
168 - Replace all printf(s) by printk(s). After all, the ncr53c8xx is
169 a driver for Linux.
170 - Perform auto-sense on COMMAND TERMINATED. Not sure it is usefull.
171 - Some other minor changes.
173 Tue Jun 4 23:00 1998 Gerard Roudier (groudier@club-internet.fr)
174 * revision 2.6n
175 - Code cleanup and simplification:
176 Remove kernel 1.2.X and 1.3.X support.
177 Remove the _old_ target capabilities table.
178 Remove the error recovery code that have'nt been really usefull.
179 Use a single alignment boundary (CACHE_LINE_SIZE) for data
180 structures.
181 - Several aggressive SCRIPTS optimizations and changes:
182 Reselect SCRIPTS code rewritten.
183 Support for selection/reselection without ATN.
184 And some others.
185 - Miscallaneous changes in the C code:
186 Count actual number of CCB queued to the controller (future use).
187 Lots of other minor changes.
189 Wed May 13 20:00 1998 Gerard Roudier (groudier@club-internet.fr)
190 * revision 2.6m
191 - Problem of missed SCSI bus reset with the 53C895 fixed by
192 Richard Waltham. The 53C895 needs about 650 us for the bus
193 mode to settle. Delays used while resetting the controller
194 and the bus have been adjusted. Thanks Richard!
195 - Some simplification for 64 bit arch done ccb address testing.
196 - Add a check of the MSG_OUT phase after Selection with ATN.
197 - The new tagged queue stuff seems ok, so some informationnal
198 message have been conditionned by verbose >= 3.
199 - Donnot reset if a SBMC interrupt reports the same bus mode.
200 - Print out the whole driver set-up. Some options were missing and
201 the print statement was misplaced for modules.
202 - Ignore a SCSI parity interrupt if the chip is not connected to
203 the SCSI bus.
205 Sat May 1 16:00 1998 Gerard Roudier (groudier@club-internet.fr)
206 * revision 2.6l
207 - Add CCB done queue support for Alpha and perhaps some other
208 architectures.
209 - Add some barriers to enforce memory ordering for x86 and
210 Alpha architectures.
211 - Fix something that looks like an old bug in the nego SIR
212 interrupt code in case of negotiation failure.
214 Sat Apr 25 21:00 1998 Gerard Roudier (groudier@club-internet.fr)
215 * revision 2.6k
216 - Remove all accesses to the on-chip RAM from the C code:
217 Use SCRIPTS to load the on-chip RAM.
218 Use SCRIPTS to repair the start queue on selection timeout.
219 Use the copy of script in main memory to calculate the chip
220 context on phase mismatch.
221 - The above allows now to use the on-chip RAM without requiring
222 to get access to the on-chip RAM from the C code. This makes
223 on-chip RAM useable for linux-1.2.13 and for Linux-Alpha for
224 instance.
225 - Some simplifications and cleanups in the SCRIPTS and C code.
226 - Buglet fixed in parity error recovery SCRIPTS (never tested).
227 - Minor updates in README.ncr53c8xx.
229 Wed Apr 15 21:00 1998 Gerard Roudier (groudier@club-internet.fr)
230 * revision 2.6j
231 - Incorporate changes from linux-2.1.95 ncr53c8xx driver version.
232 - Add SMP support for linux-2.1.95 and above.
233 - Fix a bug when QUEUE FULL is returned and no commands are
234 disconnected. This happens with Atlas I / L912 and may happen
235 with Atlas II / LXY4.
236 - Nail another one on CHECK condition when requeuing the command
237 for auto-sense.
238 - Call scsi_done() for all completed commands after interrupt
239 handling.
240 - Increase the done queue to 24 entries.
242 Sat Apr 4 20:00 1998 Gerard Roudier (groudier@club-internet.fr)
243 * revision 2.6i
244 - CTEST0 is used by the 53C885 for Power Management and
245 priority setting between the 2 functions.
246 Use SDID instead as actual target number. Just have had to
247 overwrite it with SSID on reselection.
248 - Split DATA_IN and DATA_OUT scripts into 2 sub-scripts.
249 64 segments are moved from on-chip RAM scripts.
250 If more segments, a script in main memory is used for the
251 additionnal segments.
252 - Since the SCRIPTS processor continues SCRIPTS execution after
253 having won arbitration, do some stuff prior to testing any SCSI
254 phase on reselection. This should have the vertue to process
255 scripts in parallel with the SCSI core performing selection.
256 - Increase the done queue to 12 entries.
258 Sun Mar 29 12:00 1998 Gerard Roudier (groudier@club-internet.fr)
259 * revision 2.6h
260 - Some fixes.
262 Tue Mar 26 23:00 1998 Gerard Roudier (groudier@club-internet.fr)
263 * revision 2.6g
264 - New done queue. 8 entries by default (6 always useable).
265 Can be increased if needed.
266 - Resources management using doubly linked queues.
267 - New auto-sense and QUEUE FULL handling that does not need to
268 stall the NCR queue any more.
269 - New CCB starvation avoiding algorithm.
270 - Prepare CCBs for SCSI commands that cannot be queued, instead of
271 inserting these commands into the waiting list. The waiting list
272 is now only used while resetting and when memory for CCBs is not
273 yet available?
275 Sun Feb 8 22:00 1998 Gerard Roudier (groudier@club-internet.fr)
276 * revision 2.6f
277 - Some fixes in order to really support the 53C895, at least with
278 FAST-20 devices.
279 - Heavy changes in the target/lun resources management to allow
280 the scripts to jump directly to the CCB on reselection instead
281 of walking on the lun CCBs list. Up to 32 tags per lun are now
282 supported without script processor and PCI traffic overhead.
284 Sun Jan 11 22:00 1998 Gerard Roudier (groudier@club-internet.fr)
285 * revision 2.6d
286 - new (different ?) implementation of the start queue:
287 Use a simple CALL to a launch script in the CCB.
288 - implement a minimal done queue (1 entry :-) ).
289 this avoid scanning all CCBs on INT FLY (Only scan all CCBs, on
290 overflow). Hit ratio is better than 99.9 % on my system, so no
291 need to have a larger done queue.
292 - generalization of the restart of CCB on special condition as
293 Abort, QUEUE FULL, CHECK CONDITION.
294 This has been called 'silly scheduler'.
295 - make all the profiling code conditionned by a config option.
296 This spare some PCI traffic and C code when this feature is not
297 needed.
298 - handle more cleanly the situation where direction is unknown.
299 The pointers patching is now performed by the SCRIPTS processor.
300 - remove some useless scripts instructions.
302 Ported from driver 2.5 series:
303 ------------------------------
304 - Use FAST-5 instead of SLOW for slow scsi devices according to
305 new SPI-2 draft.
306 - Make some changes in order to accomodate with 875 rev <= 3
307 device errata listing 397. Minor consequences are:
308 . Leave use of PCI Write and Invalidate under user control.
309 Now, by default the driver does not enable PCI MWI and option
310 'specf:y' is required in order to enable this feature.
311 . Memory Read Line is not enabled for 875 and 875-like chips.
312 . Programmed burst length set to 64 DWORDS (instead of 128).
313 (Note: SYMBIOS uses 32 DWORDS for the SDMS BIOS)
314 - Add 'buschk' boot option.
315 This option enables checking of SCSI BUS data lines after SCSI
316 RESET (set by default). (Submitted by Richard Waltham).
317 - Update the README file.
318 - Dispatch CONDITION MET and RESERVATION CONFLICT scsi status
319 as OK driver status.
320 - Update the README file and the Symbios NVRAM format definition
321 with removable media flags values (available with SDMS 4.09).
322 - Several PCI configuration registers fix-ups for powerpc.
323 (Patch sent by Cort).