
System Code Revision: 8ck3v06k (LVD), 8ch3v06k (HVD), 8cf3v06k (Fibre), 8ce3v06k (E-copy)


Build Date: 05-03-02

Changes to Code from v06h:
Item
Change
Impact
Date of Change
1 
Added a change to allow the drive to properly eject an expired cleaning tape.
Cleaning tape handling improvement
05/02/02
2 
Fibre channel only:
Added code to abort logical services when loss of synch is detected during an exchange
Error recovery improvement
05/02/02
3 
Fibre channel only:
Added a method to read the FAS440 Registers
Debug improvement
05/02/02



System Code Revision: 8ck3v06g (LVD), 8ch3v06g (HVD), 8cf3v06g (Fibre), 8ce3v06g (E-copy)

Build Date: 04-05-02

Changes to Code from v06f:
Item
Change
Impact
Date of Change
1 
When the drive needs cleaning, the LCD will flash CLEAN SOON on/off 6 seconds at a time whether a tape is loaded or not.
Customer request
032602
2 
 Added code to check for LEOP before allowing DFS during 22M tape write error recovery. 
Write recovery improvement
032602
3 
Changed error thresholds related to FSC 9B error reporting and write recovery.
Write recovery improvement
032602
4 
Enhanced DFS limiting on 22M tape near LEOP.
Write recovery improvement
032602
5 
Better handling of writing filemarks into PEOT.
Rare write hang.
032602
6 
On 22m tapes, move forward the LEOP warning to be consistent with other tape sizes  (500 MB).
Code improvement
032602
7 
Prevents a false failure (FSC C9) from occurring in the recovery manager when completing the hybrid clean and corrects condition that causes hang 4002.
SmartClean during tapeload improvement.  
032702
8 
Added code to check for PEOT interrupt during a motion recovery.
22m tape write recovery improvement.
032802
9 
Change unique inquiry page C7 byte 28 be reported as 0x07 instead of 0x02 for LVD drives.
Customer-unique request
040102
10 
Additional  checks added before trying a DFS when encountering THL and PBOP. 
Rare THL hang 4171 seen in Product test after a  write error
040102
11 
Returned the SmartClean maintenance interval to 100 hours. 
Drive reliability improvement
040202
12 
These changes will have the LCD display CLEAN SOON message alternate with tape action displays such as write, read or rewind. Will flash on/off every 5 - 6 seconds alternating display with whatever would be there otherwise. 
None. customer request.
040202
13 
 Added changes to the check for coasting and splice adjust to allow moving of  the splice point to prevent potential FSC 9D errors.
Write recovery improvement
040202
14 
Cleared the abort control bit for correct tape status on an Unload after a write failure.
Unload code improvement.
040202
15 
Increase the length of the 3rd try DFS.Added forced clean if threshold level falls to zero for any 1000 track sample.
Write recovery improvement
040202
16 
Modifed handling of  some PR code accounting errors.
Code improvement
040302
17 
Fixed PR shutdown issues in redirecting 2148 and 4290 hang errors.
Code improvement
040502
18 
Minor changes made to recovery manager to address hang 9304.
Rare hang seen in Firmware Lab
040502
19 
Fixed 256+ filemarks read problem seen in specialized test cases.
Rare hang seen in Firmware Lab
040502


System Code Revision: 8ck3v06f (LVD), 8ch3v06f (HVD), 8cf3v06f (Fibre), 8ce3v06f (E-copy)

Build Date: 03-22-02

Changes to Code from v06e:
Item
Change
Impact
Date of Change
1 
Change lower SER limit before posting FSC 9B to 2.30
Code improvement
032202
2 
Fixed Space timeouts related to a large number of filemarks and slow data transfer rates.
Code improvement
032202

System Code Revision: 8ck3v06d (LVD), 8ch3v06d (HVD), 8cf3v06d (Fibre), 8ce3v06d (E-copy)

Build Date: 03-15-02

Changes to Code from v06c:
Item
Change
Impact
Date of Change
1 
Changed method of initializing  the write current for all four write heads .
Write improvement
031402
2 
Disable FSC 9B.

031402
3 
Added limits to 22M tape usable area.
Write improvement.
031502


 System Code Revision: 8ck3v06e (LVD), 8ch3v06e (HVD), 8cf3v06e (Fibre), 8ce3v06e (E-copy)

Build Date: 03-21-02

Changes to Code from v06a:
Item
Change
Impact
Date of Change
1 
Fixed rare 9301 hang during SmartClean.
Code improvement.
030402
2 
Added changes for recovery partition searches.
Rare partition switch timeout.
030402
3 
Fibre/Ecopy only:  Added  changes so that sector size was same as frame size.  This resolves problems with sending out frames such as: 2048, 2048, 1024, 1536 which some hard drives cannot handle. 
Fibre-unique SanMark certification
030402
4 
Fibre only:  Making sure Xcopy.NPortPort is set to the correct value for E1 descriptors.
Fixed problem where LIP would be generated long after receiving LIP.  Also reduced time for FC bus to become active after poweron. Added fcFrameQueue initialization.
Fibre-unique issues
030402
5 
Fibre only:  REPORT_LUNS bug fixed by adding report luns CDB.
Fibre-unique issue with Report LUNs functionality
030502
6 
Fixed 2303 hang on a host reset after a timeout 
Rare hang on a host reset.
030602
7 
Added  fixes  for SE_TRACKING_LOST and SE_TRACKING_REGAINED to address a C015 hang.
Rare servo hang
030602
8 
Changed the parameter to use the last known values of the search manager. This will allow a more accurate position to issue a retry.
Space recovery improvements
030702
9 
This addresses locate/space timeout problems where physical read was not continuing to fill after releasing holes.
Rare Locate timeout issue.
030702
10 
Made changes for forced clean and smart clean history.
Recovery improvements
030702
11 
Added additional parameters to the internal trace facilities and made some efficiency improvements. 
Tracing improvements
031402
12 
Added  in saving traces and debug information during a reset or abort. 
Tracing improvements
031502
13 
Added counters for all write and read recovery actions.
Code improvement
031802
14 
Added enhancements for check of returned ILLEGAL_CMD status in recovery to address 9304 hang.
Code improvements
032002
15 
Fix for a 230B Write hang.
Rare write hang
031902
16 
Corrected a potential unload time out.
Servo error improvements.
032002
17 
Fix for a hang 6589 which could happen during a space to EOD after an unrecoverable Read Error.
Rare read hang.
032102
18 
Fixed the amber LED and LCD indications after failed codeload due to resetting of drive. Added changes so that some preliminary checking-induced codeload failures (those which do not affect DRAM codeload buffer) do not reset the drive.
Codeloading improvement
032102


System Code Revision: 8ck3v06a (LVD), 8ch3v06a (HVD), 8cf3v06a (Fibre), 8ce3v06a (E-copy)

Build Date: 03-01-02

Changes to Code from v05e:
Item
Change
Impact
Date of Change
1 
Fixed 4209 read hang  caused by the  hole release process  in read error recovery.

Rare read hang
012802
2 
Corrected ending sequence during  PEOT shutdown to prevent potential 4290 hang.
Rare read hang
012802
3 
Corrected multiscan to tracking switchover to check the free list count and not remove more entries than are available.
Rare hang
012902
4 
Changed some default values for SSI channel registers per SSI spec.
Code improvement
012902
5 
Added setting of NEED_A_SMART_CLEAN flag when drive exceeds cleaning time of 30 hours.
Recovery improvement
012902
6 
Added ability to force a DFS and only allow a max of 4 retries at doing a DFS.
Write recovery improvements
012902
7 
Added the ability for no block status to use multiscan. 
Space recovery improvements.
012902
8 
Added code to correct an unload command timeout related to attempting to write the THL if doing unload retries.  
Unload recovery improvements.
012902
9 
Prevent drum counter-rotate recovery if abort is in progress.
Servo error improvements.
012902
10 
Disable drum stick recovery if abort is in progress.
Servo error improvements.
012902
11 
Enhancements added to searching  in reverse for a filemark to address a 6588 hang. 
Space filemark improvements.
020602
12 
Prevent FSC 38 by adding search forward for a filemark as part of recovery if  previous servo movement  has positioned tape behind LBOP.
Space positioning improvements.
020602
13 
Errors which occur during the writing of PBOP or the THL will be treated as normal fail types. 
Code improvement
020602
14 
Forced DFS attempts added as part of FSC  9A recovery. Updated stats in pw_tapes_written on tape unload.
Write recovery  improvements
020602
15 
Corrects an initialization problem which could cause stack corruption. 
Some rare 23xx hangs
021202
16 
Changes added to move data from the lport bi fifo to the kodiak buffer to finish a scsi write transfer.
Rare write timeout issue.
021402
17 
Perform a pr past range shutdown if an EOD is seen that's out of range and we're not in hole mode.
Rare read hang
021502
18 
Added improvement in method to search  for  LID's during buffer fill.
Rare space timeout
021502
19 
Added change during reset process when a codeload fails. 
Customer-unique tape codeloading improvement.
021502
20 
Implement an SER  threshold measurement based on a windowed  realtime SER for all 4 tracks.  The trigger threshold is a preset number of tracks where the SER reading falls below a defined threshold. When that occurs, a forced SmartClean is requested after which another preset number of  tracks are sampled. If there is no improvement of SERs, an FSC 9B is posted.
Write recovery improvements
021502
21 
For hardware errors prevent allowing an eject without a reset.
Firmware improvement
021902
22 
Added "Illegal Tape" message when a tape is attempted to be loaded upside down.
Customer request
021902
23 
Removed references to library_mode for GE_EXB_LIBRARY as part of general  clean up of all routines that uses m_Process_Mail() with proper argument passing.
Automation enhancement request
022002
24 
Added back in the M1 tape handling routines for no LBOP processing to prevent a 2304 hang. 
Recovery improvement
022002
25 
Modified all routines to handle proper argument passing with m_Process_Mail() to prevent A802 hangs when no search fields detected.
Recovery improvement
022102
26 
Added special handling for FSC 9B in lBuild_sense_data to call it a  SK 03.
Customer-unique request
022102
27 
Change to rely on PW to give a "go no-go" instead of a limiting number of DFS.
Write recovery improvements
022102
28 
PW now tracks DFS in terms of tracks used. This removes the artificial 11 physical defects and on the 12th the posting of  FSC 9D.
Write recovery improvements
022102
29 
Prevent false illegal tape errors  when sensing a  BOT in Rewind.
Firmware improvement.
022102
30 
False Read Warning Tape Alert flag being set.
Read Tape Alert enhancements
022202
31 
Removed tape alert messages for DFS. Now will only send Critical tape alert message for either FSC 9A or 9D.
Write Tape Alert enhancements
022202
32 
The number of tracks available for DFS is now 10% of total tracks. 
Write recovery improvements
022202
33 
Changes made to non-buffered write mode to account for a pre-existing, partially-filled cu to prevent rare 2128 hang.
Firmware improvement.
022502
34 
Fixed rare problem where a prior failure resulted in the tape being left in the drive and an unload error was generated. This left the drive in a state so that the only way to remove the tape was by cycling the power.
Firmware improvement.
022602
35 
Defect Skip close to LEOP is now monitored more closely and should the DFS fail and have to be restarted the  DFS multiplier will be adjusted so as not to go past LEOP.

Write recovery improvement
022602
36 
Fibre only: Change handling of  buffer full by recovering buffer space rather than hanging.
Rare Fibre hang
022702
37 
This changes saves the current supply radius value in a sector of the flash memory on power down.   It is restored on power up.  This prevents possible  tape damage when powering up a drive with a tape still loaded in it.
Tape handling improvement
022702
38 
Ecopy only: Fixed invalid write command control bit being set.
Firmware improvement.
022702
39 
Ecopy only: Fixed incorrect handling of  a rejected command.
Firmware improvement.
022702
40 
Fixes problem with tapes not ejecting after Unload  Button aborts operation when in a write recovery.
Tape handling improvements.
022702
41 
Rare Format problem causing 9304 hang.
Firmware improvement.
022802
42 
Fixed an  event where the partition search manager aborts the search, but did not abort the physical services causing a 4303 hang. 
Rare search hang .
022802
43 
Changed WHEEWHEE_TRACKS from 25m to 10m.
Write recovery improvement
030102
44 
Implements a special DFS (up to 6 meters) to help get around a large section of  bad tape without violating the maximum SCSI command timeout window of 7 minutes.
Write recovery improvement
030102


System Code Revision: 8ck3v05e (LVD), 8ch3v05e (HVD), 8cf3v05e (Fibre), 8ce3v05e (E-copy)

Build Date: 02-22-02

Changes to Code from v05d:
Item
Change
Impact
Date of Change
1 
Fibre only: Fixes issue with SanMark re-LIP hang
SanMark certification
022202

System Code Revision: 8ck3v05d (LVD), 8ch3v05d (HVD), 8cf3v05d (Fibre), 8ce3v05d (E-copy) - EC RELEASED FOR HVD SCSI

Build Date: 01-28-02

Changes to Code from v05c:
Item
Change
Impact
Date of Change
1 
Fibre only: Fixes issue which incorrectly incremented Sequence ID on odd sector boundaries on start of new sequence.  
Firmware improvement
012302
2 
Fibre only: fixes a problem seen with LIP during writes causing a 600B hang.  
Rare Fibre hang
012302
3 
Fixes HVD codeloading problem seen by Mfg.  
Firmware improvement.
012502


Mammoth-2 Firmware Release Notes		05/22/02

Exabyte Confidential		Page 9			

