AIC/XTORE
SAS SERVER/JBOD Expander Firmware release note
Last updated:July-16th-2008
This
guide provides information on AIC/XTORE SAS products firmware Release note for
the following Models. Review this entire document before updating the firmware
|
Models
|
Brand
|
Firmware Revision
|
Remark
|
|
XJ-SA13/26-110R -x
|
AIC/ Xtore
|
xxxxxxxx
|
1U/10 SAS JBOD Series
|
|
XJ-SA13/26-224R-x
|
AIC/ Xtore
|
xxxxxxxx
|
2U/24 SAS JBOD Series
|
|
XJ-SA13/26-212R-x
|
AIC/ Xtore
|
xxxxxxxx
|
2U/12 SAS JBOD Series
|
|
XJ-SA12/24-316R-x
|
AIC/ Xtore
|
xxxxxxxx
|
3U/16 SAS JBOD Series
|
|
RSC-1A-SA1S/D-2
|
AIC/ Xtore
|
xxxxxxxx
|
1U/10 SAS Server Series
|
|
RSC-2A-SA1S/D-2
|
AIC/ Xtore
|
xxxxxxxx
|
2U/24 SAS Server Series
|
|
RSC-2EC-SA1S/D-2
|
AIC/ Xtore
|
xxxxxxxx
|
2U/12 SAS Server Series
|
|
RSC-3EC-SA1S/D-2
|
AIC/ Xtore
|
xxxxxxxx
|
3U/16 SAS Server Series
|
Firmware
Editions history:
Ø
1.14H.03 Release note (July-16th-2008)
Ø
1.14H.02 Release note (May-27th-2008)
Ø
1.14H.01 Release note (May-23rd-2008)
Ø 1.14F0.1 Release note (Jun-27th-2007)
Ø 1.09G Release note (Oct-17th-2006)
Ø 1.09D Release note (Sep-15th-2006)
Ø 1.07C Release note (July-17th-2006)
Ø 1.06d Release Note (April-11th-2006)
----------------------------------------------------------------------------------------------------------------
Release note for
1.14H.03 Date: July-16th-2008 Back to Top
Fixed fault LED issue with Areca SAS HBA
Bug Description
For now, we just ignore the SES 'rqstMissing' control bit sent from Areca HBA so the faulty LED of the drive bay won't be turned on when a drive not configured as RAID member is unplugged
Due to design limit ion in our code, the expander module at slot-A won't be able
to sync. its SAS address with the master controller module at slot-B and will
have an null SAS address.
----------------------------------------------------------------------------------------------------------------
Release note for
1.14H.02 Date: May-27th-2008 Back to Top
A serious
bug in 1.14H.01 FW for all of SAS enclosures was found and fixed in 1.14H.02.
Bug Description
In a
dual-modules setup, the expander module plugged in slot-B mistakenly insists on
taking the role as a Master enclosure controller and refuses granting the
request from expander module at slot-A for the Master controller role.
Due to design limit ion in our code, the expander module at slot-A won't be able
to sync. its SAS address with the master controller module at slot-B and will
have an null SAS address.
Note:
The above symptom manifests only when both expander modules are present when
powering up the enclosure; It won't notice the problem if only one module
installed or even another module hot plugged afterward.
----------------------------------------------------------------------------------------------------------------
Release note for 1.14H.01 Date: May-23rd-2008 Back to Top
1.14H.01 Release Highlight:
- Enhancement to the
staggering spin-up implementation
- Aggressively reducing
disk staggering spin-up delay
- All models for
2.5" drive default to 3 seconds of spin-up delay for each disk
group
- All models for
3.5" drive default to 5 seconds of spin-up delay for each disk
group
- Reorganizing and
reducing spin-up disk group
- For all models, the
number of disks grouped to be spontaneously spun up is increased (and
thus spin-up groups reduced).
- In general, the above
enhancements shall drastically speed up the power on process, with some
side effects:
- The code boot issue
with LSI 8888ELP, 8708ELP, and a Marvell RAID board, caused by their
much shorter time span in enumerating attached devices, is considered
fixed, except those models (2U24 server, 2U12 JBOD and server) with H/W
patches mandated by OEM engineer (the patches adds up 10 seconds delay
to the 7153/7154 reset sequence).
- The
increased spontaneously spun up disks will surely put more stress on the
power supply units. The implication of this will be carefully analyzed
and verified by our R/D.
- Adding a console command
'tSpinupDelay' fine tuning the spin-up delay
tSpinupDelay [delay_in_seconds
(0 - 32)]
the delay_in_seconds parameter:
can be omitted to get the current value.
can be given a value of '0' to reset to model specific system
default.
- Supporting the retrieving
of full FW version stamp
We hijacked the SCSI 'String In' Diagnostic page (Page Code 04H) to report
the full version stamp of the SEP FW:
- The returned string
will always be in the form of 'NN.nnX.nn', such as '1.14h.04' or
'1.14H.01'.
- Fixing several bugs in
controlling drive bay LED when requested to change state of a SES
ArrayDevice element
- Correcting the LED
behavior when requested to change from some states via the [RqstOk,
RqstRebuilding] request.
- Correcting the LED
behavior when requested to turn off the access LED (the blue one) with
the 'RqstOk' flag reset (set to 0).
- These bugs only
manifest themselves when special SES requests sequences are called, and:
- You may spot a few LED
behavior changes when running mainstream SAS management suits but all
seems to make the indicators more reasonable and understandable.
- Your home
brewed SES client may or may not be affected (in positive or negative
way) by the changes; carefully testing your application is advised.
- Fixing a SMP bug that
reports SATA device presence prematurely
- Fixing a bug in SES
'Download Microcode' control page
Correcting the returned status code on each received data chunk and the
final status code upon completion or aborting.
- This fix should make it
a lot easier for writing your own FW download client.
- Fixing a minor bug in
reporting the value of the 5V DC sensor upon SES requests
- The returned value of
DC 5V sensor was not normalized from the raw reading of the LM80 chip.
- Because the calculation
of the normalization is quite a heavy load to the 7153/7154 core, the bug
is fixed with a workaround:
The description of the DC 5V sensor is changed to '1.9V'; the returned
value of the nominal readings is usually in the range of 190±8.
________________________________________________________________________________________________________________________
Release note for 1.14F.01 Date: Jun-27th-2007
Back to Top
1.14F.01 Release Highlight:
- Fixed a critical bug in replying SCSI Inquiry command for the
'Device Identification' VPD page of the enclosure SES device.
- In 1.14F.00 and 1.14f.xx, the SAS
address of the enclosure SES device is not returned if this address is
not exposed to SMP client (controlled by 'tExposeELI on/off' cmd, the
default is off). This severely violates SAS specification.
- This bug may confuse host in
enumerating disk drives in a multiple expanders environment (either
cascading or not)
- With this fix, Adaptec 3085 RAID board, when connecting with
both expander modules in one enclosure, can now list correct drive
population and sustain RAID operations upon failure of a single expander
module. This may also applies to other Adaptec RAID boards with multiple
4i or 4x connectors..
1.14F Release
Highlight:
- Merge
in all low level libraries from Vitesse SDK 1.14.
You can use the 'stat' console command to verify the version of those low
level libraries.
- Formally
added customized version for OEM models:
- The
image files are packed separately in '1.15F-schg.tar.gz'.
Note: This production release should not be released to OEM until
approving of 1.14f.5 beta by them.
- Major
customization for these OEM models are:
- The
host port signal strength is boosted to support long cable (max. 6
meters)
- For
the 2U24 server model:
- 'Split-Expander'
feature (see Highlight 5) is activated by default.
- Thermal
sensitive fan control is disable and SES fan element support is
removed.
- Added
customized general OEM version the 2U24 JBOD/server and 2U12 JOBD models:
- The
image files are packed separately in '1.15F-oem.tar.gz'.
- Major
customization for these OEM models are:
- The
host port signal strength is boosted to support long cable (max. 6
meters).
- Discontinuing support of older version of backplane
because new (and fancy) drive bay LED behavior for SES Array Device
Element can't be implemented on those backplanes.
- SES-2
disk device support is reverted from 'Device Element' to 'Array Device
Element' (except on OEM model).
- Most
'Array Device Element' specific control and status functions are
supported.
- Hardware
based LED pulse train for most supported status indication.
The following table depicts the LED flash pattern for each supported control/status
field in SES-2 Enclosure Control/Status page (page code 0x02) for Array
Device Element:
|
Supported
Control
|
Supported
Status
|
LED Pulse for Status
|
Priority
|
Note
|
|
Rqst Ok
|
Ok
|
Access LED constantly ON when:
1. this bit is on, or
2. this bit being off but hot-spare bit is on
|
2
|
'Ident' bit has 1st priority in governing Access LED.
|
|
Rqst
Rsvd Device
(Reserved Device)
|
Rsvd
Device
|
Error LED flash. Pulse train segment:
000111000000
(in about 5 seconds)
|
7
|
|
|
Rqst
Hot Spare
|
Hot
Spare
|
Same as 'Rsvd Device'
|
6
|
|
|
Rqst
Cons Check
(Consistency Check)
|
Cons
Check
|
Error LED flash. Pulse train segment:
010101010101
(in about 0.502 second, e.g., fast flashing)
|
5
|
Always accompanying with Access LED flashing
Not applicable to OEM
|
|
Rqst in
Crit Array
(In Critical Array)
|
In Crit
Array
|
Error LED flash. Pulse train segment:
0110000000
(in about 0.418 second)
|
4
|
|
|
Rqst in
Failed Array
|
In
Failed Array
|
Same as 'In Critical Array'
|
3
|
|
|
Rqst
Rebuild/Remap
|
Rebuild/Remap
|
Same as 'Cons Check'
|
2
|
Always accompanying with Access LED flashing
Not applicable to OEM
|
|
Rqst
R/R Abort
(Rebuild/Remap Abort)
|
R/R
Aborted
|
Error LED flash. Pulse train segment:
10000011
(in 1 about 1 Hz)
|
1
|
|
|
Rqst
Missing
|
N/A
|
None
|
N/A
|
Event recorded without giving visual indication.
|
|
Rqst
Insert
|
Ready
to Insert
|
Error LED flash in 1Hz
|
8
|
|
|
Rqst
Remove
|
Ready
to Remove
|
Error LED flash in 1Hz
|
8
|
|
|
N/A
|
Fault
Sensed
|
Error LED constantly ON
|
8
|
|
|
Rqst
Fault
|
Fault
REQSTD
|
Error LED constantly ON
|
9
|
Usually accompanying with 'Ok' bit off and thus Access LED OFF
|
|
Rqst
Ident
|
Ident
|
Access LED flash in 1Hz
|
1
|
|
- Synchronizing
SES Enclosure Logical ID (ELI) of primary and secondary expander modules
in a single enclosure box.
SEP of both expander modules in a dual-expander enclosure are given same SES
Enclosure Logical ID that's world-wide unique. In our
implementation, this ID is the SAS address of the virtual SSP PHY of
the master expander module.
- Exposing
SES Enclosure Logical ID (ELI) to SMP client.
To work around the incompatibility of this feature with LSI non-RAID
HBA, a new console command 'tExposeELI', is added to control this
feature:
- tExposeELI
on:
enable it.
- tExposeELI
off:
disable it.
- tExposeELI
(no argument) : show the activation state.
- The
'tEnclousreState' console command is enhanced to report current SES
Enclosure Logical ID.
- Added
support of the header 'EIP' bit and each element's 'bay number' field in
response to SES request page 0Ah (Additional Element Status)
- This
is necessary so that SEP client can correctly report the bay position of
each drive. This has been verified with SAS management GUI suite from
Adaptec and LSI.
- This change may break your
customized application if your SES library doesn't take into account
of the header 'EIP' bit and react accordingly.
- The slot number field counts from 1
except on OEM models (counts from 0 per requested).
- Added
the 'Split Expander' feature.
This feature enables a single expander module to provision 2 virtual
expanders to a single HBA connected with 2 host ports.
Some
notes about the current implementation:
- This feature is in fact a hack and violating SMP and SSP
standard. It only works ad hoc with LSI HBA and RAID boards and it is
meant to work around the issue of LSI boards not capable of aggregating
SAS links across host ports.
- Split-expander
mode is disabled by default (except on specific OEM models).
A new, hidden console command 't2Expanders' is added to control the
activation of it:
- t2Expanders
SplitIt: Activated split-expander mode on next power
circle.
- t2Expanders
off:
Restored to normal mode on next power circle.
- t2Expanders
(without parameter): Display current state.
- The 'tZoneCfg' console command is enhanced to cop with
the special zoning requirement by split-expander mode:
- 'tZoneCfg
-d': display active zoning state and manual zone configuration
stored in NVRAM.
This help user to have a glance on the implicit zone configuration by
split-expander feature vs. explicit zone configuration manually set up .
- 'tZoneCfg
-s' (for manually configuring zone) is disabled when split-expander
mode is active.
- Activation status of 'split-expander' is also reported by
'tExpanderState' and 'tShowState' commands.
- This feature is only supported by expander module with 2 or
more host (up-link) ports.
- In
current implementation:
Phys for the host port #1 plus phys for the first half of drive bays;
e.g., bay #1 through #12 in the case of 24-bay box, are configured into
zone #1.
Phys for the host port #2 and phys for the last half of drive bays, e.g.,
bay #13 through #24 in the case of 24-bay box, are configured into zone
#2.
This means:
- A
HBA connecting to the expander box with single 4X port will see only
half of total drives.
- A
HBA connecting to the expander box with dual 4X ports will see 2
expanders, each containing half of total drives.
- Currently,
the SES subsystem doesn't support split-expander properly due to low
level library restriction. Some of the known issues:
- The
SEP sees drives in both zones and exposes all of them to the host. A HBA
with dual 4X connection to the box will see 2 SES enclosures (as it
should) but both enclosure are populated with duplicated drives (tested
with LSI 8480E RAID controller).
Note that this is mostly a cosmetic issue since the HBA actually knows
the exact total physical drives in the expander box and you are
provisioned with the correct number of drives for RAID configuration, at
least in LSI management suite.
- Hot-plugging
host to expander box doesn't work reliably.
For dual-port connection, you must connect both cables before powering
up the host.
- Add
more version information to the boot-up flag message and the
'tEnclosureState' command.
For beta release:
- Stamp
of normal version will be in the form of '1.1Nx.nn', such 1.14f.05;
- Stamp
of debug version will be in the form of '1.14x.nn-debug', such as
1.14f.05-debug
For production release:
- Stamp
of normal version will be in the form of '1.1Nx[.nn]', such 1.14F and
1.14F.01;
- Stamp
of debug version will be in the form of '1.14x.nn-debug', such as
1.14F-debug
- Note:
The extra sub-version stamp doesn't apply to SMP and SES response
because of the 4 bytes limitation in specification.
Release note for 1.09G
Date: Oct-17th-2006
Back to Top
Bugs fixed:
1.
With Adaptec SAS 4800/4805 RAID controllers, disk hot plug doesn't work
Symptoms:
- Disk
inserted in any slot that is empty when expander powered up won't be
recognized by the mentioned HBA.
- However,
after removing and then inserting disk in the same slot again, the disk
will be correctly recognized.
Fixed:
- The
fix is tentative though reliable and effective from my surface test; It's
just a replacement of the offensive library (libSMP) with the one copy from
Vitesse original 1.10 FW.
- Using
'stat' console command, you can get the version stamp (01.10) of 'libSMP'
Release note for 1.09D Date: Sep-15th-2006
Back to Top
1.09D FW for VSC-7153x/7154x based SAS JBOD were
put on-line.
- Due
to some critical issues related to fan control/monitoring, there will
be no 1.09D for server models except fan less models for OEM.
- Except
sub-version stamp changed from 'd' to 'D', there are following differences
from the beta 1.09d:
The syntax of
'tZoneCfg' console command was enhanced to support clearing zone configuration
of a single phy. The help message for this command was also corrected.
- For
your convenience, the release note for 1.09d beta is included.
Major Improvement in 1.09D:
- Adding
support for the revised 1U-10 JBOD:
Fully support the
redesigned fan subsystem:
- Total
4 fans instead of 3.
- Each
IOM (I/O module) controls the 2 fans directly connected to it, but every
module can monitor all 4 fans.
- 2
additional 'fan error LED' are supported to alert fan error event; one
LED for each fan set.
- In
a single IOM enclosure, a dummy I/O module mounted with 2 fans must be
installed in the empty I/O module slot and both fans must work normally
otherwise fan fault event will be fired (enclosure system error LED, fan
error LED, enclosure beeping and SES status report).
- Adding
support for dual boot with a single EEPROM chip on revised IOM:
- On
the 4 revised IOM for 1U10 and 2U12 JBOD and server, a new 2-pin jumper
header is added for selecting one of two booting FW image in a single
EEPROM chip (must be in 8 Mbit size):
|
Jumper
|
FW Image Address Space
Primary / Secondary
|
Boot Address
|
|
Open
(default)
|
BFC0:0000
/ BFC8:0000
|
Primary
at BFC00000
|
|
Close
|
BFC8:0000
/ BFC0:0000
|
Secondary
at BFC80000
|
- With
this dual boot capability, we'll have much less danger of rendering an
IOM totally non-bootable due to burning error.
When burning failed, either the Primary image is intact (not yet touched)
or the Secondary image was already updated successfully; we can just
blindly try booting from either image by opening or closing this jumper.
- With
the 'fwupdate -seconly' console command, we can burn different version of
FW onto the Secondary image location and boot it up to test new FW before
official deployment.
- A
lot of improvement in fan control and monitor:
- A
bug in setting fan speed on most server models was found and fixed:
The bug caused under-run of fans and may be the cause of overheat problem
of most of our server models. The fix has been verified OK but currently
all fans are still forced to run in max speed (In previous versions, all
fans run only at about 50% to 70% of its max capacity even we set it to
run at max speed because the base clock was set too slow).
- We
now have more specific PWM parameters based on combination of fan spec,
IOM and backplane revision. With later version of IOM and backplane, we
should now be able to get correct reading of fan RPM. With early buggy
IOM/backplane, we are more confident in determining its running or stop
state.
- Fan
status monitoring and reporting is added back to 3U server with A2
version backplane.
- Some
modification to AIC add-on console command set:
- The
'tShowState' functionality has now been distributed to two new commands
('tShowState' is still kept):
- 'tEnclosureState':
display thermal related states.
- 'tExpanderState':
display expander states (phy related)
- 'tUptime'
cmd is added to tell how long the IOM has been up and running.
* Output of 'tEnclosureState', 'tExpanderState' and 'tShowState' also has
one line reporting uptime information.
- 'tZoneConfig'
cmd is now really functional and its help message revised to better
explain usage of each sub-command.
- Note:
Because of the large scale of modification, we must test each model
intensively and carefully; bug is to be expected.
Release note for 1.07C Date: July-17th-2006 Back to Top
Major Improvement:
- Fixed
the 'PHY error 16 (24 in case of 7154 based expander) ... ' error message
while login Windows with LSI 8408/8480 RAID.
- Fix
the problem of BIOS utility of LSI 3442 and 3800 HBA not able to display
disks attached to our expander or even failed to display SAS topology.
- Added
support for the SCSI 'Report LUNS' command.
- The
added support only replies one Logical Unit (0:0 of the SEP).
- This
should fix the hidden problem of LSI 3442/3800 tooking long time to
inquiry up to 255 non-existent logical units and thus should speed up OS
boot up time.
FW Filename Naming Convention:
- For
7153 based model, the FW filename has this pattern:
eT53C-VER-bpA n_NNd.s3r
- 'T'
: could be 'd' for dual expander daughter card model or 'm'
for single expander module
- 'C'
: denotes the 7153 chip version; 'b' for 7153b, 'c' for
7153-02.
- 'VER'
: denote FW version; such as '1.07B'.
- 'n'
: denotes the backplane revision; such as '0' or '1' or '2'.
- 'NN'
: denotes the supported drive slots; such 10, 12 or 24.
Example:
- ed7153b-1.07B-bpA2_16d.s3r:
for daughter card supporting 16 drive based on 7153B chip.
- em7153c-1.07B-bpA1_12d.s3r:
for module supporting 12 drive based on 7153-02 chip.
- For
7154 based model:
Similar but simplified since all of them support 24 drive.
Release note for 1.06d Date: April-11th
-2006
Back to Top
- Added
support for new models:
- ecm7153c_a1_12d-1.06d.s3r,
ecdc7153c_a1_12d-1.06d.s3r: for 2U 12bay JBOD/Server
- ecm7153c_a1_10d-1.06d.s3r,
ecdc7153c_a1_10d-1.06d.s3r: for 2U 12bay JBOD/Server
- Continuing
improvement to SES support:
- Fixed
some bugs to better comply to SES and SSP transport spec..
- Partially
support of controlling enclosure indicator/alarm via SES:
- Fully
functional on all 7153 based models
- No
action for all 7154 based models due to hardware limitation. This will
be addressed in later version
- Fully
support of controlling access LED (blue) on disk tray for all models
- *
Support for error LED (red) control will be addressed in later version.
Some corrections to last release
announcement:
- The
release version in the subject line was incorrect; it should be 1.06d, not
1.06a.
- For
those models with two version of backplane (most of them, except the new
2U12 and 1U10), the support of SES control on the disk tray access LED only
works with version 2 of backplane; the old backplane may never be
supported since it's too difficult to make it right and reliable. Except
this, I also recommend all upgrading their old backplane for performance
issue and to enjoy planned enhancement to support different LED flash
patterns in different conditions.
As
said, this is a correction to the release information. The firmware remains
intact; you don't have to upgrade again if you had already done.