Engineering Release Notice |
Release Date: | 12-10-2009 |
OEM: | ALL_OEM |
SWR_Driver_R5_Univ | 13.14.1209.2009 |
Component: | SWR_Driver_R5_Univ |
Stream: | SWR_DRIVER_R5_UNIV_ATHENS_Integration |
Version: | 13.14.1209.2009 |
Baseline From: | SWR_DRIVER_R5_UNIV_ATHENS_10_21_2009_WINDOWS_RELEASE.8938 |
Baseline To: | SWR_DRIVER_R5_UNIV_ATHENS_12_10_2009_LINUX_RELEASE.7783 |
LSID100102202 | (TASK) | Task to make Linux release |
LSID100152546 | (TASK) | PCI Id's missing in MegaSR.inf |
LSID100152545 | (TASK) | Windows 2008 x64with event ID 129 |
LSID100132741 | (DFCT) | Windows 2008 x64with event ID 129 |
LSID100152504 | (DFCT) | PCI Id's missing in MegaSR.inf |
LSID100152557 | (CO) | Auto Generated from PR LSIP200041233 for component_SWR_Driver_R5_Univ |
CO ID: | LSID100152557 |
Headline: | Auto Generated from PR LSIP200041233 for component_SWR_Driver_R5_Univ |
Description: | LSIP200041233 Headline: Add PCI ids for Mercury ICH9R and ICH10 systems |
State: | Implement |
Associated Task: | LSID100152546 |
DFCT ID: | LSID100132741 |
Headline: | Windows 2008 x64with event ID 129 |
Description: | MEgaSR with RAID 10 (6 drives) config will create Event ID 129 under Windows 2008 x64 |
Version of Bug Reported: | 13.10.0708.2009 |
Version of Bug Fixed: | 13.14.1209.2009 |
Steps to Reproduce: | Please see attachment |
Resolution: | Fixed |
Resolution Description: | Root Cause: When there is a RAID10 IO spanning all the drives (64K*3(spans)) fails because of resource crunch in the driver the
internal data structures are not intialized correctly for retring that command. Especially the span count is not reset before going for retry, which is cuasing the span count to become more than 3 in this case!!! Fix: Reset the span count to zero before retring he command. |
Customer List: | LSI -- LSI |
Fix Impact: | Medium |
Suggested Testing: | Test: Follow the recreation steps, shouldn't see the issue. |
Child Tasks: | LSID100152545 |
DFCT ID: | LSID100152504 |
Headline: | PCI Id's missing in MegaSR.inf |
Description: | Customer reports they can't install Windows OS with their IDs, could you please help to look into this problem? ________________________________________ SVID SSID 1064E 1462 6C80 1068E 1462 6C80 id = "PCI\VEN_1000&DEV_0057&SUBSYS_6C801462", "megasr" id = "PCI\VEN_1000&DEV_0059&SUBSYS_6C801462", "megasr" (13.13.1021.2009_Windows_SWR_Driver_R5_Univ.zip) Reason: their IDs has been included in "txtsetup.oem", but "MegaSR.inf" don't. |
Version of Bug Reported: | 13.12.1009.2009 |
Version of Bug Fixed: | 13.14.1209.2009 |
Steps to Reproduce: | Try installing any windows driver using these id's. |
Resolution: | Fixed |
Resolution Description: | Added missing id |
Customer List: | APAC -- APAC |
Fix Impact: | Low |
Suggested Testing: | should load to the customer system now. |
Child Tasks: | LSID100152546 |
Task ID: | LSID100102202 |
Headline: | Task to make Linux release |
Description: | Task to make Linux release |
State: | Open |
Change Set Files: | 0 |
References: |
Task ID: | LSID100152546 |
Headline: | PCI Id's missing in MegaSR.inf |
Description: | PCI Id's missing in MegaSR.inf |
State: | Completed |
Change Set Files: | 0 |
References: | LSID100152557(CO) LSID100152504(DFCT) |
Task ID: | LSID100152545 |
Headline: | Windows 2008 x64with event ID 129 |
Description: | Windows 2008 x64with event ID 129 |
State: | Open |
Change Set Files: | 0 |
References: | LSID100132741(DFCT) |
Heading: | Known Restrictions |
Contents: | • Type "brokenmodules=ahci" line at kernel parameter line during SLES 9 & 10 installations • RHEL 3.0 64-bit versions are not supported on AMD systems. Only Intel systems are supported. • RAID 5 is not supported on XEN. • In SWR5 driver, RAID 5 feature is enabled only if iButton is present in the system. BIOS communicates the presence/absence of iButton using configuration data on disk (DDF) and through device registers from BIOS version 06.10111923. So if IButton is present then any driver version later than 10.02.1001.2007 with BIOS version later than 06.10111923 should be able to create a R5 logical drive using an operating system application. The following limitation is applicable for any driver (version less than 10.02.1001.2007) and BIOS (version less than 06.10111923) combination:- If driver does not read a valid DDF, it will disable R5 creation. Driver may not see valid DDF if (A) user does not configure using BIOS Ctrl+M; or (B) if no drives were present at the boot time and all the drives were hot-inserted after OS is booted. In both cases, user must create initial configuration using BIOS Ctrl+M utility. • A RAID 5 logical drive withstands single drive failure by reconstructing missing data from the data and parity from the remaining drives. The assumption underlying this ability is that the data in the remaining drives is consistent. This assumption becomes invalid if there is a power loss during write operation(s). If there is a loss of power during a write operation, data inconsistencies may occur because data and parity writes may have completed only partially. After subsequent reboot the logical drive has inconsistent data on that particular stripe. If a drive were to fail at this point, missing data (for the inconsistent stripe alone) cannot be reconstructed from remaining drives. To correct inconsistency, drivers starts back ground initialization (BGI) at reboot. While BGI is in progress, the inconsistent stripe(s) are vulnerable to drive failure. One way to avoid this window of vulnerability is to run check consistency on the logical drive using BIOS utility. Another way is to prevent an unplanned shutdown (e.g. using UPS). • The LITE-ON SH S752K model CD/DVD ROM device does not respond to 0x55 & 0x51 MMC commands properly. The device hangs when these commands are issued multiple times. The driver resets the device when it hangs; but reset adversely affects DVD burning. This behavior can be consistently observed using K3B DVD burning software. K3B aborts DVD writes when 0x55 is failed and device is reset. • RHEL5 installation does not automatically prefer the driver in DUD to native driver. Run replace_ahci script inside the DUD before rebooting during the last stage of installation. Refer to readme.txt under in the DUD for more details • RAID 5, by its very nature, cannot tolerate double failures. The double failure can be a combination of a single drive failure and either; more drive(s) failure or an unplanned power failure, in the RAID system. Unplanned power failure can be including but not limited to, loss of power source or an unusual system shutdown. Unplanned power failures while there are medium errors on the drives is also considered as a double failure condition. • No Linux hibernation Support. Multiple drive failure would result in a data loss condition. A single drive failure combined with any additional unplanned failure can cause loss of data integrity. To avoid loss of data integrity for the power failure case, it is highly recommended to provide an uninterrupted power supply to the server to protect the RAID 5 subsystem during the event of a degraded or rebuilding state. Medium errors on the disks, combined with an unplanned power failure can also cause loss of data integrity, but this occurrence would be very unlikely. |