IBM ServeRAID MR10M SAS Controller Firmware Update Adapters Supported: IBM ServeRAID MR10M SAS Controller (43W4339) NOTE TO SERVICE - Reference RETAIN #NA Version 11.0.1-0042 - Critical ------------------- Enhancements: Adjust various battery parameters and alerts to improve product - Reword the battery ‘Prefail‚ and ‘Fail‚ Alerts and all places it is documented - Add visibility of ‘battery nearing end of usable life‚ condition to CLI and MSM - Adjust parameters used to calculate battery capacity amount for Early Warning and EOL alerts Fixes - Fixed issue where While correcting massive UNC sectors, I/O will be interrupted causing OS becomes unusable - Fixed issue where SPD Data can get Corrupted on Boot causing the following message on boot the message displayed is "Check SDRAM Connection". - Fixed issue where firmware has Incompleted Writes During Power Fail and Recovery causing the following message on boot "Controller cache discarded due to memory/battery problems Version 11.0.1-0040 - Critical ------------------- Fixes: - Fixed an issue that was contributing to early fallout with battery packs. - Fixed an issue that was causing a loss of battery capacity. - Fixed an issue where the gas gauge was reporting capacity of 65000+mAh and then marking battery as SOH bad. - Fixed issue where multiple reboots were required after a new battery was inserted in a controller. - Fixed issue where T95: VT#00560-T95386=Battery cannot support data retention for more than 59 hours. Please replace the battery. When battery logic generates the event MR_EVT_BBU_REPLACEMENT_NEEDED_SOH_NOT_OPTIMAL, the FW prints the error, T95: EVT#00538-T95: 386=Unknown ARG type=24, code=182. - Changed post message from RAID ADAPTER Memory Error !!! Please check the SDRAM connection. If problems persist contact Tech Support. Press Any Key to Continue… to Check the SDRAM connection, if applicable, or replace the controller." Version 11.0.1-0036 - Critical ------------------- Enhancements - BBU AEN enhancements Fixes: - Fixed issue where firwmare fault could occur when a hot spare drive responds with Sense 0x02 for Write/Verify. - Fixed issue where a firwmare fault could occur when consecutive Medium Errors are found in Degraded SATA RAID5, 6 state Version 11.0.1-0033 - Critical ------------------- Fixes: - Fixed issue where SATA hard drive write cache could be enabled on a drive insert or system reboot. Version 11.0.1-0032 - Critical ------------------- Fixes: - Modified memory settings for the ServeRAID MR10k that where not optimal for the controller causing memory failues. Version 11.0.1-0031 - Critical ------------------- Fixes: - Addressed an issue where a physical drive with an excessive amount of medium errors will cause the array to fail Version 11.0.1-0030 - Critical ------------------- Fixes: - Leaving array size field blank during array creation to resolve issue where raid level would change but array size did not change with the level. Retain tip H196863. - Fixed issue where the Cursor movement is not proper while entering Security Key/PassPhrase in EFI WebBIOS environment - Fixed issue where the Visibility of webbios version doing controller selection is not there - Address compatabilty issue with uEFI WebBIOS in the x3850 x5 and HS22 x5 - Fixed issue where a fatal FW Error happens during Consistency Check operation while a patrol read is in progress. Version 11.0.1-0024 - Critical ------------------- NOTE TO SERVICE - H196421, H196426, H196374 Fixes: - Fixed issue with Firmware Hang (Fatal Error) when drive responds with Sense 0x02 for Test Unit Ready Command. - Fixed an issue where the ServeRAID-MR10ie (CIOv) battery fault indicator never clears - Added limit to quantity of consecutive relearn cycles to cache battery unit - Fixed issue where the Pass phrase can not be entered on the ServeRAID MR10is controller - Fixed issue where the user cannot enter the uEFI WebBIOS or system will hang on boot Version 11.0.1-0022 - Critical ------------------- NOTE TO SERVICE - No current tips Fixes: - Fixed an issue where systems that use DL_DOWN instead of (or in addition to) PCI reset exhibit 1800 PCI ERROR. - Fixed an issue where a non-responsive disk caused fatal firmware error - Fixed an issue with write journaling - Fixed an issue with READ XOR logic in RAID1 special operations Functionality: - Added support for dynamically toggling readjustment of PCIe maximum read request size Version 11.0.1-0017 - Suggested ------------------- NOTE TO SERVICE - No current tips Fixes: - Fixed an issue where, on shudown, firmware was not sending a command for the drive to flush its internal cache. - Removed I/O delay during device firmware update for non-disk devices. - Added check for additional error data from mode sense command. Functionality: - Added pass through command support for applications. Version 11.0.1-0014 - Critical ------------------- NOTE TO SERVICE - Reference RETAIN #H194939 Fixes: - Addressed an issue where, during I/O and manual copyback on a degraded RAID 6 virtual drive with medium errors, incorrect data could be created on the drive. - Fixed an issue where, on the ServeRAID 10ie (CIOv), a new or changed battery installation was not detected on the iBBU. - Fixed an issue where SATA drives may not be detected at POST when attached to an expander. (Retain tip H194939) - Addressed an issue where the controller may hang if a medium error occurs during a rebuild. - Fixed an issue where cache data could be lost during a reset after the express link is dropped. - Addressed an issue where a consistency check doesn't regenerate parity after hitting a double media error. - Fixed an issue where the uEFI WebBIOS can hang if the user presses the right mouse button when user input is not required. - Addressed an issue where an invalid memory or battery error is reported on boot, but does not reappear on a subsequent boot. Version 11.0.1-0012 - Suggested ------------------- NOTE TO SERVICE - No current tips Fixes: - Fixed an issue where, rarely, cached data may be lost on system shutdown or reboot. Version 11.0.1-0008 - Critical ------------------- NOTE TO SERVICE - Reference RETAIN #H194224, H194505, H194171 Fixes: - Fixed an issue which could result in incorrect data on the drives after heavy I/O. (Retain tip H194224) - Fixed an issue where, if medium errors are experienced while a background operation is running, incorrect data could be written to the drives. (Retain tip H194505) - Changed the warning message that appears when the card goes in to temporary write through mode. (Retain tip H194171) - Fixed an error where erroneous Bad Battery messages were displayed even after a Learn Cycle was performed. - Fixed an issue that resulted in SERRs on a ServeRAID-MR10M controller. - Improved the handling of bad cable scenarios which could cause system hangs. Functionality: - Added support for UEFI based systems. - Added support for Solid State Drives (SSDs). - Added tape support for the ServeRAID 10ie (CIOv) controller. - Added the ability to handle SGPIO and SES signalling on a per-port basis to handle configurations with both direct attached and expander attached drives on a single controller. - Added a field to show hard drive model names in WebBIOS. Version 9.0.1-0030 - Critical ------------------ Fixes: - Fixed an issue where, if a battery was fully discharged and was reported as 'missing', a reboot was be required before the battery would be detected as 'charging'. - Fixed an issue where a hotspare was reported as unconfirmed good after replacement. - Fixed an issue where an invalid alert 2121 was logged when a virtual drive was created. - Display and log a warning message when a consistency check is started on an uninitialized logical drive. - Fixed an issue where the message "SES enclosure 1 Recovered after fault" would be logged, after which the discovery process would take a very long time. - Fixed an issue where the multipath enclosure IDs were not updated correctly when an ESM or a drive was pulled and reinserted. - Fixed an issue where, if a unconfigured drive was pulled and reinserted, the size of the drive would be reported incorrectly. - Fixed an issue where a dedicated hotspare would become a global hotspare. This could happen after a migration or, in some circumstances, after a system reboot. - Fixed an issue where the firmware could hang when trying to view hotspare information under CTRL-R. - Fixed an issue where a rebuild would not start when a new drive was inserted. - Fixed an issue where a logical drive is still shown as optimal after a drive was pulled. - Fixed a critical firmware error that occurred when a physical drive was pulled from a RAID10 or RAID1E during heavy I/O. - Fixed an issue where a hang could occur when an ESM was removed. - Fixed an issue where, if multiple physical drives are pulled from an enclosure, additional drives might be lost. This could lead to incorrect data on the disks. - Fixed an issue where reconfiguring a virtual disk by adding a drive, or similar activity, will cause a dedicated hotspare to become unassociated. - Fixed an issue where a success code was incorrectly returned when data was sent to a failed drive. This could cause a loss of data. - Fixed an issue where, if additional drives failed in a virtual disk while a rebuild was happening, the system could hang. - Fixed an issue where daisy-chained enclosures would all be reported with the same position number. Also fixed a related issue where 'locate' did not function properly for daisy-chained enclosures. - Fixed an issue where the controller could hang if diagnostics were run while a rebuild was happening. - Fixed an issue where hotspares were lost during configuration import. - Fixed an issue where, if two enclosures were configured and the cable was pulled from each, then reattached in a swapped configuration, the system would hang on boot. - Fixed an issue where successive power cycles in a multi-enclosure configuration could result in offline drives or a system hang. - Fixed an issue where, with the controller mux set to 'Auto', configuration would be lost during reboot. - Fixed an issue where rebuilds might fail to resume after enclosure power is restored. - Corrected several issue where the consistency check was not properly handling double medium errors. - Ensured that BGI (Background Initialization) setting was retained during foreign configuration import. - Fixed an issue which caused MSM to not allow creation of additional virtual disks once 16 virtual disks existed. - Fixed an issue that caused a hang if a rebuild was in progress when a configuration was migrated from an IR to an MR controller. - Fixed an issue where load balancing would not resume properly after a cable pull. - Fixed an issue where, during a target or chip reset, the Disk Write Cache setting on a SATA drive would be reset to default, which is generally 'enabled'. - Fixed an issue where, when BGI is aborted, the status for the CC (consistency check), rather than the status for the BGI, is reported. - Fixed an issue where, if a physical disk fails, causing a logical drive to become degraded, trying to clear or change the configuration would result in a hang until the failed drive was physically removed. - Fixed a hang that was seen when a cable was pulled just before system shutdown. - Ensured that a hotspare would revert back to hotspare state after a rebuild to that hotspare failed. - Fixed several issue that could cause "Illegal Request Sense Data" messages. - Fixed an issue where a rebuild would hang if one of the logical drives on the array was deleted during the rebuild. - Fixed an issue which could result in incorrect data in a scenario where RAID 50 arrays are running in a degraded state for an extended period of time. - Fixed several issues where medium errors could result in offline drives, incorrect data on drives, failed rebuilds, or system hangs. - Fixed an issue where, when a redundant path in a multipath system fails, some drives are marked offline. - Fixed an issue which could cause hangs when replacing a controller with an existing configuration. - Fixed several grammatical and formatting issues. - Fixed an issue where importing foreign configuration under CTRL-R could cause the firmware to crash. - Fixed an issue where an unresponsive SEP could result in a "greater than 122 devices detected" message. - Fixed an issue where the, after a RAID 6 is reconstructed as a RAID 5, the configuration is reported incorrectly upon reboot. - Fixed an issue which could result in incorrect data on the drives when I/O is run while a capacity expansion is in progress. - Fixed an issue that could cause a hang when an enclosure is power cycled while I/O is running. - Fixed an issue where, if a physical drive became invalid during a rebuild to that drive, the rebuild would pause, rather than fail. - Fixed an issue where Battery Warning messages would appear, even when disabled. - Fixed an issue which resulted in "Unexpected sense" messages in the log with SATA drives. - Fixed an issue where Cache Data could be lost during a power failure. - Fixed an issue where not all SATA drives were found during discovery. - Fixed an issue which may result in a firmware hang during a reconstruction operation. - Fixed an issue where the firmware would continue to try to communicate with a bad drive. The symptom of this issue is a drive where the status light is blinking amber and the activity light is solid green. - Fixed an issue where an enclosure reboot could, very rarely, cause a condition where the driver sends a Kill Adapter signal. - Fixed an issue where, if a medium error is encountered on a source drive during a rebuild, the bad block may not get marked as bad. - Fixed an issue where running in degraded mode with medium errors could cause the firmware to hang. - Fixed an issue where running heavy I/O on daisy-chained enclosures for an extended period of time could cause a hang. - Fixed an issue where, with some hardware, the configuration is not found within the timeout time on a cold reboot. This leads to a prompt to import configuration, after which, the configuration is seen correctly. - Fixed an issue with I2C LED support. - Fixed an issue where the write cache policy would unexpectedly change from write back to write through, then would change back after about a minute. - Fixed an issue where, if a cable is pulled in a multipath environment, the alternate path is sometimes not followed. - Fixed an error in the ECC error logging which can cause a memory leakage. - Fixed an issue where the firmware will hang if discovery does not complete in two minutes. - Fixed an issue where full initialization of a large number of SATA drives would cause timeouts. - Fixed several LED issues with I2C backplanes. - Fixed an issue sensing dual I2C backplanes in a multipath environment. Functionality: - Added support for 64 virtual disks per array. - Added support for expanded RAID 1, a form of RAID 1E which supports even numbers of drives. - Added the ability to hot-swap a battery. - Added the ability to manually initiate the battery learn cycle. - Added multipathing support for SATA drives. - Added the ability to enable or disable NCQ from the CLI or WebBIOS. - Added the copyback feature, which is used when a drive goes defunct and the array rebuilds to a hotspare. Copyback allows the hotspare to rebuild to a new drive that is inserted to replace the defunct drive. - Clarified several log event messages. - Implemented load balancing when multipath SAS is used. - Added logging for some events that were not previously logged. - Increased the PHY count from 512 to 640 to support 21 enclosures. Version 8.0.1-0036 - Non-critical ------------------ Fixes: - Address issue where SATA drives attached to an expander will not respond in a specific time period causing the drives to be seen as foreign. - Add support for the IBM ServeRAID 10il SAS/SATA adapter. Version 8.0.1-0033 - Suggested ------------------ Fixes: - Fixed issue where current write policy is not changed to write back even if iBBU completes the charging in 1078 HWR. - Fixed an issue where the firmware reports "MonTask: line 3566 in file ../../raid/1078dma.c" while running reconstruction on a degraded logical drive and injecting media errors. - Fixed an issue where a controller would hang during a reboot. - Fix is done to invalidate the parity cache for the punctured block before releasing the cache line - Fix for WebBIOS hangs with greater then 5 MegaRAID adapters installed. Version 8.0.1-0029 - Critical ------------------ Fixes: - Fixed an issue where, while performing I/O during a RAID level migration or an array capacity expansion, incorrect data may be passed during a data transfer. Version 8.0.1-0027 - Critical ------------------ Fixes: - Implemented changes to ensure the firmware does not violate the DDR2 spec during the initialization sequence in a 1078 ROMB environment. - Fixed an issue where the controller enters MegaMon when a rebuilding span is removed. - Fixed an issue where the operating system may crash after an S3 resume followed by a manual restart. - Fixed an issue where a bad SAS cable, connected between two enclosures, could cause the firmware on the associated controller to crash. - Fixed a chip reset that could happen during boot sequence of RHEL 5 64-bit. - Fixed an issue where WebBIOS does not initially see external devices, and, if RAID is configured, the “missing config” prompt is shown in the BIOS. This was caused by the firmware not waiting for the second discovery attempt that is necessary to find external devices. - Fixed an issue where, on a 1078 chip, firmware was not properly marking a drive dead. Consequently, a write to that drive would appear to complete successfully, but would actually fail, therefore the data could be incorrect. - Fixed an issue where, under heavy load, with write back enabled, the following error was seen: "ERROR: Stale read check failure. Page image returned does not match previous write. Check hardware and caches. Read size was 16384" - Fixed an issue where import of foreign drive configuration would fail. - Fixed an issue where firmware could crash when a large number of hard drives is connected to a ServeRAID MR10M. - Fixed an issue where multiple reboots of cascading enclosures could result in incorrect data. - Fixed an issue where the Global Hard Drive Summary LED was not working. - Disallowed the addition of logical drives in an array that is being rebuilt. This fixes an issue where a newly created logical drive on a rebuilding array would be deleted on reboot. - Fixed an issue where the drive sizes were shown incorrectly after drive migration in an array with two or more different sized drives. - Fixed an issue where the source drive in a rebuild would be failed after several ECC errors. Functionality: - Implemented Slot ID mapping support in a mixed topology. - Implemented a performance enhancement for sequential read workloads. - Reenabled offline hard drive firmware flash. Now hard drive firmware can be flashed in either an online or an offline state. - Implemented a new event so that dirty cache will be discarded on a virtual drive that is missing or offline. Version 8.0.1-0021 ------------------ - Initial release