Patch-ID# 105356-23 Keywords: fc-al mutex ssd bp_resid multinode mhioctkown lun vold.conf timeout Synopsis: SunOS 5.6: /kernel/drv/ssd and /kernel/drv/sd patch Date: Sep/15/2003 Install Requirements: Additional instructions may be listed below Install in Single User Mode Reboot after installation See Special Install Instructions Solaris Release: 2.6 SunOS Release: 5.6 Unbundled Product: Unbundled Release: Xref: Topic: SunOS 5.6: /kernel/drv/ssd and /kernel/drv/sd patch Relevant Architectures: sparc BugId's fixed with this patch: 4015886 4058807 4074325 4079906 4094627 4095563 4100409 4111066 4113228 4114213 4117067 4125140 4128002 4128046 4132160 4148500 4155865 4161206 4170621 4174794 4204491 4212961 4223471 4224495 4235432 4236399 4236646 4240626 4248431 4250964 4251458 4252331 4256591 4271068 4271069 4282525 4289429 4291980 4295457 4297348 4328971 4337798 4356701 4367306 4384992 4397494 4511840 4524801 4530942 4636627 4739785 4765756 4851925 Changes incorporated in this version: 4851925 Patches accumulated and obsoleted by this patch: 105797-07 Patches which conflict with this patch: Patches required with this patch: Obsoleted by: Files included with this patch: /kernel/drv/sd /kernel/drv/ssd /usr/include/sys/scsi/targets/sddef.h /usr/include/sys/scsi/targets/ssddef.h /usr/lib/adb/ssa_disk Problem Description: 4851925 limiting the infinite retries for unit attention (from 105356-22) 4511840 sd error recovery for failed lun on A1000 can reduce 4739785 ssd incorrectly sets the throttle for disks of type ST336605FSUN36G and MAL3364F (from 105356-21) 4765756 I/O to Veritas volume made with A5200 by Oracle thread got stuck in biowait (from 105356-20) 4328971 pointer to a sd_thr_request struct is bogus, called from sd_mhd_resvd_recover 4636627 Failed write operation on a read-only LUN is reporting retryable 4530942 Panic in ssdstrategy due to null kstat pointers (from 105356-19) 4524801 2.6 sd patch has no fix for 4148963, instances stuck in RWAIT. (from 105356-18) 4384992 A3x00 SCSI controller NOT_READY up to 111 secs after a Bus Reset 4397494 Add the following drives to ssd_disk_table (from 105356-17) 4367306 T300 fails reserve cmd occasionally while another node is rebooting 4356701 ssd: not waiting 3 minutes for T300 BUSY status (from 105356-16) 4271069 ssd: Overloaded (Queue Full) throttle 4297348 ssd: ssddump() does not work reliably 4271068 ssd: Eliminate excessive transport rejected messages 4337798 Seagate drives have throttling problem (from 105356-15) 4295457 ssd: serial number should be obtained from the Unit Serial Number page (from 105356-14) 4289429 Sonoma results in bad dump device during dump (from 105356-13) 4236399 A3500 LUNs go offline without warning 4291980 Offlined controller causes "transport rejections" (from 105356-12) 4256591 SSD_DO_KSTATS fix (4224495) needs to be putback into sd driver 4248431 problem with sd trying to recover (from 105356-11) 4132160 ssd/sd driver retries for some non-recoverable errors. 4248431 problem with sd trying to recover 4282525 sd/ssd:Increase retries in dump routine for devices that take longer after (from 105356-10) 4223471 ssd: MUST handle T300 fail-over delay 4240626 TX kstat panic running load on FCAL 3500 (Multi-Lun) 4252331 ssd: queue full adaptive throttle ineffective 4250964 ssd: Should not perform HEROIC error recovery on unmounted T300 4251458 scsi_poll() NOT COMPATIBLE with multi-lun (RAID) devices (from 105356-09) 4236646 ssd doesn't handle bogus scsi status well. 4212961 ssd: add Not Ready Retry count for devices (A3xxx) that take long to come online 4174794 buffer overflow in sd_decode_sense causes panic handling sense errors 4224495 handle qlen == 0 case in ssd while updating kstats (from 105356-08) 4204491 ssd: Add Product String for 18G Seagate FC Drives 4148500 ssd: The device id does not get registered correctly. (from 105356-07) 4155865 ssd: 22 drive photon sees timeouts/transport errors with heavy IOs. 4128002 DR detach fails with disks under Veritas control (layered open > 0) (from 105356-06) 4074325 sd/ssd/soc/pln: DDI_SUSPEND/RESUME incomplete 4125140 ssd: DDI_SUSPEND code to handle outstanding IO will not work for sf (from 105356-05) 4128046 ssd: set pm-hardware-state to needs-suspend-resume (from 105356-04) 4015886 ssd: add support for hotplug event callbacks 4111066 ssd: bp_resid gets incorrect value while recovery for the "Soft Error" cases (from 105356-03) 4113228 Sunfire-DR never suspended pln, ssd (from 105356-02) 4094627 ssd: cannot sleep indefinitely holding the mutex, while waiting for resources 4079906 reset of first node of cluster connected to sonoma hangs other node (from 105356-01) 4058807 timeouts running async i/o test: when 2 ports of the HA connected to single IB (from 105797-07) 4235432 vold accesses cdrom continuously (from 105797-06) 4174794 buffer overflow in sd_decode_sense causes panic handling sense errors 4117067 SS4-110 paniced running sunvts: panic: recursive mutex_enter (from 105797-05) 4170621 sd: dumping broken in solaris 2.6 on Ultra-2 and Seagate ST15230W drives 4128002 DR Detach fails with disks under Veritas control (layered open > 0) (from 105797-04) 4161206 Quorum configured on the shared ccd diskgroup crashes the node. (from 105797-03) 4100409 cluster reconfiguration resulted in controller failover and devices offline 4114213 SCSI transport errors due to `timeouts` degrade system performance (from 105797-02) 4095563 Solaris 2.6 CD-ROM PREVENT bit set to disable eject if no entry in vold.conf (from 105797-01) 4079906 reset of first node of cluster connected to sonoma hangs other node Patch Installation Instructions: -------------------------------- Refer to the Install.info file within the patch for instructions on using the generic 'installpatch' and 'backoutpatch' scripts provided with each patch. Any other special or non-generic installation instructions should be described below. Special Install Instructions: ----------------------------- Reboot the system after patch installation. NOTE: This patch repairs the device identification (devid) algorithm for fibre-channel attached storage products. A devid provides a unique handle for a device which allows the Solaris kernel to identify them independently of the datapath used to access them and provides this handle to applications. This patch affects the use of T300 storage devices. T300 is a RAID controller that prior to this patch would report the same devid for all LUNs created behind the RAID controller. If you are using version 2.3 of the Alternate Pathing (AP) product, this patch prevents the loss of an existing AP 2.3 configuration with T300 devices and allows a successful upgrade to AP 2.3.1. If you are installing AP 2.3.1 this patch will allow creation of alternate paths to T300 devices. Sun has identified Sun Cluster 2.x products to require assistance in upgrading to this patch and suggests that you contact your SE in this case. Products that have been identified as NOT affected by this changes are all Veritas VxVM products, all Sun StorEdge RAID manager products and all Solstice Disk Suite products. README -- Last modified date: Monday, September 15, 2003