|Skip Navigation Links|
|Exit Print View|
|Oracle VM Server for SPARC 2.2 Release Notes Oracle VM Server for SPARC|
The following RFEs and bugs have been fixed for the Oracle VM Server for SPARC 2.2 software release:
ldmd shows MAUs warning when starting up on system which does not have any MAUs
Domain migration needs to handle multipathed disks correctly
ldm should allow cpu pinning by id
Memory DR should support > 31 memory segments configuration
RFE: support named cores via CLI for inactive domains
RFE: Disable CPU defragmentation in LDoms
Batch requests and provide both normalized and non-normalized cpu utilization to LDoms PM from HV
RFE: Logical Domains Manager SR-IOV support
RFE: Add explicit CLI to set the whole core hard cap
Implement affinity_reserve() and memory_allocate() to decouple memory routines to allow for Affinity
Whole-core constraint lost after powercycle (no HV MD property)
PM test harness has illicit knowledge of pm_cookie
RFE: Pwrcap enhancements for SPARC PAD
RFE: Add PAP SAT support to gather PPFE-PPSE pm state info for M4/T5 platforms
Support for low-CPI workloads (aka dynamic threading)
Binding a domain should use new cores if any are free; only share cores where that's not possible
Remove check of VID=0.
RFE: Link L1 memory pm support for M4/T5 platforms
Implement affinity_pm to support CPU defragmentation
Common routines for Frag and HVFrag
Remap around partial cores other than the lowest core for control domain to switch to whole-core
PRI domain service is lost after migrating back to a previous source machine
Add hvfrag support for new, per-IOS allocation class
Vds_Validation: ldmd failed to bind -f/-q guest using xml file with invalid vdsdev
avg_guest_util() and avg_cpu_util() confuse ns and ms
ldmp2v_convert should notice vntsd is in maintenance mode.
ldmconfig attempts to configure mau on T4 platform
PM does not manage vcpus in Elastic mode after a guest starts/reboots
MEMDR set-mem 8G primary fails under some memory configurations
get_state_md_template() leaks filename string
RFE: Support for migration between different CPU types and speeds
set-vsw second time core dump.
ldmd core dumps when attempting a migration when running on downrev firmware
RFE: LDoms MIB - support for the new properties such as extended-mapin-space and low CPI
prtdiag shows dual cpu speed after a few migration and cpudr operations
Incorrect return status on a failed CPUDR operation on a domain booted in single user mode.
Vdsdev device validation messages should be improved
Block 'ldm add-config' while 'ldm debug track' is in progress
Debug Logical Domains Manager hits an assertion failure when retrying a previously interrupted migration.
Primary switches to whole-core, but it does not have all vcpus of a core bound
HV mblock is modified at both ends
ldm rm-mem --auto-adj 256M primary failed on SPARC T3-1B
ldm migrate triggers free(1c5978): invalid or corrupted buffer
ldm add-spconfig -r <config> does not persist <config> over ldmd restarts
Sun Management Console operation dumps core in debug mode
Memory leaks found by parfait and ::findleaks
ldom_num_faulted_vcpus() hurts performance
With incoming_migration_enabled=false, outgoing migrations fail with version incompatibility message
LDoms MIB: memory leak in get_ldom_names
ldomCryptoResourcePool: ldomCryptoRpReserved property value is mismatch
ldmd needs an option to disable device validation
ldm add-vdisk/vnet/vcc/vsw/vds/vdpcs can result in devaliases longer than OBP supports
ds_snmp: WARNING: Received DS snmp data out of sequence with request seen on T4
ldoms-dlmd: delayed reconfiguration produces a warning in the smf log
ldmd should not include 'power-management' node for guest md
Sun4v PRI needs a property to indicate max number of hv_mblock nodes
CLI: executing 'ldm add-domain with options' but no ldm name returns no syntax error
After installing redistributable svc:/ldoms/ldmd:default is in maintenance state on ldom
ldmd crashes on removing DRM policy from the primary in delayed reconfiguration mode.
install-ldm hangs when run via absolute path from other directories
Need the lowest numbered physical strand to stay in primary domain as workaround for 7043292
Python test failures on T4 systems with no MAUs
dr_vio.py fails occasionally in _testintervnetlink()
ldm man pages says 'stop' issues 'shutdown', which it does not
The hotplug command failed to disable/poweroff a PEM (PCI ExpressModule)
ldmd spews warnings "No delayed reconfig in progress" to ldmd log
mig_conn_t leaked in mig_listen_on_port()
ldmd doesn't need to listen on legacy migration port
ldmd should send REM_RESOURCE event to PM post cpu unbind
Error message for Mapin table memory allocation failure should be improved.
Auditing changes breaks Logical Domains Manager
RFE: ldm create should have an option to specify the domain UUID
Config changes induced by memory DR do not result in the current config switching to [next-poweron].
Double free in be_hv.c
RFE: Logical Domains Manager should dynamically figure out HV MD sizes and Guest MD sizes
ldm list sometimes prints 10.0% instead of 10%
basicSuite.py: "warning: Unknown soft state 0 for LDom ldg1" messages
WARNING: lgrp_minlat_node: malformed MD, no CPUs found in latency group
XMLv3 policy output doesn't show "disabled" in the elastic mode for the domain without whole-core
ldmp2v should handle missing s2 partition
CLI for add/set/rm-core to work differently from add/set/rm-vcpu -c
cancel-reconf restore the removed vdsdev
ldmd drops core after "set-vsw inter-vnet-link=off"
Domain migration checks for available console ports only on the first vcc on the target control domain
P2V: Add support allowing the transfer of a system which has zones
sparse-core strand allocations/deallocations are seen with bits that lock-down strand 0.
In PM elastic mode, memory leak found in cpu_del()
RFE:Memory allocation should support pinning specific memory ranges
Constant CPU DR of whole cores & threading changes causes ldmd to dump core
Use "add or remove" for whole-core error messages.
Implement SSC CLIs per Functional Spec
RFE: refactor domain suspend/resume to be less migration-centric
ldmd dumps core in xmpp_close_conn()
Add manual pkg(5) dependency on system/ldoms
ldm should not allow removal of PCIe bus after a PCIe device on that bus has been removed
Problems with list-io after removing a bus
RFE: XML support for SSC CLIs
Bind of I/O domain fails, unexpected devices in list-constraints
Remove ldmconfig(1M) from ldomsmanager pkg(5) manifest
Cannot bind ldom even there are console port available in second vccs
RFE: Relax whole-core switching restrictions on bound or active domains when no cap is specified
Allow only one instance of each ldm SVR4 package be installed at a time
ldm set-vsw does not update VLAN params in database
Number of vcpus in the DB is out of sync with Logical Domains Manager after max-cores is decreased
RFE: ldmd should send an XML event when a domain is suspended/resumed during a migration
add-vdsdev with null entry for mpgroup option still creates mpgroup
XMLv3 list-bindings/list-constraints : Assertion failed: is_uuid_valid(ldp->props.uuid)
Primary mblock addresses laid out in some sequences cause endless system watchdog loops.
Domains with physical memory bindings encounter corruption during migration
Documentation should reflect that "set-vcpu -c" should not longer be used
ldmd loses UUID for control domain
Restrict ldmd -i option to debug mode
Add required dependencies to ldmd SMF service
set-vsw doesn't update the MTU
ldm rm-vdpcc doesn't update LDoms database
Inactive domains can not be bound after applying patch 147507-01/-02
Cancelling delayed reconfiguration does not restore extended-mapin-space value.
XML vdisk section should include backend information
XML should have a custom section
Adding a very large number of mblocks to a domain causes ldmd SEGV crash and restart.
ldm add-domain -i can fail but domain is created anyway
Bump up the XML Interface version number from 1.2 to 1.3
init-system prints the warning/failure message only from the first command processed
Canceling delayed reconfigurations on systems not in del reconf does not return any messages.
Primary panics on repeated delayed reconfig set/cancel of extended-mapin-space parameter.
init-system does not restore named CPU cores when threading is max-ipc.
Resource ID pools need better sparse ID allocation algorithm
Disable use of 1/2 DR from PM code
Tune the PM knobs for better responsiveness to load
Last crypto unit is auto removed without -f option when last CPUs are DR'ed out via retries.
ldmd dumps core (do_hv_pm_aync) on systems running S10 9/10
Need to document impact of vdisk timeout on mpgroup feature
Cross-CPU migration of guests with MAUs to platforms without MAUs fails
Cannot create anymore a Virt. Switch with the java API, XMPP driver refuses the operation