Patch Name: PHNE_29267 Patch Description: s700_800 11.04 (VVOS) LAN product cumulative patch Creation Date: 03/06/05 Post Date: 03/06/26 Hardware Platforms - OS Releases: s700: 11.04 s800: 11.04 Products: N/A Filesets: Networking.LAN2-KRN,fr=B.11.04,fa=HP-UX_B.11.04_64,v=HP Networking.LAN2-KRN,fr=B.11.04,fa=HP-UX_B.11.04_32,v=HP Networking.LAN-RUN,fr=B.11.04,fa=HP-UX_B.11.04_32/64,v=HP Networking.NW-ENG-A-MAN,fr=B.11.04,fa=HP-UX_B.11.04_32/64,v=HP Automatic Reboot?: Yes Status: General Release Critical: Yes PHNE_29267: PANIC ABORT Based on HP-UX Patch PHNE_28143: PANIC ABORT Symptom 2, 3, 9 fix panic. Symptom 8 fixes abort. Based on HP-UX Patch PHNE_26250: PANIC MEMORY_LEAK Symptom 1, 3, 5, 8, 25 fix panic. Symptom 16, 22 fix memory leak. PHNE_25076: PANIC HANG Based on HP-UX Patch PHNE_24413: PANIC Symptom 2 and 4 fix panics. Based on HP-UX Patch PHNE_22962: PANIC HANG Symptoms 4 Fixes a panic, 6 Fixes a hang. Based on HP-UX Patch PHNE_22353: PANIC HANG Symptoms 3, 5, 6, 7, 8 fix panics. Symptom 9 fixes a hang. PHNE_22825: OTHER PANIC CORRUPTION MEMORY_LEAK HANG Based on HP-UX Patch PHNE_21905: PANIC HANG MEMORY_LEAK CORRUPTION Symptom 2 fixes a hang due to memory leak. Symptom 7 fixes a panic due to data corruption. Based on HP-UX Patch PHNE_21217: PANIC HANG MEMORY_LEAK Symptom 4 fixes a panic. Symptom 11 fixes hang & memory leak. Based on HP-UX Patch PHNE_20657: PANIC Symptom 6 fixes a panic. Based on HP-UX Patch PHNE_19766: PANIC HANG OTHER Without this patch, packets which are sent out by an interface will not be seen by the promiscuous mode application. Symptom 2 can cause system hang or panic. Symptom 4 can cause system panic. Based on HP-UX Patch PHNE_19754: HANG Type 2 DLPI applications on ethernet links will not work without this patch. Based on HP-UX Patch PHNE_19081: PANIC PHNE_18560: ABORT OTHER PANIC CORRUPTION MEMORY_LEAK HANG Based on HP-UX Patch PHNE_18409: PANIC Based on HP-UX Patch PHNE_17113: PANIC HANG OTHER Auto-port Aggregation product depends on this patch. Symptoms 4, 8 and 13 can cause system panic. Symptom 9 causes system hang. Based on HP-UX Patch PHNE_16990: PANIC HANG Symptoms 1 and 3 are system hangs. Symptoms 2, 4, 5, 6 and 7 are system panics. Based on HP-UX Patch PHNE_15969: OTHER HP Auto-Port Aggregation Software will not work without this patch. Based on HP-UX Patch PHNE_15591: HANG OTHER Symptom 2 is system hang. With symptom 3, SNA is not able to configure Token Ring interfaces with lanadmin. Based on HP-UX Patch PHNE_14370: PANIC OTHER Symptom 1 fixes a Panic. Without fix 4, DLPI type 2 applications (e.g. SNA) won't work over non-CKO FDDI. Without fix 5, ip multicast won't work for Token Ring. Without fix 8, "lanadmin" becomes unusable on large lan configurations. Based on HP-UX Patch PHNE_14106: PANIC MEMORY_LEAK ABORT CORRUPTION Symptom 4 is corruption with in the process, but it does not affect the system as such. Symptom 3 and 6 are ABORT symptoms. Symptom 9 and 13 cause system panic. Based on HP-UX Patch PHNE_14027: OTHER This patch is being made critical since all the patches released in the 11.0 SR have to be critical. Category Tags: defect_repair enhancement general_release critical panic halts_system corruption memory_leak Path Name: /hp-ux_patches/s700_800/11.X/PHNE_29267 Symptoms: PHNE_29267: Ported HP-UX patch PHNE_28143 to VVOS Based on HP-UX patch PHNE_28143 : 1. CR: JAGae43281 SR: 8606279225 Protocol value 0x1010 unavailable for user applications. 2. CR: JAGae23942 SR: 8606259624 In a DLPI application, if DL_PROMISCON_REQ primitive is issued when it is still unbound and it is followed by DL_BIND_REQ followed by DL_PROMISCON_REQ again, system panics after the application has completed and has exit. Stack trace : panic+0x14 report_trap_or_int_and_panic+0x4c interrupt+0x1e8 $ihndlr_rtn+0x0 hp_dlpi_intr+0x24 lan2_process_packet+0x690 lan2_int_fr_rnr+0x19c lan2_isr+0x14c lasi_interrupt+0x5c mp_ext_interrupt+0x300 ivti_patch_to_nop3+0x0 idle+0xc24 swidle_exit+0x0 3. CR: JAGae22012 SR: 8606257709 For a connection oriented link over LLC2, system panics when a packet that does not contain data is received. Following is the stack trace : panic+0x14 report_trap_or_int_and_panic+0x80 interrupt+0x1d4 ihandler+0x928 t2_dlpi_data_ind+0x13c t2_dlpi_event+0x1370 t2_event+0x2678 t2_mblk_ics_put+0x104 str_spu_sw_isr+0x17c ihandler+0x904 hpstreams_select_int+0xf0 hpstreams_select+0x34 spubind_cdev_select+0x84 spec_select+0x9c vno_select+0x2c select+0xec syscall+0x480 syscallinit+0x54c 4. CR: JAGae23565 SR: 8606259247 Since DLPI patch PHNE_24413 , all frames read through DLPI coming from the FDDI cards are reported incorrectly to be group (multicast/broadcast) frames though they could be unicast. 5. CR: JAGae42209 SR: 8606278150 Incorrect DLPI debug information observed through debug tools. 6. CR: JAGae58428 SR: 8606294733 DL_HP_MULTICAST_LIST_REQ does not return IP multicast list for token ring. 7. CR: JAGae57955 SR: 8606294216 Improper padding of < 64 byte packets by NIO 10 BT driver. 8. CR: JAGae58591 SR: 8606294894 Possible core dump by lanadmin. 9. CR: JAGae65127 SR: 8606301682 Panic while changing MAC address for EISA 100BT. Based on HP-UX patch PHNE_26250 : 1. CR: JAGad92200 SR:8606223097 When system is short of memory, system will panic with data page fault after sending ioctl to the driver. Following is stack trace. panic+0x6c report_trap_or_int_and_panic+0x94 trap+0x11ec nokgdb+0x8 putq+0x40 hp_dlpi_process_wakeup+0x220 hp_dlpi_wakeup_put+0x2c streams_put+0x1d8 hp_dlpi_ilan_wakeup+0x2bc igelan_ilan_callback+0x9c0 igelan_ctl+0x458 igelan_ilan_hw_req+0x374 igelan_ilan_media_control+0xab0 igelan_ilan_ioctl+0x25c hp_dlpi_control+0x8cc hp_dlpi_wput+0x7c0 putnext+0x228 ip_wput_ctl+0x1d8 ip_wput_nondata+0x7c4 ip_wput+0x6c puthere+0x238 ar_entry_squery+0x244 ar_cmd_dispatch+0x104 ar_rput+0x68 csq_turnover_with_lock+0xa8 str_sched_mp_daemon+0x18c str_sched_daemon+0x304 im_mpnetstr+0x2c DoCalllist+0xc0 main+0x28 $vstart+0x48 istackatbase+0x84 2. CR:JAGad94736 SR:8606225662 HP-UX only allow one unbound promiscuous stream per interface, when trying to enable promiscuous mode in another stream if one is already existed, an EINVAL value will be returned instead of EBUSY. 3. CR:JAGad95328 SR:8606226258 When system is short of memory, system will panic with data page fault if one tries to join a multicast group. Following is the stack trace. panic+0x60 report_trap_or_int_and_panic+0x5c trap+0xf58 $call_trap+0x30 hp_dlpi_control+0x8d8 hp_dlpi_wput+0xf88 putnext+0xc4 ip_wput_ctl+0x78 ip_wput_nondata+0x5c8 ip_wput+0x120 puthere+0xc4 ar_entry_squery+0x12c ar_cmd_dispatch+0x9c ar_rput+0x5c puthere+0xc4 ipc_opt_set+0x7ac optcom_req+0x310 ipc_optmgmt_req+0x64 ip_wput_nondata+0x4b8 ip_wput+0x120 putnext+0xc4 udp_wput_other+0x294 udp_wput+0x434 putnext+0xc4 str_async_ioctl+0x1ec hpstreams_ioctl_int+0xd8 streams_ioctl+0x34 sosetopt+0x564 setsockopt+0xe4 syscall+0x6ec $syscallrtn+0x0 4. CR: JAGae08292 SR: 8606239879 Application which try to send data through rawdata mode stream on gelan will not work for 11.00 release. 5. CR:JAGad96423 SR:8606227362 If multiple bind and unbind to the same interface with same protocol run simutaneously, system may panic with following stack panic+0x6c report_trap_or_int_and_panic+0x94 trap+0xed4 nokgdb+0x8 lanc_remove_protocol+0xf8 hp_dlpi_ioctl+0x290 t2_unlog+0x174 hp_dlpi_unbind+0x52c hp_dlpi_wput+0xa38 putnext+0xcc putmsg_subr+0x164 putmsg+0x1bc syscall+0x750 $syscallrtn+0x0 6. CR:JAGae00121 SR:8606230883 Applications using DLPI type 2 services will fail to send out data through gigabit interface. 7. CR:JAGae00493 SR:8606231255 Possible raw data packet corruption over gelan driver. 8. CR:JAGae01356 SR:8606232120 If too many multicast addresses are enabled on one interface, getting multicast address request may return invalid value, and system may panic. 9. CR:JAGae04997 SR:8606235861 Promiscuous mode stream may receive corrupted packets. 10.CR:JAGad92458 SR:8606223361 DLPI application may see inbound packets out of order. 11.CR:JAGae05586 SR:8606236529 ER: Support igelan on 11.0 HPUX 12.CR: JAGad62560 SR: 8606193348 Interface which has multicast addresses enabled through socket will prevent APA from configuring it as primary port in LAN Monitor mode. 13.CR: JAGae08755 SR: 8606241497 Outbound packet will not be loopbacked correctly if promiscuous mode is enabled on that interface, which lead to some unexpected number of droped packet in DLPI. 14.CR: JAGae08757 SR: 8606241499 Interface which has unbound promiscuous mode enabled can be aggregated, this is not expected behaviour. 15 CR: JAGae11948 SR: 8606245478 When system is short of memory, if trying to send packet DLPI may return DL_BADADDR instead of ENOBUFS. (This apply to ARP stream also. For IP, if fastpath negotiation happened in this situation, the failed value will be DL_BADADDR instead of ENOBUFS. For ARP/IP the returned value is not visible to user) 16 CR:JAGad92395 SR:8606223292 Possible Memory leak when unbound promisc is enabled, disabled and then reenabled. 17 CR:JAGad85055 SR:8606215881 Ifconfig shows interface is up, while lanadmin may show operation status as down following a MAC address change request. 18 CR:JAGad75002 SR:8606205827 When an application (that has requested for an IP Multicast address to be listened on Token ring interface) is killed the multicast address is deleted from the listen list and when this is followed by an unplumb of the interface, DLPI returns EINVAL for deletion of the multicast Address. 19 CR:JAGad90159 SR:8606221023 When reset statistics request is received the DLPI MIB stat for discard type is nonzero. 20 CR:JAGad90136 SR:8606221000 DLPI always returns error when drivers sends a link down event. This return value is not processed by any driver as of now hence the user impact is minimal. 21 CR:JAGad87946 SR:8606218798 Incorrect Lanadmin output is seen for statistics after a reset statistics request on a APA aggregate. 22 CR:JAGad83398 SR:8606214207 Potential memory leak when DLPI application/IP stream with multicast address enabled is detached from the interface. 23 CR:JAGad71218 SR:8606202044 On a aggregate going down or coming up only the first stream bound to the aggregate recieves the link down/up notification. 24 CR:JAGad66889 SR:8606197697 Lanadmin does not report errors when setting MTU size for Gigabit Ethernet interfaces. 25 CR:JAGae07043 SR:8606238001 System panic with data Page Fault in hp_dlpi_get_mibstats(). stack trace for event 0 crash event was a panic panic+0x14 report_trap_or_int_and_panic+0x84 trap+0xd9c nokgdb+0x8 hp_dlpi_get_mibstats+0x24 hp_dlpi_wput+0xac4 putnext+0xcc putmsg_subr+0x188 putmsg+0x184 syscall+0x6f8 $syscallrtn+0x0 PHNE_25076: Ported HP-UX patch PHNE_24413 to VVOS Based on HP-UX patch PHNE_24413: 1. DLPI drops non IP multicast packets for EISA Token ring. 2. Panic in rif_putq(). 3. The fddi links dropped during blast test. 4. Spinlock/Deadlock panic for lanc_lookup_lock in lanc_remove_protocol. 5. IP multicast rcved on FDDI is broadcast forwarded back to FDDI. Based on HP-UX patch PHNE_22962: 1. DLPI does not keep track of dropped packets. 2. DLPI requeues packets in raw mode. This is not neccessary. 3. DLPI test failed for token ring interface. 4. NIO Token ring driver reports "Hard error on CTL_CCMD=0x1060" in syslog file. 5. DLPI reuses mblks without adjusting the size. 6. On recieving a badly formatted trailer protocol frame system crashes. Based on HP-UX patch PHNE_22353: 1. When the PPA number given as a lanadmin argument is wrong, lanadmin displays the information about the interface with PPA number 0. 2. When the Station Address value given as an argument to lanadmin was too large, lanadmin was not returning non-zero value indicating the failure to set the Station address. 3. The system panics when trying to run the promiscuous tests on Core Ethernet interface. 4. When one or more data lines(DI-A,DI-B, DO-A, or DO-B) of the AUI port had been severed but the other lines are still connected, LASI based systems will panic after network activity. 5. System panics with "Data Page Fault". 6. System may panic when user kills a promiscuous application. 7. Turning off the promiscuous mode panics the system. 8. System panics as it runs out of Kernel virtual space. The following panic message is displayed. "panic: Out of kernel virtual space". 9. Broadcast storm on Gigabit ports hangs system. PHNE_22825: Ported HP-UX patch PHNE_22331 to VVOS Based on HP-UX patch PHNE_22331: 1. lanscan usage string does not contain option "q". 2. ER: Need a command line option to display information which is available from lanadmin in interactive mode. 3. Incorrect reference in "SEE ALSO" section of lanadmin man page. 4. Usage strings for x/X options for lanadmin incorrect. 5. /sbin/init.d/hpether script fails if more then 10 cards are present on the system. Based on HP-UX patch PHNE_21905: 1. Data loss over serial line with unterminated Core 10BT card. 2. Under low memory conditions, HP-PB 10BT can cause system hang. 3. HP-PB 10BT reports incorrect MIB statistics causing Service Guard to misbehave. 4. Lanadmin manpage information for "-m" option is incorrect. 5. Link between landiag and lanadmin is broken in patches. 6. ER: lanscan -q is broken for NIO FDDI Fail-Over groups. 7. System panic in t2_kftslis(). 8. DLPI returns unexpected error acknowledgement. Based on HP-UX patch PHNE_21217: 1. T2 XID/TEST tests fail. 2. DLPI does not check for invalid multicast address for FDDI. 3. lanc fails to prevent setting Station Address to a multicast MAC address. 4. System panic during disable multicast request. 5. IP multicasting problem in Token Ring. 6. ER: lanc support for filtering IEEESAP_HP packets on 10.20. 7. APA_EVENT_NON_IP_LOGGED not correctly implemented for TR 8. lanadmin command fails on machines with many interfaces. 9. Q4 debug information wrong in 11.00. 10. lanadmin -A causes system panic. Based on HP-UX patch PHNE_20657: 1. ER: Support for APA in DLPI. 2. Unable to change PMTU when interface is configured as SNAP 3. Problem with DL_SUBS_BIND_REQ. 4. DLPI needs to send link up/link down messages to upper layers when MAC address changes. 5. ER: Source MAC address should be filled by DLPI on every outbound path. 6. Panic with lanc_ether_trail_mblk_intr on top of stack. 7. ER: Lan Monitor support in DLPI. 8. DLPI does not propogate LINK UP/DOWN events to all attached IP streams. 9. OTS fails to switch interfaces when a MC/Service Guard failover happens. Based on HP-UX patch PHNE_19766: 1. HP-PB 10BT and Core 10BT drivers do not receive packets they transmit, when in promiscuous mode. 2. I/O slowdown or hang because of HP-PB 10BT absorbing a large number of IOVRs. 3. Ethernet cards setup incomplete after reboot. 4. HP-PB 10BT panic during reboot. 5. Service Guard test fails on HP-PB 10BT interface. 6. 'ifconfig' on HP-PB token ring card failed. Based on HP-UX patch PHNE_19754: 1. The connection oriented DLPI applications may hang on an Ethernet link. Based on HP-UX patch PHNE_19081: 1. On a token ring LAN, an ARP packet with incorrect header fields may cause a panic. 2. The network interface status is not updated when the cable is diconnected. 3. DLPI applications failed to send data using DL_UNITDATA_REQ primitive over ethernet links which support checksum offload feature. 4. Functional addresses are not working for PCI token ring. 5. The system may panic if a TEST/XID packet is received. 6. Auto-Port aggregation product generates redundant console messages. 7. When the SNA protocol stack is in uniprocessor emulation mode, an inbound packet to one of its streams that is closing could potentially panic the system. 8. When a Core-Ethernet NIC is neither connected to a network nor terminated, then real-time applications could see performance degradation every 5 seconds. PHNE_18560: Ported HP-UX patch PHNE_18409 to VVOS Based on HP-UX patch PHNE_18409: 1. A system configured with a Core-Ethernet NIC sometimes panics with a data page fault or with the message "panic: mclfree". 2. A system panic with the message "SPINLOCK DEADLOCK" in the HP-PB Ethernet driver. Based on HP-UX patch PHNE_17113: 1. Inability to install the Auto-Port Aggregation (APA) product from the February 1999 Application Release. 2. Using the DLPI_SET_RIF_ON ioctl in a 32 bit application running on a 64 bit system results in an ENXIO error return. 3. Systems using the built-in 10 Mb/s LAN through an external MAU may see an excessive number of resets. Applications trying to transmit a packet could see an ENOLINK error return. 4. A system with HP-PB 10 Mb/s ethernet may panic under heavy load when available memory is low. 5. The MTU of built-in and HP-PB 10 Mb/s ethernet links cannot be adjusted. 6. The MIB counter ifInDiscards for HP-PB 10 Mb/s ethernet is abnormally high even though the link is operating normally. 7. lanadmin(1M) reports ifOper as down for an HP-PB 10 Mb/s ethernet interface even though the link is operational. 8. Resetting an HP-PB 10 Mb/s ethernet interface when it is hung can cause a system panic. 9. Attempting to unbind a protocol (e.g., executing ifconfig unplumb on an interface or issuing the DL_UNBIND_REQ primitive) when an inbound packet is being processed on the stream, can cause a system hang. 10. Links are unable to receive packets they transmit when in promiscuous mode. 11. "lanscan -v" dumps core if the system has the Auto-Port Aggregation (APA) product installed. 12. ping(1M) requests receive no response when source routing is turned off on Token Ring interfaces. 13. System panics on receiving a trailer encapsulated packet from mblk drivers. Based on HP-UX patch PHNE_16990: 1. An MP system under heavy networking load can panic or hang. 2. An application cannot enable more than 16 multicast addresses per stream. 3. Turning off promiscuous mode on interfaces that use sleep/wakeup can cause system panic. 4. Turning on multicast promiscuous mode on a bound stream causes all unicast packets to be dropped for that stream until the multicast promiscuous mode is turned off. 5. An inbound IP packet with no IP header causes system panic. 6. Disabling multicast addresses on an interface that uses sleep/wakeup can cause system panic. 7. Starting or stopping mrouted under heavy outbound IP traffic can cause system panic. 8. Performance of applications that use promiscuous mode is enhanced by this patch. Based on HP-UX patch PHNE_15969: 1. Support needed for HP Auto-Port Aggregation Software. 2. linkloop has hardcoded limits for -s option. 3. Needed run-time switch to turn on/off Inbound Packet Scheduling in dlpi pseudo driver. 4. With PHNE_15591, swverify(1M) on the Networking product fails with errors similar to the following: ERROR: Control file "unconfigure" should have size "545" bytes but the actual size is "555" bytes. ERROR: Control file "verify" should have size "2117" bytes but the actual size is "1807" bytes. ERROR: Fileset "Networking.LAN-RUN,l=/,r=B.11.00" had file errors. 5. Linkloop did not work with those ethernet links which support checksum offload feature. Based on HP-UX patch PHNE_15591: 1. Increased scalability by doing Inbound Packet Scheduling in DLPI. 2. System appears to hang with certain DLPI applications. 3. The lanadmin(1M) -b/-B options are not working. 4. swverify(1M) logs a large number of errors when the system does not have hardware supported by the lan2 and lan3 drivers (e.g., built-in 10 Mb/s Ethernet, HP-PB Ethernet, EISA 10 Mb/s Ethernet). An example of an error message logged is: ERROR: The LLA device files are not installed correctly. Based on HP-UX patch PHNE_14632: 1. No DELETE multicast requests were sent down to the driver when application issued a DELETE on a previously added multicast address. 2. Lanadmin fails to change MAC address for the Token Ring Interface. 3. dmesg shows debug message for the Token Ring Interface. 4. lanadmin needs easier way to reset MAC address to default. Based on HP-UX patch PHNE_14370: 1. Panic with kernel stack overflow during ServiceGuard switch over. 2. DLPI command after DL_HP_HW_RESET may get ENXIO error. 3. Need MBLK support for Token Ring Drivers. 4. DLPI type 2 won't work over FDDI devices that do not support the Checksum Offload feature (e.g. HSC and EISA FDDI) 5. IP multicast doesn't work for Token Ring devices. 6. Incorrect handling of promiscuous data by different LAN drivers. 7. IP Multicast does not work for drivers supporting checksum offload. 8. Lanadmin doesn't work on a system with large number of lan cards. Based on HP-UX patch PHNE_14106: 1. Lanscan does not recognize HyperFabric cards. 2. lanadmin does not support mode and speed changes for 100BT cards. 3. lanadmin won't work with lan0 when atm card / Hyperfabric driver is present. 4. lanadmin Unable to access PPA Number on 11.0 64 bit T600 LR+extension software bundle. 5. ATM interface named "el1000" is not displayed properly by lanscan. 6. lanscan -v report shared library error against ATM NICs. 7. HP 9000/UX ARP implementation causes caching of the least optimal path to a remote host on a multiple-ring Token Ring network. 8. DLPI performance enhancement. 9. linkloop (1m) causes memory leak and panic. 10. DL_HP_MULTICAST_REQ/ACK support for drivers supporting IP level multicasting. 11. Cannot send out files that are more than 1492 bytes over Ethernet with SNAP encapsulation. 12. The XID and TEST packet failed on DLPI type 2 service over mblk based drivers. 13. The DLPI stream bound to IEEE 802.2 saps (sap < 256) with DL_SUBS_BIND_REQ do not work correctly and could potentially panic the system. 14. linkloop returns ERANGE if MTU of link changed. Based on HP-UX patch PHNE_14027: 1. PHNE_13076 which was released in the 11.0 SR release, is not tagged as critical. A patch should have critical category tag to get released in 11.0 SR Extension Software media. Based on HP-UX patch PHNE_13076: 1. SIO/WSIO dependency is missing for HP-PB and Built-in ethernet drivers. 2. lanadmin command fails after sufficient changing of PPA/display commands when the LANG environment variable is set. 3. TCP packets are delayed. Defect Description: PHNE_29267: Ported HP-UX patch PHNE_28143 to VVOS Based on HP-UX patch PHNE_28143 : Fix Summary: For FDDI: 4 For Tokenring: 6 For Ether: 7, 9 Common to all products: 1, 2, 3, 5, 8 1. CR: JAGae43281 SR: 8606279225 Protocol value 0x1010 was being utilized for trailer encapsulation as per RFC0893. Resolution: Protocol value 0x1010 has now been released for application use. However, this restricts the maximum size of a trailer encapsulated packet from 8192 to 7680 bytes. 2. CR: JAGae23942 SR: 8606259624 Some of the function and data pointers are not properly reset. This is because information indicating that a DL_PROMISCON_REQ had been called when the application was still unbound is overwritten by the second DL_PROMISCON_REQ which was issued after a DL_BIND_REQ. As a result, the first incoming packet on the interface to which the application had been attached aborted the system. Resolution: DLPI now keeps track of the unbound DL_PROMISCON_REQ and ensures that the function and data pointers are properly reset before the application exits. 3. CR: JAGae22012 SR: 8606257709 When DLPI receives an inbound packet on a connection oriented link, it strips off the header and sends the data portion upstream to the next streams module. If there is no data associated with the packet, DLPI does not check if the memory block that should contain the data is NULL or not. Resolution: DLPI now checks if the memory block that should contain the inbound packet data is NULL before sending it upstream. 4. CR: JAGae23565 SR: 8606259247 The group bit interpretation for FDDI frames was incorrect. Resolution: The group bit interpretation for FDDI frames has been corrected. 5. CR: JAGae42209 SR: 8606278150 DLPI master config file was missing details for debug file inclusion during kernel build. Resolution: Master config file has been updated to include DLPI debug file during kernel build. 6. CR: JAGae58428 SR: 8606294733 Special case handling of DL_HP_MULTICAST_LIST_REQ for IP multicast addresses enabled on tokenring was missing. Resolution: Special case handling of DL_HP_MULTICAST_LIST_REQ for tokenring now in place. 7. CR: JAGae57955 SR: 8606294216 Padding of < 64 byte packets was not being handled by NIO 10 BT driver correctly. Resolution: Padding of < 64 byte packets is now properly done. 8. CR: JAGae58591 SR: 8606294894 lanadmin(1m) can dump core under certain circumstances. Resolution: Boundary checks are made to prevent lanadmin(1m) aborts. 9. CR: JAGae65127 SR: 8606301682 Panic was due to Data Page Fault caused by NULL pointer access. Resolution: NULL pointer access has been prevented. Based on HP-UX patch PHNE_26250: Fix Summary: For Gelan: 4, 6, 7 For Token: 18 For Igelan: 11 For Non-FDDI: 13 Common to all products: 1, 2, 3, 5, 8, 9, 10, 12, 14, 15 16, 17, 19, 20, 21, 22, 23, 24, 25 1. CR: JAGad92200 SR:8606223097 Due to shortage of memory, memory allocation may fail and pointer should be checked for NULL before accessing it. Resolution: Now each memory allocation is checked for failure and corresponding action is taken for those cases 2. CR:JAGad94736 SR:8606225662 HP-UX only allow one unbound promiscuous stream per interface When trying to enable promiscous mode on second unbound stream, application will expect an EBUSY error instead of EINVAL error. Resolution: DLPI return EBUSY for this case now. 3. CR:JAGad95328 SR:8606226258 Failed memory allocation is not checked before accessing it. Resolution: Memory allocation is checked for failure. 4. CR: JAGae08292 SR: 8606239879 For CKO driver such as gelan, DLPI need to allocate extra 8 bytes, because driver unconditionally strips the first 8 bytes even though there is no CKO info present. The problem is for rawdata mode, DLPI didn't allocate extra space before LLC header. Resolution: DLPI now leave 8 bytes before LLC header for CKO driver. 5. CR:JAGad96423 SR:8606227362 Due to the race condition bewteen bind and unbind, some data will be corrupted and lead to system panic. Resolution: New mechanism is provided to synchronize the bind and unbind operation. 6. CR:JAGae00121 SR:8606230883 Gelan driver requires DLPI leave extra space before LLC header for CKO support. DLPI type 2 implementation didn't do that before. Resolution: Extra space is allocated in the beginning of a packet by DLPI before it is sent to gigabit driver. 7. CR:JAGae00493 SR:8606231255 Gelan driver will check the packet type and reset length field of LLC header if needed. For RAW mode stream, the upper layer build the LLC header, driver should not touch the packet. DLPI should notify driver of this information. Resolution DLPI set the packet type to a specific type so that gelan driver will not modify the packet for raw mode streams 8. CR:JAGae01356 SR:8606232120 DLPI allocate a fixed size memory for multicast list, and driver doesn't check the size, and may overwrite the memory, which may corrupt or panic the system Resolution: Now driver allocate the memory according to the number of multicast address enabled in that interface 9. CR:JAGae04997 SR:8606235861 For promiscuous mode handling, DLPI duplicate the packet. And the packet may be forwarded to multiple streams. If one of the stream modifies the packet, the other stream will see a corrupted packet. Resolution: Now DLPI will copy the whole message instead of duplicating the message. 10.CR:JAGad92458 SR:8606223361 For non-IP path, the order of inbound packet was not guaranteed by DLPI due to a race condition between different threads when forwarding the packet to the upper layer. Resolution: Now DLPI alway guarantee the order of packet by eliminating the above race condition. 11.CR:JAGae05586 SR:8606236529 ER: DLPI need to support igelan for 11.0 release Resolution: DLPI now support igelan driver. 12.CR: JAGad62560 SR: 8606193348 DLPI should check whether the multicast addresses or promiscuous mode enabled on an interface belongs to IP stream, if so, we should still allow APA to configure the interface as primary interface in LAN MONITOR mode. Resolution: DLPI now only check multicast addresses and promiscuous mode for NON-IP streams. 13. CR: JAGae08755 SR: 8606241497 There is a different loopback flag for MBLK and MBUF driver, the handling of outbound packet always use MBUF flag. Resolution: DLPI checks appropriate flag for loopback. 14. CR: JAGae08757 SR: 8606241499 Incorrect flag is used to check if promiscuous mode is enabled. Resolution: Correct flag is used to check whether the promiscuous mode is enabled. 15. CR: JAGae11948 SR: 8606245478 When DLPI buildes LLC header, if there is no memory, DLPI return DL_BADADDR unconditionally. Resolution: DLPI check what causes the failure of building header and return ENOBUFS if it is because of memory shortage. 16. CR:JAGad92395 SR:8606223292 Possible Memory leak when promiscous mode is enabled disabled and the reenabled. Resolution: Handled this case by dropping the dup'd packet when promisc is disabled. 17. CR:JAGad85055 SR:8606215881 When setting MAC address, dlpi unconditionally sends link up event to upper layers. Resolution: Now DLPI checks the operation status before sending link up event. 18. CR:JAGad75002 SR:8606205827 Macro to delete multicast address (DEL_EXT_MCAST), returns EINVAL while disabling IP multicast. Resolution: This defect is because Token ring handles IP and Non-IP Multicast differently. Handled this case by checking for Token ring specific type and returning appropriate error. 19. CR:JAGad90159 SR:8606221023 When reset statistics request is received clear the DLPI MIB stat for inbound discard type. This is used for debugging and has no user visible impact. Resolution: Discard types are reset during a reset MIB statistics request. 20. CR:JAGad90136 SR:8606221000 DLPI always returns -l when a link down event is sent by driver. Resolution: Appropriate error returned to the driver on link down event. 21. CR:JAGad87946 SR:8606218798 Lanadmin output is incorrect after resetting the MIB statistics of an APA link aggregate. Resolution: DLPI statistics are reset appropriatly for an aggregate and the interface. 22. CR:JAGad83398 SR:8606214207 The multicast list is not freed during detach. Resolution: The multicast list is freed on detach request during the closing of a stream. 23. CR:JAGad71218 SR:8606202044 Defect in handling APA_EVENT_LINKAGG_DOWN and APA_EVENT_LINKAGG_UP. Resolution: Link down/up notification is now being sent to all streams bound to an aggregate on event of aggregate going down or coming up. 24. CR: JAGad66889 SR: 8606197697 hpgelan doesn't report failure to set MTU size. Resolution: Lanadmin is modified to return non-zero upon failure. 25. CR:JAGae07043 SR:8606238001 When a get MIB statistics request is received, panic in hp_dlpi_get_mibstats() since DLPI processes the request without checking the state of stream. Resolution: DLPI checks the state of stream before processing the request and returns error if stream is unattached PHNE_25076: Ported HP-UX patch PHNE_24413 to VVOS Based on HP-UX patch PHNE_24413: Fix Summary: For Token: 1,2 For FDDI: 5 Common to all products: 3, 4 1. CR:JAGad64041 SR:8606194835 DLPI drops Non-IP Multicast packets for Token Ring. Resolution: On recieving a Multicast packet for token ring DLPI now checks for IP and Non-IP multicast MAC addresses that are being listened to. 2. CR:JAGad58592 SR:8606189377 Panic in rif_putq(). Trace looks as follows: panic+0x14 report_trap_or_int_and_panic+0x80 interrupt+0x1d4 $ihndlr_rtn+0x0 rif_putq+0x1dc hp_dlpi_mblk_intr_put+0x594 streams_put_release+0x1cc hp_dlpi_mblk_intr+0x4c pcitr_recv_complete+0xe68 pcitr_isr+0x33c sapic_interrupt+0x2c mp_ext_interrupt+0x34c ivti_patch_to_nop3+0x0 idle+0x184 swidle_exit+0x0 or crash event was a panic panic+0x14 report_trap_or_int_and_panic+0x80 interrupt+0x1d4 $ihndlr_rtn+0x0 bcmp_checkitout+0x18 rif_searchq+0x38 rif_putq+0x1f0 hp_dlpi_mblk_intr_put+0x594 streams_put_release+0x1cc hp_dlpi_mblk_intr+0x4c pcitr_recv_complete+0xe68 pcitr_isr+0x33c sapic_interrupt+0x2c mp_ext_interrupt+0x34c ivti_patch_to_nop3+0x0 idle+0x954 swidle_exit+0x0 Resolution: Protected the read and write to rif_queue using locking machanism. 3. CR:JAGad64335 SR:8606195129 The fddi links dropped during blast test. Resolution: On receiving a linkup/linkdown event from the driver DLPI sends a return value back to the driver saying it is able to intimate the upper module of this event. 4. CR:JAGad54952 SR:8606185750 Spinlock/Deadlock for lanc_lookup_lock in while trying to remove a protocol that is logged. Trace looks as follows panic+0x10 too_much_time+0x238 wait_for_lock_spinner+0x2f4 wait_for_lock_4way+0x2c sl_retry+0x18 lanc_remove_protocol+0xb8 lanc_lla_close+0x2d0 closef+0x68 exit+0x214 rexit+0x20 syscall+0x75c $syscallrtn+0x0 Resolution: Handled by changing the locking implementation for logging and removing protocols. 5. CR: JAGad73103 SR:8606203925 IP multicast rcved on FDDI is broadcast forwarded back to FDDI. Resolution: MAC address of FDDI packet is checked for Multicast bit and appropriate information is forwarded to IP. Based on HP-UX patch PHNE_22962: Fix Summary: For ETHER: 6 FOR TOKEN: 3 Common to all products: 1, 2, 4, 5 1. CR: JAGad37106 SR:8606167823 DLPI does not track dropped packets. Resolution: Updated the statistics of the interface on events of packet discard and recieve. 2. CR: JAGad37107 SR:8606167824 DLPI should not attempt to requeue in raw mode Resolution: Disabled the requeuing of packets when DLPI used in raw mode. 3. CR: JAGad21664 SR:8606152325 DLPI multicast test failed for token ring interface. Resolution: This is due to call of DL_ENABMULTI_REQ failed by DLPI. This is handled by DLPI checking the right list that is passed from the driver on completion of request. 4. CR: JAGad23566 SR:8606154248 NIO Token ring driver reports "Hard error on CTL_CCMD=0x1060" in syslog. Resolution: This was due to race condition between checking the protocol logged and the driver actually logging the protocol. Handled by locking while the driver is yet to finish logging the protocol. 5. CR: JAGad41174 SR:8606171911 DLPI reuses mblks without adjusting the size when it converts struct DL_BIND_REQ to the struct DL_HP_BIND_REQ. Resolution: The size is adjusted to accommodate DL_HP_BIND_REQ struct. 6. CR:JAGad55423 SR: 8606186218 On receiving a bad formatted TRAILER protocol frame the system crashes. Resolution: The badly formatted packet is detected and dropped. Based on HP-UX patch PHNE_22353: Fix Summary: For ETHER: 3 Common to all products: 1, 2, 4, 5, 6, 7, 8, 9 1. CR: JAGad31160 SR: 8606161844 When a wrong ppa number was passed as parameter to lanadmin, it displays the value corresponding to PPA 0. 2. CR: JAGab75792 When the station Address argument passed to lanadmin is too long, lanadmin does not display any error message and does not return the value indicating the failure of the command. This was causing problem for the driver specific configuration scripts as they were not able to detect the failure but the card also was not getting configured. Resolution: lanadmin will now print an error message for the long mac address and will exit with -1. 3. CR: JAGad38176 SR: 8606168898 The system was panicing while running the promiscuous mode tests. The panic was because of a window, where the promisc stream can be freed but still hp_dlpi_intr tries to access it. Resolution: The Core Ethernet code is modified to take care of the problem by checking the content of the variable before it was being used. 4. CR JAGad35507 SR 8606166220 When transmit line being severed and the receive lines being intact, the card will still receive packets from the wire. So when the driver sends out a loopback packet in the reset path, it assumes any response to be the loopback packet response. So, it assumes success of loopback even though the packet it received was not a loopback packet. This causes the driver to come online. The next transmit will again timeout causing another reset. These continuous resets lead to the bus timeout and hence the panic. Resolution: Ensure that the packet received in lan2_process_looper is really a loopback packet. 5. CR JAGad27778 SR 8606158448 System may panic when packets with no data is passed up to the IP and when the packets have MSG_CKO flag turned on. The panic can occur when the application receives unitdata packets. Resolution: Any packet with no data is dropped at DLPI level. 6. CR JAGad39929 SR 8606170665 When a promiscuous application is killed abruptly, DLPI sends a DL_PROMISCOFF_REQ, on which driver could sleep. DLPI tries to access data during wakeup to complete the request and inadvertently accesses the element of a structure which may not be available for this particular request, causing the system to panic. Resolution: The data which DLPI accesses during wakeup is done depending on the request, instead of doing it at the common place, there by looking into data which will be available only for this request. 7. CR JAGac40857 SR 8606125467 When a application in promiscuous mode turns off the promiscuous mode or is killed, driver may still pass the promiscuous mode packet to DLPI, but the promiscuous stream may no longer be available when DLPI tries to send the packet up-stream, causing the system to panic. Resolution: The validity of the promiscuous stream is checked before DLPI sends packets upstream. 8. CR JAGad03223 SR 8606134083 System runs out of kernel virtual memory space as packets are queued up in streams above the driver. DLPI is not the primary culprit in this defect but during the time of investigation, it was found that there is a possibility of DLPI's internal queues could grow unbound. Resolution: DLPI's internal queue is flow controlled to avoid the queue from growing unbound. 9. CR JAGad09699 SR 8606140332 System may hang when there is a broadcast storm making the queues to grow unbound and eventually hanging the system. The problem was with STREAMS which can allow the queues to grow without limits. Resolution: DLPI uses new STREAMS function 'sq_canput' which flow control the streams. PHNE_22825: Ported HP-UX patch PHNE_22331 to VVOS Based on HP-UX patch PHNE_22331: Fix Summary: For ETHER : 5 Common for all products: 1,2,3,4. 1. CR: JAGac46780 SR: 8606126024 Though the support for "q" option was already added to lanscan, the usage string was not reflecting the same. Resolution: Modified the usage string to add "q" in the options available. 2. CR: JAGad09461 SR: 8606140139 There is no command line option to get the statistics and some other information related to a lan card from lanadmin. Resolution: "-g" command line option is added to display information which was ealier obtained by "display" option in interactive mode. The lanadmin manpage is also modified to give more information on new option. 3. CR: JAGaa94099 The "SEE ALSO" section of lanadmin manpage was referring to "DARPA Requests for Comments" which should be "IETF Requests for Comments" Resolution: Man page is modified. 4. CR: JAGad29655 SR: 8606160333 The lanadmin usage string for x and X option was referring to "Network Management ID" instead of PPA Number. Resolution: Modified the usage string. 5. CR:JAGad30864 SR:8606161548 The problem is due to improper lanscan -ip output for the lan cards with instance number >9. Resolution: The script has been modified to use lanscan output instead of "lanscan -ip output" to get the required information. Based on HP-UX patch PHNE_21905: Fix Summary: For ETHER: 1,2,3. For FDDI: 6. Common for all Products: 4,5,7,8. 1. CR: JAGab46943 SR:5003459701 An unterminated/disconnected Core 10BT card causes data loss over the serial port. This is because when the card is disconnected, the driver periodically tries to reset the card. The reset of the card involves many calls to busywait, which leads to poor serial I/O performance. Resolution: The resolution is to eliminate some of the busywait calls in the reset path. 2. CR: JAGad02482 SR: 8606133337 If the memory allocation fails during compression of outbound packets HP-PB 10BT driver was dropping the packet but not freeing the memory appropriately. This could lead to memory leak. Resolution: HP-PB 10BT will now free the memory appropriately when it does not get memory to compress outbound packets. 3. CR: JAGad13235 SR: 8606143902 The MIB statistics were being updated for a HP-PB 10BT card even when the cable was disconnected. This was causing the Service Guard to misbehave as it depends on the MIB statistics to detect the failure or recovery of LAN cards. Resolution: The statistics now gets updated when the packet is successfully processed and transmitted. 4. CR: JAGac88749 SR:8606129845 lanadmin manpage had incorrect information about -m option. The man page for "-m" option was as follows -m Display the current MTU size of the interface corresponding to PPA. You must have superuser privileges. Resolution: The fix is to remove "You must have superuser privileges." for -m option from lanadmin man page. 5. CR:JAGad21159/JAGad10603 SR:8606151820/8606141242 The patches on 11.0 were missing the link between lanadmin and landiag. Due to this, the modifications released in patch for lanadmin were not reflected by landiag command. Resolution: The fix is to create a hard link between lanadmin and landiag. 6. CR:JAGad04612 SR:8606135478 To fix out-bound checksum offload problem with HP-PB FDDI when aggregated, Lan Monitor applied a short term workaround. This resulted in HP-PB FDDI port based aggregates not being shown on lanscan output. Resolution: The fix introduces a new features bit - DRV_CKO_SR71_FLAVOR, for HP-PB FDDI to set and DLPI checks this features bit rather than looking for driver name to determine if SR-71 flavor of checksum offload is being used. 7. CR:JAGad14400 SR:8606145062 An operator precedence problem led to a DLPI type 2 log lock being freed twice. This caused memory corruption and eventually a data page fault panic. Resolution: The operator precedence was set right to take care of this corruption. 8. CR:JAGad13239 SR:8606143906 DLPI returns DL_ERROR_ACK in the unitdata path, which is not in conformance with DLPI standard. As this error is not expected in this path, Service Gaurd used to report strange primitive error when DL_UNITDATA_REQ fails. Resolution: The unitdata path was changed to return only DL_UDERROR_IND for failures in DL_UNITDATA_REQ. Based on HP-UX patch PHNE_21217: Fix Summary: For FDDI: 2. For ETHER: 3. For Token Ring: 4, 5. Common for all products: 1, 6, 7, 8, 9, 10. 1. JAGab69560: T2 XID/TEST scaffold tests fail, because the T2 code in DLPI, assumes that the XID/TEST data is present in first mbuf, but there is a possibility that the first mbuf only contains the LLC header and the XID data is present in second mbuf. This assumption was wrong, which created unexpected response from DLPI to an application. Resolution: The T2 code has been changed to take care of this situation, where the XID data can be present in the second mbuf. 2. JAGac88801: When Enable/Disable multicast address request was sent, DLPI does not check for invalid multicast addresses for FDDI driver, so, if an application tries to enable a invalid multicast address, on a FDDI link, it succeeds. Resolution: The code change done to resolve this defect, checks for the device type of FDDI when it tries to enable a valid multicast address. 3. JAGac95565: lanc fails to prevent setting Station Address to a multicast MAC address. As the check done to prevent this from happening, checks for the device type of DEV_8023, but some ethernet drivers set the device type to DEV_ETHER, so this checking fails to prevent setting multicast address as the station address when the drivers set the device type to DEV_ETHER. Resolution: The checking done when trying to change the station address now includes the check for device type of DEV_ETHER. 4. JAGac86212: System paniced when disable multicast request was sent to a Token Ring interface. The Token Ring mulitcast address is not stored in global list of mulitcast addresses maintained by DLPI, but when disable mulitcast address is sent, the DLPI module tries to find out the address in the list, there by manipulating invalid entries in the multicast list, thus causing the system to panic. The stack trace for the panic: 0x00050d34 (panic+0x4c) 0x0009968c (report_trap_or_int_and_panic+0x8c) 0x0009d560 (interrupt+0x3c0) 0x000b2f6c (set_protid_registers+0xa28) 0x00591dc4 (hp_dlpi_process_wakeup+0x51c) 0x00591558 (hp_dlpi_wakeup_put+0x30) 0x00630818 (streams_put+0x1e0) 0x005914c8 (hp_dlpi_wakeup+0x200) 0x008784c0 (pcitr_process_srb_response+0x238) 0x0088450c (function name is not available) 0x00883d58 (function name is not available) 0x00883878 (pcitr_isr+0x638) 0x0025c80c (sapic_interrupt+0x34) 0x0009d0b0 (mp_ext_interrupt+0x268) 0x000b2f48 (set_protid_registers+0xa04) 0x0033a36c (idle+0x115c) 0x000b1f7c (getproc_info+0x28) Resolution: The manipulation of the list is done depending on the return value of a macro which searches the list for possible match in multicast address. For TR this match operation will fail, there by preventing the list from being manipulated. 5. JAGac86794: The output routines in DLPI, do wrong checks to find the Token Ring multicast address. There are two separate checks done depending on the MAC type of the card. For token ring cards, the check is done on the functional address, but in addition to this the address is searched in DLPI global multicast address list, there by undoing the previous check. Resolution: Depending on the MAC type of the card, only one of the above mentioned check is done and ensuring the relevant multicast addresses are searched. 6. JAGad00107: ER: Support for filtering IEEESAP_HP packets in lanc. Resolution: This enhancement has been done in lanc to support filtering for IEEESAP_HP packets. 7. JAGad00334: DLPI event APA_EVENT_NON_IP_LOGGED is not correctly implemented for Token Ring interface. The dlpi_event() function uses small-endian format to check for IP multi-cast address comparison. Resolution: The function is converted to check for IP multicast addresses in wire format(big-endian) for Token Ring interfaces. 8. JAGad01906: lanadmin command has a predefined buffer size, to read data from the stream. This buffer size is insufficient for machines which mave more than 20 interfaces. When it tries to read the hwift's of the interfaces from the system, it leaves out some data on the stream head because it reads less data than what is available. This causes the subsequent operations fail. Resolution: The buffer size is redefined to accommodate larger amount of data. 9. JAGac87773: The DLPI data structure hp_dlpi_data_t has been changed, but the debug information available does not reflect this change. Resolution: Redefined the hp_dlpi_data_t to hp_dlpi_data_t_2 for debug purpose so that it can override the original definition and reflect the change in data structure. The redefinition is only for debug purpose. 10. JAGad08366: Changing the MAC address can cause system panic. The disable multicast requests are sent when the driver is still on ICS, and driver tries to hold a lock which is already held. Resolution: When the driver wakesup DLPI, it used to send upper layers of the change in link state to down, which caused the upper layer to send disable multicast requests when the driver is still on ICS. But now DLPI instead of sending the link down indication as soon as it wakes up, it puts the indication in its own queue which later can be processed by DLPI's service routine, thus avoiding the senario where the whole thread runs on ICS. Based on HP-UX patch PHNE_20657: 1. JAGac56887 ER: Changes needed in DLPI to support APA features. Resolution: Changes are made to DLPI to support some functionalities of APA. The changes done were addition of new IOCTL's APA_SET_DEFAULT_CONFIG_MODE, APA_GET_DLPI_VERSION & APA_EVENT_SET_DEFAULT_CONFIG_MODE. 2. JAGac39764: The DLPI function hp_dlpi_subs_bind() is taking the MTU value from a constant, instead of from the hw_ift structure to use for the SNAP encapsulation. The hp_dlpi_subs_bind() was taking its value from a constant SNAP8023_MTU. But when the MTU value changes, as in case of Jumbo frames, the DLPI internal data structure does not reflect the change which causes this problem. Resolution: The MTU value is taken from interface's hw_ift structure to reflect the current MTU value. 3. JAGac56915: Problem with DL_SUBS_BIND_REQ Resolution: Update the DLPI internal data structure only after the protocol logging is successful. 4. JAGac56732: When MAC/PHY address changes on a link, user had to bring the link down and then up to generate the ARP request, but this needs to be done by DLPI, because when APA is installed on the machine, and the link where the MAC address has changed is present in the aggregate as a primary link, then when user brings the link down, APA thinks that that the has gone down and tries to use the secondary as its primary link, which again, brings the secondary link down to change its MAC address, this causes a ripple effect and the aggregate goes down. Resolution: Changed DLPI to generate the ARP packet whenever MAC address changes and there by bypassing APA. 5. JAGac78533: The current implementation of DLPI fills the source MAC address for outbound packets in some of the outbound paths, but this created problem when using APA, where the aggregate had a single MAC address. When the packets from the aggregate reaches the driver, the driver again fills up the source MAC address which can differ for different links in the aggregate. This created problem with the switches. Resolution: DLPI has been modified to fill the source MAC address for every outbound packet in all the paths, and also letting the driver know that DLPI is doing so, by setting a features bit in the hwift structure of the driver. If the bit is set then, the driver knows that DLPI has filled the source MAC address and it need not fill it again. 6. JAGac39928: The operator precedence is not checked properly before trying to log the trailer protocol address in lanc_init(). Resolution: The code has been modified to explicitly enforce precedence in proper manner. 7. JAGac79105: Lan Monitor Support in DLPI: Resolution: The following changes were made to DLPI for Lan Monitor support: 1. Support for binding protocol kind LAN_CANON that will enable receiving poll packets by Lan Monitor aggregated ports. 2. Handle events DL_EVENT_INTERFACE_DOWN, DL_EVENT_INTERFACE_UP from drivers and APA_EVENT_PROTOCOL_LOGGED from Lan Monitor. 8. JAGac78287: DLPI does not propogate Link UP/DOWN events to all attached IP streams on an interface. Resolution: DLPI used to send link UP/DOWN events to the first attached IP stream on a interface. The behaviour is changed with this patch to send link UP/DOWN events to all the IP streams attached to a interface. 9. JAGac86142: OTS when used along with Service Guard fails to switch interfaces when the primary link goes down. When OTS switches from one interface to another, it tried to move all the multicast addressess from the original interface to the new one, but DLPI fails to remove all the multicast addressess enabled on the stream, but error is not returned about the failure of switching. When OTS tries to enable the same multicast address again on the new interface, the request fails, so OTS marks the interface down. DLPI fails to allocate mblk of proper size for disable multicast request. Resolution: DLPI allocates mblk of proper size of disable multicast address request. Based on HP-UX patch PHNE_19766: 1. HP-PB 10BT and Core 10BT drivers did not loopback outbound packets while in promiscuous mode. So, the stream which turned on the promiscuous mode didn't receive the outbound packets. Resolution: Drivers now will loopback the outbound packets to DLPI and DLPI will send the packets upstream. 2. HP-PB 10BT was exhausting available IOVRs resulting into unavailibility of IOVRs for use and thus causing slow I/O or hang. This happens as many small mblks are being transferred down from the TCP/ IP layers. Resolution: An additional check has been introduced to find if the number of mblks per frame coming to driver are more than 8 and compress them if greater than 8, hence eliminating the need for multiple mblks. 3. Ethernet card does not get initialized as per the settings specified in configuration file upon reboot. The problem is seen for the cards whose information in the configuration file is at index greater than the number of cards in the system. Resolution: Fix includes reading complete configuration information by correct manuplation of index counter in the init script. 4. HP-PB 10BT can panic the kernel under low memory condition at boot time. The panic happens due to access of NULL pointer returned by memory allocation function. Resolution: Fix is to check the return value from memory allocation function and leave the card in un-initialized state if the return value is NULL at boot time. 5. The Service Guard test does not pass as HP-PB 10BT driver does not send the NMV_LINKUP event. Resolution: Fix is to send the NMV_LINKUP mib event to lanc. 6. Token Ring requires token_arp module to be configured on the system to work. PHNE_18409 takes care of adding token_arp module to system file, but the superseding patches PHNE_19081 and PHNE_19754 do not take care of this. Resolution: Addition of token_arp module to system file is re-introduced in this patch to fix the problem. Based on HP-UX patch PHNE_19754: 1. Connection oriented DLPI applications on an Ethernet link could hang while waiting for a DL_DISCONNECT_IND message to arrive, because of a defect in DLPI code. Resolution: This defect has been fixed by ensuring the correct sequence of break statements. Based on HP-UX patch PHNE_19081: 1. If an arp packet is sent with incorrect header field values on a token ring LAN, it might result in a system panic. Resolution: DLPI has been fixed to check for the correctness of header fields in an arp packet on a token ring LAN. 2. A cable disconnect did not result in a link down event being sent upstream. Resolution: A link down event is now sent upstream. 3. In outbound path involving DL_UNITDATA_REQ DLPI primitive, checksum offload processing needed initialization of the fields of a structure. Resolution: This problem has been fixed by initializing the fields of the above structure. 4. When a request to enable a token ring functional address is received, DLPI fails to process it correctly. Resolution: DLPI processing of token ring functional addresses has been corrected. 5. When a type 2 DLPI connection was automatically responding to a TEST/XID packet, an incorrect typecasting of a pointer caused system panic. Resolution: Typecasting of the pointer is being done correctly. 6. Auto-port Aggregation product was sending redundant messages to the console. Resolution: These messages are no longer sent to the console. 7. The Streams framework reschedules procesing of a packet on the destination processor, if the stream is in uniprocessor emulation mode. When processing of the packet is resumed, it did not check if the stream is still open. This resulted in a system panic. Resolution: DLPI now uses the Streams interface that does these checks. 8. When a Core-Ethernet NIC is neither connected to a LAN nor terminated, the Core-Ethernet driver detects the loss of carrier and begins to continuously reset the card. As part of the resets, the driver calls the system busywait routine that makes the CPU spin for a fixed period of time, for a total of up to 20 milli-second per reset. As a result real time applications detect a 20 milli-second delay in completion of their operations, once every 5 seconds. Resolution: A global variable lan2_no_reset_on_disconnect has been made available to be set using adb(1M) command. By default this variable is set to 0 and there is no change in the behaviour of the Core-Ethernet driver. When this variable is set to a non-zero value using adb(1M), the Core-Ethernet driver will skip all driver initiated card resets. This will result in busywait routine not being called and hence avoid affecting the performance of real time applications. This variable can be set using the adb(1M) command (as super-user) : # adb -w /stand/vmunix /dev/mem # lan2_no_reset_on_disconnect/W 1 and then exit adb. When this variable is set, the Core-Ethernet driver will not be able to detect a cable reconnect event. Hence this solution should be used only when both the following conditions are true: - the Core-Ethernet NIC is not used and not connected to a network - the Core-Ethernet NIC cannot be terminated - real time application performance is impacted as described above. When normal usage of the Core-Ethernet NIC on a network is desired, the Core-Ethernet NIC should be connected to a functional network ( or terminated ) and a driver/card reset initiated using the lanadmin(1M) command. This action will restore the driver to it's default behaviour, also resetting the value of the global variable lan2_no_reset_on_disconnect to 0. PHNE_18560: Ported HP-UX patch PHNE_18409 to VVOS Based on HP-UX patch PHNE_18409: 1. The Core-Ethernet lan interface did an extra DMA to a receive frame descriptor under the following condition: when the Core-Ethernet lan interface has used up the last available receive frame descriptor, it generates a "Frame Received" interrupt and also a "Receiver Not Ready" interrupt. After processing each of these interrupts, the driver attempted to start the receive engine with a new pool of receive frame descriptors if the receive engine status showed "Not Ready". Sometimes this could result in the same pool of buffers being set up twice for DMA. This caused corruption of the 256 byte bucket. The system paniced on the next 256 byte allocation request. Resolution: The Core-Ethernet driver has been corrected to start the receive engine only on receipt of a "Receive unit Not Ready" interrupt. 2. A driver returns a negative pointer value when it can not complete a control request immediately. The DLPI checks the negative return value to unlock the driver lock. The return value needs to be a 64bit quantity on a 64bit system. One of the functions that receives this value and passes it to the DLPI did not have the right prototype. This caused the driver lock to be held forever resulting in the lock deadlock. Resolution: The lan common routine has been modified to have the correct prototype. Based on HP-UX patch PHNE_17113: 1. Enhancements to the functionality of the Auto-Port Aggregation (APA) product have a dependency on this version of DLPI. 2. DLPI, when running on a 64 bit kernel, did not recognize the ioctls DLPI_SET_RIF_ON, DLPI_SET_RIF_OFF and DLPI_RIF_INFO from 32 bit applications. 3. When the built-in 10 Mb/s LAN interface returned a "No carrier" status along with a "Late collision" on a transmit request, the driver treated the situation as a carrier loss and reset the interface. Resolution: The driver treats the situation as a late collision. 4. The driver did not check the return value from a buffer copy function. Resolution: The driver now checks for the return value and sends the packet upstream only if the copy is successful. 5. Built-in and HP-PB 10 Mb/s ethernet did not support changing the MTU. 6. The filter implementation in the HP-PB 10 Mb/s ethernet card firmware incorrectly incremented the bad address counter. Resolution: The fix is to ignore the bad address counter while computing the ifInDiscards statistic. 7. The HP-PB 10 Mb/s ethernet driver initialization code did not mark ifOper up when initialization was complete. 8. The HP-PB 10 Mb/s ethernet driver incorrectly reused memory it freed. 9. DLPI incorrectly called a STREAMS interface to clean up the queues when a protocol is unbound, an interface is detached from the stream, or a stream is closed. Resolution: DLPI no longer calls the interface. The STREAMS framework now cleans up the queues by itself. 10. Drivers did not loopback outbound packets while in promiscuous mode. DLPI did not have the filter to send such packets upstream. Resolution: DLPI now has the required filter. Drivers will individually release patches or application releases to support looping back outbound packets in promiscuous mode. This patch would be a prerequisite dependency. 11. lanscan passed an incorrect data structure to the APA lanscan shared library. 12. An ARP reply received by a Token Ring interface with source routing disabled was not sent by DLPI to the ARP resolver in a proper form. 13. The link level functionality to receive a trailer encapsulated packet was not provided for the mblk drivers. Based on HP-UX patch PHNE_16990: 1. Under heavy networking load, DLPI wakeup data structures were getting corrupted in certain situations. 2. DLPI had a limit of 16 on the number of multicast addresses that could be enabled per stream. This limit has been removed by this patch. 3. The promiscuous level was not saved correctly during sleep processing. 4. DLPI was incorrectly dropping unicast packets even for a bound stream while in multicast promiscuous mode. It is only supposed to do so for unbound streams. 5. DLPI was incorrectly assuming that every inbound packet has data after link level header. 6 An incorrect argument was passed to a macro that searches for multicast addresses in a global list. 7. DLPI was not handling errors correctly when an attempt was made to send an IP packet while a control request was pending. 8. Performance of promiscuous mode applications is enhanced by avoiding a full copy of the packet. Based on HP-UX patch PHNE_15969: 1. DLPI has been enhanced to support HP Auto-Port Aggregation Software. 2. linkloop has been modified to use mtu size limit as in dl_ppa_info. 3. It has been observed that Inbound Packet Scheduling improved network performance only when the ratio of CPUs to NIC cards is high. Hence, a run time switch has been provided to be able to turn it on/off as desired. It is turned off by default. 4. PHNE_15591 corrected several problems with the control scripts included with the Networking product. The Software Distributor product (SD-UX) overwrote some of the changes that the patch made to the control scripts. This resulted in the error messages listed in the "Symptoms" field. PHNE_15969 uses a different mechanism to modify the product control scripts, thereby eliminating the overwrite problem. 5. Checksum offload ethernet drivers expect 8 bytes of checksum information for each outbound packet. This information was not handled correctly for test packets that are used by linkloop. The problem is corrected in PHNE_15969. Based on HP-UX patch PHNE_15591: 1. A LAN interface is bound to a specific processor for handling its interrupts. Every packet coming in through this interface will be processed by that processor initially. If the time spent on processing each inbound packet is not reasonably short, the processor could become a bottleneck. Inbound Packet Scheduling allows DLPI to pick a processor that may be different from the one handling the interrupt to process the packet up the networking stack. This can result in performance and scalability improvements, especially if the processor selected is the same as the one running an application that is to receive the packet. 2. Under heavy load, a LAN driver may return an ENOBUFS error to DLPI for each outbound packet. DLPI passes these error messages upstream. If the application does not read these messages, they will remain in the Streams head and consume memory resources. This eventually consumes all available memory and the system appears to hang. 3. When -X/-x options were added to lanadmin, a defect was introduced in parsing options. Due to this, the -b/-B options fail to work. 4. The verify control scripts supplied with the Networking product checked for the existence of the hardware supported by the lan2 and lan3 drivers and logged errors, warnings, and notes if these were not found. Newer HP-9000 systems have 100Base-T interfaces, for example, and these errors/warnings/notes are inappropriate. The scripts have been changed to no longer require the presence of the earlier hardware. Based on HP-UX patch PHNE_14632: 1. Multicast Delete requests were not sent down to the hardware level by LANC. This resulted in stale multicast addresses enabled in the hardware. This has been corrected. 2. For Token Ring, if the stream is not bound, the DL_SET_PHYS_ADDR_REQ request comes with 6 bytes mac address. Currently it is rejected in hp_dlpi_control if the mac address is not 6 + RIF_SIZE. 3. Some debug printf's for a dlpi fix (in PHNE_14370) was mistakenly left in the production code. Note that the messages by default also go into syslog file. 4. lanadmin -A is enhanced to reset the MAC address to default. When the address string is DEFAULT, lanadmin restores the default address for the card. Based on HP-UX patch PHNE_14370: 1. During the Service Guard switch over from one interface to another, the transport code has a window where an IRE has been marked off, but its arp cache is around. At this time for a packet going out to this removed interface, ip would not find an IRE for it and will try to forward it. Since the arp cache entry for it can be found, the packet will be sent down with the stale mac address. When the driver gets this packet, it loops back upward, and ip in turn forwards down again. This looping repeats until Time To Live expires and the packet gets freed. Since this is an mbuf driver, the freeing will go through all the wrapper free routines, and eventually causes stack overflow panic. 2. Process issuing DL_HP_HW_RESET was getting woken up after PORT_RESET phase of reset process. If a DLPI request is issued immediately after DL_HP_HW_RESET, ENXIO may be returned if the driver has not finished the whole reset process. PORT_RESET phase is now modified to wake up all entries in cmd_q other than the one corresponding to RESET process. The RESET process will be woken up by the process_looper after the loopback test. 3. Enhanced LANC to support MBLK message buffers for Token Ring Drivers. 4. The type 2 code has always assumed that the FDDI drivers support CKO feature. So when it builds the LLC header for FDDI, the CKO info is always prepended to it. This is incorrect for some recently introduced FDDI's that do not support CKO. 5. For Token Ring, DLPI returns incorrect dl_addr_length in the dl_info_ack, thus the subsequent DL_ENABMULTI_REQ request will fail due to the wrong dl_addr_length in the request. Other fixes are also included to support ip multicast for Token Ring. For 11.0 Token Ring, since lanscan, arp output, and lanadmin all present the MAC address in wire format, the MAC address in any request to DLPI is supposed to be in wire format. However, an exception was found for the case of ip multicast. DLPI fixes include the following - Bit swap the mac address in the DL_ENABMULTI_REQ and DL_DISABMULTI_REQ before passing it to the driver. - Bit swap the mac address for the outbound and inbound ip multicast packets. - Treat Token Ring's functional address as a legal multicast address for the inbound packets. 6. The promiscuous mode data filtering and routing of the same for the "unfiltered" stream was not supported at LANC/DLPI level. Individual drivers implemented this in different ways which lead to incorrect behaviour. By supporting this at LANC/DLPI level, drivers could directly use this feature rather than implementing this at driver level. 7. If a packet is a broadcast packet or a multicast packet, DLPI takes normal path instead of fast path. In normal path, checksum offload is not handled properly. That is, it does not prepend the cko_info block to the mblk, though MSGCKO is turned on in the b_flag. When MSGCKO is turned on, IP assumes the first 8 bytes as cko_info block and uses it and this results in IP dropping the packets as bad header. 8. When there are huge number of interface cards installed on the system(For example: on 64bit system, more than 11), lanadmin does not work properly due to insufficient allocation of buffer to retrieve the information regarding all the interface cards. Based on HP-UX patch PHNE_14106: 1. Lanscan is enhanced to recognize the MAC type of HyperFabric cards. 2. An enhancement has been made in lanadmin to get and set driver specific options. This will support speed and mode changes for 100BT cards. 3. When multiple interface cards exist on the system with the same PPA number and only one of them support DLPI, but one which is not supporting the DLPI is ahead of the supporting the DLPI in the list, this problem happens. This is because, not proper conditions being used while searching the list for the required PPA. 4. When there are huge number of interface cards installed on the system(For example: on 64bit system, more than 11), lanadmin does not work properly due to insufficient allocation of buffer to retrieve the information regarding all the interface cards. 5. The ATM interfaces up to el900 are well recognized at upper level (IP), but not el1000 or more. This is because, in lanscan command, the number of digits to use for display from the PPA/NMID is hard coded as 3. 6. When ATM card is installed on a system, lanscan -v command reports that it can not open a shared library /usr/lib/lanscan/libpeatm.sl. This is because, HP DLPI do not support ATM drivers and they have their own DLPI layer and hence ATM drivers do not supply the shared library used by lanscan command. The lanscan command needs to check whether the corresponding driver supports HP DLPI or not. Only for those driver supporting HP DLPI, lanscan should open the shared library. 7. HP 9000/UX Token Ring does not conform to the IBM Token Ring Network Architecture Specification with respect to maintaining the source routing information to an off-ring system which maybe compliant with the same specification. 8. DLPI performance enhancement by code path reduction on outbound and inbound fast path (IP uses it). 9. Due to a defect in hp_dlpi_close/detach some amount of memory was not being freed. This was causing memory leak. Also in hp_dlpi_close/detach the driver was accessing memory which was already freed. This was causing a panic. 10. For drivers supporting IP level multicasting, the multicast address list is stored in a separate list. Hence dlpi should check if the driver supports IP level multicasting. If so it should return the multicast address list from this list while processing DL_HP_MULTICAST_REQ. 11. When a stream bound to SNAP for ethernet, the mtu was not adjusted and remains to be 1500 (should be 1492). The TCP and IP send data based on the wrong size of mtu, which will exceed the driver can take and cause the incomplete data to be sent and be rejected in the received site. 12. In t2_t1_xid_auto_res() routine, the PUT_MQUAD() macro is being used. This macro is of multi-line code, thus it needs a parenthesis around it. 13. In hp_dlpi_subs_bind() routine, t2_log() is called with hp_dlpi_intr() for the mblk driver, and hp_dlpi_mblk_intr() for the mbuf driver. The hp_dlpi_intr() and hp_dlpi_mblk_intr() should be reversed. 14. "linkloop" does not check the MTU on the PPA prior to sending out the test frame. If the customer has changed the MTU of the link, linkloop will fail unless the MTU size (or less) is specified via the -s parameter. This causes the MC/ServiceGuard cmlogconfig to fail when it shouldn't. The fix would be check the mtu size via DL_HP_PPA_REQ (dl_mtu) or DL_INFO_REQ (dl_max_sdu) rather than using the default size for the mac type. Based on HP-UX patch PHNE_14027: 1. PHNE_13076 is not a critical patch. However since this patch has to be released in 11.0 SR Extension media it has to be tagged critical. Hence releasing a new patch superseding PHNE_13076 and tagging it as critical. Based on HP-UX patch PHNE_13076: 1. The sio/wsio dependency is missing from the master files of HP-PB and Built-in ethernet drivers. Consequently, if other drivers, e.g., stape, which specify this dependency are removed from /stand/system, sio/wsio also disappears and an unbootable kernel results. 2. Functions in lanadmin shared libraries for HP-PB and Built-in ethernet drivers, open catalog files using catopen() but do not close the same on return. Hence, when lanadmin invokes these functions within the lanadmin shared libraries, for a reasonably large number of times, number of open files will cross maxfiles limit causing a failure in the next catopen(). However this failure occurs only when the LANG environment variable is set as that would result in some appropriate catalog file being opened. If the LANG variable is not set ( unset ) no catalog files are opened and the default messages are printed and the problem does not occur. 3. A DLPI routine saves data from an uninitialized 32bit variable, which could be a random value. If this DLPI data happens to match two particular values, then the Ethernet driver cannot successfully deliver the IP packets until this DLPI data is changed. This results in the delay in delivering TCP packets. Enhancement: Yes PHNE_29267: Ported HP-UX patch PHNE_28143 to VVOS Based on HP-UX patch PHNE_26250: This is an enhancement to support igelan driver. Additional enhancements were delivered in a patch this has superseded. Please review the Defect Description text for more information. PHNE_25076: PHNE_22825: PHNE_18560: Enhancements were delivered in these patches. Please review the Defect Description text for more information. SR: 8606194835 8606189377 8606195129 8606185750 8606203925 8606167823 8606167824 8606152325 8606154248 8606171911 8606186218 8606161844 8606168898 8606166220 8606158448 8606170665 8606125467 8606134083 8606140332 8606126024 8606140139 8606160333 8606161548 8606133337 8606143902 8606129845 8606151820 8606141242 8606135478 8606145062 8606143906 8606129886 8606130675 8606128665 8606129248 8606130949 8606131180 8606132758 8606129439 8606139097 8606126322 8606124381 8606126350 8606126167 8606127732 8606124545 8606128305 8606127485 8606128595 8606114057 8606104232 8606103762 8606101233 8606111864 5003459701 1653294488 4701431064 4701431072 5003461681 4701423897 5003451328 5003251215 5000699934 5003419895 4701400879 5003412817 1653251025 5003388603 5003387050 5003320655 1653249565 1653230078 4701374710 8606279225 8606259624 8606257709 8606259247 8606278150 8606294733 8606294216 8606294894 8606301682 8606223097 8606225662 8606226258 8606239879 8606227362 8606230883 8606231255 8606232120 8606235861 8606223361 8606236529 8606193348 8606241497 8606241499 8606245478 8606223292 8606215881 8606205827 8606221023 8606221000 8606218798 8606214207 8606202044 8606197697 8606238001 1653287052 5003446518 8606104501 8606189223 Patch Files: Networking.LAN2-KRN,fr=B.11.04,fa=HP-UX_B.11.04_64,v=HP: /usr/conf/master.d/lan /usr/conf/lib/liblan.a /usr/conf/lib/lan_dbg0.o Networking.LAN2-KRN,fr=B.11.04,fa=HP-UX_B.11.04_32,v=HP: /usr/conf/master.d/lan /usr/conf/lib/liblan.a /usr/conf/lib/lan_dbg0.o Networking.LAN-RUN,fr=B.11.04,fa=HP-UX_B.11.04_32/64,v=HP: /usr/sbin/linkloop /usr/sbin/lanscan /usr/sbin/lanadmin /usr/sbin/landiag /usr/lib/lanadmin/libdslan2.1 /usr/lib/lanadmin/libdslan3.1 /usr/lib/nls/msg/C/lanadmin.cat /usr/lib/nls/msg/C/lanscan.cat /sbin/init.d/hpether Networking.NW-ENG-A-MAN,fr=B.11.04,fa=HP-UX_B.11.04_32/64, v=HP: /usr/share/man/man1m.Z/lanadmin.1m /usr/share/man/man1m.Z/lanscan.1m what(1) Output: Networking.LAN2-KRN,fr=B.11.04,fa=HP-UX_B.11.04_64,v=HP: /usr/conf/master.d/lan: PHNE_18409 $Revision: 1.3.118.3 $ /usr/conf/lib/liblan.a: lanc_input.c: PHNE_28143 Jun 13 2003 16:30:32 lan_token.c: PHNE_18409 Jun 13 2003 16:30:42 LAN: liblan: PHNE_28143 Jun 13 2003 16:30:49 lanc.c: PHNE_28143 Jun 13 2003 16:30:49 lan_t2.c: PHNE_28143 Jun 13 2003 16:30:55 lan_dlpi.c: PHNE_28143 Jun 13 2003 16:31:54 $Source: kern/sio/lan3.c, hpuxsysio, vvos_rose, rose 0313 $ $Date: 03/06/12 08:39:03 $ $Revision: 1.8.1.10 PATCH_11.04 (PHNE_29267) $ lan2.c: PHNE_22353 Date:Jun 13 2003 16:32:36 /usr/conf/lib/lan_dbg0.o: lan_dbg0.c: PHNE_28143 DATE:Nov 7 2002 14:18:00 Networking.LAN2-KRN,fr=B.11.04,fa=HP-UX_B.11.04_32,v=HP: /usr/conf/master.d/lan: PHNE_18409 $Revision: 1.3.118.3 $ /usr/conf/lib/liblan.a: lanc_input.c: PHNE_28143 Jun 13 2003 14:16:28 lan_token.c: PHNE_18409 Jun 13 2003 14:16:42 LAN: liblan: PHNE_28143 Jun 13 2003 14:16:53 lanc.c: PHNE_28143 Jun 13 2003 14:16:53 lan_t2.c: PHNE_28143 Jun 13 2003 14:17:03 lan_dlpi.c: PHNE_28143 Jun 13 2003 14:18:00 $Source: kern/sio/lan3.c, hpuxsysio, vvos_rose, rose 0313 $ $Date: 03/06/12 08:39:03 $ $Revision: 1.8.1.10 PATCH_11.04 (PHNE_29267) $ lan2.c: PHNE_22353 Date:Jun 13 2003 14:18:44 /usr/conf/lib/lan_dbg0.o: lan_dbg0.c: PHNE_28143 DATE:Nov 7 2002 14:18:00 Networking.LAN-RUN,fr=B.11.04,fa=HP-UX_B.11.04_32/64,v=HP: /usr/sbin/linkloop: $Revision: Hewlett-Packard ISSL Level vvos_rose42 $ $Header: Hewlett-Packard ISSL Release vvos_r ose $ $Date: Fri Jun 13 13:07:13 EDT 2003 $ $Source: net/lanlink/LAN/linkloop/linkloop.c, hpuxcm dnet, vvos_rose, rose0306 $ $Date: 99/06/08 10:49:49 $ $Revision: 1.22 PATCH_11.04 (PHNE _18560) $ LAN: Version: B.11.00 $Date: 97/10/15 10:44:26 $ /usr/sbin/lanscan: lanscan.c: PHNE_22331 LAN: Version: B.11.00 $Date: 97/10/15 10:44:26 $ vmmap.o $Revision: 82.2 $ /usr/sbin/lanadmin: $Revision: Hewlett-Packard ISSL Level vvos_rose42 $ $Header: Hewlett-Packard ISSL Release vvos_r ose $ $Date: Fri Jun 13 13:07:13 EDT 2003 $ $Source: net/lanlink/LAN/lanadmin/lanadmin.c, hpuxcm dnet, vvos_rose, rose0313 $ $Date: 03/06/12 08:54:08 $ $Revision: 1.25 PATCH_11.04 (PHNE _29267) $ LAN: Version: B.11.00 $Date: 97/10/15 10:44:26 $ lanadmin_mg.c: PHNE_15969 98/09/15 /usr/sbin/landiag: $Revision: Hewlett-Packard ISSL Level vvos_rose42 $ $Header: Hewlett-Packard ISSL Release vvos_r ose $ $Date: Fri Jun 13 13:07:13 EDT 2003 $ $Source: net/lanlink/LAN/lanadmin/lanadmin.c, hpuxcm dnet, vvos_rose, rose0313 $ $Date: 03/06/12 08:54:08 $ $Revision: 1.25 PATCH_11.04 (PHNE _29267) $ LAN: Version: B.11.00 $Date: 97/10/15 10:44:26 $ lanadmin_mg.c: PHNE_15969 98/09/15 /usr/lib/lanadmin/libdslan2.1: dslan2.c for PHNE_18409 Mar 6 2003 04:04:37 /usr/lib/lanadmin/libdslan3.1: dslan3.c for PHNE_18409 Mar 6 2003 04:04:39 /usr/lib/nls/msg/C/lanadmin.cat: None /usr/lib/nls/msg/C/lanscan.cat: None /sbin/init.d/hpether: hpether: PHNE_22331 $Revision: 1.4.119.1 $ $Date: 96 /10/01 12:11:36 $ Networking.NW-ENG-A-MAN,fr=B.11.04,fa=HP-UX_B.11.04_32/64, v=HP: /usr/share/man/man1m.Z/lanadmin.1m: None /usr/share/man/man1m.Z/lanscan.1m: None cksum(1) Output: Networking.LAN2-KRN,fr=B.11.04,fa=HP-UX_B.11.04_64,v=HP: 2120775872 4094 /usr/conf/master.d/lan 514661477 723992 /usr/conf/lib/liblan.a 888823540 306144 /usr/conf/lib/lan_dbg0.o Networking.LAN2-KRN,fr=B.11.04,fa=HP-UX_B.11.04_32,v=HP: 2120775872 4094 /usr/conf/master.d/lan 2204689312 389020 /usr/conf/lib/liblan.a 2130853628 246524 /usr/conf/lib/lan_dbg0.o Networking.LAN-RUN,fr=B.11.04,fa=HP-UX_B.11.04_32/64,v=HP: 4188434844 28672 /usr/sbin/linkloop 32426244 40960 /usr/sbin/lanscan 758894702 53248 /usr/sbin/lanadmin 758894702 53248 /usr/sbin/landiag 3826492397 16384 /usr/lib/lanadmin/libdslan2.1 153275624 16384 /usr/lib/lanadmin/libdslan3.1 2340268250 6631 /usr/lib/nls/msg/C/lanadmin.cat 595530027 2173 /usr/lib/nls/msg/C/lanscan.cat 1108542347 5272 /sbin/init.d/hpether Networking.NW-ENG-A-MAN,fr=B.11.04,fa=HP-UX_B.11.04_32/64, v=HP: 813392166 5554 /usr/share/man/man1m.Z/lanadmin.1m 2207105967 1909 /usr/share/man/man1m.Z/lanscan.1m Patch Conflicts: None Patch Dependencies: s700: 11.04: PHNE_18347 s800: 11.04: PHNE_18347 Hardware Dependencies: None Other Dependencies: None Supersedes: PHNE_18560 PHNE_22825 PHNE_25076 Equivalent Patches: PHNE_28143: s700: 11.00 s800: 11.00 Patch Package Size: 800 KBytes Installation Instructions: Please review all instructions and the Hewlett-Packard SupportLine User Guide or your Hewlett-Packard support terms and conditions for precautions, scope of license, restrictions, and, limitation of liability and warranties, before installing this patch. ------------------------------------------------------------ 1. Back up your system before installing a patch. 2. Login as root. 3. Copy the patch to the /tmp directory. 4. Move to the /tmp directory and unshar the patch: cd /tmp sh PHNE_29267 5. Run swinstall to install the patch: swinstall -x autoreboot=true -x patch_match_target=true \ -s /tmp/PHNE_29267.depot By default swinstall will archive the original software in /var/adm/sw/save/PHNE_29267. If you do not wish to retain a copy of the original software, include the patch_save_files option in the swinstall command above: -x patch_save_files=false WARNING: If patch_save_files is false when a patch is installed, the patch cannot be deinstalled. Please be careful when using this feature. For future reference, the contents of the PHNE_29267.text file is available in the product readme: swlist -l product -a readme -d @ /tmp/PHNE_29267.depot To put this patch on a magnetic tape and install from the tape drive, use the command: dd if=/tmp/PHNE_29267.depot of=/dev/rmt/0m bs=2k Special Installation Instructions: None