Download Fix File

When a SSL session is established a handle is allocated and chained together with other allocated handles. This process could cause an abend when multiple SSL sessions are initiated. This zap corrects this problem by using a compare and swap to chain these handles together. A closed session is also marked for reuse and not de-chained to avoid the possibility of a corrupted SSL handle chain.

New statistical counters will be displayed when in diagnostic mode.

SSL120 Stats will contain the following counters:

A=Number of SINI successful TLS30 client sessions started

B=Number of SINI successful SSL30 client sessions started

C=Number of SINI successful server sessions started

D=Number of GHAN handles reused

E=Number of GHAN handles allocated

K=Number of RESM successful server fast resume sessions

SSL121 Stats will contain the following counters:

F=Number of FHAN handles freed because SINI failed

G=Number of CMEM handles marked free

H=Number of CMEM count of handles

I=Number of SCLS requests successfully completed

J=Number of SCLS requests failed

M=Maximum number of cached sessions allowed

In addition this fix removes the multiple initialization calls to the IBM IJBHCDRV for the crypto-express2/3 coprocessor during an applications call to the gsk_initialize(IPCRINIT) function. It is now only performed once during the first application call to gsk_initialize(IPCRINIT).

Download Fix File

This zap adds 2 new message skeletons that are required for new statistical counters messages to be issued by IPCRYPTO in the corequiste ZP15F004 zap.

Download Fix File

This zap corrects the UNKNOWN in the SEE118I message to identify the VSE/AF version that is running at the site.

Download Fix File

This zap re-issues zap 014 as a phase replacement to correct a possible abend and display the zap number in the Q VER output.

Download Fix File

A Web page can have variables, such as "&IP" to insert values that are always changing. The cleanup calculation was in error, and so an extra "x" bytes (where "x" is the length of the variable name) would be included at the end of the display.

This fix corrects the problem.

Download Fix File

The DNS client sends an upp-case request to the DNS server. For some servers, there is a new feature to respond with a CNAME record (showing the matching lower case name) followed by the actual data. This insertion of an extra halfword response (X'0005'=CNAME) caused a logical mismatch and cause the GetHostByName to fail.

This fix adds logic to notice a CNAME record and skip over it in order to point to the desired response.

You will need to recycle your stack after applying.

Download Fix File

The above diagnostic error can occur when a certificate contains a RSA-SHA256 signature. This zap allows the certificate catalog utility to successfully process a certificate containing a RSA-SHA256 signature. The following message description will be displayed on syslst when this condition is detected:

x509v3 certificate contains a 2048-bit RSASHA-256 signed

Download Fix File

The above message can occur when no more 31-bit partition getvis is available. This zap attempts to detect and correct for this problem by temporarily shutting off traffic until the system can process the queued up datagrams and then turning traffic back on.

You will need to recycle the stack after applying this fix for the update to take effect.

Download Fix File

ZP15F433 caused the corrective code in ZP15F299 to be bypassed. This zap puts the ZP15F299 correction back. The ZP15F299 zap corrected a problem when using the external message log file(MSGXLOG) for recording the output of a directory with more than 1000 entries the number of queued messages may exceeed the maximum allowed by the message writer subtask. This zap causes the message queuing routine to check the maximum number of queued messages, and wait for the queue to be processed.

Download Fix File

LE/C-based applications load the $EDCTCPV phase to use BSD processing. Corrections have been added to the BSD code and this fix incorporates them into the LE/C interface.

For additional information, see fix ZP15F438.

Download Fix File

When an external connection request to a passive open cannot be completed due to a retransmission time-out on the outbound SYN, cleanup processing may encounter an error.

This fix corrects the problem.

Download Fix File

With the release of BIM-EDIT 5.7 the returned record of a member entry was changed, and so the date was moved over by 2 bytes and the size in bytes was moved over by one byte. The result was that a DIRLIST from any client would return the right names, but odd looking dates and sizes.

This fix corrects the problem by varying the offsets, depending on the version of the record returned.

Download Fix File

After entering "DEFINE CGI,PUBLIC=name" the message shows that it defined the control block, but it never actually loads the user written CGI (the type that uses the FILEIOHD macro) into storage, and the "BEGIN" routine is never performed.

This fix corrects that problem. The user written TYPE=CGI will now be loaded into storage when it is defined.

Download Fix File

This zap in conjunction with ZP15F099 allows the syslst writing of messsages by $SOCKLST to space 1 line before the write immediate to avoid an overlayed print line. A custom $SOCKDBG.PHASE must be generated using the LSTCCW=0B to cause the space 1 line immediate. The LSTCCW=09 is the default setting for the SOCKDBG.A macro.

Download Fix File

The error message was vague, so the output messages have been updated.

This fix corrects the problem. No recycling of the stack is needed.

Download Fix File

The above diagnostic error can occur when a certificate contains a RSA-SHA256 signature. This zap adds the object identifier for this structure to be successfully processed.

Download Fix File

This zap corrects a problem with ftp using double byte character translation. A abend may occur when the DBCS translation fails. A FTP906 diagnostic error message with internal debugging information should be produced when this abend condition is detected.

This zap also corrects a problem with re-entrancy with multiple transfers failing to successfully complete.

Download Fix File

When an email address is used without an ending "@" followed by a domain name, one is generted for the user. There are two places where this is obtains, from the EMAIL DEST= command in the initialization of the stack, or from the SET DEFAULT_DOMAIN= from the same initialization command. Each of these serves different purposes.

The DEFAULT_DOMAIN was always used, and if not defined, then the EMAIL DEST= would be used. The proper behavior should be that the EMAIL DEST= should always be used first for EMAIL, and if not set, then the other will be used as an alternate container.

This fix corrects the problem of the stated default not being correctly used.

Download Fix File

When processing the VERCHECK command the input is a phase containing the table of phase names along with the corresponding SHA-1 value. If a phase name in the table cannot be located in the libdef search chain the program would immediately terminate. Although this is a significant error we should attempt to continue to check the next phase in the table. This zap will issue a message to syslst about the not found phase and continue processing the next phase in the table.

This zap also corrects an error when attempting to perform the first SHA of the phase and the crypto environment was unable to initialize the following cryptic error messages would issued:

SSL103I CIALSHPH failed RC=FFFFF3E4(MAINSHAF)

This zap changes this to make it clear to understand that the failure was due to the crypto environment failing to initialize.

For CSI licensed sites you must have a valid product key for the SSL feature product number 107.

Download Fix File

When SET DISP=RESPECT is used, it should delete a DISP=D entry and PHOLD any DISP=K entry when complete. The DISP=K entry was being deleted instead of held.

This fix corrects the problem.

Download Fix File

The DIAGNOSE PERFORM perform command can be used to gather statistics when a connection terminates but the IPT352D message contained a incorrect value for the maximum and last roundtrip times similiar to the below sample message. This zap corrects this problem.

IPT352D Roundtrip: Min: 6 ms; Max: 2147483647 ms; Last: 2147483647 ms

Download Fix File

This zap changes the Connect_Sequence On to allow the IP address pattern to be either the full 4 byte IP address match or partial matches of 3, 2, or 1 octets of the IP address specified.

Download Fix File

Extra dumps were occurring that should have only been performed with DIAG=CLEANUP active. This zap corrects this problem.

Download Fix File

When ZP15F050 or ZP15F052 have been applied against CLIENTD, the DEFINE EVENT will change "ORDER=YES" to be based on the order that the entries were added to the queue as stated in the documentation rather than the job number/segment sequence.

In order to permit the behavior to sort based on the jobnumber and segment number order, this fix will add the "ORDER=JOBNUMBER" parameter to the DEFINE EVENT command. Apply this fix if you require that behavior.

You will need to recycle the stack after applying this fix.

Download Fix File

This fix will correct the following 5 problems:

(1) The default SORTING was based on the job number and segment number rather than the "in the order that it arrived". In most cases, the former will normally match the latter, but not always. This has been corrected. For those who prefer the previous behavior, ZP15F051 will need to be applied, permitting the "ORDER=JOBNUMBER" parameter to be added to the DEFINE EVENT.

(2) Some sites reported that AutoLPR was having a SETVAR problem. This would appear to cause blanks to be embedded in the command. This too has been corrected.

(3) The default behavior of AutoFTP is to always look in the DEST field for the script name, and if one is not there, to then look in the field where the DEFINE EVENT has pointed. Because of a quirk in processing data that was created via an FTP PUT, the DEST name of "SYSTCPIP" will be treated as a non-set field.

(4) In some environments where storage compression is in use by 3rd party software, "&value" commands are corrupted. We get around this problem by not maintaining tables of options with "&".

(5) For SPRI users, an assortment of fields are not set correctly, such as the shared spooling from a GETFLD call, or the queue entry QNUM value. This caused a looping of the automated client where SINGLE=YES was not in effect. Defining the event with POWERSYSID=SPRI will cause a different logic flow to eliminate these false errors.

You will need to recycle your stack after applying this fix.

Download Fix File

The IPN230I message did not have a space that makes it difficult to read.

Download Fix File

Some sites have the experience of strings with "&" being truncated, and when obtaining the member header, the command has a "&", and is has a problem returning the header intact. This can occur for older releases of BIM-EDIT that use the current interface.

This fix corrects the problem by eliminating the prefixing problem of the "&". It also checks the version byte for older versions to make sure that the correct offest is used.

To eliminate the version problem, IPNFBIME is being distributed as a .OBJ file and will need to be relinked with the correct BIM-EDIT library to pull in the user's most current API version.

Download Fix File

On a SOCKET CLOSE, a secondary ECB would be posted, but if the CPU if very fast, it is possible that the control block will be gone before the POST takes place. Since the secondary ECB is never referenced in this module, the simple solution was to remove it from the SOCKET macro invoking the CLOSE.

When linkedited into the application, the object deck provided by this fix will correct the problem.

Download Fix File

Variables, such as "SET COPIES=&PWRCOPY" will fail because the variable "&PWRCOPY" is not resolved with the actual script. SET PAUSE would abend because of a bad numeric conversion.

This fix corrects the problems. No recycling of the stack is needed.

Download Fix File

The name of the TCP/IP initialization member is contained in the jcl PARM='INIT=xxxxxxxx' and if it was less than 8 characters the next parameter would be included into the member name causing the IPN136W error message. This zap corrects this error.

Download Fix File

This zap corrects a problem when a single partition with multiple subtasks issue the first socket request at the same time. The second subtask may overwrite the job step start time causing socket requests from the first subtask to be rejected. This zap corrects along with its prerequisite zaps corrects this problem.

Download Fix File

The automated client (CLIENTD) was not correctly using the defined SYSID correctly and was therefore always defaulting to SYSID=00.

This fix corrects that problem. There is no need to recycle the stack. If you are running AUTOSEND, then recycle AUTOSEND.

Download Fix File

When running XACLIENT, if a sending to a server failed, and DISP=Y was attempted, IPNFPOWR would branch into obsolete code that would result in an addressing exception.

This fix corrects that problem. No recycling of the stack is required.

Download Fix File

This zap implements RFC5746 to allow the usage of TLS extensions for the renegotiation security exposure.

Download Fix File

When an OPEN of a Power/VSE member fails, the cleanup routine swapped the address/length fields, and so the FREEVIS fails.

This fix corrects that problem. Recycling of the stack is required.

Download Fix File

SPRI has problems with queued requests disappearing.

SPRI does not correctly provide a unique QNUM value.

This fix will eliminate QNUM checking for SPRI users. This may cause a problem if hundreds of thousands of reports occur during the same day, but only with one report, which will work on a retry.

This only affects SPRRI users. DEFINE EVENT POWERSYSID=SPRI should be set for this to work. You will need to recycle your stack after applying this fix.

Download Fix File

The attached zap adds an additional diagnostic message for when a ftpbatch LOPEN fails.

This zap also adds a new command to allow a longer timeout value for waiting for a response from a foreign ftp daemon. The default timeout is 60 seconds but can now be longer by issuing the SET FTPCFTIM NNNNN command from a ftpbatch job. For the SET FTPCFTIM to be effective in must be issued before the LOPEN or placed in the FTPBATCH.L book.

NNNNN is the amount of time in 1/300ths of a second. The default is 18000 which is equal to 60 seconds(18000/300=60). 36000 would be equal to 120 seconds.

Upon successful completion the following message will be displayed:

IPA409I Foreign timeout set to 36000

Download Fix File

A requirement was provided that, due to behavioral changes, the default LRECL for POWER RDR output needs to be adjustable.

This fix will check if a DEFINE FILE,TYPE=POWER was also defined with "LRECL=value", and if so, THAT will be the default LRECL unless one was provided by the client (e.g., SITE LRECL will still be the override.

Apply this fix only if you need this new feature. Recycling of the stack is required.

Download Fix File

When SET TO= (or any variant) is used to deliver an email, the client was not checking the response code (2xx, 5xx) from the server, and so if it was rejected by the server, it would act as though it were ok.

This fix corrects the problem. No recycling of the stack is needed.

Download Fix File

The "PRINT /" command format was not inserting the segment number in the generated "PRINT" command, and so the first segment displayed in a PDISPLAY would be repeatedly sent if "DISP=KEEP" was in effect.

In order to eliminate this issue, the ".nnn" value has been appended to the Power report to be printed, where "nnn" is "000" by default, or the vaule in a PDISPLAY shown as "S=nnn".

In addition, the execution parameter now accepts a new command, "RETRIES=n", where "n" is a value from 0-9 (0 is the default). If a "PRINT" command failed, then it will wait 5 seconds and retry "n" times. This will help those sites where the printer sometimes needs a bit of time to "cycle down".

This fix corrects the problem. No recycling of the stack is needed.

Download Fix File

When defining an event with the new "ORDER=JOB", it would show "ORDER=YES" instead of "ORDER=JOB". (ORDER=JOB was added with ZP15F051. If you did not apply that fix, you need not apply this one.)

This fix will correct the problem. You will need to recycle the stack for it to take effect.

Download Fix File

When a Telnet menu is defined, storage will be allocated for it. When there are variables, longer strings will need to be inserted. If there is not enough room to insert the entire string, TEL952E will occur.

This fix corrects the storage allocation algorithm for menus.

You will need to recycle the stack after applying this fix.

Download Fix File

A bug only would accept 2-byte commands, ON or NO.

This fix corrects the problems. No recycling of the stack is needed.

Download Fix File

If AutoFTP processed a script and "Error:" was returned from the internal client (due to a failed or bad command), approximately 512 bytes of storage would not be recovered from 31-bit GETVIS due to the way that "QUIT" is handled within a script. This has been corrected.

While processing AutoLPR across multiple printers, if one printer ran out of paper with "SINGLE=YES", then the other printers would stop having output delivered to them as well. This has been corrected, but you should note that the MAX_LPR_EVENTS value should be raised significantly, since this backup will be keyed against it. (If you are sending output to 5 printers and one is backed up with 4 more reports, and MAX_LPR_EVENTS is set to 5, it will still hang, since there are 5 entries queued.)

This fix will correct those problems. You will need to recycle the stack for them to take effect.

Download Fix File

This zap provides a new function to allow a application to set the TCP/IP sysid to another value.

Download Fix File

The attached zap add a new function for setting the TCP/IP sysid.

After a failed SSL handshake the handle we being freevisd. but the handle was a cached entry being reused and should not be freevisd. because it was not being dechained leaving a corrupted SSL handle chain. This could cause an abend during the next SSL session negotiation. This zap will test for a reused(cached) session handle and mark it free for reuse instead of freevis'ing it.

Download Fix File

The ccblok lookup routine had the DIAG HASHING option on and caused a lot of IPN694 messages that should only be issued with DIAG HASHING active. This zap corrects this problem.

Download Fix File

An X'35' byte (TRN) is followed by a length and then data that is not to be translates. This arrives from VTAM. If the user data contains an X'35' accidentally, then the next byte will cause internal logical problems in forming the data to be delivered.

This fix corrects the problem by examining if ANY length causes a jump beyond the buffer or a negative value, and if so, it will terminate the report prematurely. Diagnostics have been enhanced to report such an issue. Recycling of the stack after applying this is recomended for it to take effect.

Download Fix File

During the startup up of a ftpbatch a failure of the EXTRACT SVA as described in ZP15F085 attempted to issue a message before the version of VSE had been set. This caused the x00 version to cause the console write routine to assume it was older that VSE/ESA 2.4 and the CCW area for it was not valid causing an invalid write to the console. This zap corrects this problem.

Download Fix File

The EXTRACT SVA request issued by ftpbatch can fail with a return code of x04 when the system getvis gate is closed. This causes the ftpbatch job to abruptly terminate. This zap allows ftpbatch to continue to run after this extract failure since the sva information is not used for anything.

Download Fix File

The "PRINT /" command format was not setting the RC if any of the reports in the group failed to be delivered, and it would continue to deliver the rest of them (which would arrive out of order).

This fix sets the RC to >0 condition and will stop when any report in the group fails to be delivered.

This fix has also added a new command to the print line: /CDISP. The valid values are "D" (default) or "K". If you use "FREE" as your job name, then "K" and "D" will both be processed, and the /CDISP parm will be ignored. DISP=* entries are always ignored.

Examples: PRINT FREE/CLASS=Q - Will print DISP=D and DISP=K for CLASS=Q PRINT ALL/CLASS=Q/CDISP=K - Will print DISP=K entries for CLASS=Q PRINT ALL/CLASS=Q - Will print DISP=D entries for CLASS=Q PRINT ALL/CLASS=Q/CDISP=D - Will print DISP=D entries for CLASS=Q PRINT FREE/CLASS=Q/CDISP=D - Will print DISP=D and K entries (CL=Q)

No recycling of the stack is needed.

Download Fix File

When DISP=RESPECT was in place, all entries would be deleted, including DISP=K. This fix will cause DISP=D to be deleted, DISP=K to become DISP=L, and it will ignore all other DISP values, acting like a DISP=KEEP

If RETIRES=n is in the CLIENT execution parameter, or if SET RETRIES=n is in the input stream, RETRIES will be properly obeyed. The value in "SET PAUSE=nn" will determine the delay between each retry.

DELAY=nnn can be put in the execution parm. It will cause a delay of "nnn" seconds to be inserted after each "PRINT/" generation.

This fix corrects the problems. No recycling of the stack is needed.

Download Fix File

If the script did not contain "QUIT", AutoFTP should force one. However, due to a pointer problem in that logic, it sends "READ" beacause "READY" was in the last receive buffer, causing hang issues.

PAGEDEF variable was incorrectly set to FORMDEF and bot were set incorrectly to 8 bytes long rather than 6 bytes long. This has been corrected.

This fix corrects the problem. You will need to recycle the stack after applying this fix.

Download Fix File

When a GET/PUT has a variable (SETVAR generated) that would extend beyond column 72, it would not convert the variable to a string, leaving it as "&FIL1" instead of "long_file_name...".

The logic has been modified to permit 125-byte commands.

You need not recycle the stack after applying this fix.

Download Fix File

If CICS was recycled with a cold restart and a terminal was in session, the initial report will not be delivered unless GPSD is deleted and then redefined.

Added logic to cause the ACB to close and reopen if this occurs. Recycling of the stack is recommended after applying this fix.

Download Fix File

There are seven conditions, failures, that if they occur, GPSD will simply terminate without specifying why. This was seen when an attempt to start up a DIRECT printer, that GPSD would terminate without providing any specifics.

For this particular user, a bad WCC would cause GPSD to shutdown without warning. It will no longer do that.

A series of new messages have been added to indicate the reason for the failure. This should provide the details needed to debug the problem. For this instance, the bad WCC is not considered an error, but a reason to release the report.

Download Fix File

If an external application issues a socket request in 24-bit mode an addressing exception may occur in the ASOCKET routine.

A CICS socket application may get an addressing exception in the ASOCKET routine.

This zap should correct these problems.

Download Fix File

When multiple "SET ATTACH" commands are used, and the last "ATTACH" was positioned in a non-POWER file-type directory (such as VSAMCAT), then the pointer to the file-type control block is in error. And if the "SET DISP=" value is anything other than KEEP, IPNACMEM will abend.

This fix corrects the problem. No recycling of the stack is needed.

Download Fix File

For the AUTOLPR portion of the problem, the length field was extended so that the entire command will be automatically generated and sent.

For the "hang" portion due to external interference, the code will now count the number of blocks that are waiting to run or are running and will ignore the "waiting to be removed" entries, which will eliminate that "hung" condition ("waiting to be removed" will never be serviced if the entry changed with an external PDELETE/PALTER. So while that small amount of storage will always be held, the processing will not stop.

This fix corrects those problems. You will need to recycle the stack after applying this fix.

Download Fix File

When using NOEJECT=YES, the Email program will strip off the first byte of the record, whether it is a real CC (control character) or not. This causes strange results for those who use generic scripts.

This fix will check if the first byte is a number, blank, "-", "+" or "-", and if not, it will disable the NOEJECT setting from being used. It is important that the COREQ fixes be applied, or the "EMAIL" command may ABEND.

Download Fix File

When using external partition getvis for socket requests a checksum is performed on the data area for socket requests before it is passed to the TCP/IP partition. Before TCP/IP starts processing the socket request the checksum is validated and if it has changed then it indicates that the application has changed the data in the socket request before the socket request processing has been posted complete. Applications should not change the passed data area until the socket request is posted complete. This zap adds dumps of the socket request block to help identify the application that needs to correct this problem.

Download Fix File

There are two problems here. The first was that the failure logic was not correctly setting a bad return code. The second is that the POWER file I/O driver was still usin a lot of 24-bit GETVIS. With the advent of z/VSE 5.x, the amount of free 24-bit storage was lessened, so it became more apparent to those who upgraded to the more recent VSE.

In the case of the stack loop, the correct return code is now set, and stack will no longer go into a loop if a GETVIS call fails.

The Power file I/O driver has been made AMODE=31,RMODE=ANY. This will reduce the heaviness of the GETVIS24 storage use. A QUERY FILEIO will show that it now resides in 31-bit (about 40k), and the use of SET DIAG=POWER will show on SYSLST that all of the buffers are also in 31-bit storage (e.g., about 167k for a simple "CD" command).

This fix corrects those issues. Recycling of the stack is required.

Download Fix File

CLIENTD, the program that performs the DEFINE EVENT had a bug in the routine that would determine if there was room in the queue of subtasks to add another. In some cases it would go into a perpetual loop, causing the stack to crash, and in others, the calculation was flawed, causing the count to be off by one, and in a single-threading environment, that would cause all event processing to stop.

The routine that caused those problems has been corrected it appears that this error was introduced in ZP15F094. If you have that zap applied, you should apply this fix.

Other unrelated changes include an additional line of information resulting from a "Q EVENT,DETAIL" command. SET DIAG=AUTO will now only show EVENT debugging. If you want to also see FTP, EMAIL, or LPR processing, you will now have to enable those debugging features as well (e.g., SET DIAG=FTP, etc.). This will reduce SYSLST output when trying to debug a generic EVENT problem.

Improved the DIAG AUTO information to see the names that are queued. Multi-threading "ready" posting for improved output (less delay between batches). Special SPRI logic added. Improved and cleaned up debugging messages. If SET SINGLEDEST=ON is enabled, then entries with matching script names will wait in line, one at a time.

Q EVENT,DETAIL is automatically issued when the event processor is being terminated. Either because of a DELETE or a SHUTDOWN. If Q EVENT,DETAIL is entered, more information is provided to debug general settings and the script name is listed next to assist in seeing "hangs".

The date/time of the generation of CLIENTD will be displayed in the detailed query.

The FTP client portion was fixed so that the QUIT will be correctly detected and the automatic "QUIT" would not be sent a second time.

You will need to recycle your stack after applying this fix.

Download Fix File

The default write to syslst is a x09 which is a write immediate without spacing and this can cause of overlay of the previously written line to syslst. This correction add a new LSTCCW= keyword that can be set to LSTCCW=OB to cause a spacing of 1 line before writing messages by the $SOCKLST program. ZP15F031 must also be applied when using this setting.

Download Fix File

In order to debug why a SETIME/DROPTIME internal request will fail, and there is no error message to display why.

Added a warning message that explains the reason that will display if one of the two conditions exist.

You will need to recycle the stack after applying this fix

Download Fix File

On a SOCKET CLOSE, a secondary ECB would be posted, but if the CPU if very fast, it is possible that the control block will be gone before the POST takes place. Since the secondary ECB is never referenced in this module, the simple solution was to remove it from the SOCKET macro invoking the CLOSE.

ZP15G055 is the same fix as this one, except that this fix will create a .PHASE version of the same .OBJ fix for those Cobol programs that are compiled with a "DYNAM" option, which requires a PHASE rather than an OBJ.

Download Fix File

A subtask attached in the CICS partition such as the IBM VSAM redirector an addressing abend can occur in the subtask due to the method used to determine that the socket request is from a CICS application. This zap corrects the problem.

Download Fix File

This appears to be a rare instance where a work register value "could" cause a 3-byte length to become a 4-byte incorrect length.

This fix corrects that problem. Recycling of the stack is required.

Download Fix File

When a subtask is attached in the CICS partition but does not run under the control of CICS we may incorrectly attempt to use CICS services that are not available. This zap corrects this problem.

Download Fix File

This zap passed back a new socket number evenn on a failed asynchronous accept.

This zap also removes a wait that could occur during close processing.

Download Fix File

The software implementation of the RSA algorithm will only support RSA 512-bit or 1024-bit key sizes. A new message:

SSL204 Hardware Crypto Express PCI required for RSA xxxx yyyy

xxxx the RSA bit size that failed and will be either 2048 or 4096. yyyy is the type of operation and will be either encrypt or decrypt.

It is recommended that the IBM crypto-express2 PCI hardware be used for RSA operations.

Download Fix File

The GETVENDORINFO call can be used to identify the stack vendor currently in use. Upon successful completion the CSI stack will return 10 character bytes as follows: CSI - 3 printable bytes C/I - 1 byte indicating a CSI or IBM licensed site(C=CSI, I=IBM) Vvrrmm - 6 bytes with the version, release, modification level of TCP/IP for VSE.

Download Fix File

The above diagnostic error can occur when a root certificate contains a RSA-SHA256 signature. This zap allows the certificate catalog utility to successfully process a certificate containing a RSA-SHA256 signature. The following message description will be displayed on syslst when this condition is detected:

x509v3 certificate contains a 2048-bit RSASHA-256 signed

Download Fix File

When using external partition getvis for socket requests a checksum is performed on the data area for socket requests before it is passed to the TCP/IP partition. Before TCP/IP starts processing the socket request the checksum is validated and if it has changed then it indicates that the application has changed the data in the socket request before the socket request processing has been posted complete. Applications should not change the passed data area until the socket request is posted complete. This zap adds dumps of the socket request block to help identify the application that needs to correct this problem.

This zap also removes unwanted diagnostic dumps that should now only be issued with DIAG SOCKET active.

Download Fix File

This zap sets a return code zero for asynchronous passive opens which can cause CICS Web Services to terminate during an accept function if a passive open terminates before the connection is fully established.

Download Fix File

This PORTQ command allows queing of incoming connection requests and is used by applications such as CICS Web Services to avoid rejected connections. This zap corrects a problem that caused the PORTQ facility to fail.

Download Fix File

The IPN411 message was not defined in the message skeleton file and is now added in this zap. This was part of a general messages clean up that corrects some inconsistent issues with messages from different programs using different substition and message types.

TCP909I message truncated under AUTOSEND. This zap was also issued as test zap 023, but now is included in this production zap(119). When AUTOSEND was producing TCP909I, the last byte would be truncated if the message was 65 bytes long. This is due to how the message processor works outside of the stack when dealing with real task identifiers.

The message was changed from "was completed" to "is complete" to have the message appear the same when under the stack or external to it.

You will need to recycle the stack for this updated message to appear. AUTOSEND will need to be recycled as well.

Download Fix File

The automated client (CLIENTD) would display message TCP905D to show the action taking place. The client type was being truncated to 4 characters. For FTP or LPR, that was not a problem. For EMAIL, it was.

This fix changes the length of the action field to 5 bytes. You will need to recyclt the stack and AUTOSEND after applying this fix.

Download Fix File

When a subtask is attached in the CICS partition but does not run under the control of CICS we may incorrectly attempt to use CICS services that are not available. This zap corrects this problem by linking in the IPNRSTUB.OBJ distributed with ZP15F107.

Download Fix File

If an external application issues a socket request in 24-bit mode an addressing exception may occur in the ASOCKET routine.

IPN960C SOCKET REQUEST REJECTED RS=BDATADDR error message may also be issued.

This zap also corrects a problem where a 24-bit application is returned to in 31-bit mode.

This zap should correct these problems.

Download Fix File

This zap will cause a IPN898I diagnostic message in the TCP/IP syslst and some diagnotics dumps.

Download Fix File

This zap corrects a reset of the BSD socket number related to problems with a asynch passive open failing in the accept function.

Download Fix File

This zap removes a givesocket error for a closed socket.

Download Fix File

The TELNET01 (CICS) application can be used to log onto other systems using the TELNET protocol (not TN3270). Part of that process includes using the SOCKET TELNET OPEN with NEGOT=YES which would pass a list of suitable telnet types. This feature stopped working in 15F and, as a result, the TELNET01 program, when trying to log onto VM, would not be usable, since a default of "line mode" would be the result.

This fix will correct the error in the ASOCKET program, which formats and passes on the OPEN request to the TELNET protocol negotiator (IPNATELN).

You will need to recycle the TCP/IP stack after applying this fix.

Download Fix File

After applying a fix to MSKELIP, which updates the message skeleton, the command processor would display IPN411I incorrectly. This is because the program that uses the message skseleton and the file that contains them were out of synch.

This fix will put CMDEXEC back into synch, and the messages will display correctly. You will need to recycle the stack after applying it.

Download Fix File

This zap corrects a problem with multiple stacks that may cause the shutdown of a stack to affect a application connected to a different stack id.

Download Fix File

This zap corrects a problem where the checking for a file type thah allows periods in the directory name may cause a change directory to fail for a file type that uses the period as a delimiter.

Download Fix File

This zap adds an event to diagnose the failure of a LCD.

Download Fix File

Commands from the console can be controlled with the SET PASSWORD command. But when using MSG xx,DATA=command the prompt for a password from the operator console was not being enforced. This zap corrects this problem although not many sites use a password for commands issued from the system console.

You will need to recycle the stack after applying this fix for the update to take effect.

Download Fix File

When a file create by GPSD is stored and transmitted to the local LPD, there is a chance that one byte deep into the report may contain a carry-over from a previous line due to a length error and the data not being cleared before being loaded.

This fix should correct that problem. You will need to recycle the stack after applying this fix.

Download Fix File

When using the EMAIL batch program, EMAIL.AUTOEXEC will be processed. If you do not want that library member to be processed, then the execution parameter "AUTOEXEC=OFF" (or "NO") should be set. That feature, however, was not working as designed, resulting in the EMAIL.AUTOEXEC being invoked in any case.

Comments in EMAIL.AUTOEXEC and SYSIPT will now be ignored.

This fix corrects that problem. No recycling of the stack is needed.