HP OpenVMS Systems Documentation
HP OpenVMS Alpha Version 7.3--2 Release Notes
5.12.2 Change in Linker Default Behavior with Library Check
Previously, the linker's check between the library and the shareable image was too sensitive. It compared against the exact date and time, signaling LINK-I-DATMISMCH, if no match was found. Now, however, it makes only the same check that the image activator does: that is, it uses the GSMATCH criteria to verify compatibility.
The old behavior (check for date and time) can be obtained by setting
the logical name LINK$SHR_DATE_CHECK.
Developers who are creating object files should be aware that the
linker's internal stack is guaranteed for only 25 elements. Any
calculations must be done within this constraint.
In releases prior to OpenVMS Version 6.1, LTDRIVER did not set the "extended DDT" bit; therefore, the POSIX function CANCEL SELECTIVE did not work with LTDRIVER. This problem has been corrected, but a restriction remains.
Although this fix allows $QIO reads and writes to be selectively
canceled, any $QIO done to the port driver (that is, with the
IO$_TTY_PORT function modifier --- such as a LAT connect $QIO)
cannot be canceled with CANCEL SELECTIVE.
OpenVMS callable mail routines are not thread-safe. Refer to the Guide to the POSIX Threads Library for more information about calling non-thread-safe routines within a threaded application.
Because callable mail context information is maintained on a per-process (rather than a per-thread) basis, multiple threads performing context-based processing must be synchronized so that only one mail context of a given type is active at once. Otherwise, one thread could corrupt another thread's mail operations.
On OpenVMS Alpha systems, there is an additional restriction when
kernel threads is enabled in a multithreaded environment. In this
environment, callable mail should be used only in the initial thread.
The following sections contain release notes pertaining to the POSIX Threads Library (formerly named DECthreads).
The following changes have been made in OpenVMS Version 7.3-2:
5.15.2 New Priority Adjustment Algorithm
As of OpenVMS Version 7.3-2, the adaptive thread scheduling behavior
that is described in the Guide to the POSIX Threads Library has been implemented with a new
priority adjustment algorithm. In some cases, the new algorithm should
help avoid problems that can arise when throughput-policy threads of
different priorities share synchronization objects. Priority adjustment
can also improve application throughput and overall system utilization.
Priority adjustment of threads with throughput scheduling policy is
automatic and transparent.
If the POSIX Threads Library detects an uncorrectable serious problem
at run time (such as data structures that have been damaged by data
corruption somewhere in the application), the library may terminate the
running image. During termination, the library may trigger creation of
a process dump file (which can subsequently be used to diagnose the
failure, by way of ANALYZE/PROCESS_DUMP). The size of such a process
dump file depends on the size of the process's address space at the
time of the failure and can be quite large.
Starting in OpenVMS Version 7.3, the POSIX Threads Library is sensitive to dynamic changes in the number of CPUs that are configured for a running multiprocessor Alpha system. When use of multiple kernel threads is enabled (by way of the LINK/THREADS_ENABLE qualifier or the THREADCP command verb) for an image, the POSIX Threads Library monitors the apparent parallelism of an application and creates multiple kernel threads up to the number of CPUs available. Each kernel thread can be scheduled by the OpenVMS executive to execute on a separate CPU and, therefore, can execute simultaneously.
While an application is running, an operator can stop or start a CPU. Such a dynamic change affects the allowable number of kernel threads that future image activations can create. It also will now affect images that are currently executing.
When a CPU is added or removed, the threads library will query for the
new number of active CPUs, and compare this to the number of kernel
threads that the process is currently using. If there are now more CPUs
than kernel threads, the library will try to spread out the existing
POSIX threads over the CPUs (creating new kernel threads as needed, now
or in the future). If there are now fewer CPUs than kernel threads, the
library will force the extra kernel threads to hibernate, and will
reschedule the POSIX threads onto the remaining kernel threads. This
will ensure that --- so far as the process is concerned --- there will
not be more kernel threads competing for CPU resources than are
The POSIX Threads Library provides enhanced data collection
capabilities to support monitoring and debugging tools. These
capabilities provide support for Visual Threads, a new debugging and
analysis tool for threaded programs on OpenVMS Alpha systems. Visual
Threads, which is licensed with OpenVMS Version 7.3, provides
monitoring, automatic debugging, and performance evaluation of
The metering capability of the POSIX Threads debugger does not work.
If you use the procedure to debug a running program that is described
in Section C.1.1 of the Guide to the POSIX Threads Library, your process could fail with an
This section contains release notes concerning privileged code and data
The method used for attaching a security profile to an I/O Request Packet (IRP) changed with Version 7.2.
In versions of OpenVMS prior to Version 7.2, the IRP structure contained the address of the processwide Access Rights Block (ARB) security structure of the requestor. Beginning with OpenVMS Alpha Version 7.2, the address of the new security profile structure (Persona Security Block, or PSB) was added to the IRP as a functional replacement of the ARB address.
The I/O subsystem maintains its access to the PSB through a reference counter within the PSB. The I/O subsystem increments this reference counter at the time of IRP creation and decrements the counter at I/O postprocessing of that IRP. When this counter reaches zero, the PSB structure is deallocated.
Device drivers that create or clone copies of IRPs to facilitate multiple I/O operations per request, and subsequently pass the copies to the I/O subsystem for postprocessing, must make code changes to account for the extra references to the PSB in these additional IRPs. This is done by passing the PSB address located in the copied IRP to the NSA_STD$REFERENCE_PSB routine. The include file and routine call for NSA_STD$REFERENCE_PSB is as follows:
Device drivers need to make this change under the following conditions:
Failure to call NSA_STD$REFERENCE_PSB in these circumstances will result in corrupt tracking information within the PSB, which can result in system failures.
If you make code changes in a device driver to call
NSA_STD$REFERENCE_PSB, you must recompile and relink the driver to run
on OpenVMS Version 7.2 or higher.
Several routines are used by privileged code to create OpenVMS fork execution threads. These routines run in system context independent of any process. There are four variations of these routines, depending on whether an immediate or queued fork is required and on which language interface is being used:
In OpenVMS V7.3-1, if SYSTEM_CHECK is set to 1, these routines check the system IPL at entry. If the IPL is below IPL$_RESCHED, the system will fail with an SPLINVIPL bugcheck.
For performance reasons, the IPL is not verified if SYSTEM_CHECK is set
to zero (the default). Incorrect code may cause the system to hang if a
reschedule to another CPU occurs during execution of these routines
from process context (for example, below IPL$_RESCHED).
The following notes pertain to LIB$FIND_IMAGE_SYMBOL.
In the OpenVMS RTL Library (LIB$) Manual, there is an error in the description of the flags argument to LIB$FIND_IMAGE_SYMBOL. Flags is documented as passed by reference. This is incorrect and returns an error message, LIB-F-INVARG, as a return value. If flags is passed by value, LIB$FIND_IMAGE_SYMBOL works as expected.
This error will be corrected in the next revision of the OpenVMS RTL Library (LIB$) Manual.
LIB$FIND_IMAGE_SYMBOL may signal a warning (LIB$_EOMWARN) to indicate that the image being activated contains modules that had compilation warnings. A condition handler used with LIB$FIND_IMAGE_SYMBOL should probably handle this as a special case.
To allow LIB$FIND_IMAGE_SYMBOL to continue execution after signaling
LIB$_EOMWARN, the condition handler should exit with SS$_CONTINUE. For
this reason, you may choose not to use LIB$SIG_TO_RET as a condition
handler for LIB$FIND_IMAGE_SYMBOL.
The following information should be added to topics in the reference section at the end of the OpenVMS RTL Screen Management (SMG$) Manual:
5.19 SORT32 Utility
The following release notes concern the SORT32 utility. Note that SORT32 is recommended as a workaround for unresolved problems with Hypersort and for functionality not implemented in Hypersort.
SORT32 has been updated to V07-006 for OpenVMS Alpha Version 7.3-2.
The version of SORT32 that shipped with OpenVMS Alpha Versions 7.3 and
7.3-1 did not properly write VFC information to VFC output files when
specification files were used. This problem has been corrected.
SORT32 does not always delete temporary work files. It's a good idea to
periodically check SYS$SCRATCH or wherever you put SORT32 work files to
see if any undeleted work files can be deleted to free up disk space.
SORT32 does not issue a diagnostic for a compound condition in a key specification file not enclosed in parentheses, such as the following:
That condition should instead be specified as:
5.19.4 SORT32 and Hypersort Performance: Working Set and Page File Quota
SORT32 and Hypersort use different sorting and work file algorithms.
Either sort utility may be faster depending on the input file and the
memory/disk/CPU configuration. Make sure that working set extent is at
most one third of page file quota with either SORT32 or Hypersort.
SORT32 and Hypersort allocate fixed-sized slots for sort work files
based on the longest record length (LRL) information in the file. To
improve sort performance, try to set LRL information in the file as
close as possible to the actual longest record length. Poor initial
performance may be the result of sorting some files produced by C
programs, because the LRL is set higher than needed (to 32767).
SORT32 work files must be redirected to directories that allow multiple
file versions that can handle the number of requested work files. This
restriction also exists in Hypersort.
The functional changes to SYS$ACM[W] described in this section are introduced in OpenVMS Version 7.3-2. In these descriptions, nonprivileged processes refer to processes running in user mode that do not have SECURITY privilege.
Refer to the SYS$ACM[W] system service description in HP OpenVMS System Services Reference Manual for
Management of Timer Queue Entries was redesigned for OpenVMS Alpha Version 7.3-1 to provide significantly higher performance for systems using many TQEs. This change is transparent to nonprivileged applications.
Privileged code can no longer manipulate TQEs directly in any form. In particular, directly accessing pointers in the TQE's queue header (TQE$L_TQFL/TQE$L_TQBL) causes an access violation in almost all cases. Privileged code may continue to use the internal routines exe_std$instimq/exe$instimq and exe_std$rmvtimq/exe$rmvtimq to enter or remove Timer Queue Entries.