Linux x86 Problems Fixed
The following problems have been fixed. More details about each item follow this list. Items new with this release are in bold.
Attaching problem fixed
Previously, attaching to a running process could cause the following fatal error:
TotalView: FATAL ERROR STARTING UP: LID for the main executable is
null
This occurred if you had recompiled the process while a copy was still active or if the proc file system had mistakenly marked it as deleted. (4656: 6.1.0-3)
CodeRoad JNI Bridge attaching problem fixed
The CodeRoad JNI Bridge no longer fails to attach to the Java Virtual Machine when running under NetBeans. This occurred when NetBeans was installed in a deep directory tree. (4741: 6.2.0-3)
configure_autostart licensing problem fixed
The configure_autostart installation script no longer fails to set up the necessary hooks to start the license manager daemon in the commonly-used runlevel 5 (graphical log in). (4611: 6.1.0-3)
File > Preferences > Formatting: Defaults Command No Longer Crashes TotalView
(4473)
GCC -gstabs compiler option support is improved
Support for executables compiled with the GNU GCC compiler with the -gstabs compilation option has been improved. (4088, 4089)
GCC 3.1 and 3.2: Opaque Type Problem When Base Class in a Different Shared Library Problem Resolved
Sometimes, when diving into a class whose base class was in a different shared library, TotalView would show the base class as an opaque type. (4466)
Lahey/Fujitsu Fortran: Assembly shown by mistake
Previously, assembly code was displayed instead of the source file when diving on F90 modules, even when the module has executable statements. (4374: 6.1.0-2)
Lahey/Fujitsu Fortran: Showing wrong source
Previously, TotalView could show source or assembly code for the wrong file when you dove or used View > Lookup Function on a Fortran 90 module. This occurred when diving on modules that contained only data (that is, it did not have any executable statements) (4158: 6.1.0-2)
Lahey/Fujitsu Fortran: Stepping problems fixed
Problems with stepping certain Lahey/Fujitsu Fortran 95 programs no longer occur. (3658: 6.2.0-3)
Miscounting processors
TotalView was miscounting processors on Hyperthreading Xeons running Red Hat 8.0. (4564: 6.2.0-0)
printf() problems
On Red Hat 7.3 with either the default packages or the updated i686 packages, calling printf() in an eval point crashed the target. (4203: 6.2.0-0)
RMS prun and Message Queue
Error messages related to and problems displaying the message queue when debugging RMS prun programs no longer occur. (4790: 6.2.0-3)
Watchpoint warning message suppressed
TotalView no longer displays the following message when you set watchpoints on long long values:
Attempt to add wrong child type to a homogeneous RowColumn widget
(4546)
