Two state variables and their related command line options enable you to filter dlopen events to plant breakpoints in the dlopened libraries only when the process stops for some other reason.
dlopen event filtering is controlled by the settings on two state variables, TV::dlopen_always_recalculate and TV::dlopen_recalculate_on_match, and their related command line options dlopen_always_recalculate and dlopen_recalculate_on_match
Three possible dlopen filtering modes are made possible by these variables: Slow, Medium and Fast.
In Fast mode, the process never stops for a dlopen event, not even "null" dlopen events. Using this option can result in significant performance gains, but may be impractical for some applications.In Medium mode, some libraries can be specified to always reevaluate their breakpoints, rather than all or none.
• Slow Mode: Reloads libraries on every dlopen event
Option:
dlopen_always_recalculate==true
Reloads libraries on every dlopen event, retaining TotalView’s traditional breakpoint reevaluation semantics. This mode is compatible with CUDA and is a good choice when your session has pending breakpoints. However, this mode does not perform or scale as well as the other modes, because it requires the TotalView client to handle every (non-null) dlopen event for every process.
If performance is not the primary concern, or the application or runtime environment does not perform many dlopen events, then this may be a good choice.
In this mode, when the target stops with a dlopen event, the server reports the event to the client, where the library list is reloaded and checked to see if any additional breakpoint locations need to be planted in the newly loaded libraries
• Medium Mode: Reports only libraries that match defined patterns on a dlopen event
Options:
dlopen_always_recalculate==false
dlopen_recalculate_on_match=="glob-list"
A glob-list is a colon-separated list of simple glob patterns used to compare and match the dlopened library. A simple glob pattern is a string, optionally ending with asterisk character ('*') For example:
This mode strikes a balance between performance and enabling breakpoints to be planted in dlopened libraries.
In Medium mode, the target process stops on every dlopen event (just as in Slow mode), but the event is not reported to the client unless one of the newly loaded libraries matches the provided pattern.
This setting requires:
— Adding the names of any dlopened libraries to the TV::dlopen_recalculate_on_match list if you want breakpoints planted in the library when the library is loaded.
— Adding "libcuda.so*" to the match list if you are debugging CUDA; otherwise TotalView will miss CUDA kernel launch events.
• Fast Mode: Does not stop for dlopen events
Options:
dlopen_always_recalculate==false
dlopen_recalculate_on_match==""
This mode provides the best performance, disallowing planting breakpoints in dlopened libraries when the library is loaded. Breakpoints are planted in the dlopened libraries only when the process stops for some other reason; however, be aware with this option that an application may have executed past the point at which you want to start debugging inside the dlopened library.
Because the debugger does not plant the dlopen breakpoint in the process, the process never stops for a dlopen event, not even "null" dlopen events. While this mode may be impractical for some applications, the performance gains are significant.
Table 4 summarizes the pros and cons of each mode.
Table 4: dlopen Event Filtering Modes
Mode/Speed
Option
Slow
dlopen_always_recalculate==true
Pros:
• Retains TotalView’s traditional breakpoint reevaluation semantics.
• Works best with pending breakpoints.
• Compatible with CUDA.
Cons:
• Does not perform or scale as well as the other modes because the TotalView client handles every (non-null) dlopen event for every process.