HP Code Advisor Diagnostics Reference Guide (5900-1865, July 2011)

1.7 +wlock Diagnostics
Following table lists diagnostic messages that cadvise emits when you use the +wlock command
line option.
Table 7 +wlock diagnostics
Diagnostic messageDiagnostic number
Trying to lock an already held lock20220
Trying to unlock a lock not held20221
Failed to release lock %s is detected at program exit20222
Trying to unlock a lock held conditionally20223
Potential access of global variable %s without any locks held20224
Potential access of global variable %s without appropriate lock held20225
1.8 +wperfadvice Diagnostics
Following table lists all the diagnostic messages that cadvise emits when you use the +wperfadvice
command line option.
Table 8 +wperfadvice diagnostics
Diagnostic messageDiagnostic number
performance advice: passing a large (%s1 byte) parameter by value is inefficient, consider
passing %s2
4319
performance advice: std::endl is expensive because it flushes the stream. If you do not want
to flush the stream replace 'std::endl' with '\\\\n' or define macro
_HP_NONSTD_FAST_IOSTREAM
4320
performance advice: Exceeding memory limits when compiling procedure %1$s. Consider
splitting it in smaller procedures.
11042
performance advice: Loop is not pipelined because of resource constraints. Consider splitting
the loop.
11043
performance advice: Loop is not pipelined because of recurrence constraints. Consider
restructuring the loop to avoid dependencies.
11044
performance advice: Minimal optimizations are performed at +O%s. Consider using an
optimization level >= 2 for better performance
20300
performance advice: Optimization is turned off for the routine '%s'. If '#pragma optimize
off' is used, consider removing it
20301
performance advice: For better performance, consider using profile-based optimization.20302
performance advice: Routine '%s' in '%s' is indirectly called %s times (%s%% of total calls).
Conditionally promoting it to a direct call could reduce indirect call overhead and enable
additional optimizations.
20310
performance advice: Loop iterated %s times during %s%% of total invocations. Multi-versioning
the loop with the constant iteration count could enable a dditional optimizations. If this loop
20311
is one of the inner loops of a loop nest, consider multi-versioning it along with the enclosing
loops.
performance advice: Parallelizing this loop will likely provide better performance. Consider
using +Oautopar to do this automatically.
20312
performance advice: Multi-versioning this loop and parallelizing the higher trip-count version
will likely provide better performance. Consider using + Oautopar to do this automatically.
20313
20 Diagnostics Categorization