Copyright (c) 2006, Sun Microsystems, Inc. All Rights Reserved.
The contents of this file are subject to the terms of the Common Development and Distribution License (the "License"). You may not use this file except in compliance with the License.
You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE or http://www.opensolaris.org/os/licensing. See the License for the specific language governing permissions and limitations under the License.
When distributing Covered Code, include this CDDL HEADER in each file and include the License file at usr/src/OPENSOLARIS.LICENSE. If applicable, add the following below this CDDL HEADER, with the fields enclosed by brackets "[]" replaced with your own identifying information: Portions Copyright [yyyy] [name of copyright owner]
c99 [ flag... ] file... -lm [ library... ] #include <fenv.h> int fex_set_log(FILE *fp);
FILE *fex_get_log(void);
int fex_set_log_depth(int depth);
int fex_get_log_depth(void);
void fex_log_entry(const char *msg);
The fex_set_log() function enables logging of retrospective diagnostic messages regarding floating point exceptions to the file specified by fp. If fp is NULL, logging is disabled. When a program starts, logging is initially disabled.
The occurrence of any of the twelve exceptions listed in fex_set_handling(3M) constitutes an event that can be logged. To prevent the log from becoming exhorbitantly long, the logging mechanism eliminates redundant entries by two methods. First, each exception is associated with a site in the program. The site is identified by the address of the instruction that caused the exception together with a stack trace. Only the first exception of a given type to occur at a given site will be logged. Second, when FEX_NONSTOP handling mode is in effect for some exception, only those occurrences of that exception that set its previously clear flag are logged. Clearing a flag using feclearexcept() allows the next occurrence of the exception to be logged provided it does not occur at a site at which it was previously logged.
Each of the different types of invalid operation exceptions can be logged at the same site. Because all invalid operation exceptions share the same flag, however, of those types for which FEX_NONSTOP mode is in effect, only the first exception to set the flag will be logged. When the invalid operation exception is raised by a call to feraiseexcept(3M) or feupdateenv(3M), which type of invalid operation is logged depends on the implementation.
If an exception results in the creation of a log entry, the entry is created at the time the exception occurs and before any exception handling actions selected with fex_set_handling() are taken. In particular, the log entry is available even if the program terminates as a result of the exception. The log entry shows the type of exception, the address of the instruction that caused it, how it will be handled, and the stack trace. If symbols are available, the address of the excepting instruction and the addresses in the stack trace are followed by the names of the corresponding symbols.
The fex_get_log() function returns the current log file.
The fex_set_log_depth() sets the maximum depth of the stack trace recorded with each exception to depth stack frames. The default depth is 100.
Thefex_get_log_depth() function returns the current maximum stack trace depth.
The fex_log_entry() function adds a user-supplied entry to the log. The entry includes the string pointed to by msg and the stack trace. Like entries for floating point exceptions, redundant user-supplied entries are eliminated: only the first user-supplied entry with a given msg to be requested from a given site will be logged. For the purpose of a user-supplied entry, the site is defined only by the stack trace, which begins with the function that called fex_log_entry().
The fex_set_log() function returns a non-zero value if logging is enabled or disabled accordingly and returns 0 otherwise. The fex_set_log_depth() returns a non-zero value if the requested stack trace depth is established (regardless of whether logging is enabled) and returns 0 otherwise.
The following example demonstrates the output generated when a floating point overflow occurs in sscanf(3C).
#include <fenv.h> int main() { double x; /* * enable logging of retrospective diagnostics */ (void) fex_set_log(stdout); /* * establish default handling for overflows */ (void) fex_set_handling(FEX_OVERFLOW, FEX_NONSTOP, NULL); /* * trigger an overflow in sscanf */ (void) sscanf("1.0e+400", "%lf", &x); return 0; }
The output from the preceding program reads:
Floating point overflow at 0xef71cac4 __base_conversion_set_exceptio n, nonstop mode 0xef71cacc __base_conversion_set_exception 0xef721820 _decimal_to_double 0xef75aba8 number 0xef75a94c __doscan_u 0xef75ecf8 sscanf 0x00010f20 main
Recompiling the program or running it on another system can produce different text addresses from those shown above.
See attributes(5) for descriptions of the following attributes:
| ATTRIBUTE TYPE | ATTRIBUTE VALUE |
| Availability | SUNWlibms, SUNWlmxs |
| Interface Stability | Stable |
| MT-Level | MT-Safe (see NOTES) |
feclearexcept(3M), fegetenv(3M), feraiseexcept(3M), feupdateenv(3M), fex_set_handling(3M), attributes(5)
Numerical Computation Guide
All threads in a process share the same log file. Each call to fex_set_log() preempts the previous one.
In addition to the log file itself, two additional file descriptors are used during the creation of a log entry in order to obtain symbol names from the executable and any shared objects it uses. These file descriptors are relinquished once the log entry is written. If the file descriptors cannot be allocated, symbols names are omitted from the stack trace.
The functions described on this page automatically install and deinstall SIGFPE handlers and set and clear the trap enable mode bits in the floating point status register as needed. If a program uses these functions and attempts to install a SIGFPE handler or control the trap enable mode bits independently, the resulting behavior is not defined.
As described in fex_set_handling(), when a handling function installed in FEX_CUSTOM mode is invoked, all exception traps are disabled (and will not be reenabled while SIGFPE is blocked). Thus, retrospective diagnostic messages are not logged for exceptions that occur within such a handler.