siginfo.h, siginfo - signal generation information
If a process is catching a signal, it might request information that tells why the system generated that signal. See sigaction(2). If a process is monitoring its children, it might receive information that tells why a child changed state. See waitid(2). In either case, the system returns the information in a structure of type siginfo_t, which includes the following information:
int si_signo /* signal number */ int si_errno /* error number */ int si_code /* signal code */ union sigval si_value /* signal value */
si_signo contains the system-generated signal number. For the waitid(2) function, si_signo is always SIGCHLD.
If si_errno is non-zero, it contains an error number associated with this signal, as defined in <errno.h> .
si_code contains a code identifying the cause of the signal.
If the value of the si_code member is SI_NOINFO, only the si_signo member of siginfo_t is meaningful, and the value of all other members is unspecified.
The psiginfo(3C) function produces messages on the standard error output describing the values contained in a siginfo_t structure.
If the value of si_code is less than or equal to 0, then the signal was generated by a user process (see kill(2), _lwp_kill(2), sigqueue(3C), sigsend(2), abort(3C), and raise(3C)) and the siginfo structure contains the following additional information:
pid_t si_pid /* sending process ID */ uid_t si_uid /* sending user ID */ ctid_t si_ctid /* sending contract ID */ zoneid_t si_zoneid /* sending zone ID */S
If the signal was generated by a user process, the following values are defined for si_code:
The signal was sent by _lwp_kill(2).
The signal was sent by sigqueue(3C).
The signal was generated by the expiration of a timer created by timer_settime(3C).
The signal was generated by the completion of an asynchronous I/O request.
The signal was generated by the arrival of a message on an empty message queue. For more information, see the mq_notify(3C) man page.
si_value contains the application specified value, which is passed to the application's signal-catching function at the time of the signal delivery if si_code is any of SI_QUEUE, SI_TIMER, SI_ASYNCHIO, or SI_MESGQ.
Non-user generated signals can arise for a number of reasons. For all of these cases, si_code contains a positive value reflecting the reason why the system generated the signal:
Signals can also be generated from the resource control subsystem. Where these signals do not already possess kernel-level siginfo codes, the siginfo si_code will be filled with SI_RCTL to indicate a kernel-generated signal from an established resource control value.
The uncatchable signals SIGSTOP and SIGKILL have undefined siginfo codes.
Signals sent with a siginfo code of SI_RCTL contain code-dependent information for kernel-generated signals:
In addition, the following signal-dependent information is available for kernel-generated signals:
Signals sent with a siginfo code of SEGV_ADIPERR or SEGV_ADIDERR contain signal-dependent information for kernel-generated SIGSEGV signals:
Signals sent with a siginfo code of SEGV_CMI contain signal-dependent information for kernel-generated SIGSEGV signals:
_lwp_kill(2), adi_version_max(2), kill(2), setrctl(2), sigaction(2), sigsend(2), waitid(2), abort(3C), aio_read(3C), mq_notify(3C), psiginfo(3C), raise(3C), sigqueue(3C), timer_create(3C), timer_settime(3C), signal.h(3HEAD)
For SIGCHLD signals, if si_code is equal to CLD_EXITED, then si_status is equal to the exit value of the process; otherwise, it is equal to the signal that caused the process to change state. For some implementations, the exact value of si_addr might not be available; in that case, si_addr is guaranteed to be on the same page as the faulting instruction or memory reference.
For SIGSEGV signals with si_code equal to SEGV_ADIPERR, the value of si_adivers will be the in-memory version of the mismatched address at the time that the kernel fetches it during the processing of the trap, not the version at the time of the trap. If a thread other than the one that executed the data access and took the trap changes the version between the time of the trap and the kernel access, si_adivers will contain that new value.
For SIGSEGV signals with si_code equal to SEGV_ADIPERR, if the in-memory version is lost due to a hardware error which cannot be corrected, or if the in-memory version is greater than the value returned by the adi_version_max(2) function, si_adivers will be set to -1.
A SIGSEGV signal with si_code equal to SEGV_ADIPERR results from an ADI exception that generates a precise trap, while a SIGSEGV signal with si_code equal to SEGV_ADIDERR results from an ADI exception that generates a disrupting trap. When si_code is equal to SEGV_ADIDERR, si_pc is -1 because the PC (program counter) where the disrupting trap occurred is not related to the store that caused the trap. si_adivers is -1 because the store address that caused the trap is not provided by the hardware. There is no reliable way for software to determine the address since an unpredictable number of instructions are executed after the store and before the trap is taken due to the nature of a disrupting trap.