int CDevWatchdog::Open( const void* devName )
{
if ( ! devName )
devName=DEV_WATCHDOG ;
m_nDevFd = open((const char*)devName, O_RDWR ) ;
if ( -1 == m_nDevFd )
{
PERR( "Unable to open device file[%s]", (const char*)devName ) ;
return false ;
}
struct watchdog_info ident ;
ioctl(m_nDevFd, WDIOC_GETSUPPORT, &ident);
if ( ! (ident.options & (WDIOF_KEEPALIVEPING|WDIOF_SETTIMEOUT) )
)
{
ERR( "features missing: KEEPALIVEPING and SETTIMEOUT");
systemf_to( 20, "log2flash 'WTD: ERR: features missing: KEEPALIVEPING and SETTIMEOUT'&");
close(m_nDevFd);
m_nDevFd = -1 ;
return false ;
}
ioctl(m_nDevFd, WDIOC_GETTIMEOUT, &m_lPingInterval);
LOG("Found [%s] with %lu seconds timeout", ident.identity, m_lPingInterval);
systemf_to( 20,"log2flash 'WTD: Found [%s] with %lu seconds timeout'&", ident.identity, m_lPingInterval);
Ping() ;
return true ;
}
int CWatchdogMngr::LoadPidList()
{
CIniParser oIniParser;
if ( !oIniParser.Load(m_pPidListFile) )
{
throw CWatchdog::exception::exception("Unable to load pidListFile") ;
}
LOG("Loaded pidListFile [%s]", m_pPidListFile);
char* currentPidList[WTD_PIDLIST_ENTRIES]={0,};
size_t currentPidListLen=0;
char pidFile[256];
for ( unsigned pos=0
; oIniParser.GetVar("WTD", "watch", pidFile, sizeof(pidFile), pos)
&& currentPidListLen < WTD_PIDLIST_ENTRIES
; ++pos )
{
void*key=pidFile;
void *el=lfind( &key,currentPidList,¤tPidListLen,sizeof(currentPidList[0]),qsortcmp ) ;
if ( NULL == el )
{
currentPidList[currentPidListLen] = strdup(pidFile);
LOG( WTD"fw.cfg:[%s]", currentPidList[currentPidListLen] );
++currentPidListLen ;
}
}
qsort( currentPidList, currentPidListLen, sizeof(currentPidList[0]), qsortcmp );
/// Check if any pidFile from m_ppPidList is missing from currentPidList.
for ( unsigned j=0; j < m_nPidListLength && m_ppPidList[j]; ++j )
{
void *el=bsearch( &m_ppPidList[j], currentPidList, currentPidListLen, sizeof(currentPidList[0]), qsortcmp) ;
if ( NULL == el )
{
LOG( WTD "STOP watching pid[%s]", m_ppPidList[j]);
systemf_to( 20, "log2flash 'WTD: STOP watching pid[%s]'&", m_ppPidList[j]);
}
}
/// Free the old list to prepare the copy of the new one.
clearPidList() ;
m_nPidListLength=currentPidListLen ;
m_ppPidList[m_nPidListLength] = 0;
LOG( WTD"WATCH:%d modules", m_nPidListLength );
if ( !m_nPidListLength )
{
return true ;
}
memmove(m_ppPidList, currentPidList, WTD_PIDLIST_ENTRIES*sizeof(currentPidList[0]) );
for ( unsigned j = 0; j< m_nPidListLength ; j++)
{
LOG( WTD "WATCH:[%s]", m_ppPidList[j]);
}
return true ;
}
int CWatchdogMngr::addFileWatch(const char* pidListFile)
{
if ( -1 == access(pidListFile, R_OK) && m_status&CFG_ENABLED )
{
PERR( WTD "Disabling WTD: addFileWatch->access");
systemf_to( 20, "log2flash 'WTD: Disabling WTD: addFileWatch(%s) access->[%s]' &", pidListFile, strerror(errno) );
m_status&=~CFG_ENABLED ;
return false ;
}
m_wd = inotify_add_watch( m_fd, pidListFile, IN_MODIFY) ;
if (m_wd < 0)
{
PERR( WTD "inotify_add_watch");
systemf_to( 20, "log2flash 'WTD: inotify_add_watch(%s) failed [%s]' &", pidListFile, strerror(errno) );
throw CWatchdog::exception::exception("Unable to add notify watch") ;
}
return true ;
}
int CWatchdogMngr::PidMissing()
{
char szBlockedModules[256]={0,} ;
if ( checkDelPidFiles((const char**)m_ppPidList,szBlockedModules) )
{
return false ;
}
ERR("Pids missing:[%s]",szBlockedModules);
systemf_to( 20, "log2flash 'WTD: ERR:Pids missing:[%s]. Force REBOOT.'&",szBlockedModules) ;
return true ;
}
//////////////////////////////////////////////////////////////////////////////
/// This closes and properly disables the watchdog
//////////////////////////////////////////////////////////////////////////////
int CDevWatchdog::Close()
{
LOG("Stop Dev watchdog.");
systemf_to( 20, "log2flash 'WTD: Stop Dev watchdog.'&");
if ( m_nDevFd <= 0 ) return m_nDevFd ;
int rv = write( m_nDevFd, "V", 1);
if ( rv != 1 )
{
static int p1; if (!p1) { printf("WTD: FAIL:Unable to write stop char to /dev/watchdog[%s]", strerror(errno));p1=1; }
FWARN("FAIL:Unable to write stop char to /dev/watchdog[%s]", strerror(errno));
systemf_to( 20, "log2flash 'WTD: FAIL:Unable to write stop char to /dev/watchdog[%s]'&", strerror(errno));
}
rv = close(m_nDevFd);
if ( rv != 0 )
{
systemf_to( 20, "log2flash 'WTD: FAIL:Unable to close /dev/watchdog[%s]'&", strerror(errno));
}
m_nDevFd = -1 ;
return rv ;
}
void CWatchdogMngr::OnQuit()
{
CSimpleTimer t ;
Ping(); //to make sure that prepare_for_reboot.sh has time to run
systemf_to( ON_QUIT_DELAY, NIVIS_FIRMWARE "prepare_for_reboot.sh&" );
LOG("Waiting[%u seconds] for prepare_for_reboot.sh",ON_QUIT_DELAY) ;
int nTimeLeftToPing = ON_QUIT_DELAY;
if (FileIsExecutable(FTP_SCRIPT))
{
nTimeLeftToPing = ON_QUIT_DELAY_FTP_LOGS;
}
t.SetTimer( nTimeLeftToPing) ;
while( ! t.IsSignaling() )
{
Ping() ;
Sleep() ;
}
Ping() ;
systemf_to( 20, "log2flash 'WTD: Finished running quit procedure.'&");
}
void CWatchdogMngr::Ping()
{
#define _USEC_IN_SEC 1000000 /* Microseconds in a second */
struct timeval now;
gettimeofday(&now, (struct timezone *)NULL);
int usecDiff = (now.tv_usec + now.tv_sec*_USEC_IN_SEC) - (m_nLastPingTime.tv_usec + m_nLastPingTime.tv_sec*_USEC_IN_SEC);
usecDiff -= m_nTimeout*_USEC_IN_SEC ;
if ( usecDiff > 600000 /* we alow a 600msec drift*/ )
{
LOG("I was %d uSeconds late on ping.", usecDiff );
systemf_to( 20, "log2flash 'WTD: I was %d uSeconds late on ping.' &", usecDiff);
}
m_nLastPingTime = now;
if ( m_nCurrentWtd < __MAX_WATCHDOGS )
m_wdg[m_nCurrentWtd]->Ping() ;
for ( uint8_t i=0; i< m_nInsertPos && m_nCurrentWtd==CWatchdog::WTD_ALL; ++i )
m_wdg[i]->Ping() ;
#undef _USEC_IN_SEC
}
////////////////////////////////////////////////////////////////////////////// /// @brief removes all the pidfiles specified, but not too frequently /// @author Mihai Buha ([email protected]) /// @param[in] p_pszNames null-terminated array of filenames /// @param[out] p_szMissNames if not null, fill with names of /// missing files. Take care to allocate enough memory for /// "Watchdog: <all filenames in p_pszNames> " or else bad /// things will happen. /// @retval true All pidfiles existed or timeout not expired /// @retval false Some pidfile was missing (controlling app was dead - /// did not create a file during latest 2 intervals) ////////////////////////////////////////////////////////////////////////////// bool CWatchdogMngr::checkDelPidFiles (const char** p_pszNames, char* p_szMissNames) { clock_t timestamp ; static clock_t sStartTime = GetClockTicks(); static clock_t last_checked; static bool last_existed = true; bool status = true; bool exists = true; timestamp = GetClockTicks(); if( timestamp < last_checked) { // large uptime overflowed the clock_t last_checked = timestamp; sStartTime = timestamp; } if( (timestamp - last_checked < WTD_PID_VRFY_INTERVAL * sysconf(_SC_CLK_TCK)) || (timestamp - sStartTime < WTD_PID_VRFY_INTERVAL * sysconf(_SC_CLK_TCK)) ) { return true; } last_checked = timestamp; for( unsigned i=0; p_pszNames[i]; ++i) { int nFileLen = GetFileLen(p_pszNames[i]); if (nFileLen == 0) { usleep(30*1000);//wait a little maybe in process of writing the pid file nFileLen = GetFileLen(p_pszNames[i]); } if ( nFileLen > 0) { unlink( p_pszNames[i]); continue; } if (nFileLen == 0) { ERR("checkDelPidFiles: pidfile[%s] len==0", p_pszNames[i]); systemf_to( 20 ,"log2flash 'WTD: ERR:checkDelPidFiles: pidfile[%s] len==0' &", p_pszNames[i]); unlink( p_pszNames[i]); } LOG( "ERR:Missing [%s]", p_pszNames[i]); if ( exists ) { exists = false; } if( p_szMissNames ) { strcat( p_szMissNames, p_pszNames[i]); strcat( p_szMissNames, ","); } } // Take snapshot if any of the pidfiles is missing. if (! exists) { system_to( 60, NIVIS_TMP"take_system_snapshot.sh "ACTIVITY_DATA_PATH"snapshot_warning.txt &"); } status = exists || last_existed; last_existed = exists; return status; }
static bool log2flash_fa(const char *fname, const char *fmt, va_list arglist)
{
bool retcode=false; // assume operation failed.
int fd;
time_t cur_time;
time_t duration=0;
struct tm *ptm; // structure to obtain date information
char printbuf[MAX_WRITE_CHAR+2]; // size + '\n' + '\0'
int bytes_to_write;
int lock_to; // time out for flock()
DPRINT( "log2flash_fa() called\n" );
fd = open(fname, O_RDWR|O_CREAT|O_NONBLOCK, 00644);
if (fd == -1) {
LOG_ERR("log2flash ERROR: Unable to open file.");
goto EXIT;
}
lock_to = FLOCK_TIMEOUT;
while( flock(fd, LOCK_EX | LOCK_NB) != 0 ) {
sleep(1);
lock_to--;
if (lock_to < 0) {
LOG_ERR("log2flash ERROR: lock timeout");
goto CLOSE_EXIT;
}
}
if ( lseek(fd, 0, SEEK_END) == -1 ) {
LOG_ERR("log2flash ERROR: unable to seek to the end of file. errno=%d", errno);
goto CLOSE_EXIT;
}
cur_time=time(NULL);
ptm=gmtime(&cur_time);
// tm_year from 1900, tm_mon [0-11]
// it takes 20 characters spaces.
sprintf(printbuf, "%d-%02d-%02d %02d:%02d:%02d ",
ptm->tm_year+1900, ptm->tm_mon+1, ptm->tm_mday,
ptm->tm_hour, ptm->tm_min, ptm->tm_sec);
// generate the buffer string.
vsnprintf(printbuf+20, MAX_WRITE_CHAR-20, fmt, arglist);
bytes_to_write = strlen(printbuf);
printbuf[bytes_to_write] = '\n';
printbuf[++bytes_to_write] = 0;
DPRINT( printbuf );
// gather statistic
struct stat f_stat;
if ( fstat(fd, &f_stat) == -1 ) {
LOG_ERR("log2flash ERROR: unable to gatter statistic for %s. errno=%d", fname, errno);
goto CLOSE_EXIT;
}
// if write failed, goto CLOSE_EXIT
if (loop_write(fd, printbuf, bytes_to_write) != true) goto CLOSE_EXIT;
if ( f_stat.st_size + strlen(printbuf) > LOG_LIMIT) { // starting to rotate
DPRINT( "File Size exceeded LOG_LIMIT - Perform rotation\n" );
char from[PATH_MAX];
char to[PATH_MAX];
for( int i=ROTATE_MAX; i>0; --i)
{
sprintf( from, "%s.%d", fname, i-1);
sprintf( to, "%s.%d", fname, i);
rename( from, to);
}
duration = time(NULL);
Copy( fname, from, LOG_LIMIT );
unlink( fname );
duration -= time(NULL);
if (FileIsExecutable(FTP_SCRIPT))
{
char sBaseFile[PATH_MAX];
char sFileName[PATH_MAX];
sprintf(sBaseFile,NIVIS_TMP"activity.log");
AttachDatetime(sBaseFile,sFileName);
Copy(from,sFileName);
systemf_to(5, FTP_SCRIPT" %s &", sFileName);
}
}
retcode=true;
CLOSE_EXIT:
flock(fd, LOCK_UN);
close(fd);
if (duration < -10) { //it really is negative :) we only convert it when we need to show it, below
log2flash("Logfile rotation took quite long: %d seconds", -duration);
}
EXIT:
return retcode;
}