Hi Mahesh, In README file, there is a note on this. Refer to "Note on memory handling for writting log record"
And caller here means the one that call the function `log_stream_write_h`, not log client. Regards, Vu > -----Original Message----- > From: A V Mahesh [mailto:mahesh.va...@oracle.com] > Sent: Wednesday, October 12, 2016 1:47 PM > To: Vu Minh Nguyen <vu.m.ngu...@dektech.com.au>; > lennart.l...@ericsson.com > Cc: opensaf-devel@lists.sourceforge.net > Subject: Re: [PATCH 1 of 1] log: write_log_record_hdl get bad file descriptor > [#2028] > > Hi Vu, > > `Return (-1) to inform that it is caller's responsibility to free the > allocated mem` is this existing practices for application ? > or new , if new do we need to document it . > > -AVM > > On 10/12/2016 10:14 AM, A V Mahesh wrote: > > ACK > > > > -AVM > > > > > > On 9/13/2016 4:19 PM, Vu Minh Nguyen wrote: > >> osaf/services/saf/logsv/lgs/lgs_filehdl.cc | 16 +++++++++++----- > >> osaf/services/saf/logsv/lgs/lgs_stream.cc | 6 ++++++ > >> 2 files changed, 17 insertions(+), 5 deletions(-) > >> > >> > >> logsv did pass the WRITE request to file handle thread even > >> the file descriptor was invalid. Also, when closing file, > >> the file handle thread did not set it to invalid. > >> > >> This patch fixes these things. > >> > >> diff --git a/osaf/services/saf/logsv/lgs/lgs_filehdl.cc > >> b/osaf/services/saf/logsv/lgs/lgs_filehdl.cc > >> --- a/osaf/services/saf/logsv/lgs/lgs_filehdl.cc > >> +++ b/osaf/services/saf/logsv/lgs/lgs_filehdl.cc > >> @@ -467,16 +467,15 @@ open_retry: > >> */ > >> int fileclose_hdl(void *indata, void *outdata, size_t max_outsize) { > >> int rc = 0; > >> - int fd; > >> + int *fd = static_cast<int *>(indata); > >> - fd = *static_cast<int *>(indata); > >> - TRACE_ENTER2("fd=%d", fd); > >> + TRACE_ENTER2("fd=%d", *fd); > >> osaf_mutex_unlock_ordie(&lgs_ftcom_mutex); /* UNLOCK critical > >> section */ > >> /* Flush and synchronize the file before closing to guaranty that > >> the file > >> * is not written to after it's closed > >> */ > >> - if ((rc = fdatasync(fd)) == -1) { > >> + if ((rc = fdatasync(*fd)) == -1) { > >> if ((errno == EROFS) || (errno == EINVAL)) { > >> TRACE("Synchronization is not supported for this file"); > >> } else { > >> @@ -485,9 +484,16 @@ int fileclose_hdl(void *indata, void *ou > >> } > >> /* Close the file */ > >> - rc = close(fd); > >> + rc = close(*fd); > >> if (rc == -1) { > >> LOG_ER("fileclose() %s",strerror(errno)); > >> + } else { > >> + // When file system is busy, operations on files will take time. > >> + // In that case, file handle thread will get timeout and the > >> `requester` > >> + // will put the `fd` into one link list to do retry next time. > >> + // But if closing file succesfully, let the `requester` knows and > >> + // no need to send `close file request` again. > >> + *fd = -1; > >> } > >> osaf_mutex_lock_ordie(&lgs_ftcom_mutex); /* LOCK after critical > >> section */ > >> diff --git a/osaf/services/saf/logsv/lgs/lgs_stream.cc > >> b/osaf/services/saf/logsv/lgs/lgs_stream.cc > >> --- a/osaf/services/saf/logsv/lgs/lgs_stream.cc > >> +++ b/osaf/services/saf/logsv/lgs/lgs_stream.cc > >> @@ -1122,6 +1122,12 @@ int log_stream_write_h(log_stream_t *str > >> if (*stream->p_fd == -1) { > >> TRACE("%s - Initiating stream files \"%s\" Failed", > >> __FUNCTION__, > >> stream->name.c_str()); > >> + // Seems file system is busy - can not create requrested files. > >> + // Let inform the log client TRY_AGAIN. > >> + // > >> + // Return (-1) to inform that it is caller's responsibility > >> + // to free the allocated memmories. > >> + return -1; > >> } else { > >> TRACE("%s - stream files initiated", __FUNCTION__); > >> } > > ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot _______________________________________________ Opensaf-devel mailing list Opensaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/opensaf-devel
