Hello Trog,
Wednesday, April 6, 2005, 12:27:13 PM, you wrote:
T> On Wed, 2005-04-06 at 10:19 +0200, Thomas Lamy wrote:
>> Joanna Roman wrote:
>> > Hi What would happen if the clamd is notified by
>> > freshdb to reload the db when the clamd is in the
>> > middle of scanning something. I have not read that
>> > part of the code yet. But if you know the answer on
>> > top of your head, pls inform me.
>> >
>> >
>> AFAIR the thread manager (main thread) wait for all workers to finish
>> (and suspends worker creation) on reload event, going back to normal
>> operation after the reload has finished.
T> Thats correct for the current release.
T> The CVS code currently uses ref counting to overcome this limitation, so
T> when a reload is requested, a new signature tree is generated, and the
T> old one is free'd when all current scanning has finished.
T> -trog
In my project called Qavird I do switching of bases the following way:
case QAVIRD_CMD_AVBASERELOAD:
{
struct Qavird_info new_state_info;
struct cl_node * new_cl_root_node = NULL;
struct cl_node * old_node = cl_root_node;
/* 1) load new database into separate memory */
.......
.......
printf("Engine rebuilt succesfully\n");
/* 2) switch engine to a new database */
pthread_rwlock_wrlock( &cl_root_node_rwlock );
{
cl_root_node = new_cl_root_node;
new_state_info.virfound = state_info.virfound;
new_state_info.fdone = state_info.fdone;
new_state_info.startup = state_info.startup;
memcpy(&state_info,&new_state_info,sizeof(struct Qavird_info));
}
pthread_rwlock_unlock( &cl_root_node_rwlock );
printf("Successfully switched to a reloaded
database\n");
/* 3) free old database */
cl_free(old_node);
retval = QAV_CLEAN;
break;
}
while all the scanning threads currently running use
pthread_rwlock_rdlock( &cl_root_rwlock ) ... pthread_rwlock_unlock(...)
clause to access root node.
So multiple scanner threads are allowed to access the tree
simultaneously while they all be prevented from accessing root node
for a short period of time while the pointers are switched.
I hope it'll help in some way when further developing clamd.
--
Best regards,
-=MikeP=- mailto:[EMAIL PROTECTED]
_______________________________________________
http://lurker.clamav.net/list/clamav-devel.html
