mjhjmtu

From what I undertstand from sync() library call is that it allows a process to flush all dirty buffers to disk

The service routine sys_sync( ) of the sync( ) system call invokes a
series of auxiliary functions:

wakeup_bdflush(0);
sync_inodes(0);
sync_supers( );
sync_filesystems(0);
sync_filesystems(1);
sync_inodes(1);

If i run sync() call once, after that there shouldn't be anything in the buffer.

I am using a tool called stress that when runs with -i flag spawn N workers spinning on sync(),

stress --i 1

This command suppose to consume high IO wait since it keep sending sync() calls,
I confirmed it using ltrace,

sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0
sync() = 0

Can someone please explain why continuous sync calls consume a lot of IO? I was assuming that after first sync call buffer should be empty.