* Stephen Smalley (sds(a)epoch.ncsc.mil) wrote:
On Tue, 2004-12-14 at 15:59, Timothy R. Chavez wrote:
> Ok, this sounds most reasonable. Thanks
What about the situation where multiple auditable objects are involved
in the syscall, whether via multi-component pathnames, multiple pathname
arguments to the syscall (e.g. rename, link), etc? Easier to just
immediately generate the object information from your hook, and then tie
all such object-based audit records to the associated syscall exit
record via the (timestamp, serial) tuples.
Or let any of those auditable objects trigger a flush on exit.
thanks,
-chris
--
Linux Security Modules
http://lsm.immunix.org http://lsm.bkbits.net