I've read on The Rootkit Arsenal book(page 269 - https://books.google.co.il/books?id=GAsuwHTquhEC) about kernel synchronization techniques.
According to what I've read, the safest/appropriate way to access a shared object in a multi CPU architecture is to issue a DPC call for every CPU core and spin over the shared object while the "main" core processing the shared object.
I don't really get It why spinning over a shared object and processing on a single cpu core is not enough, can someone explain to me?