Seccomp and deep argument inspection

Posted Jun 11, 2020 0:33 UTC (Thu) by Cyberax (✭ supporter ✭, #52523)
Parent article: Seccomp and deep argument inspection

> Sarai would like to see there at least be a path forward if requirements of that sort do arise, but seemed willing to keep things simple for now—perhaps forever.
Perhaps code can be added to flatten these syscalls into a long buffer, if this is really ever needed?

Seccomp and deep argument inspection

Posted Jun 11, 2020 2:15 UTC (Thu) by NYKevin (subscriber, #129325) [Link]

I'm not too familiar with these pointer-to-struct-to-pointer arguments, but I have to wonder if there are any structs where a given pointer field is only required to be initialized when an associated flag is set... if so, the deep copy logic would need to parse that information out, or else it would incorrectly dereference random garbage and EFAULT/SIGBUS. But that would imply that you need to know the callee's semantics, and therefore there isn't a general/declarative "recursively copy everything, and figure it out later" solution.

Regardless, this is obviously a lot more complicated than a single call to (the moral equivalent of) memcpy(), which is probably why Linus doesn't want to do it.