labath wrote:

Maybe this would be a good time to reopen that discussion. The problem with 
clone3 is that it is not sandbox-friendly. And it's not that it's just too new 
to be included in the sandbox filters. The problem is that you cannot actually 
read the `struct clone_args` from a seccomp filter -- typically you want to 
permit some clone calls (creating a new thread), but not others (creating a new 
user namespace). This is particularly annoying nowadays, as agents will 
typically run the test suite in some sort of a sandbox.

AFAICT, the "standard" solution for this is to return ENOSYS, and have the C 
library catch it, and issue a SYS_clone instead. We could do that too, but it's 
not what we've been doing so far, and at least in this case, I don't see the 
advantage of doing that over just issuing a SYS_clone (or SYS_fork) directly...

https://github.com/llvm/llvm-project/pull/242253
_______________________________________________
llvm-branch-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits

Reply via email to