On Wed, Feb 29, 2012 at 11:41 AM, Roland McGrath [off-list ref] wrote:
I don't think TIF_NOTIFY_RESUME is apropos here. That only triggers on
returning to user mode, i.e. after syscall exit. But regardless of the
exact implementation details, I don't think it will be prohibitive to add
some means by which the fast-path can back off before actual syscall entry
and go to the slow path for ptrace reporting.
Since there is no strong reason to think it can't be reorganized that way
later, I don't see any good rationale for constraining the seccomp-filter
feature definition based on a plan to optimize the implementation in the
future.
Sounds good to me. I'll move to ptrace_event and save the problem of
code organization for the future.
Thanks!