On Thu, Feb 3, 2022 at 9:40 AM Christoph Hellwig [off-list ref] wrote:

On Thu, Feb 03, 2022 at 09:32:38AM -0800, Alexei Starovoitov wrote:

quoted

These two places make it sound like it's interpreter only behavior.
I've reworded it like:
-the interpreter context is a pointer to networking packet.  ``BPF_ABS``
+the program context is a pointer to networking packet.  ``BPF_ABS``

-interpreter will abort the execution of the program.
+program execution will be aborted.

and pushed to bpf-next with the rest of patches.

The interpreter thing is actually unchanged from the old text, but I
totally agree with your fixup.  Thanks!

The old doc had:
"the interpreter will abort the execution of the program. JIT compilers
therefore must preserve this property."

Since the latter sentence is now gone the former got ambiguous.