Thomas,

On Mon, Jul 20, 2026 at 9:21 PM Thomas Gleixner [off-list ref] wrote:

Similar to what Peter and me suggested in the related discussion
vs. s390 which has a similar issue, you can just have a dedicated
syscall_return member in your pt_regs struct, which is preset to -ENOSYS
and operate on that. You have an unused padding entry there which means
it won't even change the size.

Thanks, that makes sense in principle.

One complication is that my Alpha GENERIC_ENTRY branch already reuses that
former pt_regs padding slot, while keeping the pt_regs size and stack
alignment unchanged. The extra syscall-entry bookkeeping I needed so far lives
in thread_info rather than adding more fields to pt_regs.

So the padding slot is not free in my current branch. I can still look at
whether a dedicated syscall_return scratch value would be cleaner, but the
tested version currently handles this by setting the default -ENOSYS result
only on the no-dispatch path, after generic entry processing, and only if
ptrace/seccomp/BPF did not already provide a return value.

You were on CC for the V2, no?

Sorry, I must have missed that.

The whole pile including Jinjie's seccomp bypass fix is in

    git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git core/entry

and is targeted for the 7.3 merge window.

Great!

If you want to base your stuff on that for 7.3, I can add a tag which
makes it immutable so it can be pulled into the alpha tree.

Yes, please. There is still some ongoing testing/review of my Alpha
GENERIC_ENTRY series, but assuming it is ready in time for the 7.3 merge
window, an immutable tag would be very helpful.

Magnus