On Thu, Sep 16, 2021 at 11:35:05AM +1200, Michael Schmitz wrote:
This one's a little harder - you use a 84 byte gap on each sigreturn, no matter what the frame size we need to restore. The original mangle_kernel_stack() only makes room on the stack when it has no other option (using twice as much size - correct me if I'm wrong). Ideally, we'd only leave a gap for mangle_kernel_stack() to use if the frame size requires us to do so. Working that out in asm glue would be sufficiently convoluted as to cancel out the benefits of cleaning up the C sigreturn part. Probably not worth it.
You'd need to * load the frame type from sigcontext (and deal with EFAULT, etc.) * make decision based on that * pass the type down into sigreturn(), so we wouldn't run into mismatches. And all that just to avoid a single "subtract a constant from stack pointer" insn. We are on a very shallow kernel stack here - it's a syscall entry, after all. And the stack footprint of do_sigreturn() is fairly small - e.g. stat(2) eats a lot more. We are not initializing the gap either - it's just reserved on stack; we only access it if we need to enlarge the stack frame. IOW, what would be the benefit of trying to avoid unconditional gap there?
