Re: [PATCH 05/18] Add io_uring IO interface

On 1/29/19 4:58 AM, Arnd Bergmann wrote:

On Tue, Jan 29, 2019 at 7:30 AM Christoph Hellwig [off-list ref] wrote:

quoted

On Mon, Jan 28, 2019 at 11:25:12AM -0700, Jens Axboe wrote:

quoted

quoted

Especially with poll support now in the series, don't we need a ѕigmask
argument similar to pselect/ppoll/io_pgetevents now to deal with signal
blocking during waiting for events?

Is there any way to avoid passing in the sigset_t size? If it's just a
32-bit/64-bit thing, surely the in_compat_syscall() could cover it? Or
are there other cases that need to be catered to?

As far as I can tell we never look at it, never looked at it and don't
have any plans to look at it anytime soon.  But when I tried to omit
it for io_pgetevents I got stong pushback and thus had to add the
crazy double indirection calling convention.

Deepa has recently reworked the handling for the sigset_t handling
to be more consistent. As I understand it, we only ever check the
size argument to ensure that user and kernel space agree on
the size, as it there had been some historic differences.

If you pass a signal mask to a syscall, you should now just use the
set_user_sigmask()/set_compat_user_sigmask()/restore_user_sigmask()
helpers. The compat version is required for the incompatible bit order
between 32-bit and 64-bit big-endian architectures (on little-endian, compat
and native signal masks are compatible), and to deal with the one
architecture that has an _NSIG defined to something other than 64:
MIPS uses 128 because of a historic accident.

I think Deepa originally suggested combining set_user_sigmask()
and set_compat_user_sigmask(), using an in_compat_syscall()
check. This would let us simplify a number of compat syscalls
(ppoll, ppoll_time32, pselect6, pselect6_time32, epoll_pwait()
io_pgetevents_time64, io_pgetevents_time32). I advised against
changing it at the time for consistency with other compat syscalls,
but it's something we can still do now.

That's good info. I am currently using set_user_sigmask() for it.
I'd really like to avoid having to pass in a sigset_t size for the
system call, however. What's the best way of achieving that? Can I get
away with doing something like this:

	if (in_compat_syscall()) {
		const compat_sigset_t __user *compat_sig;

		compat_sig = (const compat_sigset_t __user *) sig;
		ret = set_compat_user_sigmask(compat_sig, &ksigmask,
						&sigsaved, _NSIG_WORDS);
	} else {
		ret = set_user_sigmask(sig, &ksigmask, &sigsaved,
						_NSIG_WORDS);
	}

-- 
Jens Axboe

--
To unsubscribe, send a message with 'unsubscribe linux-aio' in
the body to majordomo@kvack.org.  For more info on Linux AIO,
see: http://www.kvack.org/aio/
Don't email: <a href=mailto:"aart@kvack.org">aart@kvack.org</a>