Re: [PATCH 0/3] syscalls: clean up stub naming convention

From: Ingo Molnar <mingo@kernel.org>
To: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: linux-kernel@vger.kernel.org, Al Viro <viro@zeniv.linux.org.uk>,
	Andi Kleen <ak@linux.intel.com>,
	Andrew Morton <akpm@linux-foundation.org>,
	Andy Lutomirski <luto@kernel.org>,
	Brian Gerst <brgerst@gmail.com>,
	Denys Vlasenko <dvlasenk@redhat.com>,
	"H. Peter Anvin" <hpa@zytor.com>, Ingo Molnar <mingo@redhat.com>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Peter Zijlstra <peterz@infradead.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	x86@kernel.org, Maninder Singh <maninder1.s@samsung.com>,
	Arnd Bergmann <arnd@arndb.de>,
	linux-arch <linux-arch@vger.kernel.org>
Subject: Re: [PATCH 0/3] syscalls: clean up stub naming convention
Date: Mon, 9 Apr 2018 09:06:11 +0200	[thread overview]
Message-ID: <20180409070611.43iq66eay2ctwv2m@gmail.com> (raw)
In-Reply-To: <20180409064511.ejhfqcurtlq2srq5@gmail.com>

* Ingo Molnar <mingo@kernel.org> wrote:

> 
> * Dominik Brodowski <linux@dominikbrodowski.net> wrote:
> 
> > On Sun, Apr 08, 2018 at 10:35:50AM +0200, Ingo Molnar wrote:
> > >   - _____sys_waitid()      # ridiculous number of underscores?
> > >   - __sys_waitid()         # too generic sounding?
> > 
> > ... and we'd need to rename internal helpers in net/
> > 
> > >   - __inline_sys_waitid()  # too long?
> > 
> > sounds acceptable, though a bit long (especially for the compat case, though
> > it doesn't really matter in the case of 
> > __inline_compat_sys_sched_rr_get_interval)
> 
> So as per the previous mail this is not just an inline function, but an active 
> type conversion wrapper that sign-extends 32-bit ints to longs, which is important 
> on some 64-bit architectures.
> 
> And that's a really non-obvious property IMO, and the name should probably reflect 
> _that_ non-obvious property, not the inlining property which is really just a 
> small detail.
> 
> I.e. how about:
> 
> 	__se_sys_waitid()
> 
> ... where 'se' stands for sign-extended, with a comment in the macro that explains 
> the prefix? (The historical abbreviation for sign extension is 'sext', which I 
> think wouldn't really be suitable these days.)

Ok, so I got confused there: I think it's the do_sys_waitid() intermediate that
is actually doing the sign-extension - and the inlined helper is what is in the 
syscall definition body.

So it's all still somewhat of a confusing misnomer: the 'do' named function is 
actually the sign-extension function variant - and the '_il' variant actually 
'does' the real work ...

I.e., old naming:

810f08d0 t     kernel_waitid	# common C function (see kernel/exit.c)

<inline>     __il_sys_waitid	# inlined helper doing the actual work
				# (takes parameters as declared)

810f1aa0 T   __do_sys_waitid	# C function calling inlined helper
				# (takes parameters of type long; casts
				#  them to the declared type)

810f1aa0 T        sys_waitid	# alias to __do_sys_waitid() (taking
				# parameters as declared), to be included
				# in syscall table

New suggested naming:

810f08d0 t     kernel_waitid	# common C function (see kernel/exit.c)

<inline>     __do_sys_waitid	# inlined helper doing the actual work
				# (takes original parameters as declared)

810f1aa0 T   __se_sys_waitid	# sign-extending C function calling inlined
				# helper (takes parameters of type long;
				# casts them to the declared type)

810f1aa0 T        sys_waitid	# alias to __se_sys_waitid() (but taking
				# original parameters as declared), to be
				# included in syscall table

Agreed?

Thanks,

	Ingo