Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 21 May 2023 10:14:26 -0700
From:      Mark Millard <marklmi@yahoo.com>
To:        hselasky@freebsd.org, dev-commits-src-main@freebsd.org
Subject:   RE: git: 805d759338a2 - main - mlx4: Move DEFINE_MUTEX() outside function body.
Message-ID:  <C375C14D-C185-415A-A2B4-82123F1EA5FF@yahoo.com>
References:  <C375C14D-C185-415A-A2B4-82123F1EA5FF.ref@yahoo.com>

next in thread | previous in thread | raw e-mail | index | archive | help
Hans Petter Selasky <hselasky_at_freebsd.org> wrote on
Date: Sun, 21 May 2023 16:57:47 UTC  :

> On 5/21/23 18:33, Jessica Clarke wrote:
> > On 21 May 2023, at 17:21, Hans Petter Selasky <hselasky@FreeBSD.org> =
wrote:
> >>
> >> The branch main has been updated by hselasky:
> >>
> >> URL: =
https://cgit.FreeBSD.org/src/commit/?id=3D805d759338a2be939fffc8bf3f25cfaa=
b981a9be
> >>
> >> commit 805d759338a2be939fffc8bf3f25cfaab981a9be
> >> Author: Hans Petter Selasky <hselasky@FreeBSD.org>
> >> AuthorDate: 2023-05-21 11:25:28 +0000
> >> Commit: Hans Petter Selasky <hselasky@FreeBSD.org>
> >> CommitDate: 2023-05-21 16:20:16 +0000
> >>
> >> mlx4: Move DEFINE_MUTEX() outside function body.
> >>
> >> Move static mutex declaration outside function body, to avoid =
global
> >> variables being declared on the stack, when using SYSINITs.
> >=20
> > What? This is nonsense. It=E2=80=99s not on the stack either way =
round.
> >=20
> > Please revert this.
> >=20
> > Jess
>=20
> Hi Jess,
>=20
> I think this is a false positive of yours. You need to look through =
all=20
> the macros used there.
>=20
> Basically DEFINE_MUTEX() expands to a bunch of structures, which are =
not=20
> in any block.

DEFINE_MUTEX is from/for linux related code:

# grep -r "define.*DEFINE_MUTEX" /usr/main-src/sys/ | more
/usr/main-src/sys/compat/linuxkpi/common/include/linux/mutex.h:#define  =
DEFINE_MUTEX(lock)                                              \

and looks like:

#define DEFINE_MUTEX(lock)                                              =
\
        mutex_t lock;                                                   =
\
        SX_SYSINIT_FLAGS(lock, &(lock).sx, mutex_name(#lock), SX_DUPOK)


> The "static" you see in patch just covers the first mutex structure.
>=20
> SYSINITs use "static" in front of all structure definitions.
> If you want to change from static structures to global symbols, then =
my=20
> change is correct.
>=20
> Before:
>=20
> static DEFINE_MUTEX(xxx);
>=20
> Expands to something like:
>=20
> static struct yyy xxx; static struct sysinit zzz; ....

Not true.

> If you want to change from "static struct sysinit zzz;" to "extern=20
> struct sysinit zzz" and also initialize the structure there, then that=20=

> won't work, based on what I currently know about C-programming. I =
tried,=20
> but clang gave me a warning about it.
>=20
>=20
>=20
> You can't declare global variables inside a function or it is not good=20=

> style.
>=20
>=20
>=20
> =46rom what I can see, this location is the only place I've come =
accross=20
> in the FreeBSD kernel, where a SYSINIT() is used inside a function, =
and=20
> I thought I would just move that outside the function instead.
>=20
> This change also allows for:
>=20
> https://reviews.freebsd.org/D40193
>=20





=3D=3D=3D
Mark Millard
marklmi at yahoo.com




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?C375C14D-C185-415A-A2B4-82123F1EA5FF>