Date: Wed, 18 May 2016 17:35:06 +0000 From: "Bjoern A. Zeeb" <bzeeb-lists@lists.zabbadoz.net> To: Ian Lepore <ian@freebsd.org> Cc: Nathan Whitehorn <nwhitehorn@freebsd.org>, Justin Hibbits <chmeeedalf@gmail.com>, Scott Long <scottl@freebsd.org>, src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: Re: svn commit: r300154 - head/sys/net Message-ID: <1EED3540-DCCF-40D2-91BA-CE0AA54C5D08@lists.zabbadoz.net> In-Reply-To: <1463592737.1180.294.camel@freebsd.org> References: <201605181545.u4IFjCKD030751@repo.freebsd.org> <20160518105033.1eae7432@zhabar.knownspace> <759d085c-a485-c2ed-5d70-26eb4d27cdc2@freebsd.org> <1463592737.1180.294.camel@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> On 18 May 2016, at 17:32 , Ian Lepore <ian@freebsd.org> wrote: >=20 > On Wed, 2016-05-18 at 10:14 -0700, Nathan Whitehorn wrote: >>=20 >> On 05/18/16 08:50, Justin Hibbits wrote: >>> On Wed, 18 May 2016 15:45:12 +0000 (UTC) >>> Scott Long <scottl@FreeBSD.org> wrote: >>>=20 >>>> Author: scottl >>>> Date: Wed May 18 15:45:12 2016 >>>> New Revision: 300154 >>>> URL: https://svnweb.freebsd.org/changeset/base/300154 >>>>=20 >>>> Log: >>>> Activate the NO_64BIT_ATOMICS code for mips and powerpc >>>>=20 >>>> Modified: >>>> head/sys/net/mp_ring.c >>>>=20 >>>> Modified: head/sys/net/mp_ring.c >>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D >>>> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D >>>> --- head/sys/net/mp_ring.c Wed May 18 15:44:45 2016 >>>> (r300153) +++ head/sys/net/mp_ring.c Wed May 18 15:45:12 >>>> 2016 (r300154) @@ -37,15 +37,17 @@ __FBSDID("$FreeBSD$"); >>>> #include <sys/malloc.h> >>>> #include <machine/cpu.h> >>>>=20 >>>> - >>>> - >>>> -#include <net/mp_ring.h> >>>> +#if defined(__powerpc__) || defined(__mips__) >>>> +#define NO_64BIT_ATOMICS >>>> +#endif >>>>=20 >>>> #if defined(__i386__) >>>> #define atomic_cmpset_acq_64 atomic_cmpset_64 >>>> #define atomic_cmpset_rel_64 atomic_cmpset_64 >>>> #endif >>>>=20 >>>> +#include <net/mp_ring.h> >>>> + >>>> union ring_state { >>>> struct { >>>> uint16_t pidx_head; >>>>=20 >>> powerpc64 defines both __powerpc__ and __powerpc64__, so you're >>> killing >>> atomics on powerpc64 with this. >>>=20 >>> - Justin >>>=20 >>=20 >> Don't all of our 64-bit platforms have 64-bit atomics? So you could >> just=20 >> #if defined(__LP64__) || defined(__i386__) ||=20 >> defined(__whatever_the_thing_is_for_mips_n32__) >> -Nathan >>=20 >=20 > It may be more complicated than that, though. armv6 can do 64-bit > atomics even tho it's 32-bit. armv4, also 32-bit, can do 64-bit > atomics in the kernel but not in userland. >=20 > Maybe machine/atomic.h needs a #define that says whether 64-bit ops = are > available in the current compilation unit. (And likewise for other = bit > sizes if we have arches that have other limitations.) Question because I didn=E2=80=99t follow the details, but how was this = solved for the COUNTERS framework? /bz=
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1EED3540-DCCF-40D2-91BA-CE0AA54C5D08>