Date: Sat, 6 Dec 2014 19:35:31 +0000 From: Andrew Turner <andrew@fubar.geek.nz> To: Dimitry Andric <dim@FreeBSD.org> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, src-committers@freebsd.org Subject: Re: svn commit: r275564 - head/sys/arm/arm Message-ID: <20141206193531.73875d39@bender.lan> In-Reply-To: <078F2487-9D4D-435C-92CC-C0E3EC7A34AE@FreeBSD.org> References: <201412061159.sB6BxZOV054978@svn.freebsd.org> <078F2487-9D4D-435C-92CC-C0E3EC7A34AE@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 6 Dec 2014 16:15:55 +0100 Dimitry Andric <dim@FreeBSD.org> wrote: > On 06 Dec 2014, at 12:59, Andrew Turner <andrew@FreeBSD.org> wrote: > > > > Author: andrew > > Date: Sat Dec 6 11:59:35 2014 > > New Revision: 275564 > > URL: https://svnweb.freebsd.org/changeset/base/275564 > > > > Log: > > Use the unified syntax when generating assembly for clang. The > > clang 3.5 integrated assembler only accepts it. > > > > MFC after: 1 week > > Sponsored by: ABT Systems Ltd > > > > Modified: > > head/sys/arm/arm/stdatomic.c > > > > Modified: head/sys/arm/arm/stdatomic.c > > ============================================================================== > > --- head/sys/arm/arm/stdatomic.c Sat Dec 6 10:05:07 > > 2014 (r275563) +++ head/sys/arm/arm/stdatomic.c Sat > > Dec 6 11:59:35 2014 (r275564) @@ -850,8 +850,13 @@ > > EMIT_FETCH_AND_OP_N(N, uintN_t, ldr, str EMIT_FETCH_AND_OP_N(N, > > uintN_t, ldr, str, fetch_and_sub, "sub") \ > > EMIT_FETCH_AND_OP_N(N, uintN_t, ldr, str, fetch_and_xor, "eor") > > > > +#ifdef __clang__ > > +EMIT_ALL_OPS_N(1, uint8_t, "ldrb", "strb", "strbeq") > > +EMIT_ALL_OPS_N(2, uint16_t, "ldrh", "strh", "strheq") > > +#else > > EMIT_ALL_OPS_N(1, uint8_t, "ldrb", "strb", "streqb") > > EMIT_ALL_OPS_N(2, uint16_t, "ldrh", "strh", "streqh") > > +#endif > > EMIT_ALL_OPS_N(4, uint32_t, "ldr", "str", "streq") > > I already had this change in the clang350-import project branch, but > unconditional. Should we just not standardize on the unified syntax, > so all these ifdefs can be avoided? Unfortunately gcc 4.2 doesn't produce unified syntax so we need to have both here. Without this the armeb build failed for me. Andrew
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20141206193531.73875d39>