Date: Tue, 14 Mar 2017 04:45:16 +0000 From: bugzilla-noreply@freebsd.org To: x11@FreeBSD.org Subject: [Bug 217771] lang/beignet: update to 1.3.1 Message-ID: <bug-217771-7141-CW2nFOk4VH@https.bugs.freebsd.org/bugzilla/> In-Reply-To: <bug-217771-7141@https.bugs.freebsd.org/bugzilla/> References: <bug-217771-7141@https.bugs.freebsd.org/bugzilla/>
next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D217771 --- Comment #7 from Jan Beich (mail not working) <jbeich@FreeBSD.org> --- Comment on attachment 180799 --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=3D180799 update lang/beignet to 1.3.1 OpenCL 2.0 requires LLVM >=3D 3.9 and libdrm >=3D 2.4.67 (i.e. ports r43146= 3 and maybe[1] drm-next). I forgot to adjust at least BUILD_DEPENDS in my version. [1] Whether drm_intel_bo_set_softpin_offset() works on FreeBSD 10.3 or 11.0= or causes a crash similar to bug 217635 is unknown. https://cgit.freedesktop.org/mesa/drm/commit/?id=3D8b4d57e7=20=20 > +OPTIONS_DEFINE=3D FP64 TEST > +FP64_DESC=3D Double precision (experimental) I don't like this style of grouping as it doesn't scale. If you have many options having to jump between descriptions and definitions when working on= a port with many can get tiring, see multimedia/ffmpeg. > +.if ${ARCH} =3D=3D amd64 > +PLIST_SUB+=3D OCL20=3D"" > +.else # ${ARCH} =3D=3D i386 > +PLIST_SUB+=3D OCL20=3D"@comment " > +.endif Conditionals break (mostly) declarative style of makefiles. Try the followi= ng instead PLIST_SUB+=3D OCL20=3D"${ARCH:Namd64:C/.+/@comment /}" > + -@(cd ${TEST_WRKSRC}.utests; . ./setenv.sh; \ Why do you need to introduce typos? $ make test WITH=3DTEST =3D=3D=3D> Testing for beignet-1.3.1 cd: /usr/ports/lang/beignet/work/Beignet-1.3.1-Source.utests: No such fil= e or directory .: cannot open ./setenv.sh: No such file or directory *** Error code 2 (ignored) $ make test WITH=3DTEST =3D=3D=3D> Testing for beignet-1.3.1 /bin/sh: ./flataddress_space: not found > +- src_data =3D (uint32_t*)memalign(base_address_alignment, buffer_sz); > +- if(!src_data) { > ++ if(posix_memalign((void**)&src_data, base_address_alignment, buffer_s= z)) { posix_memalign has greater minimal alignment than memalign or aligned_alloc= in jemalloc and glibc. And neither memalign nor posix_memalign are available on Windows. Or do you have something against using C11/C++17 features without passing corresponding -std=3D ? https://github.com/jemalloc/jemalloc/blob/4.5.0/src/jemalloc.c#L2038 https://github.com/jemalloc/jemalloc/blob/4.5.0/src/jemalloc.c#L1787 https://sourceware.org/git/?p=3Dglibc.git;a=3Dblob;f=3Dmalloc/malloc.c;hb= =3Dglibc-2.25#l3071 https://sourceware.org/git/?p=3Dglibc.git;a=3Dblob;f=3Dmalloc/malloc.c;hb= =3Dglibc-2.25#l3136 > @@ -17,7 +17,7 @@ >=20=20 > # XLib > Find_Package(X11) > -@@ -219,7 +218,7 @@ IF(OCLIcd_FOUND) > +@@ -217,7 +216,7 @@ IF(OCLIcd_FOUND) > "intel-beignet.icd.in" > "${ICD_FILE_NAME}" > ) Try to avoid excessive noise (e.g. line offsets, timestamps) in patches. > @@ -29,6 +29,8 @@ > lib/beignet/include/ocl_vload_20.h > lib/beignet/include/ocl_work_group.h > lib/beignet/include/ocl_workitem.h > +%%OCL20%%lib/beignet/beignet_20.bc > +%%OCL20%%lib/beignet/beignet_20.pch > lib/beignet/libcl.so > lib/beignet/libgbe.so > lib/beignet/libgbeinterp.so Keep sorting consistent with a version where %% macros are expanded. --=20 You are receiving this mail because: You are the assignee for the bug.=
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-217771-7141-CW2nFOk4VH>