Created attachment 160855 [details] v1 - __NR_ioprio_set on Linux is defined via <syscall.h> for a lot more architectures than those listed in ioprio.c - ioprio_set() syscall is not supported on FreeBSD even by linuxulator - Calling syscall(2) with values for different OS is dangerous e.g., __NR_ioprio_set == SYS_rfork, __NR_ioprio_get == SYS_openbsd_poll - ppc isn't a valid ARCH on FreeBSD but powerpc and powerpc64 is http://sourceforge.net/p/filebench/code/ci/fe57cd7c1d6c620020e6efb8ebcdf72b151fdf03/ ChangeLog: http://sourceforge.net/p/filebench/code/ci/1.4.9.1/log/?path= http://www.fsl.cs.sunysb.edu/~vass/filebench/ChangeLog (before 1.4.9) $ UNAME_m=powerpc UNAME_p=powerpc64 make ===> filebench-1.4.8.0.8 is only for amd64 i386 ia64 ppc, while you are running powerpc64. *** Error code 1
I can confirm is now builds and packages fine on PowerPC (locally), ia64 (eris), and sparc64 (flame). I did not review actual code changes though, just run some quick build tests as it's in line with what I'm currently doing.
One minor nit: please remove excessive caps from COMMENT: > -COMMENT= Performance Test of Filesystem I/O > +COMMENT= Performance test of filesystem I/O You might optionally replace "filesystem" with "file system", but that's too minor even for me to worry/insist about. :-)
A commit references this bug: Author: jbeich Date: Wed Sep 9 19:40:55 UTC 2015 New revision: 396529 URL: https://svnweb.freebsd.org/changeset/ports/396529 Log: benchmarks/filebench: fix cosmetic issues - Expand Linux ppc into FreeBSD powerpc and powerpc64 - Reword COMMENT based on summary in ${WRKSRC}/README - Rename files/patch-* according to |make makepatch| - Pet |portlint -C| PR: 202989 Approved by: portmgr blanket Changes: head/benchmarks/filebench/Makefile head/benchmarks/filebench/files/patch-parser__gram.c head/benchmarks/filebench/files/patch-parser_gram.c
Also builds fine for armv6 via qemu-user-static. I couldn't test aarch64 and mips due to: <jemalloc>: contrib/jemalloc/include/jemalloc/internal/tsd.h:625: Failed assertion: "tsd->state == tsd_state_reincarnated" <jemalloc>: contrib/jemalloc/include/jemalloc/internal/tsd.h:625: Failed assertion: "tsd->state == tsd_state_reincarnated" qemu: uncaught target signal 6 (Abort trap) - core dumped (In reply to Alexey Dokuchaev from comment #2) Done under blanket to avoid spoiling maintainer approval of the update.
Created attachment 160877 [details] v1.1 - Rebase after ports r396529 - Define LICENSE_FILE we have no copy of CDDL under /usr/ports/Templates/Licenses/
A commit references this bug: Author: jbeich Date: Sun Sep 27 21:58:57 UTC 2015 New revision: 398062 URL: https://svnweb.freebsd.org/changeset/ports/398062 Log: benchmarks/filebench: update to 1.4.9.1 - Drop ONLY_FOR_ARCHS as the Linux-specific code is gone - Chase new home on SourceForge - LICENSE_FILE is now shipped within distfile Changes: http://www.fsl.cs.sunysb.edu/~vass/filebench/ChangeLog (before 1.4.9) Changes: http://sourceforge.net/p/filebench/code/ci/1.4.9.1/log/ PR: 202989 Tested by: danfe (powerpc locally, ia64 on eris, sparc64 on flame) Tested by: myself (armv6 via qemu-user-static) Approved by: maintainer timeout (18 days) Changes: head/benchmarks/filebench/Makefile head/benchmarks/filebench/distinfo head/benchmarks/filebench/pkg-descr head/benchmarks/filebench/pkg-plist
Thanks. Committed.