Hi Jiri,

On Monday 12 January 2015 04:39 PM, Jiri Olsa wrote:

On Sat, Jan 10, 2015 at 04:40:52PM +0530, Vineet Gupta wrote:

quoted

This is due to duplicated unistd inclusion (via uClibc headers + kernel headers)
Also seen on ARM uClibc based tools

------- ARC build ---------->8-------------

  CC       util/evlist.o
In file included from
~/arc/k.org/arch/arc/include/uapi/asm/unistd.h:25:0,
                 from util/../perf-sys.h:10,
                 from util/../perf.h:15,
                 from util/event.h:7,
                 from util/event.c:3:
~/arc/k.org/include/uapi/asm-generic/unistd.h:906:0:
warning: "__NR_fcntl64" redefined [enabled by default]
 #define __NR_fcntl64 __NR3264_fcntl
 ^
In file included from
~/arc/gnu/INSTALL_1412-arc-2014.12-rc1/arc-snps-linux-uclibc/sysroot/usr/include/sys/syscall.h:24:0,
                 from util/../perf-sys.h:6,
----------------->8-------------------

------- ARM build ---------->8-------------

  CC FPIC  plugin_scsi.o
In file included from util/../perf-sys.h:9:0,
                 from util/../perf.h:15,
                 from util/cache.h:7,
                 from perf.c:12:
~/arc/k.org/arch/arm/include/uapi/asm/unistd.h:28:0:
warning: "__NR_restart_syscall" redefined [enabled by default]
In file included from
~/buildroot/host/usr/arm-buildroot-linux-uclibcgnueabi/sysroot/usr/include/sys/syscall.h:25:0,
                 from util/../perf-sys.h:6,
                 from util/../perf.h:15,
                 from util/cache.h:7,
                 from perf.c:12:
~/buildroot/host/usr/arm-buildroot-linux-uclibcgnueabi/sysroot/usr/include/bits/sysnum.h:17:0:
note: this is the location of the previous definition
----------------->8-------------------

Could you please point out what's the right logic for including those headers?

I'm not sure why you were including <linux/unistd.h> in first place. Vanilla
<unistd.h> shd suffice for the functions exported by that file. OTOH, if the
intent was to get SYS_xxx or NR_xxx, then per syscall man page, <sys/syscall.h> is
the recommended way.

I just tried a small test program which uses __NR_restart_syscall for both x86
(glibc) and ARC (uClibc) and including <sys/syscall.h> seems to work.

Also replacing <linux/unistd.h> with <sys/syscall.h> in the first 2 hunks of this
patch equally fixes my perf build splat. The 3rd one (perf-sys.h) already includes
<sys/syscall.h> so no change required there.

I'll send out a v3 to same effect.

I've always thought that just adding:

  #include <linux/unistd.h>

should be safe no matter what.. feels like bug in one of those
header files? (uClibc headers & kernel headers)

Well the design of asm-generic/unistd.h (no legacy syscalls ABI) which ARC uses
does require duplicate inclusion of asm/unistd.h so a simple guard in header is
not sufficient (and actually caused me embarrassing grief when I naively submitted
that into mainline some time last year)

Thx,
-Vineet

thanks for info,
jirka