Message ID | 1353248997-30763-32-git-send-email-acme@infradead.org (mailing list archive) |
---|---|
State | Not Applicable |
Headers | show |
On Sun, Nov 18, 2012 at 9:29 AM, Arnaldo Carvalho de Melo <acme@infradead.org> wrote: > From: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com> > > Use the 'unistd.h' from arch/powerpc/include/uapi to build the perf tool. This would work for powerpc, but at least s390x and ARM have the same build issue. Why would we fix this piecemeal per-arch? David Howells has a patch that _should_ fix it for all of them at once. Is there a reason that isn't going in? josh
Em Mon, Nov 19, 2012 at 08:29:09AM -0500, Josh Boyer escreveu: > On Sun, Nov 18, 2012 at 9:29 AM, Arnaldo Carvalho de Melo <acme@infradead.org> wrote: > > From: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com> > > Use the 'unistd.h' from arch/powerpc/include/uapi to build the perf tool. > > This would work for powerpc, but at least s390x and ARM have the same > build issue. Why would we fix this piecemeal per-arch? > > David Howells has a patch that _should_ fix it for all of them at once. > Is there a reason that isn't going in? David rebased his patch on top of my perf/core branch, I just tested it and all works as expected, so I'd say he should push that branch upstream as I can't just cherry pick some csets, it needs a merge that involves more than just tools/perf AFAIK. - Arnaldo
On Mon, Nov 19, 2012 at 10:07 AM, Arnaldo Carvalho de Melo <acme@ghostprotocols.net> wrote: > Em Mon, Nov 19, 2012 at 08:29:09AM -0500, Josh Boyer escreveu: >> On Sun, Nov 18, 2012 at 9:29 AM, Arnaldo Carvalho de Melo <acme@infradead.org> wrote: >> > From: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com> > >> > Use the 'unistd.h' from arch/powerpc/include/uapi to build the perf tool. >> >> This would work for powerpc, but at least s390x and ARM have the same >> build issue. Why would we fix this piecemeal per-arch? >> >> David Howells has a patch that _should_ fix it for all of them at once. >> Is there a reason that isn't going in? > > David rebased his patch on top of my perf/core branch, I just tested it > and all works as expected, so I'd say he should push that branch > upstream as I can't just cherry pick some csets, it needs a merge that > involves more than just tools/perf AFAIK. OK... is perf/core (all 32 commits) headed into 3.7? Because the perf build is broken in 3.7-rcX, and we're at the point where Linus gets kind of picky about what he accepts. So we'd need your branch and David's to fix compile issues. And then we'd need Xiao Guangrong's fixes for 'perf kvm' to only build on x86. That's quite a number of fixes we need just to get perf to compile. Is someone other than me keeping track of that? I would just like people to stop wondering when perf builds on non-x86 architectures will work. josh
diff --git a/tools/perf/perf.h b/tools/perf/perf.h index 0047264..f53ee0b 100644 --- a/tools/perf/perf.h +++ b/tools/perf/perf.h @@ -26,7 +26,7 @@ void get_term_dimensions(struct winsize *ws); #endif #ifdef __powerpc__ -#include "../../arch/powerpc/include/asm/unistd.h" +#include "../../arch/powerpc/include/uapi/asm/unistd.h" #define rmb() asm volatile ("sync" ::: "memory") #define cpu_relax() asm volatile ("" ::: "memory"); #define CPUINFO_PROC "cpu"