Hi Stephane,
attached are three patches for some issues I noticed while packaging
libpfm4 for Debian [1].
* inconsistent manpage sections
* some typos
* support for x32 (only a format string problem in an example)
Andreas
[1] https://tracker.debian.org/pkg/libpfm4
>From 09f53bdb1d11b309e2b5d098c0c81471921b49eb Mon Sep 17 00:00:00 2001
From: Andreas Beckmann <a.beckm...@fz-juelich.de>
Date: Sat, 12 Mar 2016 22:00:35 +0100
Subject: [PATCH 1/3] fix inconsistent man sections
Signed-off-by: Andreas Beckmann <a.beckm...@fz-juelich.de>
---
docs/man3/libpfm_arm_ac53.3 | 2 +-
docs/man3/libpfm_arm_ac57.3 | 2 +-
docs/man3/libpfm_arm_xgene.3 | 2 +-
3 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/docs/man3/libpfm_arm_ac53.3 b/docs/man3/libpfm_arm_ac53.3
index 319accc..09cb779 100644
--- a/docs/man3/libpfm_arm_ac53.3
+++ b/docs/man3/libpfm_arm_ac53.3
@@ -1,4 +1,4 @@
-.TH LIBPFM 4 "May, 2014" "" "Linux Programmer's Manual"
+.TH LIBPFM 3 "May, 2014" "" "Linux Programmer's Manual"
.SH NAME
libpfm_arm_ac53 - support for ARM Cortex A53 PMU
.SH SYNOPSIS
diff --git a/docs/man3/libpfm_arm_ac57.3 b/docs/man3/libpfm_arm_ac57.3
index c471ff1..feed0b5 100644
--- a/docs/man3/libpfm_arm_ac57.3
+++ b/docs/man3/libpfm_arm_ac57.3
@@ -1,4 +1,4 @@
-.TH LIBPFM 4 "May, 2014" "" "Linux Programmer's Manual"
+.TH LIBPFM 3 "May, 2014" "" "Linux Programmer's Manual"
.SH NAME
libpfm_arm_ac57 - support for Arm Cortex A57 PMU
.SH SYNOPSIS
diff --git a/docs/man3/libpfm_arm_xgene.3 b/docs/man3/libpfm_arm_xgene.3
index a0a84bc..7f12f9e 100644
--- a/docs/man3/libpfm_arm_xgene.3
+++ b/docs/man3/libpfm_arm_xgene.3
@@ -1,4 +1,4 @@
-.TH LIBPFM 4 "May, 2014" "" "Linux Programmer's Manual"
+.TH LIBPFM 3 "May, 2014" "" "Linux Programmer's Manual"
.SH NAME
libpfm_arm_ac57 - support for Applied Micro X-Gene PMU
.SH SYNOPSIS
--
2.7.0
>From 9fd328e014f7e991373061532ad77ffd46b2df64 Mon Sep 17 00:00:00 2001
From: Andreas Beckmann <a.beckm...@fz-juelich.de>
Date: Sat, 12 Mar 2016 22:02:42 +0100
Subject: [PATCH 2/3] fix some typos
Signed-off-by: Andreas Beckmann <a.beckm...@fz-juelich.de>
---
lib/events/arm_cortex_a53_events.h | 2 +-
lib/events/intel_hswep_unc_imc_events.h | 2 +-
lib/events/intel_ivbep_unc_imc_events.h | 2 +-
3 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/lib/events/arm_cortex_a53_events.h b/lib/events/arm_cortex_a53_events.h
index c0d2bb6..1d769c6 100644
--- a/lib/events/arm_cortex_a53_events.h
+++ b/lib/events/arm_cortex_a53_events.h
@@ -92,7 +92,7 @@ static const arm_entry_t arm_cortex_a53_pe[]={
{.name = "BR_IMMED_RETIRED",
.modmsk = ARMV8_ATTRS,
.code = 0x0d,
- .desc = "Software chnage of the PC, instruction architecturally executed, condition check pass"
+ .desc = "Software change of the PC, instruction architecturally executed, condition check pass"
},
{.name = "UNALIGNED_LDST_RETIRED",
.modmsk = ARMV8_ATTRS,
diff --git a/lib/events/intel_hswep_unc_imc_events.h b/lib/events/intel_hswep_unc_imc_events.h
index 7f77615..8b6f4bf 100644
--- a/lib/events/intel_hswep_unc_imc_events.h
+++ b/lib/events/intel_hswep_unc_imc_events.h
@@ -165,7 +165,7 @@ static const intel_x86_umask_t hswep_unc_m_pre_count[]={
.ucode = 0x400,
},
{ .uname = "WR",
- .udesc = "Precharhe due to write",
+ .udesc = "Precharge due to write",
.ucode = 0x800,
},
{ .uname = "BYP",
diff --git a/lib/events/intel_ivbep_unc_imc_events.h b/lib/events/intel_ivbep_unc_imc_events.h
index ba60c7e..33c6c06 100644
--- a/lib/events/intel_ivbep_unc_imc_events.h
+++ b/lib/events/intel_ivbep_unc_imc_events.h
@@ -165,7 +165,7 @@ static const intel_x86_umask_t ivbep_unc_m_pre_count[]={
.ucode = 0x400,
},
{ .uname = "WR",
- .udesc = "Precharhe due to write",
+ .udesc = "Precharge due to write",
.ucode = 0x800,
},
{ .uname = "BYP",
--
2.7.0
>From 7b4798488ae1dfe1aea9da5b180f5581e9280ae5 Mon Sep 17 00:00:00 2001
From: Andreas Beckmann <a.beckm...@fz-juelich.de>
Date: Sat, 12 Mar 2016 22:32:46 +0100
Subject: [PATCH 3/3] fix printf call with sizeof(time_t) != sizeof(long)
In environments with 32-bit long and 64-bit time_t (e.g. the x32 ABI
for x86) there is a type mismatch between the format string and its
argument. For now, always cast the timestamp to long. (Postponing this
problem until 2038 where 32-bit time_t will overflow anyway.)
Signed-off-by: Andreas Beckmann <a.beckm...@fz-juelich.de>
---
perf_examples/self_smpl_multi.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/perf_examples/self_smpl_multi.c b/perf_examples/self_smpl_multi.c
index 29b6916..0ee09e3 100644
--- a/perf_examples/self_smpl_multi.c
+++ b/perf_examples/self_smpl_multi.c
@@ -165,7 +165,7 @@ do_cycles(void)
gettimeofday(&now, NULL);
if (now.tv_sec > last.tv_sec) {
printf("%ld: myid = %3d, fd = %3d, count = %4ld, iter = %4ld, rate = %ld/Kiter\n",
- now.tv_sec - start.tv_sec,
+ (long)(now.tv_sec - start.tv_sec),
myid,
fd2ov[myid].fd,
count[myid], iter[myid],
--
2.7.0
------------------------------------------------------------------------------
Transform Data into Opportunity.
Accelerate data analysis in your applications with
Intel Data Analytics Acceleration Library.
Click to learn more.
http://pubads.g.doubleclick.net/gampad/clk?id=278785111&iu=/4140
_______________________________________________
perfmon2-devel mailing list
perfmon2-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/perfmon2-devel
