Skip to contentSkip to conversation

perf bench: fix order of arguments to memcpy_alloc_mem

PATCH In review Bruce Merry 3 participants 3 messages Jan 15, 2015 – Mar 1, 2015 perf events

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 perf bench: fix order of arguments to memcpy_alloc_mem

This was causing the destination instead of the source to be filled. As a result, the source was typically all mapped to one zero page, and hence very cacheable.

---
 tools/perf/bench/mem-memcpy.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
tools/perf/bench/mem-memcpy.c +2 -2
index 6c14afe..db1d3a2 100644
@@ -289,7 +289,7 @@ static u64 do_memcpy_cycle(const struct routine *r, size_t len, bool prefault)
 	memcpy_t fn = r->fn.memcpy;
 	int i;
 
-	memcpy_alloc_mem(&src, &dst, len);
+	memcpy_alloc_mem(&dst, &src, len);
 
 	if (prefault)
 		fn(dst, src, len);
@@ -312,7 +312,7 @@ static double do_memcpy_gettimeofday(const struct routine *r, size_t len,
 	void *src = NULL, *dst = NULL;
 	int i;
 
-	memcpy_alloc_mem(&src, &dst, len);
+	memcpy_alloc_mem(&dst, &src, len);
 
 	if (prefault)
 		fn(dst, src, len);
Signature · 1 line
1.9.1
Acked-by on v2 1/1
#1 lore raw
Quoted 23 lines · Bruce Merry

This was causing the destination instead of the source to be filled. As a result, the source was typically all mapped to one zero page, and hence very cacheable.

---
 tools/perf/bench/mem-memcpy.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
tools/perf/bench/mem-memcpy.c +1 -1
index 6c14afe..db1d3a2 100644
@@ -289,7 +289,7 @@ static u64 do_memcpy_cycle(const struct routine *r, size_t len, bool prefault)
 	memcpy_t fn = r->fn.memcpy;
 	int i;
 
-	memcpy_alloc_mem(&src, &dst, len);
+	memcpy_alloc_mem(&dst, &src, len);
 
 	if (prefault)
 		fn(dst, src, len);

tools/perf/bench/mem-memcpy.c static double do_memcpy_gettimeofday(const struct routine *r, size_t len, · line 318

Quoted 8 lines
tools/perf/bench/mem-memcpy.c +1 -1
@@ -312,7 +312,7 @@ static double do_memcpy_gettimeofday(const struct routine *r, size_t len,
 	void *src = NULL, *dst = NULL;
 	int i;
 
-	memcpy_alloc_mem(&src, &dst, len);
+	memcpy_alloc_mem(&dst, &src, len);
 
 	if (prefault)
 		fn(dst, src, len);

Thanks,

Ingo

#2 lore raw

[tip:perf/urgent] perf bench: Fix order of arguments to memcpy_alloc_mem

In reply to Bruce Merry
perf bench: Fix order of arguments to memcpy_alloc_mem
Commit-ID:  e17fdaeaec066c725f73cd3cda1feae52b2646f5
Gitweb:     http://git.kernel.org/tip/e17fdaeaec066c725f73cd3cda1feae52b2646f5
Author:     Bruce Merry <[email protected]>
AuthorDate: Thu, 15 Jan 2015 11:20:22 +0200
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Sun, 22 Feb 2015 23:10:56 -0300

perf bench: Fix order of arguments to memcpy_alloc_mem

This was causing the destination instead of the source to be filled. As a result, the source was typically all mapped to one zero page, and hence very cacheable.

---
 tools/perf/bench/mem-memcpy.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
tools/perf/bench/mem-memcpy.c +2 -2
index 6c14afe..db1d3a2 100644
@@ -289,7 +289,7 @@ static u64 do_memcpy_cycle(const struct routine *r, size_t len, bool prefault)
 	memcpy_t fn = r->fn.memcpy;
 	int i;
 
-	memcpy_alloc_mem(&src, &dst, len);
+	memcpy_alloc_mem(&dst, &src, len);
 
 	if (prefault)
 		fn(dst, src, len);
@@ -312,7 +312,7 @@ static double do_memcpy_gettimeofday(const struct routine *r, size_t len,
 	void *src = NULL, *dst = NULL;
 	int i;
 
-	memcpy_alloc_mem(&src, &dst, len);
+	memcpy_alloc_mem(&dst, &src, len);
 
 	if (prefault)
 		fn(dst, src, len);