* [PATCH v5 1/3] tools bitmap: implement bitmap_equal() operation at bitmap API
2019-12-03 11:41 [PATCH v5 0/3] perf record: adapt NUMA awareness to machines with #CPUs > 1K Alexey Budankov
@ 2019-12-03 11:43 ` Alexey Budankov
2020-01-10 17:53 ` [tip: perf/core] tools bitmap: Implement " tip-bot2 for Alexey Budankov
2019-12-03 11:44 ` [PATCH v5 2/3] perf mmap: declare type for cpu mask of arbitrary length Alexey Budankov
` (2 subsequent siblings)
3 siblings, 1 reply; 13+ messages in thread
From: Alexey Budankov @ 2019-12-03 11:43 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo
Cc: Jiri Olsa, Namhyung Kim, Alexander Shishkin, Peter Zijlstra,
Ingo Molnar, Andi Kleen, linux-kernel
Extend tools bitmap API with bitmap_equal() implementation.
The implementation has been derived from the kernel.
Extend tools bitmap API with bitmap_free() implementation for
symmetry with bitmap_alloc() function.
Signed-off-by: Alexey Budankov <alexey.budankov@linux.intel.com>
---
tools/include/linux/bitmap.h | 30 ++++++++++++++++++++++++++++++
tools/lib/bitmap.c | 15 +++++++++++++++
2 files changed, 45 insertions(+)
diff --git a/tools/include/linux/bitmap.h b/tools/include/linux/bitmap.h
index 05dca5c203f3..477a1cae513f 100644
--- a/tools/include/linux/bitmap.h
+++ b/tools/include/linux/bitmap.h
@@ -15,6 +15,8 @@ void __bitmap_or(unsigned long *dst, const unsigned long *bitmap1,
const unsigned long *bitmap2, int bits);
int __bitmap_and(unsigned long *dst, const unsigned long *bitmap1,
const unsigned long *bitmap2, unsigned int bits);
+int __bitmap_equal(const unsigned long *bitmap1,
+ const unsigned long *bitmap2, unsigned int bits);
void bitmap_clear(unsigned long *map, unsigned int start, int len);
#define BITMAP_FIRST_WORD_MASK(start) (~0UL << ((start) & (BITS_PER_LONG - 1)))
@@ -123,6 +125,15 @@ static inline unsigned long *bitmap_alloc(int nbits)
return calloc(1, BITS_TO_LONGS(nbits) * sizeof(unsigned long));
}
+/*
+ * bitmap_free - Free bitmap
+ * @bitmap: pointer to bitmap
+ */
+static inline void bitmap_free(unsigned long *bitmap)
+{
+ free(bitmap);
+}
+
/*
* bitmap_scnprintf - print bitmap list into buffer
* @bitmap: bitmap
@@ -148,4 +159,23 @@ static inline int bitmap_and(unsigned long *dst, const unsigned long *src1,
return __bitmap_and(dst, src1, src2, nbits);
}
+#ifdef __LITTLE_ENDIAN
+#define BITMAP_MEM_ALIGNMENT 8
+#else
+#define BITMAP_MEM_ALIGNMENT (8 * sizeof(unsigned long))
+#endif
+#define BITMAP_MEM_MASK (BITMAP_MEM_ALIGNMENT - 1)
+#define IS_ALIGNED(x, a) (((x) & ((typeof(x))(a) - 1)) == 0)
+
+static inline int bitmap_equal(const unsigned long *src1,
+ const unsigned long *src2, unsigned int nbits)
+{
+ if (small_const_nbits(nbits))
+ return !((*src1 ^ *src2) & BITMAP_LAST_WORD_MASK(nbits));
+ if (__builtin_constant_p(nbits & BITMAP_MEM_MASK) &&
+ IS_ALIGNED(nbits, BITMAP_MEM_ALIGNMENT))
+ return !memcmp(src1, src2, nbits / 8);
+ return __bitmap_equal(src1, src2, nbits);
+}
+
#endif /* _PERF_BITOPS_H */
diff --git a/tools/lib/bitmap.c b/tools/lib/bitmap.c
index 38494782be06..5043747ef6c5 100644
--- a/tools/lib/bitmap.c
+++ b/tools/lib/bitmap.c
@@ -71,3 +71,18 @@ int __bitmap_and(unsigned long *dst, const unsigned long *bitmap1,
BITMAP_LAST_WORD_MASK(bits));
return result != 0;
}
+
+int __bitmap_equal(const unsigned long *bitmap1,
+ const unsigned long *bitmap2, unsigned int bits)
+{
+ unsigned int k, lim = bits/BITS_PER_LONG;
+ for (k = 0; k < lim; ++k)
+ if (bitmap1[k] != bitmap2[k])
+ return 0;
+
+ if (bits % BITS_PER_LONG)
+ if ((bitmap1[k] ^ bitmap2[k]) & BITMAP_LAST_WORD_MASK(bits))
+ return 0;
+
+ return 1;
+}
--
2.20.1
^ permalink raw reply related [flat|nested] 13+ messages in thread* [tip: perf/core] tools bitmap: Implement bitmap_equal() operation at bitmap API
2019-12-03 11:43 ` [PATCH v5 1/3] tools bitmap: implement bitmap_equal() operation at bitmap API Alexey Budankov
@ 2020-01-10 17:53 ` tip-bot2 for Alexey Budankov
0 siblings, 0 replies; 13+ messages in thread
From: tip-bot2 for Alexey Budankov @ 2020-01-10 17:53 UTC (permalink / raw)
To: linux-tip-commits
Cc: Alexey Budankov, Jiri Olsa, Alexander Shishkin, Andi Kleen,
Namhyung Kim, Peter Zijlstra, Arnaldo Carvalho de Melo, x86,
LKML
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 8812ad412f851216d6c39488a7e563ccc5c604cc
Gitweb: https://git.kernel.org/tip/8812ad412f851216d6c39488a7e563ccc5c604cc
Author: Alexey Budankov <alexey.budankov@linux.intel.com>
AuthorDate: Tue, 03 Dec 2019 14:43:33 +03:00
Committer: Arnaldo Carvalho de Melo <acme@redhat.com>
CommitterDate: Mon, 06 Jan 2020 11:46:04 -03:00
tools bitmap: Implement bitmap_equal() operation at bitmap API
Extend tools bitmap API with bitmap_equal() implementation.
The implementation has been derived from the kernel.
Extend tools bitmap API with bitmap_free() implementation for symmetry
with bitmap_alloc() function.
Signed-off-by: Alexey Budankov <alexey.budankov@linux.intel.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/43757993-0b28-d8af-a6c7-ede12e3a6877@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/include/linux/bitmap.h | 30 ++++++++++++++++++++++++++++++
tools/lib/bitmap.c | 15 +++++++++++++++
2 files changed, 45 insertions(+)
diff --git a/tools/include/linux/bitmap.h b/tools/include/linux/bitmap.h
index 05dca5c..477a1ca 100644
--- a/tools/include/linux/bitmap.h
+++ b/tools/include/linux/bitmap.h
@@ -15,6 +15,8 @@ void __bitmap_or(unsigned long *dst, const unsigned long *bitmap1,
const unsigned long *bitmap2, int bits);
int __bitmap_and(unsigned long *dst, const unsigned long *bitmap1,
const unsigned long *bitmap2, unsigned int bits);
+int __bitmap_equal(const unsigned long *bitmap1,
+ const unsigned long *bitmap2, unsigned int bits);
void bitmap_clear(unsigned long *map, unsigned int start, int len);
#define BITMAP_FIRST_WORD_MASK(start) (~0UL << ((start) & (BITS_PER_LONG - 1)))
@@ -124,6 +126,15 @@ static inline unsigned long *bitmap_alloc(int nbits)
}
/*
+ * bitmap_free - Free bitmap
+ * @bitmap: pointer to bitmap
+ */
+static inline void bitmap_free(unsigned long *bitmap)
+{
+ free(bitmap);
+}
+
+/*
* bitmap_scnprintf - print bitmap list into buffer
* @bitmap: bitmap
* @nbits: size of bitmap
@@ -148,4 +159,23 @@ static inline int bitmap_and(unsigned long *dst, const unsigned long *src1,
return __bitmap_and(dst, src1, src2, nbits);
}
+#ifdef __LITTLE_ENDIAN
+#define BITMAP_MEM_ALIGNMENT 8
+#else
+#define BITMAP_MEM_ALIGNMENT (8 * sizeof(unsigned long))
+#endif
+#define BITMAP_MEM_MASK (BITMAP_MEM_ALIGNMENT - 1)
+#define IS_ALIGNED(x, a) (((x) & ((typeof(x))(a) - 1)) == 0)
+
+static inline int bitmap_equal(const unsigned long *src1,
+ const unsigned long *src2, unsigned int nbits)
+{
+ if (small_const_nbits(nbits))
+ return !((*src1 ^ *src2) & BITMAP_LAST_WORD_MASK(nbits));
+ if (__builtin_constant_p(nbits & BITMAP_MEM_MASK) &&
+ IS_ALIGNED(nbits, BITMAP_MEM_ALIGNMENT))
+ return !memcmp(src1, src2, nbits / 8);
+ return __bitmap_equal(src1, src2, nbits);
+}
+
#endif /* _PERF_BITOPS_H */
diff --git a/tools/lib/bitmap.c b/tools/lib/bitmap.c
index 3849478..5043747 100644
--- a/tools/lib/bitmap.c
+++ b/tools/lib/bitmap.c
@@ -71,3 +71,18 @@ int __bitmap_and(unsigned long *dst, const unsigned long *bitmap1,
BITMAP_LAST_WORD_MASK(bits));
return result != 0;
}
+
+int __bitmap_equal(const unsigned long *bitmap1,
+ const unsigned long *bitmap2, unsigned int bits)
+{
+ unsigned int k, lim = bits/BITS_PER_LONG;
+ for (k = 0; k < lim; ++k)
+ if (bitmap1[k] != bitmap2[k])
+ return 0;
+
+ if (bits % BITS_PER_LONG)
+ if ((bitmap1[k] ^ bitmap2[k]) & BITMAP_LAST_WORD_MASK(bits))
+ return 0;
+
+ return 1;
+}
^ permalink raw reply related [flat|nested] 13+ messages in thread
* [PATCH v5 2/3] perf mmap: declare type for cpu mask of arbitrary length
2019-12-03 11:41 [PATCH v5 0/3] perf record: adapt NUMA awareness to machines with #CPUs > 1K Alexey Budankov
2019-12-03 11:43 ` [PATCH v5 1/3] tools bitmap: implement bitmap_equal() operation at bitmap API Alexey Budankov
@ 2019-12-03 11:44 ` Alexey Budankov
2019-12-04 13:49 ` Arnaldo Carvalho de Melo
2020-01-10 17:53 ` [tip: perf/core] perf mmap: Declare " tip-bot2 for Alexey Budankov
2019-12-03 11:45 ` [PATCH v5 3/3] perf record: adapt affinity to machines with #CPUs > 1K Alexey Budankov
2019-12-03 12:17 ` [PATCH v5 0/3] perf record: adapt NUMA awareness " Jiri Olsa
3 siblings, 2 replies; 13+ messages in thread
From: Alexey Budankov @ 2019-12-03 11:44 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo
Cc: Jiri Olsa, Namhyung Kim, Alexander Shishkin, Peter Zijlstra,
Ingo Molnar, Andi Kleen, linux-kernel
Declare a dedicated struct map_cpu_mask type for cpu masks of
arbitrary length. Mask is available thru bits pointer and the
mask length is kept in nbits field. MMAP_CPU_MASK_BYTES() macro
returns mask storage size in bytes. mmap_cpu_mask__scnprintf()
function can be used to log text representation of the mask.
Signed-off-by: Alexey Budankov <alexey.budankov@linux.intel.com>
---
tools/perf/util/mmap.c | 12 ++++++++++++
tools/perf/util/mmap.h | 11 +++++++++++
2 files changed, 23 insertions(+)
diff --git a/tools/perf/util/mmap.c b/tools/perf/util/mmap.c
index 063d1b93c53d..43c12b4a3e17 100644
--- a/tools/perf/util/mmap.c
+++ b/tools/perf/util/mmap.c
@@ -23,6 +23,18 @@
#include "mmap.h"
#include "../perf.h"
#include <internal/lib.h> /* page_size */
+#include <linux/bitmap.h>
+
+#define MASK_SIZE 1023
+void mmap_cpu_mask__scnprintf(struct mmap_cpu_mask *mask, const char *tag)
+{
+ char buf[MASK_SIZE + 1];
+ size_t len;
+
+ len = bitmap_scnprintf(mask->bits, mask->nbits, buf, MASK_SIZE);
+ buf[len] = '\0';
+ pr_debug("%p: %s mask[%ld]: %s\n", mask, tag, mask->nbits, buf);
+}
size_t mmap__mmap_len(struct mmap *map)
{
diff --git a/tools/perf/util/mmap.h b/tools/perf/util/mmap.h
index bee4e83f7109..ef51667fabcb 100644
--- a/tools/perf/util/mmap.h
+++ b/tools/perf/util/mmap.h
@@ -15,6 +15,15 @@
#include "event.h"
struct aiocb;
+
+struct mmap_cpu_mask {
+ unsigned long *bits;
+ size_t nbits;
+};
+
+#define MMAP_CPU_MASK_BYTES(m) \
+ (BITS_TO_LONGS(((struct mmap_cpu_mask *)m)->nbits) * sizeof(unsigned long))
+
/**
* struct mmap - perf's ring buffer mmap details
*
@@ -52,4 +61,6 @@ int perf_mmap__push(struct mmap *md, void *to,
size_t mmap__mmap_len(struct mmap *map);
+void mmap_cpu_mask__scnprintf(struct mmap_cpu_mask *mask, const char *tag);
+
#endif /*__PERF_MMAP_H */
--
2.20.1
^ permalink raw reply related [flat|nested] 13+ messages in thread* Re: [PATCH v5 2/3] perf mmap: declare type for cpu mask of arbitrary length
2019-12-03 11:44 ` [PATCH v5 2/3] perf mmap: declare type for cpu mask of arbitrary length Alexey Budankov
@ 2019-12-04 13:49 ` Arnaldo Carvalho de Melo
2019-12-05 7:31 ` Alexey Budankov
2020-01-10 17:53 ` [tip: perf/core] perf mmap: Declare " tip-bot2 for Alexey Budankov
1 sibling, 1 reply; 13+ messages in thread
From: Arnaldo Carvalho de Melo @ 2019-12-04 13:49 UTC (permalink / raw)
To: Alexey Budankov
Cc: Jiri Olsa, Namhyung Kim, Alexander Shishkin, Peter Zijlstra,
Ingo Molnar, Andi Kleen, linux-kernel
Em Tue, Dec 03, 2019 at 02:44:18PM +0300, Alexey Budankov escreveu:
>
> Declare a dedicated struct map_cpu_mask type for cpu masks of
> arbitrary length. Mask is available thru bits pointer and the
> mask length is kept in nbits field. MMAP_CPU_MASK_BYTES() macro
> returns mask storage size in bytes. mmap_cpu_mask__scnprintf()
> function can be used to log text representation of the mask.
>
> Signed-off-by: Alexey Budankov <alexey.budankov@linux.intel.com>
> ---
> tools/perf/util/mmap.c | 12 ++++++++++++
> tools/perf/util/mmap.h | 11 +++++++++++
> 2 files changed, 23 insertions(+)
>
> diff --git a/tools/perf/util/mmap.c b/tools/perf/util/mmap.c
> index 063d1b93c53d..43c12b4a3e17 100644
> --- a/tools/perf/util/mmap.c
> +++ b/tools/perf/util/mmap.c
> @@ -23,6 +23,18 @@
> #include "mmap.h"
> #include "../perf.h"
> #include <internal/lib.h> /* page_size */
> +#include <linux/bitmap.h>
> +
> +#define MASK_SIZE 1023
> +void mmap_cpu_mask__scnprintf(struct mmap_cpu_mask *mask, const char *tag)
> +{
> + char buf[MASK_SIZE + 1];
> + size_t len;
> +
> + len = bitmap_scnprintf(mask->bits, mask->nbits, buf, MASK_SIZE);
> + buf[len] = '\0';
> + pr_debug("%p: %s mask[%ld]: %s\n", mask, tag, mask->nbits, buf);
> +}
Above should also be %zd, fixed.
- Arnaldo
^ permalink raw reply [flat|nested] 13+ messages in thread* Re: [PATCH v5 2/3] perf mmap: declare type for cpu mask of arbitrary length
2019-12-04 13:49 ` Arnaldo Carvalho de Melo
@ 2019-12-05 7:31 ` Alexey Budankov
0 siblings, 0 replies; 13+ messages in thread
From: Alexey Budankov @ 2019-12-05 7:31 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo, Jiri Olsa
Cc: Namhyung Kim, Alexander Shishkin, Peter Zijlstra, Ingo Molnar,
Andi Kleen, linux-kernel
On 04.12.2019 16:49, Arnaldo Carvalho de Melo wrote:
> Em Tue, Dec 03, 2019 at 02:44:18PM +0300, Alexey Budankov escreveu:
>>
>> Declare a dedicated struct map_cpu_mask type for cpu masks of
>> arbitrary length. Mask is available thru bits pointer and the
>> mask length is kept in nbits field. MMAP_CPU_MASK_BYTES() macro
>> returns mask storage size in bytes. mmap_cpu_mask__scnprintf()
>> function can be used to log text representation of the mask.
>>
>> Signed-off-by: Alexey Budankov <alexey.budankov@linux.intel.com>
>> ---
>> tools/perf/util/mmap.c | 12 ++++++++++++
>> tools/perf/util/mmap.h | 11 +++++++++++
>> 2 files changed, 23 insertions(+)
>>
>> diff --git a/tools/perf/util/mmap.c b/tools/perf/util/mmap.c
>> index 063d1b93c53d..43c12b4a3e17 100644
>> --- a/tools/perf/util/mmap.c
>> +++ b/tools/perf/util/mmap.c
>> @@ -23,6 +23,18 @@
>> #include "mmap.h"
>> #include "../perf.h"
>> #include <internal/lib.h> /* page_size */
>> +#include <linux/bitmap.h>
>> +
>> +#define MASK_SIZE 1023
>> +void mmap_cpu_mask__scnprintf(struct mmap_cpu_mask *mask, const char *tag)
>> +{
>> + char buf[MASK_SIZE + 1];
>> + size_t len;
>> +
>> + len = bitmap_scnprintf(mask->bits, mask->nbits, buf, MASK_SIZE);
>> + buf[len] = '\0';
>> + pr_debug("%p: %s mask[%ld]: %s\n", mask, tag, mask->nbits, buf);
>> +}
>
> Above should also be %zd, fixed.
Thanks Arnaldo, Jiri! Appreciate you collaboration and help.
~Alexey
>
> - Arnaldo
>
^ permalink raw reply [flat|nested] 13+ messages in thread
* [tip: perf/core] perf mmap: Declare type for cpu mask of arbitrary length
2019-12-03 11:44 ` [PATCH v5 2/3] perf mmap: declare type for cpu mask of arbitrary length Alexey Budankov
2019-12-04 13:49 ` Arnaldo Carvalho de Melo
@ 2020-01-10 17:53 ` tip-bot2 for Alexey Budankov
1 sibling, 0 replies; 13+ messages in thread
From: tip-bot2 for Alexey Budankov @ 2020-01-10 17:53 UTC (permalink / raw)
To: linux-tip-commits
Cc: Alexey Budankov, Jiri Olsa, Alexander Shishkin, Andi Kleen,
Namhyung Kim, Peter Zijlstra, Arnaldo Carvalho de Melo, x86,
LKML
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 9c080c0279a80057cad3dfc05d09fb283ddf72f4
Gitweb: https://git.kernel.org/tip/9c080c0279a80057cad3dfc05d09fb283ddf72f4
Author: Alexey Budankov <alexey.budankov@linux.intel.com>
AuthorDate: Tue, 03 Dec 2019 14:44:18 +03:00
Committer: Arnaldo Carvalho de Melo <acme@redhat.com>
CommitterDate: Mon, 06 Jan 2020 11:46:09 -03:00
perf mmap: Declare type for cpu mask of arbitrary length
Declare a dedicated struct map_cpu_mask type for cpu masks of arbitrary
length.
The mask is available thru bits pointer and the mask length is kept in
nbits field. MMAP_CPU_MASK_BYTES() macro returns mask storage size in
bytes.
The mmap_cpu_mask__scnprintf() function can be used to log text
representation of the mask.
Committer notes:
To print the 'nbits' struct member we must use %zd, since it is a
size_t, this fixes the build in some toolchains/arches.
Signed-off-by: Alexey Budankov <alexey.budankov@linux.intel.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/0fd2454f-477f-d15a-f4ee-79bcbd2585ff@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/util/mmap.c | 12 ++++++++++++
tools/perf/util/mmap.h | 11 +++++++++++
2 files changed, 23 insertions(+)
diff --git a/tools/perf/util/mmap.c b/tools/perf/util/mmap.c
index 063d1b9..2ee4faa 100644
--- a/tools/perf/util/mmap.c
+++ b/tools/perf/util/mmap.c
@@ -23,6 +23,18 @@
#include "mmap.h"
#include "../perf.h"
#include <internal/lib.h> /* page_size */
+#include <linux/bitmap.h>
+
+#define MASK_SIZE 1023
+void mmap_cpu_mask__scnprintf(struct mmap_cpu_mask *mask, const char *tag)
+{
+ char buf[MASK_SIZE + 1];
+ size_t len;
+
+ len = bitmap_scnprintf(mask->bits, mask->nbits, buf, MASK_SIZE);
+ buf[len] = '\0';
+ pr_debug("%p: %s mask[%zd]: %s\n", mask, tag, mask->nbits, buf);
+}
size_t mmap__mmap_len(struct mmap *map)
{
diff --git a/tools/perf/util/mmap.h b/tools/perf/util/mmap.h
index bee4e83..ef51667 100644
--- a/tools/perf/util/mmap.h
+++ b/tools/perf/util/mmap.h
@@ -15,6 +15,15 @@
#include "event.h"
struct aiocb;
+
+struct mmap_cpu_mask {
+ unsigned long *bits;
+ size_t nbits;
+};
+
+#define MMAP_CPU_MASK_BYTES(m) \
+ (BITS_TO_LONGS(((struct mmap_cpu_mask *)m)->nbits) * sizeof(unsigned long))
+
/**
* struct mmap - perf's ring buffer mmap details
*
@@ -52,4 +61,6 @@ int perf_mmap__push(struct mmap *md, void *to,
size_t mmap__mmap_len(struct mmap *map);
+void mmap_cpu_mask__scnprintf(struct mmap_cpu_mask *mask, const char *tag);
+
#endif /*__PERF_MMAP_H */
^ permalink raw reply related [flat|nested] 13+ messages in thread
* [PATCH v5 3/3] perf record: adapt affinity to machines with #CPUs > 1K
2019-12-03 11:41 [PATCH v5 0/3] perf record: adapt NUMA awareness to machines with #CPUs > 1K Alexey Budankov
2019-12-03 11:43 ` [PATCH v5 1/3] tools bitmap: implement bitmap_equal() operation at bitmap API Alexey Budankov
2019-12-03 11:44 ` [PATCH v5 2/3] perf mmap: declare type for cpu mask of arbitrary length Alexey Budankov
@ 2019-12-03 11:45 ` Alexey Budankov
2019-12-04 13:48 ` Arnaldo Carvalho de Melo
2020-01-10 17:53 ` [tip: perf/core] perf record: Adapt " tip-bot2 for Alexey Budankov
2019-12-03 12:17 ` [PATCH v5 0/3] perf record: adapt NUMA awareness " Jiri Olsa
3 siblings, 2 replies; 13+ messages in thread
From: Alexey Budankov @ 2019-12-03 11:45 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo
Cc: Jiri Olsa, Namhyung Kim, Alexander Shishkin, Peter Zijlstra,
Ingo Molnar, Andi Kleen, linux-kernel
Use struct mmap_cpu_mask type for tool's thread and mmap data
buffers to overcome current 1024 CPUs mask size limitation of
cpu_set_t type.
Currently glibc cpu_set_t type has internal mask size limit
of 1024 CPUs. Moving to struct mmap_cpu_mask type allows
overcoming that limit. tools bitmap API is used to manipulate
objects of struct mmap_cpu_mask type.
Reported-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Alexey Budankov <alexey.budankov@linux.intel.com>
---
tools/perf/builtin-record.c | 28 ++++++++++++++++++++++------
tools/perf/util/mmap.c | 28 ++++++++++++++++++++++------
tools/perf/util/mmap.h | 2 +-
3 files changed, 45 insertions(+), 13 deletions(-)
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index fb19ef63cc35..7bc83755ef8c 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -62,6 +62,7 @@
#include <linux/string.h>
#include <linux/time64.h>
#include <linux/zalloc.h>
+#include <linux/bitmap.h>
struct switch_output {
bool enabled;
@@ -93,7 +94,7 @@ struct record {
bool timestamp_boundary;
struct switch_output switch_output;
unsigned long long samples;
- cpu_set_t affinity_mask;
+ struct mmap_cpu_mask affinity_mask;
unsigned long output_max_size; /* = 0: unlimited */
};
@@ -961,10 +962,15 @@ static struct perf_event_header finished_round_event = {
static void record__adjust_affinity(struct record *rec, struct mmap *map)
{
if (rec->opts.affinity != PERF_AFFINITY_SYS &&
- !CPU_EQUAL(&rec->affinity_mask, &map->affinity_mask)) {
- CPU_ZERO(&rec->affinity_mask);
- CPU_OR(&rec->affinity_mask, &rec->affinity_mask, &map->affinity_mask);
- sched_setaffinity(0, sizeof(rec->affinity_mask), &rec->affinity_mask);
+ !bitmap_equal(rec->affinity_mask.bits, map->affinity_mask.bits,
+ rec->affinity_mask.nbits)) {
+ bitmap_zero(rec->affinity_mask.bits, rec->affinity_mask.nbits);
+ bitmap_or(rec->affinity_mask.bits, rec->affinity_mask.bits,
+ map->affinity_mask.bits, rec->affinity_mask.nbits);
+ sched_setaffinity(0, MMAP_CPU_MASK_BYTES(&rec->affinity_mask),
+ (cpu_set_t *)rec->affinity_mask.bits);
+ if (verbose == 2)
+ mmap_cpu_mask__scnprintf(&rec->affinity_mask, "thread");
}
}
@@ -2433,7 +2439,6 @@ int cmd_record(int argc, const char **argv)
# undef REASON
#endif
- CPU_ZERO(&rec->affinity_mask);
rec->opts.affinity = PERF_AFFINITY_SYS;
rec->evlist = evlist__new();
@@ -2499,6 +2504,16 @@ int cmd_record(int argc, const char **argv)
symbol__init(NULL);
+ if (rec->opts.affinity != PERF_AFFINITY_SYS) {
+ rec->affinity_mask.nbits = cpu__max_cpu();
+ rec->affinity_mask.bits = bitmap_alloc(rec->affinity_mask.nbits);
+ if (!rec->affinity_mask.bits) {
+ pr_err("Failed to allocate thread mask for %ld cpus\n", rec->affinity_mask.nbits);
+ return -ENOMEM;
+ }
+ pr_debug2("thread mask[%ld]: empty\n", rec->affinity_mask.nbits);
+ }
+
err = record__auxtrace_init(rec);
if (err)
goto out;
@@ -2613,6 +2628,7 @@ int cmd_record(int argc, const char **argv)
err = __cmd_record(&record, argc, argv);
out:
+ bitmap_free(rec->affinity_mask.bits);
evlist__delete(rec->evlist);
symbol__exit();
auxtrace_record__free(rec->itr);
diff --git a/tools/perf/util/mmap.c b/tools/perf/util/mmap.c
index 43c12b4a3e17..832d2cb94b2c 100644
--- a/tools/perf/util/mmap.c
+++ b/tools/perf/util/mmap.c
@@ -219,6 +219,8 @@ static void perf_mmap__aio_munmap(struct mmap *map __maybe_unused)
void mmap__munmap(struct mmap *map)
{
+ bitmap_free(map->affinity_mask.bits);
+
perf_mmap__aio_munmap(map);
if (map->data != NULL) {
munmap(map->data, mmap__mmap_len(map));
@@ -227,7 +229,7 @@ void mmap__munmap(struct mmap *map)
auxtrace_mmap__munmap(&map->auxtrace_mmap);
}
-static void build_node_mask(int node, cpu_set_t *mask)
+static void build_node_mask(int node, struct mmap_cpu_mask *mask)
{
int c, cpu, nr_cpus;
const struct perf_cpu_map *cpu_map = NULL;
@@ -240,17 +242,23 @@ static void build_node_mask(int node, cpu_set_t *mask)
for (c = 0; c < nr_cpus; c++) {
cpu = cpu_map->map[c]; /* map c index to online cpu index */
if (cpu__get_node(cpu) == node)
- CPU_SET(cpu, mask);
+ set_bit(cpu, mask->bits);
}
}
-static void perf_mmap__setup_affinity_mask(struct mmap *map, struct mmap_params *mp)
+static int perf_mmap__setup_affinity_mask(struct mmap *map, struct mmap_params *mp)
{
- CPU_ZERO(&map->affinity_mask);
+ map->affinity_mask.nbits = cpu__max_cpu();
+ map->affinity_mask.bits = bitmap_alloc(map->affinity_mask.nbits);
+ if (!map->affinity_mask.bits)
+ return -1;
+
if (mp->affinity == PERF_AFFINITY_NODE && cpu__max_node() > 1)
build_node_mask(cpu__get_node(map->core.cpu), &map->affinity_mask);
else if (mp->affinity == PERF_AFFINITY_CPU)
- CPU_SET(map->core.cpu, &map->affinity_mask);
+ set_bit(map->core.cpu, map->affinity_mask.bits);
+
+ return 0;
}
int mmap__mmap(struct mmap *map, struct mmap_params *mp, int fd, int cpu)
@@ -261,7 +269,15 @@ int mmap__mmap(struct mmap *map, struct mmap_params *mp, int fd, int cpu)
return -1;
}
- perf_mmap__setup_affinity_mask(map, mp);
+ if (mp->affinity != PERF_AFFINITY_SYS &&
+ perf_mmap__setup_affinity_mask(map, mp)) {
+ pr_debug2("failed to alloc mmap affinity mask, error %d\n",
+ errno);
+ return -1;
+ }
+
+ if (verbose == 2)
+ mmap_cpu_mask__scnprintf(&map->affinity_mask, "mmap");
map->core.flush = mp->flush;
diff --git a/tools/perf/util/mmap.h b/tools/perf/util/mmap.h
index ef51667fabcb..9d5f589f02ae 100644
--- a/tools/perf/util/mmap.h
+++ b/tools/perf/util/mmap.h
@@ -40,7 +40,7 @@ struct mmap {
int nr_cblocks;
} aio;
#endif
- cpu_set_t affinity_mask;
+ struct mmap_cpu_mask affinity_mask;
void *data;
int comp_level;
};
--
2.20.1
^ permalink raw reply related [flat|nested] 13+ messages in thread* Re: [PATCH v5 3/3] perf record: adapt affinity to machines with #CPUs > 1K
2019-12-03 11:45 ` [PATCH v5 3/3] perf record: adapt affinity to machines with #CPUs > 1K Alexey Budankov
@ 2019-12-04 13:48 ` Arnaldo Carvalho de Melo
2019-12-05 7:30 ` Alexey Budankov
2020-01-10 17:53 ` [tip: perf/core] perf record: Adapt " tip-bot2 for Alexey Budankov
1 sibling, 1 reply; 13+ messages in thread
From: Arnaldo Carvalho de Melo @ 2019-12-04 13:48 UTC (permalink / raw)
To: Alexey Budankov
Cc: Jiri Olsa, Namhyung Kim, Alexander Shishkin, Peter Zijlstra,
Ingo Molnar, Andi Kleen, linux-kernel
Em Tue, Dec 03, 2019 at 02:45:27PM +0300, Alexey Budankov escreveu:
>
> Use struct mmap_cpu_mask type for tool's thread and mmap data
> buffers to overcome current 1024 CPUs mask size limitation of
> cpu_set_t type.
>
> Currently glibc cpu_set_t type has internal mask size limit
> of 1024 CPUs. Moving to struct mmap_cpu_mask type allows
> overcoming that limit. tools bitmap API is used to manipulate
> objects of struct mmap_cpu_mask type.
Had to apply this to fix the build in some toolchains/arches:
[acme@quaco perf]$ git diff
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 7bc83755ef8c..4c301466101b 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -2508,10 +2508,10 @@ int cmd_record(int argc, const char **argv)
rec->affinity_mask.nbits = cpu__max_cpu();
rec->affinity_mask.bits = bitmap_alloc(rec->affinity_mask.nbits);
if (!rec->affinity_mask.bits) {
- pr_err("Failed to allocate thread mask for %ld cpus\n", rec->affinity_mask.nbits);
+ pr_err("Failed to allocate thread mask for %zd cpus\n", rec->affinity_mask.nbits);
return -ENOMEM;
}
- pr_debug2("thread mask[%ld]: empty\n", rec->affinity_mask.nbits);
+ pr_debug2("thread mask[%zd]: empty\n", rec->affinity_mask.nbits);
}
err = record__auxtrace_init(rec);
> Reported-by: Andi Kleen <ak@linux.intel.com>
> Signed-off-by: Alexey Budankov <alexey.budankov@linux.intel.com>
> ---
> tools/perf/builtin-record.c | 28 ++++++++++++++++++++++------
> tools/perf/util/mmap.c | 28 ++++++++++++++++++++++------
> tools/perf/util/mmap.h | 2 +-
> 3 files changed, 45 insertions(+), 13 deletions(-)
>
> diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
> index fb19ef63cc35..7bc83755ef8c 100644
> --- a/tools/perf/builtin-record.c
> +++ b/tools/perf/builtin-record.c
> @@ -62,6 +62,7 @@
> #include <linux/string.h>
> #include <linux/time64.h>
> #include <linux/zalloc.h>
> +#include <linux/bitmap.h>
>
> struct switch_output {
> bool enabled;
> @@ -93,7 +94,7 @@ struct record {
> bool timestamp_boundary;
> struct switch_output switch_output;
> unsigned long long samples;
> - cpu_set_t affinity_mask;
> + struct mmap_cpu_mask affinity_mask;
> unsigned long output_max_size; /* = 0: unlimited */
> };
>
> @@ -961,10 +962,15 @@ static struct perf_event_header finished_round_event = {
> static void record__adjust_affinity(struct record *rec, struct mmap *map)
> {
> if (rec->opts.affinity != PERF_AFFINITY_SYS &&
> - !CPU_EQUAL(&rec->affinity_mask, &map->affinity_mask)) {
> - CPU_ZERO(&rec->affinity_mask);
> - CPU_OR(&rec->affinity_mask, &rec->affinity_mask, &map->affinity_mask);
> - sched_setaffinity(0, sizeof(rec->affinity_mask), &rec->affinity_mask);
> + !bitmap_equal(rec->affinity_mask.bits, map->affinity_mask.bits,
> + rec->affinity_mask.nbits)) {
> + bitmap_zero(rec->affinity_mask.bits, rec->affinity_mask.nbits);
> + bitmap_or(rec->affinity_mask.bits, rec->affinity_mask.bits,
> + map->affinity_mask.bits, rec->affinity_mask.nbits);
> + sched_setaffinity(0, MMAP_CPU_MASK_BYTES(&rec->affinity_mask),
> + (cpu_set_t *)rec->affinity_mask.bits);
> + if (verbose == 2)
> + mmap_cpu_mask__scnprintf(&rec->affinity_mask, "thread");
> }
> }
>
> @@ -2433,7 +2439,6 @@ int cmd_record(int argc, const char **argv)
> # undef REASON
> #endif
>
> - CPU_ZERO(&rec->affinity_mask);
> rec->opts.affinity = PERF_AFFINITY_SYS;
>
> rec->evlist = evlist__new();
> @@ -2499,6 +2504,16 @@ int cmd_record(int argc, const char **argv)
>
> symbol__init(NULL);
>
> + if (rec->opts.affinity != PERF_AFFINITY_SYS) {
> + rec->affinity_mask.nbits = cpu__max_cpu();
> + rec->affinity_mask.bits = bitmap_alloc(rec->affinity_mask.nbits);
> + if (!rec->affinity_mask.bits) {
> + pr_err("Failed to allocate thread mask for %ld cpus\n", rec->affinity_mask.nbits);
> + return -ENOMEM;
> + }
> + pr_debug2("thread mask[%ld]: empty\n", rec->affinity_mask.nbits);
> + }
> +
> err = record__auxtrace_init(rec);
> if (err)
> goto out;
> @@ -2613,6 +2628,7 @@ int cmd_record(int argc, const char **argv)
>
> err = __cmd_record(&record, argc, argv);
> out:
> + bitmap_free(rec->affinity_mask.bits);
> evlist__delete(rec->evlist);
> symbol__exit();
> auxtrace_record__free(rec->itr);
> diff --git a/tools/perf/util/mmap.c b/tools/perf/util/mmap.c
> index 43c12b4a3e17..832d2cb94b2c 100644
> --- a/tools/perf/util/mmap.c
> +++ b/tools/perf/util/mmap.c
> @@ -219,6 +219,8 @@ static void perf_mmap__aio_munmap(struct mmap *map __maybe_unused)
>
> void mmap__munmap(struct mmap *map)
> {
> + bitmap_free(map->affinity_mask.bits);
> +
> perf_mmap__aio_munmap(map);
> if (map->data != NULL) {
> munmap(map->data, mmap__mmap_len(map));
> @@ -227,7 +229,7 @@ void mmap__munmap(struct mmap *map)
> auxtrace_mmap__munmap(&map->auxtrace_mmap);
> }
>
> -static void build_node_mask(int node, cpu_set_t *mask)
> +static void build_node_mask(int node, struct mmap_cpu_mask *mask)
> {
> int c, cpu, nr_cpus;
> const struct perf_cpu_map *cpu_map = NULL;
> @@ -240,17 +242,23 @@ static void build_node_mask(int node, cpu_set_t *mask)
> for (c = 0; c < nr_cpus; c++) {
> cpu = cpu_map->map[c]; /* map c index to online cpu index */
> if (cpu__get_node(cpu) == node)
> - CPU_SET(cpu, mask);
> + set_bit(cpu, mask->bits);
> }
> }
>
> -static void perf_mmap__setup_affinity_mask(struct mmap *map, struct mmap_params *mp)
> +static int perf_mmap__setup_affinity_mask(struct mmap *map, struct mmap_params *mp)
> {
> - CPU_ZERO(&map->affinity_mask);
> + map->affinity_mask.nbits = cpu__max_cpu();
> + map->affinity_mask.bits = bitmap_alloc(map->affinity_mask.nbits);
> + if (!map->affinity_mask.bits)
> + return -1;
> +
> if (mp->affinity == PERF_AFFINITY_NODE && cpu__max_node() > 1)
> build_node_mask(cpu__get_node(map->core.cpu), &map->affinity_mask);
> else if (mp->affinity == PERF_AFFINITY_CPU)
> - CPU_SET(map->core.cpu, &map->affinity_mask);
> + set_bit(map->core.cpu, map->affinity_mask.bits);
> +
> + return 0;
> }
>
> int mmap__mmap(struct mmap *map, struct mmap_params *mp, int fd, int cpu)
> @@ -261,7 +269,15 @@ int mmap__mmap(struct mmap *map, struct mmap_params *mp, int fd, int cpu)
> return -1;
> }
>
> - perf_mmap__setup_affinity_mask(map, mp);
> + if (mp->affinity != PERF_AFFINITY_SYS &&
> + perf_mmap__setup_affinity_mask(map, mp)) {
> + pr_debug2("failed to alloc mmap affinity mask, error %d\n",
> + errno);
> + return -1;
> + }
> +
> + if (verbose == 2)
> + mmap_cpu_mask__scnprintf(&map->affinity_mask, "mmap");
>
> map->core.flush = mp->flush;
>
> diff --git a/tools/perf/util/mmap.h b/tools/perf/util/mmap.h
> index ef51667fabcb..9d5f589f02ae 100644
> --- a/tools/perf/util/mmap.h
> +++ b/tools/perf/util/mmap.h
> @@ -40,7 +40,7 @@ struct mmap {
> int nr_cblocks;
> } aio;
> #endif
> - cpu_set_t affinity_mask;
> + struct mmap_cpu_mask affinity_mask;
> void *data;
> int comp_level;
> };
> --
> 2.20.1
>
--
- Arnaldo
^ permalink raw reply related [flat|nested] 13+ messages in thread* Re: [PATCH v5 3/3] perf record: adapt affinity to machines with #CPUs > 1K
2019-12-04 13:48 ` Arnaldo Carvalho de Melo
@ 2019-12-05 7:30 ` Alexey Budankov
0 siblings, 0 replies; 13+ messages in thread
From: Alexey Budankov @ 2019-12-05 7:30 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo
Cc: Jiri Olsa, Namhyung Kim, Alexander Shishkin, Peter Zijlstra,
Ingo Molnar, Andi Kleen, linux-kernel
On 04.12.2019 16:48, Arnaldo Carvalho de Melo wrote:
> Em Tue, Dec 03, 2019 at 02:45:27PM +0300, Alexey Budankov escreveu:
>>
>> Use struct mmap_cpu_mask type for tool's thread and mmap data
>> buffers to overcome current 1024 CPUs mask size limitation of
>> cpu_set_t type.
>>
>> Currently glibc cpu_set_t type has internal mask size limit
>> of 1024 CPUs. Moving to struct mmap_cpu_mask type allows
>> overcoming that limit. tools bitmap API is used to manipulate
>> objects of struct mmap_cpu_mask type.
>
> Had to apply this to fix the build in some toolchains/arches:
>
> [acme@quaco perf]$ git diff
> diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
> index 7bc83755ef8c..4c301466101b 100644
> --- a/tools/perf/builtin-record.c
> +++ b/tools/perf/builtin-record.c
> @@ -2508,10 +2508,10 @@ int cmd_record(int argc, const char **argv)
> rec->affinity_mask.nbits = cpu__max_cpu();
> rec->affinity_mask.bits = bitmap_alloc(rec->affinity_mask.nbits);
> if (!rec->affinity_mask.bits) {
> - pr_err("Failed to allocate thread mask for %ld cpus\n", rec->affinity_mask.nbits);
> + pr_err("Failed to allocate thread mask for %zd cpus\n", rec->affinity_mask.nbits);
> return -ENOMEM;
> }
> - pr_debug2("thread mask[%ld]: empty\n", rec->affinity_mask.nbits);
> + pr_debug2("thread mask[%zd]: empty\n", rec->affinity_mask.nbits);
> }
>
> err = record__auxtrace_init(rec);
Thank you.
~Alexey
>
>
>
>> Reported-by: Andi Kleen <ak@linux.intel.com>
>> Signed-off-by: Alexey Budankov <alexey.budankov@linux.intel.com>
>> ---
>> tools/perf/builtin-record.c | 28 ++++++++++++++++++++++------
>> tools/perf/util/mmap.c | 28 ++++++++++++++++++++++------
>> tools/perf/util/mmap.h | 2 +-
>> 3 files changed, 45 insertions(+), 13 deletions(-)
>>
>> diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
>> index fb19ef63cc35..7bc83755ef8c 100644
>> --- a/tools/perf/builtin-record.c
>> +++ b/tools/perf/builtin-record.c
>> @@ -62,6 +62,7 @@
>> #include <linux/string.h>
>> #include <linux/time64.h>
>> #include <linux/zalloc.h>
>> +#include <linux/bitmap.h>
>>
>> struct switch_output {
>> bool enabled;
>> @@ -93,7 +94,7 @@ struct record {
>> bool timestamp_boundary;
>> struct switch_output switch_output;
>> unsigned long long samples;
>> - cpu_set_t affinity_mask;
>> + struct mmap_cpu_mask affinity_mask;
>> unsigned long output_max_size; /* = 0: unlimited */
>> };
>>
>> @@ -961,10 +962,15 @@ static struct perf_event_header finished_round_event = {
>> static void record__adjust_affinity(struct record *rec, struct mmap *map)
>> {
>> if (rec->opts.affinity != PERF_AFFINITY_SYS &&
>> - !CPU_EQUAL(&rec->affinity_mask, &map->affinity_mask)) {
>> - CPU_ZERO(&rec->affinity_mask);
>> - CPU_OR(&rec->affinity_mask, &rec->affinity_mask, &map->affinity_mask);
>> - sched_setaffinity(0, sizeof(rec->affinity_mask), &rec->affinity_mask);
>> + !bitmap_equal(rec->affinity_mask.bits, map->affinity_mask.bits,
>> + rec->affinity_mask.nbits)) {
>> + bitmap_zero(rec->affinity_mask.bits, rec->affinity_mask.nbits);
>> + bitmap_or(rec->affinity_mask.bits, rec->affinity_mask.bits,
>> + map->affinity_mask.bits, rec->affinity_mask.nbits);
>> + sched_setaffinity(0, MMAP_CPU_MASK_BYTES(&rec->affinity_mask),
>> + (cpu_set_t *)rec->affinity_mask.bits);
>> + if (verbose == 2)
>> + mmap_cpu_mask__scnprintf(&rec->affinity_mask, "thread");
>> }
>> }
>>
>> @@ -2433,7 +2439,6 @@ int cmd_record(int argc, const char **argv)
>> # undef REASON
>> #endif
>>
>> - CPU_ZERO(&rec->affinity_mask);
>> rec->opts.affinity = PERF_AFFINITY_SYS;
>>
>> rec->evlist = evlist__new();
>> @@ -2499,6 +2504,16 @@ int cmd_record(int argc, const char **argv)
>>
>> symbol__init(NULL);
>>
>> + if (rec->opts.affinity != PERF_AFFINITY_SYS) {
>> + rec->affinity_mask.nbits = cpu__max_cpu();
>> + rec->affinity_mask.bits = bitmap_alloc(rec->affinity_mask.nbits);
>> + if (!rec->affinity_mask.bits) {
>> + pr_err("Failed to allocate thread mask for %ld cpus\n", rec->affinity_mask.nbits);
>> + return -ENOMEM;
>> + }
>> + pr_debug2("thread mask[%ld]: empty\n", rec->affinity_mask.nbits);
>> + }
>> +
>> err = record__auxtrace_init(rec);
>> if (err)
>> goto out;
>> @@ -2613,6 +2628,7 @@ int cmd_record(int argc, const char **argv)
>>
>> err = __cmd_record(&record, argc, argv);
>> out:
>> + bitmap_free(rec->affinity_mask.bits);
>> evlist__delete(rec->evlist);
>> symbol__exit();
>> auxtrace_record__free(rec->itr);
>> diff --git a/tools/perf/util/mmap.c b/tools/perf/util/mmap.c
>> index 43c12b4a3e17..832d2cb94b2c 100644
>> --- a/tools/perf/util/mmap.c
>> +++ b/tools/perf/util/mmap.c
>> @@ -219,6 +219,8 @@ static void perf_mmap__aio_munmap(struct mmap *map __maybe_unused)
>>
>> void mmap__munmap(struct mmap *map)
>> {
>> + bitmap_free(map->affinity_mask.bits);
>> +
>> perf_mmap__aio_munmap(map);
>> if (map->data != NULL) {
>> munmap(map->data, mmap__mmap_len(map));
>> @@ -227,7 +229,7 @@ void mmap__munmap(struct mmap *map)
>> auxtrace_mmap__munmap(&map->auxtrace_mmap);
>> }
>>
>> -static void build_node_mask(int node, cpu_set_t *mask)
>> +static void build_node_mask(int node, struct mmap_cpu_mask *mask)
>> {
>> int c, cpu, nr_cpus;
>> const struct perf_cpu_map *cpu_map = NULL;
>> @@ -240,17 +242,23 @@ static void build_node_mask(int node, cpu_set_t *mask)
>> for (c = 0; c < nr_cpus; c++) {
>> cpu = cpu_map->map[c]; /* map c index to online cpu index */
>> if (cpu__get_node(cpu) == node)
>> - CPU_SET(cpu, mask);
>> + set_bit(cpu, mask->bits);
>> }
>> }
>>
>> -static void perf_mmap__setup_affinity_mask(struct mmap *map, struct mmap_params *mp)
>> +static int perf_mmap__setup_affinity_mask(struct mmap *map, struct mmap_params *mp)
>> {
>> - CPU_ZERO(&map->affinity_mask);
>> + map->affinity_mask.nbits = cpu__max_cpu();
>> + map->affinity_mask.bits = bitmap_alloc(map->affinity_mask.nbits);
>> + if (!map->affinity_mask.bits)
>> + return -1;
>> +
>> if (mp->affinity == PERF_AFFINITY_NODE && cpu__max_node() > 1)
>> build_node_mask(cpu__get_node(map->core.cpu), &map->affinity_mask);
>> else if (mp->affinity == PERF_AFFINITY_CPU)
>> - CPU_SET(map->core.cpu, &map->affinity_mask);
>> + set_bit(map->core.cpu, map->affinity_mask.bits);
>> +
>> + return 0;
>> }
>>
>> int mmap__mmap(struct mmap *map, struct mmap_params *mp, int fd, int cpu)
>> @@ -261,7 +269,15 @@ int mmap__mmap(struct mmap *map, struct mmap_params *mp, int fd, int cpu)
>> return -1;
>> }
>>
>> - perf_mmap__setup_affinity_mask(map, mp);
>> + if (mp->affinity != PERF_AFFINITY_SYS &&
>> + perf_mmap__setup_affinity_mask(map, mp)) {
>> + pr_debug2("failed to alloc mmap affinity mask, error %d\n",
>> + errno);
>> + return -1;
>> + }
>> +
>> + if (verbose == 2)
>> + mmap_cpu_mask__scnprintf(&map->affinity_mask, "mmap");
>>
>> map->core.flush = mp->flush;
>>
>> diff --git a/tools/perf/util/mmap.h b/tools/perf/util/mmap.h
>> index ef51667fabcb..9d5f589f02ae 100644
>> --- a/tools/perf/util/mmap.h
>> +++ b/tools/perf/util/mmap.h
>> @@ -40,7 +40,7 @@ struct mmap {
>> int nr_cblocks;
>> } aio;
>> #endif
>> - cpu_set_t affinity_mask;
>> + struct mmap_cpu_mask affinity_mask;
>> void *data;
>> int comp_level;
>> };
>> --
>> 2.20.1
>>
>
^ permalink raw reply [flat|nested] 13+ messages in thread
* [tip: perf/core] perf record: Adapt affinity to machines with #CPUs > 1K
2019-12-03 11:45 ` [PATCH v5 3/3] perf record: adapt affinity to machines with #CPUs > 1K Alexey Budankov
2019-12-04 13:48 ` Arnaldo Carvalho de Melo
@ 2020-01-10 17:53 ` tip-bot2 for Alexey Budankov
1 sibling, 0 replies; 13+ messages in thread
From: tip-bot2 for Alexey Budankov @ 2020-01-10 17:53 UTC (permalink / raw)
To: linux-tip-commits
Cc: Andi Kleen, Alexey Budankov, Jiri Olsa, Alexander Shishkin,
Namhyung Kim, Peter Zijlstra, Arnaldo Carvalho de Melo, x86,
LKML
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 8384a2600c7ddfc875f64e160d8b423aca4e203a
Gitweb: https://git.kernel.org/tip/8384a2600c7ddfc875f64e160d8b423aca4e203a
Author: Alexey Budankov <alexey.budankov@linux.intel.com>
AuthorDate: Tue, 03 Dec 2019 14:45:27 +03:00
Committer: Arnaldo Carvalho de Melo <acme@redhat.com>
CommitterDate: Mon, 06 Jan 2020 11:46:09 -03:00
perf record: Adapt affinity to machines with #CPUs > 1K
Use struct mmap_cpu_mask type for the tool's thread and mmap data
buffers to overcome current 1024 CPUs mask size limitation of cpu_set_t
type.
Currently glibc's cpu_set_t type has an internal mask size limit of 1024
CPUs.
Moving to the 'struct mmap_cpu_mask' type allows overcoming that limit.
The tools bitmap API is used to manipulate objects of 'struct mmap_cpu_mask'
type.
Committer notes:
To print the 'nbits' struct member we must use %zd, since it is a
size_t, this fixes the build in some toolchains/arches.
Reported-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Alexey Budankov <alexey.budankov@linux.intel.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/96d7e2ff-ce8b-c1e0-d52c-aa59ea96f0ea@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/builtin-record.c | 28 ++++++++++++++++++++++------
tools/perf/util/mmap.c | 28 ++++++++++++++++++++++------
tools/perf/util/mmap.h | 2 +-
3 files changed, 45 insertions(+), 13 deletions(-)
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index fb19ef6..4c30146 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -62,6 +62,7 @@
#include <linux/string.h>
#include <linux/time64.h>
#include <linux/zalloc.h>
+#include <linux/bitmap.h>
struct switch_output {
bool enabled;
@@ -93,7 +94,7 @@ struct record {
bool timestamp_boundary;
struct switch_output switch_output;
unsigned long long samples;
- cpu_set_t affinity_mask;
+ struct mmap_cpu_mask affinity_mask;
unsigned long output_max_size; /* = 0: unlimited */
};
@@ -961,10 +962,15 @@ static struct perf_event_header finished_round_event = {
static void record__adjust_affinity(struct record *rec, struct mmap *map)
{
if (rec->opts.affinity != PERF_AFFINITY_SYS &&
- !CPU_EQUAL(&rec->affinity_mask, &map->affinity_mask)) {
- CPU_ZERO(&rec->affinity_mask);
- CPU_OR(&rec->affinity_mask, &rec->affinity_mask, &map->affinity_mask);
- sched_setaffinity(0, sizeof(rec->affinity_mask), &rec->affinity_mask);
+ !bitmap_equal(rec->affinity_mask.bits, map->affinity_mask.bits,
+ rec->affinity_mask.nbits)) {
+ bitmap_zero(rec->affinity_mask.bits, rec->affinity_mask.nbits);
+ bitmap_or(rec->affinity_mask.bits, rec->affinity_mask.bits,
+ map->affinity_mask.bits, rec->affinity_mask.nbits);
+ sched_setaffinity(0, MMAP_CPU_MASK_BYTES(&rec->affinity_mask),
+ (cpu_set_t *)rec->affinity_mask.bits);
+ if (verbose == 2)
+ mmap_cpu_mask__scnprintf(&rec->affinity_mask, "thread");
}
}
@@ -2433,7 +2439,6 @@ int cmd_record(int argc, const char **argv)
# undef REASON
#endif
- CPU_ZERO(&rec->affinity_mask);
rec->opts.affinity = PERF_AFFINITY_SYS;
rec->evlist = evlist__new();
@@ -2499,6 +2504,16 @@ int cmd_record(int argc, const char **argv)
symbol__init(NULL);
+ if (rec->opts.affinity != PERF_AFFINITY_SYS) {
+ rec->affinity_mask.nbits = cpu__max_cpu();
+ rec->affinity_mask.bits = bitmap_alloc(rec->affinity_mask.nbits);
+ if (!rec->affinity_mask.bits) {
+ pr_err("Failed to allocate thread mask for %zd cpus\n", rec->affinity_mask.nbits);
+ return -ENOMEM;
+ }
+ pr_debug2("thread mask[%zd]: empty\n", rec->affinity_mask.nbits);
+ }
+
err = record__auxtrace_init(rec);
if (err)
goto out;
@@ -2613,6 +2628,7 @@ int cmd_record(int argc, const char **argv)
err = __cmd_record(&record, argc, argv);
out:
+ bitmap_free(rec->affinity_mask.bits);
evlist__delete(rec->evlist);
symbol__exit();
auxtrace_record__free(rec->itr);
diff --git a/tools/perf/util/mmap.c b/tools/perf/util/mmap.c
index 2ee4faa..3b664fa 100644
--- a/tools/perf/util/mmap.c
+++ b/tools/perf/util/mmap.c
@@ -219,6 +219,8 @@ static void perf_mmap__aio_munmap(struct mmap *map __maybe_unused)
void mmap__munmap(struct mmap *map)
{
+ bitmap_free(map->affinity_mask.bits);
+
perf_mmap__aio_munmap(map);
if (map->data != NULL) {
munmap(map->data, mmap__mmap_len(map));
@@ -227,7 +229,7 @@ void mmap__munmap(struct mmap *map)
auxtrace_mmap__munmap(&map->auxtrace_mmap);
}
-static void build_node_mask(int node, cpu_set_t *mask)
+static void build_node_mask(int node, struct mmap_cpu_mask *mask)
{
int c, cpu, nr_cpus;
const struct perf_cpu_map *cpu_map = NULL;
@@ -240,17 +242,23 @@ static void build_node_mask(int node, cpu_set_t *mask)
for (c = 0; c < nr_cpus; c++) {
cpu = cpu_map->map[c]; /* map c index to online cpu index */
if (cpu__get_node(cpu) == node)
- CPU_SET(cpu, mask);
+ set_bit(cpu, mask->bits);
}
}
-static void perf_mmap__setup_affinity_mask(struct mmap *map, struct mmap_params *mp)
+static int perf_mmap__setup_affinity_mask(struct mmap *map, struct mmap_params *mp)
{
- CPU_ZERO(&map->affinity_mask);
+ map->affinity_mask.nbits = cpu__max_cpu();
+ map->affinity_mask.bits = bitmap_alloc(map->affinity_mask.nbits);
+ if (!map->affinity_mask.bits)
+ return -1;
+
if (mp->affinity == PERF_AFFINITY_NODE && cpu__max_node() > 1)
build_node_mask(cpu__get_node(map->core.cpu), &map->affinity_mask);
else if (mp->affinity == PERF_AFFINITY_CPU)
- CPU_SET(map->core.cpu, &map->affinity_mask);
+ set_bit(map->core.cpu, map->affinity_mask.bits);
+
+ return 0;
}
int mmap__mmap(struct mmap *map, struct mmap_params *mp, int fd, int cpu)
@@ -261,7 +269,15 @@ int mmap__mmap(struct mmap *map, struct mmap_params *mp, int fd, int cpu)
return -1;
}
- perf_mmap__setup_affinity_mask(map, mp);
+ if (mp->affinity != PERF_AFFINITY_SYS &&
+ perf_mmap__setup_affinity_mask(map, mp)) {
+ pr_debug2("failed to alloc mmap affinity mask, error %d\n",
+ errno);
+ return -1;
+ }
+
+ if (verbose == 2)
+ mmap_cpu_mask__scnprintf(&map->affinity_mask, "mmap");
map->core.flush = mp->flush;
diff --git a/tools/perf/util/mmap.h b/tools/perf/util/mmap.h
index ef51667..9d5f589 100644
--- a/tools/perf/util/mmap.h
+++ b/tools/perf/util/mmap.h
@@ -40,7 +40,7 @@ struct mmap {
int nr_cblocks;
} aio;
#endif
- cpu_set_t affinity_mask;
+ struct mmap_cpu_mask affinity_mask;
void *data;
int comp_level;
};
^ permalink raw reply related [flat|nested] 13+ messages in thread
* Re: [PATCH v5 0/3] perf record: adapt NUMA awareness to machines with #CPUs > 1K
2019-12-03 11:41 [PATCH v5 0/3] perf record: adapt NUMA awareness to machines with #CPUs > 1K Alexey Budankov
` (2 preceding siblings ...)
2019-12-03 11:45 ` [PATCH v5 3/3] perf record: adapt affinity to machines with #CPUs > 1K Alexey Budankov
@ 2019-12-03 12:17 ` Jiri Olsa
2019-12-03 18:36 ` Arnaldo Carvalho de Melo
3 siblings, 1 reply; 13+ messages in thread
From: Jiri Olsa @ 2019-12-03 12:17 UTC (permalink / raw)
To: Alexey Budankov
Cc: Arnaldo Carvalho de Melo, Namhyung Kim, Alexander Shishkin,
Peter Zijlstra, Ingo Molnar, Andi Kleen, linux-kernel
On Tue, Dec 03, 2019 at 02:41:29PM +0300, Alexey Budankov wrote:
>
> Current implementation of cpu_set_t type by glibc has internal cpu
> mask size limitation of no more than 1024 CPUs. This limitation confines
> NUMA awareness of Perf tool in record mode, thru --affinity option,
> to the first 1024 CPUs on machines with larger amount of CPUs.
>
> This patch set enables Perf tool to overcome 1024 CPUs limitation by
> using a dedicated struct mmap_cpu_mask type and applying tool's bitmap
> API operations to manipulate affinity masks of the tool's thread and
> the mmaped data buffers.
>
> tools bitmap API has been extended with bitmap_free() function and
> bitmap_equal() operation whose implementation is derived from the
> kernel one.
>
> ---
> Changes in v5:
> - avoided allocation of mmap affinity masks in case of
> rec->opts.affinity == PERF_AFFINITY_SYS
Acked-by: Jiri Olsa <jolsa@redhat.com>
thanks,
jirka
> Changes in v4:
> - renamed perf_mmap__print_cpu_mask() to mmap_cpu_mask__scnprintf()
> - avoided checking mask bits for NULL prior calling bitmask_free()
> - avoided thread affinity mask allocation for case of
> rec->opts.affinity == PERF_AFFINITY_SYS
> Changes in v3:
> - implemented perf_mmap__print_cpu_mask() function
> - use perf_mmap__print_cpu_mask() to log thread and mmap cpus masks
> when verbose level is equal to 2
> Changes in v2:
> - implemented bitmap_free() for symmetry with bitmap_alloc()
> - capitalized MMAP_CPU_MASK_BYTES() macro
> - returned -1 from perf_mmap__setup_affinity_mask()
> - implemented releasing of masks using bitmap_free()
> - moved debug printing under -vv option
>
> ---
> Alexey Budankov (3):
> tools bitmap: implement bitmap_equal() operation at bitmap API
> perf mmap: declare type for cpu mask of arbitrary length
> perf record: adapt affinity to machines with #CPUs > 1K
>
> tools/include/linux/bitmap.h | 30 +++++++++++++++++++++++++++
> tools/lib/bitmap.c | 15 ++++++++++++++
> tools/perf/builtin-record.c | 28 +++++++++++++++++++------
> tools/perf/util/mmap.c | 40 ++++++++++++++++++++++++++++++------
> tools/perf/util/mmap.h | 13 +++++++++++-
> 5 files changed, 113 insertions(+), 13 deletions(-)
>
> ---
> Validation:
>
> # tools/perf/perf record -vv -- ls
> Using CPUID GenuineIntel-6-5E-3
> intel_pt default config: tsc,mtc,mtc_period=3,psb_period=3,pt,branch
> nr_cblocks: 0
> affinity: SYS
> mmap flush: 1
> comp level: 0
> ------------------------------------------------------------
> perf_event_attr:
> size 120
> { sample_period, sample_freq } 4000
> sample_type IP|TID|TIME|PERIOD
> read_format ID
> disabled 1
> inherit 1
> mmap 1
> comm 1
> freq 1
> enable_on_exec 1
> task 1
> precise_ip 3
> sample_id_all 1
> exclude_guest 1
> mmap2 1
> comm_exec 1
> ksymbol 1
> bpf_event 1
> ------------------------------------------------------------
> sys_perf_event_open: pid 23718 cpu 0 group_fd -1 flags 0x8 = 4
> sys_perf_event_open: pid 23718 cpu 1 group_fd -1 flags 0x8 = 5
> sys_perf_event_open: pid 23718 cpu 2 group_fd -1 flags 0x8 = 6
> sys_perf_event_open: pid 23718 cpu 3 group_fd -1 flags 0x8 = 9
> sys_perf_event_open: pid 23718 cpu 4 group_fd -1 flags 0x8 = 10
> sys_perf_event_open: pid 23718 cpu 5 group_fd -1 flags 0x8 = 11
> sys_perf_event_open: pid 23718 cpu 6 group_fd -1 flags 0x8 = 12
> sys_perf_event_open: pid 23718 cpu 7 group_fd -1 flags 0x8 = 13
> mmap size 528384B
> 0x7f3e06e060b8: mmap mask[8]:
> 0x7f3e06e16180: mmap mask[8]:
> 0x7f3e06e26248: mmap mask[8]:
> 0x7f3e06e36310: mmap mask[8]:
> 0x7f3e06e463d8: mmap mask[8]:
> 0x7f3e06e564a0: mmap mask[8]:
> 0x7f3e06e66568: mmap mask[8]:
> 0x7f3e06e76630: mmap mask[8]:
> ------------------------------------------------------------
> perf_event_attr:
> type 1
> size 120
> config 0x9
> watermark 1
> sample_id_all 1
> bpf_event 1
> { wakeup_events, wakeup_watermark } 1
> ------------------------------------------------------------
> sys_perf_event_open: pid -1 cpu 0 group_fd -1 flags 0x8 = 14
> sys_perf_event_open: pid -1 cpu 1 group_fd -1 flags 0x8 = 15
> sys_perf_event_open: pid -1 cpu 2 group_fd -1 flags 0x8 = 16
> sys_perf_event_open: pid -1 cpu 3 group_fd -1 flags 0x8 = 17
> sys_perf_event_open: pid -1 cpu 4 group_fd -1 flags 0x8 = 18
> sys_perf_event_open: pid -1 cpu 5 group_fd -1 flags 0x8 = 19
> sys_perf_event_open: pid -1 cpu 6 group_fd -1 flags 0x8 = 20
> sys_perf_event_open: pid -1 cpu 7 group_fd -1 flags 0x8 = 21
> mmap size 528384B
> 0x7f3e0697d0b8: mmap mask[8]:
> 0x7f3e0698d180: mmap mask[8]:
> 0x7f3e0699d248: mmap mask[8]:
> 0x7f3e069ad310: mmap mask[8]:
> 0x7f3e069bd3d8: mmap mask[8]:
> 0x7f3e069cd4a0: mmap mask[8]:
> 0x7f3e069dd568: mmap mask[8]:
> 0x7f3e069ed630: mmap mask[8]:
> Synthesizing TSC conversion information
> arch copy Documentation init kernel MAINTAINERS modules.builtin.modinfo perf.data scripts System.map vmlinux
> block COPYING drivers ipc lbuild Makefile modules.order perf.data.old security tools vmlinux.o
> certs CREDITS fs Kbuild lib mm Module.symvers README sound usr
> config-5.2.7-100.fc29.x86_64 crypto include Kconfig LICENSES modules.builtin net samples stdio virt
> [ perf record: Woken up 1 times to write data ]
> Looking at the vmlinux_path (8 entries long)
> Using vmlinux for symbols
> [ perf record: Captured and wrote 0.013 MB perf.data (8 samples) ]
>
> tools/perf/perf record -vv --affinity=cpu -- ls
> thread mask[8]: empty
> Using CPUID GenuineIntel-6-5E-3
> intel_pt default config: tsc,mtc,mtc_period=3,psb_period=3,pt,branch
> nr_cblocks: 0
> affinity: CPU
> mmap flush: 1
> comp level: 0
> ------------------------------------------------------------
> perf_event_attr:
> size 120
> { sample_period, sample_freq } 4000
> sample_type IP|TID|TIME|PERIOD
> read_format ID
> disabled 1
> inherit 1
> mmap 1
> comm 1
> freq 1
> enable_on_exec 1
> task 1
> precise_ip 3
> sample_id_all 1
> exclude_guest 1
> mmap2 1
> comm_exec 1
> ksymbol 1
> bpf_event 1
> ------------------------------------------------------------
> sys_perf_event_open: pid 23713 cpu 0 group_fd -1 flags 0x8 = 4
> sys_perf_event_open: pid 23713 cpu 1 group_fd -1 flags 0x8 = 5
> sys_perf_event_open: pid 23713 cpu 2 group_fd -1 flags 0x8 = 6
> sys_perf_event_open: pid 23713 cpu 3 group_fd -1 flags 0x8 = 9
> sys_perf_event_open: pid 23713 cpu 4 group_fd -1 flags 0x8 = 10
> sys_perf_event_open: pid 23713 cpu 5 group_fd -1 flags 0x8 = 11
> sys_perf_event_open: pid 23713 cpu 6 group_fd -1 flags 0x8 = 12
> sys_perf_event_open: pid 23713 cpu 7 group_fd -1 flags 0x8 = 13
> mmap size 528384B
> 0x7f3e005bc0b8: mmap mask[8]: 0
> 0x7f3e005cc180: mmap mask[8]: 1
> 0x7f3e005dc248: mmap mask[8]: 2
> 0x7f3e005ec310: mmap mask[8]: 3
> 0x7f3e005fc3d8: mmap mask[8]: 4
> 0x7f3e0060c4a0: mmap mask[8]: 5
> 0x7f3e0061c568: mmap mask[8]: 6
> 0x7f3e0062c630: mmap mask[8]: 7
> ------------------------------------------------------------
> perf_event_attr:
> type 1
> size 120
> config 0x9
> watermark 1
> sample_id_all 1
> bpf_event 1
> { wakeup_events, wakeup_watermark } 1
> ------------------------------------------------------------
> sys_perf_event_open: pid -1 cpu 0 group_fd -1 flags 0x8 = 14
> sys_perf_event_open: pid -1 cpu 1 group_fd -1 flags 0x8 = 15
> sys_perf_event_open: pid -1 cpu 2 group_fd -1 flags 0x8 = 16
> sys_perf_event_open: pid -1 cpu 3 group_fd -1 flags 0x8 = 17
> sys_perf_event_open: pid -1 cpu 4 group_fd -1 flags 0x8 = 18
> sys_perf_event_open: pid -1 cpu 5 group_fd -1 flags 0x8 = 19
> sys_perf_event_open: pid -1 cpu 6 group_fd -1 flags 0x8 = 20
> sys_perf_event_open: pid -1 cpu 7 group_fd -1 flags 0x8 = 21
> mmap size 528384B
> 0x7f3e001330b8: mmap mask[8]:
> 0x7f3e00143180: mmap mask[8]:
> 0x7f3e00153248: mmap mask[8]:
> 0x7f3e00163310: mmap mask[8]:
> 0x7f3e001733d8: mmap mask[8]:
> 0x7f3e001834a0: mmap mask[8]:
> 0x7f3e00193568: mmap mask[8]:
> 0x7f3e001a3630: mmap mask[8]:
> Synthesizing TSC conversion information
> 0x9c9ff0: thread mask[8]: 0
> 0x9c9ff0: thread mask[8]: 1
> 0x9c9ff0: thread mask[8]: 2
> 0x9c9ff0: thread mask[8]: 3
> 0x9c9ff0: thread mask[8]: 4
> arch copy Documentation init kernel MAINTAINERS modules.builtin.modinfo perf.data scripts System.map vmlinux
> block COPYING drivers ipc lbuild Makefile modules.order perf.data.old security tools vmlinux.o
> certs CREDITS fs Kbuild lib mm Module.symvers README sound usr
> config-5.2.7-100.fc29.x86_64 crypto include Kconfig LICENSES modules.builtin net samples stdio virt
> 0x9c9ff0: thread mask[8]: 5
> 0x9c9ff0: thread mask[8]: 6
> 0x9c9ff0: thread mask[8]: 7
> 0x9c9ff0: thread mask[8]: 0
> 0x9c9ff0: thread mask[8]: 1
> 0x9c9ff0: thread mask[8]: 2
> 0x9c9ff0: thread mask[8]: 3
> 0x9c9ff0: thread mask[8]: 4
> 0x9c9ff0: thread mask[8]: 5
> 0x9c9ff0: thread mask[8]: 6
> 0x9c9ff0: thread mask[8]: 7
> [ perf record: Woken up 0 times to write data ]
> 0x9c9ff0: thread mask[8]: 0
> 0x9c9ff0: thread mask[8]: 1
> 0x9c9ff0: thread mask[8]: 2
> 0x9c9ff0: thread mask[8]: 3
> 0x9c9ff0: thread mask[8]: 4
> 0x9c9ff0: thread mask[8]: 5
> 0x9c9ff0: thread mask[8]: 6
> 0x9c9ff0: thread mask[8]: 7
> Looking at the vmlinux_path (8 entries long)
> Using vmlinux for symbols
> ...
> [ perf record: Captured and wrote 0.013 MB perf.data (10 samples) ]
>
^ permalink raw reply [flat|nested] 13+ messages in thread* Re: [PATCH v5 0/3] perf record: adapt NUMA awareness to machines with #CPUs > 1K
2019-12-03 12:17 ` [PATCH v5 0/3] perf record: adapt NUMA awareness " Jiri Olsa
@ 2019-12-03 18:36 ` Arnaldo Carvalho de Melo
0 siblings, 0 replies; 13+ messages in thread
From: Arnaldo Carvalho de Melo @ 2019-12-03 18:36 UTC (permalink / raw)
To: Jiri Olsa
Cc: Alexey Budankov, Namhyung Kim, Alexander Shishkin,
Peter Zijlstra, Ingo Molnar, Andi Kleen, linux-kernel
Em Tue, Dec 03, 2019 at 01:17:45PM +0100, Jiri Olsa escreveu:
> On Tue, Dec 03, 2019 at 02:41:29PM +0300, Alexey Budankov wrote:
> >
> > Current implementation of cpu_set_t type by glibc has internal cpu
> > mask size limitation of no more than 1024 CPUs. This limitation confines
> > NUMA awareness of Perf tool in record mode, thru --affinity option,
> > to the first 1024 CPUs on machines with larger amount of CPUs.
> >
> > This patch set enables Perf tool to overcome 1024 CPUs limitation by
> > using a dedicated struct mmap_cpu_mask type and applying tool's bitmap
> > API operations to manipulate affinity masks of the tool's thread and
> > the mmaped data buffers.
> >
> > tools bitmap API has been extended with bitmap_free() function and
> > bitmap_equal() operation whose implementation is derived from the
> > kernel one.
> >
> > ---
> > Changes in v5:
> > - avoided allocation of mmap affinity masks in case of
> > rec->opts.affinity == PERF_AFFINITY_SYS
>
> Acked-by: Jiri Olsa <jolsa@redhat.com>
Applied to my local perf/core branch, going thru tests.
- Arnaldo
> thanks,
> jirka
>
> > Changes in v4:
> > - renamed perf_mmap__print_cpu_mask() to mmap_cpu_mask__scnprintf()
> > - avoided checking mask bits for NULL prior calling bitmask_free()
> > - avoided thread affinity mask allocation for case of
> > rec->opts.affinity == PERF_AFFINITY_SYS
> > Changes in v3:
> > - implemented perf_mmap__print_cpu_mask() function
> > - use perf_mmap__print_cpu_mask() to log thread and mmap cpus masks
> > when verbose level is equal to 2
> > Changes in v2:
> > - implemented bitmap_free() for symmetry with bitmap_alloc()
> > - capitalized MMAP_CPU_MASK_BYTES() macro
> > - returned -1 from perf_mmap__setup_affinity_mask()
> > - implemented releasing of masks using bitmap_free()
> > - moved debug printing under -vv option
> >
> > ---
> > Alexey Budankov (3):
> > tools bitmap: implement bitmap_equal() operation at bitmap API
> > perf mmap: declare type for cpu mask of arbitrary length
> > perf record: adapt affinity to machines with #CPUs > 1K
> >
> > tools/include/linux/bitmap.h | 30 +++++++++++++++++++++++++++
> > tools/lib/bitmap.c | 15 ++++++++++++++
> > tools/perf/builtin-record.c | 28 +++++++++++++++++++------
> > tools/perf/util/mmap.c | 40 ++++++++++++++++++++++++++++++------
> > tools/perf/util/mmap.h | 13 +++++++++++-
> > 5 files changed, 113 insertions(+), 13 deletions(-)
> >
> > ---
> > Validation:
> >
> > # tools/perf/perf record -vv -- ls
> > Using CPUID GenuineIntel-6-5E-3
> > intel_pt default config: tsc,mtc,mtc_period=3,psb_period=3,pt,branch
> > nr_cblocks: 0
> > affinity: SYS
> > mmap flush: 1
> > comp level: 0
> > ------------------------------------------------------------
> > perf_event_attr:
> > size 120
> > { sample_period, sample_freq } 4000
> > sample_type IP|TID|TIME|PERIOD
> > read_format ID
> > disabled 1
> > inherit 1
> > mmap 1
> > comm 1
> > freq 1
> > enable_on_exec 1
> > task 1
> > precise_ip 3
> > sample_id_all 1
> > exclude_guest 1
> > mmap2 1
> > comm_exec 1
> > ksymbol 1
> > bpf_event 1
> > ------------------------------------------------------------
> > sys_perf_event_open: pid 23718 cpu 0 group_fd -1 flags 0x8 = 4
> > sys_perf_event_open: pid 23718 cpu 1 group_fd -1 flags 0x8 = 5
> > sys_perf_event_open: pid 23718 cpu 2 group_fd -1 flags 0x8 = 6
> > sys_perf_event_open: pid 23718 cpu 3 group_fd -1 flags 0x8 = 9
> > sys_perf_event_open: pid 23718 cpu 4 group_fd -1 flags 0x8 = 10
> > sys_perf_event_open: pid 23718 cpu 5 group_fd -1 flags 0x8 = 11
> > sys_perf_event_open: pid 23718 cpu 6 group_fd -1 flags 0x8 = 12
> > sys_perf_event_open: pid 23718 cpu 7 group_fd -1 flags 0x8 = 13
> > mmap size 528384B
> > 0x7f3e06e060b8: mmap mask[8]:
> > 0x7f3e06e16180: mmap mask[8]:
> > 0x7f3e06e26248: mmap mask[8]:
> > 0x7f3e06e36310: mmap mask[8]:
> > 0x7f3e06e463d8: mmap mask[8]:
> > 0x7f3e06e564a0: mmap mask[8]:
> > 0x7f3e06e66568: mmap mask[8]:
> > 0x7f3e06e76630: mmap mask[8]:
> > ------------------------------------------------------------
> > perf_event_attr:
> > type 1
> > size 120
> > config 0x9
> > watermark 1
> > sample_id_all 1
> > bpf_event 1
> > { wakeup_events, wakeup_watermark } 1
> > ------------------------------------------------------------
> > sys_perf_event_open: pid -1 cpu 0 group_fd -1 flags 0x8 = 14
> > sys_perf_event_open: pid -1 cpu 1 group_fd -1 flags 0x8 = 15
> > sys_perf_event_open: pid -1 cpu 2 group_fd -1 flags 0x8 = 16
> > sys_perf_event_open: pid -1 cpu 3 group_fd -1 flags 0x8 = 17
> > sys_perf_event_open: pid -1 cpu 4 group_fd -1 flags 0x8 = 18
> > sys_perf_event_open: pid -1 cpu 5 group_fd -1 flags 0x8 = 19
> > sys_perf_event_open: pid -1 cpu 6 group_fd -1 flags 0x8 = 20
> > sys_perf_event_open: pid -1 cpu 7 group_fd -1 flags 0x8 = 21
> > mmap size 528384B
> > 0x7f3e0697d0b8: mmap mask[8]:
> > 0x7f3e0698d180: mmap mask[8]:
> > 0x7f3e0699d248: mmap mask[8]:
> > 0x7f3e069ad310: mmap mask[8]:
> > 0x7f3e069bd3d8: mmap mask[8]:
> > 0x7f3e069cd4a0: mmap mask[8]:
> > 0x7f3e069dd568: mmap mask[8]:
> > 0x7f3e069ed630: mmap mask[8]:
> > Synthesizing TSC conversion information
> > arch copy Documentation init kernel MAINTAINERS modules.builtin.modinfo perf.data scripts System.map vmlinux
> > block COPYING drivers ipc lbuild Makefile modules.order perf.data.old security tools vmlinux.o
> > certs CREDITS fs Kbuild lib mm Module.symvers README sound usr
> > config-5.2.7-100.fc29.x86_64 crypto include Kconfig LICENSES modules.builtin net samples stdio virt
> > [ perf record: Woken up 1 times to write data ]
> > Looking at the vmlinux_path (8 entries long)
> > Using vmlinux for symbols
> > [ perf record: Captured and wrote 0.013 MB perf.data (8 samples) ]
> >
> > tools/perf/perf record -vv --affinity=cpu -- ls
> > thread mask[8]: empty
> > Using CPUID GenuineIntel-6-5E-3
> > intel_pt default config: tsc,mtc,mtc_period=3,psb_period=3,pt,branch
> > nr_cblocks: 0
> > affinity: CPU
> > mmap flush: 1
> > comp level: 0
> > ------------------------------------------------------------
> > perf_event_attr:
> > size 120
> > { sample_period, sample_freq } 4000
> > sample_type IP|TID|TIME|PERIOD
> > read_format ID
> > disabled 1
> > inherit 1
> > mmap 1
> > comm 1
> > freq 1
> > enable_on_exec 1
> > task 1
> > precise_ip 3
> > sample_id_all 1
> > exclude_guest 1
> > mmap2 1
> > comm_exec 1
> > ksymbol 1
> > bpf_event 1
> > ------------------------------------------------------------
> > sys_perf_event_open: pid 23713 cpu 0 group_fd -1 flags 0x8 = 4
> > sys_perf_event_open: pid 23713 cpu 1 group_fd -1 flags 0x8 = 5
> > sys_perf_event_open: pid 23713 cpu 2 group_fd -1 flags 0x8 = 6
> > sys_perf_event_open: pid 23713 cpu 3 group_fd -1 flags 0x8 = 9
> > sys_perf_event_open: pid 23713 cpu 4 group_fd -1 flags 0x8 = 10
> > sys_perf_event_open: pid 23713 cpu 5 group_fd -1 flags 0x8 = 11
> > sys_perf_event_open: pid 23713 cpu 6 group_fd -1 flags 0x8 = 12
> > sys_perf_event_open: pid 23713 cpu 7 group_fd -1 flags 0x8 = 13
> > mmap size 528384B
> > 0x7f3e005bc0b8: mmap mask[8]: 0
> > 0x7f3e005cc180: mmap mask[8]: 1
> > 0x7f3e005dc248: mmap mask[8]: 2
> > 0x7f3e005ec310: mmap mask[8]: 3
> > 0x7f3e005fc3d8: mmap mask[8]: 4
> > 0x7f3e0060c4a0: mmap mask[8]: 5
> > 0x7f3e0061c568: mmap mask[8]: 6
> > 0x7f3e0062c630: mmap mask[8]: 7
> > ------------------------------------------------------------
> > perf_event_attr:
> > type 1
> > size 120
> > config 0x9
> > watermark 1
> > sample_id_all 1
> > bpf_event 1
> > { wakeup_events, wakeup_watermark } 1
> > ------------------------------------------------------------
> > sys_perf_event_open: pid -1 cpu 0 group_fd -1 flags 0x8 = 14
> > sys_perf_event_open: pid -1 cpu 1 group_fd -1 flags 0x8 = 15
> > sys_perf_event_open: pid -1 cpu 2 group_fd -1 flags 0x8 = 16
> > sys_perf_event_open: pid -1 cpu 3 group_fd -1 flags 0x8 = 17
> > sys_perf_event_open: pid -1 cpu 4 group_fd -1 flags 0x8 = 18
> > sys_perf_event_open: pid -1 cpu 5 group_fd -1 flags 0x8 = 19
> > sys_perf_event_open: pid -1 cpu 6 group_fd -1 flags 0x8 = 20
> > sys_perf_event_open: pid -1 cpu 7 group_fd -1 flags 0x8 = 21
> > mmap size 528384B
> > 0x7f3e001330b8: mmap mask[8]:
> > 0x7f3e00143180: mmap mask[8]:
> > 0x7f3e00153248: mmap mask[8]:
> > 0x7f3e00163310: mmap mask[8]:
> > 0x7f3e001733d8: mmap mask[8]:
> > 0x7f3e001834a0: mmap mask[8]:
> > 0x7f3e00193568: mmap mask[8]:
> > 0x7f3e001a3630: mmap mask[8]:
> > Synthesizing TSC conversion information
> > 0x9c9ff0: thread mask[8]: 0
> > 0x9c9ff0: thread mask[8]: 1
> > 0x9c9ff0: thread mask[8]: 2
> > 0x9c9ff0: thread mask[8]: 3
> > 0x9c9ff0: thread mask[8]: 4
> > arch copy Documentation init kernel MAINTAINERS modules.builtin.modinfo perf.data scripts System.map vmlinux
> > block COPYING drivers ipc lbuild Makefile modules.order perf.data.old security tools vmlinux.o
> > certs CREDITS fs Kbuild lib mm Module.symvers README sound usr
> > config-5.2.7-100.fc29.x86_64 crypto include Kconfig LICENSES modules.builtin net samples stdio virt
> > 0x9c9ff0: thread mask[8]: 5
> > 0x9c9ff0: thread mask[8]: 6
> > 0x9c9ff0: thread mask[8]: 7
> > 0x9c9ff0: thread mask[8]: 0
> > 0x9c9ff0: thread mask[8]: 1
> > 0x9c9ff0: thread mask[8]: 2
> > 0x9c9ff0: thread mask[8]: 3
> > 0x9c9ff0: thread mask[8]: 4
> > 0x9c9ff0: thread mask[8]: 5
> > 0x9c9ff0: thread mask[8]: 6
> > 0x9c9ff0: thread mask[8]: 7
> > [ perf record: Woken up 0 times to write data ]
> > 0x9c9ff0: thread mask[8]: 0
> > 0x9c9ff0: thread mask[8]: 1
> > 0x9c9ff0: thread mask[8]: 2
> > 0x9c9ff0: thread mask[8]: 3
> > 0x9c9ff0: thread mask[8]: 4
> > 0x9c9ff0: thread mask[8]: 5
> > 0x9c9ff0: thread mask[8]: 6
> > 0x9c9ff0: thread mask[8]: 7
> > Looking at the vmlinux_path (8 entries long)
> > Using vmlinux for symbols
> > ...
> > [ perf record: Captured and wrote 0.013 MB perf.data (10 samples) ]
> >
--
- Arnaldo
^ permalink raw reply [flat|nested] 13+ messages in thread