[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20180307192838.GU3701@kernel.org>
Date: Wed, 7 Mar 2018 16:28:38 -0300
From: Arnaldo Carvalho de Melo <acme@...nel.org>
To: Jiri Olsa <jolsa@...nel.org>
Cc: lkml <linux-kernel@...r.kernel.org>,
Ingo Molnar <mingo@...nel.org>,
Namhyung Kim <namhyung@...nel.org>,
David Ahern <dsahern@...il.com>,
Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
Peter Zijlstra <a.p.zijlstra@...llo.nl>
Subject: Re: [PATCH 07/19] perf tools: Add MEM_TOPOLOGY feature to perf data
file
Em Wed, Mar 07, 2018 at 04:50:08PM +0100, Jiri Olsa escreveu:
> Adding MEM_TOPOLOGY feature to perf data file,
> that will carry physical memory map and its
> node assignments.
Good addition, applied 1-7 in this series, skipping a few after this,
processing the rest.
- Arnaldo
> The format of data in MEM_TOPOLOGY is as follows:
>
> 0 - version | for future changes
> 8 - block_size_bytes | /sys/devices/system/memory/block_size_bytes
> 16 - count | number of nodes
>
> For each node we store map of physical indexes for
> each node:
>
> 32 - node id | node index
> 40 - size | size of bitmap
> 48 - bitmap | bitmap of memory indexes that belongs to node
> | /sys/devices/system/node/node<NODE>/memory<INDEX>
>
> The MEM_TOPOLOGY could be displayed with following
> report command:
>
> $ perf report --header-only -I
> ...
> # memory nodes (nr 1, block size 0x8000000):
> # 0 [7G]: 0-23,32-69
>
> Link: http://lkml.kernel.org/n/tip-qq7sohu774wxq154n3my037z@git.kernel.org
> Signed-off-by: Jiri Olsa <jolsa@...nel.org>
> ---
> tools/include/linux/bitmap.h | 2 +-
> tools/perf/util/env.h | 9 ++
> tools/perf/util/header.c | 305 +++++++++++++++++++++++++++++++++++++++++++
> tools/perf/util/header.h | 1 +
> 4 files changed, 316 insertions(+), 1 deletion(-)
>
> diff --git a/tools/include/linux/bitmap.h b/tools/include/linux/bitmap.h
> index ca160270fdfa..63440cc8d618 100644
> --- a/tools/include/linux/bitmap.h
> +++ b/tools/include/linux/bitmap.h
> @@ -98,7 +98,7 @@ static inline int test_and_set_bit(int nr, unsigned long *addr)
>
> /**
> * bitmap_alloc - Allocate bitmap
> - * @nr: Bit to set
> + * @nbits: Number of bits
> */
> static inline unsigned long *bitmap_alloc(int nbits)
> {
> diff --git a/tools/perf/util/env.h b/tools/perf/util/env.h
> index bf970f57dce0..c4ef2e523367 100644
> --- a/tools/perf/util/env.h
> +++ b/tools/perf/util/env.h
> @@ -27,6 +27,12 @@ struct numa_node {
> struct cpu_map *map;
> };
>
> +struct memory_node {
> + u64 node;
> + u64 size;
> + unsigned long *set;
> +};
> +
> struct perf_env {
> char *hostname;
> char *os_release;
> @@ -43,6 +49,7 @@ struct perf_env {
> int nr_sibling_cores;
> int nr_sibling_threads;
> int nr_numa_nodes;
> + int nr_memory_nodes;
> int nr_pmu_mappings;
> int nr_groups;
> char *cmdline;
> @@ -54,6 +61,8 @@ struct perf_env {
> struct cpu_cache_level *caches;
> int caches_cnt;
> struct numa_node *numa_nodes;
> + struct memory_node *memory_nodes;
> + unsigned long long memory_bsize;
> };
>
> extern struct perf_env perf_env;
> diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
> index e0c3cad0fd8d..3a107e7ac135 100644
> --- a/tools/perf/util/header.c
> +++ b/tools/perf/util/header.c
> @@ -17,6 +17,7 @@
> #include <sys/stat.h>
> #include <sys/utsname.h>
> #include <linux/time64.h>
> +#include <dirent.h>
>
> #include "evlist.h"
> #include "evsel.h"
> @@ -37,6 +38,7 @@
> #include "asm/bug.h"
> #include "tool.h"
> #include "time-utils.h"
> +#include "units.h"
>
> #include "sane_ctype.h"
>
> @@ -132,6 +134,25 @@ int do_write(struct feat_fd *ff, const void *buf, size_t size)
> }
>
> /* Return: 0 if succeded, -ERR if failed. */
> +static int do_write_bitmap(struct feat_fd *ff, unsigned long *set, u64 size)
> +{
> + u64 *p = (u64 *) set;
> + int i, ret;
> +
> + ret = do_write(ff, &size, sizeof(size));
> + if (ret < 0)
> + return ret;
> +
> + for (i = 0; (u64) i < BITS_TO_U64(size); i++) {
> + ret = do_write(ff, p + i, sizeof(*p));
> + if (ret < 0)
> + return ret;
> + }
> +
> + return 0;
> +}
> +
> +/* Return: 0 if succeded, -ERR if failed. */
> int write_padded(struct feat_fd *ff, const void *bf,
> size_t count, size_t count_aligned)
> {
> @@ -243,6 +264,38 @@ static char *do_read_string(struct feat_fd *ff)
> return NULL;
> }
>
> +/* Return: 0 if succeded, -ERR if failed. */
> +static int do_read_bitmap(struct feat_fd *ff, unsigned long **pset, u64 *psize)
> +{
> + unsigned long *set;
> + u64 size, *p;
> + int i, ret;
> +
> + ret = do_read_u64(ff, &size);
> + if (ret)
> + return ret;
> +
> + set = bitmap_alloc(size);
> + if (!set)
> + return -ENOMEM;
> +
> + bitmap_zero(set, size);
> +
> + p = (u64 *) set;
> +
> + for (i = 0; (u64) i < BITS_TO_U64(size); i++) {
> + ret = do_read_u64(ff, p + i);
> + if (ret < 0) {
> + free(set);
> + return ret;
> + }
> + }
> +
> + *pset = set;
> + *psize = size;
> + return 0;
> +}
> +
> static int write_tracing_data(struct feat_fd *ff,
> struct perf_evlist *evlist)
> {
> @@ -1196,6 +1249,176 @@ static int write_sample_time(struct feat_fd *ff,
> sizeof(evlist->last_sample_time));
> }
>
> +
> +static int memory_node__read(struct memory_node *n, unsigned long index)
> +{
> + unsigned int phys, size = 0;
> + char path[PATH_MAX];
> + struct dirent *ent;
> + DIR *dir;
> +
> +#define for_each_memory(mem, dir) \
> + while ((ent = readdir(dir))) \
> + if (strcmp(ent->d_name, ".") && \
> + strcmp(ent->d_name, "..") && \
> + sscanf(ent->d_name, "memory%u", &mem) == 1)
> +
> + scnprintf(path, PATH_MAX,
> + "%s/devices/system/node/node%lu",
> + sysfs__mountpoint(), index);
> +
> + dir = opendir(path);
> + if (!dir) {
> + pr_warning("failed: cant' open memory sysfs data\n");
> + return -1;
> + }
> +
> + for_each_memory(phys, dir) {
> + size = max(phys, size);
> + }
> +
> + size++;
> +
> + n->set = bitmap_alloc(size);
> + if (!n->set) {
> + closedir(dir);
> + return -ENOMEM;
> + }
> +
> + bitmap_zero(n->set, size);
> + n->node = index;
> + n->size = size;
> +
> + rewinddir(dir);
> +
> + for_each_memory(phys, dir) {
> + set_bit(phys, n->set);
> + }
> +
> + closedir(dir);
> + return 0;
> +}
> +
> +static int memory_node__sort(const void *a, const void *b)
> +{
> + const struct memory_node *na = a;
> + const struct memory_node *nb = b;
> +
> + return na->node - nb->node;
> +}
> +
> +static int build_mem_topology(struct memory_node *nodes, u64 size, u64 *cntp)
> +{
> + char path[PATH_MAX];
> + struct dirent *ent;
> + DIR *dir;
> + u64 cnt = 0;
> + int ret = 0;
> +
> + scnprintf(path, PATH_MAX, "%s/devices/system/node/",
> + sysfs__mountpoint());
> +
> + dir = opendir(path);
> + if (!dir) {
> + pr_warning("failed: can't open node sysfs data\n");
> + return -1;
> + }
> +
> + while (!ret && (ent = readdir(dir))) {
> + unsigned int index;
> + int r;
> +
> + if (!strcmp(ent->d_name, ".") ||
> + !strcmp(ent->d_name, ".."))
> + continue;
> +
> + r = sscanf(ent->d_name, "node%u", &index);
> + if (r != 1)
> + continue;
> +
> + if (WARN_ONCE(cnt >= size,
> + "failed to write MEM_TOPOLOGY, way too many nodes\n"))
> + return -1;
> +
> + ret = memory_node__read(&nodes[cnt++], index);
> + }
> +
> + *cntp = cnt;
> + closedir(dir);
> +
> + if (!ret)
> + qsort(nodes, cnt, sizeof(nodes[0]), memory_node__sort);
> +
> + return ret;
> +}
> +
> +#define MAX_MEMORY_NODES 2000
> +
> +/*
> + * The MEM_TOPOLOGY holds physical memory map for every
> + * node in system. The format of data is as follows:
> + *
> + * 0 - version | for future changes
> + * 8 - block_size_bytes | /sys/devices/system/memory/block_size_bytes
> + * 16 - count | number of nodes
> + *
> + * For each node we store map of physical indexes for
> + * each node:
> + *
> + * 32 - node id | node index
> + * 40 - size | size of bitmap
> + * 48 - bitmap | bitmap of memory indexes that belongs to node
> + */
> +static int write_mem_topology(struct feat_fd *ff __maybe_unused,
> + struct perf_evlist *evlist __maybe_unused)
> +{
> + static struct memory_node nodes[MAX_MEMORY_NODES];
> + u64 bsize, version = 1, i, nr;
> + int ret;
> +
> + ret = sysfs__read_xll("devices/system/memory/block_size_bytes",
> + (unsigned long long *) &bsize);
> + if (ret)
> + return ret;
> +
> + ret = build_mem_topology(&nodes[0], MAX_MEMORY_NODES, &nr);
> + if (ret)
> + return ret;
> +
> + ret = do_write(ff, &version, sizeof(version));
> + if (ret < 0)
> + goto out;
> +
> + ret = do_write(ff, &bsize, sizeof(bsize));
> + if (ret < 0)
> + goto out;
> +
> + ret = do_write(ff, &nr, sizeof(nr));
> + if (ret < 0)
> + goto out;
> +
> + for (i = 0; i < nr; i++) {
> + struct memory_node *n = &nodes[i];
> +
> + #define _W(v) \
> + ret = do_write(ff, &n->v, sizeof(n->v)); \
> + if (ret < 0) \
> + goto out;
> +
> + _W(node)
> + _W(size)
> +
> + #undef _W
> +
> + ret = do_write_bitmap(ff, n->set, n->size);
> + if (ret < 0)
> + goto out;
> + }
> +
> +out:
> + return ret;
> +}
> +
> static void print_hostname(struct feat_fd *ff, FILE *fp)
> {
> fprintf(fp, "# hostname : %s\n", ff->ph->env.hostname);
> @@ -1543,6 +1766,35 @@ static void print_sample_time(struct feat_fd *ff, FILE *fp)
> fprintf(fp, "# sample duration : %10.3f ms\n", d);
> }
>
> +static void memory_node__fprintf(struct memory_node *n,
> + unsigned long long bsize, FILE *fp)
> +{
> + char buf_map[100], buf_size[50];
> + unsigned long long size;
> +
> + size = bsize * bitmap_weight(n->set, n->size);
> + unit_number__scnprintf(buf_size, 50, size);
> +
> + bitmap_scnprintf(n->set, n->size, buf_map, 100);
> + fprintf(fp, "# %3" PRIu64 " [%s]: %s\n", n->node, buf_size, buf_map);
> +}
> +
> +static void print_mem_topology(struct feat_fd *ff, FILE *fp)
> +{
> + struct memory_node *nodes;
> + int i, nr;
> +
> + nodes = ff->ph->env.memory_nodes;
> + nr = ff->ph->env.nr_memory_nodes;
> +
> + fprintf(fp, "# memory nodes (nr %d, block size 0x%llx):\n",
> + nr, ff->ph->env.memory_bsize);
> +
> + for (i = 0; i < nr; i++) {
> + memory_node__fprintf(&nodes[i], ff->ph->env.memory_bsize, fp);
> + }
> +}
> +
> static int __event_process_build_id(struct build_id_event *bev,
> char *filename,
> struct perf_session *session)
> @@ -2205,6 +2457,58 @@ static int process_sample_time(struct feat_fd *ff, void *data __maybe_unused)
> return 0;
> }
>
> +static int process_mem_topology(struct feat_fd *ff,
> + void *data __maybe_unused)
> +{
> + struct memory_node *nodes;
> + u64 version, i, nr, bsize;
> + int ret = -1;
> +
> + if (do_read_u64(ff, &version))
> + return -1;
> +
> + if (version != 1)
> + return -1;
> +
> + if (do_read_u64(ff, &bsize))
> + return -1;
> +
> + if (do_read_u64(ff, &nr))
> + return -1;
> +
> + nodes = zalloc(sizeof(*nodes) * nr);
> + if (!nodes)
> + return -1;
> +
> + for (i = 0; i < nr; i++) {
> + struct memory_node n;
> +
> + #define _R(v) \
> + if (do_read_u64(ff, &n.v)) \
> + goto out; \
> +
> + _R(node)
> + _R(size)
> +
> + #undef _R
> +
> + if (do_read_bitmap(ff, &n.set, &n.size))
> + goto out;
> +
> + nodes[i] = n;
> + }
> +
> + ff->ph->env.memory_bsize = bsize;
> + ff->ph->env.memory_nodes = nodes;
> + ff->ph->env.nr_memory_nodes = nr;
> + ret = 0;
> +
> +out:
> + if (ret)
> + free(nodes);
> + return ret;
> +}
> +
> struct feature_ops {
> int (*write)(struct feat_fd *ff, struct perf_evlist *evlist);
> void (*print)(struct feat_fd *ff, FILE *fp);
> @@ -2263,6 +2567,7 @@ static const struct feature_ops feat_ops[HEADER_LAST_FEATURE] = {
> FEAT_OPN(STAT, stat, false),
> FEAT_OPN(CACHE, cache, true),
> FEAT_OPR(SAMPLE_TIME, sample_time, false),
> + FEAT_OPR(MEM_TOPOLOGY, mem_topology, true),
> };
>
> struct header_print_data {
> diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h
> index 942bdec6d70d..90d4577a92dc 100644
> --- a/tools/perf/util/header.h
> +++ b/tools/perf/util/header.h
> @@ -36,6 +36,7 @@ enum {
> HEADER_STAT,
> HEADER_CACHE,
> HEADER_SAMPLE_TIME,
> + HEADER_MEM_TOPOLOGY,
> HEADER_LAST_FEATURE,
> HEADER_FEAT_BITS = 256,
> };
> --
> 2.13.6
Powered by blists - more mailing lists