Re: [PATCH v4 1/4] perf config: Introduce perf_config_set class

From: Taeung Song
Date: Fri Apr 01 2016 - 06:27:31 EST

Hi, Arnaldo

Thank you for your review.

On 04/01/2016 02:31 AM, Arnaldo Carvalho de Melo wrote:
Em Tue, Mar 29, 2016 at 09:43:13AM +0900, Taeung Song escreveu:
This infrastructure code was designed for
upcoming features of perf-config.

That collect config key-value pairs from user and
system config files (i.e. user wide ~/.perfconfig
and system wide $(sysconfdir)/perfconfig)
to manage perf's configs.

Cc: Namhyung Kim <namhyung@xxxxxxxxxx>
Cc: Jiri Olsa <jolsa@xxxxxxxxxx>
Signed-off-by: Taeung Song <treeze.taeung@xxxxxxxxx>
---
tools/perf/util/config.c | 171 +++++++++++++++++++++++++++++++++++++++++++++++
tools/perf/util/config.h | 26 +++++++
2 files changed, 197 insertions(+)
create mode 100644 tools/perf/util/config.h

diff --git a/tools/perf/util/config.c b/tools/perf/util/config.c
index 4e72763..2dbf47c 100644
--- a/tools/perf/util/config.c
+++ b/tools/perf/util/config.c
@@ -13,6 +13,7 @@
#include <subcmd/exec-cmd.h>
#include "util/hist.h" /* perf_hist_config */
#include "util/llvm-utils.h" /* perf_llvm_config */
+#include "config.h"

#define MAXNAME (256)

@@ -506,6 +507,176 @@ out:
return ret;
}

+static struct perf_config_section *find_section(struct list_head *sections,
+ const char *section_name)
+{
+ struct perf_config_section *section;
+
+ list_for_each_entry(section, sections, list)
+ if (!strcmp(section->name, section_name))
+ return section;
+
+ return NULL;
+}
+
+static struct perf_config_item *find_config_item(const char *name,
+ struct perf_config_section *section)
+{
+ struct perf_config_item *config_item;
+
+ list_for_each_entry(config_item, &section->config_items, list)
+ if (!strcmp(config_item->name, name))
+ return config_item;
+
+ return NULL;
+}
+
+static void find_config(struct list_head *sections,
+ struct perf_config_section **section,
+ struct perf_config_item **config_item,
+ const char *section_name, const char *name)
+{
+ *section = find_section(sections, section_name);
+
+ if (*section != NULL)
+ *config_item = find_config_item(name, *section);
+ else
+ *config_item = NULL;

+}
+
+static struct perf_config_section *add_section(struct list_head *sections,
+ const char *section_name)
+{
+ struct perf_config_section *section = zalloc(sizeof(*section));
+
+ if (!section)
+ return NULL;
+
+ INIT_LIST_HEAD(&section->config_items);
+ section->name = strdup(section_name);
+ if (!section->name) {
+ pr_err("%s: strdup failed\n", __func__);
+ free(section);
+ return NULL;
+ }
+
+ list_add_tail(&section->list, sections);
+ return section;
+}
+
+static struct perf_config_item *add_config_item(struct perf_config_section *section,
+ const char *name)
+{
+ struct perf_config_item *config_item = zalloc(sizeof(*config_item));
+
+ if (!config_item)
+ return NULL;
+
+ config_item->name = strdup(name);
+ if (!name) {

Huh? You're testing the wrong variable.


Sorry, my stupid mistake..

+ pr_err("%s: strdup failed\n", __func__);
+ goto out_err;
+ }
+
+ list_add_tail(&config_item->list, &section->config_items);
+ return config_item;
+
+out_err:
+ free(config_item);
+ return NULL;
+}
+
+static int set_value(struct perf_config_item *config_item, const char *value)
+{
+ char *val = strdup(value);
+
+ if (!val)
+ return -1;
+
+ free(config_item->value);
+ config_item->value = val;
+ return 0;
+}
+
+static int collect_config(const char *var, const char *value,
+ void *perf_config_set)
+{
+ int ret = -1;
+ char *ptr, *key;
+ char *section_name, *name;
+ struct perf_config_section *section = NULL;
+ struct perf_config_item *config_item = NULL;
+ struct perf_config_set *perf_configs = perf_config_set;
+ struct list_head *sections = &perf_configs->sections;
+
+ key = ptr = strdup(var);
+ if (!key) {
+ pr_err("%s: strdup failed\n", __func__);

pr_debug()


I'll change pr_err to pr_debug.
But why do use pr_debug at only this part ?

+ return -1;
+ }
+
+ section_name = strsep(&ptr, ".");
+ name = ptr;
+ if (name == NULL || value == NULL)
+ goto out_free;
+
+ find_config(sections, &section, &config_item, section_name, name);

This idiom is confusing, why not ditch this 'find_config()' function and
do the searches here? I.e.:

section = find_section(sections, section_name);

I got it.
I think it is needed to remove needless find_config() function as you said.

+ if (!section) {
+ section = add_section(sections, section_name);
+ if (!section)
+ goto out_free;
+ }

config_item = find_config_item(name, section);

ok.

+ if (!config_item) {
+ config_item = add_config_item(section, name);
+ if (!config_item)
+ goto out_free;
+ }
+
+ ret = set_value(config_item, value);
+ return ret;
+
+out_free:
+ free(key);
+ perf_config_set__delete(perf_configs);
+ return -1;
+}
+
+struct perf_config_set *perf_config_set__new(void)
+{
+ struct perf_config_set *perf_configs = zalloc(sizeof(*perf_configs));
+
+ if (!perf_configs)
+ return NULL;
+
+ INIT_LIST_HEAD(&perf_configs->sections);
+ perf_config(collect_config, perf_configs);
+
+ return perf_configs;
+}

Usually for these short functions we could do it more compactly as:

struct perf_config_set *perf_config_set__new(void)
{
struct perf_config_set *perf_configs = zalloc(sizeof(*perf_configs));

if (perf_configs) {
INIT_LIST_HEAD(&perf_configs->sections);
perf_config(collect_config, perf_configs);
}

return perf_configs;
}

But I'm not dwelling on this...


I got it, too!

+void perf_config_set__delete(struct perf_config_set *perf_configs)
+{
+ struct perf_config_section *section, *section_tmp;
+ struct perf_config_item *config_item, *item_tmp;
+
+ list_for_each_entry_safe(section, section_tmp,
+ &perf_configs->sections, list) {
+ list_for_each_entry_safe(config_item, item_tmp,
+ &section->config_items, list) {
+ list_del(&config_item->list);
+ free(config_item->name);
+ free(config_item->value);
+ free(config_item);
+ }
+ list_del(&section->list);
+ free(section->name);
+ free(section);
+ }
+
+ free(perf_configs);
+}

What is the problem with having perf_config_item__delete() and
perf_config_section__delete() and then have it like below, also please
rename those foo->list to foo->node.


No problem!
OK, I'll rename it.

void perf_config_item__delete(struct perf_config_item *item)
{
zfree(&item->name);
zfree(&item->value);
free(item);
}

void perf_config_section__purge(struct perf_config_section *section)
{
struct perf_config_item *item, *tmp;

list_for_each_entry_safe(item, tmp, &section->items, node) {
list_del_init(&item->node);
perf_config_item__delete(item);
}
}

void perf_config_section__delete(struct perf_config_section *section)
{
perf_config_section__purge(section);
zfree(&section->name);
free(section);
}

void perf_config_set__purge(struct perf_config_set *set)
{
struct perf_config_section *section, *tmp;

list_for_each_entry_safe(section, tmp, &set->sections, node) {
list_del_init(&section->node);
perf_config_section__delete(section);
}
}

void perf_config_set__delete(struct perf_config_set *set)
{
perf_config_set__purge(set);
free(set);
}

Using zfree() and list_del_init to NULL or poison the freed pointers
helps with debugging, please use them.

ok.


+
/*
* Call this to report error for your variable that should not
* get a boolean value (i.e. "[my] var" means "true").
diff --git a/tools/perf/util/config.h b/tools/perf/util/config.h
new file mode 100644
index 0000000..e270e51
--- /dev/null
+++ b/tools/perf/util/config.h
@@ -0,0 +1,26 @@
+#ifndef __PERF_CONFIG_H
+#define __PERF_CONFIG_H
+
+#include <stdbool.h>
+#include <linux/list.h>
+
+struct perf_config_item {
+ char *name;
+ char *value;
+ struct list_head list;

s/list/node/g

+};
+
+struct perf_config_section {
+ char *name;
+ struct list_head config_items;

s/config_items/items/g

+ struct list_head list;

s/list/node/g
+};
+
+struct perf_config_set {
+ struct list_head sections;

See? Here you did it right, no point in having it as "config_sections"


I'll rename it to 'config_sections'.

+};
+
+struct perf_config_set *perf_config_set__new(void);
+void perf_config_set__delete(struct perf_config_set *perf_configs);

void perf_config_set__delete(struct perf_config_set *set);


OK, I'll use 'set' variable name instead of perf_configs on this source file.

I'll resend this patch after modifying what you said. :-)

Thanks,
Taeung

+
+#endif /* __PERF_CONFIG_H */
--
2.5.0