* [PATCH next] perf tools: fix potential memleak in perf events parser
@ 2020-06-11 1:42 Chen Wandun
2020-06-11 13:16 ` Arnaldo Carvalho de Melo
0 siblings, 1 reply; 3+ messages in thread
From: Chen Wandun @ 2020-06-11 1:42 UTC (permalink / raw)
To: acme, linux-kernel; +Cc: peterz, mingo, cj.chengjian, chenwandun
Fix potential memory leak in function parse_events_term__sym_hw()
and parse_events_term__clone().
1. Free memory when errors occur.
2. Function new_term may return error, so it is need to free memory
when the return value is negative.
Signed-off-by: Chen Wandun <chenwandun@huawei.com>
---
tools/perf/util/parse-events.c | 40 +++++++++++++++++++++++++++++-----
1 file changed, 34 insertions(+), 6 deletions(-)
diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c
index 3decbb203846..3491c18edd71 100644
--- a/tools/perf/util/parse-events.c
+++ b/tools/perf/util/parse-events.c
@@ -2947,6 +2947,7 @@ int parse_events_term__sym_hw(struct parse_events_term **term,
.type_term = PARSE_EVENTS__TERM_TYPE_USER,
.config = config,
};
+ int ret;
if (!temp.config) {
temp.config = strdup("event");
@@ -2957,9 +2958,20 @@ int parse_events_term__sym_hw(struct parse_events_term **term,
sym = &event_symbols_hw[idx];
str = strdup(sym->symbol);
- if (!str)
+ if (!str) {
+ if (!config)
+ free(temp.config);
return -ENOMEM;
- return new_term(term, &temp, str, 0);
+ }
+
+ ret = new_term(term, &temp, str, 0);
+ if (ret < 0) {
+ free(str);
+ if (!config)
+ free(temp.config);
+ }
+
+ return ret;
}
int parse_events_term__clone(struct parse_events_term **new,
@@ -2973,19 +2985,35 @@ int parse_events_term__clone(struct parse_events_term **new,
.err_term = term->err_term,
.err_val = term->err_val,
};
+ int ret;
if (term->config) {
temp.config = strdup(term->config);
if (!temp.config)
return -ENOMEM;
}
- if (term->type_val == PARSE_EVENTS__TERM_TYPE_NUM)
- return new_term(new, &temp, NULL, term->val.num);
+ if (term->type_val == PARSE_EVENTS__TERM_TYPE_NUM) {
+ ret = new_term(new, &temp, NULL, term->val.num);
+ if (ret < 0 && term->config)
+ free(temp.config);
+ return ret;
+ }
str = strdup(term->val.str);
- if (!str)
+ if (!str) {
+ if (term->config)
+ free(temp.config);
return -ENOMEM;
- return new_term(new, &temp, str, 0);
+ }
+
+ ret = new_term(new, &temp, str, 0);
+ if (ret < 0) {
+ free(str);
+ if (term->config)
+ free(temp.config);
+ }
+
+ return ret;
}
void parse_events_term__delete(struct parse_events_term *term)
--
2.17.1
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH next] perf tools: fix potential memleak in perf events parser
2020-06-11 1:42 [PATCH next] perf tools: fix potential memleak in perf events parser Chen Wandun
@ 2020-06-11 13:16 ` Arnaldo Carvalho de Melo
2020-06-11 15:05 ` Chen Wandun
0 siblings, 1 reply; 3+ messages in thread
From: Arnaldo Carvalho de Melo @ 2020-06-11 13:16 UTC (permalink / raw)
To: Chen Wandun; +Cc: linux-kernel, peterz, mingo, cj.chengjian
Em Thu, Jun 11, 2020 at 09:42:34AM +0800, Chen Wandun escreveu:
> Fix potential memory leak in function parse_events_term__sym_hw()
> and parse_events_term__clone().
>
> 1. Free memory when errors occur.
> 2. Function new_term may return error, so it is need to free memory
> when the return value is negative.
Try to fix one thing per patch, i.e. first the most obvious one, then
the other that requires going thru the new_term() logic, i.e. first
this, which is super easy to review:
diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c
index c4906a6a9f1a..3ada3874a90a 100644
--- a/tools/perf/util/parse-events.c
+++ b/tools/perf/util/parse-events.c
@@ -2958,8 +2958,10 @@ int parse_events_term__sym_hw(struct parse_events_term **term,
sym = &event_symbols_hw[idx];
str = strdup(sym->symbol);
- if (!str)
+ if (!str) {
+ zfree(&temp.config);
return -ENOMEM;
+ }
return new_term(term, &temp, str, 0);
}
@@ -2984,8 +2986,10 @@ int parse_events_term__clone(struct parse_events_term **new,
return new_term(new, &temp, NULL, term->val.num);
str = strdup(term->val.str);
- if (!str)
+ if (!str) {
+ zfree(&temp.config);
return -ENOMEM;
+ }
return new_term(new, &temp, str, 0);
}
Then you go to the one that requires the reviewer (now or in the future,
trying to figure out why something broke) to look at the new_term()
code, ok?
- Arnaldo
> Signed-off-by: Chen Wandun <chenwandun@huawei.com>
> ---
> tools/perf/util/parse-events.c | 40 +++++++++++++++++++++++++++++-----
> 1 file changed, 34 insertions(+), 6 deletions(-)
>
> diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c
> index 3decbb203846..3491c18edd71 100644
> --- a/tools/perf/util/parse-events.c
> +++ b/tools/perf/util/parse-events.c
> @@ -2947,6 +2947,7 @@ int parse_events_term__sym_hw(struct parse_events_term **term,
> .type_term = PARSE_EVENTS__TERM_TYPE_USER,
> .config = config,
> };
> + int ret;
>
> if (!temp.config) {
> temp.config = strdup("event");
> @@ -2957,9 +2958,20 @@ int parse_events_term__sym_hw(struct parse_events_term **term,
> sym = &event_symbols_hw[idx];
>
> str = strdup(sym->symbol);
> - if (!str)
> + if (!str) {
> + if (!config)
> + free(temp.config);
> return -ENOMEM;
> - return new_term(term, &temp, str, 0);
> + }
> +
> + ret = new_term(term, &temp, str, 0);
> + if (ret < 0) {
> + free(str);
> + if (!config)
> + free(temp.config);
> + }
> +
> + return ret;
> }
>
> int parse_events_term__clone(struct parse_events_term **new,
> @@ -2973,19 +2985,35 @@ int parse_events_term__clone(struct parse_events_term **new,
> .err_term = term->err_term,
> .err_val = term->err_val,
> };
> + int ret;
>
> if (term->config) {
> temp.config = strdup(term->config);
> if (!temp.config)
> return -ENOMEM;
> }
> - if (term->type_val == PARSE_EVENTS__TERM_TYPE_NUM)
> - return new_term(new, &temp, NULL, term->val.num);
> + if (term->type_val == PARSE_EVENTS__TERM_TYPE_NUM) {
> + ret = new_term(new, &temp, NULL, term->val.num);
> + if (ret < 0 && term->config)
> + free(temp.config);
> + return ret;
> + }
>
> str = strdup(term->val.str);
> - if (!str)
> + if (!str) {
> + if (term->config)
> + free(temp.config);
> return -ENOMEM;
> - return new_term(new, &temp, str, 0);
> + }
> +
> + ret = new_term(new, &temp, str, 0);
> + if (ret < 0) {
> + free(str);
> + if (term->config)
> + free(temp.config);
> + }
> +
> + return ret;
> }
>
> void parse_events_term__delete(struct parse_events_term *term)
> --
> 2.17.1
>
--
- Arnaldo
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH next] perf tools: fix potential memleak in perf events parser
2020-06-11 13:16 ` Arnaldo Carvalho de Melo
@ 2020-06-11 15:05 ` Chen Wandun
0 siblings, 0 replies; 3+ messages in thread
From: Chen Wandun @ 2020-06-11 15:05 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo, Markus.Elfring
Cc: linux-kernel, peterz, mingo, cj.chengjian, chenwandun
在 2020/6/11 21:16, Arnaldo Carvalho de Melo 写道:
> Em Thu, Jun 11, 2020 at 09:42:34AM +0800, Chen Wandun escreveu:
>> Fix potential memory leak in function parse_events_term__sym_hw()
>> and parse_events_term__clone().
>>
>> 1. Free memory when errors occur.
>> 2. Function new_term may return error, so it is need to free memory
>> when the return value is negative.
> Try to fix one thing per patch, i.e. first the most obvious one, then
> the other that requires going thru the new_term() logic, i.e. first
> this, which is super easy to review:
> Would you like to add the tag “Fixes” to the commit message?
>
Hi, Arnaldo and Markus.
Thank you for your reply, I will update in v2
Best Regards,
Chen Wandun
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2020-06-11 15:05 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-06-11 1:42 [PATCH next] perf tools: fix potential memleak in perf events parser Chen Wandun
2020-06-11 13:16 ` Arnaldo Carvalho de Melo
2020-06-11 15:05 ` Chen Wandun
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).