git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Erik Faye-Lund <kusmabite@gmail.com>
To: git@vger.kernel.org
Cc: matthieu.moy@grenoble-inp.fr, msysgit@googlegroups.com, blees@dcon.de
Subject: Re: [PATCH v2] commit: fix memory-leak
Date: Mon, 7 Feb 2011 22:31:14 +0100	[thread overview]
Message-ID: <AANLkTimvgDjiLon-2BAwxcYOQGVE9UZPNr637o93-9HQ@mail.gmail.com> (raw)
In-Reply-To: <AANLkTikr2+OVRU6n+0tA752_x80ir9dQh65RjUp3BxPR@mail.gmail.com>

On Mon, Feb 7, 2011 at 10:12 PM, Erik Faye-Lund <kusmabite@gmail.com> wrote:
> On Mon, Feb 7, 2011 at 9:21 PM, Erik Faye-Lund <kusmabite@gmail.com> wrote:
>> The name, email and date strings are some times allocated on the
>> heap, but not free'd. Fix this by making sure they are allways
>> heap-allocated, so we can safely free the memory.
>>
>> At the same time, this fixes a problem with strict-POSIX getenv
>> implementations. POSIX says "The return value from getenv() may
>> point to static data which may be overwritten by subsequent calls
>> to getenv()", so not duplicating the strings is a potential bug.
>>
>> Signed-off-by: Erik Faye-Lund <kusmabite@gmail.com>
>> ---
>> Fixed typo in commit message, as pointed out by Matthieu Moy.
>>
>>  builtin/commit.c  |    9 ++++++---
>>  git-compat-util.h |    1 +
>>  wrapper.c         |    6 ++++++
>>  3 files changed, 13 insertions(+), 3 deletions(-)
>>
>> diff --git a/builtin/commit.c b/builtin/commit.c
>> index 03cff5a..e5a649e 100644
>> --- a/builtin/commit.c
>> +++ b/builtin/commit.c
>> @@ -465,9 +465,9 @@ static void determine_author_info(struct strbuf *author_ident)
>>  {
>>        char *name, *email, *date;
>>
>> -       name = getenv("GIT_AUTHOR_NAME");
>> -       email = getenv("GIT_AUTHOR_EMAIL");
>> -       date = getenv("GIT_AUTHOR_DATE");
>> +       name = xgetenv("GIT_AUTHOR_NAME");
>> +       email = xgetenv("GIT_AUTHOR_EMAIL");
>> +       date = xgetenv("GIT_AUTHOR_DATE");
>>
>>        if (use_message && !renew_authorship) {
>>                const char *a, *lb, *rb, *eol;
>> @@ -507,6 +507,9 @@ static void determine_author_info(struct strbuf *author_ident)
>>                date = force_date;
>>        strbuf_addstr(author_ident, fmt_ident(name, email, date,
>>                                              IDENT_ERROR_ON_NO_NAME));
>> +       free(name);
>> +       free(email);
>> +       free(date);
>
> Hmm, but I'm getting a crash here on Linux. Guess I need to debug a bit...
>

Ah, it was the force_date-assignment:
---8<---
diff --git a/builtin/commit.c b/builtin/commit.c
index e5a649e..1416c13 100644
--- a/builtin/commit.c
+++ b/builtin/commit.c
@@ -504,7 +504,7 @@ static void determine_author_info(struct strbuf
*author_ident)
 	}

 	if (force_date)
-		date = force_date;
+		date = xstrdup(force_date);
 	strbuf_addstr(author_ident, fmt_ident(name, email, date,
 					      IDENT_ERROR_ON_NO_NAME));
 	free(name);

---8<---

But now I see that I was temporarily(?) struck with insanity:
overwriting a heap-allocated pointer with another heap-allocated
pointer doesn't fix a memory-leak. So let's add some more calls to
free:

diff --git a/builtin/commit.c b/builtin/commit.c
index e5a649e..bdd0cfb 100644
--- a/builtin/commit.c
+++ b/builtin/commit.c
@@ -482,6 +482,10 @@ static void determine_author_info(struct strbuf
*author_ident)
 		if (!*lb || !*rb || !*eol)
 			die("invalid commit: %s", use_message);

+		free(name);
+		free(email);
+		free(date);
+
 		if (lb == a + strlen("\nauthor "))
 			/* \nauthor <foo@example.com> */
 			name = xcalloc(1, 1);
@@ -497,14 +501,19 @@ static void determine_author_info(struct strbuf
*author_ident)
 		const char *lb = strstr(force_author, " <");
 		const char *rb = strchr(force_author, '>');

+		free(name);
+		free(email);
+
 		if (!lb || !rb)
 			die("malformed --author parameter");
 		name = xstrndup(force_author, lb - force_author);
 		email = xstrndup(lb + 2, rb - (lb + 2));
 	}

-	if (force_date)
-		date = force_date;
+	if (force_date) {
+		free(date);
+		date = xstrdup(force_date);
+	}
 	strbuf_addstr(author_ident, fmt_ident(name, email, date,
 					      IDENT_ERROR_ON_NO_NAME));
 	free(name);

      reply	other threads:[~2011-02-07 21:31 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-02-07 18:40 [PATCH/RFC] commit: fix memory-leak Erik Faye-Lund
2011-02-07 18:48 ` Matthieu Moy
2011-02-07 19:22   ` Erik Faye-Lund
2011-02-07 20:21     ` [PATCH v2] " Erik Faye-Lund
2011-02-07 21:12       ` Erik Faye-Lund
2011-02-07 21:31         ` Erik Faye-Lund [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=AANLkTimvgDjiLon-2BAwxcYOQGVE9UZPNr637o93-9HQ@mail.gmail.com \
    --to=kusmabite@gmail.com \
    --cc=blees@dcon.de \
    --cc=git@vger.kernel.org \
    --cc=matthieu.moy@grenoble-inp.fr \
    --cc=msysgit@googlegroups.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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).