[PATCH 2/2] malloc -> calloc and fix some memory leaks.

[PATCH 2/2] malloc -> calloc and fix some memory leaks.

[Yin Kangkai]

From fd81bea16d3e1e93aebc8cb80ec16ae9c20c6e41 Mon Sep 17 00:00:00 2001
From: Yin Kangkai <kangkai.yin@intel.com>
Date: Mon, 2 Aug 2010 11:22:49 +0800
Subject: [PATCH 2/2] malloc -> calloc and fix some memory leaks.

calloc so that we do not need to memset each of them.

Signed-off-by: Yin Kangkai <kangkai.yin@intel.com>
---
 udev/udev-rules.c |   24 ++++++++++++++++--------
 udev/udevd.c      |    6 ++++--
 2 files changed, 20 insertions(+), 10 deletions(-)

diff --git a/udev/udev-rules.c b/udev/udev-rules.c
index 6d32e73..072b31b 100644
--- a/udev/udev-rules.c
+++ b/udev/udev-rules.c
@@ -1751,23 +1751,27 @@ struct udev_rules *udev_rules_new(struct udev *udev, int resolve_names)
 	struct udev_list_entry *file_loop, *file_tmp;
 	struct token end_token;
 
-	rules = malloc(sizeof(struct udev_rules));
+	rules = calloc(1, sizeof(struct udev_rules));
 	if (rules = NULL)
 		return NULL;
-	memset(rules, 0x00, sizeof(struct udev_rules));
 	rules->udev = udev;
 	rules->resolve_names = resolve_names;
 	udev_list_init(&file_list);
 
 	/* init token array and string buffer */
-	rules->tokens = malloc(PREALLOC_TOKEN * sizeof(struct token));
-	if (rules->tokens = NULL)
+	rules->tokens = calloc(PREALLOC_TOKEN, sizeof(struct token));
+	if (rules->tokens = NULL) {
+		free(rules);
 		return NULL;
+	}
 	rules->token_max = PREALLOC_TOKEN;
 
-	rules->buf = malloc(PREALLOC_STRBUF);
-	if (rules->buf = NULL)
+	rules->buf = calloc(1, PREALLOC_STRBUF);
+	if (rules->buf = NULL) {
+		free(rules->tokens);
+		free(rules);
 		return NULL;
+	}
 	rules->buf_max = PREALLOC_STRBUF;
 	/* offset 0 is always '\0' */
 	rules->buf[0] = '\0';
@@ -1775,9 +1779,13 @@ struct udev_rules *udev_rules_new(struct udev *udev, int resolve_names)
 	dbg(udev, "prealloc %zu bytes tokens (%u * %zu bytes), %zu bytes buffer\n",
 	    rules->token_max * sizeof(struct token), rules->token_max, sizeof(struct token), rules->buf_max);
 
-	rules->trie_nodes = malloc(PREALLOC_TRIE * sizeof(struct trie_node));
-	if (rules->trie_nodes = NULL)
+	rules->trie_nodes = calloc(PREALLOC_TRIE, sizeof(struct trie_node));
+	if (rules->trie_nodes = NULL) {
+		free(rules->buf);
+		free(rules->tokens);
+		free(rules);
 		return NULL;
+	}
 	rules->trie_nodes_max = PREALLOC_TRIE;
 	/* offset 0 is the trie root, with an empty string */
 	memset(rules->trie_nodes, 0x00, sizeof(struct trie_node));
diff --git a/udev/udevd.c b/udev/udevd.c
index 95d4ad8..b882479 100644
--- a/udev/udevd.c
+++ b/udev/udevd.c
@@ -227,8 +227,10 @@ static void worker_new(struct event *event)
 	udev_monitor_enable_receiving(worker_monitor);
 
 	worker = calloc(1, sizeof(struct worker));
-	if (worker = NULL)
+	if (worker = NULL) {
+		udev_monitor_unref(worker_monitor);
 		return;
+	}
 	/* worker + event reference */
 	worker->refcount = 2;
 	worker->udev = event->udev;
@@ -655,7 +657,7 @@ static int handle_inotify(struct udev *udev)
 	if ((ioctl(pfd[FD_INOTIFY].fd, FIONREAD, &nbytes) < 0) || (nbytes <= 0))
 		return 0;
 
-	buf = malloc(nbytes);
+	buf = calloc(1, nbytes);
 	if (buf = NULL) {
 		err(udev, "error getting buffer for inotify\n");
 		return -1;
-- 
1.6.5

Re: [PATCH 2/2] malloc -> calloc and fix some memory leaks.

[Kay Sievers]

On Mon, Aug 2, 2010 at 05:43, Yin Kangkai <kangkai.yin@linux.intel.com> wrote:
> From fd81bea16d3e1e93aebc8cb80ec16ae9c20c6e41 Mon Sep 17 00:00:00 2001
> From: Yin Kangkai <kangkai.yin@intel.com>
> Date: Mon, 2 Aug 2010 11:22:49 +0800
> Subject: [PATCH 2/2] malloc -> calloc and fix some memory leaks.
>
> calloc so that we do not need to memset each of them.

Any reason to change things to calloc() where no memset() is done or needed?

Kay

Re: [PATCH 2/2] malloc -> calloc and fix some memory leaks.

[Yin Kangkai]

On 2010-08-02, 11:00 +0200, Kay Sievers wrote:
> On Mon, Aug 2, 2010 at 05:43, Yin Kangkai <kangkai.yin@linux.intel.com> wrote:
> > From fd81bea16d3e1e93aebc8cb80ec16ae9c20c6e41 Mon Sep 17 00:00:00 2001
> > From: Yin Kangkai <kangkai.yin@intel.com>
> > Date: Mon, 2 Aug 2010 11:22:49 +0800
> > Subject: [PATCH 2/2] malloc -> calloc and fix some memory leaks.
> >
> > calloc so that we do not need to memset each of them.
> 
> Any reason to change things to calloc() where no memset() is done or needed?

I changed the malloc and memset pair of udev_rules allocation, and for
consistence, I finished the others :)

Maybe it's only my personal taste that I like to memset memory once I
malloc success. If this is not necessary, I can split the patch.

Thanks,
Kangkai

Re: [PATCH 2/2] malloc -> calloc and fix some memory leaks.

[Kay Sievers]

On Mon, Aug 2, 2010 at 12:00, Yin Kangkai <kangkai.yin@linux.intel.com> wrote:
> On 2010-08-02, 11:00 +0200, Kay Sievers wrote:
>> On Mon, Aug 2, 2010 at 05:43, Yin Kangkai <kangkai.yin@linux.intel.com> wrote:
>> > From fd81bea16d3e1e93aebc8cb80ec16ae9c20c6e41 Mon Sep 17 00:00:00 2001
>> > From: Yin Kangkai <kangkai.yin@intel.com>
>> > Date: Mon, 2 Aug 2010 11:22:49 +0800
>> > Subject: [PATCH 2/2] malloc -> calloc and fix some memory leaks.
>> >
>> > calloc so that we do not need to memset each of them.
>>
>> Any reason to change things to calloc() where no memset() is done or needed?
>
> I changed the malloc and memset pair of udev_rules allocation, and for
> consistence, I finished the others :)
>
> Maybe it's only my personal taste that I like to memset memory once I
> malloc success. If this is not necessary, I can split the patch.

Nah, I mean memset(), and calloc() is only necessary if the memory is
read, and not always overwritten anyway by the next operation, like
read() or similar. I don't think that needs to be 'fixed'.

Kay