[PATCH 1/1] usbip/userspace/libsrc/names.c: memory leak

[PATCH 1/1] usbip/userspace/libsrc/names.c: memory leak

[xypron.glpk]

From: Heinrich Schuchardt <xypron.glpk@gmx.de>

p is freed if NULL.
p is leaked if second calloc fails.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
---
 drivers/staging/usbip/userspace/libsrc/names.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/staging/usbip/userspace/libsrc/names.c b/drivers/staging/usbip/userspace/libsrc/names.c
index 3c8d28b..b2904e8 100644
--- a/drivers/staging/usbip/userspace/libsrc/names.c
+++ b/drivers/staging/usbip/userspace/libsrc/names.c
@@ -170,12 +170,12 @@ static void *my_malloc(size_t size)
 
 	p = calloc(1, sizeof(struct pool));
 	if (!p) {
-		free(p);
 		return NULL;
 	}
 
 	p->mem = calloc(1, size);
 	if (!p->mem)
+		free(p);
 		return NULL;
 
 	p->next = pool_head;
-- 
1.7.10.4

Re: [PATCH 1/1] usbip/userspace/libsrc/names.c: memory leak

[Greg KH]

On Mon, Jan 27, 2014 at 11:29:48PM +0100, xypron.glpk@gmx.de wrote:
> From: Heinrich Schuchardt <xypron.glpk@gmx.de>
> 
> p is freed if NULL.

Not a real problem, right?

> p is leaked if second calloc fails.

You just created a new bug in your "fix" :(

Please run your patches through scripts/checkpatch.pl, odds are, it
would have caught this error.

greg k-h

Re: [PATCH 1/1] usbip/userspace/libsrc/names.c: memory leak

[Dan Carpenter]

On Mon, Jan 27, 2014 at 11:29:48PM +0100, xypron.glpk@gmx.de wrote:
> From: Heinrich Schuchardt <xypron.glpk@gmx.de>

Fix your email account so we can get this automatically from your email.
Currently the From header on your email is just:

From: xypron.glpk@gmx.de

without your name.

> 
> p is freed if NULL.
> p is leaked if second calloc fails.
> 
> Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
> ---
>  drivers/staging/usbip/userspace/libsrc/names.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/staging/usbip/userspace/libsrc/names.c b/drivers/staging/usbip/userspace/libsrc/names.c
> index 3c8d28b..b2904e8 100644
> --- a/drivers/staging/usbip/userspace/libsrc/names.c
> +++ b/drivers/staging/usbip/userspace/libsrc/names.c
> @@ -170,12 +170,12 @@ static void *my_malloc(size_t size)
>  
>  	p = calloc(1, sizeof(struct pool));
>  	if (!p) {
> -		free(p);
>  		return NULL;
>  	}

Remove the curly braces from here since they are no longer needed.

>  
>  	p->mem = calloc(1, size);
>  	if (!p->mem)
> +		free(p);
>  		return NULL;

Add the curly braces here.  They are required.

regards,
dan carpenter

Re: [PATCH 1/1] usbip/userspace/libsrc/names.c: memory leak

[Dan Carpenter]

On Mon, Jan 27, 2014 at 02:50:04PM -0800, Greg KH wrote:
> On Mon, Jan 27, 2014 at 11:29:48PM +0100, xypron.glpk@gmx.de wrote:
> > From: Heinrich Schuchardt <xypron.glpk@gmx.de>
> > 
> > p is freed if NULL.
> 
> Not a real problem, right?
> 
> > p is leaked if second calloc fails.
> 
> You just created a new bug in your "fix" :(
> 
> Please run your patches through scripts/checkpatch.pl, odds are, it
> would have caught this error.
> 

Checkpatch doesn't catch the problems here.  I thought it would have
caught the style issue but apparently it only looks for extra curly
braces when you run it in --file mode.

Fengguang would hopefully have caught the missing curly braces bug with
Coccinelle.

regards,
dan carpenter

Re: [PATCH 1/1] usbip/userspace/libsrc/names.c: memory leak

[Greg KH]

On Tue, Jan 28, 2014 at 02:02:12AM +0300, Dan Carpenter wrote:
> On Mon, Jan 27, 2014 at 02:50:04PM -0800, Greg KH wrote:
> > On Mon, Jan 27, 2014 at 11:29:48PM +0100, xypron.glpk@gmx.de wrote:
> > > From: Heinrich Schuchardt <xypron.glpk@gmx.de>
> > > 
> > > p is freed if NULL.
> > 
> > Not a real problem, right?
> > 
> > > p is leaked if second calloc fails.
> > 
> > You just created a new bug in your "fix" :(
> > 
> > Please run your patches through scripts/checkpatch.pl, odds are, it
> > would have caught this error.
> > 
> 
> Checkpatch doesn't catch the problems here.  I thought it would have
> caught the style issue but apparently it only looks for extra curly
> braces when you run it in --file mode.

Ah, that's good to know.

> Fengguang would hopefully have caught the missing curly braces bug with
> Coccinelle.

Is Coccinelle run on the userspace .c code in the kernel?

thanks,

greg k-h

Re: [PATCH 1/1] usbip/userspace/libsrc/names.c: memory leak

[Sergei Shtylyov]

Hello.

On 01/28/2014 01:29 AM, xypron.glpk@gmx.de wrote:

> From: Heinrich Schuchardt <xypron.glpk@gmx.de>

> p is freed if NULL.

    This is not an issue.

> p is leaked if second calloc fails.
>
> Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
> ---
>   drivers/staging/usbip/userspace/libsrc/names.c |    2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/staging/usbip/userspace/libsrc/names.c b/drivers/staging/usbip/userspace/libsrc/names.c
> index 3c8d28b..b2904e8 100644
> --- a/drivers/staging/usbip/userspace/libsrc/names.c
> +++ b/drivers/staging/usbip/userspace/libsrc/names.c
> @@ -170,12 +170,12 @@ static void *my_malloc(size_t size)
>
>   	p = calloc(1, sizeof(struct pool));
>   	if (!p) {
> -		free(p);
>   		return NULL;
>   	}

    {} not needed anymore.

>
>   	p->mem = calloc(1, size);
>   	if (!p->mem)
> +		free(p);
>   		return NULL;

    How about {} around the *if* arm?

WBR, Sergei

Re: [PATCH 1/1] usbip/userspace/libsrc/names.c: memory leak

[Dan Carpenter]

On Mon, Jan 27, 2014 at 03:27:36PM -0800, Greg KH wrote:
> On Tue, Jan 28, 2014 at 02:02:12AM +0300, Dan Carpenter wrote:
> > On Mon, Jan 27, 2014 at 02:50:04PM -0800, Greg KH wrote:
> > > On Mon, Jan 27, 2014 at 11:29:48PM +0100, xypron.glpk@gmx.de wrote:
> > > > From: Heinrich Schuchardt <xypron.glpk@gmx.de>
> > > > 
> > > > p is freed if NULL.
> > > 
> > > Not a real problem, right?
> > > 
> > > > p is leaked if second calloc fails.
> > > 
> > > You just created a new bug in your "fix" :(
> > > 
> > > Please run your patches through scripts/checkpatch.pl, odds are, it
> > > would have caught this error.
> > > 
> > 
> > Checkpatch doesn't catch the problems here.  I thought it would have
> > caught the style issue but apparently it only looks for extra curly
> > braces when you run it in --file mode.
> 
> Ah, that's good to know.
> 
> > Fengguang would hopefully have caught the missing curly braces bug with
> > Coccinelle.
> 
> Is Coccinelle run on the userspace .c code in the kernel?

Hm...  I'm not sure.  Fengguang, do you know if we would have caught
the missing curly braces in this patch?

http://lkml.org/lkml/2014/1/27/460

regards,
dan carpenter

[PATCH 1/1] usbip/userspace/libsrc/names.c: memory leak

[xypron.glpk]

From: Heinrich Schuchardt <xypron.glpk@gmx.de>

revised patch

p is freed if NULL.
p is leaked if second calloc fails.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
---
 drivers/staging/usbip/userspace/libsrc/names.c |    8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/staging/usbip/userspace/libsrc/names.c b/drivers/staging/usbip/userspace/libsrc/names.c
index 3c8d28b..81ff852 100644
--- a/drivers/staging/usbip/userspace/libsrc/names.c
+++ b/drivers/staging/usbip/userspace/libsrc/names.c
@@ -169,14 +169,14 @@ static void *my_malloc(size_t size)
 	struct pool *p;
 
 	p = calloc(1, sizeof(struct pool));
-	if (!p) {
-		free(p);
+	if (!p)
 		return NULL;
-	}
 
 	p->mem = calloc(1, size);
-	if (!p->mem)
+	if (!p->mem) {
+		free(p);
 		return NULL;
+	}
 
 	p->next = pool_head;
 	pool_head = p;
-- 
1.7.10.4