Skip to contentSkip to conversation

use kzalloc instead of malloc+memset

PATCH In review Lion Vollnhals 9 participants 15 messages Sep 12, 2005 – Sep 14, 2005

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 use kzalloc instead of malloc+memset

This patch against 2.6.13-mm3 replaces malloc and memset with kzalloc in drivers/base/class.c . Furthermore this patch fixes actually two bugs:
The memset arguments were occasionally swaped and therefore wrong.

Usage of kzalloc makes this code shorter and more bugfree.

Please apply.

drivers/base/class.c +6 -8
@@ -190,12 +190,11 @@ struct class *class_create(struct module
 	struct class *cls;
 	int retval;
 
-	cls = kmalloc(sizeof(struct class), GFP_KERNEL);
+	cls = kzalloc(sizeof(struct class), GFP_KERNEL);
 	if (!cls) {
 		retval = -ENOMEM;
 		goto error;
 	}
-	memset(cls, 0x00, sizeof(struct class));
 
 	cls->name = name;
 	cls->owner = owner;
@@ -519,13 +518,13 @@ int class_device_add(struct class_device
 	/* add the needed attributes to this device */
 	if (MAJOR(class_dev->devt)) {
 		struct class_device_attribute *attr;
-		attr = kmalloc(sizeof(*attr), GFP_KERNEL);
+		attr = kzalloc(sizeof(*attr), GFP_KERNEL);
 		if (!attr) {
 			error = -ENOMEM;
 			kobject_del(&class_dev->kobj);
 			goto register_done;
 		}
-		memset(attr, sizeof(*attr), 0x00);
+		
 		attr->attr.name = "dev";
 		attr->attr.mode = S_IRUGO;
 		attr->attr.owner = parent->owner;
@@ -534,13 +533,13 @@ int class_device_add(struct class_device
 		class_device_create_file(class_dev, attr);
 		class_dev->devt_attr = attr;
 
-		attr = kmalloc(sizeof(*attr), GFP_KERNEL);
+		attr = kzalloc(sizeof(*attr), GFP_KERNEL);
 		if (!attr) {
 			error = -ENOMEM;
 			kobject_del(&class_dev->kobj);
 			goto register_done;
 		}
-		memset(attr, sizeof(*attr), 0x00);
+		
 		attr->attr.name = "sample.sh";
 		attr->attr.mode = S_IRUSR | S_IXUSR | S_IRUGO;
 		attr->attr.owner = parent->owner;
@@ -611,12 +610,11 @@ struct class_device *class_device_create
 	if (cls == NULL || IS_ERR(cls))
 		goto error;
 
-	class_dev = kmalloc(sizeof(struct class_device), GFP_KERNEL);
+	class_dev = kzalloc(sizeof(struct class_device), GFP_KERNEL);
 	if (!class_dev) {
 		retval = -ENOMEM;
 		goto error;
 	}
-	memset(class_dev, 0x00, sizeof(struct class_device));
 
 	class_dev->devt = devt;
 	class_dev->dev = device;
#1 lore raw
Quoted 2 lines · Lion Vollnhals

This patch against 2.6.13-mm3 replaces malloc and memset with kzalloc in drivers/base/class.c .

The following patch converts further kmalllocs and memsets in drivers/base/* to kzallocs.

Please apply.

drivers/base/attribute_container.c +3 -2
diff -Nurp 2.6.13-mm3/drivers/base/attribute_container.c 2.6.13-mm3-changed/drivers/base/attribute_container.c
@@ -152,12 +152,13 @@ attribute_container_add_device(struct de
 
 		if (!cont->match(cont, dev))
 			continue;
-		ic = kmalloc(sizeof(struct internal_container), GFP_KERNEL);
+		
+		ic = kzalloc(sizeof(struct internal_container), GFP_KERNEL);
 		if (!ic) {
 			dev_printk(KERN_ERR, dev, "failed to allocate class container\n");
 			continue;
 		}
-		memset(ic, 0, sizeof(struct internal_container));
+		
 		ic->cont = cont;
 		class_device_initialize(&ic->classdev);
 		ic->classdev.dev = get_device(dev);
drivers/base/firmware_class.c +3 -6
diff -Nurp 2.6.13-mm3/drivers/base/firmware_class.c 2.6.13-mm3-changed/drivers/base/firmware_class.c
@@ -301,9 +301,9 @@ fw_register_class_device(struct class_de
 			 const char *fw_name, struct device *device)
 {
 	int retval;
-	struct firmware_priv *fw_priv = kmalloc(sizeof (struct firmware_priv),
+	struct firmware_priv *fw_priv = kzalloc(sizeof (struct firmware_priv),
 						GFP_KERNEL);
-	struct class_device *class_dev = kmalloc(sizeof (struct class_device),
+	struct class_device *class_dev = kzalloc(sizeof (struct class_device),
 						 GFP_KERNEL);
 
 	*class_dev_p = NULL;
@@ -313,8 +313,6 @@ fw_register_class_device(struct class_de
 		retval = -ENOMEM;
 		goto error_kfree;
 	}
-	memset(fw_priv, 0, sizeof (*fw_priv));
-	memset(class_dev, 0, sizeof (*class_dev));
 
 	init_completion(&fw_priv->completion);
 	fw_priv->attr_data = firmware_attr_data_tmpl;
@@ -402,14 +400,13 @@ _request_firmware(const struct firmware 
 	if (!firmware_p)
 		return -EINVAL;
 
-	*firmware_p = firmware = kmalloc(sizeof (struct firmware), GFP_KERNEL);
+	*firmware_p = firmware = kzalloc(sizeof (struct firmware), GFP_KERNEL);
 	if (!firmware) {
 		printk(KERN_ERR "%s: kmalloc(struct firmware) failed\n",
 		       __FUNCTION__);
 		retval = -ENOMEM;
 		goto out;
 	}
-	memset(firmware, 0, sizeof (*firmware));
 
 	retval = fw_setup_class_device(firmware, &class_dev, name, device,
 		hotplug);
drivers/base/map.c +1 -2
diff -Nurp 2.6.13-mm3/drivers/base/map.c 2.6.13-mm3-changed/drivers/base/map.c
@@ -135,7 +135,7 @@ retry:
 struct kobj_map *kobj_map_init(kobj_probe_t *base_probe, struct semaphore *sem)
 {
 	struct kobj_map *p = kmalloc(sizeof(struct kobj_map), GFP_KERNEL);
-	struct probe *base = kmalloc(sizeof(struct probe), GFP_KERNEL);
+	struct probe *base = kzalloc(sizeof(struct probe), GFP_KERNEL);
 	int i;
 
 	if ((p == NULL) || (base == NULL)) {
@@ -144,7 +144,6 @@ struct kobj_map *kobj_map_init(kobj_prob
 		return NULL;
 	}
 
-	memset(base, 0, sizeof(struct probe));
 	base->dev = 1;
 	base->range = ~0;
 	base->get = base_probe;
drivers/base/memory.c +1 -3
diff -Nurp 2.6.13-mm3/drivers/base/memory.c 2.6.13-mm3-changed/drivers/base/memory.c
@@ -341,14 +341,12 @@ int add_memory_block(unsigned long node_
 		     unsigned long state, int phys_device)
 {
 	size_t size = sizeof(struct memory_block);
-	struct memory_block *mem = kmalloc(size, GFP_KERNEL);
+	struct memory_block *mem = kzalloc(size, GFP_KERNEL);
 	int ret = 0;
 
 	if (!mem)
 		return -ENOMEM;
 
-	memset(mem, 0, size);
-
 	mem->phys_index = __section_nr(section);
 	mem->state = state;
 	init_MUTEX(&mem->state_sem);
drivers/base/platform.c +1 -2
diff -Nurp 2.6.13-mm3/drivers/base/platform.c 2.6.13-mm3-changed/drivers/base/platform.c
@@ -225,13 +225,12 @@ struct platform_device *platform_device_
 	struct platform_object *pobj;
 	int retval;
 
-	pobj = kmalloc(sizeof(struct platform_object) + sizeof(struct resource) * num, GFP_KERNEL);
+	pobj = kzalloc(sizeof(struct platform_object) + sizeof(struct resource) * num, GFP_KERNEL);
 	if (!pobj) {
 		retval = -ENOMEM;
 		goto error;
 	}
 
-	memset(pobj, 0, sizeof(*pobj));
 	pobj->pdev.name = name;
 	pobj->pdev.id = id;
 	pobj->pdev.dev.release = platform_device_release_simple;
#2 lore raw
Quoted 15 lines · Lion Vollnhals
Quoted 2 lines · Lion Vollnhals

This patch against 2.6.13-mm3 replaces malloc and memset with kzalloc in drivers/base/class.c .

The following patch converts further kmalllocs and memsets in drivers/base/* to kzallocs.

Please apply.

diff -Nurp 2.6.13-mm3/drivers/base/platform.c 2.6.13-mm3-changed/drivers/base/platform.c

drivers/base/platform.c +0 -0
@@ -225,13 +225,12 @@ struct platform_device *platform_device_
 	struct platform_object *pobj;

drivers/base/platform.c struct platform_device *platform_device_ · line 234

Quoted 10 lines
drivers/base/platform.c +1 -2
 	int retval;
 
-	pobj = kmalloc(sizeof(struct platform_object) + sizeof(struct resource) * num, GFP_KERNEL);
+	pobj = kzalloc(sizeof(struct platform_object) + sizeof(struct resource) * num, GFP_KERNEL);
 	if (!pobj) {
 		retval = -ENOMEM;
 		goto error;
 	}
 
-	memset(pobj, 0, sizeof(*pobj));

Was this a bug or did you just introduced one?
--
vda

#3 lore raw

Denis Vlasenko napsal(a):

Quoted 34 lines · Lion Vollnhals
Quoted 33 lines · Lion Vollnhals
Quoted 6 lines

This patch against 2.6.13-mm3 replaces malloc and memset with kzalloc in drivers/base/class.c .

The following patch converts further kmalllocs and memsets in drivers/base/* to kzallocs.

Please apply.

diff -Nurp 2.6.13-mm3/drivers/base/platform.c 2.6.13-mm3-changed/drivers/base/platform.c

drivers/base/platform.c +1 -1
@@ -225,13 +225,12 @@ struct platform_device *platform_device_
 	struct platform_object *pobj;
 	int retval;
 
-	pobj = kmalloc(sizeof(struct platform_object) + sizeof(struct resource) * num, GFP_KERNEL);
+	pobj = kzalloc(sizeof(struct platform_object) + sizeof(struct resource) * num, GFP_KERNEL);
 	if (!pobj) {
 		retval = -ENOMEM;
 		goto error;
 	}
 
Quoted 7 lines

drivers/base/platform.c struct platform_device *platform_device_ · line 235

Quoted 3 lines
drivers/base/platform.c +0 -1
-	memset(pobj, 0, sizeof(*pobj));
 
 

Was this a bug or did you just introduced one?

Yes, this was a bug.

Signature · 3 lines
Jiri Slaby         www.fi.muni.cz/~xslaby
~\-/~      [email protected]      ~\-/~
241B347EC88228DE51EE A49C4A73A25004CB2A10
#4 lore raw
In reply to Lion Vollnhals

Lion Vollnhals napsal(a):

Quoted 13 lines

This patch against 2.6.13-mm3 replaces malloc and memset with kzalloc in drivers/base/class.c . Furthermore this patch fixes actually two bugs:
The memset arguments were occasionally swaped and therefore wrong.

Usage of kzalloc makes this code shorter and more bugfree.

Please apply.

drivers/base/class.c struct class *class_create(struct module · line 195

Quoted 7 lines
drivers/base/class.c +1 -1
@@ -190,12 +190,11 @@ struct class *class_create(struct module
 	struct class *cls;
 	int retval;
 
-	cls = kmalloc(sizeof(struct class), GFP_KERNEL);
+	cls = kzalloc(sizeof(struct class), GFP_KERNEL);
 
 

maybe, the better way is to write *cls' instead of struct class', better for further changes

Quoted 10 lines
drivers/base/class.c +0 -1
 	if (!cls) {
 		retval = -ENOMEM;
 		goto error;
 	}
-	memset(cls, 0x00, sizeof(struct class));
 
	cls->name = name;
	cls->owner = owner;

[snip]

drivers/base/class.c struct class_device *class_device_create · line 615

Quoted 8 lines
drivers/base/class.c +1 -1
@@ -611,12 +610,11 @@ struct class_device *class_device_create
 	if (cls == NULL || IS_ERR(cls))
 		goto error;
 
-	class_dev = kmalloc(sizeof(struct class_device), GFP_KERNEL);
+	class_dev = kzalloc(sizeof(struct class_device), GFP_KERNEL);
 
 

*class_dev' instead of struct class_device'

Quoted 10 lines
drivers/base/class.c +0 -1
 	if (!class_dev) {
 		retval = -ENOMEM;
 		goto error;
 	}
-	memset(class_dev, 0x00, sizeof(struct class_device));
 
	class_dev->devt = devt;
	class_dev->dev = device;

thanks,

Signature · 3 lines
Jiri Slaby         www.fi.muni.cz/~xslaby
~\-/~      [email protected]      ~\-/~
241B347EC88228DE51EE A49C4A73A25004CB2A10
#5 lore raw
Quoted 6 lines · Jiri Slaby

drivers/base/class.c struct class *class_create(struct module · line 193

Quoted 4 lines
drivers/base/class.c +1 -1
-      cls = kmalloc(sizeof(struct class), GFP_KERNEL);
+      cls = kzalloc(sizeof(struct class), GFP_KERNEL);

maybe, the better way is to write *cls' instead of struct class', better for further changes

Please note that some maintainers don't like it. I at least could not sneak in patches like these to drivers/usb/ because I had changed sizeof.

Pekka

#6 lore raw
Quoted 5 lines · Pekka Enberg
Quoted 4 lines · Jiri Slaby
Quoted 4 lines
drivers/base/class.c +1 -1
-      cls = kmalloc(sizeof(struct class), GFP_KERNEL);
+      cls = kzalloc(sizeof(struct class), GFP_KERNEL);
Quoted 7 lines
Quoted 2 lines

maybe, the better way is to write *cls' instead of struct class', better for further changes

Please note that some maintainers don't like it. I at least could not sneak in patches like these to drivers/usb/ because I had changed sizeof.

And given the fact that Greg maintains driver core it probably won't be accepted here either :)

FWIW I also prefer spelling out the structure I am allocating.

Signature · 1 line
Dmitry
#7 lore raw
Quoted 15 lines · Dmitry Torokhov
Quoted 5 lines · Pekka Enberg
Quoted 4 lines · Jiri Slaby
Quoted 4 lines
drivers/base/class.c +1 -1
-      cls = kmalloc(sizeof(struct class), GFP_KERNEL);
+      cls = kzalloc(sizeof(struct class), GFP_KERNEL);
Quoted 7 lines
Quoted 2 lines

maybe, the better way is to write *cls' instead of struct class', better for further changes

Please note that some maintainers don't like it. I at least could not sneak in patches like these to drivers/usb/ because I had changed sizeof.

Quoted 5 lines

And given the fact that Greg maintains driver core it probably won't be accepted here either :)

FWIW I also prefer spelling out the structure I am allocating.

It hurts readability. Quick question: is this code correct?

	dev = kmalloc(sizeof(struct net_device), GFP_KERNEL);

you don't know. You have to go hunting down the declaration of `dev' to find out.

#8 lore raw
Quoted 1 line · Dmitry Torokhov
Quoted 1 line

FWIW I also prefer spelling out the structure I am allocating.

Quoted 6 lines · Andrew Morton

It hurts readability. Quick question: is this code correct?

	dev = kmalloc(sizeof(struct net_device), GFP_KERNEL);

you don't know. You have to go hunting down the declaration of `dev' to find out.

Andrew, how about something like this?

Pekka

[PATCH] CodingStyle: memory allocation

This patch adds a new chapter on memory allocation to Documentation/CodingStyle.

---

 CodingStyle |   21 ++++++++++++++++++++-
 1 files changed, 20 insertions(+), 1 deletion(-)
Documentation/CodingStyle +20 -1
Index: 2.6-mm/Documentation/CodingStyle
===================================================================
@@ -410,7 +410,26 @@ Kernel messages do not have to be termin
 Printing numbers in parentheses (%d) adds no value and should be avoided.
 
 
-		Chapter 13: References
+		Chapter 13: Allocating memory
+
+The kernel provides the following general purpose memory allocators:
+kmalloc(), kzalloc(), kcalloc(), and vmalloc().  Please refer to the API
+documentation for further information about them.
+
+The preferred form for passing a size of a struct is the following:
+
+	p = kmalloc(sizeof(*p), ...);
+
+The alternative form where struct name is spelled out hurts readability and
+introduces an opportunity for a bug when the pointer variable type is changed
+but the corresponding sizeof that is passed to a memory allocator is not.
+
+Casting the return value which is a void pointer is redundant. The conversion
+from void pointer to any other pointer type is guaranteed by the C programming
+language.
+
+
+		Chapter 14: References
 
 The C Programming Language, Second Edition
 by Brian W. Kernighan and Dennis M. Ritchie.
#9 lore raw

Pekka J Enberg writes:

[...]

Documentation/CodingStyle Kernel messages do not have to be termin · line 421

Quoted 8 lines
Documentation/CodingStyle +8 -0
+
+The kernel provides the following general purpose memory allocators:
+kmalloc(), kzalloc(), kcalloc(), and vmalloc().  Please refer to the API
+documentation for further information about them.
+
+The preferred form for passing a size of a struct is the following:
+
+	p = kmalloc(sizeof(*p), ...);

Parentheses around *p are superfluous. See

Quoted 2 lines

The C Programming Language, Second Edition
by Brian W. Kernighan and Dennis M. Ritchie.

Nikita.

#10 lore raw
Quoted 7 lines · Pekka J Enberg
Quoted 7 lines
Documentation/CodingStyle +7 -0
+The kernel provides the following general purpose memory allocators:
+kmalloc(), kzalloc(), kcalloc(), and vmalloc().  Please refer to the API
+documentation for further information about them.
+
+The preferred form for passing a size of a struct is the following:
+
+	p = kmalloc(sizeof(*p), ...);
Quoted 1 line · Nikita Danilov

Parentheses around *p are superfluous. See

Sure but is it the preferred form for the kernel?

Pekka

#11 lore raw
In reply to Nikita Danilov
Quoted 4 lines · Nikita Danilov

Pekka J Enberg writes:

[...]

Quoted 13 lines

Documentation/CodingStyle Kernel messages do not have to be termin · line 421

Quoted 8 lines
Documentation/CodingStyle +8 -0
+
+The kernel provides the following general purpose memory allocators:
+kmalloc(), kzalloc(), kcalloc(), and vmalloc().  Please refer to the API
+documentation for further information about them.
+
+The preferred form for passing a size of a struct is the following:
+
+	p = kmalloc(sizeof(*p), ...);

Parentheses around *p are superfluous. See

Quoted 2 lines

The C Programming Language, Second Edition
by Brian W. Kernighan and Dennis M. Ritchie.

I remember that sizeof has two forms: sizeof(type) and sizeof(expr), and in one of them ()'s are optional. But I fail to remember in which one. I use ()'s always.

Thanks for refreshing my memory but I'm sure I'll forget again ;)
--
vda

#12 lore raw

Denis Vlasenko writes:

[...]

Quoted 4 lines

I remember that sizeof has two forms: sizeof(type) and sizeof(expr), and in one of them ()'s are optional. But I fail to remember in which one. I use ()'s always.

Formally speaking, sizeof have forms

sizeof(type), and

sizeof expr

it is just that expression can usually be wrapped into parentheses, like (((((0))))).

Quoted 3 lines

Thanks for refreshing my memory but I'm sure I'll forget again ;)

That's why we need more instances of sizeof expr in the kernel code, to keep your knowledge of C afresh all the time. :-)

Note that Linux doesn't follow a custom of... some other kernels to parenthesize everything to the heretical extent of writing

   if ((a == 0) && (b == 1))

or

   return (foo);
Quoted 2 lines

--
vda

Nikita.

#13 lore raw
In reply to Andrew Morton
Quoted 3 lines · Andrew

It hurts readability. Quick question: is this code correct?

	dev = kmalloc(sizeof(struct net_device), GFP_KERNEL);

And it hurts maintainability. If someone changes 'dev' so that it is no longer of type 'struct net_device', then they risk missing this allocation, and introducing what could be a nasty memory corruption kernel bug.

Signature · 3 lines
                  I won't rest till it's the best ...
                  Programmer, Linux Scalability
                  Paul Jackson <[email protected]> 1.925.600.0401
#14 lore raw
Quoted 7 lines · Paul Jackson
Quoted 3 lines · Andrew

It hurts readability. Quick question: is this code correct?

	dev = kmalloc(sizeof(struct net_device), GFP_KERNEL);

And it hurts maintainability. If someone changes 'dev' so that it is no longer of type 'struct net_device', then they

'struct net_device *'

Quoted 2 lines

risk missing this allocation, and introducing what could be a nasty memory corruption kernel bug.

SCNR,
Bernd

Signature · 3 lines
Firmix Software GmbH                   http://www.firmix.at/
mobil: +43 664 4416156                 fax: +43 1 7890849-55
          Embedded Linux Development and Services