Skip to contentSkip to conversation

USB Elan FTDI: check for workqueue creation

PATCH In review Cyrill Gorcunov 2 participants 6 messages Feb 19, 2007 – Feb 24, 2007 1 message not in the archive USB

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 USB Elan FTDI: check for workqueue creation

This patch prevents from NULL pointer usage if
workqueue creation failed.

---

 drivers/usb/misc/ftdi-elan.c |   24 +++++++++++++++++++-----
 1 files changed, 19 insertions(+), 5 deletions(-)
drivers/usb/misc/ftdi-elan.c +19 -5
index 0c1d66d..67a25e9 100644
@@ -57,9 +57,9 @@ module_param(distrust_firmware, bool, 0);
 MODULE_PARM_DESC(distrust_firmware, "true to distrust firmware power/overcurren"
         "t setup");
 extern struct platform_driver u132_platform_driver;
-static struct workqueue_struct *status_queue;
-static struct workqueue_struct *command_queue;
-static struct workqueue_struct *respond_queue;
+static struct workqueue_struct *status_queue = NULL;
+static struct workqueue_struct *command_queue = NULL;
+static struct workqueue_struct *respond_queue = NULL;
 /*
 * ftdi_module_lock exists to protect access to global variables
 *
@@ -2905,17 +2905,31 @@ static int __init ftdi_elan_init(void)
 {
         int result;
         printk(KERN_INFO "driver %s built at %s on %s\n", ftdi_elan_driver.name,
-                 __TIME__, __DATE__);
+	       __TIME__, __DATE__);
         init_MUTEX(&ftdi_module_lock);
         INIT_LIST_HEAD(&ftdi_static_list);
         status_queue = create_singlethread_workqueue("ftdi-status-control");
+	if (!status_queue)
+		goto err1;
         command_queue = create_singlethread_workqueue("ftdi-command-engine");
+	if (!command_queue)
+		goto err2;
         respond_queue = create_singlethread_workqueue("ftdi-respond-engine");
+	if (!respond_queue)
+		goto err3;
         result = usb_register(&ftdi_elan_driver);
         if (result)
                 printk(KERN_ERR "usb_register failed. Error number %d\n",
-                        result);
+		       result);
         return result;
+
+ err3:
+	destroy_workqueue(command_queue);
+ err2:
+	destroy_workqueue(status_queue);
+ err1:
+	printk(KERN_ERR "%s couldn't create workqueue\n", ftdi_elan_driver.name);
+	return -ENOMEM;
 }
 
 static void __exit ftdi_elan_exit(void)
#1 lore raw

drivers/usb/misc/ftdi-elan.c module_param(distrust_firmware, bool, 0); · line 60

Quoted 7 lines · Cyrill Gorcunov

+++ b/drivers/usb/misc/ftdi-elan.c

drivers/usb/misc/ftdi-elan.c +1 -1
@@ -57,9 +57,9 @@ module_param(distrust_firmware, bool, 0);
 MODULE_PARM_DESC(distrust_firmware, "true to distrust firmware power/overcurren"
         "t setup");
 extern struct platform_driver u132_platform_driver;
-static struct workqueue_struct *status_queue;
+static struct workqueue_struct *status_queue = NULL;

You better drop this part. Someone is bound to object.

-- Pete

#2 lore raw

On Thu, Feb 22, 2007 at 03:55:23PM -0800, Pete Zaitcev wrote:

| On Mon, 19 Feb 2007 21:15:49 +0300, Cyrill Gorcunov <[email protected]> wrote:
| 
| > +++ b/drivers/usb/misc/ftdi-elan.c
| > @@ -57,9 +57,9 @@ module_param(distrust_firmware, bool, 0);
| >  MODULE_PARM_DESC(distrust_firmware, "true to distrust firmware power/overcurren"
| >          "t setup");
| >  extern struct platform_driver u132_platform_driver;
| > -static struct workqueue_struct *status_queue;
| > +static struct workqueue_struct *status_queue = NULL;
| 
| You better drop this part. Someone is bound to object.
| 
| -- Pete
| 

Hi,

actually I don't understand why... Event on ftdi_elan_exit() status_queue is setting up to NULL. And what is bound to object? Could you write more detailed?

Cyrill

A message that is not in the archive (sent off-list or not yet received): [email protected]

#4 lore raw

On Thu, Feb 22, 2007 at 11:57:38PM -0800, Pete Zaitcev wrote:

| On Fri, 23 Feb 2007 10:53:22 +0300, Cyrill Gorcunov <[email protected]> wrote:
| 
| > | > -static struct workqueue_struct *status_queue;
| > | > +static struct workqueue_struct *status_queue = NULL;
| 
| > | You better drop this part. Someone is bound to object.
| 
| > actually I don't understand why... Event on ftdi_elan_exit()
| > status_queue is setting up to NULL. And what is bound to object?
| > Could you write more detailed?
| 
| All these variables are static, they are already initialized to zero.
| 
| -- Pete
| 

Pete,

I may be wrong, but a lot of the kernel code have static pointers initialized to NULL with explicit manner... More over I always thought that _static_ is not mean _initialized to zero_. I think _static_ is just the method to _hide_ variables in the file (as ANSI C describes). Am I wrong?

Cyrill

#5 lore raw
Quoted 5 lines · Cyrill Gorcunov

I may be wrong, but a lot of the kernel code have static pointers initialized to NULL with explicit manner... More over I always thought that _static_ is not mean _initialized to zero_. I think _static_ is just the method to _hide_ variables in the file (as ANSI C describes). Am I wrong?

I'm afraid you are wrong here. Static variables are initialized on zero in C, although I cannot provide you with a relevant quote from a standard. Just trust me for now, and resubmit the patch without the first segment... :-) You're clearly fixing a bug in it.

-- Pete

#6 lore raw

On Fri, Feb 23, 2007 at 05:43:37PM -0800, Pete Zaitcev wrote:

| On Fri, 23 Feb 2007 11:10:05 +0300, Cyrill Gorcunov <[email protected]> wrote:
| 
| > I may be wrong, but a lot of the kernel code have static pointers
| > initialized to NULL with explicit manner... More over I always thought
| > that _static_ is not mean _initialized to zero_. I think _static_ is
| > just the method to _hide_ variables in the file (as ANSI C describes).
| > Am I wrong?
| 
| I'm afraid you are wrong here. Static variables are initialized on
| zero in C, although I cannot provide you with a relevant quote from
| a standard. Just trust me for now, and resubmit the patch without
| the first segment... :-)  You're clearly fixing a bug in it.
| 
| -- Pete
| 

Hi Pete,

OK, I'll send a new one ;)

Cyrill