Re: + drivers-block-floppyc-remove-copyin-copyout-and-ecall-macros.patch added to -mm tree

Re: + drivers-block-floppyc-remove-copyin-copyout-and-ecall-macros.patch added to -mm tree

[Jiri Slaby]

On 01/27/2010 01:37 AM, akpm@linux-foundation.org wrote:
> @@ -3145,7 +3136,9 @@ static inline int raw_cmd_copyout(int cm
>  	int ret;
>  
>  	while (ptr) {
> -		COPYOUT(*ptr);
> +		ret = copy_to_user((void __user *)param, ptr, sizeof(*ptr));
> +		if (ret)
> +			return -EFAULT;
>  		param += sizeof(struct floppy_raw_cmd);
>  		if ((ptr->flags & FD_RAW_READ) && ptr->buffer_length) {
>  			if (ptr->length >= 0 &&
> @@ -3195,7 +3188,9 @@ static inline int raw_cmd_copyin(int cmd
>  		if (!ptr)
>  			return -ENOMEM;
>  		*rcmd = ptr;
> -		COPYIN(*ptr);
> +		ret = copy_from_user(ptr, (void __user *)param, sizeof(*ptr));

Actually the casts are not needed anymore, are they?

-- 
js

Re: + drivers-block-floppyc-remove-copyin-copyout-and-ecall-macros.patch added to -mm tree

[Joe Perches]

On Wed, 2010-01-27 at 10:43 +0100, Jiri Slaby wrote:
> On 01/27/2010 01:37 AM, akpm@linux-foundation.org wrote:
> > @@ -3145,7 +3136,9 @@ static inline int raw_cmd_copyout(int cm
> > -		COPYOUT(*ptr);
> > +		ret = copy_to_user((void __user *)param, ptr, sizeof(*ptr));
> > +		if (ret)
> > +			return -EFAULT;
[]
> > -		COPYIN(*ptr);
> > +		ret = copy_from_user(ptr, (void __user *)param, sizeof(*ptr));
> 
> Actually the casts are not needed anymore, are they?

There's still some char/void silliness.
It certainly could be improved/removed.

Do feel free to submit patches with your
desired level of changelog prolixity.

cheers, Joe

Re: + drivers-block-floppyc-remove-copyin-copyout-and-ecall-macros.patch added to -mm tree

[Jiri Slaby]

On 01/27/2010 03:53 PM, Joe Perches wrote:
> Do feel free to submit patches with your
> desired level of changelog prolixity.

It's not my level changelog prolixity, it's you do changes irrelevant to
your changelog and further you do not explain, why you change the code.
So (a) review process is hard and (b) nobody (mainly distro kernel
packagers) knows whether to backport some of the change to stable or not.

Look at that piece again:
Subject	[PATCH 03/24] drivers/block/floppy.c: Use pr_<level>
Date	Thu, 21 Jan 2010 20:52:33 -0800

Convert bare printk to pr_info and pr_cont
Convert printk(KERN_ERR to pr_err
...
 	if (UDP->flags & FD_DEBUG) {
-		DPRINT("reschedule timeout ");
-		printk(message, marg);
-		printk("\n");
+		DPRINT("reschedule timeout %s %d\n", message, marg);
 	}

Does it look like a "Use pr_<level>" change? No and hence my confusion
by the hunk.

-- 
js