[PATCH] Fix vde network backend in user mode linux

[PATCH] Fix vde network backend in user mode linux

[Luca Bigliardi]

* Replace kmalloc() with uml_kmalloc()
  (fix build failure)
* Remove unnecessary UM_KERN_INFO in printk()
  (don't display '<6>' while printing info)

Signed-off-by: Luca Bigliardi <shammash@artha.org>
---
 arch/um/drivers/vde_user.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/arch/um/drivers/vde_user.c b/arch/um/drivers/vde_user.c
index 56533db..c5c4325 100644
--- a/arch/um/drivers/vde_user.c
+++ b/arch/um/drivers/vde_user.c
@@ -78,7 +78,7 @@ void vde_init_libstuff(struct vde_data *vpri, struct vde_init *init)
 {
 	struct vde_open_args *args;
 
-	vpri->args = kmalloc(sizeof(struct vde_open_args), UM_GFP_KERNEL);
+	vpri->args = uml_kmalloc(sizeof(struct vde_open_args), UM_GFP_KERNEL);
 	if (vpri->args == NULL) {
 		printk(UM_KERN_ERR "vde_init_libstuff - vde_open_args "
 		       "allocation failed");
@@ -91,8 +91,8 @@ void vde_init_libstuff(struct vde_data *vpri, struct vde_init *init)
 	args->group = init->group;
 	args->mode = init->mode ? init->mode : 0700;
 
-	args->port ?  printk(UM_KERN_INFO "port %d", args->port) :
-		printk(UM_KERN_INFO "undefined port");
+	args->port ?  printk("port %d", args->port) :
+		printk("undefined port");
 }
 
 int vde_user_read(void *conn, void *buf, int len)
-- 
1.5.6.5

Re: [PATCH] Fix vde network backend in user mode linux

[Jiri Kosina]

On Mon, 2 Feb 2009, Luca Bigliardi wrote:

> * Replace kmalloc() with uml_kmalloc()
>   (fix build failure)
> * Remove unnecessary UM_KERN_INFO in printk()
>   (don't display '<6>' while printing info)
> 
> Signed-off-by: Luca Bigliardi <shammash@artha.org>
> ---
>  arch/um/drivers/vde_user.c |    6 +++---
>  1 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/um/drivers/vde_user.c b/arch/um/drivers/vde_user.c
> index 56533db..c5c4325 100644
> --- a/arch/um/drivers/vde_user.c
> +++ b/arch/um/drivers/vde_user.c
> @@ -78,7 +78,7 @@ void vde_init_libstuff(struct vde_data *vpri, struct vde_init *init)
>  {
>  	struct vde_open_args *args;
>  
> -	vpri->args = kmalloc(sizeof(struct vde_open_args), UM_GFP_KERNEL);
> +	vpri->args = uml_kmalloc(sizeof(struct vde_open_args), UM_GFP_KERNEL);
>  	if (vpri->args == NULL) {
>  		printk(UM_KERN_ERR "vde_init_libstuff - vde_open_args "
>  		       "allocation failed");
> @@ -91,8 +91,8 @@ void vde_init_libstuff(struct vde_data *vpri, struct vde_init *init)
>  	args->group = init->group;
>  	args->mode = init->mode ? init->mode : 0700;
>  
> -	args->port ?  printk(UM_KERN_INFO "port %d", args->port) :
> -		printk(UM_KERN_INFO "undefined port");
> +	args->port ?  printk("port %d", args->port) :
> +		printk("undefined port");
>  }
>  
>  int vde_user_read(void *conn, void *buf, int len)

Jeff, I don't see this in linux-next as of today. Are you going to take it 
through your tree?

-- 
Jiri Kosina
SUSE Labs

Re: [PATCH] Fix vde network backend in user mode linux

[Jiri Kosina]

[ doesn't seem to be picked up by anyone into current -next nor -mmotm, 
  but isn't appropriate for trivial; adding akpm to cc ]

On Mon, 2 Feb 2009, Luca Bigliardi wrote:

> * Replace kmalloc() with uml_kmalloc()
>   (fix build failure)
> * Remove unnecessary UM_KERN_INFO in printk()
>   (don't display '<6>' while printing info)
> 
> Signed-off-by: Luca Bigliardi <shammash@artha.org>
> ---
>  arch/um/drivers/vde_user.c |    6 +++---
>  1 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/um/drivers/vde_user.c b/arch/um/drivers/vde_user.c
> index 56533db..c5c4325 100644
> --- a/arch/um/drivers/vde_user.c
> +++ b/arch/um/drivers/vde_user.c
> @@ -78,7 +78,7 @@ void vde_init_libstuff(struct vde_data *vpri, struct vde_init *init)
>  {
>  	struct vde_open_args *args;
>  
> -	vpri->args = kmalloc(sizeof(struct vde_open_args), UM_GFP_KERNEL);
> +	vpri->args = uml_kmalloc(sizeof(struct vde_open_args), UM_GFP_KERNEL);
>  	if (vpri->args == NULL) {
>  		printk(UM_KERN_ERR "vde_init_libstuff - vde_open_args "
>  		       "allocation failed");
> @@ -91,8 +91,8 @@ void vde_init_libstuff(struct vde_data *vpri, struct vde_init *init)
>  	args->group = init->group;
>  	args->mode = init->mode ? init->mode : 0700;
>  
> -	args->port ?  printk(UM_KERN_INFO "port %d", args->port) :
> -		printk(UM_KERN_INFO "undefined port");
> +	args->port ?  printk("port %d", args->port) :
> +		printk("undefined port");
>  }
>  
>  int vde_user_read(void *conn, void *buf, int len)
> -- 
> 1.5.6.5
> 
> 

-- 
Jiri Kosina
SUSE Labs

Re: [PATCH] Fix vde network backend in user mode linux

[Américo Wang]

On Mon, Feb 16, 2009 at 01:38:46PM +0100, Jiri Kosina wrote:
>
>[ doesn't seem to be picked up by anyone into current -next nor -mmotm, 
>  but isn't appropriate for trivial; adding akpm to cc ]
>
>On Mon, 2 Feb 2009, Luca Bigliardi wrote:
>
>> * Replace kmalloc() with uml_kmalloc()
>>   (fix build failure)


Weird... I sent a patch for this several months ago! I thought
Jeff had already taken it... No?


>> * Remove unnecessary UM_KERN_INFO in printk()
>>   (don't display '<6>' while printing info)

But why only this one, not other UM_KERN_INFO in the same file?


-- 
"Against stupidity, the gods themselves, contend in vain."

Re: [PATCH] Fix vde network backend in user mode linux

[Jiri Kosina]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 737 bytes --]

On Mon, 16 Feb 2009, Américo Wang wrote:

> >[ doesn't seem to be picked up by anyone into current -next nor -mmotm, 
> >  but isn't appropriate for trivial; adding akpm to cc ]
> >> * Replace kmalloc() with uml_kmalloc()
> >>   (fix build failure)
> Weird... I sent a patch for this several months ago! I thought Jeff had 
> already taken it... No?

I am not sure whether this has been fixed in another way, but current 
linux-next still has

	void vde_init_libstuff(struct vde_data *vpri, struct vde_init *init)
	{
	        struct vde_open_args *args;
	
	        vpri->args = kmalloc(sizeof(struct vde_open_args), UM_GFP_KERNEL);

so the patch hasn't been at least applied in the form it has been 
submitted.

-- 
Jiri Kosina
SUSE Labs

Re: [PATCH] Fix vde network backend in user mode linux

[Américo Wang]

On Mon, Feb 16, 2009 at 01:51:35PM +0100, Jiri Kosina wrote:
>On Mon, 16 Feb 2009, Américo Wang wrote:
>
>> >[ doesn't seem to be picked up by anyone into current -next nor -mmotm, 
>> >  but isn't appropriate for trivial; adding akpm to cc ]
>> >> * Replace kmalloc() with uml_kmalloc()
>> >>   (fix build failure)
>> Weird... I sent a patch for this several months ago! I thought Jeff had 
>> already taken it... No?
>
>I am not sure whether this has been fixed in another way, but current 
>linux-next still has
>
>	void vde_init_libstuff(struct vde_data *vpri, struct vde_init *init)
>	{
>	        struct vde_open_args *args;
>	
>	        vpri->args = kmalloc(sizeof(struct vde_open_args), UM_GFP_KERNEL);
>
>so the patch hasn't been at least applied in the form it has been 
>submitted.

Hmmm, no, I just checked, but more than half a year ago, we already
did this: http://lkml.org/lkml/2008/7/23/142

It looks like Jeff is too busy to maintain UML patches recently...

Andrew, could you please take UML patches for us? Recently I also
saw some uml patches but no one took them (at least I don't know).

If Andrew can't do this, I would like to take those patches to send
to Linus, if others don't have objections. Andrew?

-- 
"Against stupidity, the gods themselves, contend in vain."

Re: [PATCH] Fix vde network backend in user mode linux

[Luca Bigliardi]

On Mon, Feb 16, 2009 at 08:49 PM, Américo Wang wrote:

> On Mon, Feb 16, 2009 at 01:38:46PM +0100, Jiri Kosina wrote:
> >
> >[ doesn't seem to be picked up by anyone into current -next nor -mmotm, 
> >  but isn't appropriate for trivial; adding akpm to cc ]
> >
> >On Mon, 2 Feb 2009, Luca Bigliardi wrote:
> >
> >> * Replace kmalloc() with uml_kmalloc()
> >>   (fix build failure)
> 
> 
> Weird... I sent a patch for this several months ago! I thought
> Jeff had already taken it... No?

I did as well (in June/July 08 IIRC) through usermodelinux-devel mailing
list but the patch didn't hit, so this time I've tried this way.

> >> * Remove unnecessary UM_KERN_INFO in printk()
> >>   (don't display '<6>' while printing info)
> 
> But why only this one, not other UM_KERN_INFO in the same file?

Other lines are preceded by a 'new_text_line' so vprintk 'eats' the token.


Thank you,

luca

-- 
Beware of programmers who carry screwdrivers.
                        -- Leonard Brandwein

http://shammash.homelinux.org/ - http://www.artha.org/ - http://www.yue.it/

Re: [PATCH] Fix vde network backend in user mode linux

[Américo Wang]

On Mon, Feb 16, 2009 at 02:23:45PM +0100, Luca Bigliardi wrote:
>On Mon, Feb 16, 2009 at 08:49 PM, Américo Wang wrote:
>
>> On Mon, Feb 16, 2009 at 01:38:46PM +0100, Jiri Kosina wrote:
>> >
>> >[ doesn't seem to be picked up by anyone into current -next nor -mmotm, 
>> >  but isn't appropriate for trivial; adding akpm to cc ]
>> >
>> >On Mon, 2 Feb 2009, Luca Bigliardi wrote:
>> >
>> >> * Replace kmalloc() with uml_kmalloc()
>> >>   (fix build failure)
>> 
>> 
>> Weird... I sent a patch for this several months ago! I thought
>> Jeff had already taken it... No?
>
>I did as well (in June/July 08 IIRC) through usermodelinux-devel mailing
>list but the patch didn't hit, so this time I've tried this way.

The problem is _not_ that I don't like you to fix it again, it _is_
that Jeff said he queued a same patch[1] but he didn't push it. :)

1. http://lkml.org/lkml/2008/7/23/142


>
>> >> * Remove unnecessary UM_KERN_INFO in printk()
>> >>   (don't display '<6>' while printing info)
>> 
>> But why only this one, not other UM_KERN_INFO in the same file?
>
>Other lines are preceded by a 'new_text_line' so vprintk 'eats' the token.

Ok then, thank you!

Reviewed-by: WANG Cong <wangcong@zeuux.org>


-- 
"Against stupidity, the gods themselves, contend in vain."