Re: [PATCH] mark install_page static

Re: [PATCH] mark install_page static

[Petr Vandrovec]

On  7 Sep 04 at 16:37, Christoph Hellwig wrote:
> Not used anywhere in modules and it really shouldn't either.

How are modules supposed to implement vma's populate method
without having install_page available?  And yes, there are
out of tree kernel modules which prefer fremap & populate & install_page
over creating several thousands of VMAs to get non-linear mappings.
                                            Thanks,
                                                Petr Vandrovec

> 
> --- 1.28/mm/fremap.c    2004-08-23 10:15:12 +02:00
> +++ edited/mm/fremap.c  2004-09-07 13:51:20 +02:00
> @@ -99,8 +99,6 @@
>     spin_unlock(&mm->page_table_lock);
>     return err;
>  }
> -EXPORT_SYMBOL(install_page);
> -
>  

Re: [PATCH] mark install_page static

[Christoph Hellwig]

On Tue, Sep 07, 2004 at 07:46:33PM +0200, Petr Vandrovec wrote:
> On  7 Sep 04 at 16:37, Christoph Hellwig wrote:
> > Not used anywhere in modules and it really shouldn't either.
> 
> How are modules supposed to implement vma's populate method
> without having install_page available?  And yes, there are
> out of tree kernel modules which prefer fremap & populate & install_page
> over creating several thousands of VMAs to get non-linear mappings.

Submit them fo inclusion and we can consider them.

Re: [PATCH] mark install_page static

[William Lee Irwin III]

On  7 Sep 04 at 16:37, Christoph Hellwig wrote:
>> Not used anywhere in modules and it really shouldn't either.

On Tue, Sep 07, 2004 at 07:46:33PM +0200, Petr Vandrovec wrote:
> How are modules supposed to implement vma's populate method
> without having install_page available?  And yes, there are
> out of tree kernel modules which prefer fremap & populate & install_page
> over creating several thousands of VMAs to get non-linear mappings.

I think this is serious enough to keep the export.


-- wli

[PATCH] mark install_page static

[Christoph Hellwig]

Not used anywhere in modules and it really shouldn't either.


--- 1.28/mm/fremap.c	2004-08-23 10:15:12 +02:00
+++ edited/mm/fremap.c	2004-09-07 13:51:20 +02:00
@@ -99,8 +99,6 @@
 	spin_unlock(&mm->page_table_lock);
 	return err;
 }
-EXPORT_SYMBOL(install_page);
-
 
 /*
  * Install a file pte to a given virtual memory address, release any

Re: [PATCH] mark install_page static

[Alan Cox]

On Maw, 2004-09-07 at 15:37, Christoph Hellwig wrote:
> Not used anywhere in modules and it really shouldn't either.

Doesn't that happen (conveniently from some viewpoints Im sure) to break
vmware ?

Re: [PATCH] mark install_page static

[Christoph Hellwig]

On Tue, Sep 07, 2004 at 06:09:29PM +0100, Alan Cox wrote:
> On Maw, 2004-09-07 at 15:37, Christoph Hellwig wrote:
> > Not used anywhere in modules and it really shouldn't either.
> 
> Doesn't that happen (conveniently from some viewpoints Im sure) to break
> vmware ?

It happens because Arjan & I wrote up some scripts to find dead exports.

Re: [PATCH] mark install_page static

[Alan Cox]

On Maw, 2004-09-07 at 19:12, Christoph Hellwig wrote:
> On Tue, Sep 07, 2004 at 06:09:29PM +0100, Alan Cox wrote:
> > On Maw, 2004-09-07 at 15:37, Christoph Hellwig wrote:
> > > Not used anywhere in modules and it really shouldn't either.
> > 
> > Doesn't that happen (conveniently from some viewpoints Im sure) to break
> > vmware ?
> 
> It happens because Arjan & I wrote up some scripts to find dead exports.

Sure. Most of them look good but some of them look a little dubious. Now
as I was asking - doesn't this break vmware ?

Re: [PATCH] mark install_page static

[Christoph Hellwig]

On Tue, Sep 07, 2004 at 06:25:38PM +0100, Alan Cox wrote:
> > It happens because Arjan & I wrote up some scripts to find dead exports.
> 
> Sure. Most of them look good but some of them look a little dubious. Now
> as I was asking - doesn't this break vmware ?

And the answer is:  I don't know because the vmware module license isn't
free software and thus I haven't taken a look.  Given Petr's answer
probably yes - but because it's above mentioned license I don't
particularly care.

Re: [PATCH] mark install_page static

[Randy.Dunlap]

On Tue, 7 Sep 2004 20:12:59 +0200 Christoph Hellwig wrote:

| On Tue, Sep 07, 2004 at 06:09:29PM +0100, Alan Cox wrote:
| > On Maw, 2004-09-07 at 15:37, Christoph Hellwig wrote:
| > > Not used anywhere in modules and it really shouldn't either.
| > 
| > Doesn't that happen (conveniently from some viewpoints Im sure) to break
| > vmware ?
| 
| It happens because Arjan & I wrote up some scripts to find dead exports.

Can you put those at kernelnewbies.org or janitor.kernelnewbies.org ?

--
~Randy

Re: [PATCH] mark install_page static

[Sam Ravnborg]

On Tue, Sep 07, 2004 at 12:14:43PM -0700, Randy.Dunlap wrote:
> On Tue, 7 Sep 2004 20:12:59 +0200 Christoph Hellwig wrote:
> 
> | On Tue, Sep 07, 2004 at 06:09:29PM +0100, Alan Cox wrote:
> | > On Maw, 2004-09-07 at 15:37, Christoph Hellwig wrote:
> | > > Not used anywhere in modules and it really shouldn't either.
> | > 
> | > Doesn't that happen (conveniently from some viewpoints Im sure) to break
> | > vmware ?
> | 
> | It happens because Arjan & I wrote up some scripts to find dead exports.
> 
> Can you put those at kernelnewbies.org or janitor.kernelnewbies.org ?

Keith Ownes 'namespacecheck' flags unused exported symbols as well.
And furthermore unused non-static definitions.

Present in -mm only atm.

Usage: make && make namespacecheck

Dependent on actual configuration so output should be read carefully.
Should be good on a make allmodconfig or make allyesconfig


	Sam

Re: [PATCH] mark install_page static

[Christoph Hellwig]

On Tue, Sep 07, 2004 at 11:58:49PM +0200, Sam Ravnborg wrote:
> On Tue, Sep 07, 2004 at 12:14:43PM -0700, Randy.Dunlap wrote:
> > On Tue, 7 Sep 2004 20:12:59 +0200 Christoph Hellwig wrote:
> > 
> > | On Tue, Sep 07, 2004 at 06:09:29PM +0100, Alan Cox wrote:
> > | > On Maw, 2004-09-07 at 15:37, Christoph Hellwig wrote:
> > | > > Not used anywhere in modules and it really shouldn't either.
> > | > 
> > | > Doesn't that happen (conveniently from some viewpoints Im sure) to break
> > | > vmware ?
> > | 
> > | It happens because Arjan & I wrote up some scripts to find dead exports.
> > 
> > Can you put those at kernelnewbies.org or janitor.kernelnewbies.org ?
> 
> Keith Ownes 'namespacecheck' flags unused exported symbols as well.
> And furthermore unused non-static definitions.

Well, to get halfway reliable data you need to merge the data from
multiple architectures.  And then do lots of manual grepping for
every instance to be sure.

Re: [PATCH] mark install_page static

[Herbert Poetzl]

On Tue, Sep 07, 2004 at 08:12:59PM +0200, Christoph Hellwig wrote:
> On Tue, Sep 07, 2004 at 06:09:29PM +0100, Alan Cox wrote:
> > On Maw, 2004-09-07 at 15:37, Christoph Hellwig wrote:
> > > Not used anywhere in modules and it really shouldn't either.
> > 
> > Doesn't that happen (conveniently from some viewpoints Im sure) to break
> > vmware ?
> 
> It happens because Arjan & I wrote up some scripts to find dead exports.

are those available somewhere?

sorry google didn't have a good answer for me ...

TIA,
Herbert

> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/