Userland headers available

Userland headers available

[Mariusz Mazur]

At http://ep09.pld-linux.org/~mmazur/glibc-kernel-headers/ there are userland 
headers for linux, derived from 2.6 kernels with lots of 2.4 compatibility 
fixes. CVS repo can be found at cvs.pld-linux.org/glibc-kernel-headers (anon 
and webcvs). These headers are currently used to compile a whole linux distro 
(ftp.pld-linux.org/dists/ac) for x86, sparc, amd64, alpha and ppc, but 
general fixes are applied to all archs since we never know if a new arch 
won't be added (amd64 was added just a month-two ago). #1 feature is that 
they are and will be maintained (currently three people are working on them) 
and bugs are mostly fixed instantly. Enjoy.


-- 
Każdy człowiek, który naprawdę żyje, nie ma charakteru, nie może go mieć.
Charakter jest zawsze martwy, otacza cię zgniła struktura przeniesiona z 
przeszłości. Jeżeli działasz zgodnie z charakterem wtedy nie działasz w ogóle
- jedynie mechanicznie reagujesz.                 { Osho }

Re: Userland headers available

[Daniel Jacobowitz]

On Fri, Jan 23, 2004 at 07:07:17PM +0100, Mariusz Mazur wrote:
> At http://ep09.pld-linux.org/~mmazur/glibc-kernel-headers/ there are userland 
> headers for linux, derived from 2.6 kernels with lots of 2.4 compatibility 
> fixes. CVS repo can be found at cvs.pld-linux.org/glibc-kernel-headers (anon 
> and webcvs). These headers are currently used to compile a whole linux distro 
> (ftp.pld-linux.org/dists/ac) for x86, sparc, amd64, alpha and ppc, but 
> general fixes are applied to all archs since we never know if a new arch 
> won't be added (amd64 was added just a month-two ago). #1 feature is that 
> they are and will be maintained (currently three people are working on them) 
> and bugs are mostly fixed instantly. Enjoy.

I've done precisely the same thing for Debian - if I find the time,
I'll compare...

I would really like to come up with an approach to maintain this
interface definition in the kernel source.  I'm still trying to think
of a way to do it without breaking compatibility or kernel builds.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: Userland headers available

[Mariusz Mazur]

On Friday 23 of January 2004 19:47, Daniel Jacobowitz wrote:
> I've done precisely the same thing for Debian - if I find the time,
> I'll compare...

How much testing did you have?

> I would really like to come up with an approach to maintain this
> interface definition in the kernel source.  I'm still trying to think
> of a way to do it without breaking compatibility or kernel builds.

As I really would like that (less work for me :) I do not think this is 
possible. First thing - 2.4 compatibility in 2.6 kernel would seem weird to 
say at least. Second - I've ripped out kernel code where I could and used 
glibc includes instead - this is (a) The Right Thing (tm) and (b) practically 
undoable inside kernel or would require huge amounts of work, which is really 
better of left outside.

-- 
Każdy człowiek, który naprawdę żyje, nie ma charakteru, nie może go mieć.
Charakter jest zawsze martwy, otacza cię zgniła struktura przeniesiona z 
przeszłości. Jeżeli działasz zgodnie z charakterem wtedy nie działasz w ogóle
- jedynie mechanicznie reagujesz.                 { Osho }

Re: Userland headers available

[Chris Friesen]

Daniel Jacobowitz wrote:

 > I would really like to come up with an approach to maintain this 
interface
 > definition in the kernel source.  I'm still trying to think of a
 > way to do it without breaking compatibility or kernel builds.

The obvious way is to have the kernel headers include the userland
headers, then everything below that be wrapped in "#ifdef __KERNEL__". 
Userland then includes the normal kernel headers, but only gets the 
userland-safe ones.

This sounds too easy though--I'm sure I've missed something, but I can't 
think what....

Chris


-- 
Chris Friesen                    | MailStop: 043/33/F10
Nortel Networks                  | work: (613) 765-0557
3500 Carling Avenue              | fax:  (613) 765-2986
Nepean, ON K2H 8E9 Canada        | email: cfriesen@nortelnetworks.com

Re: Userland headers available

[Chris Friesen]

Friesen, Christopher [CAR:7Q28:EXCH] wrote:

> The obvious way is to have the kernel headers include the userland
> headers, then everything below that be wrapped in "#ifdef __KERNEL__". 
> Userland then includes the normal kernel headers, but only gets the 
> userland-safe ones.

I just realized this wasn't clear.  I envision a new set of headers in 
the kernel that are clean to export to userland.  The current headers 
then include the appropriate userland-clean ones, and everything below 
that is kernel only.

This lets the kernel maintain the userland-clean headers explicitly, and 
we don't have the work of cleaning them up for glibc.

Chris



-- 
Chris Friesen                    | MailStop: 043/33/F10
Nortel Networks                  | work: (613) 765-0557
3500 Carling Avenue              | fax:  (613) 765-2986
Nepean, ON K2H 8E9 Canada        | email: cfriesen@nortelnetworks.com

Re: Userland headers available

[Sam Ravnborg]

On Fri, Jan 23, 2004 at 01:47:55PM -0500, Daniel Jacobowitz wrote:
> I would really like to come up with an approach to maintain this
> interface definition in the kernel source.  I'm still trying to think
> of a way to do it without breaking compatibility or kernel builds.

Yell if you need any kbuild related help.

	Sam

Re: Userland headers available

[jw schultz]

On Fri, Jan 23, 2004 at 02:39:57PM -0500, Chris Friesen wrote:
> Friesen, Christopher [CAR:7Q28:EXCH] wrote:
> 
> >The obvious way is to have the kernel headers include the userland
> >headers, then everything below that be wrapped in "#ifdef __KERNEL__". 
> >Userland then includes the normal kernel headers, but only gets the 
> >userland-safe ones.
> 
> I just realized this wasn't clear.  I envision a new set of headers in 
> the kernel that are clean to export to userland.  The current headers 
> then include the appropriate userland-clean ones, and everything below 
> that is kernel only.
> 
> This lets the kernel maintain the userland-clean headers explicitly, and 
> we don't have the work of cleaning them up for glibc.

This gets discussed every few months.  I think the most
recent was in August.

http://groups.google.com/groups?q=linux-kernel+include/abi&hl=en&lr=&ie=UTF-8&selm=lXHU.431.1%40gated-at.bofh.it&rnum=1

	(google linux-kernel include/abi)

-- 
________________________________________________________________
	J.W. Schultz            Pegasystems Technologies
	email address:		jw@pegasys.ws

		Remember Cernan and Schmitt

Re: Userland headers available

[H. Peter Anvin]

Followup to:  <4011788D.3070606@nortelnetworks.com>
By author:    Chris Friesen <cfriesen@nortelnetworks.com>
In newsgroup: linux.dev.kernel
>
> Friesen, Christopher [CAR:7Q28:EXCH] wrote:
> 
> > The obvious way is to have the kernel headers include the userland
> > headers, then everything below that be wrapped in "#ifdef __KERNEL__". 
> > Userland then includes the normal kernel headers, but only gets the 
> > userland-safe ones.
> 
> I just realized this wasn't clear.  I envision a new set of headers in 
> the kernel that are clean to export to userland.  The current headers 
> then include the appropriate userland-clean ones, and everything below 
> that is kernel only.
> 
> This lets the kernel maintain the userland-clean headers explicitly, and 
> we don't have the work of cleaning them up for glibc.
> 

We've referred to this for quite a while as the "ABI header project";
it's been targetted for 2.7, since it missed the 2.6 freeze.

We have set up a mailing list at:

	http://zytor.com/mailman/listinfo/linuxabi

The goal is to get a formal exportable version of the kernel ABI that
user-space libraries can use.

	-hpa

Re: Userland headers available

[Daniel Jacobowitz]

On Sat, Jan 24, 2004 at 01:38:06AM +0000, H. Peter Anvin wrote:
> Followup to:  <4011788D.3070606@nortelnetworks.com>
> By author:    Chris Friesen <cfriesen@nortelnetworks.com>
> In newsgroup: linux.dev.kernel
> >
> > Friesen, Christopher [CAR:7Q28:EXCH] wrote:
> > 
> > > The obvious way is to have the kernel headers include the userland
> > > headers, then everything below that be wrapped in "#ifdef __KERNEL__". 
> > > Userland then includes the normal kernel headers, but only gets the 
> > > userland-safe ones.
> > 
> > I just realized this wasn't clear.  I envision a new set of headers in 
> > the kernel that are clean to export to userland.  The current headers 
> > then include the appropriate userland-clean ones, and everything below 
> > that is kernel only.
> > 
> > This lets the kernel maintain the userland-clean headers explicitly, and 
> > we don't have the work of cleaning them up for glibc.
> > 
> 
> We've referred to this for quite a while as the "ABI header project";
> it's been targetted for 2.7, since it missed the 2.6 freeze.
> 
> We have set up a mailing list at:
> 
> 	http://zytor.com/mailman/listinfo/linuxabi
> 
> The goal is to get a formal exportable version of the kernel ABI that
> user-space libraries can use.

Are the list archives broken, or has there never been traffic on this
list?

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: Userland headers available

[H. Peter Anvin]

Daniel Jacobowitz wrote:
> 
> Are the list archives broken, or has there never been traffic on this
> list?
> 

There was some traffic on the klibc list, but I don't think things got 
started after the new list was created.

	-hpa