Backporting gerrit tree (from git) to 2.6.18. IT WAS: "Backport" latest (git) DCCP code to 2.6.18

Backporting gerrit tree (from git) to 2.6.18. IT WAS: "Backport" latest (git) DCCP code to 2.6.18

[@ 2007-08-18 15:44 UTC (permalink / raw)]

Arnaldo,
  as you mentioned, I'm in trouble in the kernel skb, please see below:

[sbox-SDK_ARMEL: /usr/src/kernel-source-rx-34] > make zImage
  CHK     include/linux/version.h
make[1]: `include/asm-arm/mach-types.h' is up to date.
  CHK     include/linux/utsrelease.h
  CHK     include/linux/compile.h
  CC      net/dccp/minisocks.o
net/dccp/minisocks.c: In function `dccp_check_req':
net/dccp/minisocks.c:258: warning: passing arg 1 of pointer to
function from incompatible pointer type
net/dccp/minisocks.c:258: error: too many arguments to function
make[2]: *** [net/dccp/minisocks.o] Error 1
make[1]: *** [net/dccp] Error 2
make: *** [net] Error 2

-------

the code that it is generation this error is minisocks.c file:

drop:
        if (dccp_hdr(skb)->dccph_type != DCCP_PKT_RESET)
                req->rsk_ops->send_reset(sk, skb);

The problem is related to skb and I have no idea to do from here. I
had many other problems before this, particular with the linux
workqueue implementation (include/linux/workqueue.c/h) and with
kmen_create_cache (function ccid_kmem_cache_create) but I solved them.
So, what do you suggest me to do?

Thank you,
Leandro.

2007/8/6, Arnaldo Carvalho de Melo <acme@ghostprotocols.net>:
> Em Mon, Aug 06, 2007 at 09:55:49AM -0300, ?$B%D Leandro Sales escreveu:
> > Hi folks,
> >    due to wireless card restrictions (the wifi card driver is
> > proprietary) I can't use the latest linux-omap git kernel with nokia
> > n800. In this way, the unique kernel version available for this
> > plataform is 2.6.18. Then I'd like to know if will I have problem if I
> > get the latest DCCP codes (from the git kernel) and copy them to
> > kernel 2.6.18.
> >    Just in case of you don't know if I will get any problem, I will
> > try to do this and see what happen.
>
> Its technically possible. That is what companies such as Red Hat does
> for a living. The kernel in RHEL5, the current enterprise offering from
> Red Hat is based on 2.6.18, for instance :-)
>
> The infrastructure is also not changing that much, but pay attention to
> ktime, perhaps the skb offset patches, i.e. I think it will not be just
> copying the dccp files from the top of the tree into 2.6.18 and
> rebuilding :)
>
> - Arnaldo
>


-- 
Leandro Melo de Sales.
Pervasive and Embedded Computing Laboratory
BRisa and E-Phone Projects Manager & Network Administrator
http://embedded.ufcg.edu.br/
+55 083 3310-1404 (extension 208)

"There are 10 types of people in this world: those who understand
binary, those who don't"

Re: Backporting gerrit tree (from git) to 2.6.18. IT WAS: "Backport" latest (git) DCCP code to 2.6.1

[@ 2007-08-18 16:04 UTC (permalink / raw)]

Arnaldo,

   I just change to req->rsk_ops->send_reset(skb);, instead of
req->rsk_ops->send_reset(sk, skb);

Leandro

2007/8/18, ツ Leandro Sales <leandroal@gmail.com>:
> Arnaldo,
>   as you mentioned, I'm in trouble in the kernel skb, please see below:
>
> [sbox-SDK_ARMEL: /usr/src/kernel-source-rx-34] > make zImage
>   CHK     include/linux/version.h
> make[1]: `include/asm-arm/mach-types.h' is up to date.
>   CHK     include/linux/utsrelease.h
>   CHK     include/linux/compile.h
>   CC      net/dccp/minisocks.o
> net/dccp/minisocks.c: In function `dccp_check_req':
> net/dccp/minisocks.c:258: warning: passing arg 1 of pointer to
> function from incompatible pointer type
> net/dccp/minisocks.c:258: error: too many arguments to function
> make[2]: *** [net/dccp/minisocks.o] Error 1
> make[1]: *** [net/dccp] Error 2
> make: *** [net] Error 2
>
> -------
>
> the code that it is generation this error is minisocks.c file:
>
> drop:
>         if (dccp_hdr(skb)->dccph_type != DCCP_PKT_RESET)
>                 req->rsk_ops->send_reset(sk, skb);
>
> The problem is related to skb and I have no idea to do from here. I
> had many other problems before this, particular with the linux
> workqueue implementation (include/linux/workqueue.c/h) and with
> kmen_create_cache (function ccid_kmem_cache_create) but I solved them.
> So, what do you suggest me to do?
>
> Thank you,
> Leandro.
>
> 2007/8/6, Arnaldo Carvalho de Melo <acme@ghostprotocols.net>:
> > Em Mon, Aug 06, 2007 at 09:55:49AM -0300, ?$B%D Leandro Sales escreveu:
> > > Hi folks,
> > >    due to wireless card restrictions (the wifi card driver is
> > > proprietary) I can't use the latest linux-omap git kernel with nokia
> > > n800. In this way, the unique kernel version available for this
> > > plataform is 2.6.18. Then I'd like to know if will I have problem if I
> > > get the latest DCCP codes (from the git kernel) and copy them to
> > > kernel 2.6.18.
> > >    Just in case of you don't know if I will get any problem, I will
> > > try to do this and see what happen.
> >
> > Its technically possible. That is what companies such as Red Hat does
> > for a living. The kernel in RHEL5, the current enterprise offering from
> > Red Hat is based on 2.6.18, for instance :-)
> >
> > The infrastructure is also not changing that much, but pay attention to
> > ktime, perhaps the skb offset patches, i.e. I think it will not be just
> > copying the dccp files from the top of the tree into 2.6.18 and
> > rebuilding :)
> >
> > - Arnaldo
> >
>
>
> --
> Leandro Melo de Sales.
> Pervasive and Embedded Computing Laboratory
> BRisa and E-Phone Projects Manager & Network Administrator
> http://embedded.ufcg.edu.br/
> +55 083 3310-1404 (extension 208)
>
> "There are 10 types of people in this world: those who understand
> binary, those who don't"
>


-- 
Leandro Melo de Sales.
Pervasive and Embedded Computing Laboratory
BRisa and E-Phone Projects Manager & Network Administrator
http://embedded.ufcg.edu.br/
+55 083 3310-1404 (extension 208)

"There are 10 types of people in this world: those who understand
binary, those who don't"

Re: Backporting gerrit tree (from git) to 2.6.18. IT WAS:

[Arnaldo Carvalho de Melo]

Em Sat, Aug 18, 2007 at 01:04:27PM -0300, ツ Leandro Sales escreveu:
> Arnaldo,
> 
>    I just change to req->rsk_ops->send_reset(skb);, instead of
> req->rsk_ops->send_reset(sk, skb);

OK, great, I forgot about this change in particular, the skb offset
patches I mentioned where another, unrelated change that happened in
this interim.

- Arnaldo
 
> Leandro
> 
> 2007/8/18, ツ Leandro Sales <leandroal@gmail.com>:
> > Arnaldo,
> >   as you mentioned, I'm in trouble in the kernel skb, please see below:
> >
> > [sbox-SDK_ARMEL: /usr/src/kernel-source-rx-34] > make zImage
> >   CHK     include/linux/version.h
> > make[1]: `include/asm-arm/mach-types.h' is up to date.
> >   CHK     include/linux/utsrelease.h
> >   CHK     include/linux/compile.h
> >   CC      net/dccp/minisocks.o
> > net/dccp/minisocks.c: In function `dccp_check_req':
> > net/dccp/minisocks.c:258: warning: passing arg 1 of pointer to
> > function from incompatible pointer type
> > net/dccp/minisocks.c:258: error: too many arguments to function
> > make[2]: *** [net/dccp/minisocks.o] Error 1
> > make[1]: *** [net/dccp] Error 2
> > make: *** [net] Error 2
> >
> > -------
> >
> > the code that it is generation this error is minisocks.c file:
> >
> > drop:
> >         if (dccp_hdr(skb)->dccph_type != DCCP_PKT_RESET)
> >                 req->rsk_ops->send_reset(sk, skb);
> >
> > The problem is related to skb and I have no idea to do from here. I
> > had many other problems before this, particular with the linux
> > workqueue implementation (include/linux/workqueue.c/h) and with
> > kmen_create_cache (function ccid_kmem_cache_create) but I solved them.
> > So, what do you suggest me to do?
> >
> > Thank you,
> > Leandro.
> >
> > 2007/8/6, Arnaldo Carvalho de Melo <acme@ghostprotocols.net>:
> > > Em Mon, Aug 06, 2007 at 09:55:49AM -0300, ?$B%D Leandro Sales escreveu:
> > > > Hi folks,
> > > >    due to wireless card restrictions (the wifi card driver is
> > > > proprietary) I can't use the latest linux-omap git kernel with nokia
> > > > n800. In this way, the unique kernel version available for this
> > > > plataform is 2.6.18. Then I'd like to know if will I have problem if I
> > > > get the latest DCCP codes (from the git kernel) and copy them to
> > > > kernel 2.6.18.
> > > >    Just in case of you don't know if I will get any problem, I will
> > > > try to do this and see what happen.
> > >
> > > Its technically possible. That is what companies such as Red Hat does
> > > for a living. The kernel in RHEL5, the current enterprise offering from
> > > Red Hat is based on 2.6.18, for instance :-)
> > >
> > > The infrastructure is also not changing that much, but pay attention to
> > > ktime, perhaps the skb offset patches, i.e. I think it will not be just
> > > copying the dccp files from the top of the tree into 2.6.18 and
> > > rebuilding :)

Re: Backporting gerrit tree (from git) to 2.6.18. IT WAS: "Backport" latest (git) DCCP code to 2.6.1

[@ 2007-08-19 12:38 UTC (permalink / raw)]

Are these operations equivalent? Please, tell me where can I find docs about skb

-   skb->h.raw = skb->data;

+   skb_reset_transport_header(skb);

[]s
Leandro.

2007/8/19, ツ Leandro Sales <leandroal@gmail.com>:
> Hi Arnaldo,
>       I'm stuck in the use of skb_reset_transport_header(skb); What is
> the old way make the same think of reset transport header? Do you know
> where can I find the old reference API of skb?
>
> []s
> Leandro.
>
> 2007/8/18, ツ Leandro Sales <leandroal@gmail.com>:
> > Arnaldo,
> >
> >    I just change to req->rsk_ops->send_reset(skb);, instead of
> > req->rsk_ops->send_reset(sk, skb);
> >
> > Leandro
> >
> > 2007/8/18, ツ Leandro Sales <leandroal@gmail.com>:
> > > Arnaldo,
> > >   as you mentioned, I'm in trouble in the kernel skb, please see below:
> > >
> > > [sbox-SDK_ARMEL: /usr/src/kernel-source-rx-34] > make zImage
> > >   CHK     include/linux/version.h
> > > make[1]: `include/asm-arm/mach-types.h' is up to date.
> > >   CHK     include/linux/utsrelease.h
> > >   CHK     include/linux/compile.h
> > >   CC      net/dccp/minisocks.o
> > > net/dccp/minisocks.c: In function `dccp_check_req':
> > > net/dccp/minisocks.c:258: warning: passing arg 1 of pointer to
> > > function from incompatible pointer type
> > > net/dccp/minisocks.c:258: error: too many arguments to function
> > > make[2]: *** [net/dccp/minisocks.o] Error 1
> > > make[1]: *** [net/dccp] Error 2
> > > make: *** [net] Error 2
> > >
> > > -------
> > >
> > > the code that it is generation this error is minisocks.c file:
> > >
> > > drop:
> > >         if (dccp_hdr(skb)->dccph_type != DCCP_PKT_RESET)
> > >                 req->rsk_ops->send_reset(sk, skb);
> > >
> > > The problem is related to skb and I have no idea to do from here. I
> > > had many other problems before this, particular with the linux
> > > workqueue implementation (include/linux/workqueue.c/h) and with
> > > kmen_create_cache (function ccid_kmem_cache_create) but I solved them.
> > > So, what do you suggest me to do?
> > >
> > > Thank you,
> > > Leandro.
> > >
> > > 2007/8/6, Arnaldo Carvalho de Melo <acme@ghostprotocols.net>:
> > > > Em Mon, Aug 06, 2007 at 09:55:49AM -0300, ?$B%D Leandro Sales escreveu:
> > > > > Hi folks,
> > > > >    due to wireless card restrictions (the wifi card driver is
> > > > > proprietary) I can't use the latest linux-omap git kernel with nokia
> > > > > n800. In this way, the unique kernel version available for this
> > > > > plataform is 2.6.18. Then I'd like to know if will I have problem if I
> > > > > get the latest DCCP codes (from the git kernel) and copy them to
> > > > > kernel 2.6.18.
> > > > >    Just in case of you don't know if I will get any problem, I will
> > > > > try to do this and see what happen.
> > > >
> > > > Its technically possible. That is what companies such as Red Hat does
> > > > for a living. The kernel in RHEL5, the current enterprise offering from
> > > > Red Hat is based on 2.6.18, for instance :-)
> > > >
> > > > The infrastructure is also not changing that much, but pay attention to
> > > > ktime, perhaps the skb offset patches, i.e. I think it will not be just
> > > > copying the dccp files from the top of the tree into 2.6.18 and
> > > > rebuilding :)
> > > >
> > > > - Arnaldo
> > > >
> > >
> > >
> > > --
> > > Leandro Melo de Sales.
> > > Pervasive and Embedded Computing Laboratory
> > > BRisa and E-Phone Projects Manager & Network Administrator
> > > http://embedded.ufcg.edu.br/
> > > +55 083 3310-1404 (extension 208)
> > >
> > > "There are 10 types of people in this world: those who understand
> > > binary, those who don't"
> > >
> >
> >
> > --
> > Leandro Melo de Sales.
> > Pervasive and Embedded Computing Laboratory
> > BRisa and E-Phone Projects Manager & Network Administrator
> > http://embedded.ufcg.edu.br/
> > +55 083 3310-1404 (extension 208)
> >
> > "There are 10 types of people in this world: those who understand
> > binary, those who don't"
> >
>
>
> --
> Leandro Melo de Sales.
> Pervasive and Embedded Computing Laboratory
> BRisa and E-Phone Projects Manager & Network Administrator
> http://embedded.ufcg.edu.br/
> +55 083 3310-1404 (extension 208)
>
> "There are 10 types of people in this world: those who understand
> binary, those who don't"
>

Re: Backporting gerrit tree (from git) to 2.6.18. IT WAS: "Backport" latest (git) DCCP code to 2.6.1

[Ian McDonald]

On 8/20/07, ツ Leandro Sales <leandroal@gmail.com> wrote:
> Are these operations equivalent? Please, tell me where can I find docs about skb
>
Dave Miller wrote some docs a while ago at:
http://vger.kernel.org/~davem/skb.html

-- 
Web1: http://wand.net.nz/~iam4/
Web2: http://www.jandi.co.nz
Blog: http://iansblog.jandi.co.nz

Re: Backporting gerrit tree (from git) to 2.6.18. IT WAS:

[Arnaldo Carvalho de Melo]

Em Sun, Aug 19, 2007 at 09:38:13AM -0300, ツ Leandro Sales escreveu:
> Are these operations equivalent? Please, tell me where can I find docs about skb
> 
> -   skb->h.raw = skb->data;
> 
> +   skb_reset_transport_header(skb);

That is easy as I left all the old code in include/linux/skbuff.h. Just
look at what the operation does if NET_SKBUFF_DATA_USES_OFFSET is not
defined, that, for the above api is:

static inline void skb_reset_transport_header(struct sk_buff *skb)
{
        skb->transport_header = skb->data;
}

skb->h.raw is the old name of skb->transport_header
skb->nh.raw is the old name of skb->network_header
skb->nac.raw is the old name of skb->mac_header
 
- Arnaldo