V4L hg tree fails to compile against latest stable kernel 2.6.35

V4L hg tree fails to compile against latest stable kernel 2.6.35

[VDR User]

Any idea when this will be fixed?

Re: V4L hg tree fails to compile against latest stable kernel 2.6.35

[VDR User]

On Mon, Aug 2, 2010 at 11:36 PM, VDR User <user.vdr@gmail.com> wrote:
> Any idea when this will be fixed?

Sorry, forgot to paste in the details:

make -C /lib/modules/2.6.35.amd.080210.2/build
SUBDIRS=/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l  modules
make[2]: Entering directory `/usr/src/linux-2.6.35'
  CC [M]  /usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvbdev.o
  CC [M]  /usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dmxdev.o
  CC [M]  /usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_demux.o
  CC [M]  /usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_filter.o
  CC [M]  /usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_ca_en50221.o
  CC [M]  /usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_frontend.o
  CC [M]  /usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_net.o
/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_net.c:1190:
warning: 'struct dev_mc_list' declared inside parameter list
/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_net.c:1190:
warning: its scope is only this definition or declaration, which is
probably not what you want
/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_net.c: In
function 'dvb_set_mc_filter':
/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_net.c:1197:
error: dereferencing pointer to incomplete type
/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_net.c: In
function 'wq_set_multicast_list':
/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_net.c:1247:
error: 'struct net_device' has no member named 'mc_list'
/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_net.c:1249:
error: dereferencing pointer to incomplete type
/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_net.c:1249:
warning: left-hand operand of comma expression has no effect
/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_net.c:1249:
warning: value computed is not used
/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_net.c:1250:
warning: passing argument 2 of 'dvb_set_mc_filter' from incompatible
pointer type
/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_net.c:1190:
note: expected 'struct dev_mc_list *' but argument is of type 'struct
dev_mc_list *'
make[3]: *** [/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l/dvb_net.o]
Error 1
make[2]: *** [_module_/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l]
Error 2
make[2]: Leaving directory `/usr/src/linux-2.6.35'
make[1]: *** [default] Error 2
make[1]: Leaving directory `/usr/local/dvb/v4l.source/v4l.20100802/v4l-dvb/v4l'
make: *** [all] Error 2

Re: V4L hg tree fails to compile against latest stable kernel 2.6.35

[Douglas Schilling Landgraf]

Hello Derek,

On Tue, Aug 3, 2010 at 10:37 AM, Mauro Carvalho Chehab
<mchehab@redhat.com> wrote:
> Em 03-08-2010 03:44, VDR User escreveu:
>> On Mon, Aug 2, 2010 at 11:36 PM, VDR User <user.vdr@gmail.com> wrote:
>>> Any idea when this will be fixed?
>
> It is up to Douglas to do the backport, but you can just use the latest
> git tree, where those patches are applied already at 2.6.35, at the
> branch staging/v2.6.36.

I am already working to give a full update to hg tree. Sorry this problem.

Cheers
Douglas

Re: V4L hg tree fails to compile against latest stable kernel 2.6.35

[VDR User]

On Wed, Aug 4, 2010 at 10:19 PM, Douglas Schilling Landgraf
<dougsland@gmail.com> wrote:
> Hello Derek,
>
> On Tue, Aug 3, 2010 at 10:37 AM, Mauro Carvalho Chehab
> <mchehab@redhat.com> wrote:
>> Em 03-08-2010 03:44, VDR User escreveu:
>>> On Mon, Aug 2, 2010 at 11:36 PM, VDR User <user.vdr@gmail.com> wrote:
>>>> Any idea when this will be fixed?
>>
>> It is up to Douglas to do the backport, but you can just use the latest
>> git tree, where those patches are applied already at 2.6.35, at the
>> branch staging/v2.6.36.
>
> I am already working to give a full update to hg tree. Sorry this problem.

Hi Douglas, thanks for the update!  I'll pass the word along to other
users I know who are wondering as well but who don't use the mailing
list.

A little off topic but IIRC wasn't there some talk about writing new
scripts to main keep hg more up-to-date?  If so, was that ever
started?  Otherwise, my apologies if I'm not remembering correctly.

Re: V4L hg tree fails to compile against latest stable kernel 2.6.35

[Douglas Schilling Landgraf]

Hello,

On Thu, Aug 5, 2010 at 2:44 AM, VDR User <user.vdr@gmail.com> wrote:
> On Wed, Aug 4, 2010 at 10:19 PM, Douglas Schilling Landgraf
> <dougsland@gmail.com> wrote:
>> Hello Derek,
>>
>> On Tue, Aug 3, 2010 at 10:37 AM, Mauro Carvalho Chehab
>> <mchehab@redhat.com> wrote:
>>> Em 03-08-2010 03:44, VDR User escreveu:
>>>> On Mon, Aug 2, 2010 at 11:36 PM, VDR User <user.vdr@gmail.com> wrote:
>>>>> Any idea when this will be fixed?
>>>
>>> It is up to Douglas to do the backport, but you can just use the latest
>>> git tree, where those patches are applied already at 2.6.35, at the
>>> branch staging/v2.6.36.
>>
>> I am already working to give a full update to hg tree. Sorry this problem.
>
> Hi Douglas, thanks for the update!  I'll pass the word along to other
> users I know who are wondering as well but who don't use the mailing
> list.
>
> A little off topic but IIRC wasn't there some talk about writing new
> scripts to main keep hg more up-to-date?  If so, was that ever
> started?  Otherwise, my apologies if I'm not remembering correctly.

We have a script to check if hg is synced to git tree, to extract
sequence of patches
from git tree, etc.. but not like this yet. Some time ago I and Mauro
discussed about create a script
to apply patches automatically (of course, those witch doesn't break
and need a backport manually) but I think it will
became true with this new git tree schema that Mauro implemented a few
days ago which reduces the number of branches
to follow.

Cheers
Douglas

Re: V4L hg tree fails to compile against latest stable kernel 2.6.35

[VDR User]

On Wed, Aug 4, 2010 at 10:19 PM, Douglas Schilling Landgraf
<dougsland@gmail.com> wrote:
> I am already working to give a full update to hg tree. Sorry this problem.

Hi Douglas.  Any estimate when this will be fixed?  Was hoping it was
already since new stable kernel 2.6.35.2 is out now but still the same
problem when I tried just now.

Best regards,
Derek

Re: V4L hg tree fails to compile against latest stable kernel 2.6.35

[Douglas Schilling Landgraf]

Hello Derek,

On Sat, Aug 14, 2010 at 12:46 PM, VDR User <user.vdr@gmail.com> wrote:
> On Wed, Aug 4, 2010 at 10:19 PM, Douglas Schilling Landgraf
> <dougsland@gmail.com> wrote:
>> I am already working to give a full update to hg tree. Sorry this problem.
>
> Hi Douglas.  Any estimate when this will be fixed?  Was hoping it was
> already since new stable kernel 2.6.35.2 is out now but still the same
> problem when I tried just now.

I am already working on it this weekend. I will reply this thread when finished.

Cheers
Douglas

Re: V4L hg tree fails to compile against latest stable kernel 2.6.35

[Douglas Schilling Landgraf]

Hello Derek,

On Sun, Aug 15, 2010 at 2:22 AM, Douglas Schilling Landgraf
<dougsland@gmail.com> wrote:
> Hello Derek,
>
> On Sat, Aug 14, 2010 at 12:46 PM, VDR User <user.vdr@gmail.com> wrote:
>> On Wed, Aug 4, 2010 at 10:19 PM, Douglas Schilling Landgraf
>> <dougsland@gmail.com> wrote:
>>> I am already working to give a full update to hg tree. Sorry this problem.
>>
>> Hi Douglas.  Any estimate when this will be fixed?  Was hoping it was
>> already since new stable kernel 2.6.35.2 is out now but still the same
>> problem when I tried just now.
>
> I am already working on it this weekend. I will reply this thread when finished.

2.6.35 should be working, let me know if not. Now, I need to backport
the changes to old kernels
and commit other patches in my pending list.

Cheers
Douglas

Re: V4L hg tree fails to compile against latest stable kernel 2.6.35

[VDR User]

On Sun, Aug 15, 2010 at 10:57 AM, Douglas Schilling Landgraf
<dougsland@gmail.com> wrote:
> Hello Derek,
>
>
> 2.6.35 should be working, let me know if not. Now, I need to backport
> the changes to old kernels
> and commit other patches in my pending list.

It does appear to be working so far.  Thanks for the update!  In case
a problem pops up, I'll msg again.

Best regards,
Derek

V4L hg tree fails to compile against kernel 2.6.28

[Jan Hoogenraad]

Douglas:

On compiling with  Linux  2.6.28-19-generic #62-Ubuntu

I now get:

dvb_demux.c: In function 'dvbdmx_write':
dvb_demux.c:1137: error: implicit declaration of function 'memdup_user'
dvb_demux.c:1137: warning: assignment makes pointer from integer without 
a cast

This is probably due to changeset 2ceef3d75547

which introduced the use of this function:
http://linuxtv.org/hg/v4l-dvb/diff/2ceef3d75547/linux/drivers/media/dvb/dvb-core/dvb_demux.c

This function is not available in linux/string.h in kernel 2.6.29 and lower.

http://lxr.free-electrons.com/source/include/linux/string.h?v=2.6.28

Could you please advise me on what to do ?


Douglas Schilling Landgraf wrote:
> Hello Derek,
> 
> On Sun, Aug 15, 2010 at 2:22 AM, Douglas Schilling Landgraf
> <dougsland@gmail.com> wrote:
>> Hello Derek,
>>
>> On Sat, Aug 14, 2010 at 12:46 PM, VDR User <user.vdr@gmail.com> wrote:
>>> On Wed, Aug 4, 2010 at 10:19 PM, Douglas Schilling Landgraf
>>> <dougsland@gmail.com> wrote:
>>>> I am already working to give a full update to hg tree. Sorry this problem.
>>> Hi Douglas.  Any estimate when this will be fixed?  Was hoping it was
>>> already since new stable kernel 2.6.35.2 is out now but still the same
>>> problem when I tried just now.
>> I am already working on it this weekend. I will reply this thread when finished.
> 
> 2.6.35 should be working, let me know if not. Now, I need to backport
> the changes to old kernels
> and commit other patches in my pending list.
> 
> Cheers
> Douglas
> --
> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 


-- 
Jan Hoogenraad
Hoogenraad Interface Services
Postbus 2717
3500 GS Utrecht

Re: V4L hg tree fails to compile against kernel 2.6.28

[Douglas Schilling Landgraf]

Hello Jan,

On Tue, Aug 24, 2010 at 5:04 PM, Jan Hoogenraad
<jan-conceptronic@hoogenraad.net> wrote:
> Douglas:
>
> On compiling with  Linux  2.6.28-19-generic #62-Ubuntu
>
> I now get:
>
> dvb_demux.c: In function 'dvbdmx_write':
> dvb_demux.c:1137: error: implicit declaration of function 'memdup_user'
> dvb_demux.c:1137: warning: assignment makes pointer from integer without a
> cast
>
> This is probably due to changeset 2ceef3d75547
>
> which introduced the use of this function:
> http://linuxtv.org/hg/v4l-dvb/diff/2ceef3d75547/linux/drivers/media/dvb/dvb-core/dvb_demux.c
>
> This function is not available in linux/string.h in kernel 2.6.29 and lower.
>
> http://lxr.free-electrons.com/source/include/linux/string.h?v=2.6.28
>
> Could you please advise me on what to do ?

For this issue, I have fixed right now, please try again with a
updated repository. But since I have commited several patches
yesterday,
we have another one that probably will reach your compilantion (the
kfifo one). I will work to fix this one too.

Thanks for the report.

Cheers
Douglas

Re: V4L hg tree fails to compile against kernel 2.6.28

[Douglas Schilling Landgraf]

Hi,

On Tue, Aug 24, 2010 at 11:45 PM, Douglas Schilling Landgraf
<dougsland@gmail.com> wrote:
> Hello Jan,
>
> On Tue, Aug 24, 2010 at 5:04 PM, Jan Hoogenraad
> <jan-conceptronic@hoogenraad.net> wrote:
>> Douglas:
>>
>> On compiling with  Linux  2.6.28-19-generic #62-Ubuntu
>>
>> I now get:
>>
>> dvb_demux.c: In function 'dvbdmx_write':
>> dvb_demux.c:1137: error: implicit declaration of function 'memdup_user'
>> dvb_demux.c:1137: warning: assignment makes pointer from integer without a
>> cast
>>
>> This is probably due to changeset 2ceef3d75547
>>
>> which introduced the use of this function:
>> http://linuxtv.org/hg/v4l-dvb/diff/2ceef3d75547/linux/drivers/media/dvb/dvb-core/dvb_demux.c
>>
>> This function is not available in linux/string.h in kernel 2.6.29 and lower.
>>
>> http://lxr.free-electrons.com/source/include/linux/string.h?v=2.6.28
>>
>> Could you please advise me on what to do ?
>
> For this issue, I have fixed right now, please try again with a
> updated repository. But since I have commited several patches
> yesterday,
> we have another one that probably will reach your compilantion (the
> kfifo one). I will work to fix this one too.
>
> Thanks for the report.


Ok, should be working right now, I have made the backports
needed/adjusts to compile agains 2.6.29. Let me know if you have any
problem.
Tomorrow I will continue commiting patches to get hg synced with git.

Thanks
Douglas

Re: V4L hg tree fails to compile against kernel 2.6.28

[Jan Hoogenraad]

Thanks for your help. I pulled the code

Actually, now the function definition in compat.h causes a compilation 
error: see first text below.

I fixed that by inserting
#include <linux/err.h>
at line 38 in compat.h in my local branch

After that, compilation succeeds.

Now, the device will not install, as dvb_usb cannot install anymore.
see second text below.

Therefore, I have replaced the __kmalloc_track_caller with kmalloc (as 
it was in dvb_demux.c before the change 2ceef3d75547 at line 49

can you make both changes in the hg branch ?
-----

make[2]: Entering directory `/usr/src/linux-headers-2.6.28-19-generic'
   CC [M]  tuner-xc2028.o
In file included from tuner-xc2028.c:19:
compat.h: In function 'memdup_user':
compat.h:50: error: implicit declaration of function 'ERR_PTR'
compat.h:50: warning: return makes pointer from integer without a cast
compat.h:54: warning: return makes pointer from integer without a cast
In file included from include/linux/fs.h:1891,
                  from include/linux/poll.h:11,
                  from dvbdev.h:27,
                  from dvb_frontend.h:43,
                  from tuner-xc2028.h:10,
                  from tuner-xc2028.c:22:
include/linux/err.h: At top level:
include/linux/err.h:22: error: conflicting types for 'ERR_PTR'
----
sudo modprobe dvb-usb-rtl2831u ir_protocol=2
WARNING: Error inserting dvb_usb 
(/lib/modules/2.6.28-19-generic/kernel/drivers/media/dvb/dvb-usb/dvb-usb.ko): 
Unknown symbol in module, or unknown parameter (see dmesg)
WARNING: Error inserting dvb_usb_dibusb_common 
(/lib/modules/2.6.28-19-generic/kernel/drivers/media/dvb/dvb-usb/dvb-usb-dibusb-common.ko): 
Unknown symbol in module, or unknown parameter (see dmesg)
FATAL: Error inserting dvb_usb_rtl2831u 
(/lib/modules/2.6.28-19-generic/kernel/drivers/media/dvb/rtl2831/dvb-usb-rtl2831u.ko): 
Unknown symbol in module, or unknown parameter (see dmesg)

in the system log, I find:

dvb_core: Unknown symbol __kmalloc_track_caller
dvb_core: Unknown symbol __kmalloc_track_caller




Douglas Schilling Landgraf wrote:
> Hello Jan,
> 
> On Tue, Aug 24, 2010 at 5:04 PM, Jan Hoogenraad
> <jan-conceptronic@hoogenraad.net> wrote:
>> Douglas:
>>
>> On compiling with  Linux  2.6.28-19-generic #62-Ubuntu
>>
>> I now get:
>>
>> dvb_demux.c: In function 'dvbdmx_write':
>> dvb_demux.c:1137: error: implicit declaration of function 'memdup_user'
>> dvb_demux.c:1137: warning: assignment makes pointer from integer without a
>> cast
>>
>> This is probably due to changeset 2ceef3d75547
>>
>> which introduced the use of this function:
>> http://linuxtv.org/hg/v4l-dvb/diff/2ceef3d75547/linux/drivers/media/dvb/dvb-core/dvb_demux.c
>>
>> This function is not available in linux/string.h in kernel 2.6.29 and lower.
>>
>> http://lxr.free-electrons.com/source/include/linux/string.h?v=2.6.28
>>
>> Could you please advise me on what to do ?
> 
> For this issue, I have fixed right now, please try again with a
> updated repository. But since I have commited several patches
> yesterday,
> we have another one that probably will reach your compilantion (the
> kfifo one). I will work to fix this one too.
> 
> Thanks for the report.
> 
> Cheers
> Douglas
> 


-- 
Jan Hoogenraad
Hoogenraad Interface Services
Postbus 2717
3500 GS Utrecht

Re: V4L hg tree fails to compile against kernel 2.6.28

[Douglas Schilling Landgraf]

Hello Jan,

On Wed, Aug 25, 2010 at 3:26 AM, Jan Hoogenraad
<jan-conceptronic@hoogenraad.net> wrote:
> Thanks for your help. I pulled the code
>
> Actually, now the function definition in compat.h causes a compilation
> error: see first text below.
>
> I fixed that by inserting
> #include <linux/err.h>
> at line 38 in compat.h in my local branch
>
> After that, compilation succeeds.


Thanks, applied.

> Now, the device will not install, as dvb_usb cannot install anymore.
> see second text below.
>
> Therefore, I have replaced the __kmalloc_track_caller with kmalloc (as it
> was in dvb_demux.c before the change 2ceef3d75547 at line 49
>
> can you make both changes in the hg branch ?

This one won't be required because the idea is support until kernel 2.6.26.
Please check my next email about hg x announcement.

Thanks for checking it, your help is appreciate.

Cheers
Douglas

Re: V4L hg tree fails to compile against kernel 2.6.28

[Jan Hoogenraad]

__kmalloc_track_caller is not avaiable in
2.6.28-19-generic #62-Ubuntu
which just passes your cut at 2.6.26.

Apparently, the code __kmalloc_track_caller is in the source archive for 
2.6.28, but the kernel options were not such that the symbol is available.
My guess is that it was changed in 2.6.30, where memory management was 
changed.

Thus, 2.6.28, it needs to stay kmalloc.

Douglas Schilling Landgraf wrote:
> Hello Jan,
> 
> On Wed, Aug 25, 2010 at 3:26 AM, Jan Hoogenraad
> <jan-conceptronic@hoogenraad.net> wrote:
>> Thanks for your help. I pulled the code
>>
>> Actually, now the function definition in compat.h causes a compilation
>> error: see first text below.
>>
>> I fixed that by inserting
>> #include <linux/err.h>
>> at line 38 in compat.h in my local branch
>>
>> After that, compilation succeeds.
> 
> 
> Thanks, applied.
> 
>> Now, the device will not install, as dvb_usb cannot install anymore.
>> see second text below.
>>
>> Therefore, I have replaced the __kmalloc_track_caller with kmalloc (as it
>> was in dvb_demux.c before the change 2ceef3d75547 at line 49
>>
>> can you make both changes in the hg branch ?
> 
> This one won't be required because the idea is support until kernel 2.6.26.
> Please check my next email about hg x announcement.
> 
> Thanks for checking it, your help is appreciate.
> 
> Cheers
> Douglas
> 


-- 
Jan Hoogenraad
Hoogenraad Interface Services
Postbus 2717
3500 GS Utrecht

ir_codes_dibusb_table : please make table length a constant

[Jan Hoogenraad]

During debugging a driver, I found that at least in

linux/drivers/media/dvb/dvb-usb/dibusb-mc.c	
the length of ir_codes_dibusb_table was referring to an older version.
This may have caused memory overrun problems

I have fixed this in my branch in
linux/drivers/media/dvb/dvb-usb/dibusb.h
  with a
#define IR_CODES_DIBUSB_TABLE_LEN 111

http://linuxtv.org/hg/~jhoogenraad/rtl2831-r2/rev/549b40b69fa4

I agree this is not the nicest solution, but at at least would reduce 
some problems.
I am not sure on how to create a patch for this ?
Should I make a new branch in the HG archive ?

I suggest to replate the 111-s into IR_CODES_DIBUSB_TABLE_LEN in the 
files below as well:

     * drivers/media/dvb/dvb-usb/dibusb-common.c:
           o line 330
           o line 459
     * drivers/media/dvb/dvb-usb/dibusb-mb.c:
           o line 215
           o line 299
           o line 363
           o line 420
     * drivers/media/dvb/dvb-usb/dibusb.h, line 127

-- 
Jan Hoogenraad
Hoogenraad Interface Services
Postbus 2717
3500 GS Utrecht

Re: ir_codes_dibusb_table : please make table length a constant

[Mauro Carvalho Chehab]

Em 26-08-2010 03:53, Jan Hoogenraad escreveu:
> During debugging a driver, I found that at least in
> 
> linux/drivers/media/dvb/dvb-usb/dibusb-mc.c   
> the length of ir_codes_dibusb_table was referring to an older version.
> This may have caused memory overrun problems
> 
> I have fixed this in my branch in
> linux/drivers/media/dvb/dvb-usb/dibusb.h
>  with a
> #define IR_CODES_DIBUSB_TABLE_LEN 111
> 
> http://linuxtv.org/hg/~jhoogenraad/rtl2831-r2/rev/549b40b69fa4
> 
> I agree this is not the nicest solution, but at at least would reduce some problems.
> I am not sure on how to create a patch for this ?
> Should I make a new branch in the HG archive ?
> 
> I suggest to replate the 111-s into IR_CODES_DIBUSB_TABLE_LEN in the files below as well:
> 
>     * drivers/media/dvb/dvb-usb/dibusb-common.c:
>           o line 330
>           o line 459
>     * drivers/media/dvb/dvb-usb/dibusb-mb.c:
>           o line 215
>           o line 299
>           o line 363
>           o line 420
>     * drivers/media/dvb/dvb-usb/dibusb.h, line 127
> 
The proper solution is to migrate dibusb to use the new ir-core way, just like we did at
drivers/media/dvb/dvb-usb/dib0700_devices.c:

http://git.linuxtv.org/media_tree.git?a=commitdiff;h=d700226902a62a3b6f3563782d569c0e2af74397
http://git.linuxtv.org/media_tree.git?a=commitdiff;h=5af935cc96a291f90799bf6a2587d87329a91699
http://git.linuxtv.org/media_tree.git?a=commitdiff;h=72b393106bddc9f0a1ab502b4c8c5793a0441a30

Cheers,
Mauro