devil-linux-develop Mailing List for Devil-Linux (Page 40)

I'm running a fresh 1.3 compile this morning, and just happened to
notice a weird message from coreutils.  I don't know what it means, and
I don't know why it displayed on the terminal, instead of going to LOGS:


executing coreutils with option build (in /data/build/tmp/coreutils-6.10)
*** %n in writable segment detected ***


It didn't abort, as my compile is still running.  Anyone have a clue?

 - BS

Heiko,

Heiko Zuerker wrote:
...
>> The difference is trivial:
>> build.sh ( line 403)
>> - sed -e 's/i.86/i$CONFIG_CPU/'  | sed -e 's/x86_64/i$CONFIG_CPU/'
>> + sed -e 's/i.86/i$CONFIG_CPU/g' | sed -e 's/x86_64/i$CONFIG_CPU/g'
> 
> Good catch Serge!
> I think we need the 'g' in there, since we nedd to make sure we  
> completely fool the build scripts into thinking it's a 32 bit machine.
> 
> Heiko

Thank you! I've checked it in.

--
Sincerely,
Serge Leschinsky

Quoting Serge Leschinsky <fi...@in...>:

> Hi,
>
> There is a question I cannot resolve myself. May I ask to give me a hint?
>
> Which "uname" output is more correct (from our build system's viewpoint) ?
>
> 1) We have it now:
> root:/data/build# /bin/uname -a
> Linux zab-dev.nn.t.com 2.6.24.3 #1 SMP Wed Feb 20 15:59:10 EST 2008   
> i686 x86_64
> x86_64 GNU/Linux
>
> 2) I think it's more strictly, but I'm unsure.
> root:/data/build# /bin/uname -a
> Linux zab-dev.nn.t.com 2.6.24.3 #1 SMP Wed Feb 20 15:59:10 EST 2008 i686 i686
> i686 GNU/Linux
>
> The difference is trivial:
> build.sh ( line 403)
> - sed -e 's/i.86/i$CONFIG_CPU/'  | sed -e 's/x86_64/i$CONFIG_CPU/'
> + sed -e 's/i.86/i$CONFIG_CPU/g' | sed -e 's/x86_64/i$CONFIG_CPU/g'

Good catch Serge!
I think we need the 'g' in there, since we nedd to make sure we  
completely fool the build scripts into thinking it's a 32 bit machine.

Heiko

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.

Hi,

There is a question I cannot resolve myself. May I ask to give me a hint?

Which "uname" output is more correct (from our build system's viewpoint) ?

1) We have it now:
root:/data/build# /bin/uname -a
Linux zab-dev.nn.t.com 2.6.24.3 #1 SMP Wed Feb 20 15:59:10 EST 2008 i686 x86_64
x86_64 GNU/Linux

2) I think it's more strictly, but I'm unsure.
root:/data/build# /bin/uname -a
Linux zab-dev.nn.t.com 2.6.24.3 #1 SMP Wed Feb 20 15:59:10 EST 2008 i686 i686
i686 GNU/Linux

The difference is trivial:
build.sh ( line 403)
- sed -e 's/i.86/i$CONFIG_CPU/'  | sed -e 's/x86_64/i$CONFIG_CPU/'
+ sed -e 's/i.86/i$CONFIG_CPU/g' | sed -e 's/x86_64/i$CONFIG_CPU/g'

Thank you,

--
Sincerely,
Serge Leschinsky

> Hi Heiko,
>
> Heiko Zuerker wrote:
>> Serge,
>>
>> could you please check in the ones you got the fix for?
>
> Of course, but I'd like to verify these changes before committing (the
> build is
> already started). I'll check in it tomorrow.

I like that approach, sorry for rushing you.

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Hi Heiko,

Heiko Zuerker wrote:
> Serge,
> 
> could you please check in the ones you got the fix for?

Of course, but I'd like to verify these changes before committing (the build is
already started). I'll check in it tomorrow.

--
Sincerely,
Serge Leschinsky

Serge,

could you please check in the ones you got the fix for?

Heiko

> Hi,
>
> Heiko Zuerker wrote:
>> Just got this email, but don't think I'll have too much time looking
>> into it.
>> Anybody got time to take a closer look?
> ....
>> ipsec -- required modules missing (Pluto, netkey)
>> ipsec -- too old package, should use racoon & setkey (ipsec-tools deb
>> package) instead
> As far as I understand it's about DL 1.2. There is StrongSwan package (
> not
> ipsec-tools ) and it's up-to-date (2.8.8). I can say nothing about how it
> currently works, unfortunately, but hopefully it's ok. OpenSwan worked
> perfectly.
>
>> denyhosts -- not in sync with syslog-ng (auth.log not defined in
>> syslog-ng)
>> denyhosts -- stops immediately after start, should use sshd_sentry
>> instead
> never use it, sorry
>
>> updatedb -- cannot be run at all, cannot be fixed, (unimportant small
>> prog)
> Well, at least in 1.3 it doesn't work due to few files absence.
> I guess it should help  - untested.
>
> --- findutils   12 Dec 2004 21:48:08 -0000      1.11
> +++ findutils   11 Mar 2008 10:00:25 -0000
> @@ -35,7 +35,7 @@
>                 mkdir -p $TMPDIR/usr || exit 1
>                 make install DESTDIR=$TMPDIR/ || exit 1
>                 copy_docs $TMPDIR
> -               rm -rf $TMPDIR/usr/libexec || exit 1
> +               #rm -rf $TMPDIR/usr/libexec || exit 1
>                 rm -rf $TMPDIR/usr/info || exit 1
>                 cp -dpR $TMPDIR/* $CDDIR/|| exit 1
>                 rm -rf $TMPDIR || exit 1few
>
>
>> logwatch -- some required perl modules missing (e.g. Manip.pmc),
>> though it works
>>
>> logrotate missing (e.g. from /etc/cron.daily, etc...)
> root@Devil:~ # grep logrotate /etc/crontab
> #1 0 * * *      root    test -x /usr/sbin/logrotate && /usr/sbin/logrotate
> -s
> /var/log/logrotate.status /etc/logrotate.conf
>
>
>> /etc/init.d/mountfs -- variables DISABLE_MD_AUTOSTART and
>> DISABLE_LVM_AUTOSTART are not read from  /etc/sysconfig/config
>> So LVM and RAID keeps trying to start at boot, though someone does NOT
>> want it.
> Oops... Should be fixed by the following:
>
> diff -u -r1.43 mountfs
> --- mountfs     9 Jul 2007 12:43:14 -0000       1.43
> +++ mountfs     11 Mar 2008 10:11:01 -0000
> @@ -21,6 +21,7 @@
>  #
>
>  source /etc/init.d/functions
> +source /etc/sysconfig/config
>
>  # supported LVs
>  LVMTAB="/etc/sysconfig/lvmtab"
>
>
>
>
>>
>>
>> etc ... (with minor fix> freshclam, clamav)
>>
>> I share the opinion of those who suggest upgrading kernel to 2.6.x
>> (x>20)
> Have a look at DL 1.3  ( kernel 2.6.24, ipsec-tools etc)
>
> --
> Sincerely,
> Serge Leschinsky
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Devil-linux-develop mailing list
> Dev...@li...
> https://lists.sourceforge.net/lists/listinfo/devil-linux-develop
>


-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Hi,

Heiko Zuerker wrote:
> Just got this email, but don't think I'll have too much time looking into it.
> Anybody got time to take a closer look?
....
> ipsec -- required modules missing (Pluto, netkey)
> ipsec -- too old package, should use racoon & setkey (ipsec-tools deb
> package) instead
As far as I understand it's about DL 1.2. There is StrongSwan package ( not
ipsec-tools ) and it's up-to-date (2.8.8). I can say nothing about how it
currently works, unfortunately, but hopefully it's ok. OpenSwan worked perfectly.

> denyhosts -- not in sync with syslog-ng (auth.log not defined in syslog-ng)
> denyhosts -- stops immediately after start, should use sshd_sentry instead
never use it, sorry

> updatedb -- cannot be run at all, cannot be fixed, (unimportant small prog)
Well, at least in 1.3 it doesn't work due to few files absence.
I guess it should help  - untested.

--- findutils   12 Dec 2004 21:48:08 -0000      1.11
+++ findutils   11 Mar 2008 10:00:25 -0000
@@ -35,7 +35,7 @@
                mkdir -p $TMPDIR/usr || exit 1
                make install DESTDIR=$TMPDIR/ || exit 1
                copy_docs $TMPDIR
-               rm -rf $TMPDIR/usr/libexec || exit 1
+               #rm -rf $TMPDIR/usr/libexec || exit 1
                rm -rf $TMPDIR/usr/info || exit 1
                cp -dpR $TMPDIR/* $CDDIR/|| exit 1
                rm -rf $TMPDIR || exit 1few


> logwatch -- some required perl modules missing (e.g. Manip.pmc),
> though it works
> 
> logrotate missing (e.g. from /etc/cron.daily, etc...)
root@Devil:~ # grep logrotate /etc/crontab
#1 0 * * *      root    test -x /usr/sbin/logrotate && /usr/sbin/logrotate -s
/var/log/logrotate.status /etc/logrotate.conf


> /etc/init.d/mountfs -- variables DISABLE_MD_AUTOSTART and
> DISABLE_LVM_AUTOSTART are not read from  /etc/sysconfig/config
> So LVM and RAID keeps trying to start at boot, though someone does NOT
> want it.
Oops... Should be fixed by the following:

diff -u -r1.43 mountfs
--- mountfs     9 Jul 2007 12:43:14 -0000       1.43
+++ mountfs     11 Mar 2008 10:11:01 -0000
@@ -21,6 +21,7 @@
 #

 source /etc/init.d/functions
+source /etc/sysconfig/config

 # supported LVs
 LVMTAB="/etc/sysconfig/lvmtab"




> 
> 
> etc ... (with minor fix> freshclam, clamav)
> 
> I share the opinion of those who suggest upgrading kernel to 2.6.x (x>20)
Have a look at DL 1.3  ( kernel 2.6.24, ipsec-tools etc)

--
Sincerely,
Serge Leschinsky

Just got this email, but don't think I'll have too much time looking into it.
Anybody got time to take a closer look?

Thx
Heiko

---------------------------- Original Message ----------------------------
Subject: DL improvements
From:    gy...@we...
Date:    Mon, March 10, 2008 08:16
To:      he...@de...
--------------------------------------------------------------------------

Hi head of DL,

O.K.
If I would suggest some improvements ....

Here is a list of bugs, ie services not working properly or missing ...

By the way DL is a very good frame tool, ie a very good tool to get
hands on experience with setting up a Linux box with working services.
So it is a very good tool for a student to experience with his/her Linux box.

ipsec -- required modules missing (Pluto, netkey)
ipsec -- too old package, should use racoon & setkey (ipsec-tools deb
package) instead

denyhosts -- not in sync with syslog-ng (auth.log not defined in syslog-ng)
denyhosts -- stops immediately after start, should use sshd_sentry instead

updatedb -- cannot be run at all, cannot be fixed, (unimportant small prog)

logwatch -- some required perl modules missing (e.g. Manip.pmc),
though it works

logrotate missing (e.g. from /etc/cron.daily, etc...)

postfix -- /etc/aliases should not be empty, after fix it works -:)
postfix -- in main.cf hash:/opt/majordomo/aliases unnecessary
postfix -- in main.cf alias_maps = hash:/etc/aliases (nis not needed)

/etc/init.d/mountfs -- variables DISABLE_MD_AUTOSTART and
DISABLE_LVM_AUTOSTART are not read from  /etc/sysconfig/config
So LVM and RAID keeps trying to start at boot, though someone does NOT
want it.


etc ... (with minor fix> freshclam, clamav)

I share the opinion of those who suggest upgrading kernel to 2.6.x (x>20)
Why>
--a lot of network card unsupported in 2.4.x (e.g. mcs7830 USB)
--a lot of v4l2 device not supported in 2.4.x (e.g. for motion )
--more and most recent iptables modules support
--etc ..

Nowadays a very important view is: Energy Saving, so ....
an old Laptop consumes only 10-20W, compared to a PC box (about 100W)...
(Save the nature)
so (e.g. for me) I am running DL on an old PII Laptop, but it took a
lot of time to tailor DL to my needs. E.g. by default there is no
PCMCIA enabled in DL default kernel, besides cardmgr with module
i82365 has a lot  of bug compared with yenta_socket.
yenta_socket works perfectly
\
Fortunately your custom_cd script saved me a lot of energy in building
a new img -:)
....
Thanks
Gy











----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.



-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

> I had problems compiling dhcp simmilar problem with motion
> The repository i use rel-1-2-patches.
>
> This patch seems to work for dhcp. The motion i skipped for comiling but
> seems to be the same problem
> configured with TMP directory.

Corrected in CVS

> I have createted package for chillispot & zabbix would you like this in
> devil-linux ?

Yes that would be interesting.

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Simmilar problems, with tmp dir during install

I had problems compiling dhcp simmilar problem with motion
The repository i use rel-1-2-patches.

This patch seems to work for dhcp. The motion i skipped for comiling but 
seems to be the same problem
configured with TMP directory.
I have createted package for chillispot & zabbix would you like this in 
devil-linux ?

--- /var/tmp/lfssystem/build/scripts/dhcp       2008-02-25 
16:43:21.000000000 +0100
+++ /usr/src/lfssystem/build/scripts/dhcp       2008-02-28 
09:29:18.000000000 +0100
@@ -30,15 +30,15 @@
        if [ "$CONFIG_ISC_DHCP" = "y" ]; then
                # correct pathnames
                cat > site.conf <<-EOF
-               USERBINDIR = $TMPDIR/usr/bin
-               BINDIR = $TMPDIR/usr/sbin
-               CLIENTBINDIR=$TMPDIR/sbin
-               ADMMANDIR = $TMPDIR/usr/share/man/man8
-               FFMANDIR = $TMPDIR/usr/share/man/man5
-               LIBMANDIR = $TMPDIR/usr/share/man/man3
-               USRMANDIR = $TMPDIR/usr/share/man/man1
-               LIBDIR=$TMPDIR/usr/local/lib
-               INCDIR=$TMPDIR/usr/local/include
+               USERBINDIR = /usr/bin
+               BINDIR = /usr/sbin
+               CLIENTBINDIR=/sbin
+               ADMMANDIR = /usr/share/man/man8
+               FFMANDIR = /usr/share/man/man5
+               LIBMANDIR = /usr/share/man/man3
+               USRMANDIR = /usr/share/man/man1
+               LIBDIR=/usr/local/lib
+               INCDIR=/usr/local/include
 EOF

                ./configure || exit 1
@@ -51,7 +51,7 @@
        if [ "$CONFIG_ISC_DHCP" = "y" ]; then
                rm -rf $TMPDIR || exit 1
                mkdir -p $TMPDIR/sbin || exit 1
-               make install || exit 1
+               make install DESTDIR=$TMPDIR || exit 1
                copy_docs $TMPDIR
                rm -rf $TMPDIR/usr/local || exit 1
                rm -rf $TMPDIR/usr/share || exit 1

> Hey,
>
> I screwed up the ftp repository of the 1.3 sources, please do not update
> from the FTP server until you hear from me again.
> I have to finish my test compile first before I can fix it.
>
> Sorry for that.

The FTP repository should be fine again.

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Hey,

I screwed up the ftp repository of the 1.3 sources, please do not update
from the FTP server until you hear from me again.
I have to finish my test compile first before I can fix it.

Sorry for that.

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Serge,

I don't think we have any other choice then using this patch until
something official is available.

Heiko

> Heiko,
>
> I asked my colleague (a very experienced C coder) to help with the patch.
> After
> several hours of the source code looking thru he had said, that the
> 'really
> correct patch' will take a lot of time and efforts which he can't to
> afford.
> Instead of this he suggested a 'quick fix' which allows to compile ntop
> but
> produces the compiler warning. The patch is below, as you can see he just
> removed void casting. So, it's the question - will we include this
> temporary fix
> till a proper patch appears or not?
>
> --- include/net-snmp/library/tools.h.orig    2008-02-07 14:43:10.000000000
> +0300
> +++ include/net-snmp/library/tools.h    2008-02-08 09:57:51.000000000
> +0300
> @@ -50,7 +50,7 @@
>
>  /** @def SNMP_FREE(s)
>      Frees a pointer only if it is !NULL and sets its value to NULL */
> -#define SNMP_FREE(s)    do { if (s) { free((void *)s); s=NULL; } }
> while(0)
> +#define SNMP_FREE(s)    do { if (s) { free(s); s=NULL; } } while(0)
>
>  /** @def SNMP_SWIPE_MEM(n, s)
>      Frees pointer n only if it is !NULL, sets n to s and sets s to NULL
> */
>
>
>
> Heiko Zuerker wrote:
>>> The bug explanation (it is still open):
>>>
>>> https://bugzilla.redhat.com/show_bug.cgi?id=389631
> ...
>>
>> Yeah would be nice if they fix that soon.
>>
>
> --
> Sincerely,
> Serge Leschinsky
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Devil-linux-develop mailing list
> Dev...@li...
> https://lists.sourceforge.net/lists/listinfo/devil-linux-develop
>


-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Heiko,

I asked my colleague (a very experienced C coder) to help with the patch. After
several hours of the source code looking thru he had said, that the 'really
correct patch' will take a lot of time and efforts which he can't to afford.
Instead of this he suggested a 'quick fix' which allows to compile ntop but
produces the compiler warning. The patch is below, as you can see he just
removed void casting. So, it's the question - will we include this temporary fix
till a proper patch appears or not?

--- include/net-snmp/library/tools.h.orig    2008-02-07 14:43:10.000000000 +0300
+++ include/net-snmp/library/tools.h    2008-02-08 09:57:51.000000000 +0300
@@ -50,7 +50,7 @@

 /** @def SNMP_FREE(s)
     Frees a pointer only if it is !NULL and sets its value to NULL */
-#define SNMP_FREE(s)    do { if (s) { free((void *)s); s=NULL; } } while(0)
+#define SNMP_FREE(s)    do { if (s) { free(s); s=NULL; } } while(0)

 /** @def SNMP_SWIPE_MEM(n, s)
     Frees pointer n only if it is !NULL, sets n to s and sets s to NULL */



Heiko Zuerker wrote:
>> The bug explanation (it is still open):
>>
>> https://bugzilla.redhat.com/show_bug.cgi?id=389631
...
> 
> Yeah would be nice if they fix that soon.
> 

--
Sincerely,
Serge Leschinsky

> Hi,
>
> The bug explanation (it is still open):
>
> https://bugzilla.redhat.com/show_bug.cgi?id=389631
>
>>> /usr/include/net-snmp/library/container.h: In function
>>> 'CONTAINER_FREE':
>>> /usr/include/net-snmp/library/container.h:416: error: invalid lvalue in
>>> assignment make[3]: *** [iface.lo] Error 1

Yeah would be nice if they fix that soon.

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Hi,

The bug explanation (it is still open):

https://bugzilla.redhat.com/show_bug.cgi?id=389631

>> /usr/include/net-snmp/library/container.h: In function 'CONTAINER_FREE':
>> /usr/include/net-snmp/library/container.h:416: error: invalid lvalue in
>> assignment make[3]: *** [iface.lo] Error 1


--
Serge Leschinsky

On Thu, December 20, 2007 08:44, Jeba Anandhan wrote:
> Hi Heiko,
> Could you tell me where i can find the kernel source tree for DL 1.2.9
> so that i will compile the customized driver with the kernel reference and
> port into DL 1.2.9.

We actually do not keep any old (minor) versions, since this would take up
too much space on the FTP server. We only have the current 1.2 and 1.3
sources available.

1.2.9 is pretty old. I would recommend updating to the latest version,
since there were many security fixes.

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Hi Heiko,
Could you tell me where i can find the kernel source tree for DL 1.2.9
so that i will compile the customized driver with the kernel reference
and port into DL 1.2.9.

Thanks
Jeba
On Thu, 2007-12-20 at 07:33 -0600, Heiko Zuerker wrote:
> Hi Jeba,
> 
> On Thu, December 20, 2007 05:53, Jeba Anandhan wrote:
> > Hi all,
> > I have router which runs with devil Linux distro and have eql module for
> > bonding support. In my host machine, i have compiled new eql kernel module
> > from linux-2.4.35 tree. When i try to load the compiled module in devil
> > Linux kernel, The unresolved symbols has been displayed.
> 
> < ... cut ... >
> 
> > kindly let me know why these symbols are unresolved?..
> 
> If you want to add any kind of software or (especially) Kernel modules
> then you'll have to do this within our development environment.
> 
> Take a look at our docs:
> http://www.devil-linux.org/documentation/1.2.x/ch03.html
> 
> You'll have to go through one full compile once, in order to get the full
> environment you need to add a manually compiled module.
> 
> Of course you could also add it to the build scripts and then it would
> make your life easier when we update DL.
> 

Hi Jeba,

On Thu, December 20, 2007 05:53, Jeba Anandhan wrote:
> Hi all,
> I have router which runs with devil Linux distro and have eql module for
> bonding support. In my host machine, i have compiled new eql kernel module
> from linux-2.4.35 tree. When i try to load the compiled module in devil
> Linux kernel, The unresolved symbols has been displayed.

< ... cut ... >

> kindly let me know why these symbols are unresolved?..

If you want to add any kind of software or (especially) Kernel modules
then you'll have to do this within our development environment.

Take a look at our docs:
http://www.devil-linux.org/documentation/1.2.x/ch03.html

You'll have to go through one full compile once, in order to get the full
environment you need to add a manually compiled module.

Of course you could also add it to the build scripts and then it would
make your life easier when we update DL.

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Hi all,
I have router which runs with devil Linux distro and have eql module for
bonding support. In my host machine, i have compiled new eql kernel
module from linux-2.4.35 tree. When i try to load the compiled module in
devil Linux kernel, The unresolved symbols has been displayed.

# insmod eql.o
eql.o: eql.o: unresolved symbol __kfree_skb
eql.o: eql.o: unresolved symbol kmalloc
eql.o: eql.o: unresolved symbol __global_cli
eql.o: eql.o: unresolved symbol unregister_netdev
eql.o: eql.o: unresolved symbol
del_timer                                       
eql.o: eql.o: unresolved symbol
register_netdev                                 
eql.o: eql.o: unresolved symbol
dev_queue_xmit                                  
eql.o: eql.o: unresolved symbol
kfree                                           
eql.o: eql.o: unresolved symbol
__global_save_flags                             
eql.o: eql.o: unresolved symbol
__global_restore_flags                          
eql.o: eql.o: unresolved symbol
jiffies                                         
eql.o: eql.o: unresolved symbol
printk                                          
eql.o: eql.o: unresolved symbol
add_timer                                       
eql.o: eql.o: unresolved symbol
__dev_get_by_name                               




kindly let me know why these symbols are unresolved?..

Thanks
Jeba

Heiko,

> 
> Now that's a beautiful solution!
Thank you!

> I'm too lazy to look it up now, does [[:digit:] work for double digits too ?
> 
Actually [[:digit:]] works for one digit only, but +([[:digit:]]) is a 'bash'
analog of perl expression (\d)+, i.e. one or more times.

--
Sincerely
Serge Leschinsky

On Mon, December 3, 2007 01:44, Serge Leschinsky wrote:
> Heiko,
>
> I'm sorry for delay - I forgot to pay the bill and ISP blocked my account
> from December, 1... :-\
>
>
> Heiko Zuerker wrote:
>
>> I like it, then we don't have to download that huge kernel tarball
>> every time.
> It's the first good news for me today :-)
>
>
>> It may also work if you use the $KERNELVERSION variable (may be spelled
>> a little different) for the filename, then we don't need the separate
>> folder. Something like:
>> bzcat $DL_DIR/srv/patch-$KERNELVERSION.bz2 | patch -p1 || exit 1
> :-)  I tried to do that.
> To define filename for the file I should know minor version of the kernel
> patch or add asterisk as a mask. I mean KERNELVERSION returns 2.6.23 (in
> my case) but the patch has name patch-KERNELVERSION.MINOR_VERSION.bz2
> (2.6.23.9). If I use
> the pattern patch-KERNELVERSION.*.bz2 I get false positive, so I have to
> make the pattern narrower...
>
> 'man bash' helps me :-) and I've created the next variant:
>
>
> ===================================================================
> RCS file: /cvsroot/devil-linux/build/scripts/prepare,v
> retrieving revision 1.69 diff -u -r1.69 prepare --- scripts/prepare     22
> Oct 2007 14:08:51 -0000      1.69
> +++ scripts/prepare     1 Dec 2007 11:54:19 -0000
> @@ -40,6 +40,18 @@
> exit 1 fi
>
> +# We are applying minor version kernel patch here
> +       pushd $KERNELDIR > /dev/null
> +       if [ ! -f .minor_version_patch.done ]; then
> +       shopt -s extglob
> +         for i in $(ls ../../src/patch-$KERNELVERSION.+([[:digit:]]).bz2)
>  +         do
> +          bzcat $i | patch -p1 || exit 1
> +         done
> +          touch .minor_version_patch.done
> +       popd > /dev/null
> +       fi
> +
> # let's find out if we force the usage of old kernel include files
> # this is sometimes necessary, when they change too much in the new kernel
>  KERN_INC=$(find_src_dir kernel-includes)

Now that's a beautiful solution!
I'm too lazy to look it up now, does [[:digit:] work for double digits too ?

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Heiko,

I'm sorry for delay - I forgot to pay the bill and ISP blocked my account from
December, 1... :-\

Heiko Zuerker wrote:
> I like it, then we don't have to download that huge kernel tarball every
> time.
It's the first good news for me today :-)

> It may also work if you use the $KERNELVERSION variable (may be spelled a
> little different) for the filename, then we don't need the separate
> folder.
> Something like:
> bzcat $DL_DIR/srv/patch-$KERNELVERSION.bz2 | patch -p1 || exit 1
:-)  I tried to do that.
To define filename for the file I should know minor version of the kernel patch
or add asterisk as a mask. I mean KERNELVERSION returns 2.6.23 (in my case) but
the patch has name patch-KERNELVERSION.MINOR_VERSION.bz2 (2.6.23.9). If I use
the pattern patch-KERNELVERSION.*.bz2 I get false positive, so I have to make
the pattern narrower...

'man bash' helps me :-) and I've created the next variant:

===================================================================
RCS file: /cvsroot/devil-linux/build/scripts/prepare,v
retrieving revision 1.69
diff -u -r1.69 prepare
--- scripts/prepare     22 Oct 2007 14:08:51 -0000      1.69
+++ scripts/prepare     1 Dec 2007 11:54:19 -0000
@@ -40,6 +40,18 @@
            exit 1
        fi

+# We are applying minor version kernel patch here
+       pushd $KERNELDIR > /dev/null
+       if [ ! -f .minor_version_patch.done ]; then
+       shopt -s extglob
+         for i in $(ls ../../src/patch-$KERNELVERSION.+([[:digit:]]).bz2)
+         do
+          bzcat $i | patch -p1 || exit 1
+         done
+          touch .minor_version_patch.done
+       popd > /dev/null
+       fi
+
        # let's find out if we force the usage of old kernel include files
        # this is sometimes necessary, when they change too much in the new kernel
        KERN_INC=$(find_src_dir kernel-includes)


-- 
Sincerely,
Serge Leschinsky