devil-linux-develop — developer discussions

[Devil-linux-develop] 1.2 ???

[Heiko Z. <he...@zu...>]

Howdy folks,

we were discussing if we should release 1.2 in the (very) near future.
Does anybody have any changes you want to have included?

Any comments?

Heiko

Re: [Devil-linux-develop] [PATCH] config parameter

[Heiko Z. <he...@zu...>]

Roland Pabel wrote:
> Hi,
> here is my implementation for the config boot paramter. I've tested it, and it 
> works as proposed the other day. I increased the verbosity of the output a 
> bit, which sometimes leads to lines longer than 79 characters (which was 
> avoided when possible). Is that a problem? (those devfs pathnames are just 
> too long...)
> Another thing that annoyed me: the "sleep 15" for usb-storage.  If usb-storage 
> has no devices to initialize, that can be skipped. That's what the other 
> patch is for ( but maybe someone can think of a better implementation )
> Roland

I all patch and test your stuff.

Can you also send a patch for the CHANGES file?

And yes, I don't like the 15 seconds either.

Heiko

Re: [Devil-linux-develop] apcups abort.

[Heiko Z. <he...@zu...>]

Roland Pabel wrote:
> On Thursday 08 April 2004 14:52, Bruce Smith wrote:
> 
>>Whoever checked in the apcups stuff forgot to upload a file to CVS.
>>The install script aborts on this line:
>>
>> cp -dp $MYDIR/scripts/apcupsd_apache2.config $ETCDIR/etc/apcupsd/ || exit
>>1
>>
>>because the file apcupsd_apache2.config doesn't exist anywhere.
>>
>>I'm going to comment out that line in the script for now, please
>>uncomment when the file is uploaded.
> 
> That file was in the patch I sent, I guess Heiko forgot "cvs add"...

Onces you reach 30, it just get's harder and harder....

Heiko

Re: [Devil-linux-develop] apcups abort.

[Heiko Z. <he...@zu...>]

Bruce Smith wrote:
> Whoever checked in the apcups stuff forgot to upload a file to CVS. 

That'll be me. :-((

> The install script aborts on this line:
> 
>  cp -dp $MYDIR/scripts/apcupsd_apache2.config $ETCDIR/etc/apcupsd/ || exit 1
> 
> because the file apcupsd_apache2.config doesn't exist anywhere.
> 
> I'm going to comment out that line in the script for now, please
> uncomment when the file is uploaded.
> 

I added the file to CVS.

Heiko

Re: [Devil-linux-develop] apcups abort.

[Roland P. <pa...@ta...>]

On Friday 09 April 2004 01:44, Roland Pabel wrote:
> On Thursday 08 April 2004 14:52, Bruce Smith wrote:
> > Whoever checked in the apcups stuff forgot to upload a file to CVS.
> > The install script aborts on this line:
> >
> >  cp -dp $MYDIR/scripts/apcupsd_apache2.config $ETCDIR/etc/apcupsd/ ||
> > exit 1
> >
> > because the file apcupsd_apache2.config doesn't exist anywhere.
> >
> > I'm going to comment out that line in the script for now, please
> > uncomment when the file is uploaded.
>
> That file was in the patch I sent, I guess Heiko forgot "cvs add"...
He may be forgetful, but he is fast :-)
Roland
-- 
ICQ UIN 49339118	Linux Counter #88774
GPG-Key 1024D/59C6AFA6 2003-02-07 Roland Pabel <ro...@pa...>

Re: [Devil-linux-develop] apcups abort.

[Roland P. <pa...@ta...>]

On Thursday 08 April 2004 14:52, Bruce Smith wrote:
> Whoever checked in the apcups stuff forgot to upload a file to CVS.
> The install script aborts on this line:
>
>  cp -dp $MYDIR/scripts/apcupsd_apache2.config $ETCDIR/etc/apcupsd/ || exit
> 1
>
> because the file apcupsd_apache2.config doesn't exist anywhere.
>
> I'm going to comment out that line in the script for now, please
> uncomment when the file is uploaded.
That file was in the patch I sent, I guess Heiko forgot "cvs add"...
Roland
-- 
ICQ UIN 49339118	Linux Counter #88774
GPG-Key 1024D/59C6AFA6 2003-02-07 Roland Pabel <ro...@pa...>

[Devil-linux-develop] [PATCH] config parameter

[Roland P. <ro...@pa...>]

Hi,
here is my implementation for the config boot paramter. I've tested it, and it 
works as proposed the other day. I increased the verbosity of the output a 
bit, which sometimes leads to lines longer than 79 characters (which was 
avoided when possible). Is that a problem? (those devfs pathnames are just 
too long...)
Another thing that annoyed me: the "sleep 15" for usb-storage.  If usb-storage 
has no devices to initialize, that can be skipped. That's what the other 
patch is for ( but maybe someone can think of a better implementation )
Roland
-- 
ICQ UIN 49339118	Linux Counter #88774
GPG-Key 1024D/59C6AFA6 2003-02-07 Roland Pabel <ro...@pa...>

[Devil-linux-develop] apcups abort.

[Bruce S. <bw...@ar...>]

Whoever checked in the apcups stuff forgot to upload a file to CVS. 
The install script aborts on this line:

 cp -dp $MYDIR/scripts/apcupsd_apache2.config $ETCDIR/etc/apcupsd/ || exit 1

because the file apcupsd_apache2.config doesn't exist anywhere.

I'm going to comment out that line in the script for now, please
uncomment when the file is uploaded.

 - BS

Re: [Devil-linux-develop] devices and configs

[Heiko Z. <he...@zu...>]

Roland Pabel wrote:
> Hi,
> I'm think of saving my configs on the harddisks instead of floppies, but there 
> is one thing that is not possible with config on disks now: changing them 
> easily prior to boot. Can't replace my harddisk as I do it with floppies...
> I thought of this:
> I'd add a boot parameter "config=<device>:<name of etc.tar.bz2>", i.e. 
> "config=/dev/hda1:backup.tar.bz2".
> The execution flow in linuxrc:
> 
> - device and filename given:
> search that device for file, if not found, standard behavious (load first 
> etc.tar.bz2)
> - only device given:
> device would be prepended to the list of devices in linuxrc and searched first 
> - only filename
> file would be searched on any device first, if not found, search for 
> etc.tar.bz2 on all devices in standard order
> 
> that last two cases can be easily seperated: devicenames contain "/dev/", 
> filenames *should* not...
> Comments?

I think this is a good idea.

Heiko

[Devil-linux-develop] Attention: fixed patch-o-matic in the 1.0.x series

[Friedrich L. <fl...@fl...>]

Hi!

I just fixed the patch-o-matic in the 1.0.x series. If you are now having 
troubles in the kernel compile stage, what you will hopefully not have, 
then please back out this patch and contact me.

Here the log message:

linux:~/devil-linux/lfssystem/data/build/scripts # cvs log -r1.8.2.1 
patch-o-matic

RCS file: /cvsroot/devil-linux/build/scripts/patch-o-matic,v
Working file: patch-o-matic
[...]
description:
----------------------------
revision 1.8.2.1
date: 2004/04/06 23:33:59;  author: friedl;  state: Exp;  lines: +23 -11
While searching why the pptp netfilter connection tracking modul disappeared
somewhere in the 1.0.x series, I discovered that the netfilter patch-o-matic
patches fail already pretty soon in their patch process.

This is now corrected. Hopefully we are not introducing some problems
because certain netfilter modules that should have been there, are now
really there.

In case there are troubles mail fl...@fl...
=============================================================================

-- 
MfG / Regards
Friedrich Lobenstock
____________________________________________________________________
Friedrich Lobenstock                       Linux Services Lobenstock
URL: http://www.lsl.at/                         Email: fl...@fl...
____________________________________________________________________

Re: [Devil-linux-develop] devices and configs

[Friedrich L. <fl...@fl...>]

Roland Pabel wrote on 07.04.2004 01:33 MET:
> On Wednesday 07 April 2004 01:24, Friedrich Lobenstock wrote:
> 
>>Roland Pabel wrote on 07.04.2004 01:18 MET:
> 
> [...]
> 
>>Äh, what do you mean by "any device first"? Do you intend to have some
>>function that try's all possible and impossible block devices and their
>>partitions if any?
> 
> I don't need to code that, that essentially what linuxrc does...
> "Any" here just means "scan the standard devices", which is what the 
> CONFIG_SRC_LIST variable hols in the linuxrc script. (That is now almost 
> "every devices": ide-/scsi- cdroms/disks/partition and floppies).
> What I wanted to describe was: when a filename is given, search for it first; 
> if not found, search for "etc.tar.bz2" in the 'usual places' (is that better 
> than "any device" ? ;-) )

Yes that's more wording ;-)

-- 
MfG / Regards
Friedrich Lobenstock
____________________________________________________________________
Friedrich Lobenstock                       Linux Services Lobenstock
URL: http://www.lsl.at/                         Email: fl...@fl...
____________________________________________________________________

Re: [Devil-linux-develop] devices and configs

[Roland P. <pa...@ta...>]

On Wednesday 07 April 2004 01:24, Friedrich Lobenstock wrote:
> Roland Pabel wrote on 07.04.2004 01:18 MET:
[...]
>
> =C4h, what do you mean by "any device first"? Do you intend to have some
> function that try's all possible and impossible block devices and their
> partitions if any?
I don't need to code that, that essentially what linuxrc does...
"Any" here just means "scan the standard devices", which is what the=20
CONFIG_SRC_LIST variable hols in the linuxrc script. (That is now almost=20
"every devices": ide-/scsi- cdroms/disks/partition and floppies).
What I wanted to describe was: when a filename is given, search for it firs=
t;=20
if not found, search for "etc.tar.bz2" in the 'usual places' (is that bette=
r=20
than "any device" ? ;-) )
Roland
=2D-=20
ICQ UIN 49339118	Linux Counter #88774
GPG-Key 1024D/59C6AFA6 2003-02-07 Roland Pabel <ro...@pa...>

Re: [Devil-linux-develop] devices and configs

[Friedrich L. <fl...@fl...>]

Roland Pabel wrote on 07.04.2004 01:18 MET:
> I'm think of saving my configs on the harddisks instead of floppies, but there 
> is one thing that is not possible with config on disks now: changing them 
> easily prior to boot. Can't replace my harddisk as I do it with floppies...
> I thought of this:
> I'd add a boot parameter "config=<device>:<name of etc.tar.bz2>", i.e. 
> "config=/dev/hda1:backup.tar.bz2".
> The execution flow in linuxrc:
> 
> - device and filename given:
> search that device for file, if not found, standard behavious (load first 
> etc.tar.bz2)
> - only device given:
> device would be prepended to the list of devices in linuxrc and searched first 
> - only filename
> file would be searched on any device first, if not found, search for 
> etc.tar.bz2 on all devices in standard order

Äh, what do you mean by "any device first"? Do you intend to have some 
function that try's all possible and impossible block devices and their 
partitions if any?

> that last two cases can be easily seperated: devicenames contain "/dev/", 
> filenames *should* not...

That's right.

-- 
MfG / Regards
Friedrich Lobenstock
____________________________________________________________________
Friedrich Lobenstock                       Linux Services Lobenstock
URL: http://www.lsl.at/                         Email: fl...@fl...
____________________________________________________________________

[Devil-linux-develop] devices and configs

[Roland P. <pa...@ta...>]

Hi,
I'm think of saving my configs on the harddisks instead of floppies, but there 
is one thing that is not possible with config on disks now: changing them 
easily prior to boot. Can't replace my harddisk as I do it with floppies...
I thought of this:
I'd add a boot parameter "config=<device>:<name of etc.tar.bz2>", i.e. 
"config=/dev/hda1:backup.tar.bz2".
The execution flow in linuxrc:

- device and filename given:
search that device for file, if not found, standard behavious (load first 
etc.tar.bz2)
- only device given:
device would be prepended to the list of devices in linuxrc and searched first 
- only filename
file would be searched on any device first, if not found, search for 
etc.tar.bz2 on all devices in standard order

that last two cases can be easily seperated: devicenames contain "/dev/", 
filenames *should* not...
Comments?
Roland
-- 
ICQ UIN 49339118	Linux Counter #88774
GPG-Key 1024D/59C6AFA6 2003-02-07 Roland Pabel <ro...@pa...>

Re: [Devil-linux-develop] Kernel abort - AGAIN.

[Bruce S. <bw...@ar...>]

OK, with i586-UP, PSCHED_JIFFIES, Variable HZ = Off, everything
compiled, including the kernel and Zebra.

I don't know what's the problem was before.  Weird ...

 - BS


> > mrproper/new-LFS
> > 
> > >From .config:
> > 
> > CONFIG_CPU=586
> > CONFIG_KERNEL_CPU=586
> > CONFIG_LINUX_NR_CPUS=2
> > CONFIG_LINUX_SMP=n
> > CONFIG_LINUX_VERSION=2.4
> > CONFIG_PSCHED_CLOCK_SOURCE=PSCHED_GETTIMEOFDAY
> 
> Ok, back to PSCHED_JIFFIES.
> 
> > gcc -D__KERNEL__ -I/data/build/tmp/linux-2.4.25/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -fomit-frame-pointer  -pipe -mpreferred-stack-boundary=2 -march=i586 -DMODULE -DMODVERSIONS -include /data/build/tmp/linux-2.4.25/include/linux/modversions.h  -nostdinc -iwithprefix include -DKBUILD_BASENAME=sch_htb  -c -o sch_htb.o sch_htb.c
> > sch_htb.c: In function `htb_debug_dump':
> > sch_htb.c:364: warning: statement with no effect
> > sch_htb.c: In function `htb_charge_class':
> > sch_htb.c:821: error: aggregate value used where an integer was expected
> > sch_htb.c:822: error: aggregate value used where an integer was expected
> > sch_htb.c:815: warning: statement with no effect
> > sch_htb.c: In function `htb_do_events':
> > sch_htb.c:892: error: aggregate value used where an integer was expected
> > sch_htb.c:893: error: aggregate value used where an integer was expected
> > sch_htb.c:886: warning: statement with no effect
> > make[3]: *** [sch_htb.o] Error 1
> > make[3]: Leaving directory `/data/build/tmp/linux-2.4.25/net/sched'
> > make[2]: *** [_modsubdir_sched] Error 2
> > make[2]: Leaving directory `/data/build/tmp/linux-2.4.25/net'
> > make[1]: *** [_mod_net] Error 2
> > make[1]: Leaving directory `/data/build/tmp/linux-2.4.25'
> > (END)
> 
> It seems that the htb scheduler add-on has not clean code.
> I'll try fix that when time permits.
> 
> 						Jaroslav

Re: [Devil-linux-develop] Kernel abort - AGAIN.

[<hzu...@ra...>]

On 04/06/2004 11:12:40 AM Bruce Smith wrote:
>> > gcc -D__KERNEL__ -I/data/build/tmp/linux-2.4.25/include -Wall
>-Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common
>-fomit-frame-pointer  -pipe -mpreferred-stack-boundary=2 -march=i586
>-DMODULE -DMODVERSIONS -include
>/data/build/tmp/linux-2.4.25/include/linux/modversions.h  -nostdinc
>-iwithprefix include -DKBUILD_BASENAME=sch_htb  -c -o sch_htb.o
>sch_htb.c
>> > sch_htb.c: In function `htb_debug_dump':
>> > sch_htb.c:364: warning: statement with no effect
>> > sch_htb.c: In function `htb_charge_class':
>> > sch_htb.c:821: error: aggregate value used where an integer was
>expected
>> > sch_htb.c:822: error: aggregate value used where an integer was
>expected
>> > sch_htb.c:815: warning: statement with no effect
>> > sch_htb.c: In function `htb_do_events':
>> > sch_htb.c:892: error: aggregate value used where an integer was
>expected
>> > sch_htb.c:893: error: aggregate value used where an integer was
>expected
>> > sch_htb.c:886: warning: statement with no effect
>> > make[3]: *** [sch_htb.o] Error 1
>> > make[3]: Leaving directory `/data/build/tmp/linux-2.4.25/net/sched'
>> > make[2]: *** [_modsubdir_sched] Error 2
>> > make[2]: Leaving directory `/data/build/tmp/linux-2.4.25/net'
>> > make[1]: *** [_mod_net] Error 2
>> > make[1]: Leaving directory `/data/build/tmp/linux-2.4.25'
>> > (END)
>>
>> It seems that the htb scheduler add-on has not clean code.
>> I'll try fix that when time permits.
>
>Is there something I can un-select or change so it compiles now?
>(I have no idea what a htb is :)

HTB is part of the QoS stuff and is a packet scheduler.

Heiko

Re: [Devil-linux-develop] Kernel abort - AGAIN.

[Jaroslav K. <pe...@pe...>]

On Tue, 6 Apr 2004, Bruce Smith wrote:

> > > sch_htb.c: In function `htb_do_events':
> > > sch_htb.c:892: error: aggregate value used where an integer was expected
> > > sch_htb.c:893: error: aggregate value used where an integer was expected
> > > sch_htb.c:886: warning: statement with no effect
> > 
> > It seems that the htb scheduler add-on has not clean code.
> > I'll try fix that when time permits.
> 
> Is there something I can un-select or change so it compiles now?
> (I have no idea what a htb is :)

Use PSCHED_JIFFIES as I suggested. It's the default settings in
2.4 kernels.

					Jaroslav

-----
Jaroslav Kysela <pe...@pe...>

Re: [Devil-linux-develop] Kernel abort - AGAIN.

[Bruce S. <bw...@ar...>]

> > gcc -D__KERNEL__ -I/data/build/tmp/linux-2.4.25/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -fomit-frame-pointer  -pipe -mpreferred-stack-boundary=2 -march=i586 -DMODULE -DMODVERSIONS -include /data/build/tmp/linux-2.4.25/include/linux/modversions.h  -nostdinc -iwithprefix include -DKBUILD_BASENAME=sch_htb  -c -o sch_htb.o sch_htb.c
> > sch_htb.c: In function `htb_debug_dump':
> > sch_htb.c:364: warning: statement with no effect
> > sch_htb.c: In function `htb_charge_class':
> > sch_htb.c:821: error: aggregate value used where an integer was expected
> > sch_htb.c:822: error: aggregate value used where an integer was expected
> > sch_htb.c:815: warning: statement with no effect
> > sch_htb.c: In function `htb_do_events':
> > sch_htb.c:892: error: aggregate value used where an integer was expected
> > sch_htb.c:893: error: aggregate value used where an integer was expected
> > sch_htb.c:886: warning: statement with no effect
> > make[3]: *** [sch_htb.o] Error 1
> > make[3]: Leaving directory `/data/build/tmp/linux-2.4.25/net/sched'
> > make[2]: *** [_modsubdir_sched] Error 2
> > make[2]: Leaving directory `/data/build/tmp/linux-2.4.25/net'
> > make[1]: *** [_mod_net] Error 2
> > make[1]: Leaving directory `/data/build/tmp/linux-2.4.25'
> > (END)
> 
> It seems that the htb scheduler add-on has not clean code.
> I'll try fix that when time permits.

Is there something I can un-select or change so it compiles now?
(I have no idea what a htb is :)

 - BS

Re: [Devil-linux-develop] Kernel abort - AGAIN.

[Jaroslav K. <pe...@pe...>]

On Mon, 5 Apr 2004, Bruce Smith wrote:

> mrproper/new-LFS
> 
> >From .config:
> 
> CONFIG_CPU=586
> CONFIG_KERNEL_CPU=586
> CONFIG_LINUX_NR_CPUS=2
> CONFIG_LINUX_SMP=n
> CONFIG_LINUX_VERSION=2.4
> CONFIG_PSCHED_CLOCK_SOURCE=PSCHED_GETTIMEOFDAY

Ok, back to PSCHED_JIFFIES.

> gcc -D__KERNEL__ -I/data/build/tmp/linux-2.4.25/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -fomit-frame-pointer  -pipe -mpreferred-stack-boundary=2 -march=i586 -DMODULE -DMODVERSIONS -include /data/build/tmp/linux-2.4.25/include/linux/modversions.h  -nostdinc -iwithprefix include -DKBUILD_BASENAME=sch_htb  -c -o sch_htb.o sch_htb.c
> sch_htb.c: In function `htb_debug_dump':
> sch_htb.c:364: warning: statement with no effect
> sch_htb.c: In function `htb_charge_class':
> sch_htb.c:821: error: aggregate value used where an integer was expected
> sch_htb.c:822: error: aggregate value used where an integer was expected
> sch_htb.c:815: warning: statement with no effect
> sch_htb.c: In function `htb_do_events':
> sch_htb.c:892: error: aggregate value used where an integer was expected
> sch_htb.c:893: error: aggregate value used where an integer was expected
> sch_htb.c:886: warning: statement with no effect
> make[3]: *** [sch_htb.o] Error 1
> make[3]: Leaving directory `/data/build/tmp/linux-2.4.25/net/sched'
> make[2]: *** [_modsubdir_sched] Error 2
> make[2]: Leaving directory `/data/build/tmp/linux-2.4.25/net'
> make[1]: *** [_mod_net] Error 2
> make[1]: Leaving directory `/data/build/tmp/linux-2.4.25'
> (END)

It seems that the htb scheduler add-on has not clean code.
I'll try fix that when time permits.

						Jaroslav

-----
Jaroslav Kysela <pe...@pe...>

Re: [Devil-linux-develop] Kernel abort.

[Jaroslav K. <pe...@pe...>]

On Mon, 5 Apr 2004, Bruce Smith wrote:

> > > I'm pretty sure that PSCHED_JIFFIES was set in the .config file.
> > > It worked a couple days ago on a i686-SMP compile.  All I did was
> > > switch menuconfig to a i586-UP.
> > > 
> > > The error message kind of implies that PSCHED_JIFFIES isn't supported by
> > > i586?  I just set it to PSCHED_GETTIMEOFDAY and restarted the compile.
> > > We'll see if that works, in a few hours...
> > 
> > The PSCHED_CPU is supported only when CONFIG_X86_TSC is selected in 
> > kernel so it does not work with all 586 CPUs. But there's a condition 
> > catching this situation in scripts/configuration/prepare.config . If I 
> > switch kernel CPU from 686 to 586 then PSCHED_CPU is changed to 
> > PSCHED_JIFFIES automatically. Do you have updated all files from CVS?
> 
> I'm pretty sure that I was using JIFFIES on i686-SMP before I switched
> to i586-UP.  (FWIW)

The PSCHED_CPU should never be chosen automagically.

> What's the best option to select for a i686 compile?  PSCHED_CPU?

Yes, because it has the best accuracy and is fast.

						Jaroslav

-----
Jaroslav Kysela <pe...@pe...>

Re: [Devil-linux-develop] Kernel abort - AGAIN.

[Bruce S. <bw...@ar...>]

I turned on Variable HZ and set it to 100, and the kernel compiled,
but then zebra build aborted ... (yes, newlfs/mrproper)

I'm running another compile now.  Weird ...

 - BS

Re: [Devil-linux-develop] Kernel abort - AGAIN.

[<hzu...@ra...>]

On 04/05/2004 02:02:10 PM Bruce Smith wrote:
>mrproper/new-LFS
>
>>From .config:
>
>CONFIG_CPU=586
>CONFIG_KERNEL_CPU=586
>CONFIG_LINUX_NR_CPUS=2
>CONFIG_LINUX_SMP=n
>CONFIG_LINUX_VERSION=2.4
>CONFIG_PSCHED_CLOCK_SOURCE=PSCHED_GETTIMEOFDAY
>
>
>gcc -D__KERNEL__ -I/data/build/tmp/linux-2.4.25/include -Wall
>-Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common
>-fomit-frame-pointer  -pipe -mpreferred-stack-boundary=2 -march=i586
>-DMODULE -DMODVERSIONS -include
>/data/build/tmp/linux-2.4.25/include/linux/modversions.h  -nostdinc
>-iwithprefix include -DKBUILD_BASENAME=sch_htb  -c -o sch_htb.o
>sch_htb.c
>sch_htb.c: In function `htb_debug_dump':
>sch_htb.c:364: warning: statement with no effect
>sch_htb.c: In function `htb_charge_class':
>sch_htb.c:821: error: aggregate value used where an integer was expected
>sch_htb.c:822: error: aggregate value used where an integer was expected
>sch_htb.c:815: warning: statement with no effect
>sch_htb.c: In function `htb_do_events':
>sch_htb.c:892: error: aggregate value used where an integer was expected
>sch_htb.c:893: error: aggregate value used where an integer was expected
>sch_htb.c:886: warning: statement with no effect
>make[3]: *** [sch_htb.o] Error 1
>make[3]: Leaving directory `/data/build/tmp/linux-2.4.25/net/sched'
>make[2]: *** [_modsubdir_sched] Error 2
>make[2]: Leaving directory `/data/build/tmp/linux-2.4.25/net'
>make[1]: *** [_mod_net] Error 2
>make[1]: Leaving directory `/data/build/tmp/linux-2.4.25'
>(END)

That's a question for Jaroslav.....

Heiko

Re: [Devil-linux-develop] Kernel abort - AGAIN.

[Bruce S. <bw...@ar...>]

mrproper/new-LFS

>From .config:

CONFIG_CPU=586
CONFIG_KERNEL_CPU=586
CONFIG_LINUX_NR_CPUS=2
CONFIG_LINUX_SMP=n
CONFIG_LINUX_VERSION=2.4
CONFIG_PSCHED_CLOCK_SOURCE=PSCHED_GETTIMEOFDAY


gcc -D__KERNEL__ -I/data/build/tmp/linux-2.4.25/include -Wall -Wstrict-prototypes -Wno-trigraphs -O2 -fno-strict-aliasing -fno-common -fomit-frame-pointer  -pipe -mpreferred-stack-boundary=2 -march=i586 -DMODULE -DMODVERSIONS -include /data/build/tmp/linux-2.4.25/include/linux/modversions.h  -nostdinc -iwithprefix include -DKBUILD_BASENAME=sch_htb  -c -o sch_htb.o sch_htb.c
sch_htb.c: In function `htb_debug_dump':
sch_htb.c:364: warning: statement with no effect
sch_htb.c: In function `htb_charge_class':
sch_htb.c:821: error: aggregate value used where an integer was expected
sch_htb.c:822: error: aggregate value used where an integer was expected
sch_htb.c:815: warning: statement with no effect
sch_htb.c: In function `htb_do_events':
sch_htb.c:892: error: aggregate value used where an integer was expected
sch_htb.c:893: error: aggregate value used where an integer was expected
sch_htb.c:886: warning: statement with no effect
make[3]: *** [sch_htb.o] Error 1
make[3]: Leaving directory `/data/build/tmp/linux-2.4.25/net/sched'
make[2]: *** [_modsubdir_sched] Error 2
make[2]: Leaving directory `/data/build/tmp/linux-2.4.25/net'
make[1]: *** [_mod_net] Error 2
make[1]: Leaving directory `/data/build/tmp/linux-2.4.25'
(END)

Re: [Devil-linux-develop] Kernel abort.

[Bruce S. <bw...@ar...>]

> > I'm pretty sure that PSCHED_JIFFIES was set in the .config file.
> > It worked a couple days ago on a i686-SMP compile.  All I did was
> > switch menuconfig to a i586-UP.
> > 
> > The error message kind of implies that PSCHED_JIFFIES isn't supported by
> > i586?  I just set it to PSCHED_GETTIMEOFDAY and restarted the compile.
> > We'll see if that works, in a few hours...
> 
> The PSCHED_CPU is supported only when CONFIG_X86_TSC is selected in 
> kernel so it does not work with all 586 CPUs. But there's a condition 
> catching this situation in scripts/configuration/prepare.config . If I 
> switch kernel CPU from 686 to 586 then PSCHED_CPU is changed to 
> PSCHED_JIFFIES automatically. Do you have updated all files from CVS?

I'm pretty sure that I was using JIFFIES on i686-SMP before I switched
to i586-UP.  (FWIW)

What's the best option to select for a i686 compile?  PSCHED_CPU?

> Do you have updated all files from CVS?

Yes, I updated last night, and I haven't seen any updates come across
the commit mailing list lately.

BTW, I'm actually compiling this for a friend to run on a K5-333.
Does that effect the options I'm selecting?

- BS

Re: [Devil-linux-develop] Kernel abort.

[Jaroslav K. <pe...@su...>]

On Mon, 5 Apr 2004, Bruce Smith wrote:

> > > mrproper, new LFS.
> > > Difference: I switched my config from i686 to i585 and turned off SMP.
> > > And I got the following:
> > 
> > > /data/build/tmp/linux-2.4.25/include/net/pkt_sched.h:292:2: #error PSCHED_CLOCK_SOURCE=PSCHED_CPU is not supported on this arch.
> > 
> > You have to run 'make menuconfig' to change CPU or set
> > CONFIG_PSCHED_CLOCK_SOURCE=PSCHED_JIFFIES in
> > the devil .config file.
> 
> I'm pretty sure that PSCHED_JIFFIES was set in the .config file.
> It worked a couple days ago on a i686-SMP compile.  All I did was
> switch menuconfig to a i586-UP.
> 
> The error message kind of implies that PSCHED_JIFFIES isn't supported by
> i586?  I just set it to PSCHED_GETTIMEOFDAY and restarted the compile.
> We'll see if that works, in a few hours...

The PSCHED_CPU is supported only when CONFIG_X86_TSC is selected in 
kernel so it does not work with all 586 CPUs. But there's a condition 
catching this situation in scripts/configuration/prepare.config . If I 
switch kernel CPU from 686 to 586 then PSCHED_CPU is changed to 
PSCHED_JIFFIES automatically. Do you have updated all files from CVS?

						Jaroslav

-----
Jaroslav Kysela <pe...@su...>
Linux Kernel Sound Maintainer
ALSA Project, SuSE Labs