nguyen tritai | 1 Aug 12:07 2008
Picon

Canyonlands AMCC 460EPX issues with VGA ATI Rage 128 VR

Hi folks,

I'm trying to use an ATI VGA card for video booting Canyonlands AMCC 460EPX
board.

I pacthed x86 emulator codes from Scitechsoft to initialize ATI BIOS ( I
succeeded in doing that with Sequoia AMCC 440EPX before)
but the kernel crashes and reboots after loading.

I think that it's the issue from architecture of canyonlands. Canyonlands
resources is 64 bit whereas  ATI Rage 128 is 32 bit and the code x86emulator
is for 32 bit.
Debugging the problem, it comes from the ioremap :

    if ((BIOSImagePhys = PCI_findBIOSAddr(pcidev,&BIOSImageBAR)) == 0)
        return NULL;
    if ((BIOSImage = ioremap(BIOSImagePhys,MAX_BIOSLEN)) == NULL)
        return NULL;

Did anyone see this problem before ?
I wonder if I can configure canyonlands board to use 32 bit resources or
not?

Please help me.

Thanks and best regards,

TaiNguyen

(Continue reading)

Sai Amruta | 1 Aug 13:02 2008
Picon

Linux 2.6 kernel release

Hi,
Currently i am using 2.6.19 version. I would like to upgrade to latest
version in my new project.
What is the latest stable 2.6 kernel version released by ELDK?

--Amru

Wolfgang Denk | 1 Aug 14:36 2008
Picon
Picon

Linux 2.6 kernel release

In message <dd2f3b320808010402qf5ea333t53d62d996c96abde at mail.gmail.com> you wrote:
> 
> Currently i am using 2.6.19 version. I would like to upgrade to latest
> version in my new project.
> What is the latest stable 2.6 kernel version released by ELDK?

DENX-v2.6.25.13 - see
http://git.denx.de/?p=linux-2.6-denx.git;a=commit;h=82f76148b7a14c3df8e6bdd41f362350b1f85b1e

Best regards,

Wolfgang Denk

--

-- 
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
What the gods would destroy they first submit to  an  IEEE  standards
committee.

Giuseppe Modugno | 4 Aug 15:10 2008
Picon

kill an entire process group

Dear all,

I'm using ELDK4.1 for arm platform and I'd like to kill a shell 
script with all the child processes the script had launched.

For example, if my script example.sh is:
---
#!/bin/sh
sleep 100
---
I want to kill the processes example.sh and sleep launched by 
example.sh.
The usual kill command kills only example.sh, parent process.

I wrote a small C program killpg that uses kill() function to send 
the same signal to all the child processes, by passing -pid (a 
negative number) to the function.

When I use the NFS development system, my killpg program works well 
and I can kill the parent and all the child processes.

When I copy killpg executable to my target system (based on SELF 
ramdisk), kill() function returns with an error, ESRCH (pid argument 
doesn't correspond to a real pid).

libc libraries are the same in my target and nfs development systems. 
What could cause the different behaviour?

Can someone suggest me a better method to kill parent and all the 
child processes?
(Continue reading)

Giuseppe Modugno | 4 Aug 17:25 2008
Picon

Kill strange behaviour in a script

Dear all,

I have two shell scripts, example.sh and flash.sh:
---
root:/tmpfs/tmp> cat example.sh
#!/bin/sh

while true; do
  read X
  ./flash.sh &
  FLASHPID=$!
  sleep 5
  kill $FLASHPID
done
---
root:/tmpfs/tmp> cat flash.sh
#!/bin/sh

while true; do
  echo "ON"
  sleep 1
  echo "OFF"
  sleep 1
done
---
example.sh waits for an input from the keyboard and launches the 
second script flash.sh. After some seconds, example.sh kills the 
second script that it launched.

Here what happens on my ARM system based on ELDK4.1 SELF ramdisk:
(Continue reading)

Wolfgang Denk | 4 Aug 22:36 2008
Picon
Picon

kill an entire process group

In message <48971BCF.8705.1721342 at g.modugno.elettronika.it> you wrote:
> 
> I wrote a small C program killpg that uses kill() function to send 
> the same signal to all the child processes, by passing -pid (a 
> negative number) to the function.

That means you send the signal to all processes in the process group.
This implies that there is a process group in the first place.

> When I use the NFS development system, my killpg program works well 
> and I can kill the parent and all the child processes.

In the NFS envrionment you use standard SYSV init, standard login and
standard bash.

> When I copy killpg executable to my target system (based on SELF 
> ramdisk), kill() function returns with an error, ESRCH (pid argument 
> doesn't correspond to a real pid).

In the SELF envrionment you use BusyBox init, login, and shell.

> libc libraries are the same in my target and nfs development systems. 
> What could cause the different behaviour?

The shell. The Busybox shell obviously does not create a process group
at all.

> Can someone suggest me a better method to kill parent and all the 
> child processes?

(Continue reading)

Wolfgang Denk | 4 Aug 22:40 2008
Picon
Picon

Kill strange behaviour in a script

In message <48973B9D.25371.1EE4F55 at g.modugno.elettronika.it> you wrote:
> 
> I have two shell scripts, example.sh and flash.sh:
...
> What happens? I tried both the scripts in Ubuntu on my PC and they 
> works as expected.

Keep in mind that the busybox shell is very small compared against for
example bash. You pay for this advantage in terms of memory footprint
with a lot of restrictions, especially in the fancier areas of shell
programming like support for functions, aliases, signal handling, etc.

If you want to do fancy shell stuff, the use a real shell. Yes,  bash
*is*  ten times as big as any of the Busybox shells, but you've asked
for all these features :-)

Best regards,

Wolfgang Denk

--

-- 
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Testing can show the presense of bugs, but not their absence.
                                                   -- Edsger Dijkstra

Jeffrey Haemer | 4 Aug 18:01 2008
Picon

Kill strange behaviour in a script

Giuseppe,

I just tested the possibility that you're seeing a quirk of busybox or the
busybox shell (ash).  Here's the experiment:

# install busybox and link it, locally, to "sh"
sudo apt-get install busybox
ln /usr/bin/busybox ~/Desktop/sh

# Now change both scripts to use the busybox shell
perl -i -pe "s(/bin/sh)($HOME/Desktop/sh)" example.sh flash.sh

# and execute them
./example.sh

Unfortunately, everything still works correctly.  If it's a BusyBox shell
bug, the version I just tried (v1.1.3) doesn't have it.

On Mon, Aug 4, 2008 at 9:25 AM, Giuseppe Modugno
<g.modugno at elettronika.it>wrote:

> Dear all,
>
> I have two shell scripts, example.sh and flash.sh:
> ---
> root:/tmpfs/tmp> cat example.sh
> #!/bin/sh
>
> while true; do
>  read X
(Continue reading)

Wolfgang Denk | 4 Aug 23:11 2008
Picon
Picon

Kill strange behaviour in a script

In message <5808d4420808040901t45d3428ereb1d98a35abea10f at mail.gmail.com> you wrote:
>
> Unfortunately, everything still works correctly.  If it's a BusyBox shell
> bug, the version I just tried (v1.1.3) doesn't have it.

ELDK 4.2 uses busybox-1.7.1.

> On Mon, Aug 4, 2008 at 9:25 AM, Giuseppe Modugno
> <g.modugno at elettronika.it>wrote:
> 
...

Please do not top post - full quote (see
http://en.wikipedia.org/wiki/Top_post and
http://www.netmeister.org/news/learn2quote.html).

Best regards,

Wolfgang Denk

--

-- 
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
Sometimes a man will tell his bartender things he'll never tell his doctor.
	-- Dr. Phillip Boyce, "The Menagerie" ("The Cage"),
	   stardate unknown.

Giuseppe Modugno | 5 Aug 09:15 2008
Picon

kill an entire process group

On 4 Aug 2008 at 22:36, Wolfgang Denk wrote:
> > libc libraries are the same in my target and nfs development
> > systems. What could cause the different behaviour?
> The shell. The Busybox shell obviously does not create a process
> group at all.

Oh, I understand. I was thinking that the spawn of a process 
automatically creates a process group (the parent process and the
child process). From your words I understand it's up to the shell to
create a process group when it spawns another process. Msh shell
doesn't create process groups.

> > Can someone suggest me a better method to kill parent and all the
> > child processes?
> If you don't want to create  the  process  group  and  the 
> processes yoruself, then you probably might ant to use a real shell
> (bash).

What do you mean with "create the process group and the processes
yourself"? Should I write a C program that creates a process and the
process group instead of using a script shell?

I think another way to do this is to search for the same parent ID in
the list of all processes. ps command has an option to select only 
the
processes with a certain parent PID and I kill all these processes.

But in busybox ps doesn't recognize this option, so I need a full-
features ps that depends on libproc. But I think ps+libproc is better
(in size terms) than bash.
(Continue reading)


Gmane