mkefs

QNX Newsgroups (Archives) qnx.bsp
1

Hello,
I am working on a mpc5200 (icecube) and i tried to split my image files:

  • main boot image with the core icecube drivers (mkifs)
  • driver library, shared library, etc (mkefs)

I always get messages like this
unable to start “devc-serpsc”
unable to start “pci-mgt5200”

If I put all my drivers, libs and shared libs in the main boot image with
mkifs, then I have no problem.
Could it be that the mkefs in ver 6.2.1b has problems with big endian?

Thank you

otto

2

oparz <oparz@compuserve.de> wrote:

Hello,
I am working on a mpc5200 (icecube) and i tried to split my image files:

  • main boot image with the core icecube drivers (mkifs)
  • driver library, shared library, etc (mkefs)

I always get messages like this
unable to start “devc-serpsc”
unable to start “pci-mgt5200”

If I put all my drivers, libs and shared libs in the main boot image with
mkifs, then I have no problem.
Could it be that the mkefs in ver 6.2.1b has problems with big endian?

Thank you

otto

It’s more likely a pathname issue; I’d suggest working in two steps.

-create a ‘working’ IFS image, that contains all of your drivers, etc.,
along with the flash driver, and the command line to start the flash driver
-create the ‘efs’ partition, which also contains all of the stuff you want to
load from the flash filesystem

once the image has booted from the ‘good’ IFS, take a look, and see if the
stuff you put in the efs is where you expect it to be, and is accessible,
and that your $PATH environment variable contains these directories, etc.

By the way, what error number was shown with the ‘unable to start…’ messages?
If it was (2), then that means that the boot script couldn’t find the
binaries it was trying to run, likely because either the flash driver hadn’t
finished loading yet, or because the directories they reside in on the flash
weren’t in the search path. On the other hand, if the errno was (13), it may
be simply a matter of giving executable permissions to the binaries in your
flash filesystem.

\

David Green (dgreen@qnx.com)
QNX Software Systems Ltd.
http://www.qnx.com

3

thank you,
the message is (2) and I will try it tomorrow (in two steps)


“Dave Green” <dgreen@qnx.com> schrieb im Newsbeitrag
news:ceo69o$mh1$1@inn.qnx.com…

oparz <> oparz@compuserve.de> > wrote:
Hello,
I am working on a mpc5200 (icecube) and i tried to split my image files:

  • main boot image with the core icecube drivers (mkifs)
  • driver library, shared library, etc (mkefs)

I always get messages like this
unable to start “devc-serpsc”
unable to start “pci-mgt5200”

If I put all my drivers, libs and shared libs in the main boot image
with
mkifs, then I have no problem.
Could it be that the mkefs in ver 6.2.1b has problems with big endian?

Thank you

otto

It’s more likely a pathname issue; I’d suggest working in two steps.

-create a ‘working’ IFS image, that contains all of your drivers, etc.,
along with the flash driver, and the command line to start the flash
driver
-create the ‘efs’ partition, which also contains all of the stuff you want
to
load from the flash filesystem

once the image has booted from the ‘good’ IFS, take a look, and see if the
stuff you put in the efs is where you expect it to be, and is accessible,
and that your $PATH environment variable contains these directories, etc.

By the way, what error number was shown with the ‘unable to start…’
messages?
If it was (2), then that means that the boot script couldn’t find the
binaries it was trying to run, likely because either the flash driver
hadn’t
finished loading yet, or because the directories they reside in on the
flash
weren’t in the search path. On the other hand, if the errno was (13), it
may
be simply a matter of giving executable permissions to the binaries in
your
flash filesystem.

\

David Green (> dgreen@qnx.com> )
QNX Software Systems Ltd.
http://www.qnx.com

4

“Dave Green” <dgreen@qnx.com> wrote in message
news:ceo69o$mh1$1@inn.qnx.com…

oparz <> oparz@compuserve.de> > wrote:
Hello,
I am working on a mpc5200 (icecube) and i tried to split my image files:

  • main boot image with the core icecube drivers (mkifs)
  • driver library, shared library, etc (mkefs)

I always get messages like this
unable to start “devc-serpsc”
unable to start “pci-mgt5200”

If I put all my drivers, libs and shared libs in the main boot image
with
mkifs, then I have no problem.
Could it be that the mkefs in ver 6.2.1b has problems with big endian?

Thank you

otto

It’s more likely a pathname issue; I’d suggest working in two steps.

Perhaps… but mkefs in 6.2.1b does have problem with bigendian.
Specifically, it quietly ignores +bigendian attribute.
The filesystem won’t be recognized, therefore it won’t be mounted and
nothing will be found.

There was a patch for that, but AFAIK it did not go into any official
releases. I don’t know whether it went into any BSPs other than Artesyn…

– igor

5

I have set the correct path and now I have started devb-ram as an example.
devb-ram is in the directory /test:

devb-ram ram &

I’ve got the following message:

[1] 106503
$ /test/devb-ram[1]: syntax error: `(’ unexpected
[1] + Done (1)




“Igor Kovalenko” <kovalenko@comcast.net> schrieb im Newsbeitrag
news:cev3c7$291$1@inn.qnx.com…

“Dave Green” <> dgreen@qnx.com> > wrote in message
news:ceo69o$mh1$> 1@inn.qnx.com> …
oparz <> oparz@compuserve.de> > wrote:
Hello,
I am working on a mpc5200 (icecube) and i tried to split my image
files:

  • main boot image with the core icecube drivers (mkifs)
  • driver library, shared library, etc (mkefs)

I always get messages like this
unable to start “devc-serpsc”
unable to start “pci-mgt5200”

If I put all my drivers, libs and shared libs in the main boot image
with
mkifs, then I have no problem.
Could it be that the mkefs in ver 6.2.1b has problems with big endian?

Thank you

otto

It’s more likely a pathname issue; I’d suggest working in two steps.


Perhaps… but mkefs in 6.2.1b does have problem with bigendian.
Specifically, it quietly ignores +bigendian attribute.
The filesystem won’t be recognized, therefore it won’t be mounted and
nothing will be found.

There was a patch for that, but AFAIK it did not go into any official
releases. I don’t know whether it went into any BSPs other than Artesyn…

– igor

6

oparz <oparz@compuserve.de> wrote:

I have set the correct path and now I have started devb-ram as an example.
devb-ram is in the directory /test:

devb-ram ram &

I’ve got the following message:

[1] 106503
$ /test/devb-ram[1]: syntax error: `(’ unexpected
[1] + Done (1)

OK, so you’re saying that now you can see files on the
flash filesystem (assuming /test is a directory on the
efs), so you’re not getting errno (2) any more, but the
binaries still don’t execute. The error message posted
above appears similar to what usually appears when you
attempt to run a binary compiled for a different CPU
than the one it’s executing on (running an x86 binary
on a PPC, for example). Have you tried copying a ‘known
working’ binary out of /proc/boot to your flash filesystem,
and then running it there?

ex.

cp /proc/boot/pidin /test
cd /test
…/pidin



“Igor Kovalenko” <> kovalenko@comcast.net> > schrieb im Newsbeitrag
news:cev3c7$291$> 1@inn.qnx.com> …
“Dave Green” <> dgreen@qnx.com> > wrote in message
news:ceo69o$mh1$> 1@inn.qnx.com> …
oparz <> oparz@compuserve.de> > wrote:
Hello,
I am working on a mpc5200 (icecube) and i tried to split my image
files:

  • main boot image with the core icecube drivers (mkifs)
  • driver library, shared library, etc (mkefs)

I always get messages like this
unable to start “devc-serpsc”
unable to start “pci-mgt5200”

If I put all my drivers, libs and shared libs in the main boot image
with
mkifs, then I have no problem.
Could it be that the mkefs in ver 6.2.1b has problems with big endian?

Thank you

otto

It’s more likely a pathname issue; I’d suggest working in two steps.


Perhaps… but mkefs in 6.2.1b does have problem with bigendian.
Specifically, it quietly ignores +bigendian attribute.
The filesystem won’t be recognized, therefore it won’t be mounted and
nothing will be found.

There was a patch for that, but AFAIK it did not go into any official
releases. I don’t know whether it went into any BSPs other than Artesyn…

– igor

–

David Green (dgreen@qnx.com)
QNX Software Systems Ltd.
http://www.qnx.com

7

Yes, I have copied pidin from /proc/boot to my directory /test on my flash
filesystem. I’ve got no error message when executing and the result was the
same as if I started pidin from /proc/boot

otto parzhuber
munich univ of applied sciences
oparz@compuserve.de

“Dave Green” <dgreen@qnx.com> schrieb im Newsbeitrag
news:cevt6d$ko8$1@inn.qnx.com…

oparz <> oparz@compuserve.de> > wrote:
I have set the correct path and now I have started devb-ram as an
example.
devb-ram is in the directory /test:

devb-ram ram &

I’ve got the following message:

[1] 106503
$ /test/devb-ram[1]: syntax error: `(’ unexpected
[1] + Done (1)

OK, so you’re saying that now you can see files on the
flash filesystem (assuming /test is a directory on the
efs), so you’re not getting errno (2) any more, but the
binaries still don’t execute. The error message posted
above appears similar to what usually appears when you
attempt to run a binary compiled for a different CPU
than the one it’s executing on (running an x86 binary
on a PPC, for example). Have you tried copying a ‘known
working’ binary out of /proc/boot to your flash filesystem,
and then running it there?

ex.

cp /proc/boot/pidin /test
cd /test
./pidin



“Igor Kovalenko” <> kovalenko@comcast.net> > schrieb im Newsbeitrag
news:cev3c7$291$> 1@inn.qnx.com> …
“Dave Green” <> dgreen@qnx.com> > wrote in message
news:ceo69o$mh1$> 1@inn.qnx.com> …
oparz <> oparz@compuserve.de> > wrote:
Hello,
I am working on a mpc5200 (icecube) and i tried to split my image
files:

  • main boot image with the core icecube drivers (mkifs)
  • driver library, shared library, etc (mkefs)

I always get messages like this
unable to start “devc-serpsc”
unable to start “pci-mgt5200”

If I put all my drivers, libs and shared libs in the main boot
image
with
mkifs, then I have no problem.
Could it be that the mkefs in ver 6.2.1b has problems with big
endian?

Thank you

otto

It’s more likely a pathname issue; I’d suggest working in two steps.


Perhaps… but mkefs in 6.2.1b does have problem with bigendian.
Specifically, it quietly ignores +bigendian attribute.
The filesystem won’t be recognized, therefore it won’t be mounted and
nothing will be found.

There was a patch for that, but AFAIK it did not go into any official
releases. I don’t know whether it went into any BSPs other than
Artesyn…

– igor




\

David Green (> dgreen@qnx.com> )
QNX Software Systems Ltd.
http://www.qnx.com

8

oparz <oparz@compuserve.de> wrote:

Yes, I have copied pidin from /proc/boot to my directory /test on my flash
filesystem. I’ve got no error message when executing and the result was the
same as if I started pidin from /proc/boot

This tells me that your flash filesystem is working OK, so I still
suspect that the devb-ram that you tried to run might be
compiled for some other platform than PPC. How did you
build your flash filesystem? If you used an efs build file,
can you post the build file?

otto parzhuber
munich univ of applied sciences
oparz@compuserve.de

“Dave Green” <> dgreen@qnx.com> > schrieb im Newsbeitrag
news:cevt6d$ko8$> 1@inn.qnx.com> …
oparz <> oparz@compuserve.de> > wrote:
I have set the correct path and now I have started devb-ram as an
example.
devb-ram is in the directory /test:

devb-ram ram &

I’ve got the following message:

[1] 106503
$ /test/devb-ram[1]: syntax error: `(’ unexpected
[1] + Done (1)

OK, so you’re saying that now you can see files on the
flash filesystem (assuming /test is a directory on the
efs), so you’re not getting errno (2) any more, but the
binaries still don’t execute. The error message posted
above appears similar to what usually appears when you
attempt to run a binary compiled for a different CPU
than the one it’s executing on (running an x86 binary
on a PPC, for example). Have you tried copying a ‘known
working’ binary out of /proc/boot to your flash filesystem,
and then running it there?

ex.

cp /proc/boot/pidin /test
cd /test
./pidin



“Igor Kovalenko” <> kovalenko@comcast.net> > schrieb im Newsbeitrag
news:cev3c7$291$> 1@inn.qnx.com> …
“Dave Green” <> dgreen@qnx.com> > wrote in message
news:ceo69o$mh1$> 1@inn.qnx.com> …
oparz <> oparz@compuserve.de> > wrote:
Hello,
I am working on a mpc5200 (icecube) and i tried to split my image
files:

  • main boot image with the core icecube drivers (mkifs)
  • driver library, shared library, etc (mkefs)

I always get messages like this
unable to start “devc-serpsc”
unable to start “pci-mgt5200”

If I put all my drivers, libs and shared libs in the main boot
image
with
mkifs, then I have no problem.
Could it be that the mkefs in ver 6.2.1b has problems with big
endian?

Thank you

otto

It’s more likely a pathname issue; I’d suggest working in two steps.


Perhaps… but mkefs in 6.2.1b does have problem with bigendian.
Specifically, it quietly ignores +bigendian attribute.
The filesystem won’t be recognized, therefore it won’t be mounted and
nothing will be found.

There was a patch for that, but AFAIK it did not go into any official
releases. I don’t know whether it went into any BSPs other than
Artesyn…

– igor




\

David Green (> dgreen@qnx.com> )
QNX Software Systems Ltd.
http://www.qnx.com

–

David Green (dgreen@qnx.com)
QNX Software Systems Ltd.
http://www.qnx.com

9

this is the efs buildfile that was generated by momentics:

[ block_size=64k spare_blocks=1 max_size=2M min_size=2M mount=/test
filter=“flashcmp -t ucl”]

[dperms=0777 uid=0 gid=0]

[gid=0 uid=0 perms=777] /devb-ram =
C:\QNXsdk\target\qnx6\ppcbe\sbin\devb-ram

the libs are in the boot image /proc/boot

otto parzhuber
munich univ of applied sciences
oparz@compuserve.de



“Dave Green” <dgreen@qnx.com> schrieb im Newsbeitrag
news:cfb5la$345$1@inn.qnx.com…

oparz <> oparz@compuserve.de> > wrote:
Yes, I have copied pidin from /proc/boot to my directory /test on my
flash
filesystem. I’ve got no error message when executing and the result was
the
same as if I started pidin from /proc/boot

This tells me that your flash filesystem is working OK, so I still
suspect that the devb-ram that you tried to run might be
compiled for some other platform than PPC. How did you
build your flash filesystem? If you used an efs build file,
can you post the build file?

otto parzhuber
munich univ of applied sciences
oparz@compuserve.de

“Dave Green” <> dgreen@qnx.com> > schrieb im Newsbeitrag
news:cevt6d$ko8$> 1@inn.qnx.com> …
oparz <> oparz@compuserve.de> > wrote:
I have set the correct path and now I have started devb-ram as an
example.
devb-ram is in the directory /test:

devb-ram ram &

I’ve got the following message:

[1] 106503
$ /test/devb-ram[1]: syntax error: `(’ unexpected
[1] + Done (1)

OK, so you’re saying that now you can see files on the
flash filesystem (assuming /test is a directory on the
efs), so you’re not getting errno (2) any more, but the
binaries still don’t execute. The error message posted
above appears similar to what usually appears when you
attempt to run a binary compiled for a different CPU
than the one it’s executing on (running an x86 binary
on a PPC, for example). Have you tried copying a ‘known
working’ binary out of /proc/boot to your flash filesystem,
and then running it there?

ex.

cp /proc/boot/pidin /test
cd /test
./pidin



“Igor Kovalenko” <> kovalenko@comcast.net> > schrieb im Newsbeitrag
news:cev3c7$291$> 1@inn.qnx.com> …
“Dave Green” <> dgreen@qnx.com> > wrote in message
news:ceo69o$mh1$> 1@inn.qnx.com> …
oparz <> oparz@compuserve.de> > wrote:
Hello,
I am working on a mpc5200 (icecube) and i tried to split my
image
files:

  • main boot image with the core icecube drivers (mkifs)
  • driver library, shared library, etc (mkefs)

I always get messages like this
unable to start “devc-serpsc”
unable to start “pci-mgt5200”

If I put all my drivers, libs and shared libs in the main boot
image
with
mkifs, then I have no problem.
Could it be that the mkefs in ver 6.2.1b has problems with big
endian?

Thank you

otto

It’s more likely a pathname issue; I’d suggest working in two
steps.


Perhaps… but mkefs in 6.2.1b does have problem with bigendian.
Specifically, it quietly ignores +bigendian attribute.
The filesystem won’t be recognized, therefore it won’t be mounted
and
nothing will be found.

There was a patch for that, but AFAIK it did not go into any
official
releases. I don’t know whether it went into any BSPs other than
Artesyn…

– igor




\

David Green (> dgreen@qnx.com> )
QNX Software Systems Ltd.
http://www.qnx.com


\

David Green (> dgreen@qnx.com> )
QNX Software Systems Ltd.
http://www.qnx.com

10

Otto,

The “flashcmp …” filter is causing the problem. The FLASH device
drivers in the 6.2.1B supplement no longer decompress. You’re getting
cryptic error messages because you’re trying to run a compressed file.

For compression with the new FLASH filesystem, use inflator' at run time, and deflate’ at build time. If you are building on Windows, you
need the `deflate’ utility from the patch download site.

Also, something’s wrong with using deflate' as the mkefs’ filter. I
had to deflate' files before running mkefs’. Don’t know if this is
fixed in 6.3.0.


Doug

“oparz” <oparz@compuserve.de> wrote in news:cfb8ah$58o$1@inn.qnx.com:

this is the efs buildfile that was generated by momentics:

[ block_size=64k spare_blocks=1 max_size=2M min_size=2M mount=/test
filter=“flashcmp -t ucl”]

[dperms=0777 uid=0 gid=0]

[gid=0 uid=0 perms=777] /devb-ram =
C:\QNXsdk\target\qnx6\ppcbe\sbin\devb-ram

the libs are in the boot image /proc/boot

otto parzhuber
munich univ of applied sciences
oparz@compuserve.de



“Dave Green” <> dgreen@qnx.com> > schrieb im Newsbeitrag
news:cfb5la$345$> 1@inn.qnx.com> …
oparz <> oparz@compuserve.de> > wrote:
Yes, I have copied pidin from /proc/boot to my directory /test on
my
flash
filesystem. I’ve got no error message when executing and the result
was
the
same as if I started pidin from /proc/boot

This tells me that your flash filesystem is working OK, so I still
suspect that the devb-ram that you tried to run might be
compiled for some other platform than PPC. How did you
build your flash filesystem? If you used an efs build file,
can you post the build file?

otto parzhuber
munich univ of applied sciences
oparz@compuserve.de

“Dave Green” <> dgreen@qnx.com> > schrieb im Newsbeitrag
news:cevt6d$ko8$> 1@inn.qnx.com> …
oparz <> oparz@compuserve.de> > wrote:
I have set the correct path and now I have started devb-ram as
an
example.
devb-ram is in the directory /test:

devb-ram ram &

I’ve got the following message:

[1] 106503
$ /test/devb-ram[1]: syntax error: `(’ unexpected
[1] + Done (1)

OK, so you’re saying that now you can see files on the
flash filesystem (assuming /test is a directory on the
efs), so you’re not getting errno (2) any more, but the
binaries still don’t execute. The error message posted
above appears similar to what usually appears when you
attempt to run a binary compiled for a different CPU
than the one it’s executing on (running an x86 binary
on a PPC, for example). Have you tried copying a ‘known
working’ binary out of /proc/boot to your flash filesystem,
and then running it there?

ex.

cp /proc/boot/pidin /test
cd /test
./pidin



“Igor Kovalenko” <> kovalenko@comcast.net> > schrieb im Newsbeitrag
news:cev3c7$291$> 1@inn.qnx.com> …
“Dave Green” <> dgreen@qnx.com> > wrote in message
news:ceo69o$mh1$> 1@inn.qnx.com> …
oparz <> oparz@compuserve.de> > wrote:
Hello,
I am working on a mpc5200 (icecube) and i tried to split my
image
files:

  • main boot image with the core icecube drivers (mkifs)
  • driver library, shared library, etc (mkefs)

I always get messages like this
unable to start “devc-serpsc”
unable to start “pci-mgt5200”

If I put all my drivers, libs and shared libs in the main
boot
image
with
mkifs, then I have no problem.
Could it be that the mkefs in ver 6.2.1b has problems with
big
endian?

Thank you

otto

It’s more likely a pathname issue; I’d suggest working in two
steps.


Perhaps… but mkefs in 6.2.1b does have problem with
bigendian. Specifically, it quietly ignores +bigendian
attribute. The filesystem won’t be recognized, therefore it
won’t be mounted
and
nothing will be found.

There was a patch for that, but AFAIK it did not go into any
official
releases. I don’t know whether it went into any BSPs other than
Artesyn…

– igor




\

David Green (> dgreen@qnx.com> )
QNX Software Systems Ltd.
http://www.qnx.com


\

David Green (> dgreen@qnx.com> )
QNX Software Systems Ltd.
http://www.qnx.com