There has been a posting recently.
Just search for:
"[uClinux-dev] Large Memory allocation with uclinux, ColdFire m5329"
Hope this helps,
Markus
Narasimha, Swaroop(IE10) schrieb:
Hello all
I am trying to run an application on uClinux ported on coldfire 5372
processor. Kernel version is
Zitat von rwarner <[EMAIL PROTECTED]>:
If everyone prefers the original uclinux i'll kill the uC-Linux one.
I prefer #uclinux as this is much more intuitive than the hyphenated version.
Best Regards,
Markus
___
uClinux-dev mailing list
uClinux-d
rwarner schrieb:
Not sure this will be used, guess i'll wait to see if it's useful. Have
setup an irc on freenode #uC-Linux.
*thumbsup*
Very good idea !!!
Cheers,
Markus
___
uClinux-dev mailing list
uClinux-dev@uclinux.org
http://mailman.uclinux.
Dear uClinux Users/Developers,
for a fast and performant data exchange between kernel and user space
I would like to memory map a junk of kernel memory to a user space
application. I have a simple character device and as far as I know I
can achieve this behaviour by implementing the mmap()
Martin Euredjian schrieb:
Can you clarify this statement. Do you mean to say that Coldfire is limited
by hardware to preclude setting MAX_ORDER above 11 or that some kernel work
would be required in order to go past this limit?
Well, its limited by the current implementation in the m68knommu L
Martin Euredjian schrieb:
As far as I know this is architecture dependent.
Coldfire 5373 or 5329?
Check out the kernel configuration option CONFIG_FORCE_MAX_ZONEORDER. It
can be modified on an arch-by-arch basis. If it is not defined
/include/linux/mmzone.h set's MAX_ORDER to a default of 1
Martin Euredjian schrieb:
I was told that uClinux applications can't allocate memory in chuncks larger
then 1 MiB. Is this true? I have an application that requires significanly
more than that (10 to 20 MiB) for image processing.
As far as I know this is architecture dependent.
Cheers,
Marku
Hi,
if you have arm-elf-tools installed then you should maybe try to execute
make CROSS_COMPILE=arm-elf-
Cheers,
Markus
Jeff Ramin schrieb:
Hi folks. uClinux newbie here...
I'm trying to get a uClinux emulator running on my linux machine.
I've downloaded ARMulator, and I'm able to run the l
Hi Greg,
Greg Ungerer schrieb:
The kernel can do basic profiling already. Supply a "profile=4" boot
option to it. And read out the profile information use the "readprofile"
app.
yes I know about this option. I tried with "profile=2" but readprofile
gave me only very little information. Will t
Hi Gavin,
Gavin Lambert schrieb:
While it's generally considered to be "bad style", in fact *because* of the
lack of MMU support you can usually just pass a raw pointer from kernel to
user space and the user code can just read (or even write) it directly.
yeahthought about this as well but
Mihir Sevak schrieb:
how about using netlink - socket approach ?
Hmmm.sounds quite interestingwill go through some documentation.
So you think with netlink it is possible to communicate data between
user and kernel space via a normal socket api interface?
Thanks for the hint and bes
Hi Jun,
Jun Sun schrieb:
> This might be a dumb question, but have we considered moving to gcc4.x
> for uclinux? Any pro's and con's? What would be the effort to do so?
Well, I am successfully using Codesourcery's toolchain to build uClinux
for Coldfire (m68knommu) target. The gcc is based on g
Dear uClinux Developers/Users,
I am looking for a decent way to do linux kernel profiling. I am
running uClinux-dist-20070130 (linux-2.6.19uc1) without any patches on
a Coldfire target. Any suggestions?
Best regards,
Markus Franke
___
uClinux
Dear uClinux Developers/Users,
I have a character device driver which needs to pass data to a user
space application with a high frequency. I am doing this by using the
copy_to_user() system call in the read() operation of the character
device driver. I discovered that this copy operation c
Dear uClinux Developers/Users,
I developed a linux character device driver in order to access an
external memory chip sitting on the flexbus. The data transfers are
done via the Multichannel DMA Controller.
From time to time (it's not really deterministic) my board hangs up.
Via BDM interfa
Dear uClinux developers/users,
I am working with uClinux on a Coldfire based board and I would like
to receive correct cpu usage information. I am not sure whether I can
rely on the information provided by "top" which is included in
busybox. If I sum up the cpu usage percentage of each task
Hi Philippe,
Philippe LEGROS schrieb:
I have tried to reserve it with malloc which returns NULL.
Have a look at the kernel function get_free_pages().
Best Regards,
Markus
___
uClinux-dev mailing list
uClinux-dev@uclinux.org
http://mailman.uclinux.or
Dear all,
I am wondering about the output of /proc//stat under
uClinux-dist_20070130. Column 18 resp. 19 should represent the
priority and nice values of the corresponding process.
When I start a new process, let's say a shell, the priority is set to
17 and the nice value to 0. However, som
Title: Re: [uClinux-dev] nested interrupts on m68knommu(Coldfire)
Hi Greg,
Zitat von Greg Ungerer <[EMAIL PROTECTED]>:
> No, it doesn't.
> Interrupts in uClinux are either enabled, or disabled.
> There is no use of a priority based scheme.
Thank you very much for this information.
B
Hi Greg,
Zitat von Greg Ungerer <[EMAIL PROTECTED]>:
No, it doesn't.
Interrupts in uClinux are either enabled, or disabled.
There is no use of a priority based scheme.
Thank you very much for this information.
Best regards,
Markus
___
uClinux-de
9-uc1 from
uClinux-dist-20070130.
Best regards,
Markus Franke
___
uClinux-dev mailing list
uClinux-dev@uclinux.org
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by uclinux-dev@uclinux.org
To unsubscribe
Hi Harry,
Harry Gunnarsson schrieb:
Markus,
So you are saying that including CodeSourcery's gdbserver (neither did I
know that it came with the toolchain) into the root file system and
debugging with the CodeSourcery compatible m68k-uclinux-gdb over ethernet
works out of the box without modifica
source, and built it myself.
Good luck and let us know if you have any problems
--
KB
On Fri, 2008-04-11 at 15:51 +0200, Markus Franke wrote:
Hi,
thank you very muchwill try to adapt it to my needs. By the way, I
am using gdbserver which comes along uClinux-dist-20070130 which is
pretty old I
27;t accept this as much more than a hint patch ... its not submitted
yet - but this is all what I did to extend the SH kernel and gdbserver.
--
Hope it helps.
Kieran Bingham
On Fri, 2008-04-11 at 14:39 +0200, Markus Franke wrote:
Zitat von Kieran Bingham <[EMAIL PROTECTED]>:
> I'
Zitat von Kieran Bingham <[EMAIL PROTECTED]>:
I've been talking to one of the GDB dev's on IRC, and he's pointed me
towards gdb-6.7.1/gdb/gdbserver/linux-low.c
the part that does the offsets is in linux_read_offsets(). If it doesn't
have the correct values defined for PTrace to obtain them, then
ng.c:212
212 {
(gdb)
Continuing.
Program received signal SIGTRAP, Trace/breakpoint trap.
0x03c7063e in main (argc=2, argv=0x3ccbf78) at ping.c:212
212 {
---snap---
Zitat von Markus Franke <[EMAIL PROTECTED]>:
Hi Kieran,
Zitat von Kieran Bingham <[EMAIL PROTECTED]>:
reading
Hi Kieran,
Zitat von Kieran Bingham <[EMAIL PROTECTED]>:
reading those now ... looks like you were getting identical to me ...
gdb assumes everything loads from 0x0
I've got round that by using the add-symbol-file
well, this doesn't help in my case. Besides, I provide the symbol file
alrea
Zitat von Markus Franke <[EMAIL PROTECTED]>:
HmmmI debugged arch/m68knommu/ptrace.c:arch_ptrace. As soon as I
connect from the host-gdb (by target remote :3000) to the target,
arch_ptrace is getting called with argument "addr == 56" and "request
== 3 (PTRACE_P
(struct user) - 3)
break;
---snap---
The AND-mask with 3 is due to address alignment I guess. As address
equals to 56 the "break" gets called and that's it. I guess there is
something wrong with this 56, isn't it?
Any comments on this are appreciated. Wi
world as well as withshipped applications
like ping or ftp.
With best regards,
Markus Franke
___
uClinux-dev mailing list
uClinux-dev@uclinux.org
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by uclinux-dev@uclinux.o
Zitat von David McCullough <[EMAIL PROTECTED]>:
Back when I wrote this:
http://www.ucdot.org/article.pl?sid=03/01/30/0548223
and the m68k-elf toolchain came with m68k-bdm-elf-gdb included, it was
working to a resonable extent.
I know this document very well. I tried the "m68k-bdm-e
Hi Chris,
Chris Doré schrieb:
Do you have any other m68k*gdb* execs?
Well, so far I tried the gdb from Codesourcery (based on gdb-6.6) as
well as a self compiled and bdm-patched gdb-6.3both without success
and showing the same problems. I will give m68k-bdm-elf-gdb a try
tomorrow. ;-)
feeling that
it is just a small thing which I am missunderstanding. ;-)
Thanks for any comments on this and with best regards,
Markus Franke
___
uClinux-dev mailing list
uClinux-dev@uclinux.org
http://mailman.uclinux.org/mailman/listinfo/uc
Hi Tom,
you can try the linux kernel boot option "mem=" e.g. use mem=30M in
your case. This will give uClinux only 30 MB of RAM. Another
possibility is to reduce the size of the ZONE_NORMAL memory zone in
your target specific memory configuration.
Hope this helps,
Markus
Zitat von "Curr
Zitat von David McCullough <[EMAIL PROTECTED]>:
Thats a possibility, check the code, a lot of arches reserve page 0
for trapping illegal access amoungst other things,
Well, this might be possible. I will do a
free_bootmem(PAGE_SIZE, 0x00FF - PAGE_SIZE);
to be on the safe side.
Thanks a
Hi David,
Zitat von David McCullough <[EMAIL PROTECTED]>:
Jivin Markus Franke lays it down ...
Yes, despite the code you mention below, the mem_map will get
allocated at the start of memory, the only was around this would be
to look at discontiguous memory regions, or to see if y
it perhaps have a look at:
linux-2.6.x/arch/frv/kernel/setup.c
linux-2.6.x/arch/frv/mb93090-mb00/pci-dma-nommu.c
and how it manages some consistent dma memory,
So far only kmalloc is being used. Thanks anyway for the hints above.
With best regards,
Ma
Hi David,
Zitat von David McCullough <[EMAIL PROTECTED]>:
Some likely ones to consider are:
arch/mips/mm/init.c: paging_init
arch/x86/kernel/setup_32.c: zone_sizes_init
They all seem to call:
free_area_init_nodes(max_zone_pfns);
Though you may need to make su
Hi David,
David McCullough schrieb:
bootmem perhaps ? I can't recall exactly but I think you have added a zone
without adding or freeing the pages appropriately.
Exactly. It seems that the 16 MB DMA zone is reservered correctly but
the kernel doesn't recognize this memory to be "free". How
est regards,
Markus Franke
___
uClinux-dev mailing list
uClinux-dev@uclinux.org
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by uclinux-dev@uclinux.org
To unsubscribe see:
http://mailman.uclinux.org/mailman/options/uclinux-dev
memory (e.g. @ 0x0100) and keep the ZONE_DMA zone @ 0x0 -
0x00ff?
With best regards,
Markus Franke
___
uClinux-dev mailing list
uClinux-dev@uclinux.org
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by u
Hi Greg,
Zitat von Greg Ungerer <[EMAIL PROTECTED]>:
Hi Markus,
Markus Franke wrote:
well, as far as I understand the generic DMA API in the Linux
kernel should provide an abstraction layer to driver developers and
tries to hide all the specifics about DMA implementation. That
m
set_dma_count, set_dma_address, etc. in arch/m68knommu/dma.c for my
MCF548x.
So what is the prefered way now or are both ways equal and it's just a
matter of taste?
With best regards,
Markus Franke
___
uClinux-dev mailing list
uClinux-dev@uc
Hi Gavin,
Zitat von Gavin Lambert <[EMAIL PROTECTED]>:
... and then you reply to something I quoted instead of what I said?
sorry for this. Should have read the whole thread more carefully. :-)
That's basically what I said (except in less detail); but since the OP was
talking about accessing
Hi Gavin,
Gavin Lambert schrieb:
If I only wanted to use a "chip" driver, would selecting an option in
the group labeled as "I2C bit-banging interfaces" be sufficient, or
would I need to add something else? Likewise, only using the chardev
driver, would I only need to select the "I2C Device int
ROTECTED]>:
Hi markus,
Markus Franke wrote:
Zitat von Greg Ungerer <[EMAIL PROTECTED]>:
As a general rule you want to follow the way Linux does things.
This is especially true for modern versions of the kernel (2.6.x),
and especially if you want the code to be acceptable for main
line inclus
Hi Greg,
Zitat von Greg Ungerer <[EMAIL PROTECTED]>:
As a general rule you want to follow the way Linux does things.
This is especially true for modern versions of the kernel (2.6.x),
and especially if you want the code to be acceptable for main
line inclusion.
apart from , is there any furthe
with this
generic API. I am wondering whether it might be useful if this would
be done for MCDMAAPI as well? Which one would be the prefered way for
the community? Using Freescale's MCDMAAPI or the generic Linux DMA API
or maybe both API's working together?
Regards,
Mar
gets never built. Is there any reason for this???
Regards,
Markus Franke
BTW: I already played a bit around with gdbserver. Connecting to the
target works but I when trying to set a breakpoint on main() I get
"Cannot access memory at address 0x1c". Don't know what this
Hi,
Zitat von Michael Schnell <[EMAIL PROTECTED]>:
If reading takes longer than expected, something in the hardware is
wrong. I don't suppose that the software introduces any waiting loops
on reading.
I just realised that this is another issue. As soon as I use a
printf() parametrized with
Dear Claude,
thanks for the really detailed considerations.
I don't know if you have considered this issue.
Yeah, that's indeed an issue and I am taking care for this in my dBUG
code. You are rightactually this also has to be done in the linux
MTD drivers. However, I wrote a small use
Hi Claude,
Claude schrieb:
I've looked at your mtd map and BANKSIZE is set to 4. Is your flash
really a 32 bit bus width device?
Well, I am using 2 identical flash devices (each one 16-bit). The lower
16-bits of the data-bus go to flash 1, the higher 16-bits are wired to
flash 2.
With best
Zitat von Bou <[EMAIL PROTECTED]>:
Could it be an endianess problem (create a jffs partition for big
endian while in little endian platform )??
Tried already both variants without success. Even if it is an
endianess problem, AFAIK the JFFS2 driver is capable of noticing this.
Furthermore,
Hi,
thanks for the detailed reply.
Zitat von Claude <[EMAIL PROTECTED]>:
The difference between mtd1 and mtdblock1 is that the first is a
character device whereas the second is a block device.
The mtd device is used by the flashtools such as eraseall, and
mkfs.jffs2. You can do things like:
Zitat von Detlef Vollmann <[EMAIL PROTECTED]>:
Markus Franke wrote:
What's the output of 'ls -l /dev/mtd*'?
/> ls -l /dev/mtd*
crw---1 00 90, 0 Jan 1 1970 /dev/mtd0
crw---1 00 90, 2 Jan 1 1970 /dev/mtd1
cr
Zitat von ADP <[EMAIL PROTECTED]>:
Do you have any particular flash topology, e.g. mixe nor and nand devices?
No there is just one NOR-Flash (32MB).
My romfs is mapped on /dev/mtd2. When I try to mount this partition in
the same way with:
mount -t jffs2 /dev/mtd2 /mnt
of course I get erro
Hi,
Zitat von Chris Doré <[EMAIL PROTECTED]>:
Does `cat /proc/mtd` look correct, in terms of addresses and such?
this is the /proc/mtd output. I think it looks pretty ok:
---snip---
dev:size erasesize name
mtd0: 0008 0004 "dBUG (512K)"
mtd1: 01f8 0004 "User FS (32256K)"
Hi,
Zitat von Chris Doré <[EMAIL PROTECTED]>:
Does `cat /proc/mtd` look correct, in terms of addresses and such?
this is the /proc/mtd output. I think it looks pretty ok:
---snip---
dev:size erasesize name
mtd0: 0008 0004 "dBUG (512K)"
mtd1: 01f8 0004 "User FS (32256K)"
First of all sorry for all the messages on the list. There was a
problem with my mail program.
Zitat von ADP <[EMAIL PROTECTED]>:
Use:
mount -t jffs2 /dev/mtdblock1 /mnt
Same problem. I get messages like this:
---snip---
jffs2_scan_eraseblock(): Magic bitmask 0x1985 not found at 0x:
Zitat von Gavin Lambert <[EMAIL PROTECTED]>:
It's been a while (and I'm using an old build), so I could be mistaken, but
last time I checked the JFFS2 driver did something weird with its mount
argument and parsed it directly rather than using it as a filesystem
location. Consequently you have t
Zitat von Gavin Lambert <[EMAIL PROTECTED]>:
It's been a while (and I'm using an old build), so I could be mistaken, but
last time I checked the JFFS2 driver did something weird with its mount
argument and parsed it directly rather than using it as a filesystem
location. Consequently you have t
Zitat von Gavin Lambert <[EMAIL PROTECTED]>:
It's been a while (and I'm using an old build), so I could be mistaken, but
last time I checked the JFFS2 driver did something weird with its mount
argument and parsed it directly rather than using it as a filesystem
location. Consequently you have t
- Weitergeleitete Nachricht von
[EMAIL PROTECTED] -
Datum: Thu, 24 Jan 2008 08:54:48 +0100
Von: Markus Franke <[EMAIL PROTECTED]>
Antwort an: Markus Franke <[EMAIL PROTECTED]>
Betreff: RE: [uClinux-dev] mounting JFFS2 filesystems
An: Gavin Lam
Dear uClinux Developers/Users,
I am using uClinux-dist-20070131 without any patches and have problems
when mounting JFFS2 filesystems. I have created 3
partitions(/dev/mtd0-2) on my flash device. I have successfully
created a JFFS2 filesystem image and flashed it (using Bootloader) to
the
Allon Stern schrieb:
So, I'm just getting started with uClinux on coldfire; I've got the
20061214 toolchain snapshot, the most recent on the uclinux.org website.
Where do I go from here to get later updates/patches?
You should use the toolchain from:
http://www.codesourcery.com/gnu_toolchains
Dear uClinux developers,
as far as I can see from
"uClinux-dist/arch/m68knommu/platform"-directory, the Coldfire Family
shares a lot of code with MCF5307. Bootcode/IRQ stuff/Timers etc. is
implemented there and shared among all the other MCF5xxx platforms. I
am wondering whether there mig
Hi Stuart,
Zitat von Stuart Hughes <[EMAIL PROTECTED]>:
Once you've done this you have to figure out how to add a new kernel
into uClinux-dist. The other thing to remember as that this Coldfire
has an MMU so make sure you set that appropriately so you get fork (not
vfork) etc.
that's exactly
Hi Stuart,
Zitat von Stuart Hughes <[EMAIL PROTECTED]>:
You can get a BSP here (this for mcf547x_8x):
http://www.freescale.com/webapp/sps/site/overview.jsp?nodeId=0127260061033202A9
or by going here:
http://www.bitshrine.org/
I know this BSP already.
I believe the kernel is 2.6.10. To the
n only see support for MCF5407 cpus which
are not working together with the M5474EVB. I am using a M5474EVB
together with a MCF5484 Coldfire processor.
Any hints are appreciated.
Thanks and regards,
Markus Franke
___
uClinux-dev mailing list
uClinux-dev@u
merge of uClinux and linux-2.6.
Thanks for any help on this,
Markus Franke
___
uClinux-dev mailing list
uClinux-dev@uclinux.org
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by uclinux-dev@uclinux.org
To unsubscribe
70 matches
Mail list logo