headers
Boris Itkis | 1 May 2002 16:47

Request for suggestion on a user space map-er for memory IO space

I am working on a clean way of mapping IO memory space for the ARM (Other
processor also but I am testing this on the arm). On the x86 platform there
is ioperm which allows supper-user apps to access IO space. This is not
available on the ARM since there is only one address space (On the ARM all
IO is in memory space). An equivalent system is needed for the ARM. I have
heard there are patches that modify the mmap code in the kernel so user
space apps can map IO memory space into the apps memory space. This approach
is not clean and unnecessarily complicates the sys_mmap code.

My solution is to have a file which is the IO memory space (I chose a file
instead of a device since Linux 2.4 is running low on major numbers). The
file is part of the proc file system. I have a proc entry that allows one to
list and optionally add new IO areas ( I will later add a way to optionally
delete areas/files). By having memory area (a file) that is part of the file
system a user app can mmap it as any other file in the file system. And by
having module add itself to the proc file-system all the changes are
confined to one file.

My question's:

1. I am putting all my files into root of the proc file system. Any better
suggestions?

2. What is the best name for the area/file list-er? I am currently using
"UserIoRegions".

3. Where should I but this code into the Linux directory tree? (It can be a
module or part of the kernel)

4. A related question is where to put it into the Linux config tree? (Where
(Continue reading)

Permalink | Reply | 1 comment |
headers
Russell King - ARM Linux | 1 May 2002 17:15
Picon

Re: Request for suggestion on a user space map-er for memory IO space

On Wed, May 01, 2002 at 07:47:11AM -0700, Boris Itkis wrote:
> I am working on a clean way of mapping IO memory space for the ARM (Other
> processor also but I am testing this on the arm). On the x86 platform there
> is ioperm which allows supper-user apps to access IO space. This is not
> available on the ARM since there is only one address space (On the ARM all
> IO is in memory space). An equivalent system is needed for the ARM. I have
> heard there are patches that modify the mmap code in the kernel so user
> space apps can map IO memory space into the apps memory space. This approach
> is not clean and unnecessarily complicates the sys_mmap code.

Is there something wrong with the following simple code that will map
any physical address (including memory and IO space):

void *map_memory(unsigned long phys_addr, size_t size)
{
	void *p;
	int fd;

	fd = open("/dev/mem", O_RDWR | O_SYNC);
	if (fd == -1)
		return NULL;

	p = mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, phys_addr);
	close(fd);
	if (p == (void *)-1)
		return NULL;

	return p;
}
(Continue reading)

Permalink | Reply | 0 comments |
headers
jan rinze | 2 May 2002 00:02
Picon
Favicon

cyberpro framegrabber sticks to pink...

Hi there,

the kernel 2.4.18-rmk5 and 2.4.18-rmk5-nw1 doesn't work with my videocapture
program.

machine: Netwinder rev 5. 2Mb video mem, 64 Mb SODIMM, 20 Gb IDE.

all was well with 2.4.17, I use this winder as a TV set ...
The cyberpro.c file in drivers/media/video has been altered a lot since
2.4.18.
Is there a reason for that ?

The offset of the overlay seems incorrect and the transparancy colour seems
to be converted to pink...
(the overlay starts 20% from the left of the screen and presumably the
buffer is filled with a value that translates in YuYv to pink)
No grabbing is performed.

Has this version been tested by anyone else ?

Jan Rinze.

-------------------------------------------------------------------
Subscription options: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel
FAQ/Ettiquette:       http://www.arm.linux.org.uk/armlinux/mailinglists.php
Permalink | Reply | 1 comment |
headers
Ralph Siemsen | 2 May 2002 00:52

Re: cyberpro framegrabber sticks to pink...

On Thu, May 02, 2002 at 12:02:19AM +0200, jan rinze wrote:
> 
> the kernel 2.4.18-rmk5 and 2.4.18-rmk5-nw1 doesn't work with my videocapture
> program.

Hmm, nice to see that _somebody_ actually downloads my little patches!
As for the video problem, I hadn't noticed, and will give a try with
the "standard" capture program...

-Ralph

-------------------------------------------------------------------
Subscription options: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel
FAQ/Ettiquette:       http://www.arm.linux.org.uk/armlinux/mailinglists.php
Permalink | Reply | 0 comments |
headers
rahnumai bazm | 2 May 2002 09:10
Picon
Favicon

Regarding network driver and frame buffer driver

Hi Everybody,
     We are trying to port arm-linux to one of the
assabet like board. Although we are through with it we
are facing a little problem. Till date we are using
kernel version 2.4.16. We compiled the XFbdev server
for X. But the problem is that when we run XFbdev, and
any X application we run everything comes in black and
white, and it does not display the color. Where as we
are getting the color for our boot logo during boot up
time.
       Some time back we found on the mailing list
that there was some problem with the touch screen part
in kernel version 2.4.16, it used to freeze. We also
faced this problem. It was mentioned that in 2.4.17
this problem is solved. so we downloaded the kernel
version 2.4.17 and applied the appropriate patch for
it, (since we were only interested in the touch screen
part we replace the whole of misc directory of 2.4.16
with that of 2.4.17 and recompiled the 2.4.16 kernel)
but when we tested it we got a kernel panic right
after the network initialization. 
   If we disable the frame buffer support everything
works fine. Can any body tells what could be the
problem.

With Regards

________________________________________________________________________
For live cricket scores download  Yahoo! Score Tracker
 at: http://in.sports.yahoo.com/cricket/tracker.html
(Continue reading)

Permalink | Reply | 0 comments |
headers
Russell King - ARM Linux | 2 May 2002 14:00
Picon

Re: *** URGENT *** Admin: Linux-arm / Linux-arm-kernel: Change of status

*FINAL* reminder.  You have until Midnight BST (11 hours away) to follow
the enclosed instructions.  By looks of things, I'll be unsubscribing
1146 addresses tonight.

We're at 161 replies of 1307 subscribers currently.  That's Piss Poor.
Don't whinge when you can't post messages to the list because you haven't
followed these instructions.

On Sat, Apr 27, 2002 at 08:40:10AM +0100, Russell King - ARM Linux wrote:
> A reminder.
> 
> There are over 1300 email addresses subscribed to these lists.  92 people
> have thus far replied.  The terms are: you either agree and indicate your
> agreement via the method described below, or you unsubscribe.  If you do
> neither, you will be unsubscribed on the 2nd May and *will not* be able
> to post messages to these lists.
> 
> On Thu, Apr 25, 2002 at 12:17:50PM +0100, Russell King - ARM Linux wrote:
> > Notification of change of posting requirements.
> > -----------------------------------------------
> > 
> > Following the Data Protection Act problems earlier with exim-users, it
> > has been decided that, in order to protect the list from possible legal
> > problems, it has been decided that a change in the method that these
> > lists operate.
> > 
> > Up until now, we have accepted emails from any address, even
> > unsubscribed addresses.
> > 
> > However, from this point on, you may only use your subscribed email
(Continue reading)

Permalink | Reply | 0 comments |
headers
Erik Andersen | 2 May 2002 14:22
Favicon

Re: *** URGENT *** Admin: Linux-arm / Linux-arm-kernel: Change of status

On Thu May 02, 2002 at 01:00:12PM +0100, Russell King - ARM Linux wrote:
> *FINAL* reminder.  You have until Midnight BST (11 hours away) to follow
> the enclosed instructions.  By looks of things, I'll be unsubscribing
> 1146 addresses tonight.
> 
> We're at 161 replies of 1307 subscribers currently.  That's Piss Poor.
> Don't whinge when you can't post messages to the list because you haven't
> followed these instructions.

on the other hand, dropping those that can't be bothered will
certainly help improve the signal to noise ratio....

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--

-------------------------------------------------------------------
Subscription options: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel
FAQ/Ettiquette:       http://www.arm.linux.org.uk/armlinux/mailinglists.php
Permalink | Reply | 0 comments |
headers
Ian Molton | 2 May 2002 15:50

Re: *** URGENT *** Admin: Linux-arm / Linux-arm-kernel: Change of status

Russell King - ARM Linux Awoke this dragon, who will now respond:

> *FINAL* reminder.  You have until Midnight BST (11 hours away) to follow
> the enclosed instructions.  By looks of things, I'll be unsubscribing
> 1146 addresses tonight.

Can you confirm those who /have/ agreed? I'd like to know before I get
accidentally summarily dropped.

-------------------------------------------------------------------
Subscription options: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel
FAQ/Ettiquette:       http://www.arm.linux.org.uk/armlinux/mailinglists.php
Permalink | Reply | 3 comments |
headers
Guennadi Liakhovetski | 2 May 2002 16:07
Picon
Favicon

airo_cs on ARM 2.4.13 vs. 2.5.6

Hello

Brief description of the problem: airo driver (Aironet 350 card) works
with 2.5.6 and doesn't work with 2.4.13 on a StrongARM (Trizeps) board.

Details: Loading the modules (airo.o and airo_cs.o) works fine, the
packets get transmitted, ping host&; tcpdump host host shows packets
transmitted and received, but ping reports all packets lost. Same if I
ping the board - tcpdump only sees packets received, none are transmitted.
In /var/log/messages the IO-address, reported by airo looks like a PC
(ix86) address - 0xf6000000-0xf600003f. With 2.5.6 everything works and
the address-range is 0xc2872000-0xc287203f. Any ideas?

Thanks
Guennadi
---------------------------------
Guennadi Liakhovetski, Ph.D.
DSA Daten- und Systemtechnik GmbH
Pascalstr. 28
D-52076 Aachen
Germany

-------------------------------------------------------------------
Subscription options: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel
FAQ/Ettiquette:       http://www.arm.linux.org.uk/armlinux/mailinglists.php
Permalink | Reply | 0 comments |
headers
Russell King - ARM Linux | 2 May 2002 16:28
Picon

Re: *** URGENT *** Admin: Linux-arm / Linux-arm-kernel: Change of status

On Thu, May 02, 2002 at 02:50:38PM +0100, Ian Molton wrote:
> Russell King - ARM Linux Awoke this dragon, who will now respond:
> 
> > *FINAL* reminder.  You have until Midnight BST (11 hours away) to follow
> > the enclosed instructions.  By looks of things, I'll be unsubscribing
> > 1146 addresses tonight.
> 
> Can you confirm those who /have/ agreed? I'd like to know before I get
> accidentally summarily dropped.

Posting a list would be contary to the DPA. 8/

If you sent a mail to the right address and didn't get a bounce, then
its been received.  And yes, I can confirm that yours is in there.

-------------------------------------------------------------------
Subscription options: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel
FAQ/Ettiquette:       http://www.arm.linux.org.uk/armlinux/mailinglists.php
Permalink | Reply | 2 comments |

Gmane