Re: Howto use PPC4xx memory to memory DMA

Re: Howto use PPC4xx memory to memory DMA

[- Reyneke]

Niklaus,

Here are a few things to check:

1. Your memory destination and source addresses are contiguous. You can get 
this using something like:

mem = (void*)__get_free_pages(GFP_KERNEL | GFP_DMA, get_order(size));

2. Your source and dest addresses are _bus_ addresses. You can get these by:

source = virt_to_bus((u32 *)mem);


Also, we had some trouble with the ppc4xx_ functions using 440 - it seems 
these functions are specific to the PLB4. So if you are trying to do DMA on 
say the PLB peripheral bus on a memory mapped device, you definitely have 
some trouble.

I have some code for doing both - so of you need any further help, just ask.

Regards
   Jan Reyneke

_________________________________________________________________
Get Hotmail, News, Sport and Entertainment from MSN on your mobile.  
http://www.msn.txt4content.com/

Re: Howto use PPC4xx memory to memory DMA

[T Ziomek]

On Fri, 19 Jan 2007, - Reyneke wrote:
> Here are a few things to check:
>
> 1. Your memory destination and source addresses are contiguous. You can get 
> this using something like:
>
> mem = (void*)__get_free_pages(GFP_KERNEL | GFP_DMA, get_order(size));
>
> 2. Your source and dest addresses are _bus_ addresses. You can get these by:
>
> source = virt_to_bus((u32 *)mem);

The above contradicts an email by Ben H back on 22 Dec:
<http://ozlabs.org/pipermail/linuxppc-embedded/2006-December/025546.html>
But I don't know enough to resolve the conflict...

Tom
-- 
   /"\  ASCII Ribbon Campaign   |
   \ /                          |   Email to user 'CTZ001'
    X        Against HTML       |             at 'email.mot.com'
   / \     in e-mail & news     |

Re: Howto use PPC4xx memory to memory DMA

[- Reyneke]

Yes, I know - that email is in response to a question I posted.  :)

This seems to be one of those ongoing uncertainties. It seems commonly 
advocated that you shouldn't use it, but as yet I've found no suitable 
alternative.

Some more reading on why you *should* use it:

http://ozlabs.org/pipermail/linuxppc64-dev/2004-March/001276.html

It would be nice if we could get a definitive answer to this. Anyone out 
there who knows?

Cheers
   Jan Reyneke


>From: T Ziomek <ctz001@email.mot.com>
>>
>>source = virt_to_bus((u32 *)mem);
>
>The above contradicts an email by Ben H back on 22 Dec:
><http://ozlabs.org/pipermail/linuxppc-embedded/2006-December/025546.html>
>But I don't know enough to resolve the conflict...
>
>Tom

_________________________________________________________________
Get Hotmail, News, Sport and Entertainment from MSN on your mobile.  
http://www.msn.txt4content.com/

Re: Howto use PPC4xx memory to memory DMA

[Eugene Surovegin]

On Sun, Jan 21, 2007 at 09:28:34PM +0000, - Reyneke wrote:
> 
> This seems to be one of those ongoing uncertainties. It seems commonly 
> advocated that you shouldn't use it, but as yet I've found no suitable 
> alternative.

Really?

Could you elaborate, why API described in Documentation/DMA-API.txt 
(e.g. "Streaming DMA mappings") is not suitable for you?.

-- 
Eugene

Re: Howto use PPC4xx memory to memory DMA

[padmanabha]

Hi,

I am using 2.6.23 kernel for ppc 440spe based board. When i tried  
memory to memory
  DMA ( example code taken from earlier posting, & used DMA channel  
zero ), DMA is
not happening.

Can any one help me. Any hints would be appreciated.

Thanks and Regards,
Padmanabha.s

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.

Re: Howto use PPC4xx memory to memory DMA

[Eugene Surovegin]

On Fri, Jan 19, 2007 at 11:32:04PM +0100, Alexis Berlemont wrote:
> Hi,
> 
> I would like to add some comments on this topic, I am very interested by more 
> details.
> 
> > Here are a few things to check:
> >
> > 1. Your memory destination and source addresses are contiguous. You can get
> > this using something like:
> >
> > mem = (void*)__get_free_pages(GFP_KERNEL | GFP_DMA, get_order(size));
> >
> > 2. Your source and dest addresses are _bus_ addresses. You can get these
> > by:
> >
> > source = virt_to_bus((u32 *)mem);
> 
> But you must check that your src and dst addresses does not point to 
> a "vmallocated" buffer (unless you allocated less than PAGE_SIZE and you know 
> how to convert a virtual kernel address into a physical address, for example 
> thanks to the function iopa() ). 
> Consequently, with a 2.6 kernel, the function virt_to_bus() will not work 
> properly on a virtual address.
> 
> Consequently, if the dst or the src address belongs to a buffer allocated with 
> vmalloc() or ioremap(), you will have to use the function iopa() to perform 
> the translation.
> 
> Besides, I have a question : I do not understand why the 2.6 implementation of 
> the function virt_to_bus() is different from the 2.4 version; 
> ->on 2.4, virt_to_bus() calls iopa() (if CONFIG_PPC4xx is defined),therefore 
> it works with virtual addresses (vmallocated or ioremapped);
> ->on 2.6, virt_to_bus works with logical address (kmallocated or 
> get_free_pages);
> 
> Do you know why ?

virt_to_bus was never supposed to work on anything else except for 
directly mapped kernel address. 2.6 version just fixes this for 4xx. 
And yes, virt_to_bus is deprecated.


> 
> > Ooh - and the 3d thing - cache coherence. I.e.:
> >
> >       dma_cache_inv((u32)mem, (u32)nBytes);
> >
> 
> I just add a little precision:
> 
> With PPC405 (no cache coherency integrated), you have to invalidate the cache 
> of the destination buffer (with dma_cache_inv() )and you have to write back 
> the cache of the source buffer (with dma_cache_wback() ). But if you do not 
> want to care about the direction point, you can use dma_cache_inv_wback() 
> which does both operations.
> 
> Of course, if you work with an "ioremapped" area (either as source or 
> destination buffer), this functions are useless for this buffer (ioremap() 
> maps the address range with the flag "PAGE_NO_CACHE").

Guys, please stop reinventing the wheel. There is a documented kernel 
API which takes care of coherency issues and it should be used for any 
sort of DMA (be it general purpose DMA controller or external 
DMA-capable device).

PLEASE, read Documentation/DMA-API.txt and use it to obtain DMA-able 
address for your buffer.

-- 
Eugene

Re: Howto use PPC4xx memory to memory DMA

[Alexis Berlemont]

Hi,

I would like to add some comments on this topic, I am very interested by more 
details.

> Here are a few things to check:
>
> 1. Your memory destination and source addresses are contiguous. You can get
> this using something like:
>
> mem = (void*)__get_free_pages(GFP_KERNEL | GFP_DMA, get_order(size));
>
> 2. Your source and dest addresses are _bus_ addresses. You can get these
> by:
>
> source = virt_to_bus((u32 *)mem);

But you must check that your src and dst addresses does not point to 
a "vmallocated" buffer (unless you allocated less than PAGE_SIZE and you know 
how to convert a virtual kernel address into a physical address, for example 
thanks to the function iopa() ). 
Consequently, with a 2.6 kernel, the function virt_to_bus() will not work 
properly on a virtual address.

Consequently, if the dst or the src address belongs to a buffer allocated with 
vmalloc() or ioremap(), you will have to use the function iopa() to perform 
the translation.

Besides, I have a question : I do not understand why the 2.6 implementation of 
the function virt_to_bus() is different from the 2.4 version; 
->on 2.4, virt_to_bus() calls iopa() (if CONFIG_PPC4xx is defined),therefore 
it works with virtual addresses (vmallocated or ioremapped);
->on 2.6, virt_to_bus works with logical address (kmallocated or 
get_free_pages);

Do you know why ?

> Ooh - and the 3d thing - cache coherence. I.e.:
>
>       dma_cache_inv((u32)mem, (u32)nBytes);
>

I just add a little precision:

With PPC405 (no cache coherency integrated), you have to invalidate the cache 
of the destination buffer (with dma_cache_inv() )and you have to write back 
the cache of the source buffer (with dma_cache_wback() ). But if you do not 
want to care about the direction point, you can use dma_cache_inv_wback() 
which does both operations.

Of course, if you work with an "ioremapped" area (either as source or 
destination buffer), this functions are useless for this buffer (ioremap() 
maps the address range with the flag "PAGE_NO_CACHE").

>
>
> Also, we had some trouble with the ppc4xx_ functions using 440 - it seems
> these functions are specific to the PLB4. So if you are trying to do DMA on
> say the PLB peripheral bus on a memory mapped device, you definitely have
> some trouble.

I do not understand, I would be very interested by what kind of trouble you 
had to cope with.

Alexis.

Re: Howto use PPC4xx memory to memory DMA

[Niklaus Giger]

Am Freitag, 19. Januar 2007 10:35 schrieb - Reyneke:
> Ooh - and the 3d thing - cache coherence. I.e.:
>
> 	dma_cache_inv((u32)mem, (u32)nBytes);
>
> >>Here are a few things to check:
Thanks a lot for your tips. First I called
a) __get_free_pages(GFP_KERNEL | GFP_DMA, get_order(size));
for the source and destination address.
Then before calling ppc4xx_enable_dma I added
b) ppc4xx_set_src_addr(DMA_NR, virt_to_bus(src));
c) ppc4xx_set_src_addr(DMA_NR, virt_to_bus(dst));
d) flush_dcache_all(); 
With these changes I got at least one byte copied. After fixing the 
ppc4xx_enable_dma everything worked fine. (Patch posted a few minutes ago.)

Did you manage to use DMA on PLB peripheral bus on a memory mapped dev? I 
think there one should also increment the source or destination address. But 
I have no example HW/code at hand to test these two cases too.

Best regards
-- 
Niklaus Giger

RE: Howto use PPC4xx memory to memory DMA

[- Reyneke]

Ooh - and the 3d thing - cache coherence. I.e.:

	dma_cache_inv((u32)mem, (u32)nBytes);



>>Here are a few things to check:

_________________________________________________________________
Find Love This New Year With match.com! http://msnuk.match.com

Howto use PPC4xx memory to memory DMA

[Niklaus Giger]

Hi

I ported successfully the Linux 2.6.19 kernel to our PPC405GPr based board and 
mostly everything just works as it should.

As an example I tried to make a small example program to use memory to memory 
DMA and failed miserably. I used the ppc4xx_*dma function, which are not 
referenced anywhere in a stock 2.6.19.1 kernel in the following function:

void dma_mem_to_mem(void *src, void *dst, unsigned int length,
			unsigned int use_interrupt)
{
	int res = 0;
	memset((char *)&p_init, sizeof(p_init), 0);
	p_init.polarity = 0;
	p_init.pwidth   = PW_8;
 	res = ppc4xx_init_dma_channel(DMA_NR, &p_init);
	if (res) {
		printk("%32s: nit_dma_channel return %d %d bytes dest %p\n",
			__FUNCTION__, res, length, dst);
	}
	res = ppc4xx_clr_dma_status(DMA_NR);
	if (res) { 
		printk("%32s: ppc4xx_clr_dma_status %d\n", __FUNCTION__, res);
	}

	ppc4xx_set_dma_mode(DMA_NR, DMA_MODE_MM);
	ppc4xx_set_src_addr(DMA_NR, src);
	ppc4xx_set_dst_addr(DMA_NR, dst);
	ppc4xx_set_dma_count(DMA_NR, length);
	ppc4xx_enable_dma(DMA_NR);
	if (use_interrupt) {
		res = ppc4xx_enable_dma_interrupt(DMA_NR);
	} else {
		res = ppc4xx_disable_dma_interrupt(DMA_NR);
	}
	if (res) { 
		printk("%32s: en/disable_dma_interrupt %d return %d per %d\n",
		__FUNCTION__, use_interrupt, res, 
		ppc4xx_get_peripheral_width(DMA_NR));
	}
}

However the destination memory never changed, even when calling repeatedly 
ppc4xx_get_dma_residu function correctly return first something between 
length and 0 and finally 0.

Any hints would be appreciated. As googleing around didn't allow me to find an 
answer.

Also I am a little confused as I cannot see how something dma_alloc_coherent 
should work on a PPC4xx board.

Best regards

-- 
Niklaus Giger