How does one get physical address for iorempped window?

How does one get physical address for iorempped window?

[Subodh Nijsure]

[ Perhaps not directly linux-ppc question ]

I am trying to get jffs working on my board and in trying to debug this
I need to know this.  The jffs driver requires that I map the flash address
space usinng ioremap_nocache or ioremap, when debugging I need to see exactly
which addresses its writing the CFI query commands.

My question is --

If I map memory window using ioremap_nocache() if I use virt_to_phys()
should one get back the actual physical address? If virt_to_phys is not
the correct function to use, how do I get  exact physical address?

Thanks.

/Subodh Nijsure

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: How does one get physical address for iorempped window?

[Dan Malek]

Subodh Nijsure wrote:

> If I map memory window using ioremap_nocache() if I use virt_to_phys()
> should one get back the actual physical address?

Ummmm....you give ioremap_nocache() a physical address to map, so
you have to know it........Using virt_to_phys() may not give you
the proper physical address, depending upon the platform.  I'm working
on a solution for 2.5 that will work.  Currently, it is wrapped into
a heavier weight virt_to_bus/phys() but that many not be the right
place.



	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

RE: How does one get physical address for iorempped window?

[Subodh Nijsure]

I know the physical address I should be writing to.

For debugging I was trying to print exact physical address where
write is happening .When I used virt_to_phys() on the addresse
ioremap() I was not getting back the original physical address.
Hence the question.

Basically code I have looks like this.

unsigned long physaddr;

physmap_map.map_priv_1 = (unsigned
long)ioremap_nocache(WINDOW_ADDR,WINDOW_SIZE);
printk("Physical window (0x%08X size)starting at 0x%08X is mapped to address
0x%08X \n",
            WINDOW_SIZE,WINDOW_ADDR, physmap_map.map_priv_1);
physaddr = virt_to_phys((void *)physmap_map.map_priv_1);
printk("virtual address 0x%08X maps to physical 0x%08X \n",
                        physmap_map.map_priv_1 ,physaddr);

The output I was getting is this.

Physical window (0x00400000 size) starting at 0x28000000 is mapped to
address 0xC5012000
virtual address 0xC5012000 maps to physical 0x05012000


/Subodh Nijsure


> -----Original Message-----
> From: Dan Malek [mailto:dan@mvista.com]
> Sent: Thursday, February 15, 2001 9:03 AM
> To: Subodh Nijsure
> Cc: linuxppc-embedded@lists.linuxppc.org
> Subject: Re: How does one get physical address for iorempped window?
>
>
>
> Subodh Nijsure wrote:
>
> > If I map memory window using ioremap_nocache() if I use
> virt_to_phys()
> > should one get back the actual physical address?
>
> Ummmm....you give ioremap_nocache() a physical address to map, so
> you have to know it........Using virt_to_phys() may not give you
> the proper physical address, depending upon the platform.  I'm working
> on a solution for 2.5 that will work.  Currently, it is wrapped into
> a heavier weight virt_to_bus/phys() but that many not be the right
> place.
>
>
>
> 	-- Dan
>

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: How does one get physical address for iorempped window?

[Dan Malek]

Subodh Nijsure wrote:

> ....When I used virt_to_phys() on the addresse
> ioremap() I was not getting back the original physical address.
> Hence the question.

OK.  Worded differently from my previous message.....virt_to_phys()
won't work properly on any dynamically mapped space.......


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: How does one get physical address for iorempped window?

[Matt Porter]

On Thu, Feb 15, 2001 at 09:58:59AM -0800, Subodh Nijsure wrote:
>
> I know the physical address I should be writing to.
>
> For debugging I was trying to print exact physical address where
> write is happening .When I used virt_to_phys() on the addresse
> ioremap() I was not getting back the original physical address.
> Hence the question.

Well, the base problem is that virt_to_*/*_to_virt were only ever
intended to work on addresses mapped to system RAM.  See
Documentation/IO-mapping.txt.  Why not just store argument one to
ioremap to calculate the physical addresses you are working with
in virtual space?

--
Matt Porter
mmporter@home.com
This is Linux Country. On a quiet night, you can hear Windows reboot.

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: How does one get physical address for iorempped window?

[Dan Malek]

Matt Porter wrote:

> Well, the base problem is that virt_to_*/*_to_virt were only ever
> intended to work on addresses mapped to system RAM.

I know.....But, even for someone like me that works with this
on a nearly daily basis it is sometimes confusing.  It's truly
easy to have one function that can tell you virt/phys/virt mappings,
other OS implementations have done that for years.  If you want
a "fast" RAM only mapping, use __pa()/__va(), they have been around
forever.  The virt_to_*/*_to_virt functions _should_ have become
something more generic, but they are just yet another macro name
for doing the same thing.  Now, we have gone of and created some pci_*
mapping functions, which are not at all useful for highly integrated
processors because their integrated features are not PCI devices.
Yet another set of functions for integrated processors?  I don't think so.

I think that requiring knowledge of the object attributes for a VM or
physical space so you can call the proper mapping or other management
functions is just a poor design.  If you know this information and
wish to take performance enhancement shortcuts, that's fine, but it
shouldn't be required.

I'm working on it........


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: How does one get physical address for iorempped window?

[mod+linuxppc-embedded]

>virt_to_*/*_to_virt functions _should_ have become
>something more generic, but they are just yet another macro
>name for doing the same thing.  Now, we have gone of and
>created some pci_* mapping functions, which are not at all
>useful for highly integrated processors because their
>integrated features are not PCI devices.  Yet another set
>of functions for integrated processors?  I don't think so.
>
>I think that requiring knowledge of the object attributes
>for a VM or physical space so you can call the proper
>mapping or other management functions is just a poor
>design.  If you know this information and wish to take
>performance enhancement shortcuts, that's fine, but it
>shouldn't be required.


What if you're working on a system with multiple PCI
busses (or even multiple disparate bus types) and it's
not possible to map Host memory at the same address on
all busses?  Being able to refer to something like, say,
the appropriate pci_bus* when invoking virt_to_bus()
could make it possible to accomodate those different
mappings.  I'm certainly not a fan of pointlessly
complex code, but the one-size-fits-all implementation
of virt_to_bus() seen in my 25Jan-vintage kernel sources
(it just lops off the KERNEL_OFFSET bits and adds
pci_dram_offset) won't work in situations such as those
I just described.


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: How does one get physical address for iorempped window?

[Dan Malek]

mod+linuxppc-embedded@world-f.std.com wrote:

> .....  Being able to refer to something like, say,
> the appropriate pci_bus* when invoking virt_to_bus()

Right, you have to provide enough information to get the
job done correctly, but this could be more generic.  Adding
pci_bus* to virt_to_bus is the right concept, but the wrong
implementation as I certainly want to use virt_to_bus() (or
whatever) for devices that are not on a PCI bus.

> ...  I'm certainly not a fan of pointlessly
> complex code, but the one-size-fits-all implementation
> of virt_to_bus()....

That's not a one-size-fits-all implementation :-).  It (and
virt_to_phys()) worked differently on at least three different
PowerPC implementations.  Mainly to eliminate a bunch of nested
#ifdefs and duplicated code, I now use the same implementation
on all PowerPCs, but it is just the first small step to doing
something better.


	-- Dan

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: How does one get physical address for iorempped window?

[mod+linuxppc-embedded]

>> ...  I'm certainly not a fan of pointlessly
>> complex code, but the one-size-fits-all implementation
>> of virt_to_bus()....
>
>That's not a one-size-fits-all implementation :-).


You're right - I should have said "procrustean"...   ;->


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/