📄 translating addresses in kernel space.htm
字号:
<html><head><title>Translating Addresses in Kernel Space</title>
<link rel="owner" href="mailto:">
<script language="JavaScript">
<!-- hide this
function help(message) {
self.status = message;
return true;
}
// stop hiding -->
</script></head>
<body>
<strong>The
HyperNews <a href="http://tldp.org/LDP/khg/HyperNews/get/khg.html">Linux KHG</a>
Discussion Pages</strong>
<hr>
<h2>Translating Addresses in Kernel Space</h2>
<blockquote><i>
From a message from Linus Torvalds to the linux-kernel mailing list
of 27 Sep 1996, edited.
</i></blockquote>
I'll take this opportunity to tell all device driver writers about
the ugly secrets of portability. Things are actually worse than just
physical and virtual addresses.
<p>
The aha1542 is a bus-master device, and [a patch posted to the
linux-kernel list] makes the driver give the controller the physical
address of the buffers, which is correct on x86, because all bus master
devices see the physical memory mappings directly.
</p><p>
However, on many setups, there are actually <i>three</i> different ways
of looking at memory addresses, and in this case we actually want the
third, the so-called "bus address".
</p><p>
Essentially, the three ways of addressing memory are (this is "real memory",
i.e. normal RAM; see later about other details):
</p><ul>
<li>CPU untranslated. This is the "physical" address, ie physical address
0 is what the CPU sees when it drives zeroes on the memory bus.
</li><li>CPU translated address. This is the "virtual" address, and is
completely internal to the CPU itself with the CPU doing the appropriate
translations into "CPU untranslated".
</li><li>Bus address. This is the address of memory as seen by OTHER devices,
not the CPU. Now, in theory there could be many different bus
addresses, with each device seeing memory in some device-specific way, but
happily most hardware designers aren't actually actively trying to make
things any more complex than necessary, so you can assume that all
external hardware sees the memory the same way.
</li></ul>
Now, on normal PC's, the bus address is exactly the same as the physical
address, and things are very simple indeed. However, they are that simple
because the memory and the devices share the same address space, and that is
not generally necessarily true on other PCI/ISA setups.
<p>
Now, just as an example, on the PReP (PowerPC Reference Platform), the
CPU sees a memory map something like this (this is from memory):
</p><dl>
<dt>0-2GB</dt><dd>"real memory"
</dd><dt>2GB-3GB</dt><dd>"system IO" (ie inb/out type accesses on x86)
</dd><dt>3GB-4GB</dt><dd>"IO memory" (ie shared memory over the IO bus)
</dd></dl>
Now, that looks simple enough. However, when you look at the same thing from
the viewpoint of the devices, you have the reverse, and the physical memory
address 0 actually shows up as address 2GB for any IO master.
<p>
So when the CPU wants any bus master to write to physical memory 0, it
has to give the master address <tt>0x80000000</tt> as the memory address.
</p><p>
So, for example, depending on how the kernel is actually mapped on the
PPC, you can end up with a setup like this:
</p><dl>
<dt>physical address:</dt><dd>0
</dd><dt>virtual address:</dt><dd>0xC0000000
</dd><dt>bus address:</dt><dd>0x80000000
</dd></dl>
where all the addresses actually point to the same thing, it's just seen
through different translations.
<p>
Similarly, on the alpha, the normal translation is
</p><dl>
<dt>physical address:</dt><dd>0
</dd><dt>virtual address:</dt><dd>0xfffffc0000000000
</dd><dt>bus address:</dt><dd>0x40000000
</dd></dl>
(but there are also alpha's where the physical address and the bus address
are the same).
<p>
Anyway, the way to look up all these translations, you do:
</p><pre> #include <asm/io.h>
phys_addr = virt_to_phys(virt_addr);
virt_addr = phys_to_virt(phys_addr);
bus_addr = virt_to_bus(virt_addr);
virt_addr = bus_to_virt(bus_addr);
</pre>
Now, when do you need these?
<p>
You want the <i>virtual</i> address when you are actually going to access
that pointer from the kernel. So you can have something like this (from
the aha1542 driver):
</p><pre> /*
* this is the hardware "mailbox" we use to communicate with
* the controller. The controller sees this directly.
*/
struct mailbox {
__u32 status;
__u32 bufstart;
__u32 buflen;
..
} mbox;
unsigned char * retbuffer;
/* get the address from the controller */
retbuffer = bus_to_virt(mbox.bufstart);
switch (retbuffer[0]) {
case STATUS_OK:
...
</pre>
On the other hand, you want the bus address when you have a buffer that
you want to give to the controller:
<pre> /* ask the controller to read the sense status into "sense_buffer" */
mbox.bufstart = virt_to_bus(&sense_buffer);
mbox.buflen = sizeof(sense_buffer);
mbox.status = 0;
notify_controller(&mbox);
</pre>
And you generally <i>never</i> want to use the physical address, because
you can't use that from the CPU (the CPU only uses translated virtual
addresses), and you can't use it from the bus master.
<p>
So why do we care about the physical address at all? We do need the
physical address in some cases, it's just not very often in normal code.
The physical address is needed if you use memory mappings, for example,
because the <tt>remap_page_range()</tt> mm function wants the physical
address of the memory to be remapped (the memory management layer doesn't
know about devices outside the CPU, so it shouldn't need to know about
"bus addresses" etc).
</p><p>
<b>NOTE NOTE NOTE!</b> The above is only one part of the whole
equation. The above only talks about "real memory", i.e. CPU memory, i.e. RAM.
</p><p>
There is a completely different type of memory too, and that's the "shared
memory" on the PCI or ISA bus. That's generally not RAM (although in the case
of a video graphics card it can be normal DRAM that is just used for a frame
buffer), but can be things like a packet buffer in a network card etc.
</p><p>
This memory is called "PCI memory" or "shared memory" or "IO
memory" or whatever, and there is only one way to access it: the
<tt>readb</tt>/<tt>writeb</tt> and related functions. You should never
take the address of such memory, because there is really nothing you can
do with such an address: it's not conceptually in the same memory space as
"real memory" at all, so you cannot just dereference a pointer. (Sadly,
on x86 it <i>is</i> in the same memory space, so on x86 it actually
works to just deference a pointer, but it's not portable).
</p><p>
For such memory, you can do things like
</p><dl>
<dt><b>Reading:</b>
</dt><dd><pre> /*
* read first 32 bits from ISA memory at 0xC0000, aka
* C000:0000 in DOS terms
*/
unsigned int signature = readl(0xC0000);
</pre>
</dd><dt><b>Remapping and writing:</b>
</dt><dd><pre> /*
* remap framebuffer PCI memory area at 0xFC000000,
* size 1MB, so that we can access it: We can directly
* access only the 640k-1MB area, so anything else
* has to be remapped.
*/
char * baseptr = ioremap(0xFC000000, 1024*1024);
/* write a 'A' to the offset 10 of the area */
writeb('A',baseptr+10);
/* unmap when we unload the driver */
iounmap(baseptr);
</pre>
</dd><dt><b>Copying and clearing:</b>
</dt><dd><pre> /* get the 6-byte ethernet address at ISA address E000:0040 */
memcpy_fromio(kernel_buffer, 0xE0040, 6);
/* write a packet to the driver */
memcpy_toio(0xE1000, skb->data, skb->len);
/* clear the frame buffer */
memset_io(0xA0000, 0, 0x10000);
</pre>
</dd></dl>
Ok, that just about covers the basics of accessing IO portably.
Questions? Comments? You may think that all the above is overly complex,
but one day you might find yourself with a 500MHz alpha in front of you,
and then you'll be happy that your driver works <tt>;)</tt>
<p>
Note that kernel versions 2.0.x (and earlier) mistakenly called
<tt>ioremap()</tt> "<tt>vremap()</tt>". <tt>ioremap()</tt> is the
proper name, but I didn't think straight when I wrote it originally.
People who have to support both can do something like:
</p><pre> /* support old naming sillyness */
#if LINUX_VERSION_CODE < 0x020100
#define ioremap vremap
#define iounmap vfree
#endif
</pre>
at the top of their source files, and then they can use the right
names even on 2.0.x systems.
<p>
And the above sounds worse than it really is. Most real drivers really
don't do all that complex things (or rather: the complexity is not so much
in the actual IO accesses as in error handling and timeouts etc). It's
generally not hard to fix drivers, and in many cases the code actually
looks better afterwards:
</p><pre> unsigned long signature = *(unsigned int *) 0xC0000;
</pre>
vs.
<pre> unsigned long signature = readl(0xC0000);
</pre>
I think the second version actually is more readable, no?
<p>
</p><pre> Linus</pre>
<p>
</p><p></p><hr size="3">
<p>
</p><p>
<br>
<br></p></body></html>⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -