qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: "Michael S. Tsirkin" <mst@redhat.com>
To: "Màrius Montón" <marius.monton@gmail.com>
Cc: qemu-devel@nongnu.org
Subject: [Qemu-devel] Re: Re: PCI address question
Date: Thu, 29 Oct 2009 14:12:22 +0200	[thread overview]
Message-ID: <20091029121221.GA3478@redhat.com> (raw)
In-Reply-To: <4AE97CBA.80600@gmail.com>

On Thu, Oct 29, 2009 at 12:30:02PM +0100, Màrius Montón wrote:
> Lennart Sorensen wrote:
> 
>     On Tue, Oct 27, 2009 at 04:05:25PM +0100, Màrius Montón wrote:
> 
> 
>         Màrius Montón wrote:
> 
> 
>             Hello,
> 
>             For my PCI device to QEMU, I need the real address the PCI bus is using
>             to access my device. For a IO BAR (PCI_ADDRESS_SPACE_IO), I receive the
>             real address (like 0xc200 or similar), but when registering a
>             PCI_ADDRESS_SPACE_MEM I only receive the offset to the BAR.
> 
>             How I can receive or obtaint the real address on each access to my device?
> 
>             Cjeers,
> 
>             Màrius
> 
> 
> 
>         nobody can tell me anything? :(
> 
> 
>     Add the offset to the address in the bar?
> 
> The problem is how to know what bar is accessing in case I have only one
> function for all BARs.

I expect that's unusual: different BARs usually have different
functionality. Just implement different functions and pass
calls on to a shared helper?

>     After all the OS is allowed to change your BAR if it wants to.
>     So internally the only thing that makes sense to a PCI device is the
>     offset from it's base address.
> 
>     You listen to addresses at your IO range, and at your memory BAR range.
>     What you do when you see a request for your range then depends on the
>     offset that address had from the current base.  This would also be true
>     for the IO.
> 
> I know all PCI internals, but I don't understand why for IO I receive all
> address and only the offset for MEM BARs (or I'm wrong?)
> 
> Màrius

PCI only calls a map method. I think this gets a 32 bit address, not BAR offset:

               r->addr = new_addr;
               if (r->addr != -1) {
                   r->map_func(d, i, r->addr, r->size, r->type);
               }


-- 
MST

  reply	other threads:[~2009-10-29 12:14 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-10-21  9:51 [Qemu-devel] PCI address question Màrius Montón
2009-10-27 15:05 ` [Qemu-devel] " Màrius Montón
2009-10-27 15:16   ` Lennart Sorensen
2009-10-29 11:30     ` Màrius Montón
2009-10-29 12:12       ` Michael S. Tsirkin [this message]
2009-10-29 13:05         ` Màrius Montón

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20091029121221.GA3478@redhat.com \
    --to=mst@redhat.com \
    --cc=marius.monton@gmail.com \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).