From: Anthony Liguori <anthony@codemonkey.ws>
To: Jan Kiszka <jan.kiszka@siemens.com>
Cc: Avi Kivity <avi@redhat.com>, Gleb Natapov <gleb@redhat.com>,
qemu-devel <qemu-devel@nongnu.org>
Subject: Re: [Qemu-devel] [RFC] Memory API
Date: Thu, 19 May 2011 11:36:57 -0500 [thread overview]
Message-ID: <4DD54729.8050707@codemonkey.ws> (raw)
In-Reply-To: <4DD545B2.70705@siemens.com>
On 05/19/2011 11:30 AM, Jan Kiszka wrote:
> On 2011-05-19 18:28, Gleb Natapov wrote:
>> On Thu, May 19, 2011 at 06:25:14PM +0200, Jan Kiszka wrote:
>>> On 2011-05-19 18:17, Gleb Natapov wrote:
>>>> On Thu, May 19, 2011 at 05:40:50PM +0300, Avi Kivity wrote:
>>>>> On 05/19/2011 05:37 PM, Anthony Liguori wrote:
>>>>>>
>>>>>> So.... do you do:
>>>>>>
>>>>>> isa_register_region(ISABus *bus, MemoryRegion *mr, int priority)
>>>>>> {
>>>>>> chipset_register_region(bus->chipset, mr, priority + 1);
>>>>>> }
>>>>>>
>>>>>> I don't really understand how you can fold everything into one
>>>>>> table and not allow devices to override their parents using
>>>>>> priorities.
>>>>>
>>>>> Think of how a window manager folds windows with priorities onto a
>>>>> flat framebuffer.
>>>>>
>>>>> You do a depth-first walk of the tree. For each child list, you
>>>>> iterate it from the lowest to highest priority, allowing later
>>>>> subregions override earlier subregions.
>>>>>
>>>> And how you set those priorities in a sensible way? Why two device on a
>>>> PCI bus will want to register their memory region with anything but
>>>> highest priority? And if you let PCI subsystem to assign priorities how
>>>> it will coordinate with ISA subsystem/memory controller what priorities
>>>> to assign to get meaningful system?
>>>
>>> Priorities>default will only be used for explicit overlays, e.g. RAM
>>> over MMIO in PAM regions. Non-default priorities won't be assigned to
>>> normal PCI bars or any other device's region.
>>>
>> That does not explain who and how assign those priorities in globally
>> meaningful way.
>
> There are no global priorities. Priorities are only used inside each
> level of the memory region hierarchy to generate a resulting, flattened
> view for the next higher level. At that level, everything imported from
> below has the default prio again, ie. the lowest one.
Then SMM is impossible.
Why do we need priorities at all? There should be no overlap at each
level in the hierarchy.
If you have overlapping BARs, the PCI bus will always send the request
to a single device based on something that's implementation specific.
This works because each PCI device advertises the BAR locations and
sizes in it's config space.
To dispatch a request, the PCI bus will walk the config space to find a
match. If you remove something that was previously causing an overlap,
it'll the other device will now get the I/O requests.
To model this correctly, you need to let the PCI bus decide how to
dispatch I/O requests (again, you need hierarchical dispatch).
In the absence of this, the PCI bus needs to look at all of the devices,
figure out the flat mapping, and register it. When a device is added or
removed, it needs to recalculate the flat mapping and register it.
There is no need to have centralized logic to decide this.
Regards,
Anthony Liguori
>
> Jan
>
next prev parent reply other threads:[~2011-05-19 16:37 UTC|newest]
Thread overview: 187+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-05-18 13:12 [Qemu-devel] [RFC] Memory API Avi Kivity
2011-05-18 14:05 ` Jan Kiszka
2011-05-18 14:36 ` Avi Kivity
2011-05-18 15:11 ` Jan Kiszka
2011-05-18 15:17 ` Peter Maydell
2011-05-18 15:30 ` Jan Kiszka
2011-05-18 19:10 ` Anthony Liguori
2011-05-18 19:27 ` Jan Kiszka
2011-05-18 19:34 ` Anthony Liguori
2011-05-18 20:02 ` Alex Williamson
2011-05-18 20:11 ` Jan Kiszka
2011-05-18 20:13 ` Alex Williamson
2011-05-18 20:07 ` Jan Kiszka
2011-05-18 20:41 ` Anthony Liguori
2011-05-19 8:26 ` Gleb Natapov
2011-05-19 8:30 ` Jan Kiszka
2011-05-19 8:44 ` Avi Kivity
2011-05-19 13:59 ` Anthony Liguori
2011-05-19 13:52 ` Anthony Liguori
2011-05-19 13:56 ` Avi Kivity
2011-05-19 13:57 ` Jan Kiszka
2011-05-19 14:04 ` Anthony Liguori
2011-05-19 14:06 ` Jan Kiszka
2011-05-19 14:11 ` Avi Kivity
2011-05-19 18:18 ` Anthony Liguori
2011-05-19 18:50 ` Jan Kiszka
2011-05-19 19:02 ` Anthony Liguori
2011-05-19 19:10 ` Jan Kiszka
2011-05-20 9:15 ` Avi Kivity
2011-05-20 17:30 ` Blue Swirl
2011-05-22 7:23 ` Avi Kivity
2011-05-19 6:31 ` Jan Kiszka
2011-05-19 13:23 ` Anthony Liguori
2011-05-19 13:25 ` Jan Kiszka
2011-05-19 13:26 ` Avi Kivity
2011-05-19 13:35 ` Anthony Liguori
2011-05-19 13:36 ` Jan Kiszka
2011-05-19 13:43 ` Avi Kivity
2011-05-19 13:39 ` Avi Kivity
2011-05-19 8:09 ` Avi Kivity
2011-05-18 15:23 ` Avi Kivity
2011-05-18 15:36 ` Jan Kiszka
2011-05-18 15:42 ` Avi Kivity
2011-05-18 16:00 ` Jan Kiszka
2011-05-18 16:14 ` Avi Kivity
2011-05-18 16:39 ` Jan Kiszka
2011-05-18 16:47 ` Avi Kivity
2011-05-18 17:07 ` Jan Kiszka
2011-05-18 17:15 ` Avi Kivity
2011-05-18 17:40 ` Jan Kiszka
2011-05-18 20:13 ` Richard Henderson
2011-05-19 8:04 ` Avi Kivity
2011-05-19 9:08 ` Gleb Natapov
2011-05-19 9:10 ` Avi Kivity
2011-05-19 9:14 ` Gleb Natapov
2011-05-19 11:44 ` Avi Kivity
2011-05-19 11:54 ` Gleb Natapov
2011-05-19 11:57 ` Jan Kiszka
2011-05-19 11:58 ` Gleb Natapov
2011-05-19 12:02 ` Avi Kivity
2011-05-19 12:21 ` Gleb Natapov
2011-05-19 12:02 ` Jan Kiszka
2011-05-19 11:57 ` Avi Kivity
2011-05-19 12:20 ` Jan Kiszka
2011-05-19 12:50 ` Avi Kivity
2011-05-19 12:58 ` Jan Kiszka
2011-05-19 13:00 ` Avi Kivity
2011-05-19 13:03 ` Jan Kiszka
2011-05-19 13:07 ` Avi Kivity
2011-05-19 13:26 ` Jan Kiszka
2011-05-19 13:30 ` Avi Kivity
2011-05-19 13:43 ` Jan Kiszka
2011-05-19 13:47 ` Avi Kivity
2011-05-19 13:49 ` Anthony Liguori
2011-05-19 13:53 ` Avi Kivity
2011-05-19 14:15 ` Anthony Liguori
2011-05-19 14:20 ` Jan Kiszka
2011-05-19 14:25 ` Anthony Liguori
2011-05-19 14:28 ` Jan Kiszka
2011-05-19 14:31 ` Avi Kivity
2011-05-19 14:37 ` Anthony Liguori
2011-05-19 14:40 ` Avi Kivity
2011-05-19 16:17 ` Gleb Natapov
2011-05-19 16:25 ` Jan Kiszka
2011-05-19 16:28 ` Gleb Natapov
2011-05-19 16:30 ` Jan Kiszka
2011-05-19 16:36 ` Anthony Liguori [this message]
2011-05-19 16:49 ` Jan Kiszka
2011-05-19 17:12 ` Gleb Natapov
2011-05-19 18:11 ` Jan Kiszka
2011-05-20 8:58 ` Avi Kivity
2011-05-20 8:56 ` Avi Kivity
2011-05-20 14:51 ` Anthony Liguori
2011-05-20 16:43 ` Olivier Galibert
2011-05-20 17:32 ` Anthony Liguori
2011-05-22 7:36 ` Avi Kivity
2011-05-19 16:43 ` Gleb Natapov
2011-05-19 16:51 ` Jan Kiszka
2011-05-19 16:27 ` Gleb Natapov
2011-05-20 8:59 ` Avi Kivity
2011-05-20 11:57 ` Gleb Natapov
2011-05-22 7:37 ` Avi Kivity
2011-05-22 8:06 ` Gleb Natapov
2011-05-22 8:09 ` Avi Kivity
2011-05-22 8:59 ` Gleb Natapov
2011-05-22 12:26 ` Avi Kivity
2011-05-19 16:32 ` Anthony Liguori
2011-05-19 16:35 ` Jan Kiszka
2011-05-19 16:38 ` Anthony Liguori
2011-05-19 16:50 ` Jan Kiszka
2011-05-20 9:03 ` Avi Kivity
2011-05-20 9:01 ` Avi Kivity
2011-05-20 15:33 ` Anthony Liguori
2011-05-20 15:59 ` Jan Kiszka
2011-05-22 7:38 ` Avi Kivity
2011-05-22 15:42 ` Anthony Liguori
2011-05-19 14:21 ` Avi Kivity
2011-05-19 13:44 ` Anthony Liguori
2011-05-19 13:47 ` Jan Kiszka
2011-05-19 13:50 ` Anthony Liguori
2011-05-19 13:55 ` Jan Kiszka
2011-05-19 13:55 ` Avi Kivity
2011-05-19 18:06 ` Anthony Liguori
2011-05-19 18:21 ` Jan Kiszka
2011-05-19 13:49 ` Avi Kivity
2011-05-19 9:24 ` Edgar E. Iglesias
2011-05-19 14:49 ` Peter Maydell
2011-05-18 16:33 ` Anthony Liguori
2011-05-18 16:41 ` Avi Kivity
2011-05-18 17:04 ` Anthony Liguori
2011-05-18 17:13 ` Avi Kivity
2011-05-18 16:42 ` Jan Kiszka
2011-05-18 17:05 ` Avi Kivity
2011-05-18 15:14 ` Anthony Liguori
2011-05-18 15:26 ` Avi Kivity
2011-05-18 16:21 ` Avi Kivity
2011-05-18 16:42 ` Jan Kiszka
2011-05-18 16:49 ` Avi Kivity
2011-05-18 17:11 ` Anthony Liguori
2011-05-18 17:38 ` Jan Kiszka
2011-05-18 15:08 ` Anthony Liguori
2011-05-18 15:37 ` Avi Kivity
2011-05-18 19:36 ` Jan Kiszka
2011-05-18 15:47 ` Stefan Weil
2011-05-18 16:06 ` Avi Kivity
2011-05-18 16:51 ` Richard Henderson
2011-05-18 16:53 ` Avi Kivity
2011-05-18 17:03 ` Richard Henderson
2011-05-18 15:58 ` Avi Kivity
2011-05-18 16:26 ` Richard Henderson
2011-05-18 16:37 ` Avi Kivity
2011-05-18 17:17 ` Avi Kivity
2011-05-18 19:40 ` Jan Kiszka
2011-05-19 8:06 ` Avi Kivity
2011-05-19 8:08 ` Jan Kiszka
2011-05-19 8:13 ` Avi Kivity
2011-05-19 8:25 ` Jan Kiszka
2011-05-19 8:43 ` Avi Kivity
2011-05-19 9:24 ` Jan Kiszka
2011-05-19 11:58 ` Avi Kivity
2011-05-19 13:36 ` Anthony Liguori
2011-05-19 13:37 ` Jan Kiszka
2011-05-19 13:41 ` Avi Kivity
2011-05-19 17:39 ` Gleb Natapov
2011-05-19 18:03 ` Anthony Liguori
2011-05-19 18:28 ` Gleb Natapov
2011-05-19 18:33 ` Anthony Liguori
2011-05-19 18:11 ` Jan Kiszka
2011-05-19 18:22 ` Gleb Natapov
2011-05-19 18:27 ` Jan Kiszka
2011-05-19 18:40 ` Gleb Natapov
2011-05-19 18:45 ` Jan Kiszka
2011-05-19 18:50 ` Gleb Natapov
2011-05-19 18:55 ` Jan Kiszka
2011-05-19 19:02 ` Jan Kiszka
2011-05-20 7:23 ` Gleb Natapov
2011-05-20 7:40 ` Jan Kiszka
2011-05-20 11:25 ` Gleb Natapov
2011-05-22 7:50 ` Avi Kivity
2011-05-22 8:41 ` Gleb Natapov
2011-05-22 10:53 ` Jan Kiszka
2011-05-22 11:29 ` Avi Kivity
2011-05-23 8:45 ` Gleb Natapov
2011-05-23 22:29 ` Jamie Lokier
2011-05-20 9:10 ` Avi Kivity
2011-05-20 12:08 ` Gleb Natapov
2011-05-22 7:56 ` Avi Kivity
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=4DD54729.8050707@codemonkey.ws \
--to=anthony@codemonkey.ws \
--cc=avi@redhat.com \
--cc=gleb@redhat.com \
--cc=jan.kiszka@siemens.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).