From: Anthony Liguori <anthony@codemonkey.ws>
To: Avi Kivity <avi@redhat.com>
Cc: Lucas Meneghel Rodrigues <lmr@redhat.com>,
Marcelo Tosatti <mtosatti@redhat.com>,
qemu-devel <qemu-devel@nongnu.org>,
Blue Swirl <blauwirbel@gmail.com>, Jan Kiszka <jan.kiszka@web.de>,
Gerd Hoffmann <kraxel@redhat.com>
Subject: Re: [Qemu-devel] [PATCH] pc: Clean up PIC-to-APIC IRQ path
Date: Sat, 03 Sep 2011 15:07:00 -0500 [thread overview]
Message-ID: <4E6288E4.10805@codemonkey.ws> (raw)
In-Reply-To: <4E5F1F05.7050704@redhat.com>
On 09/01/2011 12:58 AM, Avi Kivity wrote:
> On 08/31/2011 07:59 PM, Blue Swirl wrote:
>> >
>> > That makes it impossible to migrate level-triggered irq lines. Or at
>> least,
>> > the receiver has to remember the state, instead of (or in addition
>> to) the
>> > sender.
>>
>> Both ends probably need to remember the state. That should work
>> without any multiphase restores and transient suppressors.
>
> State should always correspond to real hardware state - a flip flop or
> capacitor. Input is not state, it is input.
>
>> It might be also possible to introduce stateful signal lines which
>> save and restore their state, then the receiving end could check what
>> is the current level. However, if you consider that the devices may be
>> restored in random order, if the IRQ line device happens to be
>> restored later, the receiver would still get wrong information. Adding
>> priorities could solve this, but I think stateless IRQs are the only
>> sane way.
>
> I agree that irqs should be stateless, since they don't have any memory
> associated.
In Real Life, you can tie a single bit multiple registers together with
boolean logic to form an output pin. This is essentially computed
state. If we wanted to model a stateless pin, we would need to do
something like:
struct Serial {
uint8_t thr;
uint8_t lsr;
};
static bool serial_get_irq(Serial *s) {
return (s->thr & THRE) | (s->lsr & LSRE);
}
static void serial_write(Serial *s, uint64_t addr, uint8_t value)
{
switch (addr) {
case THR:
bool old_irq = serial_get_irq(s);
s->thr = value;
if (!old_irq && serial_get_irq(s)) {
notify_edge_change(s);
}
...
}
static void serial_init(Serial *s)
{
register_pin(s, serial_get_irq);
}
Obviously, this is pretty sucky. This is what we do today but we don't
have a way to query irq value which is wrong. You could fix that by
adding the get function but that's not terribly fun. A better way:
struct Serial {
Pin irq;
uint8_t thr;
uint8_t lsr;
};
static void serial_update_irq(Serial *s)
{
pin_set_level(&s->irq, (s->thr & THRE) | (s->lsr & LSRE));
}
static void serial_write(Serial *s) {
switch (addr) {
case THR:
s->thr = value;
serial_update_irq(s);
...
}
This results in much nicer code. The edge handling can be done in
generic code which will make things more robust overall.
Regards,
Anthony Liguori
next prev parent reply other threads:[~2011-09-03 20:07 UTC|newest]
Thread overview: 68+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-08-27 14:16 [Qemu-devel] [PATCH] pc: Clean up PIC-to-APIC IRQ path Jan Kiszka
2011-08-28 7:10 ` Blue Swirl
2011-08-28 9:08 ` Jan Kiszka
2011-08-29 19:25 ` Anthony Liguori
2011-08-29 21:06 ` Jan Kiszka
2011-08-29 21:13 ` Avi Kivity
2011-08-29 21:18 ` Jan Kiszka
2011-08-30 19:19 ` Blue Swirl
2011-08-30 19:28 ` Jan Kiszka
2011-08-30 19:43 ` Blue Swirl
2011-08-31 8:25 ` Peter Maydell
2011-08-31 10:53 ` Jan Kiszka
2011-08-31 17:41 ` Blue Swirl
2011-08-31 18:17 ` Jan Kiszka
2011-08-31 19:44 ` Blue Swirl
2011-09-04 10:33 ` Blue Swirl
2011-09-04 12:25 ` Gleb Natapov
2011-09-03 19:54 ` Anthony Liguori
2011-09-04 12:13 ` Jan Kiszka
2011-09-04 13:32 ` Anthony Liguori
2011-09-04 13:36 ` Jan Kiszka
2011-09-04 13:41 ` Anthony Liguori
2011-09-04 13:49 ` Jan Kiszka
2011-09-04 13:57 ` Anthony Liguori
2011-09-04 14:37 ` Anthony Liguori
2011-09-04 15:20 ` Blue Swirl
2011-09-04 15:31 ` Anthony Liguori
2011-09-04 15:44 ` Blue Swirl
2011-09-05 10:44 ` Edgar E. Iglesias
2011-09-04 14:12 ` Avi Kivity
2011-09-04 14:43 ` Anthony Liguori
2011-09-04 15:03 ` Avi Kivity
2011-09-04 15:19 ` Anthony Liguori
2011-09-04 15:34 ` Avi Kivity
2011-09-04 15:27 ` Blue Swirl
2011-09-04 12:17 ` Avi Kivity
2011-09-04 12:37 ` Jan Kiszka
2011-09-04 12:43 ` Avi Kivity
2011-09-04 13:38 ` Anthony Liguori
2011-09-04 13:42 ` Jan Kiszka
2011-09-04 13:55 ` Anthony Liguori
2011-09-04 13:35 ` Anthony Liguori
2011-08-31 8:28 ` Avi Kivity
2011-08-31 16:59 ` Blue Swirl
2011-08-31 18:04 ` Edgar E. Iglesias
2011-08-31 18:28 ` Jan Kiszka
2011-09-01 5:58 ` Avi Kivity
2011-09-03 20:07 ` Anthony Liguori [this message]
2011-09-03 21:10 ` Blue Swirl
2011-09-03 21:41 ` Anthony Liguori
2011-09-04 9:27 ` Blue Swirl
2011-09-03 19:53 ` Anthony Liguori
2011-09-03 21:01 ` Blue Swirl
2011-09-04 14:49 ` Anthony Liguori
2011-09-05 8:38 ` Edgar E. Iglesias
2011-09-05 8:51 ` Avi Kivity
2011-09-05 9:02 ` Peter Maydell
2011-09-05 9:14 ` Avi Kivity
2011-09-05 9:22 ` Edgar E. Iglesias
2011-09-05 9:28 ` Avi Kivity
2011-09-05 10:47 ` Edgar E. Iglesias
2011-09-05 19:36 ` Blue Swirl
2011-09-06 7:46 ` Avi Kivity
2011-09-01 9:08 ` [Qemu-devel] [PATCH v2] pc: Fix and clean " Jan Kiszka
2011-09-03 8:58 ` Blue Swirl
2011-09-03 11:17 ` Jan Kiszka
2011-09-03 11:37 ` Blue Swirl
2011-09-03 18:14 ` Jan Kiszka
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=4E6288E4.10805@codemonkey.ws \
--to=anthony@codemonkey.ws \
--cc=avi@redhat.com \
--cc=blauwirbel@gmail.com \
--cc=jan.kiszka@web.de \
--cc=kraxel@redhat.com \
--cc=lmr@redhat.com \
--cc=mtosatti@redhat.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).