qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Paulo Alcantara <pcacjr@zytor.com>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: seabios@seabios.org, Paulo Alcantara <pcacjr@gmail.com>,
	qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [PATCH 1/3] ich9: add TCO interface emulation
Date: Wed, 27 May 2015 15:23:28 -0300	[thread overview]
Message-ID: <20150527152328.33a8acf5@zytor.com> (raw)
In-Reply-To: <5565B181.7020700@redhat.com>

On Wed, 27 May 2015 13:58:57 +0200
Paolo Bonzini <pbonzini@redhat.com> wrote:

> the main issue here is lack of migration support.  You need to add a
> new subsection to vmstate_ich9_pm that is migrated if the registers
> are different from the default values.

Hrm - OK. I didn't even take that into account. I'll do it.

> > +static QEMUTimer *tco_timer;
> > +static unsigned int timeouts_no;
> 
> These must not be globals.  Instead, add them to TCOIORegs.

OK.

> 
> > +static inline void tco_timer_reload(void)
> > +{
> > +    int64_t now = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);
> > +
> > +    timer_del(tco_timer);
> > +    timer_mod(tco_timer, now + tco_ticks_per_sec());
> 
> No need for del+mod.  Also, these are not tco_ticks_per_sec() but
> rather TCO_TICK_NSEC.  It can be a #define instead of a function.

OK.

> For a little extra challenge, you could the base value when the timer
> was started inside TCOIORegs.  The timer can be made to expire
> directly after 0.6*TCO_TMR seconds; you can use the base value and
> the current time to compute the value of TCO_RLD.  Something like:
> 
>     case TCO_RLD:
>         rld = tr->tco.rld;
>         /* base_clock is set to -1 in tco_timer_stop, reloaded in
>          * tco_timer_reload.
>          */
>         if (base_clock != -1) {
>             elapsed = (qemu_get_clock(QEMU_CLOCK_VIRTUAL) -
> base_clock) / TCO_TICK_NSEC;
>             rld -= MIN(elapsed, rld);
>         }
> 
> This makes it possible for QEMU to sleep most of the time when the
> guest is idle, instead of waking up every 0.6 seconds.  But feel free
> to work on this afterwards.

Nice. I liked your approach and that will definitely avoid overheard
waking up on every 0.6s. No, I will work on this approach right now so
I can sleep well.

> 
> > +void acpi_pm_tco_init(TCOIORegs *tr)
> > +{
> > +    *tr = TCO_IO_REGS_DEFAULTS_INIT();
> 
> Just inline the macro definition here.

OK.

> > +    case TCO1_CNT:
> > +        val &= TCO1_CNT_MASK;
> > +        /* TCO_LOCK bit cannot be changed once set */
> > +        tr->tco.cnt1 = (val & ~TCO_LOCK) | (tr->tco.cnt1 &
> > TCO_LOCK);
> 
> This means that TCO_LOCK is never set in tr->tco.cnt1, I think?  If
> I'm correct, it should be covered by tests.

No. The TCO_LOCK bit is set in tr->tco.cnt1, but cannot be unset
afterwards -- on reset, of course, it will. Maybe a test that covers
setting it and then unsetting it might be interesting to have in
tests/tco-test.c.

> > +#define TCO_IO_REGS_DEFAULTS_INIT()             \
> > +    (TCOIORegs) {                               \
> > +        .tco = {                                \
> > +            .rld      = TCO_RLD_DEFAULT,        \
> > +            .din      = TCO_DAT_IN_DEFAULT,     \
> > +            .dout     = TCO_DAT_OUT_DEFAULT,    \
> > +            .sts1     = TCO1_STS_DEFAULT,       \
> > +            .sts2     = TCO2_STS_DEFAULT,       \
> > +            .cnt1     = TCO1_CNT_DEFAULT,       \
> > +            .cnt2     = TCO2_CNT_DEFAULT,       \
> > +            .msg1     = TCO_MESSAGE1_DEFAULT,   \
> > +            .msg2     = TCO_MESSAGE2_DEFAULT,   \
> > +            .wdcnt    = TCO_WDCNT_DEFAULT,      \
> > +            .tmr      = TCO_TMR_DEFAULT,        \
> > +        },                                      \
> > +        .sw_irq_gen = SW_IRQ_GEN_DEFAULT        \
> > +    }
> 
> No need for this definition.  This and the *_DEFAULT definitions as
> well should be in the .c file.

OK.

> 
> > +/* tco.c */
> > +void acpi_pm_tco_init(TCOIORegs *tr);
> > +uint32_t acpi_pm_tco_ioport_readw(TCOIORegs *tr, uint32_t addr);
> > +void acpi_pm_tco_ioport_writew(TCOIORegs *tr, uint32_t addr,
> > uint32_t val); +
> > +/* As per ICH9 spec, the internal timer has an error of ~0.6s on
> > every tick */ +static inline int64_t tco_ticks_per_sec(void)
> > +{
> > +    return 600000000LL;
> > +}
> > +
> > +static inline int is_valid_tco_time(uint32_t val)
> > +{
> > +    /* values of 0 or 1 will be ignored by ICH */
> > +    return val > 1;
> > +}
> > +
> > +static inline int can_start_tco_timer(TCOIORegs *tr)
> > +{
> > +    return !(tr->tco.cnt1 & TCO_TMR_HLT) &&
> > is_valid_tco_time(tr->tco.tmr); +}
> 
> These three inlines should be in the .c file.

OK.

> 
> Don't be discouraged by the comments.  It's good work!

Alright. Thank you for the comments! I'll send a v2 soon.

Paulo

-- 
Paulo Alcantara, C.E.S.A.R
Speaking for myself only.

  reply	other threads:[~2015-05-27 18:23 UTC|newest]

Thread overview: 83+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-05-27  0:29 [Qemu-devel] [PATCH 1/3] ich9: add TCO interface emulation Paulo Alcantara
2015-05-27  0:29 ` [Qemu-devel] [PATCH 2/3] target-i386: reserve RCRB mmio space in ACPI DSDT table Paulo Alcantara
2015-05-27 12:03   ` Paolo Bonzini
2015-05-27 17:51     ` Paulo Alcantara
2015-05-28  7:13   ` [Qemu-devel] [SeaBIOS] " Gerd Hoffmann
2015-05-30 10:57     ` Paulo Alcantara
2015-06-01  7:16       ` Gerd Hoffmann
2015-06-01 11:59         ` Paulo Alcantara
2015-05-27  0:29 ` [Qemu-devel] [PATCH 3/3] tests: add testcase for TCO watchdog emulation Paulo Alcantara
2015-05-27 11:58 ` [Qemu-devel] [PATCH 1/3] ich9: add TCO interface emulation Paolo Bonzini
2015-05-27 18:23   ` Paulo Alcantara [this message]
2015-05-30 22:04 ` [Qemu-devel] [PATCH v2 " Paulo Alcantara
2015-05-30 22:04   ` [Qemu-devel] [PATCH v2 2/3] target-i386: reserve RCRB mmio space in ACPI DSDT table Paulo Alcantara
2015-05-30 22:04   ` [Qemu-devel] [PATCH v2 3/3] tests: add testcase for TCO watchdog emulation Paulo Alcantara
2015-06-01  9:07     ` Paolo Bonzini
2015-06-01  9:05   ` [Qemu-devel] [PATCH v2 1/3] ich9: add TCO interface emulation Paolo Bonzini
2015-06-01 13:38     ` Paulo Alcantara
2015-06-01 21:37       ` Paulo Alcantara
2015-06-01 23:48   ` [Qemu-devel] [PATCH v3 " Paulo Alcantara
2015-06-01 23:48     ` [Qemu-devel] [PATCH v3 2/3] target-i386: reserve RCRB mmio space in ACPI DSDT table Paulo Alcantara
2015-06-17 13:33       ` Michael S. Tsirkin
2015-06-18  2:14         ` Paulo Alcantara
2015-06-01 23:48     ` [Qemu-devel] [PATCH v3 3/3] tests: add testcase for TCO watchdog emulation Paulo Alcantara
2015-06-17 13:37       ` Michael S. Tsirkin
2015-06-18  2:23         ` Paulo Alcantara
2015-06-10 13:17     ` [Qemu-devel] [PATCH v3 1/3] ich9: add TCO interface emulation Paulo Alcantara
2015-06-17 13:27     ` Michael S. Tsirkin
2015-06-18  2:10       ` Paulo Alcantara
2015-06-22  0:37 ` [Qemu-devel] [PATCH v4 " Paulo Alcantara
2015-06-22  0:37   ` [Qemu-devel] [PATCH v4 2/3] target-i386: reserve RCRB mmio space in ACPI DSDT table Paulo Alcantara
2015-06-22  8:40     ` Michael S. Tsirkin
2015-06-22 12:53       ` Paulo Alcantara
2015-06-23 10:38     ` [Qemu-devel] [SeaBIOS] " Igor Mammedov
2015-06-23 10:58       ` Michael S. Tsirkin
2015-06-23 12:29         ` Igor Mammedov
2015-06-23 12:37           ` Michael S. Tsirkin
2015-06-23 11:15       ` Paolo Bonzini
2015-06-23 11:18       ` Paolo Bonzini
2015-06-23 12:22       ` Michael S. Tsirkin
2015-06-22  0:37   ` [Qemu-devel] [PATCH v4 3/3] tests: add testcase for TCO watchdog emulation Paulo Alcantara
2015-06-22  8:39   ` [Qemu-devel] [PATCH v4 1/3] ich9: add TCO interface emulation Michael S. Tsirkin
2015-06-22 12:30     ` Paulo Alcantara
2015-06-22 12:32       ` Paolo Bonzini
2015-06-22 12:47         ` Michael S. Tsirkin
2015-06-22 13:04           ` Paolo Bonzini
2015-06-22 13:07             ` Michael S. Tsirkin
2015-06-22 13:19               ` Paulo Alcantara
2015-06-22 13:10           ` Markus Armbruster
2015-06-22  8:43   ` Michael S. Tsirkin
2015-06-22  9:45     ` Paolo Bonzini
2015-06-22 12:11       ` Michael S. Tsirkin
2015-06-22 12:36         ` Paulo Alcantara
2015-06-22 12:44           ` Michael S. Tsirkin
2015-06-22 12:59             ` Paolo Bonzini
2015-06-22 18:29   ` Paulo Alcantara
2015-06-22 23:10 ` [Qemu-devel] [PATCH v5 " Paulo Alcantara
2015-06-22 23:10   ` [Qemu-devel] [PATCH v5 2/3] target-i386: reserve RCRB mmio space in ACPI DSDT table Paulo Alcantara
2015-06-23 14:29     ` [Qemu-devel] [SeaBIOS] " Igor Mammedov
2015-06-23 15:33       ` Michael S. Tsirkin
2015-06-23 14:39     ` Igor Mammedov
2015-06-23 15:06       ` Michael S. Tsirkin
2015-06-23 15:12         ` Igor Mammedov
2015-06-23 15:29           ` Michael S. Tsirkin
2015-06-24 15:11     ` [Qemu-devel] " Michael S. Tsirkin
2015-06-24 16:00       ` Paulo Alcantara
2015-06-24 16:04         ` Michael S. Tsirkin
2015-06-22 23:10   ` [Qemu-devel] [PATCH v5 3/3] tests: add testcase for TCO watchdog emulation Paulo Alcantara
2015-06-23  6:39   ` [Qemu-devel] [PATCH v5 1/3] ich9: add TCO interface emulation Michael S. Tsirkin
2015-06-24 18:03 ` [Qemu-devel] [PATCH v6 1/2] " Paulo Alcantara
2015-06-24 18:03   ` [Qemu-devel] [PATCH v6 2/2] tests: add testcase for TCO watchdog emulation Paulo Alcantara
2015-06-27 17:56 ` [Qemu-devel] [PATCH v7 1/3] ich9: add TCO interface emulation Paulo Alcantara
2015-06-27 17:56   ` [Qemu-devel] [PATCH v7 2/3] tests: add testcase for TCO watchdog emulation Paulo Alcantara
2015-06-27 17:56   ` [Qemu-devel] [PATCH v7 3/3] ich9: implement strap SPKR pin logic Paulo Alcantara
2015-06-28  8:37     ` Michael S. Tsirkin
2015-06-28 16:21       ` Paulo Alcantara
2015-06-28 17:58 ` [Qemu-devel] [PATCH v8 1/3] ich9: add TCO interface emulation Paulo Alcantara
2015-06-28 17:58   ` [Qemu-devel] [PATCH v8 2/3] tests: add testcase for TCO watchdog emulation Paulo Alcantara
2015-06-28 17:58   ` [Qemu-devel] [PATCH v8 3/3] ich9: implement strap SPKR pin logic Paulo Alcantara
2015-07-01 13:18     ` Paolo Bonzini
2015-07-01 13:31       ` Michael S. Tsirkin
2015-07-01 13:34         ` Paolo Bonzini
2015-07-02  1:30         ` Paulo Alcantara
2015-07-02  6:55           ` Paolo Bonzini

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=20150527152328.33a8acf5@zytor.com \
    --to=pcacjr@zytor.com \
    --cc=pbonzini@redhat.com \
    --cc=pcacjr@gmail.com \
    --cc=qemu-devel@nongnu.org \
    --cc=seabios@seabios.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).