* [PATCH RFC] x86/ACPI: allow CMOS RTC use even when ACPI says there is none
@ 2014-06-26 10:28 Jan Beulich
2014-06-26 11:12 ` Tim Deegan
0 siblings, 1 reply; 4+ messages in thread
From: Jan Beulich @ 2014-06-26 10:28 UTC (permalink / raw)
To: xen-devel
[-- Attachment #1: Type: text/plain, Size: 5218 bytes --]
HP is setting the ACPI_FADT_NO_CMOS_RTC flag on newer systems,
regardless of whether they're being booted from UEFI. Add a command
line option to allow probing for a working RTC in that case.
Note that this still has some debugging code (//temp) in it (and no
S-o-b for that reason).
--- a/docs/misc/xen-command-line.markdown
+++ b/docs/misc/xen-command-line.markdown
@@ -218,6 +218,14 @@ If set, override Xen's calculation of th
If set, override Xen's default choice for the platform timer.
+### cmos-rtc-probe
+> `= <boolean>`
+
+> Default: `false`
+
+Flag to indicate whether to probe for a CMOS Real Time Clock irrespective of
+ACPI indicating none to be there.
+
### com1,com2
> `= <baud>[/<clock_hz>][,[DPS][,[<io-base>|pci|amt][,[<irq>][,[<port-bdf>][,[<bridge-bdf>]]]]]]`
--- a/xen/arch/x86/time.c
+++ b/xen/arch/x86/time.c
@@ -654,37 +654,40 @@ mktime (unsigned int year, unsigned int
)*60 + sec; /* finally seconds */
}
-static unsigned long __get_cmos_time(void)
-{
+struct rtc_time {
unsigned int year, mon, day, hour, min, sec;
+};
- sec = CMOS_READ(RTC_SECONDS);
- min = CMOS_READ(RTC_MINUTES);
- hour = CMOS_READ(RTC_HOURS);
- day = CMOS_READ(RTC_DAY_OF_MONTH);
- mon = CMOS_READ(RTC_MONTH);
- year = CMOS_READ(RTC_YEAR);
+static void __get_cmos_time(struct rtc_time *rtc)
+{
+ rtc->sec = CMOS_READ(RTC_SECONDS);
+ rtc->min = CMOS_READ(RTC_MINUTES);
+ rtc->hour = CMOS_READ(RTC_HOURS);
+ rtc->day = CMOS_READ(RTC_DAY_OF_MONTH);
+ rtc->mon = CMOS_READ(RTC_MONTH);
+ rtc->year = CMOS_READ(RTC_YEAR);
if ( RTC_ALWAYS_BCD || !(CMOS_READ(RTC_CONTROL) & RTC_DM_BINARY) )
{
- BCD_TO_BIN(sec);
- BCD_TO_BIN(min);
- BCD_TO_BIN(hour);
- BCD_TO_BIN(day);
- BCD_TO_BIN(mon);
- BCD_TO_BIN(year);
+ BCD_TO_BIN(rtc->sec);
+ BCD_TO_BIN(rtc->min);
+ BCD_TO_BIN(rtc->hour);
+ BCD_TO_BIN(rtc->day);
+ BCD_TO_BIN(rtc->mon);
+ BCD_TO_BIN(rtc->year);
}
- if ( (year += 1900) < 1970 )
- year += 100;
-
- return mktime(year, mon, day, hour, min, sec);
+ if ( (rtc->year += 1900) < 1970 )
+ rtc->year += 100;
}
static unsigned long get_cmos_time(void)
{
unsigned long res, flags;
- int i;
+ struct rtc_time rtc;
+ unsigned int seconds = 60;
+ static bool_t __read_mostly cmos_rtc_probe;
+ boolean_param("cmos-rtc-probe", cmos_rtc_probe);
if ( efi_enabled )
{
@@ -693,23 +696,60 @@ static unsigned long get_cmos_time(void)
return res;
}
- if ( unlikely(acpi_gbl_FADT.boot_flags & ACPI_FADT_NO_CMOS_RTC) )
- panic("System without CMOS RTC must be booted from EFI");
-
- spin_lock_irqsave(&rtc_lock, flags);
-
- /* read RTC exactly on falling edge of update flag */
- for ( i = 0 ; i < 1000000 ; i++ ) /* may take up to 1 second... */
- if ( (CMOS_READ(RTC_FREQ_SELECT) & RTC_UIP) )
+ if ( likely(!(acpi_gbl_FADT.boot_flags & ACPI_FADT_NO_CMOS_RTC)) )
+ cmos_rtc_probe = 0;
+ else if ( unlikely(!cmos_rtc_probe) )
+ panic("System with no CMOS RTC advertised must be booted from EFI"
+ " or with command line option \"cmos-rtc-probe\"");
+cmos_rtc_probe = 1;//temp
+
+ for ( ; ; )
+ {
+ s_time_t start, t1, t2;
+
+ spin_lock_irqsave(&rtc_lock, flags);
+
+ /* read RTC exactly on falling edge of update flag */
+ start = NOW();
+ do { /* may take up to 1 second... */
+ t1 = NOW() - start;
+ } while ( !(CMOS_READ(RTC_FREQ_SELECT) & RTC_UIP) &&
+ t1 <= SECONDS(1) );
+
+ start = NOW();
+ do { /* must try at least 2.228 ms */
+ t2 = NOW() - start;
+ } while ( (CMOS_READ(RTC_FREQ_SELECT) & RTC_UIP) &&
+ t2 < MILLISECS(3) );
+
+ __get_cmos_time(&rtc);
+
+ spin_unlock_irqrestore(&rtc_lock, flags);
+printk("RTC: %04u-%02u-%02u %02u:%02u:%02u (%lu/%lu)\n", rtc.year, rtc.mon, rtc.day, rtc.hour, rtc.min, rtc.sec, t1, t2);//temp
+
+ if ( likely(!cmos_rtc_probe) ||
+ t1 > SECONDS(1) || t2 >= MILLISECS(3) ||
+ rtc.sec >= 60 || rtc.min >= 60 || rtc.hour >= 24 ||
+ !rtc.day || rtc.day > 31 ||
+ !rtc.mon || rtc.mon > 12 )
break;
- for ( i = 0 ; i < 1000000 ; i++ ) /* must try at least 2.228 ms */
- if ( !(CMOS_READ(RTC_FREQ_SELECT) & RTC_UIP) )
+
+ if ( seconds < 60 )
+ {
+ if ( rtc.sec != seconds )
+ cmos_rtc_probe = 0;
break;
+ }
+
+ process_pending_softirqs();
+
+ seconds = rtc.sec;
+ }
- res = __get_cmos_time();
+ if ( unlikely(cmos_rtc_probe) )
+ panic("No CMOS RTC found - system must be booted from EFI");
- spin_unlock_irqrestore(&rtc_lock, flags);
- return res;
+ return mktime(rtc.year, rtc.mon, rtc.day, rtc.hour, rtc.min, rtc.sec);
}
/***************************************************************************
[-- Attachment #2: x86-RTC-probe.patch --]
[-- Type: text/plain, Size: 5282 bytes --]
x86/ACPI: allow CMOS RTC use even when ACPI says there is none
HP is setting the ACPI_FADT_NO_CMOS_RTC flag on newer systems,
regardless of whether they're being booted from UEFI. Add a command
line option to allow probing for a working RTC in that case.
Note that this still has some debugging code (//temp) in it (and no
S-o-b for that reason).
--- a/docs/misc/xen-command-line.markdown
+++ b/docs/misc/xen-command-line.markdown
@@ -218,6 +218,14 @@ If set, override Xen's calculation of th
If set, override Xen's default choice for the platform timer.
+### cmos-rtc-probe
+> `= <boolean>`
+
+> Default: `false`
+
+Flag to indicate whether to probe for a CMOS Real Time Clock irrespective of
+ACPI indicating none to be there.
+
### com1,com2
> `= <baud>[/<clock_hz>][,[DPS][,[<io-base>|pci|amt][,[<irq>][,[<port-bdf>][,[<bridge-bdf>]]]]]]`
--- a/xen/arch/x86/time.c
+++ b/xen/arch/x86/time.c
@@ -654,37 +654,40 @@ mktime (unsigned int year, unsigned int
)*60 + sec; /* finally seconds */
}
-static unsigned long __get_cmos_time(void)
-{
+struct rtc_time {
unsigned int year, mon, day, hour, min, sec;
+};
- sec = CMOS_READ(RTC_SECONDS);
- min = CMOS_READ(RTC_MINUTES);
- hour = CMOS_READ(RTC_HOURS);
- day = CMOS_READ(RTC_DAY_OF_MONTH);
- mon = CMOS_READ(RTC_MONTH);
- year = CMOS_READ(RTC_YEAR);
+static void __get_cmos_time(struct rtc_time *rtc)
+{
+ rtc->sec = CMOS_READ(RTC_SECONDS);
+ rtc->min = CMOS_READ(RTC_MINUTES);
+ rtc->hour = CMOS_READ(RTC_HOURS);
+ rtc->day = CMOS_READ(RTC_DAY_OF_MONTH);
+ rtc->mon = CMOS_READ(RTC_MONTH);
+ rtc->year = CMOS_READ(RTC_YEAR);
if ( RTC_ALWAYS_BCD || !(CMOS_READ(RTC_CONTROL) & RTC_DM_BINARY) )
{
- BCD_TO_BIN(sec);
- BCD_TO_BIN(min);
- BCD_TO_BIN(hour);
- BCD_TO_BIN(day);
- BCD_TO_BIN(mon);
- BCD_TO_BIN(year);
+ BCD_TO_BIN(rtc->sec);
+ BCD_TO_BIN(rtc->min);
+ BCD_TO_BIN(rtc->hour);
+ BCD_TO_BIN(rtc->day);
+ BCD_TO_BIN(rtc->mon);
+ BCD_TO_BIN(rtc->year);
}
- if ( (year += 1900) < 1970 )
- year += 100;
-
- return mktime(year, mon, day, hour, min, sec);
+ if ( (rtc->year += 1900) < 1970 )
+ rtc->year += 100;
}
static unsigned long get_cmos_time(void)
{
unsigned long res, flags;
- int i;
+ struct rtc_time rtc;
+ unsigned int seconds = 60;
+ static bool_t __read_mostly cmos_rtc_probe;
+ boolean_param("cmos-rtc-probe", cmos_rtc_probe);
if ( efi_enabled )
{
@@ -693,23 +696,60 @@ static unsigned long get_cmos_time(void)
return res;
}
- if ( unlikely(acpi_gbl_FADT.boot_flags & ACPI_FADT_NO_CMOS_RTC) )
- panic("System without CMOS RTC must be booted from EFI");
-
- spin_lock_irqsave(&rtc_lock, flags);
-
- /* read RTC exactly on falling edge of update flag */
- for ( i = 0 ; i < 1000000 ; i++ ) /* may take up to 1 second... */
- if ( (CMOS_READ(RTC_FREQ_SELECT) & RTC_UIP) )
+ if ( likely(!(acpi_gbl_FADT.boot_flags & ACPI_FADT_NO_CMOS_RTC)) )
+ cmos_rtc_probe = 0;
+ else if ( unlikely(!cmos_rtc_probe) )
+ panic("System with no CMOS RTC advertised must be booted from EFI"
+ " or with command line option \"cmos-rtc-probe\"");
+cmos_rtc_probe = 1;//temp
+
+ for ( ; ; )
+ {
+ s_time_t start, t1, t2;
+
+ spin_lock_irqsave(&rtc_lock, flags);
+
+ /* read RTC exactly on falling edge of update flag */
+ start = NOW();
+ do { /* may take up to 1 second... */
+ t1 = NOW() - start;
+ } while ( !(CMOS_READ(RTC_FREQ_SELECT) & RTC_UIP) &&
+ t1 <= SECONDS(1) );
+
+ start = NOW();
+ do { /* must try at least 2.228 ms */
+ t2 = NOW() - start;
+ } while ( (CMOS_READ(RTC_FREQ_SELECT) & RTC_UIP) &&
+ t2 < MILLISECS(3) );
+
+ __get_cmos_time(&rtc);
+
+ spin_unlock_irqrestore(&rtc_lock, flags);
+printk("RTC: %04u-%02u-%02u %02u:%02u:%02u (%lu/%lu)\n", rtc.year, rtc.mon, rtc.day, rtc.hour, rtc.min, rtc.sec, t1, t2);//temp
+
+ if ( likely(!cmos_rtc_probe) ||
+ t1 > SECONDS(1) || t2 >= MILLISECS(3) ||
+ rtc.sec >= 60 || rtc.min >= 60 || rtc.hour >= 24 ||
+ !rtc.day || rtc.day > 31 ||
+ !rtc.mon || rtc.mon > 12 )
break;
- for ( i = 0 ; i < 1000000 ; i++ ) /* must try at least 2.228 ms */
- if ( !(CMOS_READ(RTC_FREQ_SELECT) & RTC_UIP) )
+
+ if ( seconds < 60 )
+ {
+ if ( rtc.sec != seconds )
+ cmos_rtc_probe = 0;
break;
+ }
+
+ process_pending_softirqs();
+
+ seconds = rtc.sec;
+ }
- res = __get_cmos_time();
+ if ( unlikely(cmos_rtc_probe) )
+ panic("No CMOS RTC found - system must be booted from EFI");
- spin_unlock_irqrestore(&rtc_lock, flags);
- return res;
+ return mktime(rtc.year, rtc.mon, rtc.day, rtc.hour, rtc.min, rtc.sec);
}
/***************************************************************************
[-- Attachment #3: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH RFC] x86/ACPI: allow CMOS RTC use even when ACPI says there is none
2014-06-26 10:28 [PATCH RFC] x86/ACPI: allow CMOS RTC use even when ACPI says there is none Jan Beulich
@ 2014-06-26 11:12 ` Tim Deegan
2014-06-26 11:41 ` Jan Beulich
0 siblings, 1 reply; 4+ messages in thread
From: Tim Deegan @ 2014-06-26 11:12 UTC (permalink / raw)
To: Jan Beulich; +Cc: xen-devel
At 11:28 +0100 on 26 Jun (1403778496), Jan Beulich wrote:
> HP is setting the ACPI_FADT_NO_CMOS_RTC flag on newer systems,
> regardless of whether they're being booted from UEFI. Add a command
> line option to allow probing for a working RTC in that case.
I like this general direction. I would add a printk in the case where
we are probing, since the probe will stall boot for >= 1 sec.
Maybe also worth checking that the times returned on the two passes
actually make sense (i.e. that the difference in RTC times matches the
difference in NOW() to within +/- 10ms).
Thirdly, since get_cmos_time() can be called more than once, I think
that just leaving cmos_rtc_probe == 0 could make us panic on the second
invocation.
Cheers,
Tim.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH RFC] x86/ACPI: allow CMOS RTC use even when ACPI says there is none
2014-06-26 11:12 ` Tim Deegan
@ 2014-06-26 11:41 ` Jan Beulich
2014-06-26 11:50 ` Tim Deegan
0 siblings, 1 reply; 4+ messages in thread
From: Jan Beulich @ 2014-06-26 11:41 UTC (permalink / raw)
To: Tim Deegan; +Cc: xen-devel
>>> On 26.06.14 at 13:12, <tim@xen.org> wrote:
> At 11:28 +0100 on 26 Jun (1403778496), Jan Beulich wrote:
>> HP is setting the ACPI_FADT_NO_CMOS_RTC flag on newer systems,
>> regardless of whether they're being booted from UEFI. Add a command
>> line option to allow probing for a working RTC in that case.
>
> I like this general direction. I would add a printk in the case where
> we are probing, since the probe will stall boot for >= 1 sec.
Right now we stall for up to 1 sec already; since the probing
doesn't happen without being asked for, I don't view it as a
problem that now we stall for up to 2 sec.
> Maybe also worth checking that the times returned on the two passes
> actually make sense (i.e. that the difference in RTC times matches the
> difference in NOW() to within +/- 10ms).
I had considered this too, but intentionally dropped the idea, so that
odd behavior when running in nested virtualization scenarios can be
avoided.
> Thirdly, since get_cmos_time() can be called more than once, I think
> that just leaving cmos_rtc_probe == 0 could make us panic on the second
> invocation.
Very good point, this definitely needs fixing. I guess I'll just amend
the condition with a system_state check.
Jan
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH RFC] x86/ACPI: allow CMOS RTC use even when ACPI says there is none
2014-06-26 11:41 ` Jan Beulich
@ 2014-06-26 11:50 ` Tim Deegan
0 siblings, 0 replies; 4+ messages in thread
From: Tim Deegan @ 2014-06-26 11:50 UTC (permalink / raw)
To: Jan Beulich; +Cc: xen-devel
At 12:41 +0100 on 26 Jun (1403782897), Jan Beulich wrote:
> >>> On 26.06.14 at 13:12, <tim@xen.org> wrote:
> > At 11:28 +0100 on 26 Jun (1403778496), Jan Beulich wrote:
> >> HP is setting the ACPI_FADT_NO_CMOS_RTC flag on newer systems,
> >> regardless of whether they're being booted from UEFI. Add a command
> >> line option to allow probing for a working RTC in that case.
> >
> > I like this general direction. I would add a printk in the case where
> > we are probing, since the probe will stall boot for >= 1 sec.
>
> Right now we stall for up to 1 sec already; since the probing
> doesn't happen without being asked for, I don't view it as a
> problem that now we stall for up to 2 sec.
Ack.
> > Maybe also worth checking that the times returned on the two passes
> > actually make sense (i.e. that the difference in RTC times matches the
> > difference in NOW() to within +/- 10ms).
>
> I had considered this too, but intentionally dropped the idea, so that
> odd behavior when running in nested virtualization scenarios can be
> avoided.
I guess so. We'd have to be running inside a hypervisor that has the
same buggy tables as these HP machines, but OTOH anything that passes
the tests you already have looks quite a lot like an RTC. :)
Tim.
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2014-06-26 11:50 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-06-26 10:28 [PATCH RFC] x86/ACPI: allow CMOS RTC use even when ACPI says there is none Jan Beulich
2014-06-26 11:12 ` Tim Deegan
2014-06-26 11:41 ` Jan Beulich
2014-06-26 11:50 ` Tim Deegan
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.