Message ID | 20100120133733.GI5238@redhat.com |
---|---|
State | New |
Headers | show |
On 01/20/2010 07:37 AM, Gleb Natapov wrote: > Windows 7 BSODs under load with HAL_RTC_IRQF_WILL_NOT_CLEAR error. > > It happens here: > hal!HalpRtcUnmaskClock: > 8281b93a 8bff mov edi,edi > 8281b93c 56 push esi > 8281b93d 33f6 xor esi,esi > 8281b93f 6a0c push 0Ch > 8281b941 e8b2ffffff call hal!CMOS_READ (8281b8f8) > 8281b946 84c0 test al,al > 8281b948 7920 jns hal!HalpRtcUnmaskClock+0x30 (8281b96a) > 8281b94a 6a0a push 0Ah > 8281b94c 46 inc esi > 8281b94d e854c8ffff call hal!KeStallExecutionProcessor (828181a6) > 8281b952 83fe64 cmp esi,64h > 8281b955 72e8 jb hal!HalpRtcUnmaskClock+0x5 (8281b93f) > 8281b957 6a00 push 0 > 8281b959 6a00 push 0 > 8281b95b 6a00 push 0 > 8281b95d 680a010000 push 10Ah > 8281b962 6a5c push 5Ch > 8281b964 ff1500c38082 call dword ptr [hal!_imp__KeBugCheckEx (8280c300)] > 8281b96a 5e pop esi > 8281b96b c3 ret > > So it loops for 100(64h) times reading register C before BSOD. Lets > reduce number of immediate reinjection well under this limit. > > Signed-off-by: Gleb Natapov<gleb@redhat.com> > Applied. Thanks. Regards, Anthony Liguori > diff --git a/hw/mc146818rtc.c b/hw/mc146818rtc.c > index e4d55c7..2616d0d 100644 > --- a/hw/mc146818rtc.c > +++ b/hw/mc146818rtc.c > @@ -30,7 +30,7 @@ > > //#define DEBUG_CMOS > > -#define RTC_REINJECT_ON_ACK_COUNT 1000 > +#define RTC_REINJECT_ON_ACK_COUNT 20 > > #define RTC_SECONDS 0 > #define RTC_SECONDS_ALARM 1 > -- > Gleb. > > > >
diff --git a/hw/mc146818rtc.c b/hw/mc146818rtc.c index e4d55c7..2616d0d 100644 --- a/hw/mc146818rtc.c +++ b/hw/mc146818rtc.c @@ -30,7 +30,7 @@ //#define DEBUG_CMOS -#define RTC_REINJECT_ON_ACK_COUNT 1000 +#define RTC_REINJECT_ON_ACK_COUNT 20 #define RTC_SECONDS 0 #define RTC_SECONDS_ALARM 1
Windows 7 BSODs under load with HAL_RTC_IRQF_WILL_NOT_CLEAR error. It happens here: hal!HalpRtcUnmaskClock: 8281b93a 8bff mov edi,edi 8281b93c 56 push esi 8281b93d 33f6 xor esi,esi 8281b93f 6a0c push 0Ch 8281b941 e8b2ffffff call hal!CMOS_READ (8281b8f8) 8281b946 84c0 test al,al 8281b948 7920 jns hal!HalpRtcUnmaskClock+0x30 (8281b96a) 8281b94a 6a0a push 0Ah 8281b94c 46 inc esi 8281b94d e854c8ffff call hal!KeStallExecutionProcessor (828181a6) 8281b952 83fe64 cmp esi,64h 8281b955 72e8 jb hal!HalpRtcUnmaskClock+0x5 (8281b93f) 8281b957 6a00 push 0 8281b959 6a00 push 0 8281b95b 6a00 push 0 8281b95d 680a010000 push 10Ah 8281b962 6a5c push 5Ch 8281b964 ff1500c38082 call dword ptr [hal!_imp__KeBugCheckEx (8280c300)] 8281b96a 5e pop esi 8281b96b c3 ret So it loops for 100(64h) times reading register C before BSOD. Lets reduce number of immediate reinjection well under this limit. Signed-off-by: Gleb Natapov <gleb@redhat.com> -- Gleb.