From: guangrong.xiao@gmail.com To: pbonzini@redhat.com, mst@redhat.com, mtosatti@redhat.com Cc: qemu-devel@nongnu.org, kvm@vger.kernel.org, yunfangtai@tencent.com, Xiao Guangrong <xiaoguangrong@tencent.com> Subject: [PATCH 3/5] mc146818rtc: properly count the time for the next interrupt Date: Wed, 12 Apr 2017 17:51:09 +0800 [thread overview] Message-ID: <20170412095111.11728-4-xiaoguangrong@tencent.com> (raw) In-Reply-To: <20170412095111.11728-1-xiaoguangrong@tencent.com> From: Tai Yunfang <yunfangtai@tencent.com> If periodic_timer_update() is called due to RegA reconfiguration, i.e, the period is updated, current time is not the start point for the next periodic timer, instead, which should start from the last interrupt, otherwise, the clock in VM will become slow This patch takes the clocks from last interrupt to current clock into account and compensates the clocks for the next interrupt, especially, if a complete interrupt was lost in this window, the time can be caught up by LOST_TICK_POLICY_SLEW [ Xiao: redesign the algorithm based on Yunfang's original work. ] Signed-off-by: Tai Yunfang <yunfangtai@tencent.com> Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com> --- hw/timer/mc146818rtc.c | 71 +++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 64 insertions(+), 7 deletions(-) diff --git a/hw/timer/mc146818rtc.c b/hw/timer/mc146818rtc.c index 649678c..3bf559d 100644 --- a/hw/timer/mc146818rtc.c +++ b/hw/timer/mc146818rtc.c @@ -161,8 +161,12 @@ static int period_code_to_clock(int period_code) return 1 << (period_code - 1); } -/* handle periodic timer */ -static void periodic_timer_update(RTCState *s, int64_t current_time) +/* + * handle periodic timer. @old_period indicates the periodic timer update + * is just due to period adjustment. + */ +static void +periodic_timer_update(RTCState *s, int64_t current_time, int old_period) { int period_code, period; int64_t cur_clock, next_irq_clock, lost_clock = 0; @@ -193,6 +197,56 @@ static void periodic_timer_update(RTCState *s, int64_t current_time) cur_clock = muldiv64(current_time, RTC_CLOCK_RATE, NANOSECONDS_PER_SECOND); + /* + * if the periodic timer's update is due to period re-configuration, + * we should count the clock since last interrupt. + */ + if (old_period) { + int64_t last_periodic_clock; + + last_periodic_clock = muldiv64(s->next_periodic_time, + RTC_CLOCK_RATE, NANOSECONDS_PER_SECOND); + /* + * if the next interrupt has not happened yet, we recall the last + * interrupt based on the original period. + */ + if (last_periodic_clock > cur_clock) { + last_periodic_clock -= period_code_to_clock(old_period); + + /* the last interrupt must have happened. */ + assert(cur_clock >= last_periodic_clock); + } + + /* calculate the clock since last interrupt. */ + lost_clock += cur_clock - last_periodic_clock; + +#ifdef TARGET_I386 + /* + * if more than period clocks were passed, i.e, the timer interrupt + * has been lost, we should catch up the time. + */ + if (s->lost_tick_policy == LOST_TICK_POLICY_SLEW && + (lost_clock / period)) { + int lost_interrupt = lost_clock / period; + + s->irq_coalesced += lost_interrupt; + lost_clock -= lost_interrupt * period; + if (lost_interrupt) { + DPRINTF_C("cmos: compensate %d interrupts, coalesced irqs " + "increased to %d\n", lost_interrupt, + s->irq_coalesced); + rtc_coalesced_timer_update(s); + } + } else +#endif + /* + * no way to compensate the interrupt if LOST_TICK_POLICY_SLEW + * is not used, we should make the time progress anyway. + */ + lost_clock = MIN(lost_clock, period); + assert(lost_clock >= 0); + } + next_irq_clock = cur_clock + period - lost_clock; s->next_periodic_time = muldiv64(next_irq_clock, NANOSECONDS_PER_SECOND, RTC_CLOCK_RATE) + 1; @@ -209,7 +263,7 @@ static void rtc_periodic_timer(void *opaque) { RTCState *s = opaque; - periodic_timer_update(s, s->next_periodic_time); + periodic_timer_update(s, s->next_periodic_time, 0); s->cmos_data[RTC_REG_C] |= REG_C_PF; if (s->cmos_data[RTC_REG_B] & REG_B_PIE) { s->cmos_data[RTC_REG_C] |= REG_C_IRQF; @@ -428,6 +482,7 @@ static void cmos_ioport_write(void *opaque, hwaddr addr, uint64_t data, unsigned size) { RTCState *s = opaque; + int cur_period; bool update_periodic_timer; if ((addr & 1) == 0) { @@ -461,6 +516,7 @@ static void cmos_ioport_write(void *opaque, hwaddr addr, } break; case RTC_REG_A: + cur_period = s->cmos_data[RTC_REG_A] & 0xf; update_periodic_timer = rtc_periodic_timer_updated_by_regA(s, data); if ((data & 0x60) == 0x60) { @@ -487,7 +543,8 @@ static void cmos_ioport_write(void *opaque, hwaddr addr, (s->cmos_data[RTC_REG_A] & REG_A_UIP); if (update_periodic_timer) { - periodic_timer_update(s, qemu_clock_get_ns(rtc_clock)); + periodic_timer_update(s, qemu_clock_get_ns(rtc_clock), + cur_period); } check_update_timer(s); @@ -523,7 +580,7 @@ static void cmos_ioport_write(void *opaque, hwaddr addr, s->cmos_data[RTC_REG_B] = data; if (update_periodic_timer) { - periodic_timer_update(s, qemu_clock_get_ns(rtc_clock)); + periodic_timer_update(s, qemu_clock_get_ns(rtc_clock), 0); } check_update_timer(s); @@ -793,7 +850,7 @@ static int rtc_post_load(void *opaque, int version_id) uint64_t now = qemu_clock_get_ns(rtc_clock); if (now < s->next_periodic_time || now > (s->next_periodic_time + get_max_clock_jump())) { - periodic_timer_update(s, qemu_clock_get_ns(rtc_clock)); + periodic_timer_update(s, qemu_clock_get_ns(rtc_clock), 0); } } @@ -858,7 +915,7 @@ static void rtc_notify_clock_reset(Notifier *notifier, void *data) int64_t now = *(int64_t *)data; rtc_set_date_from_host(ISA_DEVICE(s)); - periodic_timer_update(s, now); + periodic_timer_update(s, now, 0); check_update_timer(s); #ifdef TARGET_I386 if (s->lost_tick_policy == LOST_TICK_POLICY_SLEW) { -- 2.9.3
WARNING: multiple messages have this Message-ID (diff)
From: guangrong.xiao@gmail.com To: pbonzini@redhat.com, mst@redhat.com, mtosatti@redhat.com Cc: qemu-devel@nongnu.org, kvm@vger.kernel.org, yunfangtai@tencent.com, Xiao Guangrong <xiaoguangrong@tencent.com> Subject: [Qemu-devel] [PATCH 3/5] mc146818rtc: properly count the time for the next interrupt Date: Wed, 12 Apr 2017 17:51:09 +0800 [thread overview] Message-ID: <20170412095111.11728-4-xiaoguangrong@tencent.com> (raw) In-Reply-To: <20170412095111.11728-1-xiaoguangrong@tencent.com> From: Tai Yunfang <yunfangtai@tencent.com> If periodic_timer_update() is called due to RegA reconfiguration, i.e, the period is updated, current time is not the start point for the next periodic timer, instead, which should start from the last interrupt, otherwise, the clock in VM will become slow This patch takes the clocks from last interrupt to current clock into account and compensates the clocks for the next interrupt, especially, if a complete interrupt was lost in this window, the time can be caught up by LOST_TICK_POLICY_SLEW [ Xiao: redesign the algorithm based on Yunfang's original work. ] Signed-off-by: Tai Yunfang <yunfangtai@tencent.com> Signed-off-by: Xiao Guangrong <xiaoguangrong@tencent.com> --- hw/timer/mc146818rtc.c | 71 +++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 64 insertions(+), 7 deletions(-) diff --git a/hw/timer/mc146818rtc.c b/hw/timer/mc146818rtc.c index 649678c..3bf559d 100644 --- a/hw/timer/mc146818rtc.c +++ b/hw/timer/mc146818rtc.c @@ -161,8 +161,12 @@ static int period_code_to_clock(int period_code) return 1 << (period_code - 1); } -/* handle periodic timer */ -static void periodic_timer_update(RTCState *s, int64_t current_time) +/* + * handle periodic timer. @old_period indicates the periodic timer update + * is just due to period adjustment. + */ +static void +periodic_timer_update(RTCState *s, int64_t current_time, int old_period) { int period_code, period; int64_t cur_clock, next_irq_clock, lost_clock = 0; @@ -193,6 +197,56 @@ static void periodic_timer_update(RTCState *s, int64_t current_time) cur_clock = muldiv64(current_time, RTC_CLOCK_RATE, NANOSECONDS_PER_SECOND); + /* + * if the periodic timer's update is due to period re-configuration, + * we should count the clock since last interrupt. + */ + if (old_period) { + int64_t last_periodic_clock; + + last_periodic_clock = muldiv64(s->next_periodic_time, + RTC_CLOCK_RATE, NANOSECONDS_PER_SECOND); + /* + * if the next interrupt has not happened yet, we recall the last + * interrupt based on the original period. + */ + if (last_periodic_clock > cur_clock) { + last_periodic_clock -= period_code_to_clock(old_period); + + /* the last interrupt must have happened. */ + assert(cur_clock >= last_periodic_clock); + } + + /* calculate the clock since last interrupt. */ + lost_clock += cur_clock - last_periodic_clock; + +#ifdef TARGET_I386 + /* + * if more than period clocks were passed, i.e, the timer interrupt + * has been lost, we should catch up the time. + */ + if (s->lost_tick_policy == LOST_TICK_POLICY_SLEW && + (lost_clock / period)) { + int lost_interrupt = lost_clock / period; + + s->irq_coalesced += lost_interrupt; + lost_clock -= lost_interrupt * period; + if (lost_interrupt) { + DPRINTF_C("cmos: compensate %d interrupts, coalesced irqs " + "increased to %d\n", lost_interrupt, + s->irq_coalesced); + rtc_coalesced_timer_update(s); + } + } else +#endif + /* + * no way to compensate the interrupt if LOST_TICK_POLICY_SLEW + * is not used, we should make the time progress anyway. + */ + lost_clock = MIN(lost_clock, period); + assert(lost_clock >= 0); + } + next_irq_clock = cur_clock + period - lost_clock; s->next_periodic_time = muldiv64(next_irq_clock, NANOSECONDS_PER_SECOND, RTC_CLOCK_RATE) + 1; @@ -209,7 +263,7 @@ static void rtc_periodic_timer(void *opaque) { RTCState *s = opaque; - periodic_timer_update(s, s->next_periodic_time); + periodic_timer_update(s, s->next_periodic_time, 0); s->cmos_data[RTC_REG_C] |= REG_C_PF; if (s->cmos_data[RTC_REG_B] & REG_B_PIE) { s->cmos_data[RTC_REG_C] |= REG_C_IRQF; @@ -428,6 +482,7 @@ static void cmos_ioport_write(void *opaque, hwaddr addr, uint64_t data, unsigned size) { RTCState *s = opaque; + int cur_period; bool update_periodic_timer; if ((addr & 1) == 0) { @@ -461,6 +516,7 @@ static void cmos_ioport_write(void *opaque, hwaddr addr, } break; case RTC_REG_A: + cur_period = s->cmos_data[RTC_REG_A] & 0xf; update_periodic_timer = rtc_periodic_timer_updated_by_regA(s, data); if ((data & 0x60) == 0x60) { @@ -487,7 +543,8 @@ static void cmos_ioport_write(void *opaque, hwaddr addr, (s->cmos_data[RTC_REG_A] & REG_A_UIP); if (update_periodic_timer) { - periodic_timer_update(s, qemu_clock_get_ns(rtc_clock)); + periodic_timer_update(s, qemu_clock_get_ns(rtc_clock), + cur_period); } check_update_timer(s); @@ -523,7 +580,7 @@ static void cmos_ioport_write(void *opaque, hwaddr addr, s->cmos_data[RTC_REG_B] = data; if (update_periodic_timer) { - periodic_timer_update(s, qemu_clock_get_ns(rtc_clock)); + periodic_timer_update(s, qemu_clock_get_ns(rtc_clock), 0); } check_update_timer(s); @@ -793,7 +850,7 @@ static int rtc_post_load(void *opaque, int version_id) uint64_t now = qemu_clock_get_ns(rtc_clock); if (now < s->next_periodic_time || now > (s->next_periodic_time + get_max_clock_jump())) { - periodic_timer_update(s, qemu_clock_get_ns(rtc_clock)); + periodic_timer_update(s, qemu_clock_get_ns(rtc_clock), 0); } } @@ -858,7 +915,7 @@ static void rtc_notify_clock_reset(Notifier *notifier, void *data) int64_t now = *(int64_t *)data; rtc_set_date_from_host(ISA_DEVICE(s)); - periodic_timer_update(s, now); + periodic_timer_update(s, now, 0); check_update_timer(s); #ifdef TARGET_I386 if (s->lost_tick_policy == LOST_TICK_POLICY_SLEW) { -- 2.9.3
next prev parent reply other threads:[~2017-04-12 9:51 UTC|newest] Thread overview: 59+ messages / expand[flat|nested] mbox.gz Atom feed top 2017-04-12 9:51 [PATCH 0/5] mc146818rtc: fix Windows VM clock faster guangrong.xiao 2017-04-12 9:51 ` [Qemu-devel] " guangrong.xiao 2017-04-12 9:51 ` [PATCH 1/5] mc146818rtc: update periodic timer only if it is needed guangrong.xiao 2017-04-12 9:51 ` [Qemu-devel] " guangrong.xiao 2017-05-03 15:42 ` Paolo Bonzini 2017-05-03 15:42 ` [Qemu-devel] " Paolo Bonzini 2017-05-04 3:27 ` Xiao Guangrong 2017-05-04 3:27 ` [Qemu-devel] " Xiao Guangrong 2017-04-12 9:51 ` [PATCH 2/5] mc146818rtc: fix clock lost after scaling coalesced irq guangrong.xiao 2017-04-12 9:51 ` [Qemu-devel] " guangrong.xiao 2017-05-03 15:15 ` Paolo Bonzini 2017-05-03 15:15 ` [Qemu-devel] " Paolo Bonzini 2017-05-04 2:51 ` Xiao Guangrong 2017-05-04 2:51 ` [Qemu-devel] " Xiao Guangrong 2017-04-12 9:51 ` guangrong.xiao [this message] 2017-04-12 9:51 ` [Qemu-devel] [PATCH 3/5] mc146818rtc: properly count the time for the next interrupt guangrong.xiao 2017-05-03 15:32 ` Paolo Bonzini 2017-05-03 15:32 ` [Qemu-devel] " Paolo Bonzini 2017-05-04 2:54 ` Xiao Guangrong 2017-05-04 2:54 ` [Qemu-devel] " Xiao Guangrong 2017-05-04 12:02 ` Paolo Bonzini 2017-05-04 12:02 ` [Qemu-devel] " Paolo Bonzini 2017-04-12 9:51 ` [PATCH 4/5] mc146818rtc: move x86 specific code out of periodic_timer_update guangrong.xiao 2017-04-12 9:51 ` [Qemu-devel] " guangrong.xiao 2017-05-03 15:39 ` Paolo Bonzini 2017-05-03 15:39 ` [Qemu-devel] " Paolo Bonzini 2017-05-04 3:25 ` Xiao Guangrong 2017-05-04 3:25 ` [Qemu-devel] " Xiao Guangrong 2017-05-04 7:08 ` Paolo Bonzini 2017-05-04 7:08 ` [Qemu-devel] " Paolo Bonzini 2017-04-12 9:51 ` [PATCH 5/5] mc146818rtc: embrace all x86 specific code guangrong.xiao 2017-04-12 9:51 ` [Qemu-devel] " guangrong.xiao 2017-04-13 6:37 ` [PATCH 0/5] mc146818rtc: fix Windows VM clock faster Paolo Bonzini 2017-04-13 6:37 ` [Qemu-devel] " Paolo Bonzini 2017-04-13 8:39 ` Xiao Guangrong 2017-04-13 8:39 ` [Qemu-devel] " Xiao Guangrong 2017-04-13 8:52 ` Xiao Guangrong 2017-04-13 8:52 ` [Qemu-devel] " Xiao Guangrong 2017-04-13 9:05 ` 答复: " Zhanghailiang 2017-04-13 9:05 ` [Qemu-devel] " Zhanghailiang 2017-04-13 9:18 ` Xiao Guangrong 2017-04-13 9:18 ` [Qemu-devel] " Xiao Guangrong 2017-04-13 9:29 ` Hailiang Zhang 2017-04-13 9:29 ` [Qemu-devel] " Hailiang Zhang 2017-04-13 9:35 ` Xiao Guangrong 2017-04-13 9:35 ` [Qemu-devel] " Xiao Guangrong 2017-04-13 9:38 ` Hailiang Zhang 2017-04-13 9:38 ` [Qemu-devel] " Hailiang Zhang 2017-04-19 2:02 ` Xiao Guangrong 2017-04-19 2:02 ` [Qemu-devel] " Xiao Guangrong 2017-04-19 10:41 ` Hailiang Zhang 2017-04-19 10:41 ` [Qemu-devel] " Hailiang Zhang 2017-04-19 11:13 ` Xiao Guangrong 2017-04-19 11:13 ` [Qemu-devel] " Xiao Guangrong 2017-04-19 16:44 ` Paolo Bonzini 2017-04-19 16:44 ` [Qemu-devel] " Paolo Bonzini 2017-04-14 5:09 ` Paolo Bonzini 2017-04-14 5:09 ` [Qemu-devel] " Paolo Bonzini 2017-04-14 6:07 ` Xiao Guangrong
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=20170412095111.11728-4-xiaoguangrong@tencent.com \ --to=guangrong.xiao@gmail.com \ --cc=kvm@vger.kernel.org \ --cc=mst@redhat.com \ --cc=mtosatti@redhat.com \ --cc=pbonzini@redhat.com \ --cc=qemu-devel@nongnu.org \ --cc=xiaoguangrong@tencent.com \ --cc=yunfangtai@tencent.com \ /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: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
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.