From: Marc Zyngier <marc.zyngier@arm.com> To: Christoffer Dall <cdall@linaro.org> Cc: kvmarm@lists.cs.columbia.edu, linux-arm-kernel@lists.infradead.org, kvm@vger.kernel.org, Will Deacon <will.deacon@arm.com>, Catalin Marinas <catalin.marinas@arm.com>, Mark Rutland <mark.rutland@arm.com> Subject: Re: [PATCH v3 03/20] arm64: Use the physical counter when available for read_cycles Date: Wed, 18 Oct 2017 16:52:26 +0100 [thread overview] Message-ID: <878tg8jvwl.fsf@on-the-bus.cambridge.arm.com> (raw) In-Reply-To: <20171018113405.GA8900@cbox> (Christoffer Dall's message of "Wed, 18 Oct 2017 13:34:05 +0200") On Wed, Oct 18 2017 at 1:34:05 pm BST, Christoffer Dall <cdall@linaro.org> wrote: > On Mon, Oct 09, 2017 at 05:21:24PM +0100, Marc Zyngier wrote: >> On 23/09/17 01:41, Christoffer Dall wrote: >> > Currently get_cycles() is hardwired to arch_counter_get_cntvct() on >> > arm64, but as we move to using the physical timer for the in-kernel >> > time-keeping, we need to make that more flexible. >> > >> > First, we need to make sure the physical counter can be read on equal >> > terms to the virtual counter, which includes adding physical counter >> > read functions for timers that require errata. >> > >> > Second, we need to make a choice between reading the physical vs virtual >> > counter, depending on which timer is used for time keeping in the kernel >> > otherwise. We can do this using a static key to avoid a performance >> > penalty during runtime when reading the counter. >> > >> > Cc: Catalin Marinas <catalin.marinas@arm.com> >> > Cc: Will Deacon <will.deacon@arm.com> >> > Cc: Mark Rutland <mark.rutland@arm.com> >> > Cc: Marc Zyngier <marc.zyngier@arm.com> >> > Signed-off-by: Christoffer Dall <cdall@linaro.org> [...] >> In my reply to patch #2, I had the following hunk: >> >> @@ -310,7 +329,7 @@ static void erratum_set_next_event_tval_generic(const int access, unsigned long >> struct clock_event_device *clk) >> { >> unsigned long ctrl; >> - u64 cval = evt + arch_counter_get_cntvct(); >> + u64 cval = evt + arch_timer_read_counter(); >> >> ctrl = arch_timer_reg_read(access, ARCH_TIMER_REG_CTRL, clk); >> ctrl |= ARCH_TIMER_CTRL_ENABLE; >> >> Once we start using a different timer, this could well have an effect... >> > > Right, but wouldn't the following be a more correct way to go about it then: > > diff --git a/drivers/clocksource/arm_arch_timer.c b/drivers/clocksource/arm_arch_timer.c > index 9a7b359..07f19db 100644 > --- a/drivers/clocksource/arm_arch_timer.c > +++ b/drivers/clocksource/arm_arch_timer.c > @@ -329,16 +329,19 @@ static void erratum_set_next_event_tval_generic(const int access, unsigned long > struct clock_event_device *clk) > { > unsigned long ctrl; > - u64 cval = evt + arch_timer_read_counter(); > + u64 cval; > > ctrl = arch_timer_reg_read(access, ARCH_TIMER_REG_CTRL, clk); > ctrl |= ARCH_TIMER_CTRL_ENABLE; > ctrl &= ~ARCH_TIMER_CTRL_IT_MASK; > > - if (access == ARCH_TIMER_PHYS_ACCESS) > + if (access == ARCH_TIMER_PHYS_ACCESS) { > + cval = evt + arch_counter_get_cntpct(); > write_sysreg(cval, cntp_cval_el0); > - else > + } else { > + cval = evt + arch_counter_get_cntvct(); > write_sysreg(cval, cntv_cval_el0); > + } > > arch_timer_reg_write(access, ARCH_TIMER_REG_CTRL, ctrl, clk); > } Yup, that's much better. Thanks, M. -- Jazz is not dead, it just smell funny.
WARNING: multiple messages have this Message-ID (diff)
From: marc.zyngier@arm.com (Marc Zyngier) To: linux-arm-kernel@lists.infradead.org Subject: [PATCH v3 03/20] arm64: Use the physical counter when available for read_cycles Date: Wed, 18 Oct 2017 16:52:26 +0100 [thread overview] Message-ID: <878tg8jvwl.fsf@on-the-bus.cambridge.arm.com> (raw) In-Reply-To: <20171018113405.GA8900@cbox> (Christoffer Dall's message of "Wed, 18 Oct 2017 13:34:05 +0200") On Wed, Oct 18 2017 at 1:34:05 pm BST, Christoffer Dall <cdall@linaro.org> wrote: > On Mon, Oct 09, 2017 at 05:21:24PM +0100, Marc Zyngier wrote: >> On 23/09/17 01:41, Christoffer Dall wrote: >> > Currently get_cycles() is hardwired to arch_counter_get_cntvct() on >> > arm64, but as we move to using the physical timer for the in-kernel >> > time-keeping, we need to make that more flexible. >> > >> > First, we need to make sure the physical counter can be read on equal >> > terms to the virtual counter, which includes adding physical counter >> > read functions for timers that require errata. >> > >> > Second, we need to make a choice between reading the physical vs virtual >> > counter, depending on which timer is used for time keeping in the kernel >> > otherwise. We can do this using a static key to avoid a performance >> > penalty during runtime when reading the counter. >> > >> > Cc: Catalin Marinas <catalin.marinas@arm.com> >> > Cc: Will Deacon <will.deacon@arm.com> >> > Cc: Mark Rutland <mark.rutland@arm.com> >> > Cc: Marc Zyngier <marc.zyngier@arm.com> >> > Signed-off-by: Christoffer Dall <cdall@linaro.org> [...] >> In my reply to patch #2, I had the following hunk: >> >> @@ -310,7 +329,7 @@ static void erratum_set_next_event_tval_generic(const int access, unsigned long >> struct clock_event_device *clk) >> { >> unsigned long ctrl; >> - u64 cval = evt + arch_counter_get_cntvct(); >> + u64 cval = evt + arch_timer_read_counter(); >> >> ctrl = arch_timer_reg_read(access, ARCH_TIMER_REG_CTRL, clk); >> ctrl |= ARCH_TIMER_CTRL_ENABLE; >> >> Once we start using a different timer, this could well have an effect... >> > > Right, but wouldn't the following be a more correct way to go about it then: > > diff --git a/drivers/clocksource/arm_arch_timer.c b/drivers/clocksource/arm_arch_timer.c > index 9a7b359..07f19db 100644 > --- a/drivers/clocksource/arm_arch_timer.c > +++ b/drivers/clocksource/arm_arch_timer.c > @@ -329,16 +329,19 @@ static void erratum_set_next_event_tval_generic(const int access, unsigned long > struct clock_event_device *clk) > { > unsigned long ctrl; > - u64 cval = evt + arch_timer_read_counter(); > + u64 cval; > > ctrl = arch_timer_reg_read(access, ARCH_TIMER_REG_CTRL, clk); > ctrl |= ARCH_TIMER_CTRL_ENABLE; > ctrl &= ~ARCH_TIMER_CTRL_IT_MASK; > > - if (access == ARCH_TIMER_PHYS_ACCESS) > + if (access == ARCH_TIMER_PHYS_ACCESS) { > + cval = evt + arch_counter_get_cntpct(); > write_sysreg(cval, cntp_cval_el0); > - else > + } else { > + cval = evt + arch_counter_get_cntvct(); > write_sysreg(cval, cntv_cval_el0); > + } > > arch_timer_reg_write(access, ARCH_TIMER_REG_CTRL, ctrl, clk); > } Yup, that's much better. Thanks, M. -- Jazz is not dead, it just smell funny.
next prev parent reply other threads:[~2017-10-18 15:52 UTC|newest] Thread overview: 110+ messages / expand[flat|nested] mbox.gz Atom feed top 2017-09-23 0:41 [PATCH v3 00/20] KVM: arm/arm64: Optimize arch timer register handling Christoffer Dall 2017-09-23 0:41 ` Christoffer Dall 2017-09-23 0:41 ` [PATCH v3 01/20] irqchip/gic: Deal with broken firmware exposing only 4kB of GICv2 CPU interface Christoffer Dall 2017-09-23 0:41 ` Christoffer Dall 2017-09-23 0:41 ` [PATCH v3 02/20] arm64: Use physical counter for in-kernel reads Christoffer Dall 2017-09-23 0:41 ` Christoffer Dall 2017-10-09 16:10 ` Marc Zyngier 2017-10-09 16:10 ` Marc Zyngier 2017-10-17 15:33 ` Will Deacon 2017-10-17 15:33 ` Will Deacon 2017-10-18 10:00 ` Christoffer Dall 2017-10-18 10:00 ` Christoffer Dall 2017-09-23 0:41 ` [PATCH v3 03/20] arm64: Use the physical counter when available for read_cycles Christoffer Dall 2017-09-23 0:41 ` Christoffer Dall 2017-10-09 16:21 ` Marc Zyngier 2017-10-09 16:21 ` Marc Zyngier 2017-10-18 11:34 ` Christoffer Dall 2017-10-18 11:34 ` Christoffer Dall 2017-10-18 15:52 ` Marc Zyngier [this message] 2017-10-18 15:52 ` Marc Zyngier 2017-09-23 0:41 ` [PATCH v3 04/20] KVM: arm/arm64: Guard kvm_vgic_map_is_active against !vgic_initialized Christoffer Dall 2017-09-23 0:41 ` Christoffer Dall 2017-10-09 16:22 ` Marc Zyngier 2017-10-09 16:22 ` Marc Zyngier 2017-09-23 0:41 ` [PATCH v3 05/20] KVM: arm/arm64: Support calling vgic_update_irq_pending from irq context Christoffer Dall 2017-09-23 0:41 ` Christoffer Dall 2017-10-09 16:37 ` Marc Zyngier 2017-10-09 16:37 ` Marc Zyngier 2017-10-18 11:54 ` Christoffer Dall 2017-10-18 11:54 ` Christoffer Dall 2017-09-23 0:41 ` [PATCH v3 06/20] KVM: arm/arm64: Check that system supports split eoi/deactivate Christoffer Dall 2017-09-23 0:41 ` Christoffer Dall 2017-10-09 16:47 ` Marc Zyngier 2017-10-09 16:47 ` Marc Zyngier 2017-10-18 13:41 ` Christoffer Dall 2017-10-18 13:41 ` Christoffer Dall 2017-10-18 16:03 ` Marc Zyngier 2017-10-18 16:03 ` Marc Zyngier 2017-10-18 19:16 ` Christoffer Dall 2017-10-18 19:16 ` Christoffer Dall 2017-09-23 0:41 ` [PATCH v3 07/20] KVM: arm/arm64: Make timer_arm and timer_disarm helpers more generic Christoffer Dall 2017-09-23 0:41 ` Christoffer Dall 2017-10-09 17:05 ` Marc Zyngier 2017-10-09 17:05 ` Marc Zyngier 2017-10-18 16:47 ` Christoffer Dall 2017-10-18 16:47 ` Christoffer Dall 2017-10-18 16:53 ` Marc Zyngier 2017-10-18 16:53 ` Marc Zyngier 2017-09-23 0:41 ` [PATCH v3 08/20] KVM: arm/arm64: Rename soft timer to bg_timer Christoffer Dall 2017-09-23 0:41 ` Christoffer Dall 2017-10-09 17:06 ` Marc Zyngier 2017-10-09 17:06 ` Marc Zyngier 2017-09-23 0:41 ` [PATCH v3 09/20] KVM: arm/arm64: Use separate timer for phys timer emulation Christoffer Dall 2017-09-23 0:41 ` Christoffer Dall 2017-10-09 17:23 ` Marc Zyngier 2017-10-09 17:23 ` Marc Zyngier 2017-10-19 7:38 ` Christoffer Dall 2017-10-19 7:38 ` Christoffer Dall 2017-09-23 0:41 ` [PATCH v3 10/20] KVM: arm/arm64: Move timer/vgic flush/sync under disabled irq Christoffer Dall 2017-09-23 0:41 ` Christoffer Dall 2017-10-09 17:34 ` Marc Zyngier 2017-10-09 17:34 ` Marc Zyngier 2017-09-23 0:41 ` [PATCH v3 11/20] KVM: arm/arm64: Move timer save/restore out of the hyp code Christoffer Dall 2017-09-23 0:41 ` Christoffer Dall 2017-10-09 17:47 ` Marc Zyngier 2017-10-09 17:47 ` Marc Zyngier 2017-10-19 7:46 ` Christoffer Dall 2017-10-19 7:46 ` Christoffer Dall 2017-09-23 0:41 ` [PATCH v3 12/20] genirq: Document vcpu_info usage for percpu_devid interrupts Christoffer Dall 2017-09-23 0:41 ` Christoffer Dall 2017-10-09 17:48 ` Marc Zyngier 2017-10-09 17:48 ` Marc Zyngier 2017-09-23 0:42 ` [PATCH v3 13/20] KVM: arm/arm64: Set VCPU affinity for virt timer irq Christoffer Dall 2017-09-23 0:42 ` Christoffer Dall 2017-10-09 17:52 ` Marc Zyngier 2017-10-09 17:52 ` Marc Zyngier 2017-09-23 0:42 ` [PATCH v3 14/20] KVM: arm/arm64: Avoid timer save/restore in vcpu entry/exit Christoffer Dall 2017-09-23 0:42 ` Christoffer Dall 2017-10-10 8:47 ` Marc Zyngier 2017-10-10 8:47 ` Marc Zyngier 2017-10-19 8:15 ` Christoffer Dall 2017-10-19 8:15 ` Christoffer Dall 2017-09-23 0:42 ` [PATCH v3 15/20] KVM: arm/arm64: Support EL1 phys timer register access in set/get reg Christoffer Dall 2017-09-23 0:42 ` Christoffer Dall 2017-10-10 9:10 ` Marc Zyngier 2017-10-10 9:10 ` Marc Zyngier 2017-10-19 8:32 ` Christoffer Dall 2017-10-19 8:32 ` Christoffer Dall 2017-09-23 0:42 ` [PATCH v3 16/20] KVM: arm/arm64: Use kvm_arm_timer_set/get_reg for guest register traps Christoffer Dall 2017-09-23 0:42 ` Christoffer Dall 2017-10-10 9:12 ` Marc Zyngier 2017-10-10 9:12 ` Marc Zyngier 2017-09-23 0:42 ` [PATCH v3 17/20] KVM: arm/arm64: Move phys_timer_emulate function Christoffer Dall 2017-09-23 0:42 ` Christoffer Dall 2017-10-10 9:21 ` Marc Zyngier 2017-10-10 9:21 ` Marc Zyngier 2017-09-23 0:42 ` [PATCH v3 18/20] KVM: arm/arm64: Avoid phys timer emulation in vcpu entry/exit Christoffer Dall 2017-09-23 0:42 ` Christoffer Dall 2017-10-10 9:45 ` Marc Zyngier 2017-10-10 9:45 ` Marc Zyngier 2017-10-19 8:44 ` Christoffer Dall 2017-10-19 8:44 ` Christoffer Dall 2017-09-23 0:42 ` [PATCH v3 19/20] KVM: arm/arm64: Get rid of kvm_timer_flush_hwstate Christoffer Dall 2017-09-23 0:42 ` Christoffer Dall 2017-10-10 9:46 ` Marc Zyngier 2017-10-10 9:46 ` Marc Zyngier 2017-09-23 0:42 ` [PATCH v3 20/20] KVM: arm/arm64: Rework kvm_timer_should_fire Christoffer Dall 2017-09-23 0:42 ` Christoffer Dall 2017-10-10 9:59 ` Marc Zyngier 2017-10-10 9:59 ` Marc Zyngier
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=878tg8jvwl.fsf@on-the-bus.cambridge.arm.com \ --to=marc.zyngier@arm.com \ --cc=catalin.marinas@arm.com \ --cc=cdall@linaro.org \ --cc=kvm@vger.kernel.org \ --cc=kvmarm@lists.cs.columbia.edu \ --cc=linux-arm-kernel@lists.infradead.org \ --cc=mark.rutland@arm.com \ --cc=will.deacon@arm.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.