Re: [PATCH] irqchip/gic-v4: fix occasional VLPI drop

From: Marc Zyngier
Date: Thu Jan 24 2019 - 05:33:50 EST


On Wed, 23 Jan 2019 08:59:08 +0000,
Heyi Guo <guoheyi@xxxxxxxxxx> wrote:
>
> 1. In current implementation, every VLPI will temporarily be mapped to
> the first CPU in system (normally CPU0) and then moved to the real
> scheduled CPU later.
>
> 2. So there is a time window and a VLPI may be sent to CPU0 instead of
> the real scheduled vCPU, in a multi-CPU virtual machine.
>
> 3. However, CPU0 may have not been scheduled as a virtual CPU after
> system boots up, so the value of its GICR_VPROPBASER is unknown at
> that moment.
>
> 4. If the INTID of VLPI is larger than 2^(GICR_VPROPBASER.IDbits+1),
> while IDbits is also in unknown state, GIC will behave as if the VLPI
> is out of range and simply drop it, which results in interrupt missing
> in Guest.
>
> As no code will clear GICR_VPROPBASER at runtime, we can safely
> initialize the IDbits field at boot time for each CPU to get rid of
> this issue.
>
> We also clear Valid bit of GICR_VPENDBASER in case any ancient
> programming gets left in and causes memory corrupting.
>
> Signed-off-by: Heyi Guo <guoheyi@xxxxxxxxxx>
> Signed-off-by: Heyi Guo <heyi.guo@xxxxxxxxxx>
> ---
> drivers/irqchip/irq-gic-v3-its.c | 25 +++++++++++++++++++++++++
> 1 file changed, 25 insertions(+)
>
> diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
> index 3365d44..634c8a7 100644
> --- a/drivers/irqchip/irq-gic-v3-its.c
> +++ b/drivers/irqchip/irq-gic-v3-its.c
> @@ -2145,6 +2145,31 @@ static void its_cpu_init_lpis(void)
> val |= GICR_CTLR_ENABLE_LPIS;
> writel_relaxed(val, rbase + GICR_CTLR);
>
> + if (gic_rdists->has_vlpis) {
> + void __iomem *vlpi_base = gic_data_rdist_vlpi_base();
> +
> + /*
> + * It's possible for CPU to receive VLPIs before it is
> + * sheduled as a vPE, especially for the first CPU, and the
> + * VLPI with INTID larger than 2^(IDbits+1) will be considered
> + * as out of range and dropped by GIC.
> + * So we initialize IDbits to known value to avoid VLPI drop.
> + */
> + val = (LPI_NRBITS - 1) & GICR_VPROPBASER_IDBITS_MASK;
> + pr_debug("GICv4: CPU%d: Init IDbits to 0x%llx for GICR_VPROPBASER\n",
> + smp_processor_id(), val);
> + gits_write_vpropbaser(val, vlpi_base + GICR_VPROPBASER);
> +
> + /*
> + * Also clear Valid bit of GICR_VPENDBASER, in case some
> + * ancient programming gets left in and has possibility of
> + * corrupting memory.
> + */
> + val = gits_read_vpendbaser(vlpi_base + GICR_VPENDBASER);
> + val &= ~GICR_VPENDBASER_Valid;
> + gits_write_vpendbaser(val, vlpi_base + GICR_VPENDBASER);

I'm afraid it is not enough to just write back with the valid bit
clear. You need to make sure Dirty has been cleared before continuing,
which is why I was asking you to reuse some bits of its_vpe_deschedule.

Something like this:

diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c
index db20e992a40f..1808cb5f6a3e 100644
--- a/drivers/irqchip/irq-gic-v3-its.c
+++ b/drivers/irqchip/irq-gic-v3-its.c
@@ -2751,9 +2751,8 @@ static void its_vpe_schedule(struct its_vpe *vpe)
gits_write_vpendbaser(val, vlpi_base + GICR_VPENDBASER);
}

-static void its_vpe_deschedule(struct its_vpe *vpe)
+static u64 its_clear_vpend_valid(void __iomem *vlpi_base)
{
- void __iomem *vlpi_base = gic_data_rdist_vlpi_base();
u32 count = 1000000; /* 1s! */
bool clean;
u64 val;
@@ -2773,7 +2772,17 @@ static void its_vpe_deschedule(struct its_vpe *vpe)
}
} while (!clean && count);

- if (unlikely(!clean && !count)) {
+ return val;
+}
+
+static void its_vpe_deschedule(struct its_vpe *vpe)
+{
+ void __iomem *vlpi_base = gic_data_rdist_vlpi_base();
+ u64 val;
+
+ val = its_clear_vpend_valid(vlpi_base);
+
+ if (unlikely(val & GICR_VPENDBASER_Dirty)) {
pr_err_ratelimited("ITS virtual pending table not cleaning\n");
vpe->idai = false;
vpe->pending_last = true;

And you can the rewrite the above GICR_VPENDBASER_Valid clearing as:

val = its_clear_vpend_valid(vlpi_base);
WARN_ON(val & GICR_VPENDBASER_Dirty);

We'll at least get a warning if something goes wrong.

Thanks,

M.

--
Jazz is not dead, it just smell funny.