Re: [Xen-devel] [PATCH 2/2 V2] iommu/amd: Workaround for erratum 787

[Tim Deegan <tim@xxxxxxx>]

At 00:05 -0500 on 10 Jun (1370822751), suravee.suthikulpanit@xxxxxxx wrote:
> From: Suravee Suthikulpanit <suravee.suthikulpanit@xxxxxxx>
> 
> The IOMMU interrupt handling in bottom half must clear the PPR log interrupt
> and event log interrupt bits to re-enable the interrupt. This is done by
> writing 1 to the memory mapped register to clear the bit. Due to hardware bug,
> if the driver tries to clear this bit while the IOMMU hardware also setting
> this bit, the conflict will result with the bit being set. If the interrupt
> handling code does not make sure to clear this bit, subsequent changes in the
> event/PPR logs will no longer generating interrupts, and would result if
> buffer overflow. After clearing the bits, the driver must read back
> the register to verify.

Is there a risk of livelock here?  That is, if some device is causing a
lot of IOMMU faults, a CPU could get stuck in this loop re-enabling
interrupts as fast as they are raised.

The solution suggested in the erratum seems better: if the bit is set
after clearing, process the interrupts again (i.e. run/schedule the
top-half handler).  That way the bottom-half handler will definitely
terminate and the system can make some progress.

Cheers,

Tim.

> Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@xxxxxxx>
> ---
> V2 changes:
>       - Coding style fixes
> 
>  xen/drivers/passthrough/amd/iommu_init.c |   36 
> ++++++++++++++++++++----------
>  1 file changed, 24 insertions(+), 12 deletions(-)
> 
> diff --git a/xen/drivers/passthrough/amd/iommu_init.c 
> b/xen/drivers/passthrough/amd/iommu_init.c
> index a85c63f..048a2e6 100644
> --- a/xen/drivers/passthrough/amd/iommu_init.c
> +++ b/xen/drivers/passthrough/amd/iommu_init.c
> @@ -616,15 +616,21 @@ static void iommu_check_event_log(struct amd_iommu 
> *iommu)
>  
>      spin_lock_irqsave(&iommu->lock, flags);
>      
> -    /*check event overflow */
>      entry = readl(iommu->mmio_base + IOMMU_STATUS_MMIO_OFFSET);
> +    while (entry & (1 << IOMMU_STATUS_EVENT_LOG_INT_SHIFT))
> +    {
> +        /* Check event overflow */
> +        if ( iommu_get_bit(entry, IOMMU_STATUS_EVENT_OVERFLOW_SHIFT) )
> +            iommu_reset_log(iommu, &iommu->event_log, 
> set_iommu_event_log_control);
>  
> -    if ( iommu_get_bit(entry, IOMMU_STATUS_EVENT_OVERFLOW_SHIFT) )
> -        iommu_reset_log(iommu, &iommu->event_log, 
> set_iommu_event_log_control);
> +        /* RW1C interrupt status bit */
> +        writel((1 << IOMMU_STATUS_EVENT_LOG_INT_SHIFT),
> +            iommu->mmio_base + IOMMU_STATUS_MMIO_OFFSET);
>  
> -    /* RW1C interrupt status bit */
> -    writel((1 << IOMMU_STATUS_EVENT_LOG_INT_SHIFT),
> -        iommu->mmio_base + IOMMU_STATUS_MMIO_OFFSET);
> +        /* Workaround for erratum787:
> +         * Re-check to make sure the bit has been cleared */
> +        entry = readl(iommu->mmio_base + IOMMU_STATUS_MMIO_OFFSET);
> +    }
>  
>      spin_unlock_irqrestore(&iommu->lock, flags);
>  }
> @@ -684,15 +690,21 @@ static void iommu_check_ppr_log(struct amd_iommu *iommu)
>      
>      spin_lock_irqsave(&iommu->lock, flags);
>  
> -    /*check event overflow */
>      entry = readl(iommu->mmio_base + IOMMU_STATUS_MMIO_OFFSET);
> +    while ( entry & (1 << IOMMU_STATUS_PPR_LOG_INT_SHIFT ) )
> +    {
> +        /* Check event overflow */
> +        if ( iommu_get_bit(entry, IOMMU_STATUS_PPR_LOG_OVERFLOW_SHIFT) )
> +            iommu_reset_log(iommu, &iommu->ppr_log, 
> set_iommu_ppr_log_control);
>  
> -    if ( iommu_get_bit(entry, IOMMU_STATUS_PPR_LOG_OVERFLOW_SHIFT) )
> -        iommu_reset_log(iommu, &iommu->ppr_log, set_iommu_ppr_log_control);
> +        /* RW1C interrupt status bit */
> +        writel((1 << IOMMU_STATUS_PPR_LOG_INT_SHIFT),
> +            iommu->mmio_base + IOMMU_STATUS_MMIO_OFFSET);
>  
> -    /* RW1C interrupt status bit */
> -    writel((1 << IOMMU_STATUS_PPR_LOG_INT_SHIFT),
> -        iommu->mmio_base + IOMMU_STATUS_MMIO_OFFSET);
> +        /* Workaround for erratum787:
> +         * Re-check to make sure the bit has been cleared */
> +        entry = readl(iommu->mmio_base + IOMMU_STATUS_MMIO_OFFSET);
> +    }
>  
>      spin_unlock_irqrestore(&iommu->lock, flags);
>  }
> -- 
> 1.7.10.4
> 
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxx
> http://lists.xen.org/xen-devel

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel