Re: [PATCH] amd/iommu: fix logic bug in amd_iommu_report_page_fault()

[Joerg Roedel]

On Mon, Aug 02, 2021 at 04:30:50PM +0200, Luigi Rizzo wrote:
> Ah didn't realize that. Thank you!
> 
> Two questions on the topic:
> 1. how comes only the AMD driver is so verbose in reporting io page faults?
>    Neither intel nor other iommu drivers seem to log anything

What do you mean by 'verbose'? It is only a line per fault, and at least
the Intel driver also logs DMAR faults with one line per fault :)

> 2. Would it make sense to have a control to disable such logging,
>    either per-device or globally? Eg something like this (negative
>    logic so it must be set explicitly to disable logging).

Yes, we can talk about that. But only after the trace-event for it
landed in the code. There must be some way to report the faults and if
userspace prefers to catch them via trace-events than we can disable
printing them to the kernel log.

Regards,

	Joerg