Dear all, I have read at multiple places that we should report any case where the kernel parameter intel_iommu=igfx_off changes anything compared to intel_iommu=on. In my first message from today, I have described how intel_iommu=on breaks the AHCI / SATA driver on one of our servers. I hereby would like to report that the SATA controller in question gets initialized *correctly* if I change the kernel parameter from intel_iommu=on to intel_iommu=igfx_off. In other words, the kernel cannot initialize that controller correctly and does not detect any HDDs connected to it when booted with intel_iommu=on, but does initialize that controller and the HDDs connected correctly when booted with intel_iommu=igfx_off. As with the problem reported in my first message from today, I have verified this behavior multiple times, testing cold boots as well as warm boots. Please refer to my first message from today if you would like to know the details of the hardware and software setup and the further circumstances. I will gladly assist in debugging this if anyone is interested. Thank you very much, Binarus
