On Fri, Feb 23, 2018 at 11:27:30PM +0100, Sebastian Andrzej Siewior wrote:
> The function get_irq_table() reads/writes irq_lookup_table while holding
> the amd_iommu_devtable_lock. It also modifies
> amd_iommu_dev_table[].data[2].
> set_dte_entry() is using amd_iommu_dev_table[].data[0|1] (under the
> domain->lock) so it should be okay. The access to the iommu is
> serialized with its own (iommu's) lock.
>
> So split out get_irq_table() out of amd_iommu_devtable_lock's lock. The
> new lock is a raw_spin_lock because modify_irte_ga() is called while
> desc->lock is held (which is raw).
>
> Signed-off-by: Sebastian Andrzej Siewior <[email protected]>
This patch likely conflicts with changes already in the iommu-tree, as
it contains a patch splitting the get_irq_table() function. Can
you rebase your series against the x86/amd branch?
Thanks,
Joerg
