While no longer an immediate problem with flushes no longer timing out, errors (if any) get properly reported by iommu_flush_iotlb_{dsi,psi}(). Overwriting such an error with, perhaps, a success indicator received from another IOMMU will misguide callers. Record the first error, but don't bail from the loop (such that further necessary invalidation gets carried out on a best effort basis).
Signed-off-by: Jan Beulich <jbeul...@suse.com> --- a/xen/drivers/passthrough/vtd/iommu.c +++ b/xen/drivers/passthrough/vtd/iommu.c @@ -643,7 +643,7 @@ static int __must_check iommu_flush_iotl struct vtd_iommu *iommu; bool_t flush_dev_iotlb; int iommu_domid; - int rc = 0; + int ret = 0; /* * No need pcideves_lock here because we have flush @@ -651,6 +651,8 @@ static int __must_check iommu_flush_iotl */ for_each_drhd_unit ( drhd ) { + int rc; + iommu = drhd->iommu; if ( !test_bit(iommu->index, &hd->arch.vtd.iommu_bitmap) ) @@ -673,13 +675,12 @@ static int __must_check iommu_flush_iotl flush_dev_iotlb); if ( rc > 0 ) - { iommu_flush_write_buffer(iommu); - rc = 0; - } + else if ( !ret ) + ret = rc; } - return rc; + return ret; } static int __must_check iommu_flush_iotlb_pages(struct domain *d,