From: Jan Beulich Subject: IOMMU: default to always quarantining PCI devices XSA-302 relies on the use of libxl's "assignable-add" feature to prepare devices to be assigned to untrusted guests. Unfortunately, this is not considered a strictly required step for device assignment. The PCI passthrough documentation on the wiki describes alternate ways of preparing devices for assignment, and libvirt uses its own ways as well. Hosts where these alternate methods are used will still leave the system in a vulnerable state after the device comes back from a guest. Default to always quarantining PCI devices, but provide a command line option to revert back to prior behavior (such that people who both sufficiently trust their guests and want to be able to use devices in Dom0 again after they had been in use by a guest wouldn't need to "manually" move such devices back from DomIO to Dom0). This is XSA-306. Reported-by: Marek Marczykowski-Górecki Signed-off-by: Jan Beulich Reviewed-by: Wei Liu --- a/docs/misc/xen-command-line.pandoc +++ b/docs/misc/xen-command-line.pandoc @@ -1171,7 +1171,7 @@ detection of systems known to misbehave > Default: `new` unless directed-EOI is supported ### iommu - = List of [ , verbose, debug, force, required, + = List of [ , verbose, debug, force, required, quarantine, sharept, intremap, intpost, crash-disable, snoop, qinval, igfx, amd-iommu-perdev-intremap, dom0-{passthrough,strict} ] @@ -1209,6 +1209,12 @@ boolean (e.g. `iommu=no`) can override t will prevent Xen from booting if IOMMUs aren't discovered and enabled successfully. +* The `quarantine` boolean can be used to control Xen's behavior when + de-assigning devices from guests. If enabled (the default), Xen always + quarantines such devices; they must be explicitly assigned back to Dom0 + before they can be used there again. If disabled, Xen will only + quarantine devices the toolstack hass arranged for getting quarantined. + * The `sharept` boolean controls whether the IOMMU pagetables are shared with the CPU-side HAP pagetables, or allocated separately. Sharing reduces the memory overhead, but doesn't work in combination with CPU-side --- a/xen/drivers/passthrough/iommu.c +++ b/xen/drivers/passthrough/iommu.c @@ -30,6 +30,7 @@ bool_t __initdata iommu_enable = 1; bool_t __read_mostly iommu_enabled; bool_t __read_mostly force_iommu; bool_t __read_mostly iommu_verbose; +bool __read_mostly iommu_quarantine = true; bool_t __read_mostly iommu_igfx = 1; bool_t __read_mostly iommu_snoop = 1; bool_t __read_mostly iommu_qinval = 1; @@ -74,6 +75,8 @@ static int __init parse_iommu_param(cons else if ( (val = parse_boolean("force", s, ss)) >= 0 || (val = parse_boolean("required", s, ss)) >= 0 ) force_iommu = val; + else if ( (val = parse_boolean("quarantine", s, ss)) >= 0 ) + iommu_quarantine = val; else if ( (val = parse_boolean("igfx", s, ss)) >= 0 ) iommu_igfx = val; else if ( (val = parse_boolean("verbose", s, ss)) >= 0 ) --- a/xen/drivers/passthrough/pci.c +++ b/xen/drivers/passthrough/pci.c @@ -1548,7 +1548,8 @@ int deassign_device(struct domain *d, u1 return -ENODEV; /* De-assignment from dom_io should de-quarantine the device */ - target = (pdev->quarantine && pdev->domain != dom_io) ? + target = ((pdev->quarantine || iommu_quarantine) && + pdev->domain != dom_io) ? dom_io : hardware_domain; while ( pdev->phantom_stride ) --- a/xen/include/xen/iommu.h +++ b/xen/include/xen/iommu.h @@ -53,7 +53,7 @@ static inline bool_t dfn_eq(dfn_t x, dfn } extern bool_t iommu_enable, iommu_enabled; -extern bool_t force_iommu, iommu_verbose, iommu_igfx; +extern bool force_iommu, iommu_quarantine, iommu_verbose, iommu_igfx; extern bool_t iommu_snoop, iommu_qinval, iommu_intremap, iommu_intpost; extern bool_t iommu_hap_pt_share; extern bool_t iommu_debug;