diff mbox series

[v7.1,4/4] xen/x86: Allow stubdom access to irq created for msi.

Message ID c446916b9572a6425a994e9f8c6c995b8b38c518.1569470669.git-series.marmarek@invisiblethingslab.com (mailing list archive)
State New, archived
Headers show
Series None | expand

Commit Message

Marek Marczykowski-Górecki Sept. 26, 2019, 4:05 a.m. UTC 
Stubdomains need to be given sufficient privilege over the guest which it
provides emulation for in order for PCI passthrough to work correctly.
When a HVM domain try to enable MSI, QEMU in stubdomain calls
PHYSDEVOP_map_pirq, but later it needs to call XEN_DOMCTL_bind_pt_irq as
part of xc_domain_update_msi_irq. Give the stubdomain enough permissions
over the mapped interrupt in order to bind it successfully to it's
target domain.

This is not needed for PCI INTx, because IRQ in that case is known
beforehand and the stubdomain is given permissions over this IRQ by
libxl__device_pci_add (there's a do_pci_add against the stubdomain).

create_irq() already grant IRQ access to hardware_domain, with
assumption the device model lives there.
Modify create_irq() to take additional parameter, whether to grant
permissions to the domain creating the IRQ, which may be dom0 or a
stubdomain. Do this instead of granting access always to
hardware_domain. Save ID of the domain given permission, to revoke it in
destroy_irq() - easier and cleaner than replaying logic of create_irq()
parameter. Use domid instead of actual reference to the domain,
because it might get destroyed before destroying IRQ (stubdomain is
destroyed before its target domain). And it is not an issue,
because IRQ permissions live within domain structure, so destroying
a domain also implicitly revoke the permission.  Potential domid
reuse is detected by checking if that domain does have permission
over the IRQ being destroyed.

Then, adjust all callers to provide the parameter. In case of Xen
internal allocations, set it to false, but for domain accessible
interrupt set it to true.

Inspired by https://github.com/OpenXT/xenclient-oe/blob/5e0e7304a5a3c75ef01240a1e3673665b2aaf05e/recipes-extended/xen/files/stubdomain-msi-irq-access.patch by Eric Chanudet <chanudete@ainfosec.com>.

Signed-off-by: Simon Gaiser <simon@invisiblethingslab.com>
Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>
---
Changes in v3:
 - extend commit message
Changes in v4:
 - add missing destroy_irq on error path
Changes in v5:
 - move irq_{grant,revoke}_access() to {create,destroy}_irq(), which
   basically make it a different patch
 - add get_dm_domain() helper
 - do not give hardware_domain permission over IRQs used in Xen
   internally
 - rename create_irq argument to just 'd', to avoid confusion
   when it's called by hardware domain
 - verify that device is de-assigned before pci_remove_device call
 - save ID of domain given permission in create_irq(), to revoke it in
 destroy_irq()
 - drop domain parameter from destroy_irq() and msi_free_irq()
 - do not give hardware domain permission over IRQ created in
 iommu_set_interrupt()
Changes in v6:
 - do not give permission over hpet irq to hardware_domain
 - move creator_domid to arch_irq_desc
 - fix creator_domid initialization
 - always give current->domain permission instead of using
 get_dm_domain() helper. Analysis of all its use cases tells that it is
 the only value it returns.
 - drop unrelated change
Changes in v7:
 - Code style improvements (spaces, use %pd etc)
 - use bool parameter to create_irq, as it's only getting
 current->domain or NULL
 - remove redundant irq_access_permitted()
Changes in v7.1:
 - adjust comments, merge if
 - update commit message
---
 xen/arch/x86/hpet.c                      |  3 +-
 xen/arch/x86/irq.c                       | 42 +++++++++++++++++--------
 xen/drivers/char/ns16550.c               |  2 +-
 xen/drivers/passthrough/amd/iommu_init.c |  2 +-
 xen/drivers/passthrough/vtd/iommu.c      |  3 +-
 xen/include/asm-x86/irq.h                | 11 ++++++-
 6 files changed, 45 insertions(+), 18 deletions(-)

Comments

Jan Beulich Sept. 27, 2019, 12:27 p.m. UTC | #1 
On 26.09.2019 06:05, Marek Marczykowski-Górecki  wrote:
> Stubdomains need to be given sufficient privilege over the guest which it
> provides emulation for in order for PCI passthrough to work correctly.
> When a HVM domain try to enable MSI, QEMU in stubdomain calls
> PHYSDEVOP_map_pirq, but later it needs to call XEN_DOMCTL_bind_pt_irq as
> part of xc_domain_update_msi_irq. Give the stubdomain enough permissions
> over the mapped interrupt in order to bind it successfully to it's
> target domain.
> 
> This is not needed for PCI INTx, because IRQ in that case is known
> beforehand and the stubdomain is given permissions over this IRQ by
> libxl__device_pci_add (there's a do_pci_add against the stubdomain).
> 
> create_irq() already grant IRQ access to hardware_domain, with
> assumption the device model lives there.
> Modify create_irq() to take additional parameter, whether to grant
> permissions to the domain creating the IRQ, which may be dom0 or a
> stubdomain. Do this instead of granting access always to
> hardware_domain. Save ID of the domain given permission, to revoke it in
> destroy_irq() - easier and cleaner than replaying logic of create_irq()
> parameter. Use domid instead of actual reference to the domain,
> because it might get destroyed before destroying IRQ (stubdomain is
> destroyed before its target domain). And it is not an issue,
> because IRQ permissions live within domain structure, so destroying
> a domain also implicitly revoke the permission.  Potential domid
> reuse is detected by checking if that domain does have permission
> over the IRQ being destroyed.
> 
> Then, adjust all callers to provide the parameter. In case of Xen
> internal allocations, set it to false, but for domain accessible
> interrupt set it to true.
> 
> Inspired by https://github.com/OpenXT/xenclient-oe/blob/5e0e7304a5a3c75ef01240a1e3673665b2aaf05e/recipes-extended/xen/files/stubdomain-msi-irq-access.patch by Eric Chanudet <chanudete@ainfosec.com>.
> 
> Signed-off-by: Simon Gaiser <simon@invisiblethingslab.com>
> Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com>
> Reviewed-by: Roger Pau Monné <roger.pau@citrix.com>

Acked-by: Jan Beulich <jbeulich@suse.com>
with a couple of cosmetic things addressed, which I'll do while
committing.

Jan
diff mbox series

Patch

diff --git a/xen/arch/x86/hpet.c b/xen/arch/x86/hpet.c
index 4b08488..57f68fa 100644
--- a/xen/arch/x86/hpet.c
+++ b/xen/arch/x86/hpet.c
@@ -11,6 +11,7 @@ 
 #include <xen/softirq.h>
 #include <xen/irq.h>
 #include <xen/numa.h>
+#include <xen/sched.h>
 #include <asm/fixmap.h>
 #include <asm/div64.h>
 #include <asm/hpet.h>
@@ -368,7 +369,7 @@  static int __init hpet_assign_irq(struct hpet_event_channel *ch)
 {
     int irq;
 
-    if ( (irq = create_irq(NUMA_NO_NODE)) < 0 )
+    if ( (irq = create_irq(NUMA_NO_NODE, false)) < 0 )
         return irq;
 
     ch->msi.irq = irq;
diff --git a/xen/arch/x86/irq.c b/xen/arch/x86/irq.c
index 0ee3346..4304896 100644
--- a/xen/arch/x86/irq.c
+++ b/xen/arch/x86/irq.c
@@ -254,7 +254,8 @@  void __init clear_irq_vector(int irq)
 /*
  * Dynamic irq allocate and deallocation for MSI
  */
-int create_irq(nodeid_t node)
+
+int create_irq(nodeid_t node, bool grant_access)
 {
     int irq, ret;
     struct irq_desc *desc;
@@ -282,18 +283,23 @@  int create_irq(nodeid_t node)
         }
         ret = assign_irq_vector(irq, mask);
     }
+
+    ASSERT(desc->arch.creator_domid == DOMID_INVALID);
+
     if (ret < 0)
     {
         desc->arch.used = IRQ_UNUSED;
         irq = ret;
     }
-    else if ( hardware_domain )
+    else if ( grant_access )
     {
-        ret = irq_permit_access(hardware_domain, irq);
+        ret = irq_permit_access(current->domain, irq);
         if ( ret )
             printk(XENLOG_G_ERR
-                   "Could not grant Dom0 access to IRQ%d (error %d)\n",
-                   irq, ret);
+                   "Could not grant %pd access to IRQ%d (error %d)\n",
+                   current->domain, irq, ret);
+        else
+            desc->arch.creator_domid = current->domain->domain_id;
     }
 
     return irq;
@@ -307,14 +313,23 @@  void destroy_irq(unsigned int irq)
 
     BUG_ON(!MSI_IRQ(irq));
 
-    if ( hardware_domain )
+    if ( desc->arch.creator_domid != DOMID_INVALID )
     {
-        int err = irq_deny_access(hardware_domain, irq);
+        struct domain *d = get_domain_by_id(desc->arch.creator_domid);
 
-        if ( err )
-            printk(XENLOG_G_ERR
-                   "Could not revoke Dom0 access to IRQ%u (error %d)\n",
-                   irq, err);
+        if ( d )
+        {
+            int err = irq_deny_access(d, irq);
+            if ( err )
+                printk(XENLOG_G_ERR
+                       "Could not revoke %pd access to IRQ%u (error %d)\n",
+                       d, irq, err);
+
+            put_domain(d);
+
+        }
+
+        desc->arch.creator_domid = DOMID_INVALID;
     }
 
     spin_lock_irqsave(&desc->lock, flags);
@@ -381,6 +396,7 @@  int arch_init_one_irq_desc(struct irq_desc *desc)
 
     desc->arch.vector = IRQ_VECTOR_UNASSIGNED;
     desc->arch.old_vector = IRQ_VECTOR_UNASSIGNED;
+    desc->arch.creator_domid = DOMID_INVALID;
 
     return 0;
 }
@@ -2133,7 +2149,7 @@  int map_domain_pirq(
             spin_unlock_irqrestore(&desc->lock, flags);
 
             info = NULL;
-            irq = create_irq(NUMA_NO_NODE);
+            irq = create_irq(NUMA_NO_NODE, true);
             ret = irq >= 0 ? prepare_domain_irq_pirq(d, irq, pirq + nr, &info)
                            : irq;
             if ( ret < 0 )
@@ -2818,7 +2834,7 @@  int allocate_and_map_msi_pirq(struct domain *d, int index, int *pirq_p,
         if ( irq == -1 )
         {
     case MAP_PIRQ_TYPE_MULTI_MSI:
-            irq = create_irq(NUMA_NO_NODE);
+            irq = create_irq(NUMA_NO_NODE, true);
         }
 
         if ( irq < nr_irqs_gsi || irq >= nr_irqs )
diff --git a/xen/drivers/char/ns16550.c b/xen/drivers/char/ns16550.c
index 8667de6..fcd3979 100644
--- a/xen/drivers/char/ns16550.c
+++ b/xen/drivers/char/ns16550.c
@@ -722,7 +722,7 @@  static void __init ns16550_init_irq(struct serial_port *port)
     struct ns16550 *uart = port->uart;
 
     if ( uart->msi )
-        uart->irq = create_irq(0);
+        uart->irq = create_irq(0, false);
 #endif
 }
 
diff --git a/xen/drivers/passthrough/amd/iommu_init.c b/xen/drivers/passthrough/amd/iommu_init.c
index bb9f33e..233a8ae 100644
--- a/xen/drivers/passthrough/amd/iommu_init.c
+++ b/xen/drivers/passthrough/amd/iommu_init.c
@@ -765,7 +765,7 @@  static bool_t __init set_iommu_interrupt_handler(struct amd_iommu *iommu)
 {
     int irq, ret;
 
-    irq = create_irq(NUMA_NO_NODE);
+    irq = create_irq(NUMA_NO_NODE, false);
     if ( irq <= 0 )
     {
         dprintk(XENLOG_ERR, "IOMMU: no irqs\n");
diff --git a/xen/drivers/passthrough/vtd/iommu.c b/xen/drivers/passthrough/vtd/iommu.c
index 5d72270..24a1e92 100644
--- a/xen/drivers/passthrough/vtd/iommu.c
+++ b/xen/drivers/passthrough/vtd/iommu.c
@@ -1138,7 +1138,8 @@  static int __init iommu_set_interrupt(struct acpi_drhd_unit *drhd)
     struct irq_desc *desc;
 
     irq = create_irq(rhsa ? pxm_to_node(rhsa->proximity_domain)
-                          : NUMA_NO_NODE);
+                          : NUMA_NO_NODE,
+                     false);
     if ( irq <= 0 )
     {
         dprintk(XENLOG_ERR VTDPREFIX, "IOMMU: no irq available!\n");
diff --git a/xen/include/asm-x86/irq.h b/xen/include/asm-x86/irq.h
index bc0c0c1..a75c054 100644
--- a/xen/include/asm-x86/irq.h
+++ b/xen/include/asm-x86/irq.h
@@ -45,6 +45,11 @@  struct arch_irq_desc {
         unsigned move_cleanup_count;
         u8 move_in_progress : 1;
         s8 used;
+        /*
+         * Weak reference to domain having permission over this IRQ (which can
+         * be different from the domain actually having the IRQ assigned)
+         */
+        domid_t creator_domid;
 };
 
 /* For use with irq_desc.arch.used */
@@ -161,7 +166,11 @@  int  init_irq_data(void);
 void clear_irq_vector(int irq);
 
 int irq_to_vector(int irq);
-int create_irq(nodeid_t node);
+/*
+ * If grant_access is set the current domain is given permissions over
+ * the created IRQ.
+ */
+int create_irq(nodeid_t node, bool grant_access);
 void destroy_irq(unsigned int irq);
 int assign_irq_vector(int irq, const cpumask_t *);