Revert 23733:fbf3768e5934 "AMD IOMMU: remove global ..."

23733:fbf3768e5934 causes xen-unstable not to boot on several of the
xen.org AMD test systems.  We get an endless series of these:

  (XEN) AMD-Vi: IO_PAGE_FAULT: domain = 0, device id = 0x00a0, fault
  address = 0xfdf8f10144

I have constructed the attached patch which reverts c/s 23733
(adjusted for conflicts due to subsequent patches).  With this
reversion Xen once more boots on these machines.

23733 has been in the tree for some time now, causing this breakage,
and has already been fingered by the automatic bisector and discussed
on xen-devel as the cause of boot failures.  I think it is now time to
revert it pending a correct fix to the original problem.
Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com>

xen/drivers/passthrough/amd/iommu_acpi.c

@@ -66,8 +66,15 @@ static void __init add_ivrs_mapping_entry(
     if (ivrs_mappings[alias_id].intremap_table == NULL )
     {
          /* allocate per-device interrupt remapping table */
-         ivrs_mappings[alias_id].intremap_table =
+         if ( amd_iommu_perdev_intremap )
+             ivrs_mappings[alias_id].intremap_table =
                 amd_iommu_alloc_intremap_table();
+         else
+         {
+             if ( shared_intremap_table == NULL  )
+                 shared_intremap_table = amd_iommu_alloc_intremap_table();
+             ivrs_mappings[alias_id].intremap_table = shared_intremap_table;
+         }
     }
     /* assgin iommu hardware */
     ivrs_mappings[bdf].iommu = iommu;

xen/drivers/passthrough/amd/iommu_init.c

@@ -500,7 +500,7 @@ static void parse_event_log_entry(u32 entry[])
         /* Tell the device to stop DMAing; we can't rely on the guest to
          * control it for us. */
         for ( bdf = 0; bdf < ivrs_bdf_entries; bdf++ )
-            if ( get_requestor_id(bdf) == device_id ) 
+            if ( get_dma_requestor_id(bdf) == device_id ) 
             {
                 cword = pci_conf_read16(PCI_BUS(bdf), PCI_SLOT(bdf), 
                                 PCI_FUNC(bdf), PCI_COMMAND);
@@ -772,7 +772,8 @@ static int __init init_ivrs_mapping(void)
         ivrs_mappings[bdf].dte_ext_int_pass = IOMMU_CONTROL_DISABLED;
         ivrs_mappings[bdf].dte_init_pass = IOMMU_CONTROL_DISABLED;
 
-        spin_lock_init(&ivrs_mappings[bdf].intremap_lock);
+        if ( amd_iommu_perdev_intremap )
+            spin_lock_init(&ivrs_mappings[bdf].intremap_lock);
     }
     return 0;
 }

xen/drivers/passthrough/amd/iommu_intr.c

@@ -28,10 +28,20 @@
 #define INTREMAP_ENTRIES (1 << INTREMAP_LENGTH)
 
 int ioapic_bdf[MAX_IO_APICS];
+void *shared_intremap_table;
+static DEFINE_SPINLOCK(shared_intremap_lock);
 
 static spinlock_t* get_intremap_lock(int req_id)
 {
-    return &ivrs_mappings[req_id].intremap_lock;
+    return (amd_iommu_perdev_intremap ?
+           &ivrs_mappings[req_id].intremap_lock:
+           &shared_intremap_lock);
+}
+
+static int get_intremap_requestor_id(int bdf)
+{
+    ASSERT( bdf < ivrs_bdf_entries );
+    return ivrs_mappings[bdf].dte_requestor_id;
 }
 
 static int get_intremap_offset(u8 vector, u8 dm)
@@ -115,7 +125,7 @@ static void update_intremap_entry_from_ioapic(
     spinlock_t *lock;
     int offset;
 
-    req_id = get_requestor_id(bdf);
+    req_id = get_intremap_requestor_id(bdf);
     lock = get_intremap_lock(req_id);
 
     delivery_mode = rte->delivery_mode;
@@ -173,7 +183,7 @@ int __init amd_iommu_setup_ioapic_remapping(void)
                 continue;
             }
 
-            req_id = get_requestor_id(bdf);
+            req_id = get_intremap_requestor_id(bdf);
             lock = get_intremap_lock(req_id);
 
             delivery_mode = rte.delivery_mode;
@@ -273,13 +283,14 @@ static void update_intremap_entry_from_msi_msg(
 {
     unsigned long flags;
     u32* entry;
-    u16 bdf, req_id;
+    u16 bdf, req_id, alias_id;
     u8 delivery_mode, dest, vector, dest_mode;
     spinlock_t *lock;
     int offset;
 
     bdf = (pdev->bus << 8) | pdev->devfn;
-    req_id = get_requestor_id(bdf);
+    req_id = get_dma_requestor_id(bdf);
+    alias_id = get_intremap_requestor_id(bdf);
 
     if ( msg == NULL )
     {
@@ -288,6 +299,14 @@ static void update_intremap_entry_from_msi_msg(
         free_intremap_entry(req_id, msi_desc->remap_index);
         spin_unlock_irqrestore(lock, flags);
 
+        if ( ( req_id != alias_id ) &&
+            ivrs_mappings[alias_id].intremap_table != NULL )
+        {
+            lock = get_intremap_lock(alias_id);
+            spin_lock_irqsave(lock, flags);
+            free_intremap_entry(alias_id, msi_desc->remap_index);
+            spin_unlock_irqrestore(lock, flags);
+        }
         goto done;
     }
 
@@ -305,11 +324,30 @@ static void update_intremap_entry_from_msi_msg(
     update_intremap_entry(entry, vector, delivery_mode, dest_mode, dest);
     spin_unlock_irqrestore(lock, flags);
 
+    /*
+     * In some special cases, a pci-e device(e.g SATA controller in IDE mode)
+     * will use alias id to index interrupt remapping table.
+     * We have to setup a secondary interrupt remapping entry to satisfy those
+     * devices.
+     */
+
+    lock = get_intremap_lock(alias_id);
+    if ( ( req_id != alias_id ) &&
+        ivrs_mappings[alias_id].intremap_table != NULL )
+    {
+        spin_lock_irqsave(lock, flags);
+        entry = (u32*)get_intremap_entry(alias_id, offset);
+        update_intremap_entry(entry, vector, delivery_mode, dest_mode, dest);
+        spin_unlock_irqrestore(lock, flags);
+    }
+
 done:
     if ( iommu->enabled )
     {
         spin_lock_irqsave(&iommu->lock, flags);
         invalidate_interrupt_table(iommu, req_id);
+        if ( alias_id != req_id )
+            invalidate_interrupt_table(iommu, alias_id);
         flush_command_buffer(iommu);
         spin_unlock_irqrestore(&iommu->lock, flags);
     }

xen/drivers/passthrough/amd/iommu_map.c

@@ -719,7 +719,7 @@ static int update_paging_mode(struct domain *d, unsigned long gfn)
         for_each_pdev( d, pdev )
         {
             bdf = (pdev->bus << 8) | pdev->devfn;
-            req_id = get_requestor_id(bdf);
+            req_id = get_dma_requestor_id(bdf);
             iommu = find_iommu_for_device(bdf);
             if ( !iommu )
             {

xen/drivers/passthrough/amd/pci_amd_iommu.c

@@ -34,10 +34,25 @@ struct amd_iommu *find_iommu_for_device(int bdf)
     return ivrs_mappings[bdf].iommu;
 }
 
-int get_requestor_id(u16 bdf)
+/*
+ * Some devices will use alias id and original device id to index interrupt
+ * table and I/O page table respectively. Such devices will have
+ * both alias entry and select entry in IVRS structure.
+ *
+ * Return original device id, if device has valid interrupt remapping
+ * table setup for both select entry and alias entry.
+ */
+int get_dma_requestor_id(u16 bdf)
 {
+    int req_id;
+
     BUG_ON ( bdf >= ivrs_bdf_entries );
-    return ivrs_mappings[bdf].dte_requestor_id;
+    req_id = ivrs_mappings[bdf].dte_requestor_id;
+    if ( (ivrs_mappings[bdf].intremap_table != NULL) &&
+         (ivrs_mappings[req_id].intremap_table != NULL) )
+        req_id = bdf;
+
+    return req_id;
 }
 
 static int is_translation_valid(u32 *entry)
@@ -79,7 +94,7 @@ static void amd_iommu_setup_domain_device(
         valid = 0;
 
     /* get device-table entry */
-    req_id = get_requestor_id(bdf);
+    req_id = get_dma_requestor_id(bdf);
     dte = iommu->dev_table.buffer + (req_id * IOMMU_DEV_TABLE_ENTRY_SIZE);
 
     spin_lock_irqsave(&iommu->lock, flags);
@@ -254,7 +269,7 @@ static void amd_iommu_disable_domain_device(
     int req_id;
 
     BUG_ON ( iommu->dev_table.buffer == NULL );
-    req_id = get_requestor_id(bdf);
+    req_id = get_dma_requestor_id(bdf);
     dte = iommu->dev_table.buffer + (req_id * IOMMU_DEV_TABLE_ENTRY_SIZE);
 
     spin_lock_irqsave(&iommu->lock, flags);
@@ -316,7 +331,7 @@ static int reassign_device( struct domain *source, struct domain *target,
 static int amd_iommu_assign_device(struct domain *d, u8 bus, u8 devfn)
 {
     int bdf = (bus << 8) | devfn;
-    int req_id = get_requestor_id(bdf);
+    int req_id = get_dma_requestor_id(bdf);
 
     if ( ivrs_mappings[req_id].unity_map_enable )
     {
@@ -447,7 +462,7 @@ static int amd_iommu_group_id(u8 bus, u8 devfn)
     int rt;
     int bdf = (bus << 8) | devfn;
     rt = ( bdf < ivrs_bdf_entries ) ?
-        get_requestor_id(bdf) :
+        get_dma_requestor_id(bdf) :
         bdf;
     return rt;
 }

xen/drivers/passthrough/iommu.c

@@ -50,6 +50,7 @@ bool_t __read_mostly iommu_intremap = 1;
 bool_t __read_mostly iommu_hap_pt_share;
 bool_t __read_mostly iommu_debug;
 bool_t __read_mostly iommu_amd_perdev_vector_map = 1;
+bool_t __read_mostly amd_iommu_perdev_intremap;
 
 static void __init parse_iommu_param(char *s)
 {
@@ -76,6 +77,8 @@ static void __init parse_iommu_param(char *s)
             iommu_intremap = 0;
         else if ( !strcmp(s, "debug") )
             iommu_debug = 1;
+        else if ( !strcmp(s, "amd-iommu-perdev-intremap") )
+            amd_iommu_perdev_intremap = 1;
         else if ( !strcmp(s, "dom0-passthrough") )
             iommu_passthrough = 1;
         else if ( !strcmp(s, "dom0-strict") )

xen/include/asm-x86/hvm/svm/amd-iommu-proto.h

@@ -65,7 +65,7 @@ int amd_iommu_reserve_domain_unity_map(struct domain *domain,
 void amd_iommu_share_p2m(struct domain *d);
 
 /* device table functions */
-int get_requestor_id(u16 bdf);
+int get_dma_requestor_id(u16 bdf);
 void amd_iommu_add_dev_table_entry(
     u32 *dte, u8 sys_mgt, u8 dev_ex, u8 lint1_pass, u8 lint0_pass, 
     u8 nmi_pass, u8 ext_int_pass, u8 init_pass);
@@ -97,6 +97,7 @@ unsigned int amd_iommu_read_ioapic_from_ire(
     unsigned int apic, unsigned int reg);
 
 extern int ioapic_bdf[MAX_IO_APICS];
+extern void *shared_intremap_table;
 
 /* power management support */
 void amd_iommu_resume(void);

xen/include/xen/iommu.h

@@ -32,6 +32,7 @@ extern bool_t iommu_workaround_bios_bug, iommu_passthrough;
 extern bool_t iommu_snoop, iommu_qinval, iommu_intremap;
 extern bool_t iommu_hap_pt_share;
 extern bool_t iommu_debug;
+extern bool_t amd_iommu_perdev_intremap;
 
 extern struct rangeset *mmio_ro_ranges;