[Qemu-devel] [PATCH 10/13] iommu: Introduce IOMMU emulation infrastructure

[David Gibson <david@gibson.dropbear.id.au>]

This patch adds the basic infrastructure necessary to emulate an IOMMU
visible to the guest.  The DMAContext structure is extended with
information and a callback describing the translation, and the various
DMA functions used by devices will now perform IOMMU translation using
this callback.

Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Richard Henderson <rth@twiddle.net>

Signed-off-by: Eduard - Gabriel Munteanu <eduard.munteanu@linux360.ro>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Conflicts:

	dma.h
---
 configure     |   12 ++++
 dma-helpers.c |  191 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 dma.h         |  125 +++++++++++++++++++++++++++++++-------
 hw/qdev-dma.h |    4 +-
 4 files changed, 307 insertions(+), 25 deletions(-)

diff --git a/configure b/configure
index 8b4e3c1..139419e 100755
--- a/configure
+++ b/configure
@@ -140,6 +140,7 @@ linux_aio=""
 cap_ng=""
 attr=""
 libattr=""
+iommu="yes"
 xfs=""
 
 vhost_net="no"
@@ -795,6 +796,10 @@ for opt do
   ;;
   --enable-vhost-net) vhost_net="yes"
   ;;
+  --enable-iommu) iommu="yes"
+  ;;
+  --disable-iommu) iommu="no"
+  ;;
   --disable-opengl) opengl="no"
   ;;
   --enable-opengl) opengl="yes"
@@ -1096,6 +1101,8 @@ echo "  --enable-docs            enable documentation build"
 echo "  --disable-docs           disable documentation build"
 echo "  --disable-vhost-net      disable vhost-net acceleration support"
 echo "  --enable-vhost-net       enable vhost-net acceleration support"
+echo "  --disable-iommu          disable IOMMU emulation support"
+echo "  --enable-iommu           enable IOMMU emulation support"
 echo "  --enable-trace-backend=B Set trace backend"
 echo "                           Available backends:" $("$source_path"/scripts/tracetool --list-backends)
 echo "  --with-trace-file=NAME   Full PATH,NAME of file to store traces"
@@ -2943,6 +2950,7 @@ echo "posix_madvise     $posix_madvise"
 echo "uuid support      $uuid"
 echo "libcap-ng support $cap_ng"
 echo "vhost-net support $vhost_net"
+echo "IOMMU support     $iommu"
 echo "Trace backend     $trace_backend"
 echo "Trace output file $trace_file-<pid>"
 echo "spice support     $spice"
@@ -3828,6 +3836,10 @@ if test "$target_softmmu" = "yes" -a \( \
   echo "CONFIG_NEED_MMU=y" >> $config_target_mak
 fi
 
+if test "$iommu" = "yes" ; then
+  echo "CONFIG_IOMMU=y" >> $config_host_mak
+fi
+
 if test "$gprof" = "yes" ; then
   echo "TARGET_GPROF=yes" >> $config_target_mak
   if test "$target_linux_user" = "yes" ; then
diff --git a/dma-helpers.c b/dma-helpers.c
index 85a39b5..1d5165c 100644
--- a/dma-helpers.c
+++ b/dma-helpers.c
@@ -10,6 +10,9 @@
 #include "dma.h"
 #include "block_int.h"
 #include "trace.h"
+#include "range.h"
+
+/* #define DEBUG_IOMMU */
 
 void qemu_sglist_init(QEMUSGList *qsg, int alloc_hint, DMAContext *dma)
 {
@@ -251,3 +254,191 @@ void dma_acct_start(BlockDriverState *bs, BlockAcctCookie *cookie,
 {
     bdrv_acct_start(bs, cookie, sg->size, type);
 }
+
+#ifdef CONFIG_IOMMU
+bool iommu_dma_memory_valid(DMAContext *dma, dma_addr_t addr, dma_addr_t len,
+                            DMADirection dir)
+{
+    target_phys_addr_t paddr, plen;
+
+#ifdef DEBUG_IOMMU
+    fprintf(stderr, "dma_memory_check context=%p addr=0x" DMA_ADDR_FMT
+            " len=0x" DMA_ADDR_FMT " dir=%d\n", dma, addr, len, dir);
+#endif
+
+    while (len) {
+        if (dma->translate(dma, addr, &paddr, &plen, dir) != 0) {
+            return false;
+        }
+
+        /* The translation might be valid for larger regions. */
+        if (plen > len) {
+            plen = len;
+        }
+
+        len -= plen;
+        addr += plen;
+    }
+
+    return true;
+}
+
+int iommu_dma_memory_rw(DMAContext *dma, dma_addr_t addr,
+                        void *buf, dma_addr_t len, DMADirection dir)
+{
+    target_phys_addr_t paddr, plen;
+    int err;
+
+#ifdef DEBUG_IOMMU
+    fprintf(stderr, "dma_memory_rw context=%p addr=0x" DMA_ADDR_FMT " len=0x"
+            DMA_ADDR_FMT " dir=%d\n", dma, addr, len, dir);
+#endif
+
+    while (len) {
+        err = dma->translate(dma, addr, &paddr, &plen, dir);
+        if (err) {
+            return -1;
+        }
+
+        /* The translation might be valid for larger regions. */
+        if (plen > len) {
+            plen = len;
+        }
+
+        cpu_physical_memory_rw(paddr, buf, plen,
+                               dir == DMA_DIRECTION_FROM_DEVICE);
+
+        len -= plen;
+        addr += plen;
+        buf += plen;
+    }
+
+    return 0;
+}
+
+int iommu_dma_memory_zero(DMAContext *dma, dma_addr_t addr, dma_addr_t len)
+{
+    target_phys_addr_t paddr, plen;
+    int err;
+
+#ifdef DEBUG_IOMMU
+    fprintf(stderr, "dma_memory_zero context=%p addr=0x" DMA_ADDR_FMT
+            " len=0x" DMA_ADDR_FMT "\n", dma, addr, len);
+#endif
+
+    while (len) {
+        err = dma->translate(dma, addr, &paddr, &plen,
+                             DMA_DIRECTION_FROM_DEVICE);
+        if (err) {
+            return err;
+        }
+
+        /* The translation might be valid for larger regions. */
+        if (plen > len) {
+            plen = len;
+        }
+
+        cpu_physical_memory_zero(paddr, plen);
+
+        len -= plen;
+        addr += plen;
+    }
+
+    return 0;
+}
+
+typedef struct DMAMemoryMap DMAMemoryMap;
+struct DMAMemoryMap {
+    dma_addr_t              addr;
+    size_t                  len;
+    void                    *buf;
+    DMAInvalidateMapFunc    *invalidate;
+    void                    *invalidate_opaque;
+
+    QLIST_ENTRY(DMAMemoryMap) list;
+};
+
+void dma_context_init(DMAContext *dma, DMATranslateFunc fn)
+{
+#ifdef DEBUG_IOMMU
+    fprintf(stderr, "dma_context_init(%p, %p)\n", dma, fn);
+#endif
+    dma->translate = fn;
+    QLIST_INIT(&dma->memory_maps);
+}
+
+void dma_invalidate_memory_range(DMAContext *dma,
+                                 dma_addr_t addr, dma_addr_t len)
+{
+    DMAMemoryMap *map;
+
+    QLIST_FOREACH(map, &dma->memory_maps, list) {
+        if (ranges_overlap(addr, len, map->addr, map->len)) {
+            map->invalidate(map->invalidate_opaque);
+            QLIST_REMOVE(map, list);
+            free(map);
+        }
+    }
+}
+
+void *iommu_dma_memory_map(DMAContext *dma, DMAInvalidateMapFunc *cb,
+                           void *cb_opaque, dma_addr_t addr, dma_addr_t *len,
+                           DMADirection dir)
+{
+    int err;
+    target_phys_addr_t paddr, plen;
+    void *buf;
+
+    plen = *len;
+    err = dma->translate(dma, addr, &paddr, &plen, dir);
+    if (err) {
+        return NULL;
+    }
+
+    /*
+     * If this is true, the virtual region is contiguous,
+     * but the translated physical region isn't. We just
+     * clamp *len, much like cpu_physical_memory_map() does.
+     */
+    if (plen < *len) {
+        *len = plen;
+    }
+
+    buf = cpu_physical_memory_map(paddr, &plen,
+                                  dir == DMA_DIRECTION_FROM_DEVICE);
+    *len = plen;
+
+    /* We treat maps as remote TLBs to cope with stuff like AIO. */
+    if (cb) {
+        DMAMemoryMap *map;
+
+        map = g_malloc(sizeof(DMAMemoryMap));
+        map->addr = addr;
+        map->len = *len;
+        map->buf = buf;
+        map->invalidate = cb;
+        map->invalidate_opaque = cb_opaque;
+
+        QLIST_INSERT_HEAD(&dma->memory_maps, map, list);
+    }
+
+    return buf;
+}
+
+void iommu_dma_memory_unmap(DMAContext *dma, void *buffer, dma_addr_t len,
+                            DMADirection dir, dma_addr_t access_len)
+{
+    DMAMemoryMap *map;
+
+    cpu_physical_memory_unmap(buffer, len,
+                              dir == DMA_DIRECTION_FROM_DEVICE,
+                              access_len);
+
+    QLIST_FOREACH(map, &dma->memory_maps, list) {
+        if ((map->buf == buffer) && (map->len == len)) {
+            QLIST_REMOVE(map, list);
+            free(map);
+        }
+    }
+}
+#endif /* CONFIG_IOMMU */
diff --git a/dma.h b/dma.h
index a66e3d7..1163242 100644
--- a/dma.h
+++ b/dma.h
@@ -15,6 +15,7 @@
 #include "hw/hw.h"
 #include "block.h"
 
+typedef struct DMAContext DMAContext;
 typedef struct ScatterGatherEntry ScatterGatherEntry;
 
 typedef enum {
@@ -31,30 +32,82 @@ struct QEMUSGList {
 };
 
 #if defined(TARGET_PHYS_ADDR_BITS)
+
+#ifdef CONFIG_IOMMU
+/*
+ * When an IOMMU is present, bus addresses become distinct from
+ * CPU/memory physical addresses and may be a different size.  Because
+ * the IOVA size depends more on the bus than on the platform, we more
+ * or less have to treat these as 64-bit always to cover all (or at
+ * least most) cases.
+ */
+typedef uint64_t dma_addr_t;
+
+#define DMA_ADDR_BITS 64
+#define DMA_ADDR_FMT "%" PRIx64
+#else
 typedef target_phys_addr_t dma_addr_t;
 
 #define DMA_ADDR_BITS TARGET_PHYS_ADDR_BITS
 #define DMA_ADDR_FMT TARGET_FMT_plx
+#endif
+
+typedef int DMATranslateFunc(DMAContext *dma,
+                             dma_addr_t addr,
+                             target_phys_addr_t *paddr,
+                             target_phys_addr_t *len,
+                             DMADirection dir);
+
+typedef struct DMAContext {
+#ifdef CONFIG_IOMMU
+    DMATranslateFunc *translate;
+    QLIST_HEAD(memory_maps, DMAMemoryMap) memory_maps;
+#endif
+} DMAContext;
+
+#ifdef CONFIG_IOMMU
+static inline bool dma_has_iommu(DMAContext *dma)
+{
+    return !!dma;
+}
+#else
+static inline bool dma_has_iommu(DMAContext *dma)
+{
+    return false;
+}
+#endif
 
 typedef void DMAInvalidateMapFunc(void *);
 
 /* Checks that the given range of addresses is valid for DMA.  This is
  * useful for certain cases, but usually you should just use
  * dma_memory_{read,write}() and check for errors */
-static inline bool dma_memory_valid(DMAContext *dma, dma_addr_t addr,
-                                    dma_addr_t len, DMADirection dir)
+bool iommu_dma_memory_valid(DMAContext *dma, dma_addr_t addr, dma_addr_t len,
+                            DMADirection dir);
+static inline bool dma_memory_valid(DMAContext *dma,
+                                    dma_addr_t addr, dma_addr_t len,
+                                    DMADirection dir)
 {
-    /* Stub version, with no iommu we assume all bus addresses are valid */
-    return true;
+    if (!dma_has_iommu(dma)) {
+        return true;
+    } else {
+        return iommu_dma_memory_valid(dma, addr, len, dir);
+    }
 }
 
+int iommu_dma_memory_rw(DMAContext *dma, dma_addr_t addr,
+                        void *buf, dma_addr_t len, DMADirection dir);
 static inline int dma_memory_rw(DMAContext *dma, dma_addr_t addr,
                                 void *buf, dma_addr_t len, DMADirection dir)
 {
-    /* Stub version when we have no iommu support */
-    cpu_physical_memory_rw(addr, buf, (target_phys_addr_t)len,
-                           dir == DMA_DIRECTION_FROM_DEVICE);
-    return 0;
+    if (!dma_has_iommu(dma)) {
+        /* Fast-path for no IOMMU */
+        cpu_physical_memory_rw(addr, buf, len,
+                               dir == DMA_DIRECTION_FROM_DEVICE);
+        return 0;
+    } else {
+        return iommu_dma_memory_rw(dma, addr, buf, len, dir);
+    }
 }
 
 static inline int dma_memory_read(DMAContext *dma, dma_addr_t addr,
@@ -70,35 +123,55 @@ static inline int dma_memory_write(DMAContext *dma, dma_addr_t addr,
                          DMA_DIRECTION_FROM_DEVICE);
 }
 
+int iommu_dma_memory_zero(DMAContext *dma, dma_addr_t addr, dma_addr_t len);
 static inline int dma_memory_zero(DMAContext *dma, dma_addr_t addr,
                                   dma_addr_t len)
 {
-    /* Stub version when we have no iommu support */
-    cpu_physical_memory_zero(addr, len);
-    return 0;
+    if (!dma_has_iommu(dma)) {
+        /* Fast-path for no IOMMU */
+        cpu_physical_memory_zero(addr, len);
+        return 0;
+    } else {
+        return iommu_dma_memory_zero(dma, addr, len);
+    }
 }
 
+void *iommu_dma_memory_map(DMAContext *dma,
+                           DMAInvalidateMapFunc *cb, void *opaque,
+                           dma_addr_t addr, dma_addr_t *len,
+                           DMADirection dir);
 static inline void *dma_memory_map(DMAContext *dma,
-                                   DMAInvalidateMapFunc *cb, void *opaque,
+                                   DMAInvalidateMapFunc *cb, void *cb_opaque,
                                    dma_addr_t addr, dma_addr_t *len,
                                    DMADirection dir)
 {
-    target_phys_addr_t xlen = *len;
-    void *p;
-
-    p = cpu_physical_memory_map(addr, &xlen,
-                                dir == DMA_DIRECTION_FROM_DEVICE);
-    *len = xlen;
-    return p;
+    if (!dma_has_iommu(dma)) {
+        target_phys_addr_t xlen = *len;
+        void *p;
+
+        p = cpu_physical_memory_map(addr, &xlen,
+                                    dir == DMA_DIRECTION_FROM_DEVICE);
+        *len = xlen;
+        return p;
+    } else {
+        return iommu_dma_memory_map(dma, cb, cb_opaque, addr, len, dir);
+    }
 }
 
+void iommu_dma_memory_unmap(DMAContext *dma,
+                            void *buffer, dma_addr_t len,
+                            DMADirection dir, dma_addr_t access_len);
 static inline void dma_memory_unmap(DMAContext *dma,
                                     void *buffer, dma_addr_t len,
                                     DMADirection dir, dma_addr_t access_len)
 {
-    return cpu_physical_memory_unmap(buffer, (target_phys_addr_t)len,
-                                     dir == DMA_DIRECTION_FROM_DEVICE,
-                                     access_len);
+    if (!dma_has_iommu(dma)) {
+        return cpu_physical_memory_unmap(buffer, (target_phys_addr_t)len,
+                                         dir == DMA_DIRECTION_FROM_DEVICE,
+                                         access_len);
+    } else {
+        iommu_dma_memory_unmap(dma, buffer, len, dir, access_len);
+    }
 }
 
 #define DEFINE_LDST_DMA(_lname, _sname, _bits, _end) \
@@ -139,6 +212,14 @@ DEFINE_LDST_DMA(q, q, 64, be);
 
 #undef DEFINE_LDST_DMA
 
+#ifdef CONFIG_IOMMU
+
+void dma_context_init(DMAContext *dma, DMATranslateFunc fn);
+void dma_invalidate_memory_range(DMAContext *dma,
+                                 dma_addr_t addr, dma_addr_t len);
+
+#endif /* CONFIG_IOMMU */
+
 struct ScatterGatherEntry {
     dma_addr_t base;
     dma_addr_t len;
diff --git a/hw/qdev-dma.h b/hw/qdev-dma.h
index e407771..92a404d 100644
--- a/hw/qdev-dma.h
+++ b/hw/qdev-dma.h
@@ -1,4 +1,2 @@
-#include "qdev-addr.h"
-
 #define DEFINE_PROP_DMAADDR(_n, _s, _f, _d)                               \
-    DEFINE_PROP_TADDR(_n, _s, _f, _d)
+    DEFINE_PROP_HEX64(_n, _s, _f, _d)
-- 
1.7.9.1