pseries: Convert sPAPR TCEs to use generic IOMMU infrastructure

The pseries platform already contains an IOMMU implementation, since it is
essential for the platform's paravirtualized VIO devices.  This IOMMU
support is currently built into the implementation of the VIO "bus" and
the various VIO devices.

This patch converts this code to make use of the new common IOMMU
infrastructure.

We don't yet handle synchronization of map/unmap callbacks vs. invalidations,
this will require some complex interaction with the kernel and is not a
major concern at this stage.

Cc: Alex Graf <agraf@suse.de>

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

1 files changed, 20 insertions, 261 deletions

diff --git a/hw/spapr_vio.c b/hw/spapr_vio.c
index c8271c626c..05b55032a9 100644
--- a/hw/spapr_vio.c
+++ b/hw/spapr_vio.c
@@ -39,7 +39,6 @@
 #endif /* CONFIG_FDT */
 
 /* #define DEBUG_SPAPR */
-/* #define DEBUG_TCE */
 
 #ifdef DEBUG_SPAPR
 #define dprintf(fmt, ...) \
@@ -143,26 +142,9 @@ static int vio_make_devnode(VIOsPAPRDevice *dev,
         }
     }
 
-    if (dev->rtce_window_size) {
-        uint32_t dma_prop[] = {cpu_to_be32(dev->reg),
-                               0, 0,
-                               0, cpu_to_be32(dev->rtce_window_size)};
-
-        ret = fdt_setprop_cell(fdt, node_off, "ibm,#dma-address-cells", 2);
-        if (ret < 0) {
-            return ret;
-        }
-
-        ret = fdt_setprop_cell(fdt, node_off, "ibm,#dma-size-cells", 2);
-        if (ret < 0) {
-            return ret;
-        }
-
-        ret = fdt_setprop(fdt, node_off, "ibm,my-dma-window", dma_prop,
-                          sizeof(dma_prop));
-        if (ret < 0) {
-            return ret;
-        }
+    ret = spapr_dma_dt(fdt, node_off, "ibm,my-dma-window", dev->dma);
+    if (ret < 0) {
+        return ret;
     }
 
     if (pc->devnode) {
@@ -177,232 +159,6 @@ static int vio_make_devnode(VIOsPAPRDevice *dev,
 #endif /* CONFIG_FDT */
 
 /*
- * RTCE handling
- */
-
-static void rtce_init(VIOsPAPRDevice *dev)
-{
-    size_t size = (dev->rtce_window_size >> SPAPR_VIO_TCE_PAGE_SHIFT)
-        * sizeof(VIOsPAPR_RTCE);
-
-    if (size) {
-        dev->rtce_table = kvmppc_create_spapr_tce(dev->reg,
-                                                  dev->rtce_window_size,
-                                                  &dev->kvmtce_fd);
-
-        if (!dev->rtce_table) {
-            dev->rtce_table = g_malloc0(size);
-        }
-    }
-}
-
-static target_ulong h_put_tce(CPUPPCState *env, sPAPREnvironment *spapr,
-                              target_ulong opcode, target_ulong *args)
-{
-    target_ulong liobn = args[0];
-    target_ulong ioba = args[1];
-    target_ulong tce = args[2];
-    VIOsPAPRDevice *dev = spapr_vio_find_by_reg(spapr->vio_bus, liobn);
-    VIOsPAPR_RTCE *rtce;
-
-    if (!dev) {
-        hcall_dprintf("LIOBN 0x" TARGET_FMT_lx " does not exist\n", liobn);
-        return H_PARAMETER;
-    }
-
-    ioba &= ~(SPAPR_VIO_TCE_PAGE_SIZE - 1);
-
-#ifdef DEBUG_TCE
-    fprintf(stderr, "spapr_vio_put_tce on %s  ioba 0x" TARGET_FMT_lx
-            "  TCE 0x" TARGET_FMT_lx "\n", dev->qdev.id, ioba, tce);
-#endif
-
-    if (ioba >= dev->rtce_window_size) {
-        hcall_dprintf("Out-of-bounds IOBA 0x" TARGET_FMT_lx "\n", ioba);
-        return H_PARAMETER;
-    }
-
-    rtce = dev->rtce_table + (ioba >> SPAPR_VIO_TCE_PAGE_SHIFT);
-    rtce->tce = tce;
-
-    return H_SUCCESS;
-}
-
-int spapr_vio_check_tces(VIOsPAPRDevice *dev, target_ulong ioba,
-                         target_ulong len, enum VIOsPAPR_TCEAccess access)
-{
-    int start, end, i;
-
-    start = ioba >> SPAPR_VIO_TCE_PAGE_SHIFT;
-    end = (ioba + len - 1) >> SPAPR_VIO_TCE_PAGE_SHIFT;
-
-    for (i = start; i <= end; i++) {
-        if ((dev->rtce_table[i].tce & access) != access) {
-#ifdef DEBUG_TCE
-            fprintf(stderr, "FAIL on %d\n", i);
-#endif
-            return -1;
-        }
-    }
-
-    return 0;
-}
-
-int spapr_tce_dma_write(VIOsPAPRDevice *dev, uint64_t taddr, const void *buf,
-                        uint32_t size)
-{
-#ifdef DEBUG_TCE
-    fprintf(stderr, "spapr_tce_dma_write taddr=0x%llx size=0x%x\n",
-            (unsigned long long)taddr, size);
-#endif
-
-    /* Check for bypass */
-    if (dev->flags & VIO_PAPR_FLAG_DMA_BYPASS) {
-        cpu_physical_memory_write(taddr, buf, size);
-        return 0;
-    }
-
-    while (size) {
-        uint64_t tce;
-        uint32_t lsize;
-        uint64_t txaddr;
-
-        /* Check if we are in bound */
-        if (taddr >= dev->rtce_window_size) {
-#ifdef DEBUG_TCE
-            fprintf(stderr, "spapr_tce_dma_write out of bounds\n");
-#endif
-            return H_DEST_PARM;
-        }
-        tce = dev->rtce_table[taddr >> SPAPR_VIO_TCE_PAGE_SHIFT].tce;
-
-        /* How much til end of page ? */
-        lsize = MIN(size, ((~taddr) & SPAPR_VIO_TCE_PAGE_MASK) + 1);
-
-        /* Check TCE */
-        if (!(tce & 2)) {
-            return H_DEST_PARM;
-        }
-
-        /* Translate */
-        txaddr = (tce & ~SPAPR_VIO_TCE_PAGE_MASK) |
-            (taddr & SPAPR_VIO_TCE_PAGE_MASK);
-
-#ifdef DEBUG_TCE
-        fprintf(stderr, " -> write to txaddr=0x%llx, size=0x%x\n",
-                (unsigned long long)txaddr, lsize);
-#endif
-
-        /* Do it */
-        cpu_physical_memory_write(txaddr, buf, lsize);
-        buf += lsize;
-        taddr += lsize;
-        size -= lsize;
-    }
-    return 0;
-}
-
-int spapr_tce_dma_zero(VIOsPAPRDevice *dev, uint64_t taddr, uint32_t size)
-{
-    /* FIXME: allocating a temp buffer is nasty, but just stepping
-     * through writing zeroes is awkward.  This will do for now. */
-    uint8_t zeroes[size];
-
-#ifdef DEBUG_TCE
-    fprintf(stderr, "spapr_tce_dma_zero taddr=0x%llx size=0x%x\n",
-            (unsigned long long)taddr, size);
-#endif
-
-    memset(zeroes, 0, size);
-    return spapr_tce_dma_write(dev, taddr, zeroes, size);
-}
-
-void stb_tce(VIOsPAPRDevice *dev, uint64_t taddr, uint8_t val)
-{
-    spapr_tce_dma_write(dev, taddr, &val, sizeof(val));
-}
-
-void sth_tce(VIOsPAPRDevice *dev, uint64_t taddr, uint16_t val)
-{
-    val = tswap16(val);
-    spapr_tce_dma_write(dev, taddr, &val, sizeof(val));
-}
-
-
-void stw_tce(VIOsPAPRDevice *dev, uint64_t taddr, uint32_t val)
-{
-    val = tswap32(val);
-    spapr_tce_dma_write(dev, taddr, &val, sizeof(val));
-}
-
-void stq_tce(VIOsPAPRDevice *dev, uint64_t taddr, uint64_t val)
-{
-    val = tswap64(val);
-    spapr_tce_dma_write(dev, taddr, &val, sizeof(val));
-}
-
-int spapr_tce_dma_read(VIOsPAPRDevice *dev, uint64_t taddr, void *buf,
-                       uint32_t size)
-{
-#ifdef DEBUG_TCE
-    fprintf(stderr, "spapr_tce_dma_write taddr=0x%llx size=0x%x\n",
-            (unsigned long long)taddr, size);
-#endif
-
-    /* Check for bypass */
-    if (dev->flags & VIO_PAPR_FLAG_DMA_BYPASS) {
-        cpu_physical_memory_read(taddr, buf, size);
-        return 0;
-    }
-
-    while (size) {
-        uint64_t tce;
-        uint32_t lsize;
-        uint64_t txaddr;
-
-        /* Check if we are in bound */
-        if (taddr >= dev->rtce_window_size) {
-#ifdef DEBUG_TCE
-            fprintf(stderr, "spapr_tce_dma_read out of bounds\n");
-#endif
-            return H_DEST_PARM;
-        }
-        tce = dev->rtce_table[taddr >> SPAPR_VIO_TCE_PAGE_SHIFT].tce;
-
-        /* How much til end of page ? */
-        lsize = MIN(size, ((~taddr) & SPAPR_VIO_TCE_PAGE_MASK) + 1);
-
-        /* Check TCE */
-        if (!(tce & 1)) {
-            return H_DEST_PARM;
-        }
-
-        /* Translate */
-        txaddr = (tce & ~SPAPR_VIO_TCE_PAGE_MASK) |
-            (taddr & SPAPR_VIO_TCE_PAGE_MASK);
-
-#ifdef DEBUG_TCE
-        fprintf(stderr, " -> write to txaddr=0x%llx, size=0x%x\n",
-                (unsigned long long)txaddr, lsize);
-#endif
-        /* Do it */
-        cpu_physical_memory_read(txaddr, buf, lsize);
-        buf += lsize;
-        taddr += lsize;
-        size -= lsize;
-    }
-    return H_SUCCESS;
-}
-
-uint64_t ldq_tce(VIOsPAPRDevice *dev, uint64_t taddr)
-{
-    uint64_t val;
-
-    spapr_tce_dma_read(dev, taddr, &val, sizeof(val));
-    return tswap64(val);
-}
-
-/*
  * CRQ handling
  */
 static target_ulong h_reg_crq(CPUPPCState *env, sPAPREnvironment *spapr,
@@ -526,7 +282,7 @@ int spapr_vio_send_crq(VIOsPAPRDevice *dev, uint8_t *crq)
     }
 
     /* Maybe do a fast path for KVM just writing to the pages */
-    rc = spapr_tce_dma_read(dev, dev->crq.qladdr + dev->crq.qnext, &byte, 1);
+    rc = spapr_vio_dma_read(dev, dev->crq.qladdr + dev->crq.qnext, &byte, 1);
     if (rc) {
         return rc;
     }
@@ -534,7 +290,7 @@ int spapr_vio_send_crq(VIOsPAPRDevice *dev, uint8_t *crq)
         return 1;
     }
 
-    rc = spapr_tce_dma_write(dev, dev->crq.qladdr + dev->crq.qnext + 8,
+    rc = spapr_vio_dma_write(dev, dev->crq.qladdr + dev->crq.qnext + 8,
                              &crq[8], 8);
     if (rc) {
         return rc;
@@ -542,7 +298,7 @@ int spapr_vio_send_crq(VIOsPAPRDevice *dev, uint8_t *crq)
 
     kvmppc_eieio();
 
-    rc = spapr_tce_dma_write(dev, dev->crq.qladdr + dev->crq.qnext, crq, 8);
+    rc = spapr_vio_dma_write(dev, dev->crq.qladdr + dev->crq.qnext, crq, 8);
     if (rc) {
         return rc;
     }
@@ -560,13 +316,13 @@ int spapr_vio_send_crq(VIOsPAPRDevice *dev, uint8_t *crq)
 
 static void spapr_vio_quiesce_one(VIOsPAPRDevice *dev)
 {
-    dev->flags &= ~VIO_PAPR_FLAG_DMA_BYPASS;
+    VIOsPAPRDeviceClass *pc = VIO_SPAPR_DEVICE_GET_CLASS(dev);
+    uint32_t liobn = SPAPR_VIO_BASE_LIOBN | dev->reg;
 
-    if (dev->rtce_table) {
-        size_t size = (dev->rtce_window_size >> SPAPR_VIO_TCE_PAGE_SHIFT)
-            * sizeof(VIOsPAPR_RTCE);
-        memset(dev->rtce_table, 0, size);
+    if (dev->dma) {
+        spapr_tce_free(dev->dma);
     }
+    dev->dma = spapr_tce_new_dma_context(liobn, pc->rtce_window_size);
 
     dev->crq.qladdr = 0;
     dev->crq.qsize = 0;
@@ -593,9 +349,13 @@ static void rtas_set_tce_bypass(sPAPREnvironment *spapr, uint32_t token,
         return;
     }
     if (enable) {
-        dev->flags |= VIO_PAPR_FLAG_DMA_BYPASS;
+        spapr_tce_free(dev->dma);
+        dev->dma = NULL;
     } else {
-        dev->flags &= ~VIO_PAPR_FLAG_DMA_BYPASS;
+        VIOsPAPRDeviceClass *pc = VIO_SPAPR_DEVICE_GET_CLASS(dev);
+        uint32_t liobn = SPAPR_VIO_BASE_LIOBN | dev->reg;
+
+        dev->dma = spapr_tce_new_dma_context(liobn, pc->rtce_window_size);
     }
 
     rtas_st(rets, 0, 0);
@@ -662,6 +422,7 @@ static int spapr_vio_busdev_init(DeviceState *qdev)
 {
     VIOsPAPRDevice *dev = (VIOsPAPRDevice *)qdev;
     VIOsPAPRDeviceClass *pc = VIO_SPAPR_DEVICE_GET_CLASS(dev);
+    uint32_t liobn;
     char *id;
 
     if (dev->reg != -1) {
@@ -703,7 +464,8 @@ static int spapr_vio_busdev_init(DeviceState *qdev)
         return -1;
     }
 
-    rtce_init(dev);
+    liobn = SPAPR_VIO_BASE_LIOBN | dev->reg;
+    dev->dma = spapr_tce_new_dma_context(liobn, pc->rtce_window_size);
 
     return pc->init(dev);
 }
@@ -751,9 +513,6 @@ VIOsPAPRBus *spapr_vio_bus_init(void)
     /* hcall-vio */
     spapr_register_hypercall(H_VIO_SIGNAL, h_vio_signal);
 
-    /* hcall-tce */
-    spapr_register_hypercall(H_PUT_TCE, h_put_tce);
-
     /* hcall-crq */
     spapr_register_hypercall(H_REG_CRQ, h_reg_crq);
     spapr_register_hypercall(H_FREE_CRQ, h_free_crq);