[Qemu-devel] [PATCH v3 8/9] spapr_iommu: Introduce page_shift in sPAPRTCETable

[Alexey Kardashevskiy]

At the moment only 4K pages are supported by sPAPRTCETable. Since sPAPR
spec allows other page sizes and we are going to implement them, we need
page size to be configrable.

This adds @page_shift into sPAPRTCETable and replaces SPAPR_TCE_PAGE_SHIFT
with it where it is possible.

Signed-off-by: Alexey Kardashevskiy <address@hidden>
---
Changes:
v3:
* s/1/1ULL/ in mask operations
* defined and used IOMMU_PAGE_SIZE/SHIFT
* do not remove SPAPR_TCE_PAGE_MASK, use it to verify that
tce_list is 4K aligned as SPAPR requires

v2:
* target_ulong for "mask" replaced with hwaddr
* added page_mask and page_size local variables where possible
---
 hw/ppc/spapr_iommu.c   | 71 ++++++++++++++++++++++++++++++++------------------
 hw/ppc/spapr_pci.c     |  1 +
 hw/ppc/spapr_vio.c     |  1 +
 include/hw/ppc/spapr.h |  2 ++
 4 files changed, 49 insertions(+), 26 deletions(-)

diff --git a/hw/ppc/spapr_iommu.c b/hw/ppc/spapr_iommu.c
index c6846f7..7c4c529 100644
--- a/hw/ppc/spapr_iommu.c
+++ b/hw/ppc/spapr_iommu.c
@@ -35,6 +35,9 @@ enum sPAPRTCEAccess {
     SPAPR_TCE_RW = 3,
 };
 
+#define IOMMU_PAGE_SIZE(shift)      (1ULL << (shift))
+#define IOMMU_PAGE_MASK(shift)      (~(IOMMU_PAGE_SIZE(shift) - 1))
+
 static QLIST_HEAD(spapr_tce_tables, sPAPRTCETable) spapr_tce_tables;
 
 static sPAPRTCETable *spapr_tce_find_by_liobn(uint32_t liobn)
@@ -70,12 +73,14 @@ static IOMMUTLBEntry spapr_tce_translate_iommu(MemoryRegion 
*iommu, hwaddr addr)
 
     if (tcet->bypass) {
         ret.perm = IOMMU_RW;
-    } else if ((addr >> SPAPR_TCE_PAGE_SHIFT) < tcet->nb_table) {
+    } else if ((addr >> tcet->page_shift) < tcet->nb_table) {
         /* Check if we are in bound */
-        tce = tcet->table[addr >> SPAPR_TCE_PAGE_SHIFT];
-        ret.iova = addr & ~SPAPR_TCE_PAGE_MASK;
-        ret.translated_addr = tce & ~SPAPR_TCE_PAGE_MASK;
-        ret.addr_mask = SPAPR_TCE_PAGE_MASK;
+        hwaddr page_mask = IOMMU_PAGE_MASK(tcet->page_shift);
+
+        tce = tcet->table[addr >> tcet->page_shift];
+        ret.iova = addr & page_mask;
+        ret.translated_addr = tce & page_mask;
+        ret.addr_mask = ~page_mask;
         ret.perm = tce;
     }
     trace_spapr_iommu_xlate(tcet->liobn, addr, ret.iova, ret.perm,
@@ -113,7 +118,7 @@ static int spapr_tce_table_realize(DeviceState *dev)
     if (kvm_enabled()) {
         tcet->table = kvmppc_create_spapr_tce(tcet->liobn,
                                               tcet->nb_table <<
-                                              SPAPR_TCE_PAGE_SHIFT,
+                                              tcet->page_shift,
                                               &tcet->fd);
     }
 
@@ -136,6 +141,7 @@ static int spapr_tce_table_realize(DeviceState *dev)
 }
 
 sPAPRTCETable *spapr_tce_new_table(DeviceState *owner, uint32_t liobn,
+                                   uint32_t page_shift,
                                    uint32_t nb_table)
 {
     sPAPRTCETable *tcet;
@@ -152,6 +158,7 @@ sPAPRTCETable *spapr_tce_new_table(DeviceState *owner, 
uint32_t liobn,
 
     tcet = SPAPR_TCE_TABLE(object_new(TYPE_SPAPR_TCE_TABLE));
     tcet->liobn = liobn;
+    tcet->page_shift = page_shift;
     tcet->nb_table = nb_table;
 
     object_property_add_child(OBJECT(owner), "tce-table", OBJECT(tcet), NULL);
@@ -197,19 +204,20 @@ static target_ulong put_tce_emu(sPAPRTCETable *tcet, 
target_ulong ioba,
                                 target_ulong tce)
 {
     IOMMUTLBEntry entry;
+    hwaddr page_mask = IOMMU_PAGE_MASK(tcet->page_shift);
 
-    if ((ioba >> SPAPR_TCE_PAGE_SHIFT) >= tcet->nb_table) {
+    if ((ioba >> tcet->page_shift) >= tcet->nb_table) {
         hcall_dprintf("spapr_vio_put_tce on out-of-bounds IOBA 0x"
                       TARGET_FMT_lx "\n", ioba);
         return H_PARAMETER;
     }
 
-    tcet->table[ioba >> SPAPR_TCE_PAGE_SHIFT] = tce;
+    tcet->table[ioba >> tcet->page_shift] = tce;
 
     entry.target_as = &address_space_memory,
-    entry.iova = ioba & ~SPAPR_TCE_PAGE_MASK;
-    entry.translated_addr = tce & ~SPAPR_TCE_PAGE_MASK;
-    entry.addr_mask = SPAPR_TCE_PAGE_MASK;
+    entry.iova = ioba & page_mask;
+    entry.translated_addr = tce & page_mask;
+    entry.addr_mask = ~page_mask;
     entry.perm = tce;
     memory_region_notify_iommu(&tcet->iommu, entry);
 
@@ -229,21 +237,25 @@ static target_ulong h_put_tce_indirect(PowerPCCPU *cpu,
     target_ulong ret = H_PARAMETER;
     sPAPRTCETable *tcet = spapr_tce_find_by_liobn(liobn);
     CPUState *cs = CPU(cpu);
+    hwaddr page_mask, page_size;
 
     if (!tcet) {
         return H_PARAMETER;
     }
 
-    if (npages > 512) {
+    if ((npages > 512) || (tce_list & SPAPR_TCE_PAGE_MASK)) {
         return H_PARAMETER;
     }
 
-    ioba &= ~SPAPR_TCE_PAGE_MASK;
-    tce_list &= ~SPAPR_TCE_PAGE_MASK;
+    page_mask = IOMMU_PAGE_MASK(tcet->page_shift);
+    page_size = IOMMU_PAGE_SIZE(tcet->page_shift);
+    ioba &= page_mask;
+
+    for (i = 0; i < npages; ++i, ioba += page_size) {
+        target_ulong off = (tce_list & ~SPAPR_TCE_RW) +
+                                i * sizeof(target_ulong);
+        target_ulong tce = ldq_phys(cs->as, off);
 
-    for (i = 0; i < npages; ++i, ioba += SPAPR_TCE_PAGE_SIZE) {
-        target_ulong tce = ldq_phys(cs->as, tce_list +
-                                    i * sizeof(target_ulong));
         ret = put_tce_emu(tcet, ioba, tce);
         if (ret) {
             break;
@@ -270,6 +282,7 @@ static target_ulong h_stuff_tce(PowerPCCPU *cpu, 
sPAPREnvironment *spapr,
     target_ulong npages = args[3];
     target_ulong ret = H_PARAMETER;
     sPAPRTCETable *tcet = spapr_tce_find_by_liobn(liobn);
+    hwaddr page_mask, page_size;
 
     if (!tcet) {
         return H_PARAMETER;
@@ -279,9 +292,11 @@ static target_ulong h_stuff_tce(PowerPCCPU *cpu, 
sPAPREnvironment *spapr,
         return H_PARAMETER;
     }
 
-    ioba &= ~SPAPR_TCE_PAGE_MASK;
+    page_mask = IOMMU_PAGE_MASK(tcet->page_shift);
+    page_size = IOMMU_PAGE_SIZE(tcet->page_shift);
+    ioba &= page_mask;
 
-    for (i = 0; i < npages; ++i, ioba += SPAPR_TCE_PAGE_SIZE) {
+    for (i = 0; i < npages; ++i, ioba += page_size) {
         ret = put_tce_emu(tcet, ioba, tce_value);
         if (ret) {
             break;
@@ -301,9 +316,11 @@ static target_ulong h_put_tce(PowerPCCPU *cpu, 
sPAPREnvironment *spapr,
     target_ulong ret = H_PARAMETER;
     sPAPRTCETable *tcet = spapr_tce_find_by_liobn(liobn);
 
-    ioba &= ~(SPAPR_TCE_PAGE_SIZE - 1);
-
     if (tcet) {
+        hwaddr page_mask = IOMMU_PAGE_MASK(tcet->page_shift);
+
+        ioba &= page_mask;
+
         ret = put_tce_emu(tcet, ioba, tce);
     }
     trace_spapr_iommu_put(liobn, ioba, tce, ret);
@@ -314,13 +331,13 @@ static target_ulong h_put_tce(PowerPCCPU *cpu, 
sPAPREnvironment *spapr,
 static target_ulong get_tce_emu(sPAPRTCETable *tcet, target_ulong ioba,
                                 target_ulong *tce)
 {
-    if ((ioba >> SPAPR_TCE_PAGE_SHIFT) >= tcet->nb_table) {
+    if ((ioba >> tcet->page_shift) >= tcet->nb_table) {
         hcall_dprintf("spapr_iommu_get_tce on out-of-bounds IOBA 0x"
                       TARGET_FMT_lx "\n", ioba);
         return H_PARAMETER;
     }
 
-    *tce = tcet->table[ioba >> SPAPR_TCE_PAGE_SHIFT];
+    *tce = tcet->table[ioba >> tcet->page_shift];
 
     return H_SUCCESS;
 }
@@ -334,9 +351,11 @@ static target_ulong h_get_tce(PowerPCCPU *cpu, 
sPAPREnvironment *spapr,
     target_ulong ret = H_PARAMETER;
     sPAPRTCETable *tcet = spapr_tce_find_by_liobn(liobn);
 
-    ioba &= ~(SPAPR_TCE_PAGE_SIZE - 1);
-
     if (tcet) {
+        hwaddr page_mask = IOMMU_PAGE_MASK(tcet->page_shift);
+
+        ioba &= page_mask;
+
         ret = get_tce_emu(tcet, ioba, &tce);
         if (!ret) {
             args[0] = tce;
@@ -385,7 +404,7 @@ int spapr_tcet_dma_dt(void *fdt, int node_off, const char 
*propname,
     }
 
     return spapr_dma_dt(fdt, node_off, propname,
-                        tcet->liobn, 0, tcet->nb_table << 
SPAPR_TCE_PAGE_SHIFT);
+                        tcet->liobn, 0, tcet->nb_table << tcet->page_shift);
 }
 
 static void spapr_tce_table_class_init(ObjectClass *klass, void *data)
diff --git a/hw/ppc/spapr_pci.c b/hw/ppc/spapr_pci.c
index fdd4c07..c9850d4 100644
--- a/hw/ppc/spapr_pci.c
+++ b/hw/ppc/spapr_pci.c
@@ -656,6 +656,7 @@ static void spapr_phb_finish_realize(sPAPRPHBState *sphb, 
Error **errp)
     sPAPRTCETable *tcet;
 
     tcet = spapr_tce_new_table(DEVICE(sphb), sphb->dma_liobn,
+                               SPAPR_TCE_PAGE_SHIFT,
                                0x40000000 >> SPAPR_TCE_PAGE_SHIFT);
     if (!tcet) {
         error_setg(errp, "Unable to create TCE table for %s",
diff --git a/hw/ppc/spapr_vio.c b/hw/ppc/spapr_vio.c
index b84e481..d7e9e6a 100644
--- a/hw/ppc/spapr_vio.c
+++ b/hw/ppc/spapr_vio.c
@@ -457,6 +457,7 @@ static int spapr_vio_busdev_init(DeviceState *qdev)
     if (pc->rtce_window_size) {
         uint32_t liobn = SPAPR_VIO_BASE_LIOBN | dev->reg;
         dev->tcet = spapr_tce_new_table(qdev, liobn,
+                                        SPAPR_TCE_PAGE_SHIFT,
                                         pc->rtce_window_size >>
                                         SPAPR_TCE_PAGE_SHIFT);
         address_space_init(&dev->as, spapr_tce_get_iommu(dev->tcet), qdev->id);
diff --git a/include/hw/ppc/spapr.h b/include/hw/ppc/spapr.h
index 7a0bcc5..38e61cf 100644
--- a/include/hw/ppc/spapr.h
+++ b/include/hw/ppc/spapr.h
@@ -398,6 +398,7 @@ struct sPAPRTCETable {
     DeviceState parent;
     uint32_t liobn;
     uint32_t nb_table;
+    uint32_t page_shift;
     uint64_t *table;
     bool bypass;
     int fd;
@@ -409,6 +410,7 @@ void spapr_events_init(sPAPREnvironment *spapr);
 void spapr_events_fdt_skel(void *fdt, uint32_t epow_irq);
 int spapr_h_cas_compose_response(target_ulong addr, target_ulong size);
 sPAPRTCETable *spapr_tce_new_table(DeviceState *owner, uint32_t liobn,
+                                   uint32_t page_shift,
                                    uint32_t nb_table);
 MemoryRegion *spapr_tce_get_iommu(sPAPRTCETable *tcet);
 void spapr_tce_set_bypass(sPAPRTCETable *tcet, bool bypass);
-- 
1.8.4.rc4