[Qemu-devel] [Applied PATCH] Split TLB addend and target_phys_addr_t

[Paul Brook]

I think I got all the tcg backend bits right, but can't test most of them. 
Please test and report any problems.

Historically the qemu tlb "addend" field was used for both RAM and IO accesses,
so needed to be able to hold both host addresses (unsigned long) and guest
physical addresses (target_phys_addr_t).  However since the introduction of
the iotlb field it has only been used for RAM accesses.

This means we can change the type of addend to unsigned long, and remove
associated hacks in the big-endian TCG backends.

We can also remove the host dependence from target_phys_addr_t.

Signed-off-by: Paul Brook <address@hidden>
---
 configure              |    3 ---
 cpu-defs.h             |   17 +++++++----------
 exec.c                 |    2 +-
 softmmu_template.h     |   28 ++++++++++++++++------------
 targphys.h             |    5 +----
 tcg/mips/tcg-target.c  |    4 ++--
 tcg/ppc/tcg-target.c   |   12 ++----------
 tcg/ppc64/tcg-target.c |   12 ++----------
 tcg/sparc/tcg-target.c |    8 +-------
 9 files changed, 32 insertions(+), 59 deletions(-)

diff --git a/configure b/configure
index 73d566f..1d5fb17 100755
--- a/configure
+++ b/configure
@@ -2538,9 +2538,6 @@ if [ "$TARGET_ABI_DIR" = "" ]; then
   TARGET_ABI_DIR=$TARGET_ARCH
 fi
 echo "TARGET_ABI_DIR=$TARGET_ABI_DIR" >> $config_target_mak
-if [ $target_phys_bits -lt $hostlongbits ] ; then
-  target_phys_bits=$hostlongbits
-fi
 case "$target_arch2" in
   i386|x86_64)
     if test "$xen" = "yes" -a "$target_softmmu" = "yes" ; then
diff --git a/cpu-defs.h b/cpu-defs.h
index 2e94585..0140596 100644
--- a/cpu-defs.h
+++ b/cpu-defs.h
@@ -76,7 +76,7 @@ typedef uint64_t target_ulong;
 #define CPU_TLB_BITS 8
 #define CPU_TLB_SIZE (1 << CPU_TLB_BITS)
 
-#if TARGET_PHYS_ADDR_BITS == 32 && TARGET_LONG_BITS == 32
+#if HOST_LONG_BITS == 32 && TARGET_LONG_BITS == 32
 #define CPU_TLB_ENTRY_BITS 4
 #else
 #define CPU_TLB_ENTRY_BITS 5
@@ -92,21 +92,18 @@ typedef struct CPUTLBEntry {
     target_ulong addr_read;
     target_ulong addr_write;
     target_ulong addr_code;
-    /* Addend to virtual address to get physical address.  IO accesses
+    /* Addend to virtual address to get host address.  IO accesses
        use the corresponding iotlb value.  */
-#if TARGET_PHYS_ADDR_BITS == 64
-    /* on i386 Linux make sure it is aligned */
-    target_phys_addr_t addend __attribute__((aligned(8)));
-#else
-    target_phys_addr_t addend;
-#endif
+    unsigned long addend;
     /* padding to get a power of two size */
     uint8_t dummy[(1 << CPU_TLB_ENTRY_BITS) - 
                   (sizeof(target_ulong) * 3 + 
-                   ((-sizeof(target_ulong) * 3) & (sizeof(target_phys_addr_t) 
- 1)) + 
-                   sizeof(target_phys_addr_t))];
+                   ((-sizeof(target_ulong) * 3) & (sizeof(unsigned long) - 1)) 
+ 
+                   sizeof(unsigned long))];
 } CPUTLBEntry;
 
+extern int CPUTLBEntry_wrong_size[sizeof(CPUTLBEntry) == (1 << 
CPU_TLB_ENTRY_BITS) ? 1 : -1];
+
 #define CPU_COMMON_TLB \
     /* The meaning of the MMU modes is defined in the target code. */   \
     CPUTLBEntry tlb_table[NB_MMU_MODES][CPU_TLB_SIZE];                  \
diff --git a/exec.c b/exec.c
index 04e74d2..33854e1 100644
--- a/exec.c
+++ b/exec.c
@@ -2188,7 +2188,7 @@ void tlb_set_page(CPUState *env, target_ulong vaddr,
     unsigned int index;
     target_ulong address;
     target_ulong code_address;
-    target_phys_addr_t addend;
+    unsigned long addend;
     CPUTLBEntry *te;
     CPUWatchpoint *wp;
     target_phys_addr_t iotlb;
diff --git a/softmmu_template.h b/softmmu_template.h
index 2f37c34..c2df9ec 100644
--- a/softmmu_template.h
+++ b/softmmu_template.h
@@ -87,7 +87,8 @@ DATA_TYPE REGPARM glue(glue(__ld, SUFFIX), 
MMUSUFFIX)(target_ulong addr,
     DATA_TYPE res;
     int index;
     target_ulong tlb_addr;
-    target_phys_addr_t addend;
+    target_phys_addr_t ioaddr;
+    unsigned long addend;
     void *retaddr;
 
     /* test if there is match for unaligned or IO access */
@@ -101,8 +102,8 @@ DATA_TYPE REGPARM glue(glue(__ld, SUFFIX), 
MMUSUFFIX)(target_ulong addr,
             if ((addr & (DATA_SIZE - 1)) != 0)
                 goto do_unaligned_access;
             retaddr = GETPC();
-            addend = env->iotlb[mmu_idx][index];
-            res = glue(io_read, SUFFIX)(addend, addr, retaddr);
+            ioaddr = env->iotlb[mmu_idx][index];
+            res = glue(io_read, SUFFIX)(ioaddr, addr, retaddr);
         } else if (((addr & ~TARGET_PAGE_MASK) + DATA_SIZE - 1) >= 
TARGET_PAGE_SIZE) {
             /* slow unaligned access (it spans two pages or IO) */
         do_unaligned_access:
@@ -143,7 +144,8 @@ static DATA_TYPE glue(glue(slow_ld, SUFFIX), 
MMUSUFFIX)(target_ulong addr,
 {
     DATA_TYPE res, res1, res2;
     int index, shift;
-    target_phys_addr_t addend;
+    target_phys_addr_t ioaddr;
+    unsigned long addend;
     target_ulong tlb_addr, addr1, addr2;
 
     index = (addr >> TARGET_PAGE_BITS) & (CPU_TLB_SIZE - 1);
@@ -154,8 +156,8 @@ static DATA_TYPE glue(glue(slow_ld, SUFFIX), 
MMUSUFFIX)(target_ulong addr,
             /* IO access */
             if ((addr & (DATA_SIZE - 1)) != 0)
                 goto do_unaligned_access;
-            addend = env->iotlb[mmu_idx][index];
-            res = glue(io_read, SUFFIX)(addend, addr, retaddr);
+            ioaddr = env->iotlb[mmu_idx][index];
+            res = glue(io_read, SUFFIX)(ioaddr, addr, retaddr);
         } else if (((addr & ~TARGET_PAGE_MASK) + DATA_SIZE - 1) >= 
TARGET_PAGE_SIZE) {
         do_unaligned_access:
             /* slow unaligned access (it spans two pages) */
@@ -224,7 +226,8 @@ void REGPARM glue(glue(__st, SUFFIX), 
MMUSUFFIX)(target_ulong addr,
                                                  DATA_TYPE val,
                                                  int mmu_idx)
 {
-    target_phys_addr_t addend;
+    target_phys_addr_t ioaddr;
+    unsigned long addend;
     target_ulong tlb_addr;
     void *retaddr;
     int index;
@@ -238,8 +241,8 @@ void REGPARM glue(glue(__st, SUFFIX), 
MMUSUFFIX)(target_ulong addr,
             if ((addr & (DATA_SIZE - 1)) != 0)
                 goto do_unaligned_access;
             retaddr = GETPC();
-            addend = env->iotlb[mmu_idx][index];
-            glue(io_write, SUFFIX)(addend, val, addr, retaddr);
+            ioaddr = env->iotlb[mmu_idx][index];
+            glue(io_write, SUFFIX)(ioaddr, val, addr, retaddr);
         } else if (((addr & ~TARGET_PAGE_MASK) + DATA_SIZE - 1) >= 
TARGET_PAGE_SIZE) {
         do_unaligned_access:
             retaddr = GETPC();
@@ -277,7 +280,8 @@ static void glue(glue(slow_st, SUFFIX), 
MMUSUFFIX)(target_ulong addr,
                                                    int mmu_idx,
                                                    void *retaddr)
 {
-    target_phys_addr_t addend;
+    target_phys_addr_t ioaddr;
+    unsigned long addend;
     target_ulong tlb_addr;
     int index, i;
 
@@ -289,8 +293,8 @@ static void glue(glue(slow_st, SUFFIX), 
MMUSUFFIX)(target_ulong addr,
             /* IO access */
             if ((addr & (DATA_SIZE - 1)) != 0)
                 goto do_unaligned_access;
-            addend = env->iotlb[mmu_idx][index];
-            glue(io_write, SUFFIX)(addend, val, addr, retaddr);
+            ioaddr = env->iotlb[mmu_idx][index];
+            glue(io_write, SUFFIX)(ioaddr, val, addr, retaddr);
         } else if (((addr & ~TARGET_PAGE_MASK) + DATA_SIZE - 1) >= 
TARGET_PAGE_SIZE) {
         do_unaligned_access:
             /* XXX: not efficient, but simple */
diff --git a/targphys.h b/targphys.h
index 99ab23c..95648d6 100644
--- a/targphys.h
+++ b/targphys.h
@@ -5,10 +5,7 @@
 
 #ifdef TARGET_PHYS_ADDR_BITS
 /* target_phys_addr_t is the type of a physical address (its size can
-   be different from 'target_ulong'). We have sizeof(target_phys_addr)
-   = max(sizeof(unsigned long),
-   sizeof(size_of_target_physical_address)) because we must pass a
-   host pointer to memory operations in some cases */
+   be different from 'target_ulong').  */
 
 #if TARGET_PHYS_ADDR_BITS == 32
 typedef uint32_t target_phys_addr_t;
diff --git a/tcg/mips/tcg-target.c b/tcg/mips/tcg-target.c
index 10b9fc1..351efb1 100644
--- a/tcg/mips/tcg-target.c
+++ b/tcg/mips/tcg-target.c
@@ -867,7 +867,7 @@ static void tcg_out_qemu_ld(TCGContext *s, const TCGArg 
*args,
     reloc_pc16(label1_ptr, (tcg_target_long) s->code_ptr);
 
     tcg_out_opc_imm(s, OPC_LW, TCG_REG_A0, TCG_REG_A0,
-                    offsetof(CPUState, tlb_table[mem_index][0].addend) + 
addr_meml);
+                    offsetof(CPUState, tlb_table[mem_index][0].addend));
     tcg_out_opc_reg(s, OPC_ADDU, TCG_REG_V0, TCG_REG_A0, addr_regl);
 #else
     if (GUEST_BASE == (int16_t)GUEST_BASE) {
@@ -1054,7 +1054,7 @@ static void tcg_out_qemu_st(TCGContext *s, const TCGArg 
*args,
     reloc_pc16(label1_ptr, (tcg_target_long) s->code_ptr);
 
     tcg_out_opc_imm(s, OPC_LW, TCG_REG_A0, TCG_REG_A0,
-                    offsetof(CPUState, tlb_table[mem_index][0].addend) + 
addr_meml);
+                    offsetof(CPUState, tlb_table[mem_index][0].addend));
     tcg_out_opc_reg(s, OPC_ADDU, TCG_REG_A0, TCG_REG_A0, addr_regl);
 #else
     if (GUEST_BASE == (int16_t)GUEST_BASE) {
diff --git a/tcg/ppc/tcg-target.c b/tcg/ppc/tcg-target.c
index 18fdd27..609035a 100644
--- a/tcg/ppc/tcg-target.c
+++ b/tcg/ppc/tcg-target.c
@@ -37,14 +37,6 @@ static uint8_t *tb_ret_addr;
 
 #define FAST_PATH
 
-#ifdef CONFIG_SOFTMMU
-#if TARGET_PHYS_ADDR_BITS <= 32
-#define ADDEND_OFFSET 0
-#else
-#define ADDEND_OFFSET 4
-#endif
-#endif
-
 #ifndef GUEST_BASE
 #define GUEST_BASE 0
 #endif
@@ -648,7 +640,7 @@ static void tcg_out_qemu_ld (TCGContext *s, const TCGArg 
*args, int opc)
     tcg_out32 (s, (LWZ
                    | RT (r0)
                    | RA (r0)
-                   | (ADDEND_OFFSET + offsetof (CPUTLBEntry, addend)
+                   | (offsetof (CPUTLBEntry, addend)
                       - offsetof (CPUTLBEntry, addr_read))
                    ));
     /* r0 = env->tlb_table[mem_index][index].addend */
@@ -847,7 +839,7 @@ static void tcg_out_qemu_st (TCGContext *s, const TCGArg 
*args, int opc)
     tcg_out32 (s, (LWZ
                    | RT (r0)
                    | RA (r0)
-                   | (ADDEND_OFFSET + offsetof (CPUTLBEntry, addend)
+                   | (offsetof (CPUTLBEntry, addend)
                       - offsetof (CPUTLBEntry, addr_write))
                    ));
     /* r0 = env->tlb_table[mem_index][index].addend */
diff --git a/tcg/ppc64/tcg-target.c b/tcg/ppc64/tcg-target.c
index 4d6e68c..2725c6e 100644
--- a/tcg/ppc64/tcg-target.c
+++ b/tcg/ppc64/tcg-target.c
@@ -28,14 +28,6 @@ static uint8_t *tb_ret_addr;
 
 #define FAST_PATH
 
-#ifdef CONFIG_SOFTMMU
-#if TARGET_PHYS_ADDR_BITS == 32
-#define LD_ADDEND LWZ
-#else
-#define LD_ADDEND LD
-#endif
-#endif
-
 #if TARGET_LONG_BITS == 32
 #define LD_ADDR LWZU
 #define CMP_L 0
@@ -684,7 +676,7 @@ static void tcg_out_qemu_ld (TCGContext *s, const TCGArg 
*args, int opc)
 #endif
 
     /* r0 now contains &env->tlb_table[mem_index][index].addr_read */
-    tcg_out32 (s, (LD_ADDEND
+    tcg_out32 (s, (LD
                    | RT (r0)
                    | RA (r0)
                    | (offsetof (CPUTLBEntry, addend)
@@ -812,7 +804,7 @@ static void tcg_out_qemu_st (TCGContext *s, const TCGArg 
*args, int opc)
     reloc_pc14 (label1_ptr, (tcg_target_long) s->code_ptr);
 #endif
 
-    tcg_out32 (s, (LD_ADDEND
+    tcg_out32 (s, (LD
                    | RT (r0)
                    | RA (r0)
                    | (offsetof (CPUTLBEntry, addend)
diff --git a/tcg/sparc/tcg-target.c b/tcg/sparc/tcg-target.c
index 6d8410c..e460d44 100644
--- a/tcg/sparc/tcg-target.c
+++ b/tcg/sparc/tcg-target.c
@@ -726,13 +726,7 @@ static const void * const qemu_st_helpers[4] = {
 #endif
 
 #if defined(CONFIG_SOFTMMU)
-#if TARGET_PHYS_ADDR_BITS == 32
-#define TARGET_ADDEND_LD_OP LDUW
-#else
-#define TARGET_ADDEND_LD_OP LDX
-#endif
-#else
-#if TARGET_ABI_BITS == 32
+#if HOST_LONG_BITS == 32
 #define TARGET_ADDEND_LD_OP LDUW
 #else
 #define TARGET_ADDEND_LD_OP LDX
-- 
1.7.0.3