[Qemu-devel] [PATCH 5/5] tcg/arm: improve direct jump

[Aurelien Jarno]

Use ldr pc, [pc, #-4] kind of branch for direct jump. This removes the
need to flush the icache on TB linking, and allow to remove the limit
on the code generation buffer.

This improves the boot-up speed of a MIPS guest by 11%.

Cc: Andrzej Zaborowski <address@hidden>
Cc: Peter Maydell <address@hidden>
Signed-off-by: Aurelien Jarno <address@hidden>
---
 exec-all.h           |   24 ++++--------------------
 exec.c               |    4 ----
 tcg/arm/tcg-target.c |    7 +------
 3 files changed, 5 insertions(+), 30 deletions(-)

diff --git a/exec-all.h b/exec-all.h
index 6516da0..662b916 100644
--- a/exec-all.h
+++ b/exec-all.h
@@ -224,26 +224,10 @@ static inline void tb_set_jmp_target1(uintptr_t jmp_addr, 
uintptr_t addr)
 #elif defined(__arm__)
 static inline void tb_set_jmp_target1(uintptr_t jmp_addr, uintptr_t addr)
 {
-#if !QEMU_GNUC_PREREQ(4, 1)
-    register unsigned long _beg __asm ("a1");
-    register unsigned long _end __asm ("a2");
-    register unsigned long _flg __asm ("a3");
-#endif
-
-    /* we could use a ldr pc, [pc, #-4] kind of branch and avoid the flush */
-    *(uint32_t *)jmp_addr =
-        (*(uint32_t *)jmp_addr & ~0xffffff)
-        | (((addr - (jmp_addr + 8)) >> 2) & 0xffffff);
-
-#if QEMU_GNUC_PREREQ(4, 1)
-    __builtin___clear_cache((char *) jmp_addr, (char *) jmp_addr + 4);
-#else
-    /* flush icache */
-    _beg = jmp_addr;
-    _end = jmp_addr + 4;
-    _flg = 0;
-    __asm __volatile__ ("swi 0x9f0002" : : "r" (_beg), "r" (_end), "r" (_flg));
-#endif
+    /* Patch the branch destination. It uses a ldr pc, [pc, #-4] kind
+       of branch so we write absolute address and we don't need to
+       flush icache. */
+    *(uint32_t *)jmp_addr = addr;
 }
 #elif defined(__sparc__)
 void tb_set_jmp_target1(uintptr_t jmp_addr, uintptr_t addr);
diff --git a/exec.c b/exec.c
index 7899042..8d115ac 100644
--- a/exec.c
+++ b/exec.c
@@ -546,10 +546,6 @@ static void code_gen_alloc(unsigned long tb_size)
         start = (void *) 0x40000000UL;
         if (code_gen_buffer_size > (512 * 1024 * 1024))
             code_gen_buffer_size = (512 * 1024 * 1024);
-#elif defined(__arm__)
-        /* Keep the buffer no bigger than 16MB to branch between blocks */
-        if (code_gen_buffer_size > 16 * 1024 * 1024)
-            code_gen_buffer_size = 16 * 1024 * 1024;
 #elif defined(__s390x__)
         /* Map the buffer so that we can use direct calls and branches.  */
         /* We have a +- 4GB range on the branches; leave some slop.  */
diff --git a/tcg/arm/tcg-target.c b/tcg/arm/tcg-target.c
index fafbd5d..e04cfa7 100644
--- a/tcg/arm/tcg-target.c
+++ b/tcg/arm/tcg-target.c
@@ -1501,14 +1501,9 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode 
opc,
     case INDEX_op_goto_tb:
         if (s->tb_jmp_offset) {
             /* Direct jump method */
-#if defined(USE_DIRECT_JUMP)
-            s->tb_jmp_offset[args[0]] = s->code_ptr - s->code_buf;
-            tcg_out_b_noaddr(s, COND_AL);
-#else
             tcg_out_ld32_12(s, COND_AL, TCG_REG_PC, TCG_REG_PC, -4);
             s->tb_jmp_offset[args[0]] = s->code_ptr - s->code_buf;
-            tcg_out32(s, 0);
-#endif
+            tcg_out32(s, (int)s->code_ptr + 4);
         } else {
             /* Indirect jump method */
 #if 1
-- 
1.7.10.4