We had a hack for arm and sparc, allocating code_gen_prologue to a
special section. Which, honestly does no good under certain cases.
We've already got limits on code_gen_buffer_size to ensure that all
TBs can use direct branches between themselves; reuse this limit to
ensure the prologue is also reachable.
As a bonus, we get to avoid marking a page of the main executable's
data segment as executable.
Signed-off-by: Richard Henderson <r...@twiddle.net>
---
exec.c | 30 +++++++++++-------------------
tcg/tcg.h | 2 +-
2 files changed, 12 insertions(+), 20 deletions(-)
diff --git a/exec.c b/exec.c
index 5e33a3d..8958b28 100644
--- a/exec.c
+++ b/exec.c
@@ -86,22 +86,7 @@ static int nb_tbs;
/* any access to the tbs or the page table must use this lock */
spinlock_t tb_lock = SPIN_LOCK_UNLOCKED;
-#if defined(__arm__) || defined(__sparc__)
-/* The prologue must be reachable with a direct jump. ARM and Sparc64
- have limited branch ranges (possibly also PPC) so place it in a
- section close to code segment. */
-#define code_gen_section \
- __attribute__((__section__(".gen_code"))) \
- __attribute__((aligned (32)))
-#elif defined(_WIN32) && !defined(_WIN64)
-#define code_gen_section \
- __attribute__((aligned (16)))
-#else
-#define code_gen_section \
- __attribute__((aligned (32)))
-#endif
-
-uint8_t code_gen_prologue[1024] code_gen_section;
+uint8_t *code_gen_prologue;
static uint8_t *code_gen_buffer;
static size_t code_gen_buffer_size;
/* threshold to flush the translated code buffer */
@@ -221,7 +206,7 @@ static int tb_flush_count;
static int tb_phys_invalidate_count;
#ifdef _WIN32
-static void map_exec(void *addr, long size)
+static inline void map_exec(void *addr, long size)
{
DWORD old_protect;
VirtualProtect(addr, size,
@@ -229,7 +214,7 @@ static void map_exec(void *addr, long size)
}
#else
-static void map_exec(void *addr, long size)
+static inline void map_exec(void *addr, long size)
{
unsigned long start, end, page_size;
@@ -621,7 +606,14 @@ static inline void code_gen_alloc(size_t tb_size)
exit(1);
}
- map_exec(code_gen_prologue, sizeof(code_gen_prologue));
+ /* Steal room for the prologue at the end of the buffer. This ensures
+ (via the MAX_CODE_GEN_BUFFER_SIZE limits above) that direct branches
+ from TB's to the prologue are going to be in range. It also means
+ that we don't need to mark (additional) portions of the data segment
+ as executable. */
+ code_gen_prologue = code_gen_buffer + code_gen_buffer_size - 1024;
+ code_gen_buffer_size -= 1024;
+
code_gen_buffer_max_size = code_gen_buffer_size -
(TCG_MAX_OP_SIZE * OPC_BUF_SIZE);
code_gen_max_blocks = code_gen_buffer_size / CODE_GEN_AVG_BLOCK_SIZE;
diff --git a/tcg/tcg.h b/tcg/tcg.h
index 7bafe0e..45e94f5 100644
--- a/tcg/tcg.h
+++ b/tcg/tcg.h
@@ -616,7 +616,7 @@ TCGv_i64 tcg_const_i64(int64_t val);
TCGv_i32 tcg_const_local_i32(int32_t val);
TCGv_i64 tcg_const_local_i64(int64_t val);
-extern uint8_t code_gen_prologue[];
+extern uint8_t *code_gen_prologue;
/* TCG targets may use a different definition of tcg_qemu_tb_exec. */
#if !defined(tcg_qemu_tb_exec)
--
1.7.11.7
