At present, the "average" guestimate of TB size is way too small, leading
to many unused entries in the pre-allocated TB array. For a guest with 1GB
ram, we're currently allocating 256MB for the array.
Survey arm, alpha, aarch64, ppc, sparc, i686, x86_64 guests running on
x86_64 and ppc64 hosts and select a new average. The size of the array
drops to 81MB with no more flushing than before.
Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Richard Henderson <rth@twiddle.net>
#define CODE_GEN_PHYS_HASH_BITS 15
#define CODE_GEN_PHYS_HASH_SIZE (1 << CODE_GEN_PHYS_HASH_BITS)
-/* estimated block size for TB allocation */
-/* XXX: use a per code average code fragment size and modulate it
- according to the host CPU */
+/* Estimated block size for TB allocation. */
+/* ??? The following is based on a 2015 survey of x86_64 host output.
+ Better would seem to be some sort of dynamically sized TB array,
+ adapting to the block sizes actually being produced. */
#if defined(CONFIG_SOFTMMU)
-#define CODE_GEN_AVG_BLOCK_SIZE 128
+#define CODE_GEN_AVG_BLOCK_SIZE 400
#else
-#define CODE_GEN_AVG_BLOCK_SIZE 64
+#define CODE_GEN_AVG_BLOCK_SIZE 150
#endif
#if defined(__arm__) || defined(_ARCH_PPC) \