Changes from v1:
* Split into teeny weeny pieces.
* It turns out the sparc_tr_tb_stop hunk of v1 was buggy,
in that things that are not simple branches use DYNAMIC_PC,
e.g. the RETT (return from trap) instruction.
Introduce DYNAMIC_PC_LOOKUP to distinguish the couple of
places where we have a dynamic pc, but no other change
of state (conditional branches, JMPL, RETURN).
* Drop the change for WRFPRS, because it's too infrequent.
The WRASI change affects memcpy/memset, so that's more important.
Boots Mark's sol8 install cdrom. :-)
Top of the profile changes from
41.55% qemu-system-sparc [.] cpu_exec_loop
14.02% qemu-system-sparc [.] cpu_tb_exec
8.74% qemu-system-sparc [.] tb_lookup
2.11% qemu-system-sparc [.] tcg_splitwx_to_rw
1.63% memfd:tcg-jit (deleted) [.] 0x0000000000000004
to
31.59% qemu-system-sparc [.] helper_lookup_tb_ptr
17.79% qemu-system-sparc [.] tb_lookup
5.38% qemu-system-sparc [.] compute_all_sub
2.38% qemu-system-sparc [.] helper_compute_psr
2.36% qemu-system-sparc [.] helper_check_align
1.79% memfd:tcg-jit (deleted) [.] 0x000000000063fc8e
This probably indicates that cpu_get_tb_cpu_state could be
improved to not consume so much overhead.
Richard Henderson (8):
target/sparc: Use tcg_gen_lookup_and_goto_ptr in gen_goto_tb
target/sparc: Fix npc comparison in sparc_tr_insn_start
target/sparc: Drop inline markers from translate.c
target/sparc: Introduce DYNAMIC_PC_LOOKUP
target/sparc: Use DYNAMIC_PC_LOOKUP for conditional branches
target/sparc: Use DYNAMIC_PC_LOOKUP for JMPL
target/sparc: Use DYNAMIC_PC_LOOKUP for v9 RETURN
target/sparc: Use tcg_gen_lookup_and_goto_ptr for v9 WRASI
target/sparc/translate.c | 410 ++++++++++++++++++++++-----------------
1 file changed, 233 insertions(+), 177 deletions(-)