The xtensa frontend calls get_page_addr_code() from its itlb_hit_test helper function. This function is really part of the TCG core's internals, and calling it from a target helper makes it awkward to make changes to that core code. It also means that we don't pass the correct retaddr to tlb_fill(), so we won't correctly handle the case where an exception is generated.
The helper is used for the instructions IHI, IHU and IPFL. Change it to call cpu_ldb_code_ra() instead. Signed-off-by: Peter Maydell <[email protected]> --- This retains the current behaviour that all these insns may cause exceptions both for MMU permissions checks failures and also for "resulting physaddr doesn't point at memory". My reading of the ISA manual is that this isn't strictly correct for IHI and IHU, which ought to only cause MMU exceptions but not actually do a memory access. If we wanted to fix that, the right thing would be to split them into their own helper function, which could then just do a tlb_fill(). Tagged as RFC because I don't have any xtensa test images. My motivation here is that at some point I'd like us to support execution from arbitrary MMIO/small MPU regions/etc, for which purpose get_page_addr_code() will change to return -1 to mean "this isn't RAM, load a single insn into a throwaway TB and execute it"... target/xtensa/op_helper.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/target/xtensa/op_helper.c b/target/xtensa/op_helper.c index 8a8c763c631..34dc1eb68fe 100644 --- a/target/xtensa/op_helper.c +++ b/target/xtensa/op_helper.c @@ -465,7 +465,11 @@ void HELPER(check_interrupts)(CPUXtensaState *env) void HELPER(itlb_hit_test)(CPUXtensaState *env, uint32_t vaddr) { - get_page_addr_code(env, vaddr); + /* + * Attempt the memory load; we don't care about the result but + * only the side-effects (ie any MMU or other exception) + */ + cpu_ldub_code_ra(env, vaddr, GETPC()); } /*! -- 2.17.1
