On 12.01.2018 13:54, David Hildenbrand wrote:
> CC == 2 can only happen due to a protection exception, not if memory is
> not available (PGM_ADDRESSING). So all PGM_ADDRESSING exceptions have to
> be forwarded to the guest.
>
> Since the initial definition of TEST PROTECTION, we now read globals
> (e.g. PSW mask), so we have to correctly mark the instruction
> (otherwise, e.g. booting fedora 27 fails).
>
> Also, the architecture explicitly specifies which exceptions are
> forwarded to the guest, this makes the code a little nicer.
>
> Signed-off-by: David Hildenbrand <da...@redhat.com>
> ---
> target/s390x/cpu.h | 2 ++
> target/s390x/helper.h | 2 +-
> target/s390x/EM_helper.c | 41 +++++++++++++++++++----------------------
> 3 files changed, 22 insertions(+), 23 deletions(-)
>
> diff --git a/target/s390x/cpu.h b/target/s390x/cpu.h
> index 1a8b6b9ae9..915bccbc75 100644
> --- a/target/s390x/cpu.h
> +++ b/target/s390x/cpu.h
> @@ -759,6 +759,8 @@ int s390_cpu_virt_mem_rw(S390CPU *cpu, vaddr laddr,
> uint8_t ar, void *hostbuf,
> s390_cpu_virt_mem_rw(cpu, laddr, ar, dest, len, false)
> #define s390_cpu_virt_mem_write(cpu, laddr, ar, dest, len) \
> s390_cpu_virt_mem_rw(cpu, laddr, ar, dest, len, true)
> +#define s390_cpu_virt_mem_check_read(cpu, laddr, ar, len) \
> + s390_cpu_virt_mem_rw(cpu, laddr, ar, NULL, len, false)
> #define s390_cpu_virt_mem_check_write(cpu, laddr, ar, len) \
> s390_cpu_virt_mem_rw(cpu, laddr, ar, NULL, len, true)
> void s390_cpu_virt_mem_handle_exc(S390CPU *cpu, uintptr_t ra);
> diff --git a/target/s390x/helper.h b/target/s390x/helper.h
> index 26c1b07b44..59a1d9869b 100644
> --- a/target/s390x/helper.h
> +++ b/target/s390x/helper.h
> @@ -137,7 +137,7 @@ DEF_HELPER_FLAGS_4(lctlg, TCG_CALL_NO_WG, void, env, i32,
> i64, i32)
> DEF_HELPER_FLAGS_4(stctl, TCG_CALL_NO_WG, void, env, i32, i64, i32)
> DEF_HELPER_FLAGS_4(stctg, TCG_CALL_NO_WG, void, env, i32, i64, i32)
> DEF_HELPER_FLAGS_2(testblock, TCG_CALL_NO_WG, i32, env, i64)
> -DEF_HELPER_FLAGS_3(tprot, TCG_CALL_NO_RWG, i32, env, i64, i64)
> +DEF_HELPER_FLAGS_3(tprot, TCG_CALL_NO_WG, i32, env, i64, i64)
> DEF_HELPER_FLAGS_2(iske, TCG_CALL_NO_RWG_SE, i64, env, i64)
> DEF_HELPER_FLAGS_3(sske, TCG_CALL_NO_RWG, void, env, i64, i64)
> DEF_HELPER_FLAGS_2(rrbe, TCG_CALL_NO_RWG, i32, env, i64)
> diff --git a/target/s390x/mem_helper.c b/target/s390x/mem_helper.c
> index 359e446c6f..c957febc6d 100644
> --- a/target/s390x/mem_helper.c
> +++ b/target/s390x/mem_helper.c
> @@ -1730,34 +1730,31 @@ uint32_t HELPER(tprot)(CPUS390XState *env, uint64_t
> a1, uint64_t a2)
> /* Fetching permitted; storing permitted */
> return 0;
> }
> +
> + if (env->int_pgm_code == PGM_PROTECTION) {
> + /* retry if reading is possible */
> + cs->exception_index = 0;
> + if (!s390_cpu_virt_mem_check_read(cpu, a1, 0, 1)) {
> + /* Fetching permitted; storing not permitted */
> + return 1;
> + }
> + }
> +
> switch (env->int_pgm_code) {
> case PGM_PROTECTION:
> - /* Fetching permitted; storing not permitted */
> - cs->exception_index = 0;
> - return 1;
> - case PGM_ADDRESSING:
> /* Fetching not permitted; storing not permitted */
> cs->exception_index = 0;
> return 2;
> - case PGM_ASCE_TYPE:
> - case PGM_REG_FIRST_TRANS:
> - case PGM_REG_SEC_TRANS:
> - case PGM_REG_THIRD_TRANS:
> - case PGM_SEGMENT_TRANS:
> - case PGM_PAGE_TRANS:
> - case PGM_ALET_SPEC:
> - case PGM_ALEN_SPEC:
> - case PGM_ALE_SEQ:
> - case PGM_ASTE_VALID:
> - case PGM_ASTE_SEQ:
> - case PGM_EXT_AUTH:
> - /* Translation not available */
> - cs->exception_index = 0;
> - return 3;
> + case PGM_ADDRESSING:
> + case PGM_TRANS_SPEC:
> + /* exceptions forwarded to the guest */
> + s390_cpu_virt_mem_handle_exc(cpu, GETPC());
> + return 0;
> }
> - /* any other exception is forwarded to the guest */
> - s390_cpu_virt_mem_handle_exc(cpu, GETPC());
> - return 0;
> +
> + /* Translation not available */
> + cs->exception_index = 0;
> + return 3;
> }
>
> /* insert storage key extended */
>
Ping, this is certainly better than the current (queued) state. Would
love to see some review. But if nobody volunteers to read the PoP, I
think my current use case (Linux and kvm-unit-tests) have to be
sufficient for now.
--
Thanks,
David / dhildenb
