| commit | b5e85329026115b5a679849f45e7c19c2714e4fd | |
| author | David Hildenbrand <david@redhat.com> | |
| Fri, 12 Jan 2018 12:54:52 +0000 (12 13:54 +0100) | ||
| committer | Cornelia Huck <cohuck@redhat.com> | |
| Mon, 22 Jan 2018 10:04:52 +0000 (22 11:04 +0100) | ||
| tree | 4e1b8e2008a670e87e2b4c65b50c9c4ab301788a | treesnapshot (tar.gz zip) |
| parent | 46fa893355e0bd88f3c59b886f0d75cbd5f0bbbe | commitdiff |
s390x/tcg: fixup TEST PROTECTION
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 <david@redhat.com>
Message-Id: <20180112125452.8569-1-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
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 <david@redhat.com>
Message-Id: <20180112125452.8569-1-david@redhat.com>
Signed-off-by: Cornelia Huck <cohuck@redhat.com>
| target/s390x/cpu.h | diffblobblamehistory | |
| target/s390x/helper.h | diffblobblamehistory | |
| target/s390x/mem_helper.c | diffblobblamehistory |