| commit | becf8217deb2afc347d5172d9f30c8a8964b8b27 | |
| author | David Hildenbrand <david@redhat.com> | |
| Fri, 9 Jun 2017 14:21:56 +0000 (9 16:21 +0200) | ||
| committer | Richard Henderson <rth@twiddle.net> | |
| Tue, 13 Jun 2017 18:09:39 +0000 (13 11:09 -0700) | ||
| tree | 47f7da6a3210b63ee7346bbf5f143abc9358d7f3 | treesnapshot (tar.gz zip) |
| parent | 465aec461726d5157d458acee94584f8cdfbe4b7 | commitdiff |
target/s390x: rework PGM interrupt psw.addr handling
We can tell from the program interrupt code, whether a program interrupt
has to forward the address in the PGM new PSW
(suppressing/terminated/completed) to point at the next instruction, or
if it is nullifying and the PSW address does not have to be incremented.
So let's not modify the PSW address outside of the injection path and
handle this internally. We just have to handle instruction length
auto detection if no valid instruction length can be provided.
This should fix various program interrupt injection paths, where the
PSW was not properly forwarded.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20170609142156.18767-3-david@redhat.com>
Signed-off-by: Richard Henderson <rth@twiddle.net>
We can tell from the program interrupt code, whether a program interrupt
has to forward the address in the PGM new PSW
(suppressing/terminated/completed) to point at the next instruction, or
if it is nullifying and the PSW address does not have to be incremented.
So let's not modify the PSW address outside of the injection path and
handle this internally. We just have to handle instruction length
auto detection if no valid instruction length can be provided.
This should fix various program interrupt injection paths, where the
PSW was not properly forwarded.
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20170609142156.18767-3-david@redhat.com>
Signed-off-by: Richard Henderson <rth@twiddle.net>