[Qemu-devel] [PULL for-2.4 02/11] css: mss/mcss-e vs. migration

[Cornelia Huck]

Our main channel_subsys structure is not a device (yet), but we need
to setup mss/mcss-e again if the guest had enabled it before. Use
a hack that should catch most configurations (assuming that the guest
will have enabled at least one device in higher subchannel sets or
channel subsystems if it enabled the functionality.)

Signed-off-by: Cornelia Huck <address@hidden>
---
 hw/s390x/css.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/hw/s390x/css.c b/hw/s390x/css.c
index 5561d80..a9cf3d7 100644
--- a/hw/s390x/css.c
+++ b/hw/s390x/css.c
@@ -1464,6 +1464,21 @@ int subch_device_load(SubchDev *s, QEMUFile *f)
     }
     s->ccw_fmt_1 = qemu_get_byte(f);
     s->ccw_no_data_cnt = qemu_get_byte(f);
+    /*
+     * Hack alert. We don't migrate the channel subsystem status (no
+     * device!), but we need to find out if the guest enabled mss/mcss-e.
+     * If the subchannel is enabled, it certainly was able to access it,
+     * so adjust the max_ssid/max_cssid values for relevant ssid/cssid
+     * values. This is not watertight, but better than nothing.
+     */
+    if (s->curr_status.pmcw.flags & PMCW_FLAGS_MASK_ENA) {
+        if (s->ssid) {
+            channel_subsys->max_ssid = MAX_SSID;
+        }
+        if (s->cssid != channel_subsys->default_cssid) {
+            channel_subsys->max_cssid = MAX_CSSID;
+        }
+    }
     return 0;
 }
 
-- 
2.4.5