The migration dest will use only the value of low_bandwidth to restore the surfaces with lossy settings. But the server estimates low_bandwidth on each connection, which might not be the same after each migration. Older origin servers will still be vulnerable to assert() on destination. To mitigate the issue, let's replace the assert with a warning. https://bugzilla.redhat.com/show_bug.cgi?id=956345 --- server/red_worker.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/server/red_worker.c b/server/red_worker.c index 8ba8070..5c17b97 100644 --- a/server/red_worker.c +++ b/server/red_worker.c @@ -8770,7 +8770,10 @@ static void display_channel_marshall_migrate_data(RedChannelClient *rcc, MIGRATE_DATA_DISPLAY_MAX_CACHE_CLIENTS == MAX_CACHE_CLIENTS); display_data.message_serial = red_channel_client_get_message_serial(rcc); - display_data.low_bandwidth_setting = display_channel_client_is_low_bandwidth(dcc); + + display_data.low_bandwidth_setting = + display_channel_client_is_low_bandwidth(dcc) | + display_channel->enable_jpeg; display_data.pixmap_cache_freezer = pixmap_cache_freeze(dcc->pixmap_cache); display_data.pixmap_cache_id = dcc->pixmap_cache->id; @@ -10110,7 +10113,7 @@ static void display_channel_client_restore_surface(DisplayChannelClient *dcc, ui { /* we don't process commands till we receive the migration data, thus, * we should have not sent any surface to the client. */ - spice_assert(!dcc->surface_client_created[surface_id]); + spice_warn_if_fail(!dcc->surface_client_created[surface_id]); dcc->surface_client_created[surface_id] = TRUE; } -- 1.8.2.1.342.gfa7285d _______________________________________________ Spice-devel mailing list Spice-devel@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/spice-devel