On 2026-03-13 06:44, Andrew Cooper wrote:
On 13/03/2026 8:33 am, Jiri Slaby wrote:
On 12. 03. 26, 18:38, Jason Andryuk wrote:
--- a/include/xen/interface/io/console.h
+++ b/include/xen/interface/io/console.h
@@ -19,6 +19,19 @@ struct xencons_interface {
char out[2048];
XENCONS_RING_IDX in_cons, in_prod;
XENCONS_RING_IDX out_cons, out_prod;
+/*
+ * Flag values signaling from backend to frontend whether the
console is
+ * connected. i.e. Whether it will be serviced and emptied.
+ *
+ * The flag starts as disconnected.
+ */
+#define XENCONSOLE_DISCONNECTED 1
+/*
+ * The flag is set to connected when the backend connects and the
console
+ * will be serviced.
+ */
+#define XENCONSOLE_CONNECTED 0
This all should be an enum. And you can document it using kernel-doc
properly then.
+ uint8_t connection;
And type check as well.
This is a non-Linux header file being re-sync'd with it's original source.
It describes an ABI between VMs, where things like enum are forbidden.
Yes, it is as Andrew wrote. I included "Update the console.h header to
bring in the new field." in the commit message to try in indicate that.
"Sync console.h from the xen repo to bring in the new field." would
better explain its origin.
I only brought in the needed part. There is an emacs variable block
that is not synced for other headers. There is also an ifdef block for
XEN_WANT_FLEX_CONSOLE_RING that isn't used by linux, which I did not sync.
Thanks,
Jason
