On 8/3/23 11:59, Jan Beulich wrote:
On 03.08.2023 17:52, Daniel P. Smith wrote:
On 8/1/23 21:06, Stefano Stabellini wrote:
On Tue, 1 Aug 2023, Daniel P. Smith wrote:
--- a/xen/include/xen/sched.h
+++ b/xen/include/xen/sched.h
@@ -474,9 +474,8 @@ struct domain
uint8_t role;
#define CAP_CONSOLE_IO (1U<<0)
#define CAP_DISABLE_CPU_FAULT (1U<<1)
- uint8_t capabilities;
- /* Is this guest being debugged by dom0? */
- bool debugger_attached;
+#define CAP_DEBUGGER_ATTACH (1U<<2)
+ uint16_t capabilities;
No need to switch to uint16_t just yet?
I know space is tight in struct domain, wanted to reclaim the freed
space into capabilities (or roles). One thing I was considering if
enough space could be found is instead replace it all with a pointer to
a new struct that held these values. It would allow using heap space and
future growth of the structure. As of this patch, it is consuming 5
bytes and would need to find an additional 3 bytes. Is there a
willingness to entertain such an approach?
Usually we do such conversion when data belonging to a subsystem (for lack
of a better term) grows big enough, not right away when eventual data size
isn't even known yet.
Right and if there is a desire to instead follow the suggestion in my
reply to Stefano on patch 4, that approach could be sizeable. Larger
than 40 bits/flags, that I am less confident about.
v/r,
dps
