On 04/05/2011 03:16 PM, David Gibson wrote:
On Tue, Apr 05, 2011 at 10:05:12AM +0200, Alexander Graf wrote:
On 05.04.2011, at 07:12, David Gibson wrote:
[snip]
+struct icp_state *xics_system_init(int nr_irqs)
{
+ CPUState *env;
+ int max_server_num;
int i;
struct icp_state *icp;
struct ics_state *ics;
+ max_server_num = -1;
+ for (env = first_cpu; env != NULL; env = env->next_cpu) {
+ if (env->cpu_index> max_server_num) {
+ max_server_num = env->cpu_index;
+ }
+ }
+
icp = qemu_mallocz(sizeof(*icp));
- icp->nr_servers = nr_servers;
- icp->ss = qemu_mallocz(nr_servers * sizeof(struct icp_server_state));
+ icp->nr_servers = max_server_num + 1;
+ icp->ss = qemu_mallocz(icp->nr_servers*sizeof(struct icp_server_state));
+
+ for (i = 0; i< icp->nr_servers; i++) {
+ icp->ss[i].mfrr = 0xff;
Is ss always big enough to hold all CPUs + 1?
Well, it should be. We allocated it just above based on
icp->nr_servers, and nr_servers is the largest cpu_index + 1, computer
just above that
Ah, there it is :). I missed that part. Will apply the patches to my tree.
Alex
