Hi Andre,
On 13/02/18 18:17, Andre Przywara wrote:
On 13/02/18 16:52, Julien Grall wrote:
+struct vgic_register_region {
+ unsigned int reg_offset;
+ unsigned int len;
+ unsigned int bits_per_irq;
+ unsigned int access_flags;
+ union
+ {
+ unsigned long (*read)(struct vcpu *vcpu, paddr_t addr,
+ unsigned int len);
+ unsigned long (*its_read)(struct domain *d, struct vgic_its
*its,
+ paddr_t addr, unsigned int len);
+ };
+ union
+ {
+ void (*write)(struct vcpu *vcpu, paddr_t addr,
+ unsigned int len, unsigned long val);
+ void (*its_write)(struct domain *d, struct vgic_its *its,
+ paddr_t addr, unsigned int len,
+ unsigned long val);
+ };
+ unsigned long (*uaccess_read)(struct vcpu *vcpu, paddr_t addr,
+ unsigned int len);
+ union
+ {
+ void (*uaccess_write)(struct vcpu *vcpu, paddr_t addr,
+ unsigned int len, unsigned long val);
+ int (*uaccess_its_write)(struct domain *d, struct vgic_its *its,
+ paddr_t addr, unsigned int len,
+ unsigned long val);
+ };
I don't think uaccess helpers makes sense for Xen.
True, I was unsure about whether to keep them. I have the gut feeling we
need it later when we want to suspend/resume the VGIC, so removing
everything and then simplifying the code afterwards might bite us in the
future.
So as long as it doesn't really hurt, I am tempted to keep that code in,
which also keeps it closer the the KVM implementation.
I don't want to see code that is going to potentially rot. If we really
need it, we can add it afterwards.
Cheers,
--
Julien Grall
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
