Hello Andrew Jones,
The patch 14c47b7530e2: "kvm: selftests: introduce ucall" from Sep
18, 2018, leads to the following static checker warning:
./tools/testing/selftests/kvm/lib/ucall.c:61 ucall_init()
warn: always true condition '(gpa >= 0) => (0-u64max >= 0)'
./tools/testing/selftests/kvm/lib/ucall.c
28 void ucall_init(struct kvm_vm *vm, ucall_type_t type, void *arg)
29 {
30 ucall_type = type;
31 sync_global_to_guest(vm, ucall_type);
32
33 if (type == UCALL_PIO)
34 return;
35
36 if (type == UCALL_MMIO) {
37 vm_paddr_t gpa, start, end, step;
vm_paddr_t is a u64.
38 bool ret;
39
40 if (arg) {
41 gpa = (vm_paddr_t)arg;
42 ret = ucall_mmio_init(vm, gpa);
43 TEST_ASSERT(ret, "Can't set ucall mmio address to %lx", gpa);
44 return;
45 }
46
47 /*
48 * Find an address within the allowed virtual address space,
49 * that does _not_ have a KVM memory region associated with it.
50 * Identity mapping an address like this allows the guest to
51 * access it, but as KVM doesn't know what to do with it, it
52 * will assume it's something userspace handles and exit with
53 * KVM_EXIT_MMIO. Well, at least that's how it works for AArch64.
54 * Here we start with a guess that the addresses around two
55 * thirds of the VA space are unmapped and then work both down
56 * and up from there in 1/6 VA space sized steps.
57 */
58 start = 1ul << (vm->va_bits * 2 / 3);
59 end = 1ul << vm->va_bits;
60 step = 1ul << (vm->va_bits / 6);
61 for (gpa = start; gpa >= 0; gpa -= step) {
^^^^^^^^
So this doesn't work.
62 if (ucall_mmio_init(vm, gpa & ~(vm->page_size - 1)))
63 return;
64 }
65 for (gpa = start + step; gpa < end; gpa += step) {
66 if (ucall_mmio_init(vm, gpa & ~(vm->page_size - 1)))
67 return;
68 }
69 TEST_ASSERT(false, "Can't find a ucall mmio address");
70 }
71 }
regards,
dan carpenter
