We were using the wrong types in kvm_get_dirty_pages_log_range(). This would break dirty logging if the region to log resides at a physical address above 32-bit, which can happen with 64-bit guest but also with 32-bit guests that support >32-bit physical addresses such as ppc BookE.
Signed-off-by: Benjamin Herrenschmidt <b...@kernel.crashing.org> --- kvm-all.c | 7 ++++--- 1 files changed, 4 insertions(+), 3 deletions(-) diff --git a/kvm-all.c b/kvm-all.c index 4c466d6..91b0694 100644 --- a/kvm-all.c +++ b/kvm-all.c @@ -328,13 +328,14 @@ static int kvm_set_migration_log(int enable) } /* get kvm's dirty pages bitmap and update qemu's */ -static int kvm_get_dirty_pages_log_range(unsigned long start_addr, +static int kvm_get_dirty_pages_log_range(target_phys_addr_t start_addr, unsigned long *bitmap, - unsigned long offset, + target_phys_addr_t offset, unsigned long mem_size) { unsigned int i, j; - unsigned long page_number, addr, addr1, c; + unsigned long page_number, c; + target_phys_addr_t addr, addr1; ram_addr_t ram_addr; unsigned int len = ((mem_size / TARGET_PAGE_SIZE) + HOST_LONG_BITS - 1) / HOST_LONG_BITS;