Re: [PATCH] drm/xen-front: fix pointer casts

[Juergen Gross]

On 23/05/18 12:00, Oleksandr Andrushchenko wrote:
> On 05/23/2018 12:19 PM, Juergen Gross wrote:
>> On 21/05/18 09:39, Oleksandr Andrushchenko wrote:
>>> From: Oleksandr Andrushchenko <oleksandr_andrushchenko@xxxxxxxx>
>>>
>>> Building for a 32-bit target results in warnings from casting
>>> between a 32-bit pointer and a 64-bit integer. Fix the warnings
>>> by casting those pointers to uintptr_t first.
>>>
>>> Signed-off-by: Oleksandr Andrushchenko
>>> <oleksandr_andrushchenko@xxxxxxxx>
>>> ---
>>> Â drivers/gpu/drm/xen/xen_drm_front.hÂÂÂÂÂÂ | 4 ++--
>>> Â drivers/gpu/drm/xen/xen_drm_front_shbuf.c | 2 +-
>>> Â 2 files changed, 3 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/xen/xen_drm_front.h
>>> b/drivers/gpu/drm/xen/xen_drm_front.h
>>> index 2c2479b571ae..8e15dbebc4ba 100644
>>> --- a/drivers/gpu/drm/xen/xen_drm_front.h
>>> +++ b/drivers/gpu/drm/xen/xen_drm_front.h
>>> @@ -126,12 +126,12 @@ struct xen_drm_front_drm_info {
>>> Â Â static inline u64 xen_drm_front_fb_to_cookie(struct
>>> drm_framebuffer *fb)
>>> Â {
>>> -ÂÂÂ return (u64)fb;
>>> +ÂÂÂ return (u64)(uintptr_t)fb;
>> Do you really still need the cast to u64?
> Indeed, I can remove (u64) now, thank you
>>
>>> Â }
>>> Â Â static inline u64 xen_drm_front_dbuf_to_cookie(struct
>>> drm_gem_object *gem_obj)
>>> Â {
>>> -ÂÂÂ return (u64)gem_obj;
>>> +ÂÂÂ return (u64)(uintptr_t)gem_obj;
>>> Â }
>>> Â Â int xen_drm_front_mode_set(struct xen_drm_front_drm_pipeline
>>> *pipeline,
>>> diff --git a/drivers/gpu/drm/xen/xen_drm_front_shbuf.c
>>> b/drivers/gpu/drm/xen/xen_drm_front_shbuf.c
>>> index 8099cb343ae3..47fc93847765 100644
>>> --- a/drivers/gpu/drm/xen/xen_drm_front_shbuf.c
>>> +++ b/drivers/gpu/drm/xen/xen_drm_front_shbuf.c
>>> @@ -122,7 +122,7 @@ static void guest_calc_num_grefs(struct
>>> xen_drm_front_shbuf *buf)
>>> Â }
>>> Â Â #define xen_page_to_vaddr(page) \
>>> -ÂÂÂÂÂÂÂ ((phys_addr_t)pfn_to_kaddr(page_to_xen_pfn(page)))
>>> +ÂÂÂÂÂÂÂ ((phys_addr_t)(uintptr_t)pfn_to_kaddr(page_to_xen_pfn(page)))
>> phys_addr_t for a virtual address?
> This is because the resulting value is then passed to gnttab_set_map_op/
> gnttab_set_unmap_op which expects host address to be passed as
> phys_addr_t addr :(

Okay, but this means the compiler should do the necessary cast when
you drop the cast to phys_addr_t in xen_page_to_vaddr(), as the cast
to uintptr_t is already producing an unsigned integer value which can
easily be promoted to more bits, right?


Juergen