Am 10.06.21 um 10:20 schrieb He Ying:
> When compiling the kernel for MIPS with CONFIG_DRM_AMDGPU = y, errors are
> encountered as follows:
>
> drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.o: In function `amdgpu_vram_mgr_new':
> amdgpu_vram_mgr.c:(.text+0x740): undefined reference to `__udivdi3'
>
> Making a 64 bit division by a/b (a is uint64_t) is not supported by default
> in linux kernel space. Instead, using do_div is OK for this situation. For
> this problem, using DIV_ROUND_UP_ULL instead of DIV_ROUND_UP is better.
Already fixed by this patch in drm-next:
commit 691cf8cd7a531dbfcc29d09a23c509a86fd9b24f
Author: Dave Airlie <[email protected]>
Date: Thu Jun 10 12:59:00 2021 +1000
drm/amdgpu: use correct rounding macro for 64-bit
This fixes 32-bit arm build due to lack of 64-bit divides.
Regards,
Christian.
>
> Fixes: 6a7f76e70fac ("drm/amdgpu: add VRAM manager v2")
> Reported-by: Hulk Robot <[email protected]>
> Signed-off-by: He Ying <[email protected]>
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
> index 9a6df02477ce..436ec246a7da 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
> @@ -407,7 +407,7 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man,
> #endif
> pages_per_node = max_t(uint32_t, pages_per_node,
> tbo->page_alignment);
> - num_nodes = DIV_ROUND_UP(PFN_UP(mem_bytes), pages_per_node);
> + num_nodes = DIV_ROUND_UP_ULL(PFN_UP(mem_bytes), pages_per_node);
> }
>
> node = kvmalloc(struct_size(node, mm_nodes, num_nodes),
Hello,
在 2021/6/10 16:20, Christian König 写道:
>
>
> Am 10.06.21 um 10:20 schrieb He Ying:
>> When compiling the kernel for MIPS with CONFIG_DRM_AMDGPU = y, errors
>> are
>> encountered as follows:
>>
>> drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.o: In function
>> `amdgpu_vram_mgr_new':
>> amdgpu_vram_mgr.c:(.text+0x740): undefined reference to `__udivdi3'
>>
>> Making a 64 bit division by a/b (a is uint64_t) is not supported by
>> default
>> in linux kernel space. Instead, using do_div is OK for this
>> situation. For
>> this problem, using DIV_ROUND_UP_ULL instead of DIV_ROUND_UP is better.
>
> Already fixed by this patch in drm-next:
>
> commit 691cf8cd7a531dbfcc29d09a23c509a86fd9b24f
> Author: Dave Airlie <[email protected]>
> Date: Thu Jun 10 12:59:00 2021 +1000
>
> drm/amdgpu: use correct rounding macro for 64-bit
>
> This fixes 32-bit arm build due to lack of 64-bit divides.
OK. Sigh.
Before sending my patch, I searched patches in https://lore.kernel.org/lkml/
but I didn't find this patch. How can I find whether my patch is duplicated
before sending it? Any suggestions?
Thanks.
>
> Regards,
> Christian.
>
>>
>> Fixes: 6a7f76e70fac ("drm/amdgpu: add VRAM manager v2")
>> Reported-by: Hulk Robot <[email protected]>
>> Signed-off-by: He Ying <[email protected]>
>> ---
>> drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
>> b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
>> index 9a6df02477ce..436ec246a7da 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c
>> @@ -407,7 +407,7 @@ static int amdgpu_vram_mgr_new(struct
>> ttm_resource_manager *man,
>> #endif
>> pages_per_node = max_t(uint32_t, pages_per_node,
>> tbo->page_alignment);
>> - num_nodes = DIV_ROUND_UP(PFN_UP(mem_bytes), pages_per_node);
>> + num_nodes = DIV_ROUND_UP_ULL(PFN_UP(mem_bytes),
>> pages_per_node);
>> }
>> node = kvmalloc(struct_size(node, mm_nodes, num_nodes),
>
> .