On Mon, Jul 19, 2021 at 8:36 PM Yunsheng Lin <linyunsheng@xxxxxxxxxx> wrote:
>
> For 32 bit systems with 64 bit dma, dma_addr[1] is used to
> store the upper 32 bit dma addr, those system should be rare
> those days.
>
> For normal system, the dma_addr[1] in 'struct page' is not
> used, so we can reuse dma_addr[1] for storing frag count,
> which means how many frags this page might be splited to.
>
> In order to simplify the page frag support in the page pool,
> the PAGE_POOL_DMA_USE_PP_FRAG_COUNT macro is added to indicate
> the 32 bit systems with 64 bit dma, and the page frag support
> in page pool is disabled for such system.
>
> The newly added page_pool_set_frag_count() is called to reserve
> the maximum frag count before any page frag is passed to the
> user. The page_pool_atomic_sub_frag_count_return() is called
> when user is done with the page frag.
>
> Signed-off-by: Yunsheng Lin <linyunsheng@xxxxxxxxxx>
> ---
> include/linux/mm_types.h | 18 +++++++++++++-----
> include/net/page_pool.h | 41 ++++++++++++++++++++++++++++++++++-------
> net/core/page_pool.c | 4 ++++
> 3 files changed, 51 insertions(+), 12 deletions(-)
>
<snip>
> +static inline long page_pool_atomic_sub_frag_count_return(struct page *page,
> + long nr)
> +{
> + long frag_count = atomic_long_read(&page->pp_frag_count);
> + long ret;
> +
> + if (frag_count == nr)
> + return 0;
> +
> + ret = atomic_long_sub_return(nr, &page->pp_frag_count);
> + WARN_ON(ret < 0);
> + return ret;
> }
>
So this should just be an atomic_long_sub_return call. You should get
rid of the atomic_long_read portion of this as it can cover up
reference count errors.
