On Mon, 21 Aug 2023 at 11:38, Ilias Apalodimas
[off-list ref] wrote:

resending for the mailing list apologies for the noise.


On Sat, 19 Aug 2023 at 00:51, Jakub Kicinski [off-list ref] wrote:

quoted

On Fri, 18 Aug 2023 09:12:09 +0300 Ilias Apalodimas wrote:

quoted

quoted

Right, IIUC we don't have enough space to fit dma_addr_t and the
refcount, but if we store the dma addr on a shifted u32 instead
of using dma_addr_t explicitly - the refcount should fit?

struct page looks like this:

unsigned long dma_addr;
union {
      unsigned long dma_addr_upper;
      atomic_long_t pp_frag_count;
};

I could be completely misunderstanding the problem.

You aren't!

quoted

Let me show you the diff of what I was thinking more or less.

diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index 5e74ce4a28cd..58ffa8dc745f 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h

@@ -126,11 +126,6 @@ struct page {
                        unsigned long _pp_mapping_pad;
                        unsigned long dma_addr;
                        union {
-                               /**
-                                * dma_addr_upper: might require a 64-bit
-                                * value on 32-bit architectures.
-                                */
-                               unsigned long dma_addr_upper;
                                /**
                                 * For frag page support, not supported in
                                 * 32-bit architectures with 64-bit DMA.

diff --git a/include/net/page_pool/helpers.h b/include/net/page_pool/helpers.h
index 94231533a369..6f87a0fa2178 100644
--- a/include/net/page_pool/helpers.h
+++ b/include/net/page_pool/helpers.h

@@ -212,16 +212,24 @@ static inline dma_addr_t page_pool_get_dma_addr(struct page *page)
        dma_addr_t ret = page->dma_addr;

        if (PAGE_POOL_DMA_USE_PP_FRAG_COUNT)
-               ret |= (dma_addr_t)page->dma_addr_upper << 16 << 16;
+               ret <<= PAGE_SHIFT;

        return ret;
 }

-static inline void page_pool_set_dma_addr(struct page *page, dma_addr_t addr)
+static inline bool page_pool_set_dma_addr(struct page *page, dma_addr_t addr)
 {
+       bool failed = false;
+
        page->dma_addr = addr;
-       if (PAGE_POOL_DMA_USE_PP_FRAG_COUNT)
-               page->dma_addr_upper = upper_32_bits(addr);
+       if (PAGE_POOL_DMA_USE_PP_FRAG_COUNT) {
+               page->dma_addr >>= PAGE_SHIFT;
+               /* We assume page alignment to shave off bottom bits,
+                * if this "compression" doesn't work we need to drop.
+                */
+               failed = addr != page->dma_addr << PAGE_SHIFT;
+       }
+       return failed;
 }

 static inline bool page_pool_put(struct page_pool *pool)

diff --git a/net/core/page_pool.c b/net/core/page_pool.c
index 77cb75e63aca..9ea42e242a89 100644
--- a/net/core/page_pool.c
+++ b/net/core/page_pool.c

@@ -211,10 +211,6 @@ static int page_pool_init(struct page_pool *pool,
                 */
        }

-       if (PAGE_POOL_DMA_USE_PP_FRAG_COUNT &&
-           pool->p.flags & PP_FLAG_PAGE_FRAG)
-               return -EINVAL;
-
 #ifdef CONFIG_PAGE_POOL_STATS
        pool->recycle_stats = alloc_percpu(struct page_pool_recycle_stats);
        if (!pool->recycle_stats)

@@ -359,12 +355,19 @@ static bool page_pool_dma_map(struct page_pool *pool, struct page *page)
        if (dma_mapping_error(pool->p.dev, dma))
                return false;

-       page_pool_set_dma_addr(page, dma);
+       if (page_pool_set_dma_addr(page, dma))
+               goto unmap_failed;

        if (pool->p.flags & PP_FLAG_DMA_SYNC_DEV)
                page_pool_dma_sync_for_device(pool, page, pool->p.max_len);

        return true;
+
+unmap_failed:
+       dma_unmap_page_attrs(pool->p.dev, dma,
+                            PAGE_SIZE << pool->p.order, pool->p.dma_dir,
+                            DMA_ATTR_SKIP_CPU_SYNC | DMA_ATTR_WEAK_ORDERING);
+       return false;
 }

That seems reasonable and would work for pages > 4k as well. But is
16TB enough?  I am more familiar with embedded than large servers,
which do tend to scale that high.

Right never respond before coffee .... I think this is reasonable overall.

Thanks
/Ilias

Regards
/Ilias

quoted

 static void page_pool_set_pp_info(struct page_pool *pool,