If dma_direct_alloc() alloc memory in size of 64MB, the inner function
dma_common_contiguous_remap() will allocate 128KB memory by invoking 
the function kmalloc_array(). and the kmalloc_array seems to fail to try to
allocate 128KB mem. work around by doing kvmalloc_array instead.

Signed-off-by: Gao Xu <gaoxu2@hihonor.com>
---
 kernel/dma/remap.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/dma/remap.c b/kernel/dma/remap.c index b45266680..27596f3b4 100644
--- a/kernel/dma/remap.c
+++ b/kernel/dma/remap.c
@@ -43,13 +43,13 @@ void *dma_common_contiguous_remap(struct page *page, size_t size,
 	void *vaddr;
 	int i;
 
-	pages = kmalloc_array(count, sizeof(struct page *), GFP_KERNEL);
+	pages = kvmalloc_array(count, sizeof(struct page *), GFP_KERNEL);
 	if (!pages)
 		return NULL;
 	for (i = 0; i < count; i++)
 		pages[i] = nth_page(page, i);
 	vaddr = vmap(pages, count, VM_DMA_COHERENT, prot);
-	kfree(pages);
+	kvfree(pages);
 
 	return vaddr;
 }
--
2.17.1

On Sat, May 6, 2023 at 12:33 AM gaoxu 00016977 <gaoxu2@hihonor.com> wrote:
>
> If dma_direct_alloc() alloc memory in size of 64MB, the inner function
> dma_common_contiguous_remap() will allocate 128KB memory by invoking
> the function kmalloc_array(). and the kmalloc_array seems to fail to try to
> allocate 128KB mem. work around by doing kvmalloc_array instead.

On systems like Android it's not unusual for order 5 allocation to
fail. The change seems safe to me with no downside since kvmalloc will
try kmalloc first.

>
> Signed-off-by: Gao Xu <gaoxu2@hihonor.com>

Reviewed-by: Suren Baghdasaryan <surenb@google.com>

> ---
>  kernel/dma/remap.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/dma/remap.c b/kernel/dma/remap.c index b45266680..27596f3b4 100644
> --- a/kernel/dma/remap.c
> +++ b/kernel/dma/remap.c
> @@ -43,13 +43,13 @@ void *dma_common_contiguous_remap(struct page *page, size_t size,
>         void *vaddr;
>         int i;
>
> -       pages = kmalloc_array(count, sizeof(struct page *), GFP_KERNEL);
> +       pages = kvmalloc_array(count, sizeof(struct page *), GFP_KERNEL);
>         if (!pages)
>                 return NULL;
>         for (i = 0; i < count; i++)
>                 pages[i] = nth_page(page, i);
>         vaddr = vmap(pages, count, VM_DMA_COHERENT, prot);
> -       kfree(pages);
> +       kvfree(pages);
>
>         return vaddr;
>  }
> --
> 2.17.1
>

Please help to review, thanks.

-----Original Message-----
From: gaoxu 00016977 
Sent: 2023年5月6日 15:34
To: 'hch@lst.de' <hch@lst.de>; 'm.szyprowski@samsung.com' <m.szyprowski@samsung.com>
Cc: 'robin.murphy@arm.com' <robin.murphy@arm.com>; 'iommu@lists.linux.dev' <iommu@lists.linux.dev>; 'linux-kernel@vger.kernel.org' <linux-kernel@vger.kernel.org>; 'surenb@google.com' <surenb@google.com>; yipengxiang 00013268 <yipengxiang@hihonor.com>; wangbintian 00013160 <bintian.wang@hihonor.com>; hanfeng 00012985 <feng.han@hihonor.com>
Subject: [PATCH] dma-remap: Use kvmalloc_array/kvfree for larger dma memory remap

If dma_direct_alloc() alloc memory in size of 64MB, the inner function
dma_common_contiguous_remap() will allocate 128KB memory by invoking the function kmalloc_array(). and the kmalloc_array seems to fail to try to allocate 128KB mem. work around by doing kvmalloc_array instead.

Signed-off-by: Gao Xu <gaoxu2@hihonor.com>
---
 kernel/dma/remap.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/dma/remap.c b/kernel/dma/remap.c index b45266680..27596f3b4 100644
--- a/kernel/dma/remap.c
+++ b/kernel/dma/remap.c
@@ -43,13 +43,13 @@ void *dma_common_contiguous_remap(struct page *page, size_t size,
 	void *vaddr;
 	int i;
 
-	pages = kmalloc_array(count, sizeof(struct page *), GFP_KERNEL);
+	pages = kvmalloc_array(count, sizeof(struct page *), GFP_KERNEL);
 	if (!pages)
 		return NULL;
 	for (i = 0; i < count; i++)
 		pages[i] = nth_page(page, i);
 	vaddr = vmap(pages, count, VM_DMA_COHERENT, prot);
-	kfree(pages);
+	kvfree(pages);
 
 	return vaddr;
 }
--
2.17.1