Re: [PATCH v4 4/4] dma-buf: heaps: add chunk heap to dmabuf heaps

From: Minchan Kim <minchan@kernel.org>
To: John Stultz <john.stultz@linaro.org>
Cc: Andrew Morton <akpm@linux-foundation.org>,
	linux-mm <linux-mm@kvack.org>,
	LKML <linux-kernel@vger.kernel.org>,
	Hyesoo Yu <hyesoo.yu@samsung.com>,
	david@redhat.com, Michal Hocko <mhocko@suse.com>,
	Suren Baghdasaryan <surenb@google.com>,
	KyongHo Cho <pullip.cho@samsung.com>,
	John Dias <joaodias@google.com>,
	Hridya Valsaraju <hridya@google.com>,
	Sumit Semwal <sumit.semwal@linaro.org>,
	linux-media <linux-media@vger.kernel.org>,
	"open list:OPEN FIRMWARE AND FLATTENED DEVICE TREE BINDINGS" 
	<devicetree@vger.kernel.org>,
	Christoph Hellwig <hch@infradead.org>,
	Rob Herring <robh+dt@kernel.org>,
	"moderated list:DMA BUFFER SHARING FRAMEWORK" 
	<linaro-mm-sig@lists.linaro.org>
Subject: Re: [PATCH v4 4/4] dma-buf: heaps: add chunk heap to dmabuf heaps
Date: Tue, 26 Jan 2021 11:24:43 -0800	[thread overview]
Message-ID: <YBBse31hUxQcglig@google.com> (raw)
In-Reply-To: <CALAqxLWJDo=pM8bvt9YWrzJ+VyK5gojoL-v7ch1fQC=cjGwJOw@mail.gmail.com>

On Mon, Jan 25, 2021 at 11:32:57PM -0800, John Stultz wrote:
> On Thu, Jan 21, 2021 at 9:55 AM Minchan Kim <minchan@kernel.org> wrote:
> 
> Hey Minchan,
>   Thanks for sending this out! I'm still working through testing with
> this patch set, so I may have some more feedback tomorrow, but a few
> quick items I did hit below.
> 
> > +
> > +#define CHUNK_PREFIX "chunk-"
> > +
> > +static int register_chunk_heap(struct chunk_heap *chunk_heap_info)
> > +{
> > +       struct dma_heap_export_info exp_info;
> > +       const char *name = cma_get_name(chunk_heap_info->cma);
> > +       size_t len = strlen(CHUNK_PREFIX) + strlen(name) + 1;
> > +       char *buf = kmalloc(len, GFP_KERNEL);
> > +
> > +       if (!buf)
> > +               return -ENOMEM;
> > +
> > +       sprintf(buf, CHUNK_PREFIX"%s", cma_get_name(chunk_heap_info->cma));
> > +       buf[len] = '\0';
> > +
> > +       exp_info.name = buf;
> > +       exp_info.name = cma_get_name(chunk_heap_info->cma);
> 
> I think you intended to delete this line, as it's overwriting your
> prefixed name.

Hi John,

You're right. Will fix it.

> 
> > +       exp_info.ops = &chunk_heap_ops;
> > +       exp_info.priv = chunk_heap_info;
> > +
> > +       chunk_heap_info->heap = dma_heap_add(&exp_info);
> > +       if (IS_ERR(chunk_heap_info->heap)) {
> > +               kfree(buf);
> > +               return PTR_ERR(chunk_heap_info->heap);
> > +       }
> > +
> > +       return 0;
> > +}
> > +
> > +static int __init chunk_heap_init(void)
> > +{
> > +       unsigned int i;
> > +
> > +       for (i = 0; i < chunk_heap_count; i++)
> > +               register_chunk_heap(&chunk_heaps[i]);
> > +
> > +       return 0;
> > +}
> > +module_init(chunk_heap_init);
> > +
> > +#ifdef CONFIG_OF_EARLY_FLATTREE
> > +
> > +static int __init dmabuf_chunk_heap_area_init(struct reserved_mem *rmem)
> > +{
> > +       int ret;
> > +       struct cma *cma;
> > +       struct chunk_heap *chunk_heap_info;
> > +       const __be32 *chunk_order;
> > +
> > +       phys_addr_t align = PAGE_SIZE << max(MAX_ORDER - 1, pageblock_order);
> > +       phys_addr_t mask = align - 1;
> > +
> > +       if ((rmem->base & mask) || (rmem->size & mask)) {
> > +               pr_err("Incorrect alignment for CMA region\n");
> > +               return -EINVAL;
> 
> Passing this check can be tough if you're using dynamically assigned
> rmem, so it might be helpful for debugging to print the base/size/mask
> values?

Let me fold this into next respin.

diff --git a/drivers/dma-buf/heaps/chunk_heap.c b/drivers/dma-buf/heaps/chunk_heap.c
index 6fe8e69d108f..cc2ed5341b54 100644
--- a/drivers/dma-buf/heaps/chunk_heap.c
+++ b/drivers/dma-buf/heaps/chunk_heap.c
@@ -456,7 +456,8 @@ static int __init dmabuf_chunk_heap_area_init(struct reserved_mem *rmem)
        phys_addr_t mask = align - 1;

        if ((rmem->base & mask) || (rmem->size & mask)) {
-               pr_err("Incorrect alignment for CMA region\n");
+               pr_err("Incorrect alignment for CMA region: base %pa size %pa mask %pa\n",
+                               rmem->base, rmem->size, mask);
                return -EINVAL;
        }

Thanks for the review, John!
