Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Subject: Re: [PATCH 6/9] nouveau: simplify nouveau_dmem_migrate_vma
To:     Christoph Hellwig <hch@lst.de>,
        =?UTF-8?B?SsOpcsO0bWUgR2xpc3Nl?= <jglisse@redhat.com>,
        Jason Gunthorpe <jgg@mellanox.com>,
        Ben Skeggs <bskeggs@redhat.com>
CC:     Bharata B Rao <bharata@linux.ibm.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        <linux-mm@kvack.org>, <nouveau@lists.freedesktop.org>,
        <dri-devel@lists.freedesktop.org>, <linux-kernel@vger.kernel.org>
References: <20190729142843.22320-1-hch@lst.de>
 <20190729142843.22320-7-hch@lst.de>
From:   Ralph Campbell <rcampbell@nvidia.com>
Message-ID: <ceb593fb-840b-c915-be8e-77d056e26628@nvidia.com>
Date:   Mon, 29 Jul 2019 16:27:58 -0700
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101
 Thunderbird/60.7.0
MIME-Version: 1.0
In-Reply-To: <20190729142843.22320-7-hch@lst.de>
Content-Type: text/plain; charset="utf-8"; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


On 7/29/19 7:28 AM, Christoph Hellwig wrote:
> Factor the main copy page to vram routine out into a helper that acts
> on a single page and which doesn't require the nouveau_dmem_migrate
> structure for argument passing.  As an added benefit the new version
> only allocates the dma address array once and reuses it for each
> subsequent chunk of work.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

Reviewed-by: Ralph Campbell <rcampbell@nvidia.com>

> ---
>   drivers/gpu/drm/nouveau/nouveau_dmem.c | 185 ++++++++-----------------
>   1 file changed, 56 insertions(+), 129 deletions(-)
> 
> diff --git a/drivers/gpu/drm/nouveau/nouveau_dmem.c b/drivers/gpu/drm/nouveau/nouveau_dmem.c
> index 036e6c07d489..6cb930755970 100644
> --- a/drivers/gpu/drm/nouveau/nouveau_dmem.c
> +++ b/drivers/gpu/drm/nouveau/nouveau_dmem.c
> @@ -44,8 +44,6 @@
>   #define DMEM_CHUNK_SIZE (2UL << 20)
>   #define DMEM_CHUNK_NPAGES (DMEM_CHUNK_SIZE >> PAGE_SHIFT)
>   
> -struct nouveau_migrate;
> -
>   enum nouveau_aper {
>   	NOUVEAU_APER_VIRT,
>   	NOUVEAU_APER_VRAM,
> @@ -86,15 +84,6 @@ static inline struct nouveau_dmem *page_to_dmem(struct page *page)
>   	return container_of(page->pgmap, struct nouveau_dmem, pagemap);
>   }
>   
> -struct nouveau_migrate {
> -	struct vm_area_struct *vma;
> -	struct nouveau_drm *drm;
> -	struct nouveau_fence *fence;
> -	unsigned long npages;
> -	dma_addr_t *dma;
> -	unsigned long dma_nr;
> -};
> -
>   static unsigned long nouveau_dmem_page_addr(struct page *page)
>   {
>   	struct nouveau_dmem_chunk *chunk = page->zone_device_data;
> @@ -569,131 +558,67 @@ nouveau_dmem_init(struct nouveau_drm *drm)
>   	drm->dmem = NULL;
>   }
>   
> -static void
> -nouveau_dmem_migrate_alloc_and_copy(struct vm_area_struct *vma,
> -				    const unsigned long *src_pfns,
> -				    unsigned long *dst_pfns,
> -				    unsigned long start,
> -				    unsigned long end,
> -				    struct nouveau_migrate *migrate)
> +static unsigned long nouveau_dmem_migrate_copy_one(struct nouveau_drm *drm,
> +		struct vm_area_struct *vma, unsigned long addr,
> +		unsigned long src, dma_addr_t *dma_addr)
>   {
> -	struct nouveau_drm *drm = migrate->drm;
>   	struct device *dev = drm->dev->dev;
> -	unsigned long addr, i, npages = 0;
> -	nouveau_migrate_copy_t copy;
> -	int ret;
> -
> -	/* First allocate new memory */
> -	for (addr = start, i = 0; addr < end; addr += PAGE_SIZE, i++) {
> -		struct page *dpage, *spage;
> -
> -		dst_pfns[i] = 0;
> -		spage = migrate_pfn_to_page(src_pfns[i]);
> -		if (!spage || !(src_pfns[i] & MIGRATE_PFN_MIGRATE))
> -			continue;
> -
> -		dpage = nouveau_dmem_page_alloc_locked(drm);
> -		if (!dpage)
> -			continue;
> -
> -		dst_pfns[i] = migrate_pfn(page_to_pfn(dpage)) |
> -			      MIGRATE_PFN_LOCKED |
> -			      MIGRATE_PFN_DEVICE;
> -		npages++;
> -	}
> -
> -	if (!npages)
> -		return;
> -
> -	/* Allocate storage for DMA addresses, so we can unmap later. */
> -	migrate->dma = kmalloc(sizeof(*migrate->dma) * npages, GFP_KERNEL);
> -	if (!migrate->dma)
> -		goto error;
> -	migrate->dma_nr = 0;
> -
> -	/* Copy things over */
> -	copy = drm->dmem->migrate.copy_func;
> -	for (addr = start, i = 0; addr < end; addr += PAGE_SIZE, i++) {
> -		struct page *spage, *dpage;
> -
> -		dpage = migrate_pfn_to_page(dst_pfns[i]);
> -		if (!dpage || dst_pfns[i] == MIGRATE_PFN_ERROR)
> -			continue;
> -
> -		spage = migrate_pfn_to_page(src_pfns[i]);
> -		if (!spage || !(src_pfns[i] & MIGRATE_PFN_MIGRATE)) {
> -			nouveau_dmem_page_free_locked(drm, dpage);
> -			dst_pfns[i] = 0;
> -			continue;
> -		}
> -
> -		migrate->dma[migrate->dma_nr] =
> -			dma_map_page_attrs(dev, spage, 0, PAGE_SIZE,
> -					   PCI_DMA_BIDIRECTIONAL,
> -					   DMA_ATTR_SKIP_CPU_SYNC);
> -		if (dma_mapping_error(dev, migrate->dma[migrate->dma_nr])) {
> -			nouveau_dmem_page_free_locked(drm, dpage);
> -			dst_pfns[i] = 0;
> -			continue;
> -		}
> -
> -		ret = copy(drm, 1, NOUVEAU_APER_VRAM,
> -				nouveau_dmem_page_addr(dpage),
> -				NOUVEAU_APER_HOST,
> -				migrate->dma[migrate->dma_nr++]);
> -		if (ret) {
> -			nouveau_dmem_page_free_locked(drm, dpage);
> -			dst_pfns[i] = 0;
> -			continue;
> -		}
> -	}
> +	struct page *dpage, *spage;
>   
> -	nouveau_fence_new(drm->dmem->migrate.chan, false, &migrate->fence);
> +	spage = migrate_pfn_to_page(src);
> +	if (!spage || !(src & MIGRATE_PFN_MIGRATE))
> +		goto out;
>   
> -	return;
> +	dpage = nouveau_dmem_page_alloc_locked(drm);
> +	if (!dpage)
> +		return 0;
>   
> -error:
> -	for (addr = start, i = 0; addr < end; addr += PAGE_SIZE, ++i) {
> -		struct page *page;
> +	*dma_addr = dma_map_page(dev, spage, 0, PAGE_SIZE, DMA_BIDIRECTIONAL);
> +	if (dma_mapping_error(dev, *dma_addr))
> +		goto out_free_page;
>   
> -		if (!dst_pfns[i] || dst_pfns[i] == MIGRATE_PFN_ERROR)
> -			continue;
> +	if (drm->dmem->migrate.copy_func(drm, 1, NOUVEAU_APER_VRAM,
> +			nouveau_dmem_page_addr(dpage), NOUVEAU_APER_HOST,
> +			*dma_addr))
> +		goto out_dma_unmap;
>   
> -		page = migrate_pfn_to_page(dst_pfns[i]);
> -		dst_pfns[i] = MIGRATE_PFN_ERROR;
> -		if (page == NULL)
> -			continue;
> +	return migrate_pfn(page_to_pfn(dpage)) |
> +		MIGRATE_PFN_LOCKED | MIGRATE_PFN_DEVICE;
>   
> -		__free_page(page);
> -	}
> +out_dma_unmap:
> +	dma_unmap_page(dev, *dma_addr, PAGE_SIZE, DMA_BIDIRECTIONAL);
> +out_free_page:
> +	nouveau_dmem_page_free_locked(drm, dpage);
> +out:
> +	return 0;
>   }
>   
> -static void
> -nouveau_dmem_migrate_finalize_and_map(struct nouveau_migrate *migrate)
> +static void nouveau_dmem_migrate_chunk(struct migrate_vma *args,
> +		struct nouveau_drm *drm, dma_addr_t *dma_addrs)
>   {
> -	struct nouveau_drm *drm = migrate->drm;
> +	struct nouveau_fence *fence;
> +	unsigned long addr = args->start, nr_dma = 0, i;
> +
> +	for (i = 0; addr < args->end; i++) {
> +		args->dst[i] = nouveau_dmem_migrate_copy_one(drm, args->vma,
> +				addr, args->src[i], &dma_addrs[nr_dma]);
> +		if (args->dst[i])
> +			nr_dma++;
> +		addr += PAGE_SIZE;
> +	}
>   
> -	nouveau_dmem_fence_done(&migrate->fence);
> +	nouveau_fence_new(drm->dmem->migrate.chan, false, &fence);
> +	migrate_vma_pages(args);
> +	nouveau_dmem_fence_done(&fence);
>   
> -	while (migrate->dma_nr--) {
> -		dma_unmap_page(drm->dev->dev, migrate->dma[migrate->dma_nr],
> -			       PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
> +	while (nr_dma--) {
> +		dma_unmap_page(drm->dev->dev, dma_addrs[nr_dma], PAGE_SIZE,
> +				DMA_BIDIRECTIONAL);
>   	}
> -	kfree(migrate->dma);
> -
>   	/*
> -	 * FIXME optimization: update GPU page table to point to newly
> -	 * migrated memory.
> +	 * FIXME optimization: update GPU page table to point to newly migrated
> +	 * memory.
>   	 */
> -}
> -
> -static void nouveau_dmem_migrate_chunk(struct migrate_vma *args,
> -		struct nouveau_migrate *migrate)
> -{
> -	nouveau_dmem_migrate_alloc_and_copy(args->vma, args->src, args->dst,
> -			args->start, args->end, migrate);
> -	migrate_vma_pages(args);
> -	nouveau_dmem_migrate_finalize_and_map(migrate);
>   	migrate_vma_finalize(args);
>   }
>   
> @@ -705,38 +630,40 @@ nouveau_dmem_migrate_vma(struct nouveau_drm *drm,
>   {
>   	unsigned long npages = (end - start) >> PAGE_SHIFT;
>   	unsigned long max = min(SG_MAX_SINGLE_ALLOC, npages);
> +	dma_addr_t *dma_addrs;
>   	struct migrate_vma args = {
>   		.vma		= vma,
>   		.start		= start,
>   	};
> -	struct nouveau_migrate migrate = {
> -		.drm		= drm,
> -		.vma		= vma,
> -		.npages		= npages,
> -	};
>   	unsigned long c, i;
>   	int ret = -ENOMEM;
>   
> -	args.src = kzalloc(sizeof(long) * max, GFP_KERNEL);
> +	args.src = kcalloc(max, sizeof(args.src), GFP_KERNEL);
>   	if (!args.src)
>   		goto out;
> -	args.dst = kzalloc(sizeof(long) * max, GFP_KERNEL);
> +	args.dst = kcalloc(max, sizeof(args.dst), GFP_KERNEL);
>   	if (!args.dst)
>   		goto out_free_src;
>   
> +	dma_addrs = kmalloc_array(max, sizeof(*dma_addrs), GFP_KERNEL);
> +	if (!dma_addrs)
> +		goto out_free_dst;
> +
>   	for (i = 0; i < npages; i += c) {
>   		c = min(SG_MAX_SINGLE_ALLOC, npages);
>   		args.end = start + (c << PAGE_SHIFT);
>   		ret = migrate_vma_setup(&args);
>   		if (ret)
> -			goto out_free_dst;
> +			goto out_free_dma;
>   
>   		if (args.cpages)
> -			nouveau_dmem_migrate_chunk(&args, &migrate);
> +			nouveau_dmem_migrate_chunk(&args, drm, dma_addrs);
>   		args.start = args.end;
>   	}
>   
>   	ret = 0;
> +out_free_dma:
> +	kfree(dma_addrs);
>   out_free_dst:
>   	kfree(args.dst);
>   out_free_src:
>