From: Robin Murphy <robin.murphy@arm.com>
To: Christoph Hellwig <hch@lst.de>
Cc: Joerg Roedel <joro@8bytes.org>,
Catalin Marinas <catalin.marinas@arm.com>,
Will Deacon <will.deacon@arm.com>,
Tom Lendacky <thomas.lendacky@amd.com>,
iommu@lists.linux-foundation.org,
linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH 12/21] dma-iommu: factor atomic pool allocations into helpers
Date: Tue, 9 Apr 2019 18:59:32 +0100 [thread overview]
Message-ID: <b3f80a11-1504-e8f9-4438-92bcd5f3df7f@arm.com> (raw)
In-Reply-To: <20190327080448.5500-13-hch@lst.de>
On 27/03/2019 08:04, Christoph Hellwig wrote:
> This keeps the code together and will simplify compiling the code
> out on architectures that are always dma coherent.
And this is where things take a turn in the direction I just can't get
on with - I'm looking at the final result and the twisty maze of little
disjoint helpers all overlapping each other in functionality is really
difficult to follow. And I would *much* rather have things rely on
compile-time constant optimisation than spend the future having to fix
the #ifdefed parts for arm64 whenever x86-centric changes fail to test them.
Conceptually, everything except the iommu_dma_alloc_remap() case is more
or less just dma_direct_alloc() with an IOMMU mapping on top - if we
could pass that an internal flag to say "don't fail or bounce because of
masks" it seems like that approach could end up being quite a bit
simpler (I did once have lofty plans to refactor the old arm64 code in
such a way...)
Robin.
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---
> drivers/iommu/dma-iommu.c | 51 +++++++++++++++++++++++++++++----------
> 1 file changed, 38 insertions(+), 13 deletions(-)
>
> diff --git a/drivers/iommu/dma-iommu.c b/drivers/iommu/dma-iommu.c
> index 2013c650718a..8ec69176673d 100644
> --- a/drivers/iommu/dma-iommu.c
> +++ b/drivers/iommu/dma-iommu.c
> @@ -673,6 +673,35 @@ static int iommu_dma_get_sgtable_remap(struct sg_table *sgt, void *cpu_addr,
> GFP_KERNEL);
> }
>
> +static void iommu_dma_free_pool(struct device *dev, size_t size,
> + void *vaddr, dma_addr_t dma_handle)
> +{
> + __iommu_dma_unmap(iommu_get_domain_for_dev(dev), dma_handle, size);
> + dma_free_from_pool(vaddr, PAGE_ALIGN(size));
> +}
> +
> +static void *iommu_dma_alloc_pool(struct device *dev, size_t size,
> + dma_addr_t *dma_handle, gfp_t gfp, unsigned long attrs)
> +{
> + bool coherent = dev_is_dma_coherent(dev);
> + struct page *page;
> + void *vaddr;
> +
> + vaddr = dma_alloc_from_pool(PAGE_ALIGN(size), &page, gfp);
> + if (!vaddr)
> + return NULL;
> +
> + *dma_handle = __iommu_dma_map(dev, page_to_phys(page), size,
> + dma_info_to_prot(DMA_BIDIRECTIONAL, coherent, attrs),
> + iommu_get_domain_for_dev(dev));
> + if (*dma_handle == DMA_MAPPING_ERROR) {
> + dma_free_from_pool(vaddr, PAGE_ALIGN(size));
> + return NULL;
> + }
> +
> + return vaddr;
> +}
> +
> static void iommu_dma_sync_single_for_cpu(struct device *dev,
> dma_addr_t dma_handle, size_t size, enum dma_data_direction dir)
> {
> @@ -981,21 +1010,18 @@ static void *iommu_dma_alloc(struct device *dev, size_t size,
> * get the virtually contiguous buffer we need by way of a
> * physically contiguous allocation.
> */
> - if (coherent) {
> - page = alloc_pages(gfp, get_order(size));
> - addr = page ? page_address(page) : NULL;
> - } else {
> - addr = dma_alloc_from_pool(size, &page, gfp);
> - }
> - if (!addr)
> + if (!coherent)
> + return iommu_dma_alloc_pool(dev, iosize, handle, gfp,
> + attrs);
> +
> + page = alloc_pages(gfp, get_order(size));
> + if (!page)
> return NULL;
>
> + addr = page_address(page);
> *handle = __iommu_dma_map_page(dev, page, 0, iosize, ioprot);
> if (*handle == DMA_MAPPING_ERROR) {
> - if (coherent)
> - __free_pages(page, get_order(size));
> - else
> - dma_free_from_pool(addr, size);
> + __free_pages(page, get_order(size));
> addr = NULL;
> }
> } else if (attrs & DMA_ATTR_FORCE_CONTIGUOUS) {
> @@ -1049,8 +1075,7 @@ static void iommu_dma_free(struct device *dev, size_t size, void *cpu_addr,
> * Hence how dodgy the below logic looks...
> */
> if (dma_in_atomic_pool(cpu_addr, size)) {
> - __iommu_dma_unmap_page(dev, handle, iosize, 0, 0);
> - dma_free_from_pool(cpu_addr, size);
> + iommu_dma_free_pool(dev, size, cpu_addr, handle);
> } else if (attrs & DMA_ATTR_FORCE_CONTIGUOUS) {
> struct page *page = vmalloc_to_page(cpu_addr);
>
>
next prev parent reply other threads:[~2019-04-09 17:59 UTC|newest]
Thread overview: 61+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-03-27 8:04 implement generic dma_map_ops for IOMMUs v2 [rebase + resend] Christoph Hellwig
2019-03-27 8:04 ` [PATCH 01/21] arm64/iommu: handle non-remapped addresses in ->mmap and ->get_sgtable Christoph Hellwig
2019-04-05 17:16 ` Robin Murphy
2019-03-27 8:04 ` [PATCH 02/21] arm64/iommu: improve mmap bounds checking Christoph Hellwig
2019-04-05 17:30 ` Robin Murphy
2019-04-07 6:59 ` Christoph Hellwig
2019-04-09 15:12 ` Robin Murphy
2019-04-09 17:09 ` Christoph Hellwig
2019-03-27 8:04 ` [PATCH 03/21] dma-mapping: add a Kconfig symbol to indicated arch_dma_prep_coherent presence Christoph Hellwig
2019-04-05 17:41 ` Robin Murphy
2019-03-27 8:04 ` [PATCH 04/21] dma-iommu: cleanup dma-iommu.h Christoph Hellwig
2019-04-05 17:42 ` Robin Murphy
2019-04-09 17:10 ` Christoph Hellwig
2019-03-27 8:04 ` [PATCH 05/21] dma-iommu: remove the flush_page callback Christoph Hellwig
2019-04-05 17:46 ` Robin Murphy
2019-03-27 8:04 ` [PATCH 06/21] dma-iommu: use for_each_sg in iommu_dma_alloc Christoph Hellwig
2019-04-05 18:08 ` Robin Murphy
2019-03-27 8:04 ` [PATCH 07/21] dma-iommu: move the arm64 wrappers to common code Christoph Hellwig
2019-04-09 15:07 ` Robin Murphy
2019-04-09 17:15 ` Christoph Hellwig
2019-04-09 17:23 ` Christoph Hellwig
2019-04-09 17:33 ` Robin Murphy
2019-03-27 8:04 ` [PATCH 08/21] dma-iommu: refactor iommu_dma_mmap Christoph Hellwig
2019-04-09 15:29 ` Robin Murphy
2019-04-09 17:25 ` Christoph Hellwig
2019-03-27 8:04 ` [PATCH 09/21] dma-iommu: refactor iommu_dma_get_sgtable Christoph Hellwig
2019-04-09 15:49 ` Robin Murphy
2019-04-09 17:26 ` Christoph Hellwig
2019-03-27 8:04 ` [PATCH 10/21] dma-iommu: move __iommu_dma_map Christoph Hellwig
2019-04-09 15:54 ` Robin Murphy
2019-03-27 8:04 ` [PATCH 11/21] dma-iommu: refactor page array remap helpers Christoph Hellwig
2019-04-09 16:38 ` Robin Murphy
2019-03-27 8:04 ` [PATCH 12/21] dma-iommu: factor atomic pool allocations into helpers Christoph Hellwig
2019-04-09 17:59 ` Robin Murphy [this message]
2019-04-10 6:11 ` Christoph Hellwig
2019-04-17 6:33 ` Christoph Hellwig
2019-04-17 11:54 ` Robin Murphy
2019-04-18 15:06 ` Robin Murphy
2019-04-18 16:35 ` Christoph Hellwig
2019-04-18 16:41 ` Robin Murphy
2019-04-19 9:07 ` Christoph Hellwig
2019-04-23 9:48 ` Robin Murphy
2019-04-18 18:15 ` Robin Murphy
2019-04-19 8:23 ` Christoph Hellwig
2019-04-23 10:01 ` Robin Murphy
2019-04-23 14:52 ` Christoph Hellwig
2019-04-29 11:49 ` Christoph Hellwig
2019-04-29 12:02 ` Robin Murphy
2019-04-22 18:03 ` Christoph Hellwig
2019-03-27 8:04 ` [PATCH 13/21] dma-iommu: factor contiguous " Christoph Hellwig
2019-03-27 8:04 ` [PATCH 14/21] dma-iommu: refactor iommu_dma_free Christoph Hellwig
2019-03-27 8:04 ` [PATCH 15/21] dma-iommu: don't remap contiguous allocations for coherent devices Christoph Hellwig
2019-03-27 8:04 ` [PATCH 16/21] dma-iommu: factor contiguous remapped allocations into helpers Christoph Hellwig
2019-03-27 8:04 ` [PATCH 17/21] dma-iommu: refactor iommu_dma_alloc Christoph Hellwig
2019-03-27 8:04 ` [PATCH 18/21] dma-iommu: don't depend on CONFIG_DMA_DIRECT_REMAP Christoph Hellwig
2019-03-27 8:04 ` [PATCH 19/21] dma-iommu: switch copyright boilerplace to SPDX Christoph Hellwig
2019-03-27 8:04 ` [PATCH 20/21] arm64: switch copyright boilerplace to SPDX in dma-mapping.c Christoph Hellwig
2019-04-01 6:28 ` Mukesh Ojha
2019-04-01 9:39 ` Robin Murphy
2019-03-27 8:04 ` [PATCH 21/21] arm64: trim includes " Christoph Hellwig
-- strict thread matches above, loose matches on Subject: below --
2019-02-13 18:28 implement generic dma_map_ops for IOMMUs v2 Christoph Hellwig
2019-02-13 18:29 ` [PATCH 12/21] dma-iommu: factor atomic pool allocations into helpers Christoph Hellwig
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=b3f80a11-1504-e8f9-4438-92bcd5f3df7f@arm.com \
--to=robin.murphy@arm.com \
--cc=catalin.marinas@arm.com \
--cc=hch@lst.de \
--cc=iommu@lists.linux-foundation.org \
--cc=joro@8bytes.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=thomas.lendacky@amd.com \
--cc=will.deacon@arm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).