From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id F30DC1DB150; Wed, 15 Jan 2025 08:17:30 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=10.30.226.201 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1736929052; cv=none; b=Y/zBce6A2YuCB7Wwb36PZweG4WOEh4vIBW/6KXdRAZ/h+vh29ZzT/5b3gcA+gWLMZxWEto0Cc4NdjBz1SA3ZaUjMRgcajoGfi9cTZTXQPqVWpBcjwevoKKXsJgdksWzSH9heuW+h8VOLyZG5LD1V/uFpbL/BRiUnChJelBtNIzo= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1736929052; c=relaxed/simple; bh=Yu7S/jLyUGhfFZjVFlk4eprtZYRxfla2Nw7nZEhy9no=; h=Date:From:To:Cc:Subject:Message-ID:References:MIME-Version: Content-Type:Content-Disposition:In-Reply-To; b=nHYM0FRi8hR27xAay804JFfYHPqR3c82VSXMvKjfyg+QXHjtLGrSfPvg8YvzraFY5uzrD+nzXRJ6oU7mjTcCMC3SERRuwDd4M0aNXS3AyQiTjotk7MOkQPXV/DVsOZp9i4u8hGXSvOLfaJewEvIVI8kh67Xrz2ckXvM+4NASUl4= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b=jtS4YogD; arc=none smtp.client-ip=10.30.226.201 Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="jtS4YogD" Received: by smtp.kernel.org (Postfix) with ESMTPSA id DCD6DC4CEDF; Wed, 15 Jan 2025 08:17:29 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1736929050; bh=Yu7S/jLyUGhfFZjVFlk4eprtZYRxfla2Nw7nZEhy9no=; h=Date:From:To:Cc:Subject:References:In-Reply-To:From; b=jtS4YogDlGBJKPE4uxCOLjxH7XCi5QYsOo4leoD+xgmN8tj70qVx5829WIg+sShgX yiakEsBadvxORmzL2tNvh64PTbxN2Yn7ZjPr4SYb84VfV3NxFD7kGP/KdoZOcX+tov rQJGthlfV+pWswtBJZw2US2akwQ3E9tQt/jwG/ICtUFlz/2ySVXe68KJv7W8xdw1OU pIcG2l3K1LSF8572QwfNaFSMofL4ExapBuk12HqOU0whbiq9SaXUul7IQT4rV47kiH Y6/VezMg1RWwWJ+XN6OfK0HDAZ0irI3FF3+Tw83vfWul8FBVAnUy0tZrZ3zZK1Ybu0 cyRrXOGYV/nDQ== Date: Wed, 15 Jan 2025 10:17:26 +0200 From: Leon Romanovsky To: Robin Murphy Cc: Jens Axboe , Jason Gunthorpe , Joerg Roedel , Will Deacon , Christoph Hellwig , Sagi Grimberg , Keith Busch , Bjorn Helgaas , Logan Gunthorpe , Yishai Hadas , Shameer Kolothum , Kevin Tian , Alex Williamson , Marek Szyprowski , =?iso-8859-1?B?Suly9G1l?= Glisse , Andrew Morton , Jonathan Corbet , linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org, linux-block@vger.kernel.org, linux-rdma@vger.kernel.org, iommu@lists.linux.dev, linux-nvme@lists.infradead.org, linux-pci@vger.kernel.org, kvm@vger.kernel.org, linux-mm@kvack.org, Randy Dunlap Subject: Re: [PATCH v5 05/17] dma-mapping: Provide an interface to allow allocate IOVA Message-ID: <20250115081726.GK3146852@unreal> References: Precedence: bulk X-Mailing-List: linux-doc@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: On Tue, Jan 14, 2025 at 08:50:28PM +0000, Robin Murphy wrote: > On 17/12/2024 1:00 pm, Leon Romanovsky wrote: > > From: Leon Romanovsky > > > > The existing .map_page() callback provides both allocating of IOVA > > and linking DMA pages. That combination works great for most of the > > callers who use it in control paths, but is less effective in fast > > paths where there may be multiple calls to map_page(). > > > > These advanced callers already manage their data in some sort of > > database and can perform IOVA allocation in advance, leaving range > > linkage operation to be in fast path. > > > > Provide an interface to allocate/deallocate IOVA and next patch > > link/unlink DMA ranges to that specific IOVA. > > > > In the new API a DMA mapping transaction is identified by a > > struct dma_iova_state, which holds some recomputed information > > for the transaction which does not change for each page being > > mapped, so add a check if IOVA can be used for the specific > > transaction. > > > > The API is exported from dma-iommu as it is the only implementation > > supported, the namespace is clearly different from iommu_* functions > > which are not allowed to be used. This code layout allows us to save > > function call per API call used in datapath as well as a lot of boilerplate > > code. > > > > Reviewed-by: Christoph Hellwig > > Signed-off-by: Leon Romanovsky > > --- > > drivers/iommu/dma-iommu.c | 74 +++++++++++++++++++++++++++++++++++++ > > include/linux/dma-mapping.h | 49 ++++++++++++++++++++++++ > > 2 files changed, 123 insertions(+) > > > > diff --git a/drivers/iommu/dma-iommu.c b/drivers/iommu/dma-iommu.c > > index 853247c42f7d..5906b47a300c 100644 > > --- a/drivers/iommu/dma-iommu.c > > +++ b/drivers/iommu/dma-iommu.c > > @@ -1746,6 +1746,80 @@ size_t iommu_dma_max_mapping_size(struct device *dev) > > return SIZE_MAX; > > } > > +/** > > + * dma_iova_try_alloc - Try to allocate an IOVA space > > + * @dev: Device to allocate the IOVA space for > > + * @state: IOVA state > > + * @phys: physical address > > + * @size: IOVA size > > + * > > + * Check if @dev supports the IOVA-based DMA API, and if yes allocate IOVA space > > + * for the given base address and size. > > + * > > + * Note: @phys is only used to calculate the IOVA alignment. Callers that always > > + * do PAGE_SIZE aligned transfers can safely pass 0 here. > > + * > > + * Returns %true if the IOVA-based DMA API can be used and IOVA space has been > > + * allocated, or %false if the regular DMA API should be used. > > + */ > > +bool dma_iova_try_alloc(struct device *dev, struct dma_iova_state *state, > > + phys_addr_t phys, size_t size) > > +{ > > + struct iommu_domain *domain = iommu_get_dma_domain(dev); > > + struct iommu_dma_cookie *cookie = domain->iova_cookie; > > + struct iova_domain *iovad = &cookie->iovad; > > + size_t iova_off = iova_offset(iovad, phys); > > + dma_addr_t addr; > > + > > + memset(state, 0, sizeof(*state)); > > + if (!use_dma_iommu(dev)) > > + return false; > > Can you guess why that return won't ever be taken? I will move references to pointers after this check, but like Christoph said, this "if ..." is taken a lot and we didn't see any issues with inbox GCC versions. > > > + if (static_branch_unlikely(&iommu_deferred_attach_enabled) && > > + iommu_deferred_attach(dev, iommu_get_domain_for_dev(dev))) > > + return false; > > + > > + if (WARN_ON_ONCE(!size)) > > + return false; > > + if (WARN_ON_ONCE(size & DMA_IOVA_USE_SWIOTLB)) > > This looks weird. Why would a caller ever set an effectively-private flag in > the first place? If it's actually supposed to be a maximum size check, > please make it look like a maximum size check. It is in-kernel API and the idea is to warn developers of something that is not right and not perform defensive programming by doing size checks. <...> > > +/** > > + * dma_iova_free - Free an IOVA space > > + * @dev: Device to free the IOVA space for > > + * @state: IOVA state > > + * > > + * Undoes a successful dma_try_iova_alloc(). > > + * > > + * Note that all dma_iova_link() calls need to be undone first. For callers > > + * that never call dma_iova_unlink(), dma_iova_destroy() can be used instead > > + * which unlinks all ranges and frees the IOVA space in a single efficient > > + * operation. > > That's only true if they *also* call dma_iova_link() in just the right way > too. We can update the comment section to any wording, feel free to propose. > > > + */ > > +void dma_iova_free(struct device *dev, struct dma_iova_state *state) > > +{ > > + struct iommu_domain *domain = iommu_get_dma_domain(dev); > > + struct iommu_dma_cookie *cookie = domain->iova_cookie; > > + struct iova_domain *iovad = &cookie->iovad; > > + size_t iova_start_pad = iova_offset(iovad, state->addr); > > + size_t size = dma_iova_size(state); > > + > > + iommu_dma_free_iova(cookie, state->addr - iova_start_pad, > > + iova_align(iovad, size + iova_start_pad), NULL); > > +} > > +EXPORT_SYMBOL_GPL(dma_iova_free); > > + > > void iommu_setup_dma_ops(struct device *dev) > > { > > struct iommu_domain *domain = iommu_get_domain_for_dev(dev); > > diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h > > index b79925b1c433..55899d65668b 100644 > > --- a/include/linux/dma-mapping.h > > +++ b/include/linux/dma-mapping.h > > @@ -7,6 +7,8 @@ > > #include > > #include > > #include > > +#include > > +#include > > Why are these being pulled in here? It is rebase leftover. > > > /** > > * List of possible attributes associated with a DMA mapping. The semantics > > @@ -72,6 +74,21 @@ > > #define DMA_BIT_MASK(n) (((n) == 64) ? ~0ULL : ((1ULL<<(n))-1)) > > +struct dma_iova_state { > > + dma_addr_t addr; > > + size_t __size; > > +}; > > + > > +/* > > + * Use the high bit to mark if we used swiotlb for one or more ranges. > > + */ > > +#define DMA_IOVA_USE_SWIOTLB (1ULL << 63) > > This will give surprising results for 32-bit size_t (in fact I guess it > might fire some build warnings already). We got none, I will change to u64. > > Thanks, > Robin. Thanks