[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20090128115403R.fujita.tomonori@lab.ntt.co.jp>
Date: Wed, 28 Jan 2009 11:53:45 +0900
From: FUJITA Tomonori <fujita.tomonori@....ntt.co.jp>
To: ink@...assic.park.msu.ru
Cc: akpm@...ux-foundation.org, rth@...ddle.net,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH] alpha: compile fixes
On Tue, 27 Jan 2009 11:38:23 +0300
Ivan Kokshaysky <ink@...assic.park.msu.ru> wrote:
> - jensen build: fix conflicting declarations for pci_alloc_consistent()
> and undefined virt_to_phys();
>
> - SMP: arch/alpha/kernel/smp.c:124: warning: passing argument 2
> of '__cpu_test_and_set' discards qualifiers from pointer target type
> Interestingly, this only happens with gcc-4.2; gcc <= 4.1 and gcc-4.3
> are OK. Fixed with extra assignment.
>
> Signed-off-by: Ivan Kokshaysky <ink@...assic.park.msu.ru>
> ---
> arch/alpha/include/asm/dma-mapping.h | 2 ++
> arch/alpha/kernel/pci-noop.c | 3 ++-
> arch/alpha/kernel/smp.c | 3 ++-
> 3 files changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/arch/alpha/include/asm/dma-mapping.h b/arch/alpha/include/asm/dma-mapping.h
> index a5801ae..04eb568 100644
> --- a/arch/alpha/include/asm/dma-mapping.h
> +++ b/arch/alpha/include/asm/dma-mapping.h
> @@ -29,6 +29,8 @@
>
> #else /* no PCI - no IOMMU. */
>
> +#include <asm/io.h> /* for virt_to_phys() */
> +
> struct scatterlist;
> void *dma_alloc_coherent(struct device *dev, size_t size,
> dma_addr_t *dma_handle, gfp_t gfp);
> diff --git a/arch/alpha/kernel/pci-noop.c b/arch/alpha/kernel/pci-noop.c
> index 8ac0831..c19a376 100644
> --- a/arch/alpha/kernel/pci-noop.c
> +++ b/arch/alpha/kernel/pci-noop.c
> @@ -109,7 +109,8 @@ sys_pciconfig_write(unsigned long bus, unsigned long dfn,
> /* Stubs for the routines in pci_iommu.c: */
>
> void *
> -pci_alloc_consistent(struct pci_dev *pdev, size_t size, dma_addr_t *dma_addrp)
> +__pci_alloc_consistent(struct pci_dev *pdev, size_t size,
> + dma_addr_t *dma_addrp, gfp_t gfp)
> {
> return NULL;
> }
A patch to change Alpha to handle two dma mapping implementations
(pci-noop and the IOMMU) in the standard way (like X86, POWER, IA64,
etc do) is acceptable?
pci-noop has something like struct dma_map_ops nommu_dma_ops (similar
to pci-nommu.c in arch/x86/kernel/pci-nommu.c) and pci_iommu.c has
struct dma_map_ops iommu_dma_ops. Then at startup, the kernel properly
sets dma_ops pointer to nommu_dma_ops or iommu_dma_ops.
Then we can handle the dma mapping operations in the consistent way,
e.g., dma_map_sg can be just `return ops->map_sg(dev, sg, nents,
dir)`.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists