From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
To: <linuxppc-dev@ozlabs.org>
Subject: [PATCH 28/32] powerpc: Make cell use direct DMA ops
Date: Fri, 10 Nov 2006 18:45:08 +1100 [thread overview]
Message-ID: <20061110074511.B55D367F6A@ozlabs.org> (raw)
In-Reply-To: <1163144683.554182.685908332811.qpush@grosgo>
Now that the direct DMA ops supports an offset, we use that instead
of defining our own.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
arch/powerpc/platforms/cell/iommu.c | 79 ++----------------------------------
1 file changed, 6 insertions(+), 73 deletions(-)
Index: linux-cell/arch/powerpc/platforms/cell/iommu.c
===================================================================
--- linux-cell.orig/arch/powerpc/platforms/cell/iommu.c 2006-11-07 16:45:08.000000000 +1100
+++ linux-cell/arch/powerpc/platforms/cell/iommu.c 2006-11-07 16:48:00.000000000 +1100
@@ -46,8 +46,6 @@
#include "iommu.h"
-static dma_addr_t cell_dma_valid = SPIDER_DMA_VALID;
-
static inline unsigned long
get_iopt_entry(unsigned long real_address, unsigned long ioid,
unsigned long prot)
@@ -417,83 +415,18 @@ static int cell_map_iommu(void)
return 1;
}
-static void *cell_alloc_coherent(struct device *hwdev, size_t size,
- dma_addr_t *dma_handle, gfp_t flag)
-{
- void *ret;
-
- ret = (void *)__get_free_pages(flag, get_order(size));
- if (ret != NULL) {
- memset(ret, 0, size);
- *dma_handle = virt_to_abs(ret) | cell_dma_valid;
- }
- return ret;
-}
-
-static void cell_free_coherent(struct device *hwdev, size_t size,
- void *vaddr, dma_addr_t dma_handle)
-{
- free_pages((unsigned long)vaddr, get_order(size));
-}
-
-static dma_addr_t cell_map_single(struct device *hwdev, void *ptr,
- size_t size, enum dma_data_direction direction)
-{
- return virt_to_abs(ptr) | cell_dma_valid;
-}
-
-static void cell_unmap_single(struct device *hwdev, dma_addr_t dma_addr,
- size_t size, enum dma_data_direction direction)
-{
-}
-
-static int cell_map_sg(struct device *hwdev, struct scatterlist *sg,
- int nents, enum dma_data_direction direction)
-{
- int i;
-
- for (i = 0; i < nents; i++, sg++) {
- sg->dma_address = (page_to_phys(sg->page) + sg->offset)
- | cell_dma_valid;
- sg->dma_length = sg->length;
- }
-
- return nents;
-}
-
-static void cell_unmap_sg(struct device *hwdev, struct scatterlist *sg,
- int nents, enum dma_data_direction direction)
-{
-}
-
-static int cell_dma_supported(struct device *dev, u64 mask)
-{
- return mask < 0x100000000ull;
-}
-
-static struct dma_mapping_ops cell_iommu_ops = {
- .alloc_coherent = cell_alloc_coherent,
- .free_coherent = cell_free_coherent,
- .map_single = cell_map_single,
- .unmap_single = cell_unmap_single,
- .map_sg = cell_map_sg,
- .unmap_sg = cell_unmap_sg,
- .dma_supported = cell_dma_supported,
-};
-
void cell_init_iommu(void)
{
int setup_bus = 0;
- /* If we have an Axon bridge, clear the DMA valid mask. This is fairly
- * hackish but will work well enough until we have proper iommu code.
- */
- if (of_find_node_by_name(NULL, "axon"))
- cell_dma_valid = 0;
-
if (of_find_node_by_path("/mambo")) {
pr_info("Not using iommu on systemsim\n");
} else {
+ /* If we don't have an Axon bridge, we assume we have a
+ * spider which requires a DMA offset
+ */
+ if (of_find_node_by_name(NULL, "axon") == NULL)
+ dma_direct_offset = SPIDER_DMA_VALID;
if (!(of_chosen &&
get_property(of_chosen, "linux,iommu-off", NULL)))
@@ -509,5 +442,5 @@ void cell_init_iommu(void)
}
}
- pci_dma_ops = &cell_iommu_ops;
+ pci_dma_ops = &dma_direct_ops;
}
next prev parent reply other threads:[~2006-11-10 7:45 UTC|newest]
Thread overview: 36+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-11-10 7:44 [PATCH 0/32] My current serie of patches for 2.6.20 for review Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 1/32] ibmveth: Remove ibmveth "liobn" field Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 2/32] Call platform_notify_remove later Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 3/32] Driver core: add notification of bus events Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 4/32] arch provides generic iomap missing accessors Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 5/32] Add arch specific dev_sysdata to struct device Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 6/32] Change ACPI to use dev_sysdata instead of firmware_data Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 7/32] powerpc: Make pci_read_irq_line the default Benjamin Herrenschmidt
2006-11-14 5:47 ` Zang Roy-r61911
2006-11-14 5:48 ` Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 8/32] powerpc: Remove ppc_md.pci_map_irq & ppc_swizzle for ARCH=powerpc Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 9/32] powerpc: Generic DCR infrastructure Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 10/32] powerpc: Make EMAC use generic DCR access methods Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 11/32] powerpc: Support for DCR based MPIC Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 12/32] powerpc: Improve MPIC driver auto-configuration from DT Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 13/32] powerpc: Native cell support for MPIC in southbridge Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 14/32] powerpc: Souped-up of_platform_device support Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 16/32] powerpc: Refactor 64 bits DMA operations Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 15/32] powerpc: Hook of_platform_bus_probe with cell Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 17/32] powerpc: Add DMA ops support for of_plaform_device to Cell Benjamin Herrenschmidt
2006-11-10 7:44 ` [PATCH 18/32] powerpc: Resolve the parent address of a PCI bus range Benjamin Herrenschmidt
2006-11-10 7:45 ` [PATCH 19/32] powerpc: Resolve the BUID fir RTAS PCI config space accesses Benjamin Herrenschmidt
2006-11-10 7:45 ` [PATCH 20/32] powerpc: Add "parent" struct device for PCI host bridges Benjamin Herrenschmidt
2006-11-10 7:45 ` [PATCH 21/32] powerpc: Generic OF platform driver " Benjamin Herrenschmidt
2006-11-10 7:45 ` [PATCH 23/32] powerpc: Allow hooking of PCI MMIO & PIO accessors on 64 bits Benjamin Herrenschmidt
2006-11-10 7:45 ` [PATCH 22/32] powerpc: Cell fixup DMA offset for new southbridge Benjamin Herrenschmidt
2006-11-10 7:45 ` [PATCH 24/32] powerpc: Cell "Spider" MMIO workarounds Benjamin Herrenschmidt
2006-11-10 7:45 ` [PATCH 26/32] powerpc: Add an optional offset to direct DMA on 64 bits Benjamin Herrenschmidt
2006-11-10 7:45 ` [PATCH 25/32] powerpc: spider uses low level BE MMIO accessors Benjamin Herrenschmidt
2006-11-10 7:45 ` [PATCH 27/32] powerpc: Make direct DMA use node local allocations Benjamin Herrenschmidt
2006-11-10 7:45 ` Benjamin Herrenschmidt [this message]
2006-11-10 7:45 ` [PATCH 29/32] powerpc: Cell iommu support Benjamin Herrenschmidt
2006-11-10 7:45 ` [PATCH 30/32] powerpc: remove ioremap64 and fixup_bigphys_addr Benjamin Herrenschmidt
2006-11-10 7:45 ` [PATCH 31/32] powerpc: Merge 32 and 64 bits asm-powerpc/io.h Benjamin Herrenschmidt
2006-11-10 7:45 ` [PATCH 32/32] powerpc: Fix a typo in new style SPE mapping code Benjamin Herrenschmidt
-- strict thread matches above, loose matches on Subject: below --
2006-11-11 6:24 [PATCH 0/32] My current serie of patches for 2.6.20 for review Benjamin Herrenschmidt
2006-11-11 6:25 ` [PATCH 28/32] powerpc: Make cell use direct DMA ops Benjamin Herrenschmidt
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=20061110074511.B55D367F6A@ozlabs.org \
--to=benh@kernel.crashing.org \
--cc=linuxppc-dev@ozlabs.org \
/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).