From 70a7d3cc1308a55104fbe505d76f2aca8a4cf53e Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Mon, 22 Dec 2008 10:26:05 -0800 Subject: swiotlb: add hwdev to swiotlb_phys_to_bus() / swiotlb_sg_to_bus() Impact: extend functions with a (yet unused) parameter, update callsites Some architectures need it - in preparation for highmem swiotlb. Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- lib/swiotlb.c | 53 ++++++++++++++++++++++------------------------------- 1 file changed, 22 insertions(+), 31 deletions(-) (limited to 'lib') diff --git a/lib/swiotlb.c b/lib/swiotlb.c index fa2dc4e5f9b..3657da8ebbc 100644 --- a/lib/swiotlb.c +++ b/lib/swiotlb.c @@ -126,7 +126,7 @@ void * __weak swiotlb_alloc(unsigned order, unsigned long nslabs) return (void *)__get_free_pages(GFP_DMA | __GFP_NOWARN, order); } -dma_addr_t __weak swiotlb_phys_to_bus(phys_addr_t paddr) +dma_addr_t __weak swiotlb_phys_to_bus(struct device *hwdev, phys_addr_t paddr) { return paddr; } @@ -136,9 +136,10 @@ phys_addr_t __weak swiotlb_bus_to_phys(dma_addr_t baddr) return baddr; } -static dma_addr_t swiotlb_virt_to_bus(volatile void *address) +static dma_addr_t swiotlb_virt_to_bus(struct device *hwdev, + volatile void *address) { - return swiotlb_phys_to_bus(virt_to_phys(address)); + return swiotlb_phys_to_bus(hwdev, virt_to_phys(address)); } static void *swiotlb_bus_to_virt(dma_addr_t address) @@ -151,35 +152,23 @@ int __weak swiotlb_arch_range_needs_mapping(void *ptr, size_t size) return 0; } -static dma_addr_t swiotlb_sg_to_bus(struct scatterlist *sg) +static dma_addr_t swiotlb_sg_to_bus(struct device *hwdev, struct scatterlist *sg) { - return swiotlb_phys_to_bus(page_to_phys(sg_page(sg)) + sg->offset); + return swiotlb_phys_to_bus(hwdev, page_to_phys(sg_page(sg)) + sg->offset); } static void swiotlb_print_info(unsigned long bytes) { phys_addr_t pstart, pend; - dma_addr_t bstart, bend; pstart = virt_to_phys(io_tlb_start); pend = virt_to_phys(io_tlb_end); - bstart = swiotlb_phys_to_bus(pstart); - bend = swiotlb_phys_to_bus(pend); - printk(KERN_INFO "Placing %luMB software IO TLB between %p - %p\n", bytes >> 20, io_tlb_start, io_tlb_end); - if (pstart != bstart || pend != bend) - printk(KERN_INFO "software IO TLB at phys %#llx - %#llx" - " bus %#llx - %#llx\n", - (unsigned long long)pstart, - (unsigned long long)pend, - (unsigned long long)bstart, - (unsigned long long)bend); - else - printk(KERN_INFO "software IO TLB at phys %#llx - %#llx\n", - (unsigned long long)pstart, - (unsigned long long)pend); + printk(KERN_INFO "software IO TLB at phys %#llx - %#llx\n", + (unsigned long long)pstart, + (unsigned long long)pend); } /* @@ -406,7 +395,7 @@ map_single(struct device *hwdev, struct swiotlb_phys_addr buffer, size_t size, i struct swiotlb_phys_addr slot_buf; mask = dma_get_seg_boundary(hwdev); - start_dma_addr = swiotlb_virt_to_bus(io_tlb_start) & mask; + start_dma_addr = swiotlb_virt_to_bus(hwdev, io_tlb_start) & mask; offset_slots = ALIGN(start_dma_addr, 1 << IO_TLB_SHIFT) >> IO_TLB_SHIFT; @@ -585,7 +574,9 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size, dma_mask = hwdev->coherent_dma_mask; ret = (void *)__get_free_pages(flags, order); - if (ret && !is_buffer_dma_capable(dma_mask, swiotlb_virt_to_bus(ret), size)) { + if (ret && + !is_buffer_dma_capable(dma_mask, swiotlb_virt_to_bus(hwdev, ret), + size)) { /* * The allocated memory isn't reachable by the device. * Fall back on swiotlb_map_single(). @@ -609,7 +600,7 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size, } memset(ret, 0, size); - dev_addr = swiotlb_virt_to_bus(ret); + dev_addr = swiotlb_virt_to_bus(hwdev, ret); /* Confirm address can be DMA'd by device */ if (!is_buffer_dma_capable(dma_mask, dev_addr, size)) { @@ -669,7 +660,7 @@ dma_addr_t swiotlb_map_single_attrs(struct device *hwdev, void *ptr, size_t size, int dir, struct dma_attrs *attrs) { - dma_addr_t dev_addr = swiotlb_virt_to_bus(ptr); + dma_addr_t dev_addr = swiotlb_virt_to_bus(hwdev, ptr); void *map; struct swiotlb_phys_addr buffer; @@ -694,7 +685,7 @@ swiotlb_map_single_attrs(struct device *hwdev, void *ptr, size_t size, map = io_tlb_overflow_buffer; } - dev_addr = swiotlb_virt_to_bus(map); + dev_addr = swiotlb_virt_to_bus(hwdev, map); /* * Ensure that the address returned is DMA'ble @@ -840,7 +831,7 @@ swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl, int nelems, BUG_ON(dir == DMA_NONE); for_each_sg(sgl, sg, nelems, i) { - dev_addr = swiotlb_sg_to_bus(sg); + dev_addr = swiotlb_sg_to_bus(hwdev, sg); if (range_needs_mapping(sg_virt(sg), sg->length) || address_needs_mapping(hwdev, dev_addr, sg->length)) { void *map; @@ -856,7 +847,7 @@ swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl, int nelems, sgl[0].dma_length = 0; return 0; } - sg->dma_address = swiotlb_virt_to_bus(map); + sg->dma_address = swiotlb_virt_to_bus(hwdev, map); } else sg->dma_address = dev_addr; sg->dma_length = sg->length; @@ -886,7 +877,7 @@ swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl, BUG_ON(dir == DMA_NONE); for_each_sg(sgl, sg, nelems, i) { - if (sg->dma_address != swiotlb_sg_to_bus(sg)) + if (sg->dma_address != swiotlb_sg_to_bus(hwdev, sg)) unmap_single(hwdev, swiotlb_bus_to_virt(sg->dma_address), sg->dma_length, dir); else if (dir == DMA_FROM_DEVICE) @@ -919,7 +910,7 @@ swiotlb_sync_sg(struct device *hwdev, struct scatterlist *sgl, BUG_ON(dir == DMA_NONE); for_each_sg(sgl, sg, nelems, i) { - if (sg->dma_address != swiotlb_sg_to_bus(sg)) + if (sg->dma_address != swiotlb_sg_to_bus(hwdev, sg)) sync_single(hwdev, swiotlb_bus_to_virt(sg->dma_address), sg->dma_length, dir, target); else if (dir == DMA_FROM_DEVICE) @@ -944,7 +935,7 @@ swiotlb_sync_sg_for_device(struct device *hwdev, struct scatterlist *sg, int swiotlb_dma_mapping_error(struct device *hwdev, dma_addr_t dma_addr) { - return (dma_addr == swiotlb_virt_to_bus(io_tlb_overflow_buffer)); + return (dma_addr == swiotlb_virt_to_bus(hwdev, io_tlb_overflow_buffer)); } /* @@ -956,7 +947,7 @@ swiotlb_dma_mapping_error(struct device *hwdev, dma_addr_t dma_addr) int swiotlb_dma_supported(struct device *hwdev, u64 mask) { - return swiotlb_virt_to_bus(io_tlb_end - 1) <= mask; + return swiotlb_virt_to_bus(hwdev, io_tlb_end - 1) <= mask; } EXPORT_SYMBOL(swiotlb_map_single); -- cgit v1.2.3-70-g09d2 From bc40ac66988a7721f2a244b6df65f8c13d16479c Mon Sep 17 00:00:00 2001 From: Becky Bruce Date: Mon, 22 Dec 2008 10:26:08 -0800 Subject: swiotlb: store phys address in io_tlb_orig_addr array Impact: refactor code, cleanup When we enable swiotlb for platforms that support HIGHMEM, we can no longer store the virtual address of the original dma buffer, because that buffer might not have a permament mapping. Change the swiotlb code to instead store the physical address of the original buffer. Signed-off-by: Becky Bruce Signed-off-by: Ingo Molnar --- lib/swiotlb.c | 120 +++++++++++++++------------------------------------------- 1 file changed, 30 insertions(+), 90 deletions(-) (limited to 'lib') diff --git a/lib/swiotlb.c b/lib/swiotlb.c index 3657da8ebbc..98a7a4450e0 100644 --- a/lib/swiotlb.c +++ b/lib/swiotlb.c @@ -89,10 +89,7 @@ static unsigned int io_tlb_index; * We need to save away the original address corresponding to a mapped entry * for the sync operations. */ -static struct swiotlb_phys_addr { - struct page *page; - unsigned int offset; -} *io_tlb_orig_addr; +static phys_addr_t *io_tlb_orig_addr; /* * Protect the above data structures in the map and unmap calls @@ -204,7 +201,7 @@ swiotlb_init_with_default_size(size_t default_size) for (i = 0; i < io_tlb_nslabs; i++) io_tlb_list[i] = IO_TLB_SEGSIZE - OFFSET(i, IO_TLB_SEGSIZE); io_tlb_index = 0; - io_tlb_orig_addr = alloc_bootmem(io_tlb_nslabs * sizeof(struct swiotlb_phys_addr)); + io_tlb_orig_addr = alloc_bootmem(io_tlb_nslabs * sizeof(phys_addr_t)); /* * Get the overflow emergency buffer @@ -278,12 +275,14 @@ swiotlb_late_init_with_default_size(size_t default_size) io_tlb_list[i] = IO_TLB_SEGSIZE - OFFSET(i, IO_TLB_SEGSIZE); io_tlb_index = 0; - io_tlb_orig_addr = (struct swiotlb_phys_addr *)__get_free_pages(GFP_KERNEL, - get_order(io_tlb_nslabs * sizeof(struct swiotlb_phys_addr))); + io_tlb_orig_addr = (phys_addr_t *) + __get_free_pages(GFP_KERNEL, + get_order(io_tlb_nslabs * + sizeof(phys_addr_t))); if (!io_tlb_orig_addr) goto cleanup3; - memset(io_tlb_orig_addr, 0, io_tlb_nslabs * sizeof(struct swiotlb_phys_addr)); + memset(io_tlb_orig_addr, 0, io_tlb_nslabs * sizeof(phys_addr_t)); /* * Get the overflow emergency buffer @@ -298,8 +297,8 @@ swiotlb_late_init_with_default_size(size_t default_size) return 0; cleanup4: - free_pages((unsigned long)io_tlb_orig_addr, get_order(io_tlb_nslabs * - sizeof(char *))); + free_pages((unsigned long)io_tlb_orig_addr, + get_order(io_tlb_nslabs * sizeof(phys_addr_t))); io_tlb_orig_addr = NULL; cleanup3: free_pages((unsigned long)io_tlb_list, get_order(io_tlb_nslabs * @@ -330,59 +329,11 @@ static int is_swiotlb_buffer(char *addr) return addr >= io_tlb_start && addr < io_tlb_end; } -static struct swiotlb_phys_addr swiotlb_bus_to_phys_addr(char *dma_addr) -{ - int index = (dma_addr - io_tlb_start) >> IO_TLB_SHIFT; - struct swiotlb_phys_addr buffer = io_tlb_orig_addr[index]; - buffer.offset += (long)dma_addr & ((1 << IO_TLB_SHIFT) - 1); - buffer.page += buffer.offset >> PAGE_SHIFT; - buffer.offset &= PAGE_SIZE - 1; - return buffer; -} - -static void -__sync_single(struct swiotlb_phys_addr buffer, char *dma_addr, size_t size, int dir) -{ - if (PageHighMem(buffer.page)) { - size_t len, bytes; - char *dev, *host, *kmp; - - len = size; - while (len != 0) { - unsigned long flags; - - bytes = len; - if ((bytes + buffer.offset) > PAGE_SIZE) - bytes = PAGE_SIZE - buffer.offset; - local_irq_save(flags); /* protects KM_BOUNCE_READ */ - kmp = kmap_atomic(buffer.page, KM_BOUNCE_READ); - dev = dma_addr + size - len; - host = kmp + buffer.offset; - if (dir == DMA_FROM_DEVICE) - memcpy(host, dev, bytes); - else - memcpy(dev, host, bytes); - kunmap_atomic(kmp, KM_BOUNCE_READ); - local_irq_restore(flags); - len -= bytes; - buffer.page++; - buffer.offset = 0; - } - } else { - void *v = page_address(buffer.page) + buffer.offset; - - if (dir == DMA_TO_DEVICE) - memcpy(dma_addr, v, size); - else - memcpy(v, dma_addr, size); - } -} - /* * Allocates bounce buffer and returns its kernel virtual address. */ static void * -map_single(struct device *hwdev, struct swiotlb_phys_addr buffer, size_t size, int dir) +map_single(struct device *hwdev, phys_addr_t phys, size_t size, int dir) { unsigned long flags; char *dma_addr; @@ -392,7 +343,6 @@ map_single(struct device *hwdev, struct swiotlb_phys_addr buffer, size_t size, i unsigned long mask; unsigned long offset_slots; unsigned long max_slots; - struct swiotlb_phys_addr slot_buf; mask = dma_get_seg_boundary(hwdev); start_dma_addr = swiotlb_virt_to_bus(hwdev, io_tlb_start) & mask; @@ -477,15 +427,10 @@ found: * This is needed when we sync the memory. Then we sync the buffer if * needed. */ - slot_buf = buffer; - for (i = 0; i < nslots; i++) { - slot_buf.page += slot_buf.offset >> PAGE_SHIFT; - slot_buf.offset &= PAGE_SIZE - 1; - io_tlb_orig_addr[index+i] = slot_buf; - slot_buf.offset += 1 << IO_TLB_SHIFT; - } + for (i = 0; i < nslots; i++) + io_tlb_orig_addr[index+i] = phys + (i << IO_TLB_SHIFT); if (dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL) - __sync_single(buffer, dma_addr, size, DMA_TO_DEVICE); + memcpy(dma_addr, phys_to_virt(phys), size); return dma_addr; } @@ -499,17 +444,17 @@ unmap_single(struct device *hwdev, char *dma_addr, size_t size, int dir) unsigned long flags; int i, count, nslots = ALIGN(size, 1 << IO_TLB_SHIFT) >> IO_TLB_SHIFT; int index = (dma_addr - io_tlb_start) >> IO_TLB_SHIFT; - struct swiotlb_phys_addr buffer = swiotlb_bus_to_phys_addr(dma_addr); + phys_addr_t phys = io_tlb_orig_addr[index]; /* * First, sync the memory before unmapping the entry */ - if ((dir == DMA_FROM_DEVICE) || (dir == DMA_BIDIRECTIONAL)) + if (phys && ((dir == DMA_FROM_DEVICE) || (dir == DMA_BIDIRECTIONAL))) /* * bounce... copy the data back into the original buffer * and * delete the bounce buffer. */ - __sync_single(buffer, dma_addr, size, DMA_FROM_DEVICE); + memcpy(phys_to_virt(phys), dma_addr, size); /* * Return the buffer to the free list by setting the corresponding @@ -541,18 +486,21 @@ static void sync_single(struct device *hwdev, char *dma_addr, size_t size, int dir, int target) { - struct swiotlb_phys_addr buffer = swiotlb_bus_to_phys_addr(dma_addr); + int index = (dma_addr - io_tlb_start) >> IO_TLB_SHIFT; + phys_addr_t phys = io_tlb_orig_addr[index]; + + phys += ((unsigned long)dma_addr & ((1 << IO_TLB_SHIFT) - 1)); switch (target) { case SYNC_FOR_CPU: if (likely(dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL)) - __sync_single(buffer, dma_addr, size, DMA_FROM_DEVICE); + memcpy(phys_to_virt(phys), dma_addr, size); else BUG_ON(dir != DMA_TO_DEVICE); break; case SYNC_FOR_DEVICE: if (likely(dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL)) - __sync_single(buffer, dma_addr, size, DMA_TO_DEVICE); + memcpy(dma_addr, phys_to_virt(phys), size); else BUG_ON(dir != DMA_FROM_DEVICE); break; @@ -591,10 +539,7 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size, * swiotlb_map_single(), which will grab memory from * the lowest available address range. */ - struct swiotlb_phys_addr buffer; - buffer.page = virt_to_page(NULL); - buffer.offset = 0; - ret = map_single(hwdev, buffer, size, DMA_FROM_DEVICE); + ret = map_single(hwdev, 0, size, DMA_FROM_DEVICE); if (!ret) return NULL; } @@ -662,7 +607,6 @@ swiotlb_map_single_attrs(struct device *hwdev, void *ptr, size_t size, { dma_addr_t dev_addr = swiotlb_virt_to_bus(hwdev, ptr); void *map; - struct swiotlb_phys_addr buffer; BUG_ON(dir == DMA_NONE); /* @@ -677,9 +621,7 @@ swiotlb_map_single_attrs(struct device *hwdev, void *ptr, size_t size, /* * Oh well, have to allocate and map a bounce buffer. */ - buffer.page = virt_to_page(ptr); - buffer.offset = (unsigned long)ptr & ~PAGE_MASK; - map = map_single(hwdev, buffer, size, dir); + map = map_single(hwdev, virt_to_phys(ptr), size, dir); if (!map) { swiotlb_full(hwdev, size, dir, 1); map = io_tlb_overflow_buffer; @@ -824,20 +766,18 @@ swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl, int nelems, int dir, struct dma_attrs *attrs) { struct scatterlist *sg; - struct swiotlb_phys_addr buffer; - dma_addr_t dev_addr; int i; BUG_ON(dir == DMA_NONE); for_each_sg(sgl, sg, nelems, i) { - dev_addr = swiotlb_sg_to_bus(hwdev, sg); - if (range_needs_mapping(sg_virt(sg), sg->length) || + void *addr = sg_virt(sg); + dma_addr_t dev_addr = swiotlb_virt_to_bus(hwdev, addr); + + if (range_needs_mapping(addr, sg->length) || address_needs_mapping(hwdev, dev_addr, sg->length)) { - void *map; - buffer.page = sg_page(sg); - buffer.offset = sg->offset; - map = map_single(hwdev, buffer, sg->length, dir); + void *map = map_single(hwdev, sg_phys(sg), + sg->length, dir); if (!map) { /* Don't panic here, we expect map_sg users to do proper error handling. */ -- cgit v1.2.3-70-g09d2 From fb05a37929e0cd99016b4f5e5a5ef077fb10a947 Mon Sep 17 00:00:00 2001 From: Becky Bruce Date: Mon, 22 Dec 2008 10:26:09 -0800 Subject: swiotlb: add support for systems with highmem Impact: extend code for highmem - existing users unaffected On highmem systems, the original dma buffer might not have a virtual mapping - we need to kmap it in to perform the bounce. Extract the code that does the actual copy into a function that does the kmap if highmem is enabled, and default to the normal swiotlb memcpy if not. [ ported by Jeremy Fitzhardinge ] Signed-off-by: Becky Bruce Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- lib/swiotlb.c | 68 ++++++++++++++++++++++++++++++++++++++++++++--------------- 1 file changed, 51 insertions(+), 17 deletions(-) (limited to 'lib') diff --git a/lib/swiotlb.c b/lib/swiotlb.c index 98a7a4450e0..785046e4c3a 100644 --- a/lib/swiotlb.c +++ b/lib/swiotlb.c @@ -14,6 +14,7 @@ * 04/07/.. ak Better overflow handling. Assorted fixes. * 05/09/10 linville Add support for syncing ranges, support syncing for * DMA_BIDIRECTIONAL mappings, miscellaneous cleanup. + * 08/12/11 beckyb Add highmem support */ #include @@ -24,6 +25,7 @@ #include #include #include +#include #include #include #include @@ -149,11 +151,6 @@ int __weak swiotlb_arch_range_needs_mapping(void *ptr, size_t size) return 0; } -static dma_addr_t swiotlb_sg_to_bus(struct device *hwdev, struct scatterlist *sg) -{ - return swiotlb_phys_to_bus(hwdev, page_to_phys(sg_page(sg)) + sg->offset); -} - static void swiotlb_print_info(unsigned long bytes) { phys_addr_t pstart, pend; @@ -329,6 +326,47 @@ static int is_swiotlb_buffer(char *addr) return addr >= io_tlb_start && addr < io_tlb_end; } +/* + * Bounce: copy the swiotlb buffer back to the original dma location + */ +static void swiotlb_bounce(phys_addr_t phys, char *dma_addr, size_t size, + enum dma_data_direction dir) +{ + unsigned long pfn = PFN_DOWN(phys); + + if (PageHighMem(pfn_to_page(pfn))) { + /* The buffer does not have a mapping. Map it in and copy */ + unsigned int offset = phys & ~PAGE_MASK; + char *buffer; + unsigned int sz = 0; + unsigned long flags; + + while (size) { + sz = min(PAGE_SIZE - offset, size); + + local_irq_save(flags); + buffer = kmap_atomic(pfn_to_page(pfn), + KM_BOUNCE_READ); + if (dir == DMA_TO_DEVICE) + memcpy(dma_addr, buffer + offset, sz); + else + memcpy(buffer + offset, dma_addr, sz); + kunmap_atomic(buffer, KM_BOUNCE_READ); + local_irq_restore(flags); + + size -= sz; + pfn++; + dma_addr += sz; + offset = 0; + } + } else { + if (dir == DMA_TO_DEVICE) + memcpy(dma_addr, phys_to_virt(phys), size); + else + memcpy(phys_to_virt(phys), dma_addr, size); + } +} + /* * Allocates bounce buffer and returns its kernel virtual address. */ @@ -430,7 +468,7 @@ found: for (i = 0; i < nslots; i++) io_tlb_orig_addr[index+i] = phys + (i << IO_TLB_SHIFT); if (dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL) - memcpy(dma_addr, phys_to_virt(phys), size); + swiotlb_bounce(phys, dma_addr, size, DMA_TO_DEVICE); return dma_addr; } @@ -450,11 +488,7 @@ unmap_single(struct device *hwdev, char *dma_addr, size_t size, int dir) * First, sync the memory before unmapping the entry */ if (phys && ((dir == DMA_FROM_DEVICE) || (dir == DMA_BIDIRECTIONAL))) - /* - * bounce... copy the data back into the original buffer * and - * delete the bounce buffer. - */ - memcpy(phys_to_virt(phys), dma_addr, size); + swiotlb_bounce(phys, dma_addr, size, DMA_FROM_DEVICE); /* * Return the buffer to the free list by setting the corresponding @@ -494,13 +528,13 @@ sync_single(struct device *hwdev, char *dma_addr, size_t size, switch (target) { case SYNC_FOR_CPU: if (likely(dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL)) - memcpy(phys_to_virt(phys), dma_addr, size); + swiotlb_bounce(phys, dma_addr, size, DMA_FROM_DEVICE); else BUG_ON(dir != DMA_TO_DEVICE); break; case SYNC_FOR_DEVICE: if (likely(dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL)) - memcpy(dma_addr, phys_to_virt(phys), size); + swiotlb_bounce(phys, dma_addr, size, DMA_TO_DEVICE); else BUG_ON(dir != DMA_FROM_DEVICE); break; @@ -817,11 +851,11 @@ swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl, BUG_ON(dir == DMA_NONE); for_each_sg(sgl, sg, nelems, i) { - if (sg->dma_address != swiotlb_sg_to_bus(hwdev, sg)) + if (sg->dma_address != swiotlb_virt_to_bus(hwdev, sg_virt(sg))) unmap_single(hwdev, swiotlb_bus_to_virt(sg->dma_address), sg->dma_length, dir); else if (dir == DMA_FROM_DEVICE) - dma_mark_clean(swiotlb_bus_to_virt(sg->dma_address), sg->dma_length); + dma_mark_clean(sg_virt(sg), sg->dma_length); } } EXPORT_SYMBOL(swiotlb_unmap_sg_attrs); @@ -850,11 +884,11 @@ swiotlb_sync_sg(struct device *hwdev, struct scatterlist *sgl, BUG_ON(dir == DMA_NONE); for_each_sg(sgl, sg, nelems, i) { - if (sg->dma_address != swiotlb_sg_to_bus(hwdev, sg)) + if (sg->dma_address != swiotlb_virt_to_bus(hwdev, sg_virt(sg))) sync_single(hwdev, swiotlb_bus_to_virt(sg->dma_address), sg->dma_length, dir, target); else if (dir == DMA_FROM_DEVICE) - dma_mark_clean(swiotlb_bus_to_virt(sg->dma_address), sg->dma_length); + dma_mark_clean(sg_virt(sg), sg->dma_length); } } -- cgit v1.2.3-70-g09d2 From ac86ccc637458dea53077823f98bc3e1b874d478 Mon Sep 17 00:00:00 2001 From: FUJITA Tomonori Date: Sun, 28 Dec 2008 15:02:06 +0900 Subject: swiotlb: remove unnecessary declaration Impact: cleanup Signed-off-by: FUJITA Tomonori Signed-off-by: Ingo Molnar --- lib/swiotlb.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'lib') diff --git a/lib/swiotlb.c b/lib/swiotlb.c index 785046e4c3a..5a7d0a4a9da 100644 --- a/lib/swiotlb.c +++ b/lib/swiotlb.c @@ -22,7 +22,6 @@ #include #include #include -#include #include #include #include @@ -777,8 +776,6 @@ swiotlb_sync_single_range_for_device(struct device *hwdev, dma_addr_t dev_addr, SYNC_FOR_DEVICE); } -void swiotlb_unmap_sg_attrs(struct device *, struct scatterlist *, int, int, - struct dma_attrs *); /* * Map a set of buffers described by scatterlist in streaming mode for DMA. * This is the scatter-gather version of the above swiotlb_map_single -- cgit v1.2.3-70-g09d2 From 874d6a9555516cbac828131d573ce3d16ce47164 Mon Sep 17 00:00:00 2001 From: FUJITA Tomonori Date: Sun, 28 Dec 2008 15:02:07 +0900 Subject: swiotlb: clean up EXPORT_SYMBOL usage Impact: cleanup swiotlb uses EXPORT_SYMBOL in an inconsistent way. Some functions use EXPORT_SYMBOL at the end of functions. Some use it at the end of swiotlb.c. This cleans up swiotlb to use EXPORT_SYMBOL in a consistent way (at the end of functions). Signed-off-by: FUJITA Tomonori Signed-off-by: Ingo Molnar --- lib/swiotlb.c | 28 ++++++++++++++-------------- 1 file changed, 14 insertions(+), 14 deletions(-) (limited to 'lib') diff --git a/lib/swiotlb.c b/lib/swiotlb.c index 5a7d0a4a9da..ef3cfe58a15 100644 --- a/lib/swiotlb.c +++ b/lib/swiotlb.c @@ -593,6 +593,7 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size, *dma_handle = dev_addr; return ret; } +EXPORT_SYMBOL(swiotlb_alloc_coherent); void swiotlb_free_coherent(struct device *hwdev, size_t size, void *vaddr, @@ -605,6 +606,7 @@ swiotlb_free_coherent(struct device *hwdev, size_t size, void *vaddr, /* DMA_TO_DEVICE to avoid memcpy in unmap_single */ unmap_single(hwdev, vaddr, size, DMA_TO_DEVICE); } +EXPORT_SYMBOL(swiotlb_free_coherent); static void swiotlb_full(struct device *dev, size_t size, int dir, int do_panic) @@ -677,6 +679,7 @@ swiotlb_map_single(struct device *hwdev, void *ptr, size_t size, int dir) { return swiotlb_map_single_attrs(hwdev, ptr, size, dir, NULL); } +EXPORT_SYMBOL(swiotlb_map_single); /* * Unmap a single streaming mode DMA translation. The dma_addr and size must @@ -706,6 +709,8 @@ swiotlb_unmap_single(struct device *hwdev, dma_addr_t dev_addr, size_t size, { return swiotlb_unmap_single_attrs(hwdev, dev_addr, size, dir, NULL); } +EXPORT_SYMBOL(swiotlb_unmap_single); + /* * Make physical memory consistent for a single streaming mode DMA translation * after a transfer. @@ -735,6 +740,7 @@ swiotlb_sync_single_for_cpu(struct device *hwdev, dma_addr_t dev_addr, { swiotlb_sync_single(hwdev, dev_addr, size, dir, SYNC_FOR_CPU); } +EXPORT_SYMBOL(swiotlb_sync_single_for_cpu); void swiotlb_sync_single_for_device(struct device *hwdev, dma_addr_t dev_addr, @@ -742,6 +748,7 @@ swiotlb_sync_single_for_device(struct device *hwdev, dma_addr_t dev_addr, { swiotlb_sync_single(hwdev, dev_addr, size, dir, SYNC_FOR_DEVICE); } +EXPORT_SYMBOL(swiotlb_sync_single_for_device); /* * Same as above, but for a sub-range of the mapping. @@ -767,6 +774,7 @@ swiotlb_sync_single_range_for_cpu(struct device *hwdev, dma_addr_t dev_addr, swiotlb_sync_single_range(hwdev, dev_addr, offset, size, dir, SYNC_FOR_CPU); } +EXPORT_SYMBOL_GPL(swiotlb_sync_single_range_for_cpu); void swiotlb_sync_single_range_for_device(struct device *hwdev, dma_addr_t dev_addr, @@ -775,6 +783,7 @@ swiotlb_sync_single_range_for_device(struct device *hwdev, dma_addr_t dev_addr, swiotlb_sync_single_range(hwdev, dev_addr, offset, size, dir, SYNC_FOR_DEVICE); } +EXPORT_SYMBOL_GPL(swiotlb_sync_single_range_for_device); /* * Map a set of buffers described by scatterlist in streaming mode for DMA. @@ -833,6 +842,7 @@ swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems, { return swiotlb_map_sg_attrs(hwdev, sgl, nelems, dir, NULL); } +EXPORT_SYMBOL(swiotlb_map_sg); /* * Unmap a set of streaming mode DMA translations. Again, cpu read rules @@ -863,6 +873,7 @@ swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sgl, int nelems, { return swiotlb_unmap_sg_attrs(hwdev, sgl, nelems, dir, NULL); } +EXPORT_SYMBOL(swiotlb_unmap_sg); /* * Make physical memory consistent for a set of streaming mode DMA translations @@ -895,6 +906,7 @@ swiotlb_sync_sg_for_cpu(struct device *hwdev, struct scatterlist *sg, { swiotlb_sync_sg(hwdev, sg, nelems, dir, SYNC_FOR_CPU); } +EXPORT_SYMBOL(swiotlb_sync_sg_for_cpu); void swiotlb_sync_sg_for_device(struct device *hwdev, struct scatterlist *sg, @@ -902,12 +914,14 @@ swiotlb_sync_sg_for_device(struct device *hwdev, struct scatterlist *sg, { swiotlb_sync_sg(hwdev, sg, nelems, dir, SYNC_FOR_DEVICE); } +EXPORT_SYMBOL(swiotlb_sync_sg_for_device); int swiotlb_dma_mapping_error(struct device *hwdev, dma_addr_t dma_addr) { return (dma_addr == swiotlb_virt_to_bus(hwdev, io_tlb_overflow_buffer)); } +EXPORT_SYMBOL(swiotlb_dma_mapping_error); /* * Return whether the given device DMA address mask can be supported @@ -920,18 +934,4 @@ swiotlb_dma_supported(struct device *hwdev, u64 mask) { return swiotlb_virt_to_bus(hwdev, io_tlb_end - 1) <= mask; } - -EXPORT_SYMBOL(swiotlb_map_single); -EXPORT_SYMBOL(swiotlb_unmap_single); -EXPORT_SYMBOL(swiotlb_map_sg); -EXPORT_SYMBOL(swiotlb_unmap_sg); -EXPORT_SYMBOL(swiotlb_sync_single_for_cpu); -EXPORT_SYMBOL(swiotlb_sync_single_for_device); -EXPORT_SYMBOL_GPL(swiotlb_sync_single_range_for_cpu); -EXPORT_SYMBOL_GPL(swiotlb_sync_single_range_for_device); -EXPORT_SYMBOL(swiotlb_sync_sg_for_cpu); -EXPORT_SYMBOL(swiotlb_sync_sg_for_device); -EXPORT_SYMBOL(swiotlb_dma_mapping_error); -EXPORT_SYMBOL(swiotlb_alloc_coherent); -EXPORT_SYMBOL(swiotlb_free_coherent); EXPORT_SYMBOL(swiotlb_dma_supported); -- cgit v1.2.3-70-g09d2 From ea319518ba3de282c13ae1cf4bf2215c5e03e67e Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 26 Dec 2008 15:08:55 +0100 Subject: locking, percpu counters: introduce separate lock classes Impact: fix lockdep false positives Classify percpu_counter instances similar to regular lock objects -- that is, per instantiation site. The networking code has increased its use of percpu_counters, which leads to false positives if they are treated as a single class. Signed-off-by: Peter Zijlstra Signed-off-by: Ingo Molnar --- include/linux/percpu_counter.h | 14 ++++++++++---- lib/percpu_counter.c | 18 ++++-------------- lib/proportions.c | 6 +++--- mm/backing-dev.c | 2 +- 4 files changed, 18 insertions(+), 22 deletions(-) (limited to 'lib') diff --git a/include/linux/percpu_counter.h b/include/linux/percpu_counter.h index 9007ccdfc11..96bdde36599 100644 --- a/include/linux/percpu_counter.h +++ b/include/linux/percpu_counter.h @@ -30,8 +30,16 @@ struct percpu_counter { #define FBC_BATCH (NR_CPUS*4) #endif -int percpu_counter_init(struct percpu_counter *fbc, s64 amount); -int percpu_counter_init_irq(struct percpu_counter *fbc, s64 amount); +int __percpu_counter_init(struct percpu_counter *fbc, s64 amount, + struct lock_class_key *key); + +#define percpu_counter_init(fbc, value) \ + ({ \ + static struct lock_class_key __key; \ + \ + __percpu_counter_init(fbc, value, &__key); \ + }) + void percpu_counter_destroy(struct percpu_counter *fbc); void percpu_counter_set(struct percpu_counter *fbc, s64 amount); void __percpu_counter_add(struct percpu_counter *fbc, s64 amount, s32 batch); @@ -85,8 +93,6 @@ static inline int percpu_counter_init(struct percpu_counter *fbc, s64 amount) return 0; } -#define percpu_counter_init_irq percpu_counter_init - static inline void percpu_counter_destroy(struct percpu_counter *fbc) { } diff --git a/lib/percpu_counter.c b/lib/percpu_counter.c index a8663890a88..c7fe2e4e8ed 100644 --- a/lib/percpu_counter.c +++ b/lib/percpu_counter.c @@ -71,11 +71,11 @@ s64 __percpu_counter_sum(struct percpu_counter *fbc) } EXPORT_SYMBOL(__percpu_counter_sum); -static struct lock_class_key percpu_counter_irqsafe; - -int percpu_counter_init(struct percpu_counter *fbc, s64 amount) +int __percpu_counter_init(struct percpu_counter *fbc, s64 amount, + struct lock_class_key *key) { spin_lock_init(&fbc->lock); + lockdep_set_class(&fbc->lock, key); fbc->count = amount; fbc->counters = alloc_percpu(s32); if (!fbc->counters) @@ -87,17 +87,7 @@ int percpu_counter_init(struct percpu_counter *fbc, s64 amount) #endif return 0; } -EXPORT_SYMBOL(percpu_counter_init); - -int percpu_counter_init_irq(struct percpu_counter *fbc, s64 amount) -{ - int err; - - err = percpu_counter_init(fbc, amount); - if (!err) - lockdep_set_class(&fbc->lock, &percpu_counter_irqsafe); - return err; -} +EXPORT_SYMBOL(__percpu_counter_init); void percpu_counter_destroy(struct percpu_counter *fbc) { diff --git a/lib/proportions.c b/lib/proportions.c index 4f387a643d7..7367f2b727d 100644 --- a/lib/proportions.c +++ b/lib/proportions.c @@ -83,11 +83,11 @@ int prop_descriptor_init(struct prop_descriptor *pd, int shift) pd->index = 0; pd->pg[0].shift = shift; mutex_init(&pd->mutex); - err = percpu_counter_init_irq(&pd->pg[0].events, 0); + err = percpu_counter_init(&pd->pg[0].events, 0); if (err) goto out; - err = percpu_counter_init_irq(&pd->pg[1].events, 0); + err = percpu_counter_init(&pd->pg[1].events, 0); if (err) percpu_counter_destroy(&pd->pg[0].events); @@ -191,7 +191,7 @@ int prop_local_init_percpu(struct prop_local_percpu *pl) spin_lock_init(&pl->lock); pl->shift = 0; pl->period = 0; - return percpu_counter_init_irq(&pl->events, 0); + return percpu_counter_init(&pl->events, 0); } void prop_local_destroy_percpu(struct prop_local_percpu *pl) diff --git a/mm/backing-dev.c b/mm/backing-dev.c index f2e574dbc30..f3b12585782 100644 --- a/mm/backing-dev.c +++ b/mm/backing-dev.c @@ -220,7 +220,7 @@ int bdi_init(struct backing_dev_info *bdi) bdi->max_prop_frac = PROP_FRAC_BASE; for (i = 0; i < NR_BDI_STAT_ITEMS; i++) { - err = percpu_counter_init_irq(&bdi->bdi_stat[i], 0); + err = percpu_counter_init(&bdi->bdi_stat[i], 0); if (err) goto err; } -- cgit v1.2.3-70-g09d2