AcoSail/kernel_optimize_test
8
0
Fork 0
forked from luck/tmp_suning_uos_patched
Code Pull Requests Activity

iommu/amd: Make use of the generic IOVA allocator

Browse Source 
Remove the old address allocation code and make use of the
generic IOVA allocator that is also used by other dma-ops
implementations.

Signed-off-by: Joerg Roedel <jroedel@suse.de>
This commit is contained in:
Joerg Roedel 2016-07-05 14:23:01 +02:00
parent 518d9b4503
commit 256e4621c2
1 changed files with 26 additions and 157 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

183
drivers/iommu/amd_iommu.c
View File

@ -1649,167 +1649,32 @@ static int alloc_new_range(struct dma_ops_domain *dma_dom,
return -ENOMEM; return -ENOMEM;
} }
static dma_addr_t dma_ops_aperture_alloc(struct dma_ops_domain *dom, static unsigned long dma_ops_alloc_iova(struct device *dev,
struct aperture_range *range, struct dma_ops_domain *dma_dom,
unsigned long pages, unsigned int pages, u64 dma_mask)
unsigned long dma_mask,
unsigned long boundary_size,
unsigned long align_mask,
bool trylock)
{ {
unsigned long offset, limit, flags; unsigned long pfn = 0;
dma_addr_t address;
bool flush = false;
offset = range->offset >> PAGE_SHIFT; pages = __roundup_pow_of_two(pages);
limit = iommu_device_max_index(APERTURE_RANGE_PAGES, offset,
dma_mask >> PAGE_SHIFT);
if (trylock) { if (dma_mask > DMA_BIT_MASK(32))
if (!spin_trylock_irqsave(&range->bitmap_lock, flags)) pfn = alloc_iova_fast(&dma_dom->iovad, pages,
return -1; IOVA_PFN(DMA_BIT_MASK(32)));
} else {
spin_lock_irqsave(&range->bitmap_lock, flags);
}
address = iommu_area_alloc(range->bitmap, limit, range->next_bit, if (!pfn)
pages, offset, boundary_size, align_mask); pfn = alloc_iova_fast(&dma_dom->iovad, pages, IOVA_PFN(dma_mask));
if (address == -1) {
/* Nothing found, retry one time */
address = iommu_area_alloc(range->bitmap, limit,
0, pages, offset, boundary_size,
align_mask);
flush = true;
}
if (address != -1) return (pfn << PAGE_SHIFT);
range->next_bit = address + pages;
spin_unlock_irqrestore(&range->bitmap_lock, flags);
if (flush) {
domain_flush_tlb(&dom->domain);
domain_flush_complete(&dom->domain);
}
return address;
} }
static unsigned long dma_ops_area_alloc(struct device *dev, static void dma_ops_free_iova(struct dma_ops_domain *dma_dom,
struct dma_ops_domain *dom, unsigned long address,
unsigned int pages, unsigned int pages)
unsigned long align_mask,
u64 dma_mask)
{ {
unsigned long boundary_size, mask; pages = __roundup_pow_of_two(pages);
unsigned long address = -1; address >>= PAGE_SHIFT;
bool first = true;
u32 start, i;
preempt_disable();
mask = dma_get_seg_boundary(dev);
again:
start = this_cpu_read(*dom->next_index);
/* Sanity check - is it really necessary? */
if (unlikely(start > APERTURE_MAX_RANGES)) {
start = 0;
this_cpu_write(*dom->next_index, 0);
}
boundary_size = mask + 1 ? ALIGN(mask + 1, PAGE_SIZE) >> PAGE_SHIFT :
1UL << (BITS_PER_LONG - PAGE_SHIFT);
for (i = 0; i < APERTURE_MAX_RANGES; ++i) {
struct aperture_range *range;
int index;
index = (start + i) % APERTURE_MAX_RANGES;
range = dom->aperture[index];
if (!range || range->offset >= dma_mask)
continue;
address = dma_ops_aperture_alloc(dom, range, pages,
dma_mask, boundary_size,
align_mask, first);
if (address != -1) {
address = range->offset + (address << PAGE_SHIFT);
this_cpu_write(*dom->next_index, index);
break;
}
}
if (address == -1 && first) {
first = false;
goto again;
}
preempt_enable();
return address;
}
static unsigned long dma_ops_alloc_addresses(struct device *dev,
struct dma_ops_domain *dom,
unsigned int pages,
unsigned long align_mask,
u64 dma_mask)
{
unsigned long address = -1;
while (address == -1) {
address = dma_ops_area_alloc(dev, dom, pages,
align_mask, dma_mask);
if (address == -1 && alloc_new_range(dom, false, GFP_ATOMIC))
break;
}
if (unlikely(address == -1))
address = DMA_ERROR_CODE;
WARN_ON((address + (PAGE_SIZE*pages)) > dom->aperture_size);
return address;
}
/*
* The address free function.
*
* called with domain->lock held
*/
static void dma_ops_free_addresses(struct dma_ops_domain *dom,
unsigned long address,
unsigned int pages)
{
unsigned i = address >> APERTURE_RANGE_SHIFT;
struct aperture_range *range = dom->aperture[i];
unsigned long flags;
BUG_ON(i >= APERTURE_MAX_RANGES || range == NULL);
#ifdef CONFIG_IOMMU_STRESS
if (i < 4)
return;
#endif
if (amd_iommu_unmap_flush) {
domain_flush_tlb(&dom->domain);
domain_flush_complete(&dom->domain);
}
address = (address % APERTURE_RANGE_SIZE) >> PAGE_SHIFT;
spin_lock_irqsave(&range->bitmap_lock, flags);
if (address + pages > range->next_bit)
range->next_bit = address + pages;
bitmap_clear(range->bitmap, address, pages);
spin_unlock_irqrestore(&range->bitmap_lock, flags);
free_iova_fast(&dma_dom->iovad, address, pages);
} }
/**************************************************************************** /****************************************************************************
@ -2586,9 +2451,7 @@ static dma_addr_t __map_single(struct device *dev,
if (align) if (align)
align_mask = (1UL << get_order(size)) - 1; align_mask = (1UL << get_order(size)) - 1;
address = dma_ops_alloc_addresses(dev, dma_dom, pages, align_mask, address = dma_ops_alloc_iova(dev, dma_dom, pages, dma_mask);
dma_mask);
if (address == DMA_ERROR_CODE) if (address == DMA_ERROR_CODE)
goto out; goto out;
@ -2626,7 +2489,10 @@ static dma_addr_t __map_single(struct device *dev,
iommu_unmap_page(&dma_dom->domain, start, PAGE_SIZE); iommu_unmap_page(&dma_dom->domain, start, PAGE_SIZE);
} }
dma_ops_free_addresses(dma_dom, address, pages); domain_flush_tlb(&dma_dom->domain);
domain_flush_complete(&dma_dom->domain);
dma_ops_free_iova(dma_dom, address, pages);
return DMA_ERROR_CODE; return DMA_ERROR_CODE;
} }
@ -2658,7 +2524,10 @@ static void __unmap_single(struct dma_ops_domain *dma_dom,
start += PAGE_SIZE; start += PAGE_SIZE;
} }
dma_ops_free_addresses(dma_dom, dma_addr, pages); domain_flush_tlb(&dma_dom->domain);
domain_flush_complete(&dma_dom->domain);
dma_ops_free_iova(dma_dom, dma_addr, pages);
} }
/* /*