linux/arch/parisc/kernel/pci-dma.c

603 lines
16 KiB
C
Raw Normal View History

/*
** PARISC 1.1 Dynamic DMA mapping support.
** This implementation is for PA-RISC platforms that do not support
** I/O TLBs (aka DMA address translation hardware).
** See Documentation/DMA-mapping.txt for interface definitions.
**
** (c) Copyright 1999,2000 Hewlett-Packard Company
** (c) Copyright 2000 Grant Grundler
** (c) Copyright 2000 Philipp Rumpf <prumpf@tux.org>
** (c) Copyright 2000 John Marvin
**
** "leveraged" from 2.3.47: arch/ia64/kernel/pci-dma.c.
** (I assume it's from David Mosberger-Tang but there was no Copyright)
**
** AFAIK, all PA7100LC and PA7300LC platforms can use this code.
**
** - ggg
*/
#include <linux/init.h>
#include <linux/mm.h>
#include <linux/pci.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/slab.h>
#include <linux/string.h>
#include <linux/types.h>
#include <linux/scatterlist.h>
#include <asm/cacheflush.h>
#include <asm/dma.h> /* for DMA_CHUNK_SIZE */
#include <asm/io.h>
#include <asm/page.h> /* get_order */
#include <asm/pgalloc.h>
#include <asm/uaccess.h>
#include <asm/tlbflush.h> /* for purge_tlb_*() macros */
static struct proc_dir_entry * proc_gsc_root __read_mostly = NULL;
static unsigned long pcxl_used_bytes __read_mostly = 0;
static unsigned long pcxl_used_pages __read_mostly = 0;
extern unsigned long pcxl_dma_start; /* Start of pcxl dma mapping area */
static spinlock_t pcxl_res_lock;
static char *pcxl_res_map;
static int pcxl_res_hint;
static int pcxl_res_size;
#ifdef DEBUG_PCXL_RESOURCE
#define DBG_RES(x...) printk(x)
#else
#define DBG_RES(x...)
#endif
/*
** Dump a hex representation of the resource map.
*/
#ifdef DUMP_RESMAP
static
void dump_resmap(void)
{
u_long *res_ptr = (unsigned long *)pcxl_res_map;
u_long i = 0;
printk("res_map: ");
for(; i < (pcxl_res_size / sizeof(unsigned long)); ++i, ++res_ptr)
printk("%08lx ", *res_ptr);
printk("\n");
}
#else
static inline void dump_resmap(void) {;}
#endif
static int pa11_dma_supported( struct device *dev, u64 mask)
{
return 1;
}
static inline int map_pte_uncached(pte_t * pte,
unsigned long vaddr,
unsigned long size, unsigned long *paddr_ptr)
{
unsigned long end;
unsigned long orig_vaddr = vaddr;
vaddr &= ~PMD_MASK;
end = vaddr + size;
if (end > PMD_SIZE)
end = PMD_SIZE;
do {
if (!pte_none(*pte))
printk(KERN_ERR "map_pte_uncached: page already exists\n");
set_pte(pte, __mk_pte(*paddr_ptr, PAGE_KERNEL_UNC));
purge_tlb_start();
pdtlb_kernel(orig_vaddr);
purge_tlb_end();
vaddr += PAGE_SIZE;
orig_vaddr += PAGE_SIZE;
(*paddr_ptr) += PAGE_SIZE;
pte++;
} while (vaddr < end);
return 0;
}
static inline int map_pmd_uncached(pmd_t * pmd, unsigned long vaddr,
unsigned long size, unsigned long *paddr_ptr)
{
unsigned long end;
unsigned long orig_vaddr = vaddr;
vaddr &= ~PGDIR_MASK;
end = vaddr + size;
if (end > PGDIR_SIZE)
end = PGDIR_SIZE;
do {
[PATCH] mm: init_mm without ptlock First step in pushing down the page_table_lock. init_mm.page_table_lock has been used throughout the architectures (usually for ioremap): not to serialize kernel address space allocation (that's usually vmlist_lock), but because pud_alloc,pmd_alloc,pte_alloc_kernel expect caller holds it. Reverse that: don't lock or unlock init_mm.page_table_lock in any of the architectures; instead rely on pud_alloc,pmd_alloc,pte_alloc_kernel to take and drop it when allocating a new one, to check lest a racing task already did. Similarly no page_table_lock in vmalloc's map_vm_area. Some temporary ugliness in __pud_alloc and __pmd_alloc: since they also handle user mms, which are converted only by a later patch, for now they have to lock differently according to whether or not it's init_mm. If sources get muddled, there's a danger that an arch source taking init_mm.page_table_lock will be mixed with common source also taking it (or neither take it). So break the rules and make another change, which should break the build for such a mismatch: remove the redundant mm arg from pte_alloc_kernel (ppc64 scrapped its distinct ioremap_mm in 2.6.13). Exceptions: arm26 used pte_alloc_kernel on user mm, now pte_alloc_map; ia64 used pte_alloc_map on init_mm, now pte_alloc_kernel; parisc had bad args to pmd_alloc and pte_alloc_kernel in unused USE_HPPA_IOREMAP code; ppc64 map_io_page forgot to unlock on failure; ppc mmu_mapin_ram and ppc64 im_free took page_table_lock for no good reason. Signed-off-by: Hugh Dickins <hugh@veritas.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-30 01:16:21 +00:00
pte_t * pte = pte_alloc_kernel(pmd, vaddr);
if (!pte)
return -ENOMEM;
if (map_pte_uncached(pte, orig_vaddr, end - vaddr, paddr_ptr))
return -ENOMEM;
vaddr = (vaddr + PMD_SIZE) & PMD_MASK;
orig_vaddr += PMD_SIZE;
pmd++;
} while (vaddr < end);
return 0;
}
static inline int map_uncached_pages(unsigned long vaddr, unsigned long size,
unsigned long paddr)
{
pgd_t * dir;
unsigned long end = vaddr + size;
dir = pgd_offset_k(vaddr);
do {
pmd_t *pmd;
pmd = pmd_alloc(NULL, dir, vaddr);
if (!pmd)
return -ENOMEM;
if (map_pmd_uncached(pmd, vaddr, end - vaddr, &paddr))
return -ENOMEM;
vaddr = vaddr + PGDIR_SIZE;
dir++;
} while (vaddr && (vaddr < end));
return 0;
}
static inline void unmap_uncached_pte(pmd_t * pmd, unsigned long vaddr,
unsigned long size)
{
pte_t * pte;
unsigned long end;
unsigned long orig_vaddr = vaddr;
if (pmd_none(*pmd))
return;
if (pmd_bad(*pmd)) {
pmd_ERROR(*pmd);
pmd_clear(pmd);
return;
}
pte = pte_offset_map(pmd, vaddr);
vaddr &= ~PMD_MASK;
end = vaddr + size;
if (end > PMD_SIZE)
end = PMD_SIZE;
do {
pte_t page = *pte;
pte_clear(&init_mm, vaddr, pte);
purge_tlb_start();
pdtlb_kernel(orig_vaddr);
purge_tlb_end();
vaddr += PAGE_SIZE;
orig_vaddr += PAGE_SIZE;
pte++;
if (pte_none(page) || pte_present(page))
continue;
printk(KERN_CRIT "Whee.. Swapped out page in kernel page table\n");
} while (vaddr < end);
}
static inline void unmap_uncached_pmd(pgd_t * dir, unsigned long vaddr,
unsigned long size)
{
pmd_t * pmd;
unsigned long end;
unsigned long orig_vaddr = vaddr;
if (pgd_none(*dir))
return;
if (pgd_bad(*dir)) {
pgd_ERROR(*dir);
pgd_clear(dir);
return;
}
pmd = pmd_offset(dir, vaddr);
vaddr &= ~PGDIR_MASK;
end = vaddr + size;
if (end > PGDIR_SIZE)
end = PGDIR_SIZE;
do {
unmap_uncached_pte(pmd, orig_vaddr, end - vaddr);
vaddr = (vaddr + PMD_SIZE) & PMD_MASK;
orig_vaddr += PMD_SIZE;
pmd++;
} while (vaddr < end);
}
static void unmap_uncached_pages(unsigned long vaddr, unsigned long size)
{
pgd_t * dir;
unsigned long end = vaddr + size;
dir = pgd_offset_k(vaddr);
do {
unmap_uncached_pmd(dir, vaddr, end - vaddr);
vaddr = vaddr + PGDIR_SIZE;
dir++;
} while (vaddr && (vaddr < end));
}
#define PCXL_SEARCH_LOOP(idx, mask, size) \
for(; res_ptr < res_end; ++res_ptr) \
{ \
if(0 == ((*res_ptr) & mask)) { \
*res_ptr |= mask; \
idx = (int)((u_long)res_ptr - (u_long)pcxl_res_map); \
pcxl_res_hint = idx + (size >> 3); \
goto resource_found; \
} \
}
#define PCXL_FIND_FREE_MAPPING(idx, mask, size) { \
u##size *res_ptr = (u##size *)&(pcxl_res_map[pcxl_res_hint & ~((size >> 3) - 1)]); \
u##size *res_end = (u##size *)&pcxl_res_map[pcxl_res_size]; \
PCXL_SEARCH_LOOP(idx, mask, size); \
res_ptr = (u##size *)&pcxl_res_map[0]; \
PCXL_SEARCH_LOOP(idx, mask, size); \
}
unsigned long
pcxl_alloc_range(size_t size)
{
int res_idx;
u_long mask, flags;
unsigned int pages_needed = size >> PAGE_SHIFT;
mask = (u_long) -1L;
mask >>= BITS_PER_LONG - pages_needed;
DBG_RES("pcxl_alloc_range() size: %d pages_needed %d pages_mask 0x%08lx\n",
size, pages_needed, mask);
spin_lock_irqsave(&pcxl_res_lock, flags);
if(pages_needed <= 8) {
PCXL_FIND_FREE_MAPPING(res_idx, mask, 8);
} else if(pages_needed <= 16) {
PCXL_FIND_FREE_MAPPING(res_idx, mask, 16);
} else if(pages_needed <= 32) {
PCXL_FIND_FREE_MAPPING(res_idx, mask, 32);
} else {
panic("%s: pcxl_alloc_range() Too many pages to map.\n",
__FILE__);
}
dump_resmap();
panic("%s: pcxl_alloc_range() out of dma mapping resources\n",
__FILE__);
resource_found:
DBG_RES("pcxl_alloc_range() res_idx %d mask 0x%08lx res_hint: %d\n",
res_idx, mask, pcxl_res_hint);
pcxl_used_pages += pages_needed;
pcxl_used_bytes += ((pages_needed >> 3) ? (pages_needed >> 3) : 1);
spin_unlock_irqrestore(&pcxl_res_lock, flags);
dump_resmap();
/*
** return the corresponding vaddr in the pcxl dma map
*/
return (pcxl_dma_start + (res_idx << (PAGE_SHIFT + 3)));
}
#define PCXL_FREE_MAPPINGS(idx, m, size) \
u##size *res_ptr = (u##size *)&(pcxl_res_map[(idx) + (((size >> 3) - 1) & (~((size >> 3) - 1)))]); \
/* BUG_ON((*res_ptr & m) != m); */ \
*res_ptr &= ~m;
/*
** clear bits in the pcxl resource map
*/
static void
pcxl_free_range(unsigned long vaddr, size_t size)
{
u_long mask, flags;
unsigned int res_idx = (vaddr - pcxl_dma_start) >> (PAGE_SHIFT + 3);
unsigned int pages_mapped = size >> PAGE_SHIFT;
mask = (u_long) -1L;
mask >>= BITS_PER_LONG - pages_mapped;
DBG_RES("pcxl_free_range() res_idx: %d size: %d pages_mapped %d mask 0x%08lx\n",
res_idx, size, pages_mapped, mask);
spin_lock_irqsave(&pcxl_res_lock, flags);
if(pages_mapped <= 8) {
PCXL_FREE_MAPPINGS(res_idx, mask, 8);
} else if(pages_mapped <= 16) {
PCXL_FREE_MAPPINGS(res_idx, mask, 16);
} else if(pages_mapped <= 32) {
PCXL_FREE_MAPPINGS(res_idx, mask, 32);
} else {
panic("%s: pcxl_free_range() Too many pages to unmap.\n",
__FILE__);
}
pcxl_used_pages -= (pages_mapped ? pages_mapped : 1);
pcxl_used_bytes -= ((pages_mapped >> 3) ? (pages_mapped >> 3) : 1);
spin_unlock_irqrestore(&pcxl_res_lock, flags);
dump_resmap();
}
static int proc_pcxl_dma_show(struct seq_file *m, void *v)
{
#if 0
u_long i = 0;
unsigned long *res_ptr = (u_long *)pcxl_res_map;
#endif
unsigned long total_pages = pcxl_res_size << 3; /* 8 bits per byte */
seq_printf(m, "\nDMA Mapping Area size : %d bytes (%ld pages)\n",
PCXL_DMA_MAP_SIZE, total_pages);
seq_printf(m, "Resource bitmap : %d bytes\n", pcxl_res_size);
seq_puts(m, " total: free: used: % used:\n");
seq_printf(m, "blocks %8d %8ld %8ld %8ld%%\n", pcxl_res_size,
pcxl_res_size - pcxl_used_bytes, pcxl_used_bytes,
(pcxl_used_bytes * 100) / pcxl_res_size);
seq_printf(m, "pages %8ld %8ld %8ld %8ld%%\n", total_pages,
total_pages - pcxl_used_pages, pcxl_used_pages,
(pcxl_used_pages * 100 / total_pages));
#if 0
seq_puts(m, "\nResource bitmap:");
for(; i < (pcxl_res_size / sizeof(u_long)); ++i, ++res_ptr) {
if ((i & 7) == 0)
seq_puts(m,"\n ");
seq_printf(m, "%s %08lx", buf, *res_ptr);
}
#endif
seq_putc(m, '\n');
return 0;
}
static int proc_pcxl_dma_open(struct inode *inode, struct file *file)
{
return single_open(file, proc_pcxl_dma_show, NULL);
}
static const struct file_operations proc_pcxl_dma_ops = {
.owner = THIS_MODULE,
.open = proc_pcxl_dma_open,
.read = seq_read,
.llseek = seq_lseek,
.release = single_release,
};
static int __init
pcxl_dma_init(void)
{
if (pcxl_dma_start == 0)
return 0;
spin_lock_init(&pcxl_res_lock);
pcxl_res_size = PCXL_DMA_MAP_SIZE >> (PAGE_SHIFT + 3);
pcxl_res_hint = 0;
pcxl_res_map = (char *)__get_free_pages(GFP_KERNEL,
get_order(pcxl_res_size));
memset(pcxl_res_map, 0, pcxl_res_size);
proc_gsc_root = proc_mkdir("gsc", NULL);
if (!proc_gsc_root)
printk(KERN_WARNING
"pcxl_dma_init: Unable to create gsc /proc dir entry\n");
else {
struct proc_dir_entry* ent;
ent = create_proc_entry("pcxl_dma", 0, proc_gsc_root);
if (ent)
ent->proc_fops = &proc_pcxl_dma_ops;
else
printk(KERN_WARNING
"pci-dma.c: Unable to create pcxl_dma /proc entry.\n");
}
return 0;
}
__initcall(pcxl_dma_init);
static void * pa11_dma_alloc_consistent (struct device *dev, size_t size, dma_addr_t *dma_handle, gfp_t flag)
{
unsigned long vaddr;
unsigned long paddr;
int order;
order = get_order(size);
size = 1 << (order + PAGE_SHIFT);
vaddr = pcxl_alloc_range(size);
paddr = __get_free_pages(flag, order);
flush_kernel_dcache_range(paddr, size);
paddr = __pa(paddr);
map_uncached_pages(vaddr, size, paddr);
*dma_handle = (dma_addr_t) paddr;
#if 0
/* This probably isn't needed to support EISA cards.
** ISA cards will certainly only support 24-bit DMA addressing.
** Not clear if we can, want, or need to support ISA.
*/
if (!dev || *dev->coherent_dma_mask < 0xffffffff)
gfp |= GFP_DMA;
#endif
return (void *)vaddr;
}
static void pa11_dma_free_consistent (struct device *dev, size_t size, void *vaddr, dma_addr_t dma_handle)
{
int order;
order = get_order(size);
size = 1 << (order + PAGE_SHIFT);
unmap_uncached_pages((unsigned long)vaddr, size);
pcxl_free_range((unsigned long)vaddr, size);
free_pages((unsigned long)__va(dma_handle), order);
}
static dma_addr_t pa11_dma_map_single(struct device *dev, void *addr, size_t size, enum dma_data_direction direction)
{
if (direction == DMA_NONE) {
printk(KERN_ERR "pa11_dma_map_single(PCI_DMA_NONE) called by %p\n", __builtin_return_address(0));
BUG();
}
flush_kernel_dcache_range((unsigned long) addr, size);
return virt_to_phys(addr);
}
static void pa11_dma_unmap_single(struct device *dev, dma_addr_t dma_handle, size_t size, enum dma_data_direction direction)
{
if (direction == DMA_NONE) {
printk(KERN_ERR "pa11_dma_unmap_single(PCI_DMA_NONE) called by %p\n", __builtin_return_address(0));
BUG();
}
if (direction == DMA_TO_DEVICE)
return;
/*
* For PCI_DMA_FROMDEVICE this flush is not necessary for the
* simple map/unmap case. However, it IS necessary if if
* pci_dma_sync_single_* has been called and the buffer reused.
*/
flush_kernel_dcache_range((unsigned long) phys_to_virt(dma_handle), size);
return;
}
static int pa11_dma_map_sg(struct device *dev, struct scatterlist *sglist, int nents, enum dma_data_direction direction)
{
int i;
if (direction == DMA_NONE)
BUG();
for (i = 0; i < nents; i++, sglist++ ) {
unsigned long vaddr = sg_virt_addr(sglist);
sg_dma_address(sglist) = (dma_addr_t) virt_to_phys(vaddr);
sg_dma_len(sglist) = sglist->length;
flush_kernel_dcache_range(vaddr, sglist->length);
}
return nents;
}
static void pa11_dma_unmap_sg(struct device *dev, struct scatterlist *sglist, int nents, enum dma_data_direction direction)
{
int i;
if (direction == DMA_NONE)
BUG();
if (direction == DMA_TO_DEVICE)
return;
/* once we do combining we'll need to use phys_to_virt(sg_dma_address(sglist)) */
for (i = 0; i < nents; i++, sglist++ )
flush_kernel_dcache_range(sg_virt_addr(sglist), sglist->length);
return;
}
static void pa11_dma_sync_single_for_cpu(struct device *dev, dma_addr_t dma_handle, unsigned long offset, size_t size, enum dma_data_direction direction)
{
if (direction == DMA_NONE)
BUG();
flush_kernel_dcache_range((unsigned long) phys_to_virt(dma_handle) + offset, size);
}
static void pa11_dma_sync_single_for_device(struct device *dev, dma_addr_t dma_handle, unsigned long offset, size_t size, enum dma_data_direction direction)
{
if (direction == DMA_NONE)
BUG();
flush_kernel_dcache_range((unsigned long) phys_to_virt(dma_handle) + offset, size);
}
static void pa11_dma_sync_sg_for_cpu(struct device *dev, struct scatterlist *sglist, int nents, enum dma_data_direction direction)
{
int i;
/* once we do combining we'll need to use phys_to_virt(sg_dma_address(sglist)) */
for (i = 0; i < nents; i++, sglist++ )
flush_kernel_dcache_range(sg_virt_addr(sglist), sglist->length);
}
static void pa11_dma_sync_sg_for_device(struct device *dev, struct scatterlist *sglist, int nents, enum dma_data_direction direction)
{
int i;
/* once we do combining we'll need to use phys_to_virt(sg_dma_address(sglist)) */
for (i = 0; i < nents; i++, sglist++ )
flush_kernel_dcache_range(sg_virt_addr(sglist), sglist->length);
}
struct hppa_dma_ops pcxl_dma_ops = {
.dma_supported = pa11_dma_supported,
.alloc_consistent = pa11_dma_alloc_consistent,
.alloc_noncoherent = pa11_dma_alloc_consistent,
.free_consistent = pa11_dma_free_consistent,
.map_single = pa11_dma_map_single,
.unmap_single = pa11_dma_unmap_single,
.map_sg = pa11_dma_map_sg,
.unmap_sg = pa11_dma_unmap_sg,
.dma_sync_single_for_cpu = pa11_dma_sync_single_for_cpu,
.dma_sync_single_for_device = pa11_dma_sync_single_for_device,
.dma_sync_sg_for_cpu = pa11_dma_sync_sg_for_cpu,
.dma_sync_sg_for_device = pa11_dma_sync_sg_for_device,
};
static void *fail_alloc_consistent(struct device *dev, size_t size,
dma_addr_t *dma_handle, gfp_t flag)
{
return NULL;
}
static void *pa11_dma_alloc_noncoherent(struct device *dev, size_t size,
dma_addr_t *dma_handle, gfp_t flag)
{
void *addr;
addr = (void *)__get_free_pages(flag, get_order(size));
if (addr)
*dma_handle = (dma_addr_t)virt_to_phys(addr);
return addr;
}
static void pa11_dma_free_noncoherent(struct device *dev, size_t size,
void *vaddr, dma_addr_t iova)
{
free_pages((unsigned long)vaddr, get_order(size));
return;
}
struct hppa_dma_ops pcx_dma_ops = {
.dma_supported = pa11_dma_supported,
.alloc_consistent = fail_alloc_consistent,
.alloc_noncoherent = pa11_dma_alloc_noncoherent,
.free_consistent = pa11_dma_free_noncoherent,
.map_single = pa11_dma_map_single,
.unmap_single = pa11_dma_unmap_single,
.map_sg = pa11_dma_map_sg,
.unmap_sg = pa11_dma_unmap_sg,
.dma_sync_single_for_cpu = pa11_dma_sync_single_for_cpu,
.dma_sync_single_for_device = pa11_dma_sync_single_for_device,
.dma_sync_sg_for_cpu = pa11_dma_sync_sg_for_cpu,
.dma_sync_sg_for_device = pa11_dma_sync_sg_for_device,
};