mirror of
https://github.com/AuxXxilium/linux_dsm_epyc7002.git
synced 2024-12-28 11:18:45 +07:00
dfaad39cec
Switch to the generic noncoherent direct mapping implementation. This removes the previous sync_single_for_device implementation, which looks bogus given that no syncing is happening in the similar but more important map_single case. Signed-off-by: Christoph Hellwig <hch@lst.de> Acked-by: Sam Ravnborg <sam@ravnborg.org> Signed-off-by: David S. Miller <davem@davemloft.net>
561 lines
14 KiB
C
561 lines
14 KiB
C
// SPDX-License-Identifier: GPL-2.0
|
|
/*
|
|
* ioport.c: Simple io mapping allocator.
|
|
*
|
|
* Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
|
|
* Copyright (C) 1995 Miguel de Icaza (miguel@nuclecu.unam.mx)
|
|
*
|
|
* 1996: sparc_free_io, 1999: ioremap()/iounmap() by Pete Zaitcev.
|
|
*
|
|
* 2000/01/29
|
|
* <rth> zait: as long as pci_alloc_consistent produces something addressable,
|
|
* things are ok.
|
|
* <zaitcev> rth: no, it is relevant, because get_free_pages returns you a
|
|
* pointer into the big page mapping
|
|
* <rth> zait: so what?
|
|
* <rth> zait: remap_it_my_way(virt_to_phys(get_free_page()))
|
|
* <zaitcev> Hmm
|
|
* <zaitcev> Suppose I did this remap_it_my_way(virt_to_phys(get_free_page())).
|
|
* So far so good.
|
|
* <zaitcev> Now, driver calls pci_free_consistent(with result of
|
|
* remap_it_my_way()).
|
|
* <zaitcev> How do you find the address to pass to free_pages()?
|
|
* <rth> zait: walk the page tables? It's only two or three level after all.
|
|
* <rth> zait: you have to walk them anyway to remove the mapping.
|
|
* <zaitcev> Hmm
|
|
* <zaitcev> Sounds reasonable
|
|
*/
|
|
|
|
#include <linux/module.h>
|
|
#include <linux/sched.h>
|
|
#include <linux/kernel.h>
|
|
#include <linux/errno.h>
|
|
#include <linux/types.h>
|
|
#include <linux/ioport.h>
|
|
#include <linux/mm.h>
|
|
#include <linux/slab.h>
|
|
#include <linux/pci.h> /* struct pci_dev */
|
|
#include <linux/proc_fs.h>
|
|
#include <linux/seq_file.h>
|
|
#include <linux/scatterlist.h>
|
|
#include <linux/dma-noncoherent.h>
|
|
#include <linux/of_device.h>
|
|
|
|
#include <asm/io.h>
|
|
#include <asm/vaddrs.h>
|
|
#include <asm/oplib.h>
|
|
#include <asm/prom.h>
|
|
#include <asm/page.h>
|
|
#include <asm/pgalloc.h>
|
|
#include <asm/dma.h>
|
|
#include <asm/iommu.h>
|
|
#include <asm/io-unit.h>
|
|
#include <asm/leon.h>
|
|
|
|
const struct sparc32_dma_ops *sparc32_dma_ops;
|
|
|
|
/* This function must make sure that caches and memory are coherent after DMA
|
|
* On LEON systems without cache snooping it flushes the entire D-CACHE.
|
|
*/
|
|
static inline void dma_make_coherent(unsigned long pa, unsigned long len)
|
|
{
|
|
if (sparc_cpu_model == sparc_leon) {
|
|
if (!sparc_leon3_snooping_enabled())
|
|
leon_flush_dcache_all();
|
|
}
|
|
}
|
|
|
|
static void __iomem *_sparc_ioremap(struct resource *res, u32 bus, u32 pa, int sz);
|
|
static void __iomem *_sparc_alloc_io(unsigned int busno, unsigned long phys,
|
|
unsigned long size, char *name);
|
|
static void _sparc_free_io(struct resource *res);
|
|
|
|
static void register_proc_sparc_ioport(void);
|
|
|
|
/* This points to the next to use virtual memory for DVMA mappings */
|
|
static struct resource _sparc_dvma = {
|
|
.name = "sparc_dvma", .start = DVMA_VADDR, .end = DVMA_END - 1
|
|
};
|
|
/* This points to the start of I/O mappings, cluable from outside. */
|
|
/*ext*/ struct resource sparc_iomap = {
|
|
.name = "sparc_iomap", .start = IOBASE_VADDR, .end = IOBASE_END - 1
|
|
};
|
|
|
|
/*
|
|
* Our mini-allocator...
|
|
* Boy this is gross! We need it because we must map I/O for
|
|
* timers and interrupt controller before the kmalloc is available.
|
|
*/
|
|
|
|
#define XNMLN 15
|
|
#define XNRES 10 /* SS-10 uses 8 */
|
|
|
|
struct xresource {
|
|
struct resource xres; /* Must be first */
|
|
int xflag; /* 1 == used */
|
|
char xname[XNMLN+1];
|
|
};
|
|
|
|
static struct xresource xresv[XNRES];
|
|
|
|
static struct xresource *xres_alloc(void) {
|
|
struct xresource *xrp;
|
|
int n;
|
|
|
|
xrp = xresv;
|
|
for (n = 0; n < XNRES; n++) {
|
|
if (xrp->xflag == 0) {
|
|
xrp->xflag = 1;
|
|
return xrp;
|
|
}
|
|
xrp++;
|
|
}
|
|
return NULL;
|
|
}
|
|
|
|
static void xres_free(struct xresource *xrp) {
|
|
xrp->xflag = 0;
|
|
}
|
|
|
|
/*
|
|
* These are typically used in PCI drivers
|
|
* which are trying to be cross-platform.
|
|
*
|
|
* Bus type is always zero on IIep.
|
|
*/
|
|
void __iomem *ioremap(phys_addr_t offset, size_t size)
|
|
{
|
|
char name[14];
|
|
|
|
sprintf(name, "phys_%08x", (u32)offset);
|
|
return _sparc_alloc_io(0, (unsigned long)offset, size, name);
|
|
}
|
|
EXPORT_SYMBOL(ioremap);
|
|
|
|
/*
|
|
* Complementary to ioremap().
|
|
*/
|
|
void iounmap(volatile void __iomem *virtual)
|
|
{
|
|
unsigned long vaddr = (unsigned long) virtual & PAGE_MASK;
|
|
struct resource *res;
|
|
|
|
/*
|
|
* XXX Too slow. Can have 8192 DVMA pages on sun4m in the worst case.
|
|
* This probably warrants some sort of hashing.
|
|
*/
|
|
if ((res = lookup_resource(&sparc_iomap, vaddr)) == NULL) {
|
|
printk("free_io/iounmap: cannot free %lx\n", vaddr);
|
|
return;
|
|
}
|
|
_sparc_free_io(res);
|
|
|
|
if ((char *)res >= (char*)xresv && (char *)res < (char *)&xresv[XNRES]) {
|
|
xres_free((struct xresource *)res);
|
|
} else {
|
|
kfree(res);
|
|
}
|
|
}
|
|
EXPORT_SYMBOL(iounmap);
|
|
|
|
void __iomem *of_ioremap(struct resource *res, unsigned long offset,
|
|
unsigned long size, char *name)
|
|
{
|
|
return _sparc_alloc_io(res->flags & 0xF,
|
|
res->start + offset,
|
|
size, name);
|
|
}
|
|
EXPORT_SYMBOL(of_ioremap);
|
|
|
|
void of_iounmap(struct resource *res, void __iomem *base, unsigned long size)
|
|
{
|
|
iounmap(base);
|
|
}
|
|
EXPORT_SYMBOL(of_iounmap);
|
|
|
|
/*
|
|
* Meat of mapping
|
|
*/
|
|
static void __iomem *_sparc_alloc_io(unsigned int busno, unsigned long phys,
|
|
unsigned long size, char *name)
|
|
{
|
|
static int printed_full;
|
|
struct xresource *xres;
|
|
struct resource *res;
|
|
char *tack;
|
|
int tlen;
|
|
void __iomem *va; /* P3 diag */
|
|
|
|
if (name == NULL) name = "???";
|
|
|
|
if ((xres = xres_alloc()) != NULL) {
|
|
tack = xres->xname;
|
|
res = &xres->xres;
|
|
} else {
|
|
if (!printed_full) {
|
|
printk("ioremap: done with statics, switching to malloc\n");
|
|
printed_full = 1;
|
|
}
|
|
tlen = strlen(name);
|
|
tack = kmalloc(sizeof (struct resource) + tlen + 1, GFP_KERNEL);
|
|
if (tack == NULL) return NULL;
|
|
memset(tack, 0, sizeof(struct resource));
|
|
res = (struct resource *) tack;
|
|
tack += sizeof (struct resource);
|
|
}
|
|
|
|
strlcpy(tack, name, XNMLN+1);
|
|
res->name = tack;
|
|
|
|
va = _sparc_ioremap(res, busno, phys, size);
|
|
/* printk("ioremap(0x%x:%08lx[0x%lx])=%p\n", busno, phys, size, va); */ /* P3 diag */
|
|
return va;
|
|
}
|
|
|
|
/*
|
|
*/
|
|
static void __iomem *
|
|
_sparc_ioremap(struct resource *res, u32 bus, u32 pa, int sz)
|
|
{
|
|
unsigned long offset = ((unsigned long) pa) & (~PAGE_MASK);
|
|
|
|
if (allocate_resource(&sparc_iomap, res,
|
|
(offset + sz + PAGE_SIZE-1) & PAGE_MASK,
|
|
sparc_iomap.start, sparc_iomap.end, PAGE_SIZE, NULL, NULL) != 0) {
|
|
/* Usually we cannot see printks in this case. */
|
|
prom_printf("alloc_io_res(%s): cannot occupy\n",
|
|
(res->name != NULL)? res->name: "???");
|
|
prom_halt();
|
|
}
|
|
|
|
pa &= PAGE_MASK;
|
|
srmmu_mapiorange(bus, pa, res->start, resource_size(res));
|
|
|
|
return (void __iomem *)(unsigned long)(res->start + offset);
|
|
}
|
|
|
|
/*
|
|
* Complementary to _sparc_ioremap().
|
|
*/
|
|
static void _sparc_free_io(struct resource *res)
|
|
{
|
|
unsigned long plen;
|
|
|
|
plen = resource_size(res);
|
|
BUG_ON((plen & (PAGE_SIZE-1)) != 0);
|
|
srmmu_unmapiorange(res->start, plen);
|
|
release_resource(res);
|
|
}
|
|
|
|
#ifdef CONFIG_SBUS
|
|
|
|
void sbus_set_sbus64(struct device *dev, int x)
|
|
{
|
|
printk("sbus_set_sbus64: unsupported\n");
|
|
}
|
|
EXPORT_SYMBOL(sbus_set_sbus64);
|
|
|
|
/*
|
|
* Allocate a chunk of memory suitable for DMA.
|
|
* Typically devices use them for control blocks.
|
|
* CPU may access them without any explicit flushing.
|
|
*/
|
|
static void *sbus_alloc_coherent(struct device *dev, size_t len,
|
|
dma_addr_t *dma_addrp, gfp_t gfp,
|
|
unsigned long attrs)
|
|
{
|
|
struct platform_device *op = to_platform_device(dev);
|
|
unsigned long len_total = PAGE_ALIGN(len);
|
|
unsigned long va;
|
|
struct resource *res;
|
|
int order;
|
|
|
|
/* XXX why are some lengths signed, others unsigned? */
|
|
if (len <= 0) {
|
|
return NULL;
|
|
}
|
|
/* XXX So what is maxphys for us and how do drivers know it? */
|
|
if (len > 256*1024) { /* __get_free_pages() limit */
|
|
return NULL;
|
|
}
|
|
|
|
order = get_order(len_total);
|
|
va = __get_free_pages(gfp, order);
|
|
if (va == 0)
|
|
goto err_nopages;
|
|
|
|
if ((res = kzalloc(sizeof(struct resource), GFP_KERNEL)) == NULL)
|
|
goto err_nomem;
|
|
|
|
if (allocate_resource(&_sparc_dvma, res, len_total,
|
|
_sparc_dvma.start, _sparc_dvma.end, PAGE_SIZE, NULL, NULL) != 0) {
|
|
printk("sbus_alloc_consistent: cannot occupy 0x%lx", len_total);
|
|
goto err_nova;
|
|
}
|
|
|
|
// XXX The sbus_map_dma_area does this for us below, see comments.
|
|
// srmmu_mapiorange(0, virt_to_phys(va), res->start, len_total);
|
|
/*
|
|
* XXX That's where sdev would be used. Currently we load
|
|
* all iommu tables with the same translations.
|
|
*/
|
|
if (sbus_map_dma_area(dev, dma_addrp, va, res->start, len_total) != 0)
|
|
goto err_noiommu;
|
|
|
|
res->name = op->dev.of_node->name;
|
|
|
|
return (void *)(unsigned long)res->start;
|
|
|
|
err_noiommu:
|
|
release_resource(res);
|
|
err_nova:
|
|
kfree(res);
|
|
err_nomem:
|
|
free_pages(va, order);
|
|
err_nopages:
|
|
return NULL;
|
|
}
|
|
|
|
static void sbus_free_coherent(struct device *dev, size_t n, void *p,
|
|
dma_addr_t ba, unsigned long attrs)
|
|
{
|
|
struct resource *res;
|
|
struct page *pgv;
|
|
|
|
if ((res = lookup_resource(&_sparc_dvma,
|
|
(unsigned long)p)) == NULL) {
|
|
printk("sbus_free_consistent: cannot free %p\n", p);
|
|
return;
|
|
}
|
|
|
|
if (((unsigned long)p & (PAGE_SIZE-1)) != 0) {
|
|
printk("sbus_free_consistent: unaligned va %p\n", p);
|
|
return;
|
|
}
|
|
|
|
n = PAGE_ALIGN(n);
|
|
if (resource_size(res) != n) {
|
|
printk("sbus_free_consistent: region 0x%lx asked 0x%zx\n",
|
|
(long)resource_size(res), n);
|
|
return;
|
|
}
|
|
|
|
release_resource(res);
|
|
kfree(res);
|
|
|
|
pgv = virt_to_page(p);
|
|
sbus_unmap_dma_area(dev, ba, n);
|
|
|
|
__free_pages(pgv, get_order(n));
|
|
}
|
|
|
|
/*
|
|
* Map a chunk of memory so that devices can see it.
|
|
* CPU view of this memory may be inconsistent with
|
|
* a device view and explicit flushing is necessary.
|
|
*/
|
|
static dma_addr_t sbus_map_page(struct device *dev, struct page *page,
|
|
unsigned long offset, size_t len,
|
|
enum dma_data_direction dir,
|
|
unsigned long attrs)
|
|
{
|
|
void *va = page_address(page) + offset;
|
|
|
|
/* XXX why are some lengths signed, others unsigned? */
|
|
if (len <= 0) {
|
|
return 0;
|
|
}
|
|
/* XXX So what is maxphys for us and how do drivers know it? */
|
|
if (len > 256*1024) { /* __get_free_pages() limit */
|
|
return 0;
|
|
}
|
|
return mmu_get_scsi_one(dev, va, len);
|
|
}
|
|
|
|
static void sbus_unmap_page(struct device *dev, dma_addr_t ba, size_t n,
|
|
enum dma_data_direction dir, unsigned long attrs)
|
|
{
|
|
mmu_release_scsi_one(dev, ba, n);
|
|
}
|
|
|
|
static int sbus_map_sg(struct device *dev, struct scatterlist *sg, int n,
|
|
enum dma_data_direction dir, unsigned long attrs)
|
|
{
|
|
mmu_get_scsi_sgl(dev, sg, n);
|
|
return n;
|
|
}
|
|
|
|
static void sbus_unmap_sg(struct device *dev, struct scatterlist *sg, int n,
|
|
enum dma_data_direction dir, unsigned long attrs)
|
|
{
|
|
mmu_release_scsi_sgl(dev, sg, n);
|
|
}
|
|
|
|
static void sbus_sync_sg_for_cpu(struct device *dev, struct scatterlist *sg,
|
|
int n, enum dma_data_direction dir)
|
|
{
|
|
BUG();
|
|
}
|
|
|
|
static void sbus_sync_sg_for_device(struct device *dev, struct scatterlist *sg,
|
|
int n, enum dma_data_direction dir)
|
|
{
|
|
BUG();
|
|
}
|
|
|
|
static int sbus_dma_supported(struct device *dev, u64 mask)
|
|
{
|
|
return 0;
|
|
}
|
|
|
|
static const struct dma_map_ops sbus_dma_ops = {
|
|
.alloc = sbus_alloc_coherent,
|
|
.free = sbus_free_coherent,
|
|
.map_page = sbus_map_page,
|
|
.unmap_page = sbus_unmap_page,
|
|
.map_sg = sbus_map_sg,
|
|
.unmap_sg = sbus_unmap_sg,
|
|
.sync_sg_for_cpu = sbus_sync_sg_for_cpu,
|
|
.sync_sg_for_device = sbus_sync_sg_for_device,
|
|
.dma_supported = sbus_dma_supported,
|
|
};
|
|
|
|
static int __init sparc_register_ioport(void)
|
|
{
|
|
register_proc_sparc_ioport();
|
|
|
|
return 0;
|
|
}
|
|
|
|
arch_initcall(sparc_register_ioport);
|
|
|
|
#endif /* CONFIG_SBUS */
|
|
|
|
|
|
/* Allocate and map kernel buffer using consistent mode DMA for a device.
|
|
* hwdev should be valid struct pci_dev pointer for PCI devices.
|
|
*/
|
|
void *arch_dma_alloc(struct device *dev, size_t size, dma_addr_t *dma_handle,
|
|
gfp_t gfp, unsigned long attrs)
|
|
{
|
|
unsigned long len_total = PAGE_ALIGN(size);
|
|
void *va;
|
|
struct resource *res;
|
|
int order;
|
|
|
|
if (size == 0) {
|
|
return NULL;
|
|
}
|
|
if (size > 256*1024) { /* __get_free_pages() limit */
|
|
return NULL;
|
|
}
|
|
|
|
order = get_order(len_total);
|
|
va = (void *) __get_free_pages(gfp, order);
|
|
if (va == NULL) {
|
|
printk("%s: no %ld pages\n", __func__, len_total>>PAGE_SHIFT);
|
|
goto err_nopages;
|
|
}
|
|
|
|
if ((res = kzalloc(sizeof(struct resource), GFP_KERNEL)) == NULL) {
|
|
printk("%s: no core\n", __func__);
|
|
goto err_nomem;
|
|
}
|
|
|
|
if (allocate_resource(&_sparc_dvma, res, len_total,
|
|
_sparc_dvma.start, _sparc_dvma.end, PAGE_SIZE, NULL, NULL) != 0) {
|
|
printk("%s: cannot occupy 0x%lx", __func__, len_total);
|
|
goto err_nova;
|
|
}
|
|
srmmu_mapiorange(0, virt_to_phys(va), res->start, len_total);
|
|
|
|
*dma_handle = virt_to_phys(va);
|
|
return (void *) res->start;
|
|
|
|
err_nova:
|
|
kfree(res);
|
|
err_nomem:
|
|
free_pages((unsigned long)va, order);
|
|
err_nopages:
|
|
return NULL;
|
|
}
|
|
|
|
/* Free and unmap a consistent DMA buffer.
|
|
* cpu_addr is what was returned arch_dma_alloc, size must be the same as what
|
|
* was passed into arch_dma_alloc, and likewise dma_addr must be the same as
|
|
* what *dma_ndler was set to.
|
|
*
|
|
* References to the memory and mappings associated with cpu_addr/dma_addr
|
|
* past this call are illegal.
|
|
*/
|
|
void arch_dma_free(struct device *dev, size_t size, void *cpu_addr,
|
|
dma_addr_t dma_addr, unsigned long attrs)
|
|
{
|
|
struct resource *res;
|
|
|
|
if ((res = lookup_resource(&_sparc_dvma,
|
|
(unsigned long)cpu_addr)) == NULL) {
|
|
printk("%s: cannot free %p\n", __func__, cpu_addr);
|
|
return;
|
|
}
|
|
|
|
if (((unsigned long)cpu_addr & (PAGE_SIZE-1)) != 0) {
|
|
printk("%s: unaligned va %p\n", __func__, cpu_addr);
|
|
return;
|
|
}
|
|
|
|
size = PAGE_ALIGN(size);
|
|
if (resource_size(res) != size) {
|
|
printk("%s: region 0x%lx asked 0x%zx\n", __func__,
|
|
(long)resource_size(res), size);
|
|
return;
|
|
}
|
|
|
|
dma_make_coherent(dma_addr, size);
|
|
srmmu_unmapiorange((unsigned long)cpu_addr, size);
|
|
|
|
release_resource(res);
|
|
kfree(res);
|
|
free_pages((unsigned long)phys_to_virt(dma_addr), get_order(size));
|
|
}
|
|
|
|
/* IIep is write-through, not flushing on cpu to device transfer. */
|
|
|
|
void arch_sync_dma_for_cpu(struct device *dev, phys_addr_t paddr,
|
|
size_t size, enum dma_data_direction dir)
|
|
{
|
|
if (dir != PCI_DMA_TODEVICE)
|
|
dma_make_coherent(paddr, PAGE_ALIGN(size));
|
|
}
|
|
|
|
const struct dma_map_ops *dma_ops = &sbus_dma_ops;
|
|
EXPORT_SYMBOL(dma_ops);
|
|
|
|
#ifdef CONFIG_PROC_FS
|
|
|
|
static int sparc_io_proc_show(struct seq_file *m, void *v)
|
|
{
|
|
struct resource *root = m->private, *r;
|
|
const char *nm;
|
|
|
|
for (r = root->child; r != NULL; r = r->sibling) {
|
|
if ((nm = r->name) == NULL) nm = "???";
|
|
seq_printf(m, "%016llx-%016llx: %s\n",
|
|
(unsigned long long)r->start,
|
|
(unsigned long long)r->end, nm);
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
#endif /* CONFIG_PROC_FS */
|
|
|
|
static void register_proc_sparc_ioport(void)
|
|
{
|
|
#ifdef CONFIG_PROC_FS
|
|
proc_create_single_data("io_map", 0, NULL, sparc_io_proc_show,
|
|
&sparc_iomap);
|
|
proc_create_single_data("dvma_map", 0, NULL, sparc_io_proc_show,
|
|
&_sparc_dvma);
|
|
#endif
|
|
}
|