linux/arch/mips/include/asm/page.h

262 lines
7.2 KiB
C
Raw Normal View History

/*
* This file is subject to the terms and conditions of the GNU General Public
* License. See the file "COPYING" in the main directory of this archive
* for more details.
*
* Copyright (C) 1994 - 1999, 2000, 03 Ralf Baechle
* Copyright (C) 1999, 2000 Silicon Graphics, Inc.
*/
#ifndef _ASM_PAGE_H
#define _ASM_PAGE_H
#include <spaces.h>
#include <linux/const.h>
#include <linux/kernel.h>
#include <asm/mipsregs.h>
/*
* PAGE_SHIFT determines the page size
*/
#ifdef CONFIG_PAGE_SIZE_4KB
#define PAGE_SHIFT 12
#endif
#ifdef CONFIG_PAGE_SIZE_8KB
#define PAGE_SHIFT 13
#endif
#ifdef CONFIG_PAGE_SIZE_16KB
#define PAGE_SHIFT 14
#endif
#ifdef CONFIG_PAGE_SIZE_32KB
#define PAGE_SHIFT 15
#endif
#ifdef CONFIG_PAGE_SIZE_64KB
#define PAGE_SHIFT 16
#endif
#define PAGE_SIZE (_AC(1,UL) << PAGE_SHIFT)
#define PAGE_MASK (~((1 << PAGE_SHIFT) - 1))
/*
* This is used for calculating the real page sizes
* for FTLB or VTLB + FTLB configurations.
*/
static inline unsigned int page_size_ftlb(unsigned int mmuextdef)
{
switch (mmuextdef) {
case MIPS_CONF4_MMUEXTDEF_FTLBSIZEEXT:
if (PAGE_SIZE == (1 << 30))
return 5;
if (PAGE_SIZE == (1llu << 32))
return 6;
if (PAGE_SIZE > (256 << 10))
return 7; /* reserved */
/* fall through */
case MIPS_CONF4_MMUEXTDEF_VTLBSIZEEXT:
return (PAGE_SHIFT - 10) / 2;
default:
panic("Invalid FTLB configuration with Conf4_mmuextdef=%d value\n",
mmuextdef >> 14);
}
}
#ifdef CONFIG_MIPS_HUGE_TLB_SUPPORT
#define HPAGE_SHIFT (PAGE_SHIFT + PAGE_SHIFT - 3)
#define HPAGE_SIZE (_AC(1,UL) << HPAGE_SHIFT)
#define HPAGE_MASK (~(HPAGE_SIZE - 1))
#define HUGETLB_PAGE_ORDER (HPAGE_SHIFT - PAGE_SHIFT)
#else /* !CONFIG_MIPS_HUGE_TLB_SUPPORT */
#define HPAGE_SHIFT ({BUILD_BUG(); 0; })
#define HPAGE_SIZE ({BUILD_BUG(); 0; })
#define HPAGE_MASK ({BUILD_BUG(); 0; })
#define HUGETLB_PAGE_ORDER ({BUILD_BUG(); 0; })
#endif /* CONFIG_MIPS_HUGE_TLB_SUPPORT */
#include <linux/pfn.h>
extern void build_clear_page(void);
extern void build_copy_page(void);
/*
* It's normally defined only for FLATMEM config but it's
* used in our early mem init code for all memory models.
* So always define it.
*/
#define ARCH_PFN_OFFSET PFN_UP(PHYS_OFFSET)
extern void clear_page(void * page);
extern void copy_page(void * to, void * from);
extern unsigned long shm_align_mask;
static inline unsigned long pages_do_alias(unsigned long addr1,
unsigned long addr2)
{
return (addr1 ^ addr2) & shm_align_mask;
}
struct page;
static inline void clear_user_page(void *addr, unsigned long vaddr,
struct page *page)
{
extern void (*flush_data_cache_page)(unsigned long addr);
clear_page(addr);
[MIPS] Retire flush_icache_page from mm use. On the 34K the redundant cache operations were causing excessive stalls resulting in realtime code running on the second VPE missing its deadline. For all other platforms this patch is just a significant performance improvment as illustrated by below benchmark numbers. Processor, Processes - times in microseconds - smaller is better ------------------------------------------------------------------------------ Host OS Mhz null null open slct sig sig fork exec sh call I/O stat clos TCP inst hndl proc proc proc --------- ------------- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- 25Kf 2.6.18-rc4 533 0.49 1.16 7.57 33.4 30.5 1.34 12.4 5497 17.K 54.K 25Kf 2.6.18-rc4-p 533 0.49 1.16 6.68 23.0 30.7 1.36 8.55 5030 16.K 48.K 4Kc 2.6.18-rc4 80 4.21 15.0 131. 289. 261. 16.5 258. 18.K 70.K 227K 4Kc 2.6.18-rc4-p 80 4.34 13.1 128. 285. 262. 18.2 258. 12.K 52.K 176K 34Kc 2.6.18-rc4 40 5.01 14.0 61.6 90.0 477. 17.9 94.7 29.K 108K 342K 34Kc 2.6.18-rc4-p 40 4.98 13.9 61.2 89.7 475. 17.6 93.7 8758 44.K 158K BCM1480 2.6.18-rc4 700 0.28 0.60 3.68 5.92 16.0 0.78 5.08 931. 3163 15.K BCM1480 2.6.18-rc4-p 700 0.28 0.61 3.65 5.85 16.0 0.79 5.20 395. 1464 8385 TX49-16K 2.6.18-rc3 197 0.73 2.41 19.0 37.8 82.9 2.94 17.5 4438 14.K 56.K TX49-16K 2.6.18-rc3-p 197 0.73 2.40 19.9 36.3 82.9 2.94 23.4 2577 9103 38.K TX49-32K 2.6.18-rc3 396 0.36 1.19 6.80 11.8 41.0 1.46 8.17 2738 8465 32.K TX49-32K 2.6.18-rc3-p 396 0.36 1.19 6.82 10.2 41.0 1.46 8.18 1330 4638 18.K Original patch by me with enhancements by Atsushi Nemoto. Signed-off-by: Ralf Baechle <ralf@linux-mips.org> Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
2006-08-12 15:40:08 +00:00
if (pages_do_alias((unsigned long) addr, vaddr & PAGE_MASK))
flush_data_cache_page((unsigned long)addr);
}
struct vm_area_struct;
extern void copy_user_highpage(struct page *to, struct page *from,
unsigned long vaddr, struct vm_area_struct *vma);
#define __HAVE_ARCH_COPY_USER_HIGHPAGE
/*
* These are used to make use of C type-checking..
*/
#ifdef CONFIG_PHYS_ADDR_T_64BIT
#ifdef CONFIG_CPU_MIPS32
typedef struct { unsigned long pte_low, pte_high; } pte_t;
#define pte_val(x) ((x).pte_low | ((unsigned long long)(x).pte_high << 32))
#define __pte(x) ({ pte_t __pte = {(x), ((unsigned long long)(x)) >> 32}; __pte; })
#else
typedef struct { unsigned long long pte; } pte_t;
#define pte_val(x) ((x).pte)
#define __pte(x) ((pte_t) { (x) } )
#endif
#else
typedef struct { unsigned long pte; } pte_t;
#define pte_val(x) ((x).pte)
#define __pte(x) ((pte_t) { (x) } )
#endif
CONFIG_HIGHPTE vs. sub-page page tables. Background: I've implemented 1K/2K page tables for s390. These sub-page page tables are required to properly support the s390 virtualization instruction with KVM. The SIE instruction requires that the page tables have 256 page table entries (pte) followed by 256 page status table entries (pgste). The pgstes are only required if the process is using the SIE instruction. The pgstes are updated by the hardware and by the hypervisor for a number of reasons, one of them is dirty and reference bit tracking. To avoid wasting memory the standard pte table allocation should return 1K/2K (31/64 bit) and 2K/4K if the process is using SIE. Problem: Page size on s390 is 4K, page table size is 1K or 2K. That means the s390 version for pte_alloc_one cannot return a pointer to a struct page. Trouble is that with the CONFIG_HIGHPTE feature on x86 pte_alloc_one cannot return a pointer to a pte either, since that would require more than 32 bit for the return value of pte_alloc_one (and the pte * would not be accessible since its not kmapped). Solution: The only solution I found to this dilemma is a new typedef: a pgtable_t. For s390 pgtable_t will be a (pte *) - to be introduced with a later patch. For everybody else it will be a (struct page *). The additional problem with the initialization of the ptl lock and the NR_PAGETABLE accounting is solved with a constructor pgtable_page_ctor and a destructor pgtable_page_dtor. The page table allocation and free functions need to call these two whenever a page table page is allocated or freed. pmd_populate will get a pgtable_t instead of a struct page pointer. To get the pgtable_t back from a pmd entry that has been installed with pmd_populate a new function pmd_pgtable is added. It replaces the pmd_page call in free_pte_range and apply_to_pte_range. Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: <linux-arch@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-08 12:22:04 +00:00
typedef struct page *pgtable_t;
/*
* Right now we don't support 4-level pagetables, so all pud-related
* definitions come from <asm-generic/pgtable-nopud.h>.
*/
/*
* Finall the top of the hierarchy, the pgd
*/
typedef struct { unsigned long pgd; } pgd_t;
#define pgd_val(x) ((x).pgd)
#define __pgd(x) ((pgd_t) { (x) } )
/*
* Manipulate page protection bits
*/
typedef struct { unsigned long pgprot; } pgprot_t;
#define pgprot_val(x) ((x).pgprot)
#define __pgprot(x) ((pgprot_t) { (x) } )
/*
* On R4000-style MMUs where a TLB entry is mapping a adjacent even / odd
* pair of pages we only have a single global bit per pair of pages. When
* writing to the TLB make sure we always have the bit set for both pages
* or none. This macro is used to access the `buddy' of the pte we're just
* working on.
*/
#define ptep_buddy(x) ((pte_t *)((unsigned long)(x) ^ sizeof(pte_t)))
/*
* __pa()/__va() should be used only during mem init.
*/
MIPS: Use CPHYSADDR to implement mips32 __pa Use CPHYSADDR to implement the __pa macro converting from a virtual to a physical address for MIPS32, much as is already done for MIPS64 (though without the complication of having both compatibility & XKPHYS segments). This allows for __pa to work regardless of whether the address being translated is in kseg0 or kseg1, unlike the previous subtraction based approach which only worked for addresses in kseg0. Working for kseg1 addresses is important if __pa is used on addresses allocated by dma_alloc_coherent, where on systems with non-coherent I/O we provide addresses in kseg1. If this address is then used with dma_map_single_attrs then it is provided to virt_to_page, which in turn calls virt_to_phys which is a wrapper around __pa. The result is that we end up with a physical address 0x20000000 bytes (ie. the size of kseg0) too high. In addition to providing consistency with MIPS64 & fixing the kseg1 case above this has the added bonus of generating smaller code for systems implementing MIPS32r2 & beyond, where a single ext instruction can extract the physical address rather than needing to load an immediate into a temp register & subtract it. This results in ~1.3KB savings for a boston_defconfig kernel adjusted to set CONFIG_32BIT=y. This patch does not change the EVA case, which may or may not have similar issues around handling both cached & uncached addresses but is beyond the scope of this patch. Signed-off-by: Paul Burton <paul.burton@imgtec.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/13836/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2016-08-02 10:40:57 +00:00
static inline unsigned long ___pa(unsigned long x)
{
if (IS_ENABLED(CONFIG_64BIT)) {
MIPS: Use CPHYSADDR to implement mips32 __pa Use CPHYSADDR to implement the __pa macro converting from a virtual to a physical address for MIPS32, much as is already done for MIPS64 (though without the complication of having both compatibility & XKPHYS segments). This allows for __pa to work regardless of whether the address being translated is in kseg0 or kseg1, unlike the previous subtraction based approach which only worked for addresses in kseg0. Working for kseg1 addresses is important if __pa is used on addresses allocated by dma_alloc_coherent, where on systems with non-coherent I/O we provide addresses in kseg1. If this address is then used with dma_map_single_attrs then it is provided to virt_to_page, which in turn calls virt_to_phys which is a wrapper around __pa. The result is that we end up with a physical address 0x20000000 bytes (ie. the size of kseg0) too high. In addition to providing consistency with MIPS64 & fixing the kseg1 case above this has the added bonus of generating smaller code for systems implementing MIPS32r2 & beyond, where a single ext instruction can extract the physical address rather than needing to load an immediate into a temp register & subtract it. This results in ~1.3KB savings for a boston_defconfig kernel adjusted to set CONFIG_32BIT=y. This patch does not change the EVA case, which may or may not have similar issues around handling both cached & uncached addresses but is beyond the scope of this patch. Signed-off-by: Paul Burton <paul.burton@imgtec.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/13836/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2016-08-02 10:40:57 +00:00
/*
* For MIPS64 the virtual address may either be in one of
* the compatibility segements ckseg0 or ckseg1, or it may
* be in xkphys.
*/
return x < CKSEG0 ? XPHYSADDR(x) : CPHYSADDR(x);
}
if (!IS_ENABLED(CONFIG_EVA)) {
MIPS: Use CPHYSADDR to implement mips32 __pa Use CPHYSADDR to implement the __pa macro converting from a virtual to a physical address for MIPS32, much as is already done for MIPS64 (though without the complication of having both compatibility & XKPHYS segments). This allows for __pa to work regardless of whether the address being translated is in kseg0 or kseg1, unlike the previous subtraction based approach which only worked for addresses in kseg0. Working for kseg1 addresses is important if __pa is used on addresses allocated by dma_alloc_coherent, where on systems with non-coherent I/O we provide addresses in kseg1. If this address is then used with dma_map_single_attrs then it is provided to virt_to_page, which in turn calls virt_to_phys which is a wrapper around __pa. The result is that we end up with a physical address 0x20000000 bytes (ie. the size of kseg0) too high. In addition to providing consistency with MIPS64 & fixing the kseg1 case above this has the added bonus of generating smaller code for systems implementing MIPS32r2 & beyond, where a single ext instruction can extract the physical address rather than needing to load an immediate into a temp register & subtract it. This results in ~1.3KB savings for a boston_defconfig kernel adjusted to set CONFIG_32BIT=y. This patch does not change the EVA case, which may or may not have similar issues around handling both cached & uncached addresses but is beyond the scope of this patch. Signed-off-by: Paul Burton <paul.burton@imgtec.com> Cc: Dan Williams <dan.j.williams@intel.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org Patchwork: https://patchwork.linux-mips.org/patch/13836/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2016-08-02 10:40:57 +00:00
/*
* We're using the standard MIPS32 legacy memory map, ie.
* the address x is going to be in kseg0 or kseg1. We can
* handle either case by masking out the desired bits using
* CPHYSADDR.
*/
return CPHYSADDR(x);
}
/*
* EVA is in use so the memory map could be anything, making it not
* safe to just mask out bits.
*/
return x - PAGE_OFFSET + PHYS_OFFSET;
}
#define __pa(x) ___pa((unsigned long)(x))
#define __va(x) ((void *)((unsigned long)(x) + PAGE_OFFSET - PHYS_OFFSET))
#include <asm/io.h>
/*
* RELOC_HIDE was originally added by 6007b903dfe5f1d13e0c711ac2894bdd4a61b1ad
* (lmo) rsp. 8431fd094d625b94d364fe393076ccef88e6ce18 (kernel.org). The
* discussion can be found in lkml posting
* <a2ebde260608230500o3407b108hc03debb9da6e62c@mail.gmail.com> which is
* archived at http://lists.linuxcoding.com/kernel/2006-q3/msg17360.html
*
* It is unclear if the misscompilations mentioned in
* http://lkml.org/lkml/2010/8/8/138 also affect MIPS so we keep this one
* until GCC 3.x has been retired before we can apply
* https://patchwork.linux-mips.org/patch/1541/
*/
#ifndef __pa_symbol
#define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0))
#endif
#define pfn_to_kaddr(pfn) __va((pfn) << PAGE_SHIFT)
#ifdef CONFIG_FLATMEM
static inline int pfn_valid(unsigned long pfn)
{
/* avoid <linux/mm.h> include hell */
extern unsigned long max_mapnr;
unsigned long pfn_offset = ARCH_PFN_OFFSET;
return pfn >= pfn_offset && pfn < max_mapnr;
}
#elif defined(CONFIG_SPARSEMEM)
/* pfn_valid is defined in linux/mmzone.h */
#elif defined(CONFIG_NEED_MULTIPLE_NODES)
#define pfn_valid(pfn) \
({ \
unsigned long __pfn = (pfn); \
int __n = pfn_to_nid(__pfn); \
((__n >= 0) ? (__pfn < NODE_DATA(__n)->node_start_pfn + \
NODE_DATA(__n)->node_spanned_pages) \
: 0); \
})
#endif
#define virt_to_pfn(kaddr) PFN_DOWN(virt_to_phys((void *)(kaddr)))
#define virt_to_page(kaddr) pfn_to_page(virt_to_pfn(kaddr))
extern int __virt_addr_valid(const volatile void *kaddr);
#define virt_addr_valid(kaddr) \
__virt_addr_valid((const volatile void *) (kaddr))
#define VM_DATA_DEFAULT_FLAGS \
(VM_READ | VM_WRITE | \
((current->personality & READ_IMPLIES_EXEC) ? VM_EXEC : 0) | \
VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC)
#define UNCAC_ADDR(addr) ((addr) - PAGE_OFFSET + UNCAC_BASE)
#define CAC_ADDR(addr) ((addr) - UNCAC_BASE + PAGE_OFFSET)
#include <asm-generic/memory_model.h>
#include <asm-generic/getorder.h>
#endif /* _ASM_PAGE_H */