x86: Full support of PAT
Subsystems (5)
Juergen Gross
maintainercover x86: Full support of PAT
The x86 architecture offers via the PAT (Page Attribute Table) a way to specify different caching modes in page table entries. The PAT MSR contains 8 entries each specifying one of 6 possible cache modes. A pte references one of those entries via 3 bits: _PAGE_PAT, _PAGE_PWT and _PAGE_PCD.
The Linux kernel currently supports only 4 different cache modes. The PAT MSR is set up in a way that the setting of _PAGE_PAT in a pte doesn't matter: the top 4 entries in the PAT MSR are the same as the 4 lower entries.
This results in the kernel not supporting e.g. write-through mode. Especially this cache mode would speed up drivers of video cards which now have to use uncached accesses.
OTOH some old processors (Pentium) don't support PAT correctly and the Xen hypervisor has been using a different PAT MSR configuration for some time now and can't change that as this setting is part of the ABI.
This patch set abstracts the cache mode from the pte and introduces tables to translate between cache mode and pte bits (the default cache mode "write back" is hard-wired to PAT entry 0). The tables are statically initialized with values being compatible to old processors and current usage. As soon as the PAT MSR is changed (or - in case of Xen - is read at boot time) the tables are changed accordingly. Requests of mappings with special cache modes are always possible now, in case they are not supported there will be a fallback to a compatible but slower mode.
Summing it up, this patch set adds the following features:
- capability to support WT and WP cache modes on processors with full PAT support
- processors with no or uncorrect PAT support are still working as today, even if WT or WP cache mode are selected by drivers for some pages
- reduction of Xen special handling regarding cache mode
Changes in V5:
- split up first patch as requested by Ingo Molnar and Thomas Gleixner
- add a helper function in pat_init_cache_modes() as requested by Ingo Molnar
Changes in V4:
- rebased to 3.18-rc2
Changes in V3:
- corrected two minor nits (UC_MINUS, again) detected by Toshi Kani
Changes in V2:
- simplified handling of PAT MSR write under Xen as suggested by David Vrabel
- removed resetting of pat_enabled under Xen
- two small corrections requested by Toshi Kani (UC_MINUS cache mode in vermilion driver, fix 32 bit kernel build failure)
- correct build error on non-x86 arch by moving definition of update_cache_mode_entry() to x86 specific header
Changes since RFC:
- renamed functions and variables as suggested by Toshi Kani
- corrected cache mode bits for WT and WP
- modified handling of PAT MSR write under Xen as suggested by Jan Beulich
Juergen Gross (17):
x86: Make page cache mode a real type
x86: Use new cache mode type in include/asm/fb.h
x86: Use new cache mode type in drivers/video/fbdev/gbefb.c
x86: Use new cache mode type in drivers/video/fbdev/vermilion
x86: Use new cache mode type in arch/x86/pci
x86: Use new cache mode type in arch/x86/mm/init_64.c
x86: Use new cache mode type in asm/pgtable.h
x86: Use new cache mode type in mm/iomap_32.c
x86: Use new cache mode type in track_pfn_remap() and
track_pfn_insert()
x86: Use new cache mode type in setting page attributes
x86: Use new cache mode type in mm/ioremap.c
x86: Use new cache mode type in memtype related functions
x86: Clean up pgtable_types.h
x86: Support PAT bit in pagetable dump for lower levels
x86: Respect PAT bit when copying pte values between large and normal
pages
x86: Enable PAT to use cache mode translation tables
xen: Support Xen pv-domains using PAT
arch/x86/include/asm/cacheflush.h | 38 ++++--- arch/x86/include/asm/fb.h | 6 +- arch/x86/include/asm/io.h | 2 +- arch/x86/include/asm/pat.h | 7 +- arch/x86/include/asm/pgtable.h | 19 ++-- arch/x86/include/asm/pgtable_types.h | 96 ++++++++++++---- arch/x86/mm/dump_pagetables.c | 24 ++-- arch/x86/mm/init.c | 37 +++++++ arch/x86/mm/init_64.c | 9 +- arch/x86/mm/iomap_32.c | 12 +- arch/x86/mm/ioremap.c | 63 ++++++----- arch/x86/mm/mm_internal.h | 2 + arch/x86/mm/pageattr.c | 84 ++++++++------ arch/x86/mm/pat.c | 176 +++++++++++++++++++----------- arch/x86/mm/pat_internal.h | 22 ++-- arch/x86/mm/pat_rbtree.c | 8 +- arch/x86/pci/i386.c | 4 +- arch/x86/xen/enlighten.c | 25 ++--- arch/x86/xen/mmu.c | 47 +------- arch/x86/xen/xen-ops.h | 1 - drivers/video/fbdev/gbefb.c | 3 +- drivers/video/fbdev/vermilion/vermilion.c | 6 +- 22 files changed, 412 insertions(+), 279 deletions(-)
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 1/17 x86: Make page cache mode a real type
At the moment there are a lot of places that handle setting or getting the page cache mode by treating the pgprot bits equal to the cache mode. This is only true because there are a lot of assumptions about the setup of the PAT MSR. Otherwise the cache type needs to get translated into pgprot bits and vice versa.
This patch tries to prepare for that by introducing a separate type for the cache mode and adding functions to translate between those and pgprot values.
To avoid too much performance penalty the translation between cache mode and pgprot values is done via tables which contain the relevant information. Write-back cache mode is hard-wired to be 0, all other modes are configurable via those tables. For large pages there are translation functions as the PAT bit is located at different positions in the ptes of 4k and large pages.
--- arch/x86/include/asm/pgtable_types.h | 73 +++++++++++++++++++++++++++++++++++- arch/x86/mm/init.c | 29 ++++++++++++++ 2 files changed, 101 insertions(+), 1 deletion(-)
arch/x86/include/asm/pgtable_types.h +72 -1
@@ -128,12 +128,34 @@ _PAGE_SOFT_DIRTY | _PAGE_NUMA)#define _HPAGE_CHG_MASK (_PAGE_CHG_MASK | _PAGE_PSE | _PAGE_NUMA) -#define _PAGE_CACHE_MASK (_PAGE_PCD | _PAGE_PWT)#define _PAGE_CACHE_WB (0)#define _PAGE_CACHE_WC (_PAGE_PWT)#define _PAGE_CACHE_UC_MINUS (_PAGE_PCD)#define _PAGE_CACHE_UC (_PAGE_PCD | _PAGE_PWT) +/* +* The cache modes defined here are used to translate between pure SW usage +* and the HW defined cache mode bits and/or PAT entries. +* +* The resulting bits for PWT, PCD and PAT should be chosen in a way +* to have the WB mode at index 0 (all bits clear). This is the default +* right now and likely would break too much if changed. +*/ +#ifndef __ASSEMBLY__ +enum page_cache_mode { + _PAGE_CACHE_MODE_WB =0 , + _PAGE_CACHE_MODE_WC =1 , + _PAGE_CACHE_MODE_UC_MINUS =2 , + _PAGE_CACHE_MODE_UC =3 , + _PAGE_CACHE_MODE_WT =4 , + _PAGE_CACHE_MODE_WP =5 , + _PAGE_CACHE_MODE_NUM =8 +}; +#endif + +#define _PAGE_CACHE_MASK (_PAGE_PAT | _PAGE_PCD | _PAGE_PWT) +#define _PAGE_NOCACHE (cachemode2protval(_PAGE_CACHE_MODE_UC)) +#define PAGE_NONE __pgprot(_PAGE_PROTNONE | _PAGE_ACCESSED)#define PAGE_SHARED __pgprot(_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | \ _PAGE_ACCESSED | _PAGE_NX)
@@ -341,6 +363,55 @@ static inline pmdval_t pmdnuma_flags(pmd_t pmd)#define pgprot_val (x) ((x).pgprot)#define __pgprot (x) ((pgprot_t) { (x) } ) +extern uint16_t __cachemode2pte_tbl[_PAGE_CACHE_MODE_NUM]; +extern uint8_t __pte2cachemode_tbl[8 ]; + +#define __pte2cm_idx (cb) \ + ((((cb) >> (_PAGE_BIT_PAT - 2)) & 4) | \ + (((cb) >> (_PAGE_BIT_PCD - 1)) & 2) | \ + (((cb) >> _PAGE_BIT_PWT) & 1)) + +static inline unsigned long cachemode2protval (enum page_cache_mode pcm) +{ +if (likely (pcm ==0 )) +return 0 ; +return __cachemode2pte_tbl[pcm]; +} +static inline pgprot_t cachemode2pgprot (enum page_cache_mode pcm) +{ +return __pgprot (cachemode2protval (pcm)); +} +static inline enum page_cache_mode pgprot2cachemode (pgprot_t pgprot) +{ +unsigned long masked; + + masked =pgprot_val (pgprot) & _PAGE_CACHE_MASK; +if (likely (masked ==0 )) +return 0 ; +return __pte2cachemode_tbl[__pte2cm_idx (masked)]; +} +static inline pgprot_t pgprot_4k_2_large (pgprot_t pgprot) +{ +pgprot_t new; +unsigned long val; + + val =pgprot_val (pgprot); +pgprot_val (new) = (val & ~(_PAGE_PAT | _PAGE_PAT_LARGE)) | + ((val & _PAGE_PAT) << (_PAGE_BIT_PAT_LARGE - _PAGE_BIT_PAT)); +return new; +} +static inline pgprot_t pgprot_large_2_4k (pgprot_t pgprot) +{ +pgprot_t new; +unsigned long val; + + val =pgprot_val (pgprot); +pgprot_val (new) = (val & ~(_PAGE_PAT | _PAGE_PAT_LARGE)) | + ((val & _PAGE_PAT_LARGE) >> + (_PAGE_BIT_PAT_LARGE - _PAGE_BIT_PAT)); +return new; +} +typedef struct page *pgtable_t ;
arch/x86/mm/init.c +29 -0
@@ -27,6 +27,35 @@#include "mm_internal.h" +/* +* Tables translating between page_cache_type_t and pte encoding. +* Minimal supported modes are defined statically, modified if more supported +* cache modes are available. +* Index into __cachemode2pte_tbl is the cachemode. +* Index into __pte2cachemode_tbl are the caching attribute bits of the pte +* (_PAGE_PWT, _PAGE_PCD, _PAGE_PAT) at index bit positions 0, 1, 2. +*/ +uint16_t __cachemode2pte_tbl[_PAGE_CACHE_MODE_NUM] = { + [_PAGE_CACHE_MODE_WB] =0 , + [_PAGE_CACHE_MODE_WC] = _PAGE_PWT, + [_PAGE_CACHE_MODE_UC_MINUS] = _PAGE_PCD, + [_PAGE_CACHE_MODE_UC] = _PAGE_PCD | _PAGE_PWT, + [_PAGE_CACHE_MODE_WT] = _PAGE_PCD, + [_PAGE_CACHE_MODE_WP] = _PAGE_PCD, +}; +EXPORT_SYMBOL_GPL (__cachemode2pte_tbl); +uint8_t __pte2cachemode_tbl[8 ] = { + [__pte2cm_idx (0 )] = _PAGE_CACHE_MODE_WB, + [__pte2cm_idx (_PAGE_PWT)] = _PAGE_CACHE_MODE_WC, + [__pte2cm_idx (_PAGE_PCD)] = _PAGE_CACHE_MODE_UC_MINUS, + [__pte2cm_idx (_PAGE_PWT | _PAGE_PCD)] = _PAGE_CACHE_MODE_UC, + [__pte2cm_idx (_PAGE_PAT)] = _PAGE_CACHE_MODE_WB, + [__pte2cm_idx (_PAGE_PWT | _PAGE_PAT)] = _PAGE_CACHE_MODE_WC, + [__pte2cm_idx (_PAGE_PCD | _PAGE_PAT)] = _PAGE_CACHE_MODE_UC_MINUS, + [__pte2cm_idx (_PAGE_PWT | _PAGE_PCD | _PAGE_PAT)] = _PAGE_CACHE_MODE_UC, +}; +EXPORT_SYMBOL_GPL (__pte2cachemode_tbl); +static unsigned long __initdata pgt_buf_start;static unsigned long __initdata pgt_buf_end;static unsigned long __initdata pgt_buf_top;
Signature · 1 line
1.8.4.5
Borislav Petkov
maintainerRe: [PATCH 01/17] x86: Make page cache mode a real type
Quoted 17 lines · Juergen Gross
At the moment there are a lot of places that handle setting or getting the page cache mode by treating the pgprot bits equal to the cache mode. This is only true because there are a lot of assumptions about the setup of the PAT MSR. Otherwise the cache type needs to get translated into pgprot bits and vice versa.
This patch tries to prepare for that by introducing a separate type for the cache mode and adding functions to translate between those and pgprot values.
To avoid too much performance penalty the translation between cache mode and pgprot values is done via tables which contain the relevant information. Write-back cache mode is hard-wired to be 0, all other modes are configurable via those tables. For large pages there are translation functions as the PAT bit is located at different positions in the ptes of 4k and large pages.
Quoted 2 lines
Signed-off-by Stefan Bader <[email protected]>Signed-off-by Juergen Gross <[email protected]>
Just a clarification question: how is one to understand this attribution here? Is Stefan the original author, was he a reviewer, or? Because this SOB chain is misleading...
Signature · 5 lines
Regards/Gruss,
Boris.
Sent from a fat crate under my desk. Formatting is fine.
--Re: [PATCH 01/17] x86: Make page cache mode a real type
Quoted 18 lines · Borislav Petkov
Quoted 17 lines · Juergen Gross
At the moment there are a lot of places that handle setting or getting the page cache mode by treating the pgprot bits equal to the cache mode. This is only true because there are a lot of assumptions about the setup of the PAT MSR. Otherwise the cache type needs to get translated into pgprot bits and vice versa.
This patch tries to prepare for that by introducing a separate type for the cache mode and adding functions to translate between those and pgprot values.
To avoid too much performance penalty the translation between cache mode and pgprot values is done via tables which contain the relevant information. Write-back cache mode is hard-wired to be 0, all other modes are configurable via those tables. For large pages there are translation functions as the PAT bit is located at different positions in the ptes of 4k and large pages.
Quoted 6 lines
Quoted 2 lines
Signed-off-by Stefan Bader <[email protected]>Signed-off-by Juergen Gross <[email protected]>Just a clarification question: how is one to understand this attribution here? Is Stefan the original author, was he a reviewer, or? Because this SOB chain is misleading...
He was the first then Juergen took over. Should the SOB order be inversed?
Quoted 7 lines
Signature · 6 lines
-- Regards/Gruss, Boris. Sent from a fat crate under my desk. Formatting is fine. --
Borislav Petkov
maintainerRe: [PATCH 01/17] x86: Make page cache mode a real type
Quoted 8 lines · Konrad Rzeszutek Wilk
Quoted 6 lines
Quoted 2 lines
Signed-off-by Stefan Bader <[email protected]>Signed-off-by Juergen Gross <[email protected]>Just a clarification question: how is one to understand this attribution here? Is Stefan the original author, was he a reviewer, or? Because this SOB chain is misleading...
He was the first then Juergen took over. Should the SOB order be inversed?
Well, I'd say:
Originally-by: Stefan...
or
Based-on-patch-by: Stefan...
or something like that. For more inspiration, one could do something like that:
$ git log | grep -Eio "original.*by:" | sort | uniq original bug can be reproduced and the bug fix can be verified by: Original-by: original by or fixed by: Original-code-by: Original-coded-by: original driver was written by: Original i2c code by: Original-Idea-and-Signed-off-by: Original idea by: Original-idea-by: originally by: Originally by: Originally-by: originally made by: Originally reported by: Originally-Reported-by: originally submitted by: originally written by: Original partial fix Signed-off-by: original patch authored by: original patch by: original-patch-by: Original patch by: Original-patch-by: Original-Patch-By: Original patches by: original-screenplay-by: Original-Signed-off-by: Original-version-by: original work completed by:
Signature · 5 lines
Regards/Gruss,
Boris.
Sent from a fat crate under my desk. Formatting is fine.
--Juergen Gross
maintainerRe: [PATCH 01/17] x86: Make page cache mode a real type
Quoted 10 lines · Borislav Petkov
Quoted 8 lines · Konrad Rzeszutek Wilk
Quoted 6 lines
Quoted 2 lines
Signed-off-by Stefan Bader <[email protected]>Signed-off-by Juergen Gross <[email protected]>Just a clarification question: how is one to understand this attribution here? Is Stefan the original author, was he a reviewer, or? Because this SOB chain is misleading...
He was the first then Juergen took over. Should the SOB order be inversed?
Quoted 7 lines
Well, I'd say:
Originally-by: Stefan...
or
Based-on-patch-by: Stefan...
I think this would fit best.
Shall I respin another version modifying the SOB to above?
Juergen
Re: [PATCH 01/17] x86: Make page cache mode a real type
Quoted 15 lines · Juergen Gross
Quoted 14 lines · Borislav Petkov
Quoted 8 lines · Konrad Rzeszutek Wilk
Quoted 6 lines
Quoted 2 lines
Signed-off-by Stefan Bader <[email protected]>Signed-off-by Juergen Gross <[email protected]>Just a clarification question: how is one to understand this attribution here? Is Stefan the original author, was he a reviewer, or? Because this SOB chain is misleading...
He was the first then Juergen took over. Should the SOB order be inversed?
Well, I'd say:
Originally-by: Stefan...
Quoted 7 lines
Quoted 3 lines
or
Based-on-patch-by: Stefan...
I think this would fit best.
Shall I respin another version modifying the SOB to above?
That would be appreciated.
Thanks,
tglx
Re: [PATCH 01/17] x86: Make page cache mode a real type
In reply to Borislav PetkovQuoted 18 lines · Borislav Petkov
Quoted 17 lines · Juergen Gross
At the moment there are a lot of places that handle setting or getting the page cache mode by treating the pgprot bits equal to the cache mode. This is only true because there are a lot of assumptions about the setup of the PAT MSR. Otherwise the cache type needs to get translated into pgprot bits and vice versa.
This patch tries to prepare for that by introducing a separate type for the cache mode and adding functions to translate between those and pgprot values.
To avoid too much performance penalty the translation between cache mode and pgprot values is done via tables which contain the relevant information. Write-back cache mode is hard-wired to be 0, all other modes are configurable via those tables. For large pages there are translation functions as the PAT bit is located at different positions in the ptes of 4k and large pages.
Quoted 6 lines
Quoted 2 lines
Signed-off-by Stefan Bader <[email protected]>Signed-off-by Juergen Gross <[email protected]>Just a clarification question: how is one to understand this attribution here? Is Stefan the original author, was he a reviewer, or? Because this SOB chain is misleading...
Bah, wanted to ask that as well. Then got lost in the review and forgot....
Thanks,
tglx
Juergen Gross
maintainerPATCH 2/17 x86: Use new cache mode type in include/asm/fb.h
In reply to Juergen GrossInstead of directly using cache mode bits in the pte switch to usage of the new cache mode type.
--- arch/x86/include/asm/fb.h | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-)
arch/x86/include/asm/fb.h +5 -1
@@ -8,8 +8,12 @@static inline void fb_pgprotect (struct file *file,struct vm_area_struct *vma,unsigned long off) { +unsigned long prot; + + prot =pgprot_val (vma->vm_page_prot) & ~_PAGE_CACHE_MASK;if (boot_cpu_data.x86 >3 ) -pgprot_val (vma->vm_page_prot) |= _PAGE_PCD; +pgprot_val (vma->vm_page_prot) = + prot |cachemode2protval (_PAGE_CACHE_MODE_UC_MINUS); }extern int fb_is_primary_device (struct fb_info *info);
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 3/17 x86: Use new cache mode type in drivers/video/fbdev/gbefb.c
In reply to Juergen GrossInstead of directly using the cache mode bits in the pte switch to using the cache mode type.
--- drivers/video/fbdev/gbefb.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/video/fbdev/gbefb.c +2 -1
@@ -54,7 +54,8 @@ struct gbefb_par {#endif #endif #ifdef CONFIG_X86 -#define pgprot_fb (_prot) ((_prot) | _PAGE_PCD) +#define pgprot_fb (_prot) (((_prot) & ~_PAGE_CACHE_MASK) | \ + cachemode2protval(_PAGE_CACHE_MODE_UC_MINUS))#endif /*
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 4/17 x86: Use new cache mode type in drivers/video/fbdev/vermilion
In reply to Juergen GrossInstead of directly using the cache mode bits in the pte switch to using the cache mode type.
--- drivers/video/fbdev/vermilion/vermilion.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/video/fbdev/vermilion/vermilion.c +4 -2
@@ -1003,13 +1003,15 @@ static int vmlfb_mmap(struct fb_info *info, struct vm_area_struct *vma) struct vml_info *vinfo = container_of(info, struct vml_info, info);unsigned long offset = vma->vm_pgoff <<PAGE_SHIFT ;int ret; +unsigned long prot; ret =vmlfb_vram_offset (vinfo, offset);if (ret)return -EINVAL ; -pgprot_val (vma->vm_page_prot) |= _PAGE_PCD; -pgprot_val (vma->vm_page_prot) &= ~_PAGE_PWT; + prot =pgprot_val (vma->vm_page_prot) & ~_PAGE_CACHE_MASK; +pgprot_val (vma->vm_page_prot) = + prot |cachemode2protval (_PAGE_CACHE_MODE_UC_MINUS);return vm_iomap_memory (vma, vinfo->vram_start, vinfo->vram_contig_size);
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 5/17 x86: Use new cache mode type in arch/x86/pci
In reply to Juergen GrossInstead of directly using the cache mode bits in the pte switch to using the cache mode type.
--- arch/x86/pci/i386.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
arch/x86/pci/i386.c +2 -2
@@ -433,14 +433,14 @@ int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma,return -EINVAL ;if (pat_enabled && write_combine) - prot |= _PAGE_CACHE_WC; + prot |=cachemode2protval (_PAGE_CACHE_MODE_WC);else if (pat_enabled || boot_cpu_data.x86 >3 )/* * ioremap() and ioremap_nocache() defaults to UC MINUS for now. * To avoid attribute conflicts, request UC MINUS here * as well. */ - prot |= _PAGE_CACHE_UC_MINUS; + prot |=cachemode2protval (_PAGE_CACHE_MODE_UC_MINUS); vma->vm_page_prot =__pgprot (prot);
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 6/17 x86: Use new cache mode type in arch/x86/mm/init_64.c
In reply to Juergen GrossInstead of directly using the cache mode bits in the pte switch to using the cache mode type.
--- arch/x86/mm/init_64.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-)
arch/x86/mm/init_64.c +6 -3
@@ -338,12 +338,15 @@ pte_t * __init populate_extra_pte(unsigned long vaddr)* Create large page table mappings for a range of physical addresses. */ static void __init __init_extra_mapping (unsigned long phys,unsigned long size, -pgprot_t prot) +enum page_cache_mode cache) {pgd_t *pgd;pud_t *pud;pmd_t *pmd; +pgprot_t prot; +pgprot_val (prot) =pgprot_val (PAGE_KERNEL_LARGE ) | +pgprot_val (pgprot_4k_2_large (cachemode2pgprot (cache)));BUG_ON ((phys & ~PMD_MASK ) || (size & ~PMD_MASK ));for (; size; phys +=PMD_SIZE , size -=PMD_SIZE ) { pgd =pgd_offset_k ((unsigned long )__va (phys));
@@ -366,12 +369,12 @@ static void __init __init_extra_mapping(unsigned long phys, unsigned long size,void __init init_extra_mapping_wb (unsigned long phys,unsigned long size) { -__init_extra_mapping (phys, size,PAGE_KERNEL_LARGE ); +__init_extra_mapping (phys, size, _PAGE_CACHE_MODE_WB); }void __init init_extra_mapping_uc (unsigned long phys,unsigned long size) { -__init_extra_mapping (phys, size,PAGE_KERNEL_LARGE_NOCACHE ); +__init_extra_mapping (phys, size, _PAGE_CACHE_MODE_UC); }/*
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 7/17 x86: Use new cache mode type in asm/pgtable.h
In reply to Juergen GrossInstead of directly using the cache mode bits in the pte switch to using the cache mode type. This requires changing some callers of is_new_memtype_allowed() to be changed as well.
--- arch/x86/include/asm/pgtable.h | 19 ++++++++++--------- arch/x86/mm/ioremap.c | 3 ++- arch/x86/mm/pat.c | 8 ++++++-- 3 files changed, 18 insertions(+), 12 deletions(-)
arch/x86/include/asm/pgtable.h +10 -9
@@ -9,9 +9,10 @@/* * Macro to mark a page protection value as UC- */ -#define pgprot_noncached (prot) \ - ((boot_cpu_data.x86 > 3) \ - ? (__pgprot(pgprot_val(prot) | _PAGE_CACHE_UC_MINUS)) \ +#define pgprot_noncached (prot) \ + ((boot_cpu_data.x86 > 3) \ + ? (__pgprot(pgprot_val(prot) | \ + cachemode2protval(_PAGE_CACHE_MODE_UC_MINUS))) \ : (prot))#ifndef __ASSEMBLY__
@@ -404,8 +405,8 @@ static inline pgprot_t pgprot_modify(pgprot_t oldprot, pgprot_t newprot)#define canon_pgprot (p) __pgprot(massage_pgprot(p))static inline int is_new_memtype_allowed (u64 paddr,unsigned long size, -unsigned long flags, -unsigned long new_flags) +enum page_cache_mode pcm, +enum page_cache_mode new_pcm) {/* * PAT type is always WB for untracked ranges, so no need to check.
@@ -419,10 +420,10 @@ static inline int is_new_memtype_allowed(u64 paddr, unsigned long size,* - request is uncached, return cannot be write-back * - request is write-combine, return cannot be write-back */ -if ((flags == _PAGE_CACHE_UC_MINUS && - new_flags == _PAGE_CACHE_WB) || - (flags == _PAGE_CACHE_WC && - new_flags == _PAGE_CACHE_WB)) { +if ((pcm == _PAGE_CACHE_MODE_UC_MINUS && + new_pcm == _PAGE_CACHE_MODE_WB) || + (pcm == _PAGE_CACHE_MODE_WC && + new_pcm == _PAGE_CACHE_MODE_WB)) {return 0 ; }
arch/x86/mm/ioremap.c +2 -1
@@ -142,7 +142,8 @@ static void __iomem *__ioremap_caller(resource_size_t phys_addr,if (prot_val != new_prot_val) {if (!is_new_memtype_allowed (phys_addr, size, - prot_val, new_prot_val)) { +pgprot2cachemode (__pgprot (prot_val)), +pgprot2cachemode (__pgprot (new_prot_val)))) { printk(KERN_ERR "ioremap error for 0x%llx-0x%llx, requested 0x%lx, got 0x%lx\n", (unsigned long long)phys_addr,
arch/x86/mm/pat.c +6 -2
@@ -455,7 +455,9 @@ int io_reserve_memtype(resource_size_t start, resource_size_t end,if (ret) goto out_err; -if (!is_new_memtype_allowed (start, size, req_type, new_type)) +if (!is_new_memtype_allowed (start, size, +pgprot2cachemode (__pgprot (req_type)), +pgprot2cachemode (__pgprot (new_type)))) goto out_free;if (kernel_map_sync_memtype (start, size, new_type) <0 )
@@ -630,7 +632,9 @@ static int reserve_pfn_range(u64 paddr, unsigned long size, pgprot_t *vma_prot,if (flags != want_flags) {if (strict_prot || - !is_new_memtype_allowed (paddr, size, want_flags, flags)) { + !is_new_memtype_allowed (paddr, size, +pgprot2cachemode (__pgprot (want_flags)), +pgprot2cachemode (__pgprot (flags)))) {free_memtype (paddr, paddr + size); printk(KERN_ERR "%s:%d map pfn expected mapping type %s" " for [mem %#010Lx-%#010Lx], got %s\n",
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 8/17 x86: Use new cache mode type in mm/iomap_32.c
In reply to Juergen GrossInstead of directly using the cache mode bits in the pte switch to using the cache mode type. This requires to change io_reserve_memtype() as well.
--- arch/x86/include/asm/pat.h | 2 +- arch/x86/mm/iomap_32.c | 12 +++++++----- arch/x86/mm/pat.c | 18 ++++++++++-------- 3 files changed, 18 insertions(+), 14 deletions(-)
arch/x86/include/asm/pat.h +1 -1
@@ -20,7 +20,7 @@ extern int kernel_map_sync_memtype(u64 base, unsigned long size, unsigned long flag);int io_reserve_memtype (resource_size_t start,resource_size_t end, -unsigned long *type); +enum page_cache_mode *pcm);void io_free_memtype (resource_size_t start,resource_size_t end);
arch/x86/mm/iomap_32.c +7 -5
@@ -33,17 +33,17 @@ static int is_io_mapping_possible(resource_size_t base, unsigned long size)int iomap_create_wc (resource_size_t base,unsigned long size,pgprot_t *prot) { -unsigned long flag = _PAGE_CACHE_WC; +enum page_cache_mode pcm = _PAGE_CACHE_MODE_WC;int ret;if (!is_io_mapping_possible (base, size))return -EINVAL ; - ret =io_reserve_memtype (base, base + size, &flag); + ret =io_reserve_memtype (base, base + size, &pcm);if (ret)return ret; - *prot =__pgprot (__PAGE_KERNEL | flag); + *prot =__pgprot (__PAGE_KERNEL |cachemode2protval (pcm));return 0 ; }EXPORT_SYMBOL_GPL (iomap_create_wc);
@@ -82,8 +82,10 @@ iomap_atomic_prot_pfn(unsigned long pfn, pgprot_t prot)* MTRR is UC or WC. UC_MINUS gets the real intention, of the * user, which is "WC if the MTRR is WC, UC if you can't do that." */ -if (!pat_enabled &&pgprot_val (prot) ==pgprot_val (PAGE_KERNEL_WC )) - prot =PAGE_KERNEL_UC_MINUS ; +if (!pat_enabled &&pgprot_val (prot) == + (__PAGE_KERNEL |cachemode2protval (_PAGE_CACHE_MODE_WC))) + prot =__pgprot (__PAGE_KERNEL | +cachemode2protval (_PAGE_CACHE_MODE_UC_MINUS));return (void __force __iomem *)kmap_atomic_prot_pfn (pfn, prot); }
arch/x86/mm/pat.c +10 -8
@@ -442,25 +442,27 @@ static unsigned long lookup_memtype(u64 paddr)* On failure, returns non-zero */ int io_reserve_memtype (resource_size_t start,resource_size_t end, -unsigned long *type) +enum page_cache_mode *type) {resource_size_t size = end - start; -unsigned long req_type = *type; -unsigned long new_type; +enum page_cache_mode req_type = *type; +enum page_cache_mode new_type; +unsigned long new_prot;int ret;WARN_ON_ONCE (iomem_map_sanity_check (start, size)); - ret =reserve_memtype (start, end, req_type, &new_type); + ret =reserve_memtype (start, end,cachemode2protval (req_type), + &new_prot);if (ret) goto out_err; -if (!is_new_memtype_allowed (start, size, -pgprot2cachemode (__pgprot (req_type)), -pgprot2cachemode (__pgprot (new_type)))) + new_type =pgprot2cachemode (__pgprot (new_prot)); + +if (!is_new_memtype_allowed (start, size, req_type, new_type)) goto out_free; -if (kernel_map_sync_memtype (start, size, new_type) <0 ) +if (kernel_map_sync_memtype (start, size, new_prot) <0 ) goto out_free; *type = new_type;
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 9/17 x86: Use new cache mode type in track_pfn_remap() and track_pfn_insert()
In reply to Juergen GrossInstead of directly using the cache mode bits in the pte switch to using the cache mode type. As those are the main callers of lookup_memtype(), change this as well.
--- arch/x86/mm/pat.c | 32 ++++++++++++++++---------------- 1 file changed, 16 insertions(+), 16 deletions(-)
arch/x86/mm/pat.c +16 -16
@@ -394,12 +394,12 @@ int free_memtype(u64 start, u64 end)* * Only to be called when PAT is enabled * -* Returns _PAGE_CACHE_WB, _PAGE_CACHE_WC, _PAGE_CACHE_UC_MINUS or -* _PAGE_CACHE_UC +* Returns _PAGE_CACHE_MODE_WB, _PAGE_CACHE_MODE_WC, _PAGE_CACHE_MODE_UC_MINUS +* or _PAGE_CACHE_MODE_UC */ -static unsigned long lookup_memtype (u64 paddr) +static enum page_cache_mode lookup_memtype (u64 paddr) { -int rettype = _PAGE_CACHE_WB; +enum page_cache_mode rettype = _PAGE_CACHE_MODE_WB;struct memtype *entry;if (x86_platform.is_untracked_pat_range (paddr, paddr +PAGE_SIZE ))
@@ -408,13 +408,13 @@ static unsigned long lookup_memtype(u64 paddr)if (pat_pagerange_is_ram (paddr, paddr +PAGE_SIZE )) {struct page *page; page =pfn_to_page (paddr >>PAGE_SHIFT ); - rettype =get_page_memtype (page); + rettype =pgprot2cachemode (__pgprot (get_page_memtype (page)));/* * -1 from get_page_memtype() implies RAM page is in its * default state and not reserved, and hence of type WB */ if (rettype ==-1 ) - rettype = _PAGE_CACHE_WB; + rettype = _PAGE_CACHE_MODE_WB;return rettype; }
@@ -423,9 +423,9 @@ static unsigned long lookup_memtype(u64 paddr) entry =rbt_memtype_lookup (paddr);if (entry !=NULL ) - rettype = entry->type; + rettype =pgprot2cachemode (__pgprot (entry->type));else - rettype = _PAGE_CACHE_UC_MINUS; + rettype = _PAGE_CACHE_MODE_UC_MINUS;spin_unlock (&memtype_lock);return rettype;
@@ -613,7 +613,7 @@ static int reserve_pfn_range(u64 paddr, unsigned long size, pgprot_t *vma_prot,if (!pat_enabled)return 0 ; - flags =lookup_memtype (paddr); + flags =cachemode2protval (lookup_memtype (paddr));if (want_flags != flags) { printk(KERN_WARNING "%s:%d map pfn RAM range req %s for [mem %#010Lx-%#010Lx], got %s\n", current->comm, current->pid,
@@ -715,7 +715,7 @@ int track_pfn_remap(struct vm_area_struct *vma, pgprot_t *prot, unsigned long pfn, unsigned long addr, unsigned long size) { resource_size_t paddr = (resource_size_t)pfn << PAGE_SHIFT; -unsigned long flags; +enum page_cache_mode pcm;/* reserve the whole chunk starting from paddr */ if (addr == vma->vm_start && size == (vma->vm_end - vma->vm_start)) {
@@ -734,18 +734,18 @@ int track_pfn_remap(struct vm_area_struct *vma, pgprot_t *prot,* For anything smaller than the vma size we set prot based on the * lookup. */ - flags =lookup_memtype (paddr); + pcm =lookup_memtype (paddr);/* Check memtype for the remaining pages */ while (size >PAGE_SIZE ) { size -=PAGE_SIZE ; paddr +=PAGE_SIZE ; -if (flags !=lookup_memtype (paddr)) +if (pcm !=lookup_memtype (paddr))return -EINVAL ; } *prot =__pgprot ((pgprot_val (vma->vm_page_prot) & (~_PAGE_CACHE_MASK)) | - flags); +cachemode2protval (pcm));return 0 ; }
@@ -753,15 +753,15 @@ int track_pfn_remap(struct vm_area_struct *vma, pgprot_t *prot,int track_pfn_insert (struct vm_area_struct *vma,pgprot_t *prot,unsigned long pfn) { -unsigned long flags; +enum page_cache_mode pcm;if (!pat_enabled)return 0 ;/* Set prot based on lookup */ - flags =lookup_memtype ((resource_size_t )pfn <<PAGE_SHIFT ); + pcm =lookup_memtype ((resource_size_t )pfn <<PAGE_SHIFT ); *prot =__pgprot ((pgprot_val (vma->vm_page_prot) & (~_PAGE_CACHE_MASK)) | - flags); +cachemode2protval (pcm));return 0 ; }
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 10/17 x86: Use new cache mode type in setting page attributes
In reply to Juergen GrossInstead of directly using the cache mode bits in the pte switch to using the cache mode type in the functiuons for modifying page attributes.
--- arch/x86/mm/pageattr.c | 54 +++++++++++++++++++++++++++----------------------- 1 file changed, 29 insertions(+), 25 deletions(-)
arch/x86/mm/pageattr.c +29 -25
@@ -1304,12 +1304,6 @@ static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias)return 0 ; } -static inline int cache_attr (pgprot_t attr) -{ -return pgprot_val (attr) & - (_PAGE_PAT | _PAGE_PAT_LARGE | _PAGE_PWT | _PAGE_PCD); -} - static int change_page_attr_set_clr(unsigned long *addr, int numpages, pgprot_t mask_set, pgprot_t mask_clr, int force_split, int in_flag,
@@ -1390,7 +1384,7 @@ static int change_page_attr_set_clr(unsigned long *addr, int numpages,* No need to flush, when we did not set any of the caching * attributes: */ - cache =cache_attr (mask_set); + cache = !!pgprot2cachemode (mask_set);/* * On success we use CLFLUSH, when the CPU supports it to
@@ -1445,7 +1439,8 @@ int _set_memory_uc(unsigned long addr, int numpages)* for now UC MINUS. see comments in ioremap_nocache() */ return change_page_attr_set (&addr, numpages, -__pgprot (_PAGE_CACHE_UC_MINUS),0 ); +cachemode2pgprot (_PAGE_CACHE_MODE_UC_MINUS), +0 ); }int set_memory_uc (unsigned long addr,int numpages)
@@ -1456,7 +1451,7 @@ int set_memory_uc(unsigned long addr, int numpages)* for now UC MINUS. see comments in ioremap_nocache() */ ret =reserve_memtype (__pa (addr),__pa (addr) + numpages *PAGE_SIZE , - _PAGE_CACHE_UC_MINUS,NULL ); + _PAGE_CACHE_UC_MINUS,NULL );if (ret) goto out_err;
@@ -1474,7 +1469,7 @@ out_err:EXPORT_SYMBOL (set_memory_uc);static int _set_memory_array (unsigned long *addr,int addrinarray, -unsigned long new_type) +enum page_cache_mode new_type) {int i, j;int ret;
@@ -1484,17 +1479,19 @@ static int _set_memory_array(unsigned long *addr, int addrinarray,*/ for (i =0 ; i < addrinarray; i++) { ret =reserve_memtype (__pa (addr[i]),__pa (addr[i]) +PAGE_SIZE , - new_type,NULL ); +cachemode2protval (new_type),NULL );if (ret) goto out_free; } ret =change_page_attr_set (addr, addrinarray, -__pgprot (_PAGE_CACHE_UC_MINUS),1 ); +cachemode2pgprot (_PAGE_CACHE_MODE_UC_MINUS), +1 ); -if (!ret && new_type == _PAGE_CACHE_WC) +if (!ret && new_type == _PAGE_CACHE_MODE_WC) ret =change_page_attr_set_clr (addr, addrinarray, -__pgprot (_PAGE_CACHE_WC), +cachemode2pgprot ( + _PAGE_CACHE_MODE_WC),__pgprot (_PAGE_CACHE_MASK),0 ,CPA_ARRAY ,NULL );if (ret)
@@ -1511,13 +1508,13 @@ out_free:int set_memory_array_uc (unsigned long *addr,int addrinarray) { -return _set_memory_array (addr, addrinarray, _PAGE_CACHE_UC_MINUS); +return _set_memory_array (addr, addrinarray, _PAGE_CACHE_MODE_UC_MINUS); }EXPORT_SYMBOL (set_memory_array_uc);int set_memory_array_wc (unsigned long *addr,int addrinarray) { -return _set_memory_array (addr, addrinarray, _PAGE_CACHE_WC); +return _set_memory_array (addr, addrinarray, _PAGE_CACHE_MODE_WC); }EXPORT_SYMBOL (set_memory_array_wc);
@@ -1527,10 +1524,12 @@ int _set_memory_wc(unsigned long addr, int numpages)unsigned long addr_copy = addr; ret =change_page_attr_set (&addr, numpages, -__pgprot (_PAGE_CACHE_UC_MINUS),0 ); +cachemode2pgprot (_PAGE_CACHE_MODE_UC_MINUS), +0 );if (!ret) { ret =change_page_attr_set_clr (&addr_copy, numpages, -__pgprot (_PAGE_CACHE_WC), +cachemode2pgprot ( + _PAGE_CACHE_MODE_WC),__pgprot (_PAGE_CACHE_MASK),0 ,0 ,NULL ); }
@@ -1564,6 +1563,7 @@ EXPORT_SYMBOL(set_memory_wc);int _set_memory_wb (unsigned long addr,int numpages) { +/* WB cache mode is hard wired to all cache attribute bits being 0 */ return change_page_attr_clear (&addr, numpages,__pgprot (_PAGE_CACHE_MASK),0 ); }
@@ -1586,6 +1586,7 @@ int set_memory_array_wb(unsigned long *addr, int addrinarray)int i;int ret; +/* WB cache mode is hard wired to all cache attribute bits being 0 */ ret =change_page_attr_clear (addr, addrinarray,__pgprot (_PAGE_CACHE_MASK),1 );if (ret)
@@ -1648,7 +1649,7 @@ int set_pages_uc(struct page *page, int numpages)EXPORT_SYMBOL (set_pages_uc);static int _set_pages_array (struct page **pages,int addrinarray, -unsigned long new_type) +enum page_cache_mode new_type) {unsigned long start;unsigned long end;
@@ -1661,15 +1662,17 @@ static int _set_pages_array(struct page **pages, int addrinarray,continue ; start =page_to_pfn (pages[i]) <<PAGE_SHIFT ; end = start +PAGE_SIZE ; -if (reserve_memtype (start, end, new_type,NULL )) +if (reserve_memtype (start, end,cachemode2protval (new_type), +NULL )) goto err_out; } ret =cpa_set_pages_array (pages, addrinarray, -__pgprot (_PAGE_CACHE_UC_MINUS)); -if (!ret && new_type == _PAGE_CACHE_WC) +cachemode2pgprot (_PAGE_CACHE_MODE_UC_MINUS)); +if (!ret && new_type == _PAGE_CACHE_MODE_WC) ret =change_page_attr_set_clr (NULL , addrinarray, -__pgprot (_PAGE_CACHE_WC), +cachemode2pgprot ( + _PAGE_CACHE_MODE_WC),__pgprot (_PAGE_CACHE_MASK),0 ,CPA_PAGES_ARRAY , pages);if (ret)
@@ -1689,13 +1692,13 @@ err_out:int set_pages_array_uc (struct page **pages,int addrinarray) { -return _set_pages_array (pages, addrinarray, _PAGE_CACHE_UC_MINUS); +return _set_pages_array (pages, addrinarray, _PAGE_CACHE_MODE_UC_MINUS); }EXPORT_SYMBOL (set_pages_array_uc);int set_pages_array_wc (struct page **pages,int addrinarray) { -return _set_pages_array (pages, addrinarray, _PAGE_CACHE_WC); +return _set_pages_array (pages, addrinarray, _PAGE_CACHE_MODE_WC); }EXPORT_SYMBOL (set_pages_array_wc);
@@ -1714,6 +1717,7 @@ int set_pages_array_wb(struct page **pages, int addrinarray)unsigned long end;int i; +/* WB cache mode is hard wired to all cache attribute bits being 0 */ retval =cpa_clear_pages_array (pages, addrinarray,__pgprot (_PAGE_CACHE_MASK));if (retval)
Signature · 1 line
1.8.4.5
Re: [PATCH 10/17] x86: Use new cache mode type in setting page attributes
Quoted 16 lines · Juergen Gross
arch/x86/mm/pageattr.c +0
-6@@ -1304,12 +1304,6 @@ static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias)return 0 ; } -static inline int cache_attr (pgprot_t attr) -{ -return pgprot_val (attr) & - (_PAGE_PAT | _PAGE_PAT_LARGE | _PAGE_PWT | _PAGE_PCD); -} - static int change_page_attr_set_clr(unsigned long *addr, int numpages, pgprot_t mask_set, pgprot_t mask_clr, int force_split, int in_flag,
arch/x86/mm/pageattr.c static int change_page_attr_set_clr(unsigned long *addr, int numpages, · line 1387
Quoted 5 lines
arch/x86/mm/pageattr.c +1
-1@@ -1390,7 +1384,7 @@ static int change_page_attr_set_clr(unsigned long *addr, int numpages,* No need to flush, when we did not set any of the caching * attributes: */ - cache =cache_attr (mask_set); + cache = !!pgprot2cachemode (mask_set);
So this loses _PAGE_PAT_LARGE, right ?
arch/x86/mm/pageattr.c int set_memory_uc(unsigned long addr, int numpages) · line 1454
Quoted 7 lines
arch/x86/mm/pageattr.c +1
-1int set_memory_uc (unsigned long addr,int numpages)@@ -1456,7 +1451,7 @@ int set_memory_uc(unsigned long addr, int numpages)* for now UC MINUS. see comments in ioremap_nocache() */ ret =reserve_memtype (__pa (addr),__pa (addr) + numpages *PAGE_SIZE , - _PAGE_CACHE_UC_MINUS,NULL ); + _PAGE_CACHE_UC_MINUS,NULL );
That should be in the patch which added the _PAGE_CACHE_UC_MINUS
arch/x86/mm/pageattr.c EXPORT_SYMBOL(set_memory_wc); · line 1566
Quoted 3 lines
arch/x86/mm/pageattr.c +1
-0int _set_memory_wb (unsigned long addr,int numpages) { +/* WB cache mode is hard wired to all cache attribute bits being 0 */
I like the comment, but shouldn't we compile time check that assumption somewhere?
arch/x86/mm/pageattr.c EXPORT_SYMBOL(set_memory_wc); · line 1568
Quoted 2 lines
arch/x86/mm/pageattr.c +0
-0return change_page_attr_clear (&addr, numpages,__pgprot (_PAGE_CACHE_MASK),0 );
Thanks,
tglx
Juergen Gross
maintainerRe: [PATCH 10/17] x86: Use new cache mode type in setting page attributes
Quoted 16 lines · Thomas Gleixner
Quoted 15 lines · Juergen Gross
arch/x86/mm/pageattr.c +0
-6@@ -1304,12 +1304,6 @@ static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias)return 0 ; } -static inline int cache_attr (pgprot_t attr) -{ -return pgprot_val (attr) & - (_PAGE_PAT | _PAGE_PAT_LARGE | _PAGE_PWT | _PAGE_PCD); -} - static int change_page_attr_set_clr(unsigned long *addr, int numpages, pgprot_t mask_set, pgprot_t mask_clr, int force_split, int in_flag,
Quoted 8 lines
arch/x86/mm/pageattr.c static int change_page_attr_set_clr(unsigned long *addr, int numpages, · line 1387
Quoted 6 lines
arch/x86/mm/pageattr.c +1
-1@@ -1390,7 +1384,7 @@ static int change_page_attr_set_clr(unsigned long *addr, int numpages,* No need to flush, when we did not set any of the caching * attributes: */ - cache =cache_attr (mask_set); + cache = !!pgprot2cachemode (mask_set);So this loses
_PAGE_PAT_LARGE, right ?
change_page_attr_set_clr() is never called with _PAGE_PAT_LARGE set in mask, so this is no problem.
BTW: correct handling of the PAT bit for large pages is added in patch 15. There have been places in the kernel respecting _PAGE_PAT_LARGE, but handling has never been complete up to now.
Quoted 11 lines
arch/x86/mm/pageattr.c int set_memory_uc(unsigned long addr, int numpages) · line 1454
Quoted 8 lines
arch/x86/mm/pageattr.c +1
-1int set_memory_uc (unsigned long addr,int numpages)@@ -1456,7 +1451,7 @@ int set_memory_uc(unsigned long addr, int numpages)* for now UC MINUS. see comments in ioremap_nocache() */ ret =reserve_memtype (__pa (addr),__pa (addr) + numpages *PAGE_SIZE , - _PAGE_CACHE_UC_MINUS,NULL ); + _PAGE_CACHE_UC_MINUS,NULL );That should be in the patch which added the
_PAGE_CACHE_UC_MINUS
Quoted 6 lines
arch/x86/mm/pageattr.c EXPORT_SYMBOL(set_memory_wc); · line 1566
Quoted 3 lines
arch/x86/mm/pageattr.c +1
-0int _set_memory_wb (unsigned long addr,int numpages) { +/* WB cache mode is hard wired to all cache attribute bits being 0 */ I like the comment, but shouldn't we compile time check that assumption somewhere?
There is a comment in patch 1 where the page_cache_mode enum is set up. The translation functions between page_cache_mode and protection values have a special check for "0" built in. Isn't this enough?
BTW: How would you check this assumption at compile time? The translation between WB cache mode and protection values is done only dynamically...
Thanks for the review,
Juergen
Re: [PATCH 10/17] x86: Use new cache mode type in setting page attributes
Quoted 28 lines · Juergen Gross
Quoted 18 lines · Thomas Gleixner
Quoted 17 lines · Juergen Gross
arch/x86/mm/pageattr.c +0
-6@@ -1304,12 +1304,6 @@ static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias) return 0; } -static inline int cache_attr (pgprot_t attr) -{ -return pgprot_val (attr) & - (_PAGE_PAT | _PAGE_PAT_LARGE | _PAGE_PWT | _PAGE_PCD); -} - static int change_page_attr_set_clr(unsigned long *addr, int numpages, pgprot_t mask_set, pgprot_t mask_clr,int
force_split, intin_flag,arch/x86/mm/pageattr.c +0
-0@@ -1390,7 +1384,7 @@ static int change_page_attr_set_clr(unsigned longQuoted 8 lines
arch/x86/mm/pageattr.c static int change_page_attr_set_clr(unsigned long *addr, int numpages, · line 1387
Quoted 6 lines
arch/x86/mm/pageattr.c +1
-1*addr, int numpages, * No need to flush, when we did not set any of the caching * attributes: */ - cache =cache_attr (mask_set); + cache = !!pgprot2cachemode (mask_set);So this loses
_PAGE_PAT_LARGE, right ?
Quoted 2 lines
change_page_attr_set_clr()is never called with_PAGE_PAT_LARGEset in mask, so this is no problem.
Ok. The we should have a separate patch which removes it first with a proper explanation and then the one which uses pgprot2cachemode().
Quoted 3 lines
BTW: correct handling of the PAT bit for large pages is added in patch 15. There have been places in the kernel respecting
_PAGE_PAT_LARGE, but handling has never been complete up to now.
Yes. I've seen that.
Quoted 7 lines
Quoted 6 lines
arch/x86/mm/pageattr.c EXPORT_SYMBOL(set_memory_wc); · line 1566
Quoted 3 lines
arch/x86/mm/pageattr.c +1
-0int _set_memory_wb (unsigned long addr,int numpages) { +/* WB cache mode is hard wired to all cache attribute bits being 0 */ I like the comment, but shouldn't we compile time check that assumption somewhere?
Quoted 7 lines
There is a comment in patch 1 where the
page_cache_modeenum is set up. The translation functions betweenpage_cache_modeand protection values have a special check for "0" built in. Isn't this enough?BTW: How would you check this assumption at compile time? The translation between WB cache mode and protection values is done only dynamically...
Indeed.
Thanks,
tglx
Juergen Gross
maintainerPATCH 11/17 x86: Use new cache mode type in mm/ioremap.c
In reply to Juergen GrossInstead of directly using the cache mode bits in the pte switch to using the cache mode type.
--- arch/x86/include/asm/io.h | 2 +- arch/x86/include/asm/pat.h | 2 +- arch/x86/mm/ioremap.c | 65 +++++++++++++++++++++++++--------------------- arch/x86/mm/pat.c | 12 +++++---- 4 files changed, 44 insertions(+), 37 deletions(-)
arch/x86/include/asm/io.h +1 -1
@@ -314,7 +314,7 @@ extern void *xlate_dev_mem_ptr(unsigned long phys);extern void unxlate_dev_mem_ptr (unsigned long phys,void *addr);extern int ioremap_change_attr (unsigned long vaddr,unsigned long size, -unsigned long prot_val); +enum page_cache_mode pcm);extern void __iomem *ioremap_wc (resource_size_t offset,unsigned long size);extern bool is_early_ioremap_ptep (pte_t *ptep);
arch/x86/include/asm/pat.h +1 -1
@@ -17,7 +17,7 @@ extern int reserve_memtype(u64 start, u64 end,extern int free_memtype (u64 start,u64 end);extern int kernel_map_sync_memtype (u64 base,unsigned long size, -unsigned long flag); +enum page_cache_mode pcm);int io_reserve_memtype (resource_size_t start,resource_size_t end,enum page_cache_mode *pcm);
arch/x86/mm/ioremap.c +35 -30
149 lines not on this page · raw message
arch/x86/mm/pat.c +7 -5
@@ -462,7 +462,7 @@ int io_reserve_memtype(resource_size_t start, resource_size_t end,if (!is_new_memtype_allowed (start, size, req_type, new_type)) goto out_free; -if (kernel_map_sync_memtype (start, size, new_prot) <0 ) +if (kernel_map_sync_memtype (start, size, new_type) <0 ) goto out_free; *type = new_type;
@@ -560,7 +560,8 @@ int phys_mem_access_prot_allowed(struct file *file, unsigned long pfn,* Change the memory type for the physial address range in kernel identity * mapping space if that range is a part of identity map. */ -int kernel_map_sync_memtype (u64 base,unsigned long size,unsigned long flags) +int kernel_map_sync_memtype (u64 base,unsigned long size, +enum page_cache_mode pcm) {unsigned long id_sz;
@@ -578,11 +579,11 @@ int kernel_map_sync_memtype(u64 base, unsigned long size, unsigned long flags) __pa(high_memory) - base : size; -if (ioremap_change_attr ((unsigned long )__va (base), id_sz, flags) <0 ) { +if (ioremap_change_attr ((unsigned long )__va (base), id_sz, pcm) <0 ) {printk (KERN_INFO "%s:%d ioremap_change_attr failed %s " "for [mem %#010Lx-%#010Lx]\n" , current->comm, current->pid, -cattr_name (flags), +cattr_name (cachemode2protval (pcm)), base, (unsigned long long )(base + size-1 ));return -EINVAL ; }
@@ -656,7 +657,8 @@ static int reserve_pfn_range(u64 paddr, unsigned long size, pgprot_t *vma_prot, flags); } -if (kernel_map_sync_memtype (paddr, size, flags) <0 ) { +if (kernel_map_sync_memtype (paddr, size, +pgprot2cachemode (__pgprot (flags))) <0 ) {free_memtype (paddr, paddr + size);return -EINVAL ; }
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 12/17 x86: Use new cache mode type in memtype related functions
In reply to Juergen GrossInstead of directly using the cache mode bits in the pte switch to using the cache mode type.
--- arch/x86/include/asm/cacheflush.h | 38 ++++++++------ arch/x86/include/asm/pat.h | 2 +- arch/x86/mm/ioremap.c | 5 +- arch/x86/mm/pageattr.c | 9 ++-- arch/x86/mm/pat.c | 102 ++++++++++++++++++-------------------- arch/x86/mm/pat_internal.h | 22 ++++---- arch/x86/mm/pat_rbtree.c | 8 +-- 7 files changed, 96 insertions(+), 90 deletions(-)
arch/x86/include/asm/cacheflush.h +24 -14
@@ -9,10 +9,10 @@/* * X86 PAT uses page flags WC and Uncached together to keep track of * memory type of pages that have backing page struct. X86 PAT supports 3 -* different memory types, _PAGE_CACHE_WB, _PAGE_CACHE_WC and -* _PAGE_CACHE_UC_MINUS and fourth state where page's memory type has not +* different memory types, _PAGE_CACHE_MODE_WB, _PAGE_CACHE_MODE_WC and +* _PAGE_CACHE_MODE_UC_MINUS and fourth state where page's memory type has not * been changed from its default (value of -1 used to denote this). -* Note we do not support _PAGE_CACHE_UC here. +* Note we do not support _PAGE_CACHE_MODE_UC here. */ #define _PGMT_DEFAULT 0
@@ -22,36 +22,40 @@#define _PGMT_MASK (1UL << PG_uncached | 1UL << PG_arch_1)#define _PGMT_CLEAR_MASK (~_PGMT_MASK) -static inline unsigned long get_page_memtype (struct page *pg) +static inline enum page_cache_mode get_page_memtype (struct page *pg) {unsigned long pg_flags = pg->flags & _PGMT_MASK;if (pg_flags == _PGMT_DEFAULT)return -1 ;else if (pg_flags == _PGMT_WC) -return _PAGE_CACHE_WC; +return _PAGE_CACHE_MODE_WC;else if (pg_flags == _PGMT_UC_MINUS) -return _PAGE_CACHE_UC_MINUS; +return _PAGE_CACHE_MODE_UC_MINUS;else -return _PAGE_CACHE_WB; +return _PAGE_CACHE_MODE_WB; } -static inline void set_page_memtype (struct page *pg,unsigned long memtype) +static inline void set_page_memtype (struct page *pg, +enum page_cache_mode memtype) { -unsigned long memtype_flags = _PGMT_DEFAULT; +unsigned long memtype_flags;unsigned long old_flags;unsigned long new_flags;switch (memtype) { -case _PAGE_CACHE_WC: +case _PAGE_CACHE_MODE_WC: memtype_flags = _PGMT_WC;break ; -case _PAGE_CACHE_UC_MINUS: +case _PAGE_CACHE_MODE_UC_MINUS: memtype_flags = _PGMT_UC_MINUS;break ; -case _PAGE_CACHE_WB: +case _PAGE_CACHE_MODE_WB: memtype_flags = _PGMT_WB;break ; +default : + memtype_flags = _PGMT_DEFAULT; +break ; } do {
@@ -60,8 +64,14 @@ static inline void set_page_memtype(struct page *pg, unsigned long memtype) } while (cmpxchg(&pg->flags, old_flags, new_flags) != old_flags); }#else -static inline unsigned long get_page_memtype (struct page *pg) {return -1 ; } -static inline void set_page_memtype (struct page *pg,unsigned long memtype) { } +static inline enum page_cache_mode get_page_memtype (struct page *pg) +{ +return -1 ; +} +static inline void set_page_memtype (struct page *pg, +enum page_cache_mode memtype) +{ +}#endif /*
arch/x86/include/asm/pat.h +1 -1
@@ -13,7 +13,7 @@ static const int pat_enabled;extern void pat_init (void );extern int reserve_memtype (u64 start,u64 end, -unsigned long req_type,unsigned long *ret_type); +enum page_cache_mode req_pcm,enum page_cache_mode *ret_pcm);extern int free_memtype (u64 start,u64 end); extern int kernel_map_sync_memtype(u64 base, unsigned long size,
arch/x86/mm/ioremap.c +1 -4
@@ -83,7 +83,6 @@ static void __iomem *__ioremap_caller(resource_size_t phys_addr,const unsigned long unaligned_size = size;struct vm_struct *area;enum page_cache_mode new_pcm; -unsigned long new_prot_val;pgprot_t prot;int retval;void __iomem *ret_addr;
@@ -135,14 +134,12 @@ static void __iomem *__ioremap_caller(resource_size_t phys_addr, size =PAGE_ALIGN (last_addr+1 ) - phys_addr; retval =reserve_memtype (phys_addr, (u64 )phys_addr + size, -cachemode2protval (pcm), &new_prot_val); + pcm, &new_pcm);if (retval) {printk (KERN_ERR "ioremap reserve_memtype failed %d\n" , retval);return NULL ; } - new_pcm =pgprot2cachemode (__pgprot (new_prot_val)); -if (pcm != new_pcm) {if (!is_new_memtype_allowed (phys_addr, size, pcm, new_pcm)) { printk(KERN_ERR
arch/x86/mm/pageattr.c +4 -5
@@ -1451,7 +1451,7 @@ int set_memory_uc(unsigned long addr, int numpages)* for now UC MINUS. see comments in ioremap_nocache() */ ret =reserve_memtype (__pa (addr),__pa (addr) + numpages *PAGE_SIZE , - _PAGE_CACHE_UC_MINUS,NULL ); + _PAGE_CACHE_MODE_UC_MINUS,NULL );if (ret) goto out_err;
@@ -1479,7 +1479,7 @@ static int _set_memory_array(unsigned long *addr, int addrinarray,*/ for (i =0 ; i < addrinarray; i++) { ret =reserve_memtype (__pa (addr[i]),__pa (addr[i]) +PAGE_SIZE , -cachemode2protval (new_type),NULL ); + new_type,NULL );if (ret) goto out_free; }
@@ -1544,7 +1544,7 @@ int set_memory_wc(unsigned long addr, int numpages)return set_memory_uc (addr, numpages); ret =reserve_memtype (__pa (addr),__pa (addr) + numpages *PAGE_SIZE , - _PAGE_CACHE_WC,NULL ); + _PAGE_CACHE_MODE_WC,NULL );if (ret) goto out_err;
@@ -1662,8 +1662,7 @@ static int _set_pages_array(struct page **pages, int addrinarray,continue ; start =page_to_pfn (pages[i]) <<PAGE_SHIFT ; end = start +PAGE_SIZE ; -if (reserve_memtype (start, end,cachemode2protval (new_type), -NULL )) +if (reserve_memtype (start, end, new_type,NULL )) goto err_out; }
arch/x86/mm/pat.c +49 -53
267 lines not on this page · raw message
arch/x86/mm/pat_internal.h +12 -10
@@ -10,30 +10,32 @@ struct memtype {u64 start;u64 end;u64 subtree_max_end; -unsigned long type; +enum page_cache_mode type;struct rb_node rb; }; -static inline char *cattr_name (unsigned long flags) +static inline char *cattr_name (enum page_cache_mode pcm) { -switch (flags & _PAGE_CACHE_MASK) { -case _PAGE_CACHE_UC:return "uncached" ; -case _PAGE_CACHE_UC_MINUS:return "uncached-minus" ; -case _PAGE_CACHE_WB:return "write-back" ; -case _PAGE_CACHE_WC:return "write-combining" ; -default :return "broken" ; +switch (pcm) { +case _PAGE_CACHE_MODE_UC:return "uncached" ; +case _PAGE_CACHE_MODE_UC_MINUS:return "uncached-minus" ; +case _PAGE_CACHE_MODE_WB:return "write-back" ; +case _PAGE_CACHE_MODE_WC:return "write-combining" ; +case _PAGE_CACHE_MODE_WT:return "write-through" ; +case _PAGE_CACHE_MODE_WP:return "write-protected" ; +default :return "broken" ; } }#ifdef CONFIG_X86_PAT extern int rbt_memtype_check_insert (struct memtype *new, -unsigned long *new_type); +enum page_cache_mode *new_type);extern struct memtype *rbt_memtype_erase (u64 start,u64 end);extern struct memtype *rbt_memtype_lookup (u64 addr);extern int rbt_memtype_copy_nth_element (struct memtype *out,loff_t pos);#else static inline int rbt_memtype_check_insert (struct memtype *new, -unsigned long *new_type) +enum page_cache_mode *new_type) {return 0 ; }static inline struct memtype *rbt_memtype_erase (u64 start,u64 end) {return NULL ; }
arch/x86/mm/pat_rbtree.c +5 -3
@@ -122,11 +122,12 @@ static struct memtype *memtype_rb_exact_match(struct rb_root *root,static int memtype_rb_check_conflict (struct rb_root *root,u64 start,u64 end, -unsigned long reqtype,unsigned long *newtype) +enum page_cache_mode reqtype, +enum page_cache_mode *newtype) {struct rb_node *node;struct memtype *match; -int found_type = reqtype; +enum page_cache_mode found_type = reqtype; match =memtype_rb_lowest_match (&memtype_rbroot, start, end);if (match ==NULL )
@@ -187,7 +188,8 @@ static void memtype_rb_insert(struct rb_root *root, struct memtype *newdata)rb_insert_augmented (&newdata->rb, root, &memtype_rb_augment_cb); } -int rbt_memtype_check_insert (struct memtype *new,unsigned long *ret_type) +int rbt_memtype_check_insert (struct memtype *new, +enum page_cache_mode *ret_type) {int err =0 ;
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 13/17 x86: Clean up pgtable_types.h
In reply to Juergen GrossRemove no longer used defines from pgtable_types.h as they are not used any longer.
Switch __PAGE_KERNEL_NOCACHE to use cache mode type instead of pte bits.
--- arch/x86/include/asm/pgtable_types.h | 21 +-------------------- 1 file changed, 1 insertion(+), 20 deletions(-)
arch/x86/include/asm/pgtable_types.h +1 -20
@@ -128,11 +128,6 @@ _PAGE_SOFT_DIRTY | _PAGE_NUMA)#define _HPAGE_CHG_MASK (_PAGE_CHG_MASK | _PAGE_PSE | _PAGE_NUMA) -#define _PAGE_CACHE_WB (0) -#define _PAGE_CACHE_WC (_PAGE_PWT) -#define _PAGE_CACHE_UC_MINUS (_PAGE_PCD) -#define _PAGE_CACHE_UC (_PAGE_PCD | _PAGE_PWT) -/* * The cache modes defined here are used to translate between pure SW usage * and the HW defined cache mode bits and/or PAT entries.
@@ -178,41 +173,27 @@ enum page_cache_mode {#define __PAGE_KERNEL_RO (__PAGE_KERNEL & ~_PAGE_RW)#define __PAGE_KERNEL_RX (__PAGE_KERNEL_EXEC & ~_PAGE_RW) -#define __PAGE_KERNEL_EXEC_NOCACHE (__PAGE_KERNEL_EXEC | _PAGE_PCD | _PAGE_PWT) -#define __PAGE_KERNEL_WC (__PAGE_KERNEL | _PAGE_CACHE_WC) -#define __PAGE_KERNEL_NOCACHE (__PAGE_KERNEL | _PAGE_PCD | _PAGE_PWT) -#define __PAGE_KERNEL_UC_MINUS (__PAGE_KERNEL | _PAGE_PCD) +#define __PAGE_KERNEL_NOCACHE (__PAGE_KERNEL | _PAGE_NOCACHE)#define __PAGE_KERNEL_VSYSCALL (__PAGE_KERNEL_RX | _PAGE_USER)#define __PAGE_KERNEL_VVAR (__PAGE_KERNEL_RO | _PAGE_USER) -#define __PAGE_KERNEL_VVAR_NOCACHE (__PAGE_KERNEL_VVAR | _PAGE_PCD | _PAGE_PWT)#define __PAGE_KERNEL_LARGE (__PAGE_KERNEL | _PAGE_PSE) -#define __PAGE_KERNEL_LARGE_NOCACHE (__PAGE_KERNEL | _PAGE_CACHE_UC | _PAGE_PSE)#define __PAGE_KERNEL_LARGE_EXEC (__PAGE_KERNEL_EXEC | _PAGE_PSE)#define __PAGE_KERNEL_IO (__PAGE_KERNEL)#define __PAGE_KERNEL_IO_NOCACHE (__PAGE_KERNEL_NOCACHE) -#define __PAGE_KERNEL_IO_UC_MINUS (__PAGE_KERNEL_UC_MINUS) -#define __PAGE_KERNEL_IO_WC (__PAGE_KERNEL_WC)#define PAGE_KERNEL __pgprot(__PAGE_KERNEL)#define PAGE_KERNEL_RO __pgprot(__PAGE_KERNEL_RO)#define PAGE_KERNEL_EXEC __pgprot(__PAGE_KERNEL_EXEC)#define PAGE_KERNEL_RX __pgprot(__PAGE_KERNEL_RX) -#define PAGE_KERNEL_WC __pgprot(__PAGE_KERNEL_WC)#define PAGE_KERNEL_NOCACHE __pgprot(__PAGE_KERNEL_NOCACHE) -#define PAGE_KERNEL_UC_MINUS __pgprot(__PAGE_KERNEL_UC_MINUS) -#define PAGE_KERNEL_EXEC_NOCACHE __pgprot(__PAGE_KERNEL_EXEC_NOCACHE)#define PAGE_KERNEL_LARGE __pgprot(__PAGE_KERNEL_LARGE) -#define PAGE_KERNEL_LARGE_NOCACHE __pgprot(__PAGE_KERNEL_LARGE_NOCACHE)#define PAGE_KERNEL_LARGE_EXEC __pgprot(__PAGE_KERNEL_LARGE_EXEC)#define PAGE_KERNEL_VSYSCALL __pgprot(__PAGE_KERNEL_VSYSCALL)#define PAGE_KERNEL_VVAR __pgprot(__PAGE_KERNEL_VVAR) -#define PAGE_KERNEL_VVAR_NOCACHE __pgprot(__PAGE_KERNEL_VVAR_NOCACHE)#define PAGE_KERNEL_IO __pgprot(__PAGE_KERNEL_IO)#define PAGE_KERNEL_IO_NOCACHE __pgprot(__PAGE_KERNEL_IO_NOCACHE) -#define PAGE_KERNEL_IO_UC_MINUS __pgprot(__PAGE_KERNEL_IO_UC_MINUS) -#define PAGE_KERNEL_IO_WC __pgprot(__PAGE_KERNEL_IO_WC)/* xwr */ #define __P000 PAGE_NONE
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 14/17 x86: Support PAT bit in pagetable dump for lower levels
In reply to Juergen GrossDumping page table protection bits is not correct for entries on levels 2 and 3 regarding the PAT bit, which is at a different position as on level 4.
--- arch/x86/mm/dump_pagetables.c | 24 +++++++++++------------- 1 file changed, 11 insertions(+), 13 deletions(-)
arch/x86/mm/dump_pagetables.c +11 -13
@@ -126,7 +126,7 @@ static void printk_prot(struct seq_file *m, pgprot_t prot, int level, bool dmsg)if (!pgprot_val (prot)) {/* Not present */ -pt_dump_cont_printf (m, dmsg," " ); +pt_dump_cont_printf (m, dmsg," " ); }else {if (pr & _PAGE_USER)pt_dump_cont_printf (m, dmsg,"USR " );
@@ -145,18 +145,16 @@ static void printk_prot(struct seq_file *m, pgprot_t prot, int level, bool dmsg)else pt_dump_cont_printf (m, dmsg," " ); -/* Bit 9 has a different meaning on level 3 vs 4 */ -if (level <=3 ) { -if (pr & _PAGE_PSE) -pt_dump_cont_printf (m, dmsg,"PSE " ); -else -pt_dump_cont_printf (m, dmsg," " ); - }else { -if (pr & _PAGE_PAT) -pt_dump_cont_printf (m, dmsg,"pat " ); -else -pt_dump_cont_printf (m, dmsg," " ); - } +/* Bit 7 has a different meaning on level 3 vs 4 */ +if (level <=3 && pr & _PAGE_PSE) +pt_dump_cont_printf (m, dmsg,"PSE " ); +else +pt_dump_cont_printf (m, dmsg," " ); +if ((level ==4 && pr & _PAGE_PAT) || + ((level ==3 || level ==2 ) && pr & _PAGE_PAT_LARGE)) +pt_dump_cont_printf (m, dmsg,"pat " ); +else +pt_dump_cont_printf (m, dmsg," " );if (pr & _PAGE_GLOBAL)pt_dump_cont_printf (m, dmsg,"GLB " );else
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 15/17 x86: Respect PAT bit when copying pte values between large and normal pages
In reply to Juergen GrossThe PAT bit in the ptes is not moved to the correct position when copying page protection attributes between entries of different sized pages. Translate the ptes according to their page size.
--- arch/x86/mm/pageattr.c | 33 +++++++++++++++++++++++---------- 1 file changed, 23 insertions(+), 10 deletions(-)
arch/x86/mm/pageattr.c +23 -10
@@ -485,14 +485,23 @@ try_preserve_large_page(pte_t *kpte, unsigned long address,/* * We are safe now. Check whether the new pgprot is the same: +* Convert protection attributes to 4k-format, as cpa->mask* are set +* up accordingly. */ old_pte = *kpte; - old_prot = req_prot =pte_pgprot (old_pte); + old_prot = req_prot =pgprot_large_2_4k (pte_pgprot (old_pte));pgprot_val (req_prot) &= ~pgprot_val (cpa->mask_clr);pgprot_val (req_prot) |=pgprot_val (cpa->mask_set);/* +* req_prot is in format of 4k pages. It must be converted to large +* page format: the caching mode includes the PAT bit located at +* different bit positions in the two formats. +*/ + req_prot =pgprot_4k_2_large (req_prot); + +/* * Set the PSE and GLOBAL flags only if the PRESENT flag is * set otherwise pmd_present/pmd_huge will return true even on * a non present pmd. The canon_pgprot will clear _PAGE_GLOBAL
@@ -585,13 +594,10 @@ __split_large_page(struct cpa_data *cpa, pte_t *kpte, unsigned long address,paravirt_alloc_pte (&init_mm,page_to_pfn (base)); ref_prot =pte_pgprot (pte_clrhuge (*kpte)); -/* -* If we ever want to utilize the PAT bit, we need to -* update this function to make sure it's converted from -* bit 12 to bit 7 when we cross from the 2MB level to -* the 4K level: -*/ -WARN_ON_ONCE (pgprot_val (ref_prot) & _PAGE_PAT_LARGE); + +/* promote PAT bit to correct position */ +if (level ==PG_LEVEL_2M ) + ref_prot =pgprot_large_2_4k (ref_prot);#ifdef CONFIG_X86_64 if (level ==PG_LEVEL_1G ) {
@@ -879,6 +885,7 @@ static int populate_pmd(struct cpa_data *cpa, {unsigned int cur_pages =0 ;pmd_t *pmd; +pgprot_t pmd_pgprot;/* * Not on a 2M boundary?
@@ -910,6 +917,8 @@ static int populate_pmd(struct cpa_data *cpa,if (num_pages == cur_pages)return cur_pages; + pmd_pgprot =pgprot_4k_2_large (pgprot); +while (end - start >=PMD_SIZE ) {/*
@@ -921,7 +930,8 @@ static int populate_pmd(struct cpa_data *cpa, pmd =pmd_offset (pud, start); -set_pmd (pmd,__pmd (cpa->pfn | _PAGE_PSE |massage_pgprot (pgprot))); +set_pmd (pmd,__pmd (cpa->pfn | _PAGE_PSE | +massage_pgprot (pmd_pgprot))); start +=PMD_SIZE ; cpa->pfn +=PMD_SIZE ;
@@ -949,6 +959,7 @@ static int populate_pud(struct cpa_data *cpa, unsigned long start, pgd_t *pgd,pud_t *pud;unsigned long end;int cur_pages =0 ; +pgprot_t pud_pgprot; end = start + (cpa->numpages <<PAGE_SHIFT );
@@ -986,12 +997,14 @@ static int populate_pud(struct cpa_data *cpa, unsigned long start, pgd_t *pgd,return cur_pages; pud =pud_offset (pgd, start); + pud_pgprot =pgprot_4k_2_large (pgprot);/* * Map everything starting from the Gb boundary, possibly with 1G pages */ while (end - start >=PUD_SIZE ) { -set_pud (pud,__pud (cpa->pfn | _PAGE_PSE |massage_pgprot (pgprot))); +set_pud (pud,__pud (cpa->pfn | _PAGE_PSE | +massage_pgprot (pud_pgprot))); start +=PUD_SIZE ; cpa->pfn +=PUD_SIZE ;
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 16/17 x86: Enable PAT to use cache mode translation tables
In reply to Juergen GrossUpdate the translation tables from cache mode to pgprot values according to the PAT settings. This enables changing the cache attributes of a PAT index in just one place without having to change at the users side.
With this change it is possible to use the same kernel with different PAT configurations, e.g. supporting Xen.
--- arch/x86/include/asm/pat.h | 1 + arch/x86/include/asm/pgtable_types.h | 4 +++ arch/x86/mm/init.c | 8 ++++++ arch/x86/mm/mm_internal.h | 2 ++ arch/x86/mm/pat.c | 50 ++++++++++++++++++++++++++++++++++-- 5 files changed, 63 insertions(+), 2 deletions(-)
arch/x86/include/asm/pat.h +1 -0
@@ -11,6 +11,7 @@ static const int pat_enabled;#endif extern void pat_init (void ); +void pat_init_cache_modes (void );extern int reserve_memtype (u64 start,u64 end,enum page_cache_mode req_pcm,enum page_cache_mode *ret_pcm);
arch/x86/include/asm/pgtable_types.h +4 -0
@@ -351,6 +351,10 @@ extern uint8_t __pte2cachemode_tbl[8]; ((((cb) >> (_PAGE_BIT_PAT -2 )) &4 ) | \ (((cb) >> (_PAGE_BIT_PCD -1 )) &2 ) | \ (((cb) >> _PAGE_BIT_PWT) & 1)) +#define __cm_idx2pte (i) \ + ((((i) & 4) << (_PAGE_BIT_PAT - 2)) | \ + (((i) & 2) << (_PAGE_BIT_PCD - 1)) | \ + (((i) & 1) << _PAGE_BIT_PWT))static inline unsigned long cachemode2protval (enum page_cache_mode pcm) {
arch/x86/mm/init.c +8 -0
@@ -716,3 +716,11 @@ void __init zone_sizes_init(void)free_area_init_nodes (max_zone_pfns); } +void update_cache_mode_entry (unsigned entry,enum page_cache_mode cache) +{ +/* entry 0 MUST be WB (hardwired to speed up translations) */ +BUG_ON (!entry && cache != _PAGE_CACHE_MODE_WB); + + __cachemode2pte_tbl[cache] =__cm_idx2pte (entry); + __pte2cachemode_tbl[entry] = cache; +}
arch/x86/mm/mm_internal.h +2 -0
@@ -16,4 +16,6 @@ void zone_sizes_init(void);extern int after_bootmem; +void update_cache_mode_entry (unsigned entry,enum page_cache_mode cache); +#endif /* __X86_MM_INTERNAL_H */
arch/x86/mm/pat.c +48 -2
68 lines not on this page · raw message
Signature · 1 line
1.8.4.5
Juergen Gross
maintainerPATCH 17/17 xen: Support Xen pv-domains using PAT
In reply to Juergen GrossWith the dynamical mapping between cache modes and pgprot values it is now possible to use all cache modes via the Xen hypervisor PAT settings in a pv domain.
All to be done is to read the PAT configuration MSR and set up the translation tables accordingly.
--- arch/x86/xen/enlighten.c | 25 +++++++------------------ arch/x86/xen/mmu.c | 47 +---------------------------------------------- arch/x86/xen/xen-ops.h | 1 - 3 files changed, 8 insertions(+), 65 deletions(-)
arch/x86/xen/enlighten.c +7 -18
@@ -1100,12 +1100,6 @@ static int xen_write_msr_safe(unsigned int msr, unsigned low, unsigned high)/* Fast syscall setup is all done in hypercalls, so these are all ignored. Stub them out here to stop Xen console noise. */ -break ; - -case MSR_IA32_CR_PAT : -if (smp_processor_id () ==0 ) -xen_set_pat (((u64 )high <<32 ) | low); -break ;default : ret =native_write_msr_safe (msr, low, high);
@@ -1561,10 +1555,6 @@ asmlinkage __visible void __init xen_start_kernel(void)/* Prevent unwanted bits from being set in PTEs. */ __supported_pte_mask &= ~_PAGE_GLOBAL; -#if 0 - if (!xen_initial_domain()) -#endif - __supported_pte_mask &= ~(_PAGE_PWT | _PAGE_PCD);/* * Prevent page tables from being allocated in highmem, even
@@ -1618,14 +1608,6 @@ asmlinkage __visible void __init xen_start_kernel(void)*/ acpi_numa =-1 ;#endif -#ifdef CONFIG_X86_PAT -/* -* For right now disable the PAT. We should remove this once -* git commit 8eaffa67b43e99ae581622c5133e20b0f48bcef1 -* (xen/pat: Disable PAT support for now) is reverted. -*/ - pat_enabled =0 ; -#endif /* Don't do the full vcpu_info placement stuff until we have a possible map and a non-dummy shared_info. */ per_cpu (xen_vcpu,0 ) = &HYPERVISOR_shared_info->vcpu_info[0 ];
@@ -1636,6 +1618,13 @@ asmlinkage __visible void __init xen_start_kernel(void)xen_raw_console_write ("mapping kernel into physical memory\n" );xen_setup_kernel_pagetable ((pgd_t *)xen_start_info->pt_base, xen_start_info->nr_pages); +/* +* Modify the cache mode translation tables to match Xen's PAT +* configuration. +*/ + +pat_init_cache_modes (); +/* keep using Xen gdt for now; no urgent need to change it */ #ifdef CONFIG_X86_32
arch/x86/xen/mmu.c +1 -46
61 lines not on this page · raw message
arch/x86/xen/xen-ops.h +0 -1
@@ -33,7 +33,6 @@ extern unsigned long xen_max_p2m_pfn;void xen_mm_pin_all (void );void xen_mm_unpin_all (void ); -void xen_set_pat (u64 );char *__init xen_memory_setup (void );char *xen_auto_xlated_memory_setup (void );
Signature · 1 line
1.8.4.5
Quoted 3 lines · Juergen Gross
Changes in V5:
- split up first patch as requested by Ingo Molnar and Thomas Gleixner
- add a helper function inpat_init_cache_modes()as requested by Ingo Molnar
The structure of this series looks great to me now.
Just to make sure: the series is fully bisectable, it is supposed to build and boot at every patch in the series, right?
If nobody objects then I'll start applying this series in a couple of days.
Thanks,
Ingo
Juergen Gross
maintainerQuoted 11 lines · Ingo Molnar
Quoted 3 lines · Juergen Gross
Changes in V5:
- split up first patch as requested by Ingo Molnar and Thomas Gleixner
- add a helper function inpat_init_cache_modes()as requested by Ingo MolnarThe structure of this series looks great to me now.
Just to make sure: the series is fully bisectable, it is supposed to build and boot at every patch in the series, right?
Right. I've built after each single patch and booted a test system with the new kernels.
Quoted 2 lines
If nobody objects then I'll start applying this series in a couple of days.
Great, thank you very much!
Juergen
Thomas Gleixner
Reviewed-byQuoted 8 lines · Ingo Molnar
Quoted 3 lines · Juergen Gross
Changes in V5:
- split up first patch as requested by Ingo Molnar and Thomas Gleixner
- add a helper function inpat_init_cache_modes()as requested by Ingo MolnarThe structure of this series looks great to me now.
And it actually was a pleasure to review.
Quoted 5 lines
Just to make sure: the series is fully bisectable, it is supposed to build and boot at every patch in the series, right?
If nobody objects then I'll start applying this series in a couple of days.
I did not find anything except the question about losing the large pat bit. If that's resolved its good to go.
Please add
to the whole series.
Thanks,
tglx
Juergen Gross
maintainerQuoted 11 lines · Thomas Gleixner
Quoted 8 lines · Ingo Molnar
Quoted 3 lines · Juergen Gross
Changes in V5:
- split up first patch as requested by Ingo Molnar and Thomas Gleixner
- add a helper function inpat_init_cache_modes()as requested by Ingo MolnarThe structure of this series looks great to me now.
And it actually was a pleasure to review.
You're welcome.
Quoted 7 lines
Quoted 6 lines
Just to make sure: the series is fully bisectable, it is supposed to build and boot at every patch in the series, right?
If nobody objects then I'll start applying this series in a couple of days.
Quoted 8 lines
I did not find anything except the question about losing the large pat bit. If that's resolved its good to go.
Please add
Reviewed-by Thomas Gleixner <[email protected]>to the whole series.
Thank you,
Juergen