arm64: add early_ioremap support
Subsystems (2)
PATCH 1/1 arm64: add early_ioremap support
EFI runtime support needs to make temporary short term mappings at early boot time before ioremap is available. This patch adds support for such early mappings based on "fixmap" support found in a number of other architectures. This also allows for a number of permanent mappings which may need to persist beyond paging_init(). This implementation uses the 2MiB of address space currently dedicated to the earlyprintk console registers. This allows for mapping up to 32 64K pages or 512 4K pages.
--- Documentation/arm64/memory.txt | 2 +- arch/arm64/include/asm/fixmap.h | 117 ++++++++++++++++ arch/arm64/include/asm/io.h | 6 + arch/arm64/include/asm/memory.h | 2 +- arch/arm64/kernel/early_printk.c | 8 +- arch/arm64/kernel/head.S | 9 +- arch/arm64/kernel/setup.c | 3 + arch/arm64/mm/ioremap.c | 285 +++++++++++++++++++++++++++++++++++++++ arch/arm64/mm/mmu.c | 41 ------ 9 files changed, 422 insertions(+), 51 deletions(-) create mode 100644 arch/arm64/include/asm/fixmap.h
Documentation/arm64/memory.txt +1 -1
@@ -35,7 +35,7 @@ ffffffbc00000000 ffffffbdffffffff 8GB vmemmap ffffffbe00000000 ffffffbffbbfffff ~8GB [guard, future vmmemap] -ffffffbffbc00000 ffffffbffbdfffff 2MB earlyprintk device +ffffffbffbc00000 ffffffbffbdfffff 2MB fixed mappings ffffffbffbe00000 ffffffbffbe0ffff 64KB PCI I/O space
arch/arm64/include/asm/fixmap.h new file +117 -0
117 lines not on this page · raw message
arch/arm64/include/asm/io.h +6 -0
@@ -236,6 +236,12 @@ extern void __iounmap(volatile void __iomem *addr);#define ioremap_cached (addr, size) __ioremap((addr), (size), __pgprot(PROT_NORMAL))#define iounmap __iounmap +extern void __iomem *early_ioremap (resource_size_t phys_addr, +unsigned long size); +extern void __iomem *early_memremap (resource_size_t phys_addr, +unsigned long size); +extern void early_iounmap (void __iomem *addr,unsigned long size); +#define PROT_SECT_DEFAULT (PMD_TYPE_SECT | PMD_SECT_AF)#define PROT_SECT_DEVICE_nGnRE (PROT_SECT_DEFAULT | PTE_PXN | PTE_UXN | PMD_ATTRINDX(MT_DEVICE_nGnRE))
arch/arm64/include/asm/memory.h +1 -1
@@ -43,7 +43,7 @@#define PAGE_OFFSET UL(0xffffffc000000000)#define MODULES_END (PAGE_OFFSET)#define MODULES_VADDR (MODULES_END - SZ_64M) -#define EARLYCON_IOBASE (MODULES_VADDR - SZ_4M) +#define FIXADDR_TOP (MODULES_VADDR - SZ_2M - SZ_64K)#define VA_BITS (39)#define TASK_SIZE_64 (UL(1) << VA_BITS)
arch/arm64/kernel/early_printk.c +6 -2
@@ -26,6 +26,8 @@#include <linux/amba/serial.h> #include <linux/serial_reg.h> +#include <asm/fixmap.h> +static void __iomem *early_base;static void (*printch)(char ch);
@@ -141,8 +143,10 @@ static int __init setup_early_printk(char *buf) }/* no options parsing yet */ -if (paddr) - early_base =early_io_map (paddr,EARLYCON_IOBASE ); +if (paddr) { +set_fixmap_io (FIX_EARLYCON , paddr); + early_base = (void __iomem *)fix_to_virt (FIX_EARLYCON ); + } printch = match->printch; early_console = &early_console_dev;
arch/arm64/kernel/head.S +3 -6
@@ -484,7 +484,7 @@ ENDPROC(__calc_phys_offset) * - identity mapping to enable the MMU (low address, TTBR0) * - first few MB of the kernel linear mapping to jump to once the MMU has * been enabled, including the FDT blob (TTBR1) - * - UART mapping if CONFIG_EARLY_PRINTK is enabled (TTBR1) + * - pgd entry for fixed mappings */ __create_page_tables: pgtbl x25, x26, x24 // idmap_pg_dir and swapper_pg_dir addresses
@@ -537,15 +537,12 @@ __create_page_tables: sub x6, x6, #1 // inclusive range create_block_map x0, x7, x3, x5, x6 1: -#ifdef CONFIG_EARLY_PRINTK /* - * Create the pgd entry for the UART mapping. The full mapping is done - * later based earlyprintk kernel parameter. + * Create the pgd entry for the fixed mappings. */ - ldr x5, =EARLYCON_IOBASE // UART virtual address + ldr x5, =FIXADDR_TOP // Fixed mapping virtual address add x0, x26, #2 * PAGE_SIZE // section table address create_pgd_entry x26, x0, x5, x6, x7 -#endif ret ENDPROC(__create_page_tables) .ltorg
arch/arm64/kernel/setup.c +3 -0
@@ -42,6 +42,7 @@#include <linux/of_fdt.h> #include <linux/of_platform.h> +#include <asm/fixmap.h> #include <asm/cputype.h> #include <asm/elf.h> #include <asm/cputable.h>
@@ -252,6 +253,8 @@ void __init setup_arch(char **cmdline_p) *cmdline_p = boot_command_line; +early_ioremap_init (); +parse_early_param ();arm64_memblock_init ();
arch/arm64/mm/ioremap.c +285 -0
294 lines not on this page · raw message
arch/arm64/mm/mmu.c +0 -41
47 lines not on this page · raw message
Signature · 1 line
1.8.3.1
Catalin Marinas
maintainerHi Mark,
arch/arm64/include/asm/fixmap.h line 10
Quoted 13 lines
arch/arm64/include/asm/fixmap.h new file +10
-0@@ -0,0 +1,117 @@ +/* +* fixmap.h: compile-time virtual memory allocation +* +* 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) 1998 Ingo Molnar +* +*/
I can see several architectures having very similar macros/functions in fixmap.h. It would make sense to create a generic fixmap.h holding at least the fix_to_virt and related macros, FIXADDR_START etc. with enum fixed_addresses in arch code.
Quoted 11 lines
arch/arm64/kernel/setup.c +1
-0@@ -42,6 +42,7 @@#include <linux/of_fdt.h> #include <linux/of_platform.h> +#include <asm/fixmap.h> #include <asm/cputype.h> #include <asm/elf.h> #include <asm/cputable.h>
arch/arm64/kernel/setup.c void __init setup_arch(char **cmdline_p) · line 258
Quoted 6 lines
arch/arm64/kernel/setup.c +2
-0@@ -252,6 +253,8 @@ void __init setup_arch(char **cmdline_p) *cmdline_p = boot_command_line; +early_ioremap_init (); +parse_early_param ();
Should the early_ioremap_init() call happen after parse_early_param()? Is early_ioremap_debug initialised already?
Quoted 22 lines
arch/arm64/mm/ioremap.c +9
-0@@ -25,6 +25,10 @@#include <linux/vmalloc.h> #include <linux/io.h> +#include <asm/fixmap.h> +#include <asm/tlbflush.h> +#include <asm/pgalloc.h> +static void __iomem *__ioremap_caller (phys_addr_t phys_addr,size_t size,pgprot_t prot,void *caller) {@@ -82,3 +86,284 @@ void __iounmap(volatile void __iomem *io_addr)vunmap (addr); }EXPORT_SYMBOL (__iounmap); + +static int early_ioremap_debug__initdata ; + +static int __init early_ioremap_debug_setup (char *str) +{
arch/arm64/mm/ioremap.c void __iounmap(volatile void __iomem *io_addr) · line 103
Quoted 10 lines
arch/arm64/mm/ioremap.c +10
-0+ early_ioremap_debug =1 ; + +return 0 ; +} +early_param ("early_ioremap_debug" , early_ioremap_debug_setup); + +static int after_paging_init__initdata ; +#ifndef CONFIG_ARM64_64K_PAGES +static pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)] __page_aligned_bss; +#endif
bm_pte[PTRS_PER_PTE];
arch/arm64/mm/ioremap.c void __iounmap(volatile void __iomem *io_addr) · line 107
Quoted 4 lines
arch/arm64/mm/ioremap.c +4
-0+ +static inline pmd_t *__init early_ioremap_pmd (unsigned long addr) +{ +pgd_t *pgd = &swapper_pg_dir[pgd_index (addr)];
pgd_offset_k(addr);
Quoted 4 lines
arch/arm64/mm/ioremap.c +4
-0+pud_t *pud =pud_offset (pgd, addr); +pmd_t *pmd =pmd_offset (pud, addr); + +return pmd;
arch/arm64/mm/ioremap.c void __iounmap(volatile void __iomem *io_addr) · line 121
Quoted 10 lines
arch/arm64/mm/ioremap.c +10
-0+} + +static inline pte_t *__init early_ioremap_pte (unsigned long addr) +{ +#ifdef CONFIG_ARM64_64K_PAGES +pmd_t *pmd =early_ioremap_pmd (addr); +return pte_offset_kernel (pmd, addr); +#else +return &bm_pte[pte_index (addr)]; +#endif
If we populate the pmd correctly with 4K pages (and I think we do in early_ioremap_init()), can we not just use this function without the #ifdef-else part (always pte_offset_kernel())?
Quoted 7 lines
arch/arm64/mm/ioremap.c +7
-0+static unsigned long slot_virt[FIX_BTMAPS_SLOTS ]__initdata ; + +void __init early_ioremap_init (void ) +{ +pmd_t *pmd; +int i; +
arch/arm64/mm/ioremap.c void __iounmap(volatile void __iomem *io_addr) · line 140
Quoted 10 lines
arch/arm64/mm/ioremap.c +10
-0+if (early_ioremap_debug) +pr_info ("early_ioremap_init()\n" ); + +for (i =0 ; i <FIX_BTMAPS_SLOTS ; i++) + slot_virt[i] =__fix_to_virt (FIX_BTMAP_BEGIN -NR_FIX_BTMAPS *i); + + pmd =early_ioremap_pmd (fix_to_virt (FIX_BTMAP_BEGIN )); +#ifndef CONFIG_ARM64_64K_PAGES +/* need to populate pmd for 4k pagesize only */ +memset (bm_pte,0 ,sizeof (bm_pte));
Do we need memset() here? bm_pte[] is placed in the .bss section.
arch/arm64/mm/ioremap.c void __iounmap(volatile void __iomem *io_addr) · line 151
Quoted 11 lines
arch/arm64/mm/ioremap.c +11
-0+pmd_populate_kernel (&init_mm, pmd, bm_pte); +#endif + +/* +* The boot-ioremap range spans multiple pmds, for which +* we are not prepared: +*/ +#define __FIXADDR_TOP (-PAGE_SIZE) +BUILD_BUG_ON ((__fix_to_virt (FIX_BTMAP_BEGIN ) >>PMD_SHIFT ) + != (__fix_to_virt (FIX_BTMAP_END ) >>PMD_SHIFT )); +#undef __FIXADDR_TOP
Why this #define/#undef? FIXADDR_TOP is statically defined.
Quoted 12 lines
arch/arm64/mm/ioremap.c +12
-0+void __init __set_fixmap (enum fixed_addresses idx, +phys_addr_t phys,pgprot_t flags) +{ +unsigned long addr =__fix_to_virt (idx); +pte_t *pte; + +if (idx >= __end_of_fixed_addresses) { +BUG (); +return ; + } +if (after_paging_init) { +WARN_ON (1 );
arch/arm64/mm/ioremap.c void __iounmap(volatile void __iomem *io_addr) · line 193
Quoted 10 lines
arch/arm64/mm/ioremap.c +10
-0+return ; + } + + pte =early_ioremap_pte (addr); + +if (pgprot_val (flags)) +set_pte (pte,pfn_pte (phys >>PAGE_SHIFT , flags)); +else +pte_clear (&init_mm, addr, pte); +flush_tlb_kernel_range (addr, addr+PAGE_SIZE );
Would __set_fixmap be used to change valid ptes? If not, we could keep the flush_tlb_kernel_range() call only under the 'else' block.
As I was going through the patch, I realised that the early_ioremap() looks really to the x86 implementation. Can we move it into a library to be shared between the two?
Signature · 1 line
Catalin
Quoted 12 lines · Catalin Marinas
Hi Mark,
Quoted 10 lines
arch/arm64/include/asm/fixmap.h new file +7
-0@@ -0,0 +1,117 @@ +/* +* fixmap.h: compile-time virtual memory allocation +* +* 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. +*
Quoted 8 lines
arch/arm64/include/asm/fixmap.h line 10
Quoted 3 lines
arch/arm64/include/asm/fixmap.h new file +3
-0+* Copyright (C) 1998 Ingo Molnar +* +*/ I can see several architectures having very similar macros/functions in
fixmap.h. It would make sense to create a genericfixmap.hholding at least thefix_to_virtand related macros,FIXADDR_STARTetc. with enumfixed_addressesin arch code.
Right, there are a quite a few using similar fixmap code. I actually looked at consolidating them, but I gave up so that I could get something working for EFI work in progress. But yeah, I think I will step back and have another go at it. I think I can get at a few of the architectures using common code and make it so others can be converted as desired.
Quoted 13 lines
Quoted 13 lines
arch/arm64/kernel/setup.c +1
-0@@ -42,6 +42,7 @@#include <linux/of_fdt.h> #include <linux/of_platform.h> +#include <asm/fixmap.h> #include <asm/cputype.h> #include <asm/elf.h> #include <asm/cputable.h> @@ -252,6 +253,8 @@ void __init setup_arch(char **cmdline_p)
Quoted 8 lines
arch/arm64/kernel/setup.c void __init setup_arch(char **cmdline_p) · line 258
Quoted 5 lines
arch/arm64/kernel/setup.c +2
-0*cmdline_p = boot_command_line; +early_ioremap_init (); +parse_early_param ();Should the
early_ioremap_init()call happen afterparse_early_param()? Isearly_ioremap_debuginitialised already?
No, early_ioremap_debug is not initialized which makes the debug prints in early_ioremap_init() unreachable. I'll fix that. On x86, the init comes first, but there is other code between that and parse_early_param() which uses early_ioremap(). Not the case for arm64.
Quoted 27 lines
Quoted 27 lines
arch/arm64/mm/ioremap.c +13
-0@@ -25,6 +25,10 @@#include <linux/vmalloc.h> #include <linux/io.h> +#include <asm/fixmap.h> +#include <asm/tlbflush.h> +#include <asm/pgalloc.h> +static void __iomem *__ioremap_caller (phys_addr_t phys_addr,size_t size,pgprot_t prot,void *caller) {@@ -82,3 +86,284 @@ void __iounmap(volatile void __iomem *io_addr)vunmap (addr); }EXPORT_SYMBOL (__iounmap); + +static int early_ioremap_debug__initdata ; + +static int __init early_ioremap_debug_setup (char *str) +{ + early_ioremap_debug =1 ; + +return 0 ; +}
Quoted 8 lines
arch/arm64/mm/ioremap.c void __iounmap(volatile void __iomem *io_addr) · line 103
Quoted 6 lines
arch/arm64/mm/ioremap.c +6
-0+early_param ("early_ioremap_debug" , early_ioremap_debug_setup); + +static int after_paging_init__initdata ; +#ifndef CONFIG_ARM64_64K_PAGES +static pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)] __page_aligned_bss; +#endif bm_pte[PTRS_PER_PTE];
ok
Quoted 7 lines
arch/arm64/mm/ioremap.c void __iounmap(volatile void __iomem *io_addr) · line 107
Quoted 5 lines
arch/arm64/mm/ioremap.c +4
-0+ +static inline pmd_t *__init early_ioremap_pmd (unsigned long addr) +{ +pgd_t *pgd = &swapper_pg_dir[pgd_index (addr)];
pgd_offset_k(addr);
ok
Quoted 11 lines
Quoted 11 lines
arch/arm64/mm/ioremap.c +10
-0+pud_t *pud =pud_offset (pgd, addr); +pmd_t *pmd =pmd_offset (pud, addr); + +return pmd; +} + +static inline pte_t *__init early_ioremap_pte (unsigned long addr) +{ +#ifdef CONFIG_ARM64_64K_PAGES +pmd_t *pmd =early_ioremap_pmd (addr);
Quoted 8 lines
arch/arm64/mm/ioremap.c void __iounmap(volatile void __iomem *io_addr) · line 121
Quoted 4 lines
arch/arm64/mm/ioremap.c +4
-0+return pte_offset_kernel (pmd, addr); +#else +return &bm_pte[pte_index (addr)]; +#endif If we populate the pmd correctly with 4K pages (and I think we do in
early_ioremap_init()), can we not just use this function without the #ifdef-else part (alwayspte_offset_kernel())?
Ah, yes. Nice!
Quoted 12 lines
Quoted 12 lines
arch/arm64/mm/ioremap.c +11
-0+static unsigned long slot_virt[FIX_BTMAPS_SLOTS ]__initdata ; + +void __init early_ioremap_init (void ) +{ +pmd_t *pmd; +int i; + +if (early_ioremap_debug) +pr_info ("early_ioremap_init()\n" ); + +for (i =0 ; i <FIX_BTMAPS_SLOTS ; i++)
Quoted 8 lines
arch/arm64/mm/ioremap.c void __iounmap(volatile void __iomem *io_addr) · line 140
Quoted 6 lines
arch/arm64/mm/ioremap.c +6
-0+ slot_virt[i] =__fix_to_virt (FIX_BTMAP_BEGIN -NR_FIX_BTMAPS *i); + + pmd =early_ioremap_pmd (fix_to_virt (FIX_BTMAP_BEGIN )); +#ifndef CONFIG_ARM64_64K_PAGES +/* need to populate pmd for 4k pagesize only */ +memset (bm_pte,0 ,sizeof (bm_pte));Do we need
memset()here? bm_pte[] is placed in the .bss section.
No we don't. If .bss isn't cleared here, we have bigger problems.
Quoted 14 lines
arch/arm64/mm/ioremap.c void __iounmap(volatile void __iomem *io_addr) · line 151
Quoted 12 lines
arch/arm64/mm/ioremap.c +11
-0+pmd_populate_kernel (&init_mm, pmd, bm_pte); +#endif + +/* +* The boot-ioremap range spans multiple pmds, for which +* we are not prepared: +*/ +#define __FIXADDR_TOP (-PAGE_SIZE) +BUILD_BUG_ON ((__fix_to_virt (FIX_BTMAP_BEGIN ) >>PMD_SHIFT ) + != (__fix_to_virt (FIX_BTMAP_END ) >>PMD_SHIFT )); +#undef __FIXADDR_TOPWhy this #define/#undef?
FIXADDR_TOPis statically defined.
Right. Copy and paste from x86 where is isn't static, so the trick to trigger a BUILD_BUG_ON isn't needed here.
Quoted 18 lines
Quoted 18 lines
arch/arm64/mm/ioremap.c +17
-0+void __init __set_fixmap (enum fixed_addresses idx, +phys_addr_t phys,pgprot_t flags) +{ +unsigned long addr =__fix_to_virt (idx); +pte_t *pte; + +if (idx >= __end_of_fixed_addresses) { +BUG (); +return ; + } +if (after_paging_init) { +WARN_ON (1 ); +return ; + } + + pte =early_ioremap_pte (addr); +
Quoted 8 lines
arch/arm64/mm/ioremap.c void __iounmap(volatile void __iomem *io_addr) · line 193
Quoted 5 lines
arch/arm64/mm/ioremap.c +5
-0+if (pgprot_val (flags)) +set_pte (pte,pfn_pte (phys >>PAGE_SHIFT , flags)); +else +pte_clear (&init_mm, addr, pte); +flush_tlb_kernel_range (addr, addr+PAGE_SIZE );Would
__set_fixmapbe used to change valid ptes? If not, we could keep theflush_tlb_kernel_range()call only under the 'else' block.
A valid pte shouldn't be updated except for the unmap case. Nothing prevents it though. We could do the tlb flush only in the else block and add a BUG_ON to catch attempts to modify an already used pte.
Quoted 5 lines
As I was going through the patch, I realised that the
early_ioremap()looks really to the x86 implementation. Can we move it into a library to be shared between the two?
Yes, that's where I started. I will look at consolidating the common bits.
Thanks.
--Mark