From: Linus Walleij <linus@xxxxxxxxxxxxxxxxxxxxx> This adds the TCM interface to Linux, when active, it will detect and report TCM memories and sizes early in boot if present, introduce generic TCM memory handling, provide a generic TCM memory pool and select TCM memory for the U300 platform. I use arch/arm/include/asm/cputype.h that defines a CPUID_TCM register that you can read out from the system control coprocessor. Documentation from ARM can be found at http://infocenter.arm.com, search for "TCM Status Register" to see documents for all CPUs. Reading this register you can determine if ITCM (bit 0) and/or DTCM (bit 16) is present in the machine. This will be used to implement power management (sleep mode) for the U300 machines. Tell me if you think this should go into /proc/cpuinfo as well. Currently it doesn't. Caveats: I use the addresses at 0xffff2000 and 0xffff4000 in the U300 architecture for mapping the TCM with a one-to-one memory map. This is "reserved" in Russells map, good suggestions on where else to put this 0x4000 block is welcome. No special handling of BSS (uninitialized) sections. They get converted to generic DATA (and thus zero-initialized). Signed-off-by: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx> --- ChangeLog v1->v2 - Fixed linker file, added explicit AT command to make the linker stop moaning about dot being moved backwards. - Adjusted TCM sections in linkfile to the PAGE_SIZE (4096) so they can be free():ed by free_initmem. - Modified free_initmem() to also free the TCM sections from the linkfile. - Some checkpatch stuff etc. - See earlier mail Patch [0/1] v1 for an introduction. --- Documentation/arm/tcm.txt | 139 +++++++++++++++++ arch/arm/Kconfig | 5 + arch/arm/include/asm/cputype.h | 5 + arch/arm/include/asm/tcm.h | 31 ++++ arch/arm/kernel/Makefile | 1 + arch/arm/kernel/setup.c | 2 + arch/arm/kernel/tcm.c | 246 ++++++++++++++++++++++++++++++ arch/arm/kernel/tcm.h | 17 ++ arch/arm/kernel/vmlinux.lds.S | 57 +++++++ arch/arm/mach-u300/include/mach/memory.h | 8 + arch/arm/mm/init.c | 8 + 11 files changed, 519 insertions(+), 0 deletions(-) create mode 100644 Documentation/arm/tcm.txt create mode 100644 arch/arm/include/asm/tcm.h create mode 100644 arch/arm/kernel/tcm.c create mode 100644 arch/arm/kernel/tcm.h diff --git a/Documentation/arm/tcm.txt b/Documentation/arm/tcm.txt new file mode 100644 index 0000000..7554511 --- /dev/null +++ b/Documentation/arm/tcm.txt @@ -0,0 +1,139 @@ +ARM TCM (Tightly-Coupled Memory) handling in Linux +---- +Written by Linus Walleij <linus.walleij@xxxxxxxxxxxxxx> + +Some ARM SoC:s have a so-called TCM (Tightly-Coupled Memory). +This is usually just a few 4-64 KiB inside the ARM processor. +Due to being embedded inside the CPU The TCM has a +Harvard-architecture, so there is an ITCM (instruction TCM) +and a DTCM (data TCM). The DTCM can not contain any +instructions, but the ITCM can actually contain data. +The size of DTCM or ITCM is minimum 4KiB so the typical +minimum configuration is 4KiB ITCM and 4KiB DTCM. + +ARM CPU:s have special registers to read out status, physical +location and size of TCM memories. arch/arm/include/asm/cputype.h +defines a CPUID_TCM register that you can read out from the +system control coprocessor. Documentation from ARM can be found +at http://infocenter.arm.com, search for "TCM Status Register" +to see documents for all CPUs. Reading this register you can +determine if ITCM (bit 0) and/or DTCM (bit 16) is present in the +machine. + +There is further a TCM region register (search for "TCM Region +Registers" that can report and modify the location size of TCM +memories at runtime. This is used to read out and modify +TCM location and size. + +TCM is used for a few things: + +- FIQ and other interrupt handlers that need deterministic + timing and cannot wait for cache misses. + +- Idle loops where all external RAM is set to self-refresh + retention mode, so only on-chip RAM is accessible by + the CPU and then we hang inside ITCM waiting for an + interrupt. + +- Other operations which implies shutting off or reconfiguring + the external RAM controller. + +Code executing inside the ITCM does not "see" any MMU mappings +and e.g. register accesses must be made to physical addresses. + +There is an interface for using TCM on the ARM architecture +in <asm/tcm.h>. Using this interface it is possible to: + +- Define the physical address and size of ITCM and DTCM. + +- Be able to tag functions to be compiled into ITCM. + +- Be able to tag data and constants to be allocated to + ITCM. + +- Have the remaining TCM RAM added to a special + allocation pool with gen_pool_create() and gen_pool_add() + and provice tcm_alloc() and tcm_free() for this + memory. Such a heap is great for things like saving + device state when shutting off device power domains. + +A machine that has TCM memory shall select HAVE_TCM in +arch/arm/Kconfig for itself, and then the +rest of the functionality will depend on the physical +location and size of ITCM and DTCM to be defined in +mach/memory.h for the machine. Code that needs to use +TCM shall #include <asm/tcm.h> If the TCM is not located +at the place given in memory.h it will be moved using +the TCM Regiom registers. + +Functions to go into itcm can be tagged like this: +int __tcmfunc foo(int bar); + +Variables to go into dtcm can be tagged like this: +int __tcmdata foo; + +Constants can be tagged like this: +int __tcmconst foo; + +To put assembler into TCM just use +.section ".tcm.text" or .section ".tcm.data" +respectively. + +Example code: + +#include <asm/tcm.h> + +/* Uninitialized data */ +static u32 __tcmdata tcmvar; +/* Initialized data */ +static u32 __tcmdata tcmassigned = 0x2BADBABEU; +/* Constant */ +static const u32 __tcmconst tcmconst = 0xCAFEBABEU; + +static void __tcmlocalfunc tcm_to_tcm(void) +{ + int i; + for (i = 0; i < 100; i++) + tcmvar ++; +} + +static void __tcmfunc hello_tcm(void) +{ + /* Some abstract code that runs in ITCM */ + int i; + for (i = 0; i < 100; i++) { + tcmvar ++; + } + tcm_to_tcm(); +} + +static void __init test_tcm(void) +{ + u32 *tcmem; + int i; + + hello_tcm(); + printk("Hello TCM executed from ITCM RAM\n"); + + printk("TCM variable from testrun: %u @ %p\n", tcmvar, &tcmvar); + tcmvar = 0xDEADBEEFU; + printk("TCM variable: 0x%x @ %p\n", tcmvar, &tcmvar); + + printk("TCM assigned variable: 0x%x @ %p\n", tcmassigned, &tcmassigned); + + printk("TCM constant: 0x%x @ %p\n", tcmconst, &tcmconst); + + /* Allocate some TCM memory from the pool */ + tcmem = tcm_alloc(20); + if (tcmem) { + printk("TCM Allocated 20 bytes of TCM @ %p\n", tcmem); + tcmem[0] = 0xDEADBEEFU; + tcmem[1] = 0x2BADBABEU; + tcmem[2] = 0xCAFEBABEU; + tcmem[3] = 0xDEADBEEFU; + tcmem[4] = 0x2BADBABEU; + for (i = 0; i < 5; i++) + printk("TCM tcmem[%d] = %08x\n", i, tcmem[i]); + tcm_free(tcmem, 20); + } +} diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig index aef63c8..d2253f4 100644 --- a/arch/arm/Kconfig +++ b/arch/arm/Kconfig @@ -50,6 +50,10 @@ config MMU bool default y +config HAVE_TCM + bool + select GENERIC_ALLOCATOR + config NO_IOPORT bool @@ -602,6 +606,7 @@ config ARCH_U300 bool "ST-Ericsson U300 Series" depends on MMU select CPU_ARM926T + select HAVE_TCM select ARM_AMBA select ARM_VIC select GENERIC_TIME diff --git a/arch/arm/include/asm/cputype.h b/arch/arm/include/asm/cputype.h index b3e656c..54a1f1d 100644 --- a/arch/arm/include/asm/cputype.h +++ b/arch/arm/include/asm/cputype.h @@ -63,6 +63,11 @@ static inline unsigned int __attribute_const__ read_cpuid_cachetype(void) return read_cpuid(CPUID_CACHETYPE); } +static inline unsigned int __attribute_const__ read_cpuid_tcmstatus(void) +{ + return read_cpuid(CPUID_TCM); +} + /* * Intel's XScale3 core supports some v6 features (supersections, L2) * but advertises itself as v5 as it does not support the v6 ISA. For diff --git a/arch/arm/include/asm/tcm.h b/arch/arm/include/asm/tcm.h new file mode 100644 index 0000000..5929ef5 --- /dev/null +++ b/arch/arm/include/asm/tcm.h @@ -0,0 +1,31 @@ +/* + * + * Copyright (C) 2008-2009 ST-Ericsson AB + * License terms: GNU General Public License (GPL) version 2 + * + * Author: Rickard Andersson <rickard.andersson@xxxxxxxxxxxxxx> + * Author: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx> + * + */ +#ifndef __ASMARM_TCM_H +#define __ASMARM_TCM_H + +#ifndef CONFIG_HAVE_TCM +#error "You should not be including tcm.h unless you have a TCM!" +#endif + +#include <linux/compiler.h> + +/* Tag variables with this */ +#define __tcmdata __section(.tcm.data) +/* Tag constants with this */ +#define __tcmconst __section(.tcm.rodata) +/* Tag functions inside TCM called from outside TCM with this */ +#define __tcmfunc __attribute__((long_call)) __section(.tcm.text) noinline +/* Tag function inside TCM called from inside TCM with this */ +#define __tcmlocalfunc __section(.tcm.text) + +void *tcm_alloc(size_t len); +void tcm_free(void *addr, size_t len); + +#endif diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile index ff89d0b..9961878 100644 --- a/arch/arm/kernel/Makefile +++ b/arch/arm/kernel/Makefile @@ -32,6 +32,7 @@ obj-$(CONFIG_OABI_COMPAT) += sys_oabi-compat.o obj-$(CONFIG_ARM_THUMBEE) += thumbee.o obj-$(CONFIG_KGDB) += kgdb.o obj-$(CONFIG_ARM_UNWIND) += unwind.o +obj-$(CONFIG_HAVE_TCM) += tcm.o obj-$(CONFIG_CRUNCH) += crunch.o crunch-bits.o AFLAGS_crunch-bits.o := -Wa,-mcpu=ep9312 diff --git a/arch/arm/kernel/setup.c b/arch/arm/kernel/setup.c index bc5e412..efd8af9 100644 --- a/arch/arm/kernel/setup.c +++ b/arch/arm/kernel/setup.c @@ -44,6 +44,7 @@ #include "compat.h" #include "atags.h" +#include "tcm.h" #ifndef MEM_SIZE #define MEM_SIZE (16*1024*1024) @@ -735,6 +736,7 @@ void __init setup_arch(char **cmdline_p) #endif cpu_init(); + tcm_init(); /* * Set up various architecture-specific pointers diff --git a/arch/arm/kernel/tcm.c b/arch/arm/kernel/tcm.c new file mode 100644 index 0000000..e503038 --- /dev/null +++ b/arch/arm/kernel/tcm.c @@ -0,0 +1,246 @@ +/* + * Copyright (C) 2008-2009 ST-Ericsson AB + * License terms: GNU General Public License (GPL) version 2 + * TCM memory handling for ARM systems + * + * Author: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx> + * Author: Rickard Andersson <rickard.andersson@xxxxxxxxxxxxxx> + */ +#include <linux/init.h> +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/stddef.h> +#include <linux/ioport.h> +#include <linux/genalloc.h> +#include <linux/string.h> /* memcpy */ +#include <asm/page.h> /* PAGE_SHIFT */ +#include <asm/cputype.h> +#include <asm/mach/map.h> +#include <mach/memory.h> +#include "tcm.h" + +/* Scream and warn about misuse */ +#if !defined(ITCM_OFFSET) || !defined(ITCM_END) || \ + !defined(DTCM_OFFSET) || !defined(DTCM_END) +#error "TCM support selected but offsets not defined!" +#endif + +static struct gen_pool *tcm_pool; + +/* TCM section definitions from the linker */ +extern char __itcm_start, __sitcm_text, __eitcm_text; +extern char __dtcm_start, __sdtcm_data, __edtcm_data; + +/* + * TCM memory resources + */ +static struct resource dtcm_res = { + .name = "DTCM RAM", + .start = DTCM_OFFSET, + .end = DTCM_END, + .flags = IORESOURCE_MEM +}; + +static struct resource itcm_res = { + .name = "ITCM RAM", + .start = ITCM_OFFSET, + .end = ITCM_END, + .flags = IORESOURCE_MEM +}; + +static struct map_desc dtcm_iomap[] __initdata = { + { + .virtual = DTCM_OFFSET, + .pfn = __phys_to_pfn(DTCM_OFFSET), + .length = (DTCM_END - DTCM_OFFSET + 1), + .type = MT_UNCACHED + } +}; + +static struct map_desc itcm_iomap[] __initdata = { + { + .virtual = ITCM_OFFSET, + .pfn = __phys_to_pfn(ITCM_OFFSET), + .length = (ITCM_END - ITCM_OFFSET + 1), + .type = MT_UNCACHED + } +}; + +/* + * Allocate a chunk of TCM memory + */ +void *tcm_alloc(size_t len) +{ + unsigned long vaddr; + + if (!tcm_pool) + return NULL; + + vaddr = gen_pool_alloc(tcm_pool, len); + if (!vaddr) + return NULL; + + return (void *) vaddr; +} +EXPORT_SYMBOL(tcm_alloc); + +/* + * Free a chunk of TCM memory + */ +void tcm_free(void *addr, size_t len) +{ + gen_pool_free(tcm_pool, (unsigned long) addr, len); +} +EXPORT_SYMBOL(tcm_free); + + +static void __init setup_tcm_bank(u8 type, u32 offset, u32 expected_size) +{ + const int tcm_sizes[16] = { 0, -1, -1, 4, 8, 16, 32, 64, 128, + 256, 512, 1024, -1, -1, -1, -1 }; + u32 tcm_region; + int tcm_size; + + /* Read the special TCM region register c9, 0 */ + if (!type) + asm("mrc p15, 0, %0, c9, c1, 0" + : "=r" (tcm_region)); + else + asm("mrc p15, 0, %0, c9, c1, 1" + : "=r" (tcm_region)); + + tcm_size = tcm_sizes[(tcm_region >> 2) & 0x0f]; + if (tcm_size < 0) { + pr_err("CPU: %sTCM of unknown size!\n", + type ? "I" : "D"); + } else { + pr_info("CPU: found %sTCM %dk @ %08x, %senabled\n", + type ? "I" : "D", + tcm_size, + (tcm_region & 0xfffff000U), + (tcm_region & 1) ? "" : "not "); + } + + if (tcm_size != expected_size) { + pr_crit("CPU: %sTCM was detected %dk but expected %dk!\n", + type ? "I" : "D", + tcm_size, + expected_size); + /* Adjust to the expected size? what can we do... */ + } + + /* Force move the TCM bank to where we want it, enable */ + tcm_region = offset | (tcm_region & 0x00000ffeU) | 1; + + if (!type) + asm("mcr p15, 0, %0, c9, c1, 0" + : /* No output operands */ + : "r" (tcm_region)); + else + asm("mcr p15, 0, %0, c9, c1, 1" + : /* No output operands */ + : "r" (tcm_region)); + + pr_debug("CPU: moved %sTCM %dk to %08x, enabled\n", + type ? "I" : "D", + tcm_size, + (tcm_region & 0xfffff000U)); +} + +/* + * This initializes the TCM memory + */ +void __init tcm_init(void) +{ + u32 tcm_status = read_cpuid_tcmstatus(); + char *start; + char *end; + char *ram; + + /* Setup DTCM if present */ + if (tcm_status & (1 << 16)) { + setup_tcm_bank(0, DTCM_OFFSET, + (DTCM_END - DTCM_OFFSET + 1) >> 10); + request_resource(&iomem_resource, &dtcm_res); + iotable_init(dtcm_iomap, 1); + /* Copy data from RAM to DTCM */ + start = &__sdtcm_data; + end = &__edtcm_data; + ram = &__dtcm_start; + memcpy(start, ram, (end-start)); + pr_debug("CPU DTCM: copied data from %p - %p\n", start, end); + } + + /* Setup ITCM if present */ + if (tcm_status & 1) { + setup_tcm_bank(1, ITCM_OFFSET, + (ITCM_END - ITCM_OFFSET + 1) >> 10); + request_resource(&iomem_resource, &itcm_res); + iotable_init(itcm_iomap, 1); + /* Copy code from RAM to ITCM */ + start = &__sitcm_text; + end = &__eitcm_text; + ram = &__itcm_start; + memcpy(start, ram, (end-start)); + pr_debug("CPU ITCM: copied code from %p - %p\n", start, end); + } +} + +/* + * This creates the TCM memory pool and has to be done later, + * during the core_initicalls, since the allocator is not yet + * up and running when the first initialization runs. + */ +static int __init setup_tcm_pool(void) +{ + u32 tcm_status = read_cpuid_tcmstatus(); + u32 dtcm_pool_start = (u32) &__edtcm_data; + u32 itcm_pool_start = (u32) &__eitcm_text; + int ret; + + /* + * Set up malloc pool, 2^2 = 4 bytes granularity since + * the TCM is sometimes just 4 KiB. NB: pages and cache + * line alignments does not matter in TCM! + */ + tcm_pool = gen_pool_create(2, -1); + + pr_debug("Setting up TCM memory pool\n"); + + /* Add the rest of DTCM to the TCM pool */ + if (tcm_status & (1 << 16)) { + if (dtcm_pool_start < DTCM_END) { + ret = gen_pool_add(tcm_pool, dtcm_pool_start, + DTCM_END - dtcm_pool_start + 1, -1); + if (ret) { + pr_err("CPU DTCM: could not add DTCM " \ + "remainder to pool!\n"); + return ret; + } + pr_debug("CPU DTCM: Added %08x bytes @ %08x to " \ + "the TCM memory pool\n", + DTCM_END - dtcm_pool_start + 1, + dtcm_pool_start); + } + } + + /* Add the rest of ITCM to the TCM pool */ + if (tcm_status & 1) { + if (itcm_pool_start < ITCM_END) { + ret = gen_pool_add(tcm_pool, itcm_pool_start, + ITCM_END - itcm_pool_start + 1, -1); + if (ret) { + pr_err("CPU ITCM: could not add ITCM " \ + "remainder to pool!\n"); + return ret; + } + pr_debug("CPU ITCM: Added %08x bytes @ %08x to " \ + "the TCM memory pool\n", + ITCM_END - itcm_pool_start + 1, + itcm_pool_start); + } + } + return 0; +} + +core_initcall(setup_tcm_pool); diff --git a/arch/arm/kernel/tcm.h b/arch/arm/kernel/tcm.h new file mode 100644 index 0000000..8015ad4 --- /dev/null +++ b/arch/arm/kernel/tcm.h @@ -0,0 +1,17 @@ +/* + * Copyright (C) 2008-2009 ST-Ericsson AB + * License terms: GNU General Public License (GPL) version 2 + * TCM memory handling for ARM systems + * + * Author: Linus Walleij <linus.walleij@xxxxxxxxxxxxxx> + * Author: Rickard Andersson <rickard.andersson@xxxxxxxxxxxxxx> + */ + +#ifdef CONFIG_HAVE_TCM +void __init tcm_init(void); +#else +/* No TCM support, just blank inlines to be optimized out */ +inline void tcm_init(void) +{ +} +#endif diff --git a/arch/arm/kernel/vmlinux.lds.S b/arch/arm/kernel/vmlinux.lds.S index 6937102..3f58c56 100644 --- a/arch/arm/kernel/vmlinux.lds.S +++ b/arch/arm/kernel/vmlinux.lds.S @@ -198,6 +198,63 @@ SECTIONS } _edata_loc = __data_loc + SIZEOF(.data); +#ifdef CONFIG_HAVE_TCM + /* + * We align everything to a page boundary so we can + * free it after init has commenced and TCM contents have + * been copied to its destination. + */ + .tcm_start : { + . = ALIGN(PAGE_SIZE); + __tcm_start = .; + __itcm_start = .; + } + + /* + * Link these to the ITCM RAM + * Put VMA to the TCM address and LMA to the common RAM + * and we'll upload the contents from RAM to TCM and free + * the used RAM after that. + */ + .text_itcm ITCM_OFFSET : AT(__itcm_start) + { + __sitcm_text = .; + *(.tcm.text) + *(.tcm.rodata) + . = ALIGN(4); + __eitcm_text = .; + } + + /* + * Reset the dot pointer, this is needed to create the + * relative __dtcm_start below (to be used as extern in code). + */ + . = ADDR(.tcm_start) + SIZEOF(.tcm_start) + SIZEOF(.text_itcm); + + .dtcm_start : { + __dtcm_start = .; + } + + /* TODO: add remainder of ITCM as well, that can be used for data! */ + .data_dtcm DTCM_OFFSET : AT(__dtcm_start) + { + . = ALIGN(4); + __sdtcm_data = .; + *(.tcm.data) + . = ALIGN(4); + __edtcm_data = .; + } + + /* Reset the dot pointer or the linker gets confused */ + . = ADDR(.dtcm_start) + SIZEOF(.data_dtcm); + + /* End marker for freeing TCM copy in linked object */ + .tcm_end : AT(ADDR(.dtcm_start) + SIZEOF(.data_dtcm)){ + . = ALIGN(PAGE_SIZE); + __tcm_end = .; + } +#endif + .bss : { __bss_start = .; /* BSS */ *(.bss) diff --git a/arch/arm/mach-u300/include/mach/memory.h b/arch/arm/mach-u300/include/mach/memory.h index bf134bc..ab000df 100644 --- a/arch/arm/mach-u300/include/mach/memory.h +++ b/arch/arm/mach-u300/include/mach/memory.h @@ -35,6 +35,14 @@ #endif /* + * TCM memory whereabouts + */ +#define ITCM_OFFSET 0xffff2000 +#define ITCM_END 0xffff3fff +#define DTCM_OFFSET 0xffff4000 +#define DTCM_END 0xffff5fff + +/* * We enable a real big DMA buffer if need be. */ #define CONSISTENT_DMA_SIZE SZ_4M diff --git a/arch/arm/mm/init.c b/arch/arm/mm/init.c index 8277802..994873a 100644 --- a/arch/arm/mm/init.c +++ b/arch/arm/mm/init.c @@ -553,6 +553,14 @@ void __init mem_init(void) void free_initmem(void) { +#ifdef CONFIG_HAVE_TCM + extern char *__tcm_start, *__tcm_end; + + totalram_pages += free_area(__phys_to_pfn(__pa(__tcm_start)), + __phys_to_pfn(__pa(__tcm_end)), + "TCM link"); +#endif + if (!machine_is_integrator() && !machine_is_cintegrator()) totalram_pages += free_area(__phys_to_pfn(__pa(__init_begin)), __phys_to_pfn(__pa(__init_end)), -- 1.6.2.2 -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
