// SPDX-License-Identifier: GPL-2.0-or-later /* * Copyright (C) 2004 Benjamin Herrenschmidt, IBM Corp. * */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #undef DEBUG #ifdef DEBUG #define DBG(fmt...) printk(fmt) #else #define DBG(fmt...) #endif /* Max supported size for symbol names */ #define MAX_SYMNAME 64 /* The alignment of the vDSO */ #define VDSO_ALIGNMENT (1 << 16) static void *vdso32_kbase; extern char vdso32_start, vdso32_end; extern char vdso64_start, vdso64_end; static void *vdso64_kbase = &vdso64_start; static int vdso_ready; /* * The vdso data page (aka. systemcfg for old ppc64 fans) is here. * Once the early boot kernel code no longer needs to muck around * with it, it will become dynamically allocated */ static union { struct vdso_arch_data data; u8 page[PAGE_SIZE]; } vdso_data_store __page_aligned_data; struct vdso_arch_data *vdso_data = &vdso_data_store.data; /* Format of the patch table */ struct vdso_patch_def { unsigned long ftr_mask, ftr_value; const char *gen_name; const char *fix_name; }; /* Table of functions to patch based on the CPU type/revision * * Currently, we only change sync_dicache to do nothing on processors * with a coherent icache */ static struct vdso_patch_def vdso_patches[] = { }; /* * Some infos carried around for each of them during parsing at * boot time. */ struct lib32_elfinfo { Elf32_Ehdr *hdr; /* ptr to ELF */ Elf32_Sym *dynsym; /* ptr to .dynsym section */ unsigned long dynsymsize; /* size of .dynsym section */ char *dynstr; /* ptr to .dynstr section */ unsigned long text; /* offset of .text section in .so */ }; struct lib64_elfinfo { Elf64_Ehdr *hdr; Elf64_Sym *dynsym; unsigned long dynsymsize; char *dynstr; unsigned long text; }; static int vdso_mremap(const struct vm_special_mapping *sm, struct vm_area_struct *new_vma, unsigned long text_size) { unsigned long new_size = new_vma->vm_end - new_vma->vm_start; if (new_size != text_size + PAGE_SIZE) return -EINVAL; current->mm->context.vdso = (void __user *)new_vma->vm_start + PAGE_SIZE; return 0; } static int vdso32_mremap(const struct vm_special_mapping *sm, struct vm_area_struct *new_vma) { return vdso_mremap(sm, new_vma, &vdso32_end - &vdso32_start); } static int vdso64_mremap(const struct vm_special_mapping *sm, struct vm_area_struct *new_vma) { return vdso_mremap(sm, new_vma, &vdso64_end - &vdso64_start); } static struct vm_special_mapping vdso32_spec __ro_after_init = { .name = "[vdso]", .mremap = vdso32_mremap, }; static struct vm_special_mapping vdso64_spec __ro_after_init = { .name = "[vdso]", .mremap = vdso64_mremap, }; /* * This is called from binfmt_elf, we create the special vma for the * vDSO and insert it into the mm struct tree */ static int __arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp) { struct mm_struct *mm = current->mm; struct vm_special_mapping *vdso_spec; struct vm_area_struct *vma; unsigned long vdso_size; unsigned long vdso_base; if (is_32bit_task()) { vdso_spec = &vdso32_spec; vdso_size = &vdso32_end - &vdso32_start; vdso_base = VDSO32_MBASE; } else { vdso_spec = &vdso64_spec; vdso_size = &vdso64_end - &vdso64_start; /* * On 64bit we don't have a preferred map address. This * allows get_unmapped_area to find an area near other mmaps * and most likely share a SLB entry. */ vdso_base = 0; } /* Add a page to the vdso size for the data page */ vdso_size += PAGE_SIZE; /* * pick a base address for the vDSO in process space. We try to put it * at vdso_base which is the "natural" base for it, but we might fail * and end up putting it elsewhere. * Add enough to the size so that the result can be aligned. */ vdso_base = get_unmapped_area(NULL, vdso_base, vdso_size + ((VDSO_ALIGNMENT - 1) & PAGE_MASK), 0, 0); if (IS_ERR_VALUE(vdso_base)) return vdso_base; /* Add required alignment. */ vdso_base = ALIGN(vdso_base, VDSO_ALIGNMENT); /* * Put vDSO base into mm struct. We need to do this before calling * install_special_mapping or the perf counter mmap tracking code * will fail to recognise it as a vDSO. */ mm->context.vdso = (void __user *)vdso_base + PAGE_SIZE; /* * our vma flags don't have VM_WRITE so by default, the process isn't * allowed to write those pages. * gdb can break that with ptrace interface, and thus trigger COW on * those pages but it's then your responsibility to never do that on * the "data" page of the vDSO or you'll stop getting kernel updates * and your nice userland gettimeofday will be totally dead. * It's fine to use that for setting breakpoints in the vDSO code * pages though. */ vma = _install_special_mapping(mm, vdso_base, vdso_size, VM_READ | VM_EXEC | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC, vdso_spec); return PTR_ERR_OR_ZERO(vma); } int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp) { struct mm_struct *mm = current->mm; int rc; mm->context.vdso = NULL; if (!vdso_ready) return 0; if (mmap_write_lock_killable(mm)) return -EINTR; rc = __arch_setup_additional_pages(bprm, uses_interp); if (rc) mm->context.vdso = NULL; mmap_write_unlock(mm); return rc; } #ifdef CONFIG_VDSO32 static void * __init find_section32(Elf32_Ehdr *ehdr, const char *secname, unsigned long *size) { Elf32_Shdr *sechdrs; unsigned int i; char *secnames; /* Grab section headers and strings so we can tell who is who */ sechdrs = (void *)ehdr + ehdr->e_shoff; secnames = (void *)ehdr + sechdrs[ehdr->e_shstrndx].sh_offset; /* Find the section they want */ for (i = 1; i < ehdr->e_shnum; i++) { if (strcmp(secnames+sechdrs[i].sh_name, secname) == 0) { if (size) *size = sechdrs[i].sh_size; return (void *)ehdr + sechdrs[i].sh_offset; } } *size = 0; return NULL; } static Elf32_Sym * __init find_symbol32(struct lib32_elfinfo *lib, const char *symname) { unsigned int i; char name[MAX_SYMNAME], *c; for (i = 0; i < (lib->dynsymsize / sizeof(Elf32_Sym)); i++) { if (lib->dynsym[i].st_name == 0) continue; strlcpy(name, lib->dynstr + lib->dynsym[i].st_name, MAX_SYMNAME); c = strchr(name, '@'); if (c) *c = 0; if (strcmp(symname, name) == 0) return &lib->dynsym[i]; } return NULL; } static int __init vdso_do_func_patch32(struct lib32_elfinfo *v32, struct lib64_elfinfo *v64, const char *orig, const char *fix) { Elf32_Sym *sym32_gen, *sym32_fix; sym32_gen = find_symbol32(v32, orig); if (sym32_gen == NULL) { printk(KERN_ERR "vDSO32: Can't find symbol %s !\n", orig); return -1; } if (fix == NULL) { sym32_gen->st_name = 0; return 0; } sym32_fix = find_symbol32(v32, fix); if (sym32_fix == NULL) { printk(KERN_ERR "vDSO32: Can't find symbol %s !\n", fix); return -1; } sym32_gen->st_value = sym32_fix->st_value; sym32_gen->st_size = sym32_fix->st_size; sym32_gen->st_info = sym32_fix->st_info; sym32_gen->st_other = sym32_fix->st_other; sym32_gen->st_shndx = sym32_fix->st_shndx; return 0; } #else /* !CONFIG_VDSO32 */ static int __init vdso_do_func_patch32(struct lib32_elfinfo *v32, struct lib64_elfinfo *v64, const char *orig, const char *fix) { return 0; } #endif /* CONFIG_VDSO32 */ #ifdef CONFIG_PPC64 static void * __init find_section64(Elf64_Ehdr *ehdr, const char *secname, unsigned long *size) { Elf64_Shdr *sechdrs; unsigned int i; char *secnames; /* Grab section headers and strings so we can tell who is who */ sechdrs = (void *)ehdr + ehdr->e_shoff; secnames = (void *)ehdr + sechdrs[ehdr->e_shstrndx].sh_offset; /* Find the section they want */ for (i = 1; i < ehdr->e_shnum; i++) { if (strcmp(secnames+sechdrs[i].sh_name, secname) == 0) { if (size) *size = sechdrs[i].sh_size; return (void *)ehdr + sechdrs[i].sh_offset; } } if (size) *size = 0; return NULL; } static Elf64_Sym * __init find_symbol64(struct lib64_elfinfo *lib, const char *symname) { unsigned int i; char name[MAX_SYMNAME], *c; for (i = 0; i < (lib->dynsymsize / sizeof(Elf64_Sym)); i++) { if (lib->dynsym[i].st_name == 0) continue; strlcpy(name, lib->dynstr + lib->dynsym[i].st_name, MAX_SYMNAME); c = strchr(name, '@'); if (c) *c = 0; if (strcmp(symname, name) == 0) return &lib->dynsym[i]; } return NULL; } static int __init vdso_do_func_patch64(struct lib32_elfinfo *v32, struct lib64_elfinfo *v64, const char *orig, const char *fix) { Elf64_Sym *sym64_gen, *sym64_fix; sym64_gen = find_symbol64(v64, orig); if (sym64_gen == NULL) { printk(KERN_ERR "vDSO64: Can't find symbol %s !\n", orig); return -1; } if (fix == NULL) { sym64_gen->st_name = 0; return 0; } sym64_fix = find_symbol64(v64, fix); if (sym64_fix == NULL) { printk(KERN_ERR "vDSO64: Can't find symbol %s !\n", fix); return -1; } sym64_gen->st_value = sym64_fix->st_value; sym64_gen->st_size = sym64_fix->st_size; sym64_gen->st_info = sym64_fix->st_info; sym64_gen->st_other = sym64_fix->st_other; sym64_gen->st_shndx = sym64_fix->st_shndx; return 0; } #endif /* CONFIG_PPC64 */ #define VDSO_DO_FIXUPS(type, value, bits, sec) do { \ void *__start = (void *)VDSO##bits##_SYMBOL(&vdso##bits##_start, sec##_start); \ void *__end = (void *)VDSO##bits##_SYMBOL(&vdso##bits##_start, sec##_end); \ \ do_##type##_fixups((value), __start, __end); \ } while (0) static __init int vdso_do_find_sections(struct lib32_elfinfo *v32, struct lib64_elfinfo *v64) { void *sect; /* * Locate symbol tables & text section */ #ifdef CONFIG_VDSO32 v32->dynsym = find_section32(v32->hdr, ".dynsym", &v32->dynsymsize); v32->dynstr = find_section32(v32->hdr, ".dynstr", NULL); if (v32->dynsym == NULL || v32->dynstr == NULL) { printk(KERN_ERR "vDSO32: required symbol section not found\n"); return -1; } sect = find_section32(v32->hdr, ".text", NULL); if (sect == NULL) { printk(KERN_ERR "vDSO32: the .text section was not found\n"); return -1; } v32->text = sect - vdso32_kbase; #endif #ifdef CONFIG_PPC64 v64->dynsym = find_section64(v64->hdr, ".dynsym", &v64->dynsymsize); v64->dynstr = find_section64(v64->hdr, ".dynstr", NULL); if (v64->dynsym == NULL || v64->dynstr == NULL) { printk(KERN_ERR "vDSO64: required symbol section not found\n"); return -1; } sect = find_section64(v64->hdr, ".text", NULL); if (sect == NULL) { printk(KERN_ERR "vDSO64: the .text section was not found\n"); return -1; } v64->text = sect - vdso64_kbase; #endif /* CONFIG_PPC64 */ return 0; } static __init int vdso_fixup_features(struct lib32_elfinfo *v32, struct lib64_elfinfo *v64) { #ifdef CONFIG_PPC64 VDSO_DO_FIXUPS(feature, cur_cpu_spec->cpu_features, 64, ftr_fixup); VDSO_DO_FIXUPS(feature, cur_cpu_spec->mmu_features, 64, mmu_ftr_fixup); VDSO_DO_FIXUPS(feature, powerpc_firmware_features, 64, fw_ftr_fixup); VDSO_DO_FIXUPS(lwsync, cur_cpu_spec->cpu_features, 64, lwsync_fixup); #endif /* CONFIG_PPC64 */ #ifdef CONFIG_VDSO32 VDSO_DO_FIXUPS(feature, cur_cpu_spec->cpu_features, 32, ftr_fixup); VDSO_DO_FIXUPS(feature, cur_cpu_spec->mmu_features, 32, mmu_ftr_fixup); #ifdef CONFIG_PPC64 VDSO_DO_FIXUPS(feature, powerpc_firmware_features, 32, fw_ftr_fixup); #endif /* CONFIG_PPC64 */ VDSO_DO_FIXUPS(lwsync, cur_cpu_spec->cpu_features, 32, lwsync_fixup); #endif return 0; } static __init int vdso_fixup_alt_funcs(struct lib32_elfinfo *v32, struct lib64_elfinfo *v64) { int i; for (i = 0; i < ARRAY_SIZE(vdso_patches); i++) { struct vdso_patch_def *patch = &vdso_patches[i]; int match = (cur_cpu_spec->cpu_features & patch->ftr_mask) == patch->ftr_value; if (!match) continue; DBG("replacing %s with %s...\n", patch->gen_name, patch->fix_name ? "NONE" : patch->fix_name); /* * Patch the 32 bits and 64 bits symbols. Note that we do not * patch the "." symbol on 64 bits. * It would be easy to do, but doesn't seem to be necessary, * patching the OPD symbol is enough. */ vdso_do_func_patch32(v32, v64, patch->gen_name, patch->fix_name); #ifdef CONFIG_PPC64 vdso_do_func_patch64(v32, v64, patch->gen_name, patch->fix_name); #endif /* CONFIG_PPC64 */ } return 0; } static __init int vdso_setup(void) { struct lib32_elfinfo v32; struct lib64_elfinfo v64; v32.hdr = vdso32_kbase; v64.hdr = vdso64_kbase; if (vdso_do_find_sections(&v32, &v64)) return -1; if (vdso_fixup_features(&v32, &v64)) return -1; if (vdso_fixup_alt_funcs(&v32, &v64)) return -1; return 0; } /* * Called from setup_arch to initialize the bitmap of available * syscalls in the systemcfg page */ static void __init vdso_setup_syscall_map(void) { unsigned int i; for (i = 0; i < NR_syscalls; i++) { if (sys_call_table[i] != (unsigned long)&sys_ni_syscall) vdso_data->syscall_map[i >> 5] |= 0x80000000UL >> (i & 0x1f); if (IS_ENABLED(CONFIG_COMPAT) && compat_sys_call_table[i] != (unsigned long)&sys_ni_syscall) vdso_data->compat_syscall_map[i >> 5] |= 0x80000000UL >> (i & 0x1f); } } #ifdef CONFIG_PPC64 int vdso_getcpu_init(void) { unsigned long cpu, node, val; /* * SPRG_VDSO contains the CPU in the bottom 16 bits and the NUMA node * in the next 16 bits. The VDSO uses this to implement getcpu(). */ cpu = get_cpu(); WARN_ON_ONCE(cpu > 0xffff); node = cpu_to_node(cpu); WARN_ON_ONCE(node > 0xffff); val = (cpu & 0xffff) | ((node & 0xffff) << 16); mtspr(SPRN_SPRG_VDSO_WRITE, val); get_paca()->sprg_vdso = val; put_cpu(); return 0; } /* We need to call this before SMP init */ early_initcall(vdso_getcpu_init); #endif static struct page ** __init vdso_setup_pages(void *start, void *end) { int i; struct page **pagelist; int pages = (end - start) >> PAGE_SHIFT; pagelist = kcalloc(pages + 1, sizeof(struct page *), GFP_KERNEL); if (!pagelist) panic("%s: Cannot allocate page list for VDSO", __func__); pagelist[0] = virt_to_page(vdso_data); for (i = 0; i < pages; i++) pagelist[i + 1] = virt_to_page(start + i * PAGE_SIZE); return pagelist; } static int __init vdso_init(void) { #ifdef CONFIG_PPC64 /* * Fill up the "systemcfg" stuff for backward compatibility */ strcpy((char *)vdso_data->eye_catcher, "SYSTEMCFG:PPC64"); vdso_data->version.major = SYSTEMCFG_MAJOR; vdso_data->version.minor = SYSTEMCFG_MINOR; vdso_data->processor = mfspr(SPRN_PVR); /* * Fake the old platform number for pSeries and add * in LPAR bit if necessary */ vdso_data->platform = 0x100; if (firmware_has_feature(FW_FEATURE_LPAR)) vdso_data->platform |= 1; vdso_data->physicalMemorySize = memblock_phys_mem_size(); vdso_data->dcache_size = ppc64_caches.l1d.size; vdso_data->dcache_line_size = ppc64_caches.l1d.line_size; vdso_data->icache_size = ppc64_caches.l1i.size; vdso_data->icache_line_size = ppc64_caches.l1i.line_size; vdso_data->dcache_block_size = ppc64_caches.l1d.block_size; vdso_data->icache_block_size = ppc64_caches.l1i.block_size; vdso_data->dcache_log_block_size = ppc64_caches.l1d.log_block_size; vdso_data->icache_log_block_size = ppc64_caches.l1i.log_block_size; #endif /* CONFIG_PPC64 */ vdso32_kbase = &vdso32_start; vdso_setup_syscall_map(); /* * Initialize the vDSO images in memory, that is do necessary * fixups of vDSO symbols, locate trampolines, etc... */ if (vdso_setup()) { printk(KERN_ERR "vDSO setup failure, not enabled !\n"); return 0; } if (IS_ENABLED(CONFIG_VDSO32)) vdso32_spec.pages = vdso_setup_pages(&vdso32_start, &vdso32_end); if (IS_ENABLED(CONFIG_PPC64)) vdso64_spec.pages = vdso_setup_pages(&vdso64_start, &vdso64_end); smp_wmb(); vdso_ready = 1; return 0; } arch_initcall(vdso_init);