diff options
Diffstat (limited to 'src/shared/elf-util.c')
| -rw-r--r-- | src/shared/elf-util.c | 899 |
1 files changed, 899 insertions, 0 deletions
diff --git a/src/shared/elf-util.c b/src/shared/elf-util.c new file mode 100644 index 0000000..24ed16e --- /dev/null +++ b/src/shared/elf-util.c @@ -0,0 +1,899 @@ +/* SPDX-License-Identifier: LGPL-2.1-or-later */ + +#if HAVE_ELFUTILS + +#include <dwarf.h> +#include <elfutils/libdwelf.h> +#include <elfutils/libdwfl.h> +#include <libelf.h> +#include <sys/prctl.h> +#include <sys/resource.h> +#include <sys/types.h> +#include <unistd.h> + +#include "alloc-util.h" +#include "dlfcn-util.h" +#include "elf-util.h" +#include "errno-util.h" +#include "escape.h" +#include "fileio.h" +#include "fd-util.h" +#include "format-util.h" +#include "hexdecoct.h" +#include "io-util.h" +#include "macro.h" +#include "memstream-util.h" +#include "process-util.h" +#include "rlimit-util.h" +#include "string-util.h" + +#define FRAMES_MAX 64 +#define THREADS_MAX 64 +#define ELF_PACKAGE_METADATA_ID 0xcafe1a7e + +/* The amount of data we're willing to write to each of the output pipes. */ +#define COREDUMP_PIPE_MAX (1024*1024U) + +static void *dw_dl = NULL; +static void *elf_dl = NULL; + +/* libdw symbols */ +Dwarf_Attribute *(*sym_dwarf_attr_integrate)(Dwarf_Die *, unsigned int, Dwarf_Attribute *); +const char *(*sym_dwarf_diename)(Dwarf_Die *); +const char *(*sym_dwarf_formstring)(Dwarf_Attribute *); +int (*sym_dwarf_getscopes)(Dwarf_Die *, Dwarf_Addr, Dwarf_Die **); +int (*sym_dwarf_getscopes_die)(Dwarf_Die *, Dwarf_Die **); +Elf *(*sym_dwelf_elf_begin)(int); +#if HAVE_DWELF_ELF_E_MACHINE_STRING +const char *(*sym_dwelf_elf_e_machine_string)(int); +#endif +ssize_t (*sym_dwelf_elf_gnu_build_id)(Elf *, const void **); +int (*sym_dwarf_tag)(Dwarf_Die *); +Dwfl_Module *(*sym_dwfl_addrmodule)(Dwfl *, Dwarf_Addr); +Dwfl *(*sym_dwfl_begin)(const Dwfl_Callbacks *); +int (*sym_dwfl_build_id_find_elf)(Dwfl_Module *, void **, const char *, Dwarf_Addr, char **, Elf **); +int (*sym_dwfl_core_file_attach)(Dwfl *, Elf *); +int (*sym_dwfl_core_file_report)(Dwfl *, Elf *, const char *); +void (*sym_dwfl_end)(Dwfl *); +const char *(*sym_dwfl_errmsg)(int); +int (*sym_dwfl_errno)(void); +bool (*sym_dwfl_frame_pc)(Dwfl_Frame *, Dwarf_Addr *, bool *); +ptrdiff_t (*sym_dwfl_getmodules)(Dwfl *, int (*)(Dwfl_Module *, void **, const char *, Dwarf_Addr, void *), void *, ptrdiff_t); +int (*sym_dwfl_getthreads)(Dwfl *, int (*)(Dwfl_Thread *, void *), void *); +Dwarf_Die *(*sym_dwfl_module_addrdie)(Dwfl_Module *, Dwarf_Addr, Dwarf_Addr *); +const char *(*sym_dwfl_module_addrname)(Dwfl_Module *, GElf_Addr); +int (*sym_dwfl_module_build_id)(Dwfl_Module *, const unsigned char **, GElf_Addr *); +Elf *(*sym_dwfl_module_getelf)(Dwfl_Module *, GElf_Addr *); +const char *(*sym_dwfl_module_info)(Dwfl_Module *, void ***, Dwarf_Addr *, Dwarf_Addr *, Dwarf_Addr *, Dwarf_Addr *, const char **, const char **); +int (*sym_dwfl_offline_section_address)(Dwfl_Module *, void **, const char *, Dwarf_Addr, const char *, GElf_Word, const GElf_Shdr *, Dwarf_Addr *); +int (*sym_dwfl_report_end)(Dwfl *, int (*)(Dwfl_Module *, void *, const char *, Dwarf_Addr, void *), void *); +int (*sym_dwfl_standard_find_debuginfo)(Dwfl_Module *, void **, const char *, Dwarf_Addr, const char *, const char *, GElf_Word, char **); +int (*sym_dwfl_thread_getframes)(Dwfl_Thread *, int (*)(Dwfl_Frame *, void *), void *); +pid_t (*sym_dwfl_thread_tid)(Dwfl_Thread *); + +/* libelf symbols */ +Elf *(*sym_elf_begin)(int, Elf_Cmd, Elf *); +int (*sym_elf_end)(Elf *); +Elf_Data *(*sym_elf_getdata_rawchunk)(Elf *, int64_t, size_t, Elf_Type); +GElf_Ehdr *(*sym_gelf_getehdr)(Elf *, GElf_Ehdr *); +int (*sym_elf_getphdrnum)(Elf *, size_t *); +const char *(*sym_elf_errmsg)(int); +int (*sym_elf_errno)(void); +Elf *(*sym_elf_memory)(char *, size_t); +unsigned int (*sym_elf_version)(unsigned int); +GElf_Phdr *(*sym_gelf_getphdr)(Elf *, int, GElf_Phdr *); +size_t (*sym_gelf_getnote)(Elf_Data *, size_t, GElf_Nhdr *, size_t *, size_t *); + +int dlopen_dw(void) { + int r; + + r = dlopen_many_sym_or_warn( + &dw_dl, "libdw.so.1", LOG_DEBUG, + DLSYM_ARG(dwarf_getscopes), + DLSYM_ARG(dwarf_getscopes_die), + DLSYM_ARG(dwarf_tag), + DLSYM_ARG(dwarf_attr_integrate), + DLSYM_ARG(dwarf_formstring), + DLSYM_ARG(dwarf_diename), + DLSYM_ARG(dwelf_elf_gnu_build_id), + DLSYM_ARG(dwelf_elf_begin), +#if HAVE_DWELF_ELF_E_MACHINE_STRING + DLSYM_ARG(dwelf_elf_e_machine_string), +#endif + DLSYM_ARG(dwfl_addrmodule), + DLSYM_ARG(dwfl_frame_pc), + DLSYM_ARG(dwfl_module_addrdie), + DLSYM_ARG(dwfl_module_addrname), + DLSYM_ARG(dwfl_module_info), + DLSYM_ARG(dwfl_module_build_id), + DLSYM_ARG(dwfl_module_getelf), + DLSYM_ARG(dwfl_begin), + DLSYM_ARG(dwfl_core_file_report), + DLSYM_ARG(dwfl_report_end), + DLSYM_ARG(dwfl_getmodules), + DLSYM_ARG(dwfl_core_file_attach), + DLSYM_ARG(dwfl_end), + DLSYM_ARG(dwfl_errmsg), + DLSYM_ARG(dwfl_errno), + DLSYM_ARG(dwfl_build_id_find_elf), + DLSYM_ARG(dwfl_standard_find_debuginfo), + DLSYM_ARG(dwfl_thread_tid), + DLSYM_ARG(dwfl_thread_getframes), + DLSYM_ARG(dwfl_getthreads), + DLSYM_ARG(dwfl_offline_section_address)); + if (r <= 0) + return r; + + return 1; +} + +int dlopen_elf(void) { + int r; + + r = dlopen_many_sym_or_warn( + &elf_dl, "libelf.so.1", LOG_DEBUG, + DLSYM_ARG(elf_begin), + DLSYM_ARG(elf_end), + DLSYM_ARG(elf_getphdrnum), + DLSYM_ARG(elf_getdata_rawchunk), + DLSYM_ARG(elf_errmsg), + DLSYM_ARG(elf_errno), + DLSYM_ARG(elf_memory), + DLSYM_ARG(elf_version), + DLSYM_ARG(gelf_getehdr), + DLSYM_ARG(gelf_getphdr), + DLSYM_ARG(gelf_getnote)); + if (r <= 0) + return r; + + return 1; +} + +typedef struct StackContext { + MemStream m; + Dwfl *dwfl; + Elf *elf; + unsigned n_thread; + unsigned n_frame; + JsonVariant **package_metadata; + Set **modules; +} StackContext; + +static void stack_context_done(StackContext *c) { + assert(c); + + memstream_done(&c->m); + + if (c->dwfl) { + sym_dwfl_end(c->dwfl); + c->dwfl = NULL; + } + + if (c->elf) { + sym_elf_end(c->elf); + c->elf = NULL; + } +} + +DEFINE_TRIVIAL_CLEANUP_FUNC_FULL(Elf *, sym_elf_end, NULL); + +static int frame_callback(Dwfl_Frame *frame, void *userdata) { + StackContext *c = ASSERT_PTR(userdata); + Dwarf_Addr pc, pc_adjusted; + const char *fname = NULL, *symbol = NULL; + Dwfl_Module *module; + bool is_activation; + uint64_t module_offset = 0; + + assert(frame); + + if (c->n_frame >= FRAMES_MAX) + return DWARF_CB_ABORT; + + if (!sym_dwfl_frame_pc(frame, &pc, &is_activation)) + return DWARF_CB_ABORT; + + pc_adjusted = pc - (is_activation ? 0 : 1); + + module = sym_dwfl_addrmodule(c->dwfl, pc_adjusted); + if (module) { + Dwarf_Addr start, bias = 0; + Dwarf_Die *cudie; + + cudie = sym_dwfl_module_addrdie(module, pc_adjusted, &bias); + if (cudie) { + _cleanup_free_ Dwarf_Die *scopes = NULL; + int n; + + n = sym_dwarf_getscopes(cudie, pc_adjusted - bias, &scopes); + if (n > 0) + for (Dwarf_Die *s = scopes; s && s < scopes + n; s++) { + Dwarf_Attribute *a, space; + + if (!IN_SET(sym_dwarf_tag(s), DW_TAG_subprogram, DW_TAG_inlined_subroutine, DW_TAG_entry_point)) + continue; + + a = sym_dwarf_attr_integrate(s, DW_AT_MIPS_linkage_name, &space); + if (!a) + a = sym_dwarf_attr_integrate(s, DW_AT_linkage_name, &space); + if (a) + symbol = sym_dwarf_formstring(a); + if (!symbol) + symbol = sym_dwarf_diename(s); + + if (symbol) + break; + } + } + + if (!symbol) + symbol = sym_dwfl_module_addrname(module, pc_adjusted); + + fname = sym_dwfl_module_info(module, NULL, &start, NULL, NULL, NULL, NULL, NULL); + module_offset = pc - start; + } + + if (c->m.f) + fprintf(c->m.f, "#%-2u 0x%016" PRIx64 " %s (%s + 0x%" PRIx64 ")\n", c->n_frame, (uint64_t) pc, strna(symbol), strna(fname), module_offset); + c->n_frame++; + + return DWARF_CB_OK; +} + +static int thread_callback(Dwfl_Thread *thread, void *userdata) { + StackContext *c = ASSERT_PTR(userdata); + pid_t tid; + + assert(thread); + + if (c->n_thread >= THREADS_MAX) + return DWARF_CB_ABORT; + + if (c->n_thread != 0 && c->m.f) + fputc('\n', c->m.f); + + c->n_frame = 0; + + if (c->m.f) { + tid = sym_dwfl_thread_tid(thread); + fprintf(c->m.f, "Stack trace of thread " PID_FMT ":\n", tid); + } + + if (sym_dwfl_thread_getframes(thread, frame_callback, c) < 0) + return DWARF_CB_ABORT; + + c->n_thread++; + + return DWARF_CB_OK; +} + +static char* build_package_reference( + const char *type, + const char *name, + const char *version, + const char *arch) { + + /* Construct an identifier for a specific version of the package. The syntax is most suitable for + * rpm: the resulting string can be used directly in queries and rpm/dnf/yum commands. For dpkg and + * other systems, it might not be usable directly, but users should still be able to figure out the + * meaning. + */ + + return strjoin(type ?: "package", + " ", + name, + + version ? "-" : "", + strempty(version), + + /* arch is meaningful even without version, so always print it */ + arch ? "." : "", + strempty(arch)); +} + +static void report_module_metadata(StackContext *c, const char *name, JsonVariant *metadata) { + assert(c); + assert(name); + + if (!c->m.f) + return; + + fprintf(c->m.f, "Module %s", name); + + if (metadata) { + const char + *build_id = json_variant_string(json_variant_by_key(metadata, "buildId")), + *type = json_variant_string(json_variant_by_key(metadata, "type")), + *package = json_variant_string(json_variant_by_key(metadata, "name")), + *version = json_variant_string(json_variant_by_key(metadata, "version")), + *arch = json_variant_string(json_variant_by_key(metadata, "architecture")); + + if (package) { + /* Version/architecture is only meaningful with a package name. + * Skip the detailed fields if package is unknown. */ + _cleanup_free_ char *id = build_package_reference(type, package, version, arch); + fprintf(c->m.f, " from %s", strnull(id)); + } + + if (build_id && !(package && version)) + fprintf(c->m.f, ", build-id=%s", build_id); + } + + fputs("\n", c->m.f); +} + +static int parse_package_metadata(const char *name, JsonVariant *id_json, Elf *elf, bool *ret_interpreter_found, StackContext *c) { + bool interpreter_found = false; + size_t n_program_headers; + int r; + + assert(name); + assert(elf); + assert(c); + + /* When iterating over PT_LOAD we will visit modules more than once */ + if (set_contains(*c->modules, name)) + return 0; + + r = sym_elf_getphdrnum(elf, &n_program_headers); + if (r < 0) /* Not the handle we are looking for - that's ok, skip it */ + return 0; + + /* Iterate over all program headers in that ELF object. These will have been copied by + * the kernel verbatim when the core file is generated. */ + for (size_t i = 0; i < n_program_headers; ++i) { + GElf_Phdr mem, *program_header; + GElf_Nhdr note_header; + Elf_Data *data; + + /* Package metadata is in PT_NOTE headers. */ + program_header = sym_gelf_getphdr(elf, i, &mem); + if (!program_header || (program_header->p_type != PT_NOTE && program_header->p_type != PT_INTERP)) + continue; + + if (program_header->p_type == PT_INTERP) { + interpreter_found = true; + continue; + } + + /* Fortunately there is an iterator we can use to walk over the + * elements of a PT_NOTE program header. We are interested in the + * note with type. */ + data = sym_elf_getdata_rawchunk(elf, + program_header->p_offset, + program_header->p_filesz, + ELF_T_NHDR); + if (!data) + continue; + + for (size_t note_offset = 0, name_offset, desc_offset; + note_offset < data->d_size && + (note_offset = sym_gelf_getnote(data, note_offset, ¬e_header, &name_offset, &desc_offset)) > 0;) { + + _cleanup_(json_variant_unrefp) JsonVariant *v = NULL, *w = NULL; + const char *payload = (const char *)data->d_buf + desc_offset; + + if (note_header.n_namesz == 0 || note_header.n_descsz == 0) + continue; + + /* Package metadata might have different owners, but the + * magic ID is always the same. */ + if (note_header.n_type != ELF_PACKAGE_METADATA_ID) + continue; + + _cleanup_free_ char *payload_0suffixed = NULL; + assert(note_offset > desc_offset); + size_t payload_len = note_offset - desc_offset; + + /* If we are lucky and the payload is NUL-padded, we don't need to copy the string. + * But if happens to go all the way until the end of the buffer, make a copy. */ + if (payload[payload_len-1] != '\0') { + payload_0suffixed = memdup_suffix0(payload, payload_len); + if (!payload_0suffixed) + return log_oom(); + payload = payload_0suffixed; + } + + r = json_parse(payload, 0, &v, NULL, NULL); + if (r < 0) { + _cleanup_free_ char *esc = cescape(payload); + return log_error_errno(r, "json_parse on \"%s\" failed: %m", strnull(esc)); + } + + /* If we have a build-id, merge it in the same JSON object so that it appears all + * nicely together in the logs/metadata. */ + if (id_json) { + r = json_variant_merge_object(&v, id_json); + if (r < 0) + return log_error_errno(r, "json_variant_merge of package meta with buildId failed: %m"); + } + + /* Pretty-print to the buffer, so that the metadata goes as plaintext in the + * journal. */ + report_module_metadata(c, name, v); + + /* Then we build a new object using the module name as the key, and merge it + * with the previous parses, so that in the end it all fits together in a single + * JSON blob. */ + r = json_build(&w, JSON_BUILD_OBJECT(JSON_BUILD_PAIR(name, JSON_BUILD_VARIANT(v)))); + if (r < 0) + return log_error_errno(r, "Failed to build JSON object: %m"); + + r = json_variant_merge_object(c->package_metadata, w); + if (r < 0) + return log_error_errno(r, "json_variant_merge of package meta with buildId failed: %m"); + + /* Finally stash the name, so we avoid double visits. */ + r = set_put_strdup(c->modules, name); + if (r < 0) + return log_error_errno(r, "set_put_strdup failed: %m"); + + if (ret_interpreter_found) + *ret_interpreter_found = interpreter_found; + + return 1; + } + } + + if (ret_interpreter_found) + *ret_interpreter_found = interpreter_found; + + /* Didn't find package metadata for this module - that's ok, just go to the next. */ + return 0; +} + +/* Get the build-id out of an ELF object or a dwarf core module. */ +static int parse_buildid(Dwfl_Module *mod, Elf *elf, const char *name, StackContext *c, JsonVariant **ret_id_json) { + _cleanup_(json_variant_unrefp) JsonVariant *id_json = NULL; + const unsigned char *id; + GElf_Addr id_vaddr; + ssize_t id_len; + int r; + + assert(mod || elf); + assert(name); + assert(c); + + if (mod) + id_len = sym_dwfl_module_build_id(mod, &id, &id_vaddr); + else + id_len = sym_dwelf_elf_gnu_build_id(elf, (const void **)&id); + if (id_len <= 0) { + /* If we don't find a build-id, note it in the journal message, and try + * anyway to find the package metadata. It's unlikely to have the latter + * without the former, but there's no hard rule. */ + if (c->m.f) + fprintf(c->m.f, "Module %s without build-id.\n", name); + } else { + /* We will later parse package metadata json and pass it to our caller. Prepare the + * build-id in json format too, so that it can be appended and parsed cleanly. It + * will then be added as metadata to the journal message with the stack trace. */ + r = json_build(&id_json, JSON_BUILD_OBJECT(JSON_BUILD_PAIR("buildId", JSON_BUILD_HEX(id, id_len)))); + if (r < 0) + return log_error_errno(r, "json_build on buildId failed: %m"); + } + + if (ret_id_json) + *ret_id_json = TAKE_PTR(id_json); + + return 0; +} + +static int module_callback(Dwfl_Module *mod, void **userdata, const char *name, Dwarf_Addr start, void *arg) { + _cleanup_(json_variant_unrefp) JsonVariant *id_json = NULL; + StackContext *c = ASSERT_PTR(arg); + size_t n_program_headers; + GElf_Addr bias; + int r; + Elf *elf; + + assert(mod); + + if (!name) + name = "(unnamed)"; /* For logging purposes */ + + /* We are iterating on each "module", which is what dwfl calls ELF objects contained in the + * core file, and extracting the build-id first and then the package metadata. + * We proceed in a best-effort fashion - not all ELF objects might contain both or either. + * The build-id is easy, as libdwfl parses it during the sym_dwfl_core_file_report() call and + * stores it separately in an internal library struct. */ + r = parse_buildid(mod, NULL, name, c, &id_json); + if (r < 0) + return DWARF_CB_ABORT; + + /* The .note.package metadata is more difficult. From the module, we need to get a reference + * to the ELF object first. We might be lucky and just get it from elfutils. */ + elf = sym_dwfl_module_getelf(mod, &bias); + if (elf) { + r = parse_package_metadata(name, id_json, elf, NULL, c); + if (r < 0) + return DWARF_CB_ABORT; + if (r > 0) + return DWARF_CB_OK; + } else + elf = c->elf; + + /* We did not get the ELF object, or it's just a reference to the core. That is likely + * because we didn't get direct access to the executable, and the version of elfutils does + * not yet support parsing it out of the core file directly. + * So fallback to manual extraction - get the PT_LOAD section from the core, + * and if it's the right one we can interpret it as an Elf object, and parse + * its notes manually. */ + + r = sym_elf_getphdrnum(elf, &n_program_headers); + if (r < 0) { + log_warning("Could not parse number of program headers from core file: %s", + sym_elf_errmsg(-1)); /* -1 retrieves the most recent error */ + report_module_metadata(c, name, id_json); + + return DWARF_CB_OK; + } + + for (size_t i = 0; i < n_program_headers; ++i) { + GElf_Phdr mem, *program_header; + Elf_Data *data; + GElf_Addr end_of_segment; + + /* The core file stores the ELF files in the PT_LOAD segment. */ + program_header = sym_gelf_getphdr(elf, i, &mem); + if (!program_header || program_header->p_type != PT_LOAD) + continue; + + /* Check that the end of segment is a valid address. */ + if (__builtin_add_overflow(program_header->p_vaddr, program_header->p_memsz, &end_of_segment)) { + log_error("Abort due to corrupted core dump, end of segment address %#zx + %#zx overflows", (size_t)program_header->p_vaddr, (size_t)program_header->p_memsz); + return DWARF_CB_ABORT; + } + + /* This PT_LOAD segment doesn't contain the start address, so it can't be the module we are looking for. */ + if (start < program_header->p_vaddr || start >= end_of_segment) + continue; + + /* Now get a usable Elf reference, and parse the notes from it. */ + data = sym_elf_getdata_rawchunk(elf, + program_header->p_offset, + program_header->p_filesz, + ELF_T_NHDR); + if (!data) + continue; + + _cleanup_(sym_elf_endp) Elf *memelf = sym_elf_memory(data->d_buf, data->d_size); + if (!memelf) + continue; + r = parse_package_metadata(name, id_json, memelf, NULL, c); + if (r < 0) + return DWARF_CB_ABORT; + if (r > 0) + break; + } + + return DWARF_CB_OK; +} + +static int parse_core(int fd, const char *executable, char **ret, JsonVariant **ret_package_metadata) { + + const Dwfl_Callbacks callbacks = { + .find_elf = sym_dwfl_build_id_find_elf, + .section_address = sym_dwfl_offline_section_address, + .find_debuginfo = sym_dwfl_standard_find_debuginfo, + }; + + _cleanup_(json_variant_unrefp) JsonVariant *package_metadata = NULL; + _cleanup_set_free_ Set *modules = NULL; + _cleanup_(stack_context_done) StackContext c = { + .package_metadata = &package_metadata, + .modules = &modules, + }; + int r; + + assert(fd >= 0); + + if (lseek(fd, 0, SEEK_SET) < 0) + return log_warning_errno(errno, "Failed to seek to beginning of the core file: %m"); + + if (ret && !memstream_init(&c.m)) + return log_oom(); + + sym_elf_version(EV_CURRENT); + + c.elf = sym_elf_begin(fd, ELF_C_READ_MMAP, NULL); + if (!c.elf) + return log_warning_errno(SYNTHETIC_ERRNO(EINVAL), "Could not parse core file, elf_begin() failed: %s", sym_elf_errmsg(sym_elf_errno())); + + c.dwfl = sym_dwfl_begin(&callbacks); + if (!c.dwfl) + return log_warning_errno(SYNTHETIC_ERRNO(EINVAL), "Could not parse core file, dwfl_begin() failed: %s", sym_dwfl_errmsg(sym_dwfl_errno())); + + if (sym_dwfl_core_file_report(c.dwfl, c.elf, executable) < 0) + return log_warning_errno(SYNTHETIC_ERRNO(EINVAL), "Could not parse core file, dwfl_core_file_report() failed: %s", sym_dwfl_errmsg(sym_dwfl_errno())); + + if (sym_dwfl_report_end(c.dwfl, NULL, NULL) != 0) + return log_warning_errno(SYNTHETIC_ERRNO(EINVAL), "Could not parse core file, dwfl_report_end() failed: %s", sym_dwfl_errmsg(sym_dwfl_errno())); + + if (sym_dwfl_getmodules(c.dwfl, &module_callback, &c, 0) < 0) + return log_warning_errno(SYNTHETIC_ERRNO(EINVAL), "Could not parse core file, dwfl_getmodules() failed: %s", sym_dwfl_errmsg(sym_dwfl_errno())); + + if (sym_dwfl_core_file_attach(c.dwfl, c.elf) < 0) + return log_warning_errno(SYNTHETIC_ERRNO(EINVAL), "Could not parse core file, dwfl_core_file_attach() failed: %s", sym_dwfl_errmsg(sym_dwfl_errno())); + + if (sym_dwfl_getthreads(c.dwfl, thread_callback, &c) < 0) + return log_warning_errno(SYNTHETIC_ERRNO(EINVAL), "Could not parse core file, dwfl_getthreads() failed: %s", sym_dwfl_errmsg(sym_dwfl_errno())); + + if (ret) { + r = memstream_finalize(&c.m, ret, NULL); + if (r < 0) + return log_warning_errno(r, "Could not parse core file, flushing file buffer failed: %m"); + } + + if (ret_package_metadata) + *ret_package_metadata = TAKE_PTR(package_metadata); + + return 0; +} + +static int parse_elf(int fd, const char *executable, char **ret, JsonVariant **ret_package_metadata) { + _cleanup_(json_variant_unrefp) JsonVariant *package_metadata = NULL, *elf_metadata = NULL; + _cleanup_set_free_ Set *modules = NULL; + _cleanup_(stack_context_done) StackContext c = { + .package_metadata = &package_metadata, + .modules = &modules, + }; + const char *elf_type; + GElf_Ehdr elf_header; + int r; + + assert(fd >= 0); + + if (lseek(fd, 0, SEEK_SET) < 0) + return log_warning_errno(errno, "Failed to seek to beginning of the ELF file: %m"); + + if (ret && !memstream_init(&c.m)) + return log_oom(); + + sym_elf_version(EV_CURRENT); + + c.elf = sym_elf_begin(fd, ELF_C_READ_MMAP, NULL); + if (!c.elf) + return log_warning_errno(SYNTHETIC_ERRNO(EINVAL), "Could not parse ELF file, elf_begin() failed: %s", sym_elf_errmsg(sym_elf_errno())); + + if (!sym_gelf_getehdr(c.elf, &elf_header)) + return log_warning_errno(SYNTHETIC_ERRNO(EINVAL), "Could not parse ELF file, gelf_getehdr() failed: %s", sym_elf_errmsg(sym_elf_errno())); + + if (elf_header.e_type == ET_CORE) { + _cleanup_free_ char *out = NULL; + + r = parse_core(fd, executable, ret ? &out : NULL, &package_metadata); + if (r < 0) + return log_warning_errno(r, "Failed to inspect core file: %m"); + + if (out) + fprintf(c.m.f, "%s", out); + + elf_type = "coredump"; + } else { + _cleanup_(json_variant_unrefp) JsonVariant *id_json = NULL; + const char *e = executable ?: "(unnamed)"; + bool interpreter_found = false; + + r = parse_buildid(NULL, c.elf, e, &c, &id_json); + if (r < 0) + return log_warning_errno(r, "Failed to parse build-id of ELF file: %m"); + + r = parse_package_metadata(e, id_json, c.elf, &interpreter_found, &c); + if (r < 0) + return log_warning_errno(r, "Failed to parse package metadata of ELF file: %m"); + + /* If we found a build-id and nothing else, return at least that. */ + if (!package_metadata && id_json) { + r = json_build(&package_metadata, JSON_BUILD_OBJECT(JSON_BUILD_PAIR(e, JSON_BUILD_VARIANT(id_json)))); + if (r < 0) + return log_warning_errno(r, "Failed to build JSON object: %m"); + } + + if (interpreter_found) + elf_type = "executable"; + else + elf_type = "library"; + } + + /* Note that e_type is always DYN for both executables and libraries, so we can't tell them apart from the header, + * but we will search for the PT_INTERP section when parsing the metadata. */ + r = json_build(&elf_metadata, JSON_BUILD_OBJECT(JSON_BUILD_PAIR("elfType", JSON_BUILD_STRING(elf_type)))); + if (r < 0) + return log_warning_errno(r, "Failed to build JSON object: %m"); + +#if HAVE_DWELF_ELF_E_MACHINE_STRING + const char *elf_architecture = sym_dwelf_elf_e_machine_string(elf_header.e_machine); + if (elf_architecture) { + _cleanup_(json_variant_unrefp) JsonVariant *json_architecture = NULL; + + r = json_build(&json_architecture, + JSON_BUILD_OBJECT(JSON_BUILD_PAIR("elfArchitecture", JSON_BUILD_STRING(elf_architecture)))); + if (r < 0) + return log_warning_errno(r, "Failed to build JSON object: %m"); + + r = json_variant_merge_object(&elf_metadata, json_architecture); + if (r < 0) + return log_warning_errno(r, "Failed to merge JSON objects: %m"); + + if (ret) + fprintf(c.m.f, "ELF object binary architecture: %s\n", elf_architecture); + } +#endif + + /* We always at least have the ELF type, so merge that (and possibly the arch). */ + r = json_variant_merge_object(&elf_metadata, package_metadata); + if (r < 0) + return log_warning_errno(r, "Failed to merge JSON objects: %m"); + + if (ret) { + r = memstream_finalize(&c.m, ret, NULL); + if (r < 0) + return log_warning_errno(r, "Could not parse ELF file, flushing file buffer failed: %m"); + } + + if (ret_package_metadata) + *ret_package_metadata = TAKE_PTR(elf_metadata); + + return 0; +} + +int parse_elf_object(int fd, const char *executable, bool fork_disable_dump, char **ret, JsonVariant **ret_package_metadata) { + _cleanup_close_pair_ int error_pipe[2] = EBADF_PAIR, + return_pipe[2] = EBADF_PAIR, + json_pipe[2] = EBADF_PAIR; + _cleanup_(json_variant_unrefp) JsonVariant *package_metadata = NULL; + _cleanup_free_ char *buf = NULL; + int r; + + assert(fd >= 0); + + r = dlopen_dw(); + if (r < 0) + return r; + + r = dlopen_elf(); + if (r < 0) + return r; + + r = RET_NERRNO(pipe2(error_pipe, O_CLOEXEC|O_NONBLOCK)); + if (r < 0) + return r; + + if (ret) { + r = RET_NERRNO(pipe2(return_pipe, O_CLOEXEC|O_NONBLOCK)); + if (r < 0) + return r; + } + + if (ret_package_metadata) { + r = RET_NERRNO(pipe2(json_pipe, O_CLOEXEC|O_NONBLOCK)); + if (r < 0) + return r; + } + + /* Parsing possibly malformed data is crash-happy, so fork. In case we crash, + * the core file will not be lost, and the messages will still be attached to + * the journal. Reading the elf object might be slow, but it still has an upper + * bound since the core files have an upper size limit. It's also not doing any + * system call or interacting with the system in any way, besides reading from + * the file descriptor and writing into these four pipes. */ + r = safe_fork_full("(sd-parse-elf)", + NULL, + (int[]){ fd, error_pipe[1], return_pipe[1], json_pipe[1] }, + 4, + FORK_RESET_SIGNALS|FORK_CLOSE_ALL_FDS|FORK_NEW_MOUNTNS|FORK_MOUNTNS_SLAVE|FORK_NEW_USERNS|FORK_WAIT|FORK_REOPEN_LOG, + NULL); + if (r < 0) { + if (r == -EPROTO) { /* We should have the errno from the child, but don't clobber original error */ + int e, k; + + k = read(error_pipe[0], &e, sizeof(e)); + if (k < 0 && errno != EAGAIN) /* Pipe is non-blocking, EAGAIN means there's nothing */ + return -errno; + if (k == sizeof(e)) + return e; /* propagate error sent to us from child */ + if (k != 0) + return -EIO; + } + + return r; + } + if (r == 0) { + /* We want to avoid loops, given this can be called from systemd-coredump */ + if (fork_disable_dump) { + r = RET_NERRNO(prctl(PR_SET_DUMPABLE, 0)); + if (r < 0) + goto child_fail; + } + + r = parse_elf(fd, executable, ret ? &buf : NULL, ret_package_metadata ? &package_metadata : NULL); + if (r < 0) + goto child_fail; + + if (buf) { + size_t len = strlen(buf); + + if (len > COREDUMP_PIPE_MAX) { + /* This is iffy. A backtrace can be a few hundred kilobytes, but too much is + * too much. Let's log a warning and ignore the rest. */ + log_warning("Generated backtrace is %zu bytes (more than the limit of %u bytes), backtrace will be truncated.", + len, COREDUMP_PIPE_MAX); + len = COREDUMP_PIPE_MAX; + } + + /* Bump the space for the returned string. + * Failure is ignored, because partial output is still useful. */ + (void) fcntl(return_pipe[1], F_SETPIPE_SZ, len); + + r = loop_write(return_pipe[1], buf, len); + if (r == -EAGAIN) + log_warning("Write failed, backtrace will be truncated."); + else if (r < 0) + goto child_fail; + + return_pipe[1] = safe_close(return_pipe[1]); + } + + if (package_metadata) { + _cleanup_fclose_ FILE *json_out = NULL; + + /* Bump the space for the returned string. We don't know how much space we'll need in + * advance, so we'll just try to write as much as possible and maybe fail later. */ + (void) fcntl(json_pipe[1], F_SETPIPE_SZ, COREDUMP_PIPE_MAX); + + json_out = take_fdopen(&json_pipe[1], "w"); + if (!json_out) { + r = -errno; + goto child_fail; + } + + r = json_variant_dump(package_metadata, JSON_FORMAT_FLUSH, json_out, NULL); + if (r < 0) + log_warning_errno(r, "Failed to write JSON package metadata, ignoring: %m"); + } + + _exit(EXIT_SUCCESS); + + child_fail: + (void) write(error_pipe[1], &r, sizeof(r)); + _exit(EXIT_FAILURE); + } + + error_pipe[1] = safe_close(error_pipe[1]); + return_pipe[1] = safe_close(return_pipe[1]); + json_pipe[1] = safe_close(json_pipe[1]); + + if (ret) { + _cleanup_fclose_ FILE *in = NULL; + + in = take_fdopen(&return_pipe[0], "r"); + if (!in) + return -errno; + + r = read_full_stream(in, &buf, NULL); + if (r < 0) + return r; + } + + if (ret_package_metadata) { + _cleanup_fclose_ FILE *json_in = NULL; + + json_in = take_fdopen(&json_pipe[0], "r"); + if (!json_in) + return -errno; + + r = json_parse_file(json_in, NULL, 0, &package_metadata, NULL, NULL); + if (r < 0 && r != -ENODATA) /* ENODATA: json was empty, so we got nothing, but that's ok */ + log_warning_errno(r, "Failed to read or parse json metadata, ignoring: %m"); + } + + if (ret) + *ret = TAKE_PTR(buf); + if (ret_package_metadata) + *ret_package_metadata = TAKE_PTR(package_metadata); + + return 0; +} + +#endif |