/* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2008 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ /* * Copyright (c) 1988 AT&T * All Rights Reserved */ /* * Run time linker common setup. * * Called from _setup to get the process going at startup. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "_rtld.h" #include "_audit.h" #include "_elf.h" #include "_a.out.h" #include "msg.h" extern int _end, _edata, _etext; extern void _init(void); extern int _brk_unlocked(void *); #ifndef SGS_PRE_UNIFIED_PROCESS /* needed for _brk_unlocked() */ void *_nd = &_end; #endif /* * Define for the executable's interpreter. * Usually it is ld.so.1, but for the first release of ICL binaries * it is libc.so.1. We keep this information so that we don't end * up mapping libc twice if it is the interpreter. */ static Interp _interp; static int preload(const char *str, Rt_map *lmp) { Rt_map *clmp = lmp; char *objs, *ptr, *next; Word lmflags = lml_main.lm_flags; uint_t flags; DBG_CALL(Dbg_util_nl(&lml_main, DBG_NL_STD)); if ((objs = strdup(str)) == 0) return (0); /* * Establish the flags for loading each object. If we're called via * lddstub, then the first shared object is the object being inspected * by ldd(1). This object should not be marked as an interposer, as * it is intended to act like the first object of the process. */ if ((lmflags & LML_FLG_TRC_ENABLE) && (FLAGS1(lmp) & FL1_RT_LDDSTUB)) flags = FLG_RT_PRELOAD; else flags = (FLG_RT_PRELOAD | FLG_RT_OBJINTPO); ptr = strtok_r(objs, MSG_ORIG(MSG_STR_DELIMIT), &next); do { Pnode *pnp; Rt_map *nlmp = 0; DBG_CALL(Dbg_file_preload(&lml_main, ptr)); /* * If this a secure application, then preload errors are * reduced to warnings, as the errors are non-fatal. */ if (rtld_flags & RT_FL_SECURE) rtld_flags2 |= RT_FL2_FTL2WARN; if ((pnp = expand_paths(clmp, ptr, PN_FLG_EXTLOAD, 0)) != 0) nlmp = load_one(&lml_main, ALIST_OFF_DATA, pnp, clmp, MODE(lmp), flags, 0, NULL); if (pnp) remove_pnode(pnp); if (rtld_flags & RT_FL_SECURE) rtld_flags2 &= ~RT_FL2_FTL2WARN; if (nlmp && (bind_one(clmp, nlmp, BND_NEEDED) == 0)) nlmp = 0; /* * Establish state for the next preloadable object. If no * error occurred with loading this object, indicate that this * link-map list contains an interposer. */ flags |= FLG_RT_OBJINTPO; if (nlmp == 0) { if ((lmflags & LML_FLG_TRC_ENABLE) || (rtld_flags & RT_FL_SECURE)) continue; else return (0); } lml_main.lm_flags |= LML_FLG_INTRPOSE; /* * If we're tracing shared objects via lddstub, establish a * binding between the initial shared object and lddstub so that * the shared object isn't called out from unused() processing. * After the first object is loaded increment the caller to the * initial preloaded object to provide intuitive ldd -v and -s * diagnostics */ if ((lmflags & LML_FLG_TRC_ENABLE) && (FLAGS1(lmp) & FL1_RT_LDDSTUB)) { if ((lmp == clmp) && (lmflags & (LML_FLG_TRC_UNREF | LML_FLG_TRC_UNUSED))) { if (bind_one(clmp, nlmp, BND_REFER) == 0) continue; } clmp = NEXT_RT_MAP(lmp); } } while ((ptr = strtok_r(NULL, MSG_ORIG(MSG_STR_DELIMIT), &next)) != NULL); free(objs); return (1); } Rt_map * setup(char **envp, auxv_t *auxv, Word _flags, char *_platform, int _syspagsz, char *_rtldname, Dyn *dyn_ptr, ulong_t ld_base, ulong_t interp_base, int fd, Phdr *phdr, char *execname, char **argv, int dz_fd, uid_t uid, uid_t euid, gid_t gid, gid_t egid, void *aoutdyn, int auxflags, uint_t hwcap_1) { Rt_map *rlmp, *mlmp, **tobj = 0; Ehdr *ehdr; rtld_stat_t status; int features = 0, ldsoexec = 0; size_t eaddr, esize; char *str, *argvname; Mmap *mmaps; Word lmflags; /* * Now that ld.so has relocated itself, initialize our own 'environ' so * as to establish an address suitable for libc's hardware mul/div * magic (libc/sparc/crt/hwmuldiv.o). */ _environ = (char **)((ulong_t)auxv - sizeof (char *)); _init(); _environ = envp; /* * Far the most common application execution revolves around appending * the application name to the users PATH definition, thus a full name * is passed to exec() which will in turn be returned via * AT_SUN_EXECNAME. Applications may also be invoked from the current * working directory, or via a relative name. * * Determine whether the kernel has supplied a AT_SUN_EXECNAME aux * vector. This vector points to the full pathname, on the stack, of * the object that started the process. If this is null, then * AT_SUN_EXECNAME isn't supported (if the pathname exceeded the system * limit (PATH_MAX) the exec would have failed). This flag is used to * determine whether we can call resolvepath(). */ if (execname) rtld_flags |= RT_FL_EXECNAME; /* * Determine how ld.so.1 has been executed. */ if ((fd == -1) && (phdr == 0)) { /* * If we received neither the AT_EXECFD nor the AT_PHDR aux * vector, ld.so.1 must have been invoked directly from the * command line. */ ldsoexec = 1; /* * AT_SUN_EXECNAME provides the most precise name, if it is * available, otherwise fall back to argv[0]. At this time, * there is no process name. */ if (execname) rtldname = execname; else if (argv[0]) rtldname = argv[0]; else rtldname = (char *)MSG_INTL(MSG_STR_UNKNOWN); } else { /* * Otherwise, we have a standard process. AT_SUN_EXECNAME * provides the most precise name, if it is available, * otherwise fall back to argv[0]. Provided the application * is already mapped, the process is the application, so * simplify the application name for use in any diagnostics. */ if (execname) argvname = execname; else if (argv[0]) argvname = execname = argv[0]; else argvname = execname = (char *)MSG_INTL(MSG_STR_UNKNOWN); if (fd == -1) { if ((str = strrchr(argvname, '/')) != 0) procname = ++str; else procname = argvname; } /* * At this point, we don't know the runtime linkers full path * name. The _rtldname passed to us is the SONAME of the * runtime linker, which is typically /lib/ld.so.1 no matter * what the full path is. Use this for now, we'll reset the * runtime linkers name once the application is analyzed. */ if (_rtldname) { if ((str = strrchr(_rtldname, '/')) != 0) rtldname = ++str; else rtldname = _rtldname; } else rtldname = (char *)MSG_INTL(MSG_STR_UNKNOWN); } /* * Initialize any global variables. */ at_flags = _flags; if (dz_fd != FD_UNAVAIL) dz_init(dz_fd); platform = _platform; /* * If pagesize is unspecified find its value. */ if ((syspagsz = _syspagsz) == 0) syspagsz = _sysconfig(_CONFIG_PAGESIZE); fmap_setup(); /* * Add the unused portion of the last data page to the free space list. * The page size must be set before doing this. Here, _end refers to * the end of the runtime linkers bss. Note that we do not use the * unused data pages from any included .so's to supplement this free * space as badly behaved .os's may corrupt this data space, and in so * doing ruin our data. */ eaddr = S_DROUND((size_t)&_end); esize = eaddr % syspagsz; if (esize) { esize = syspagsz - esize; addfree((void *)eaddr, esize); } /* * Establish initial link-map list flags, and link-map list alists. */ if (alist_append(&lml_main.lm_lists, 0, sizeof (Lm_cntl), AL_CNT_LMLISTS) == 0) return (0); lml_main.lm_flags |= LML_FLG_BASELM; lml_main.lm_lmid = LM_ID_BASE; lml_main.lm_lmidstr = (char *)MSG_ORIG(MSG_LMID_BASE); if (alist_append(&lml_rtld.lm_lists, 0, sizeof (Lm_cntl), AL_CNT_LMLISTS) == 0) return (0); lml_rtld.lm_flags |= (LML_FLG_RTLDLM | LML_FLG_NOAUDIT | LML_FLG_HOLDLOCK); lml_rtld.lm_lmid = LM_ID_LDSO; lml_rtld.lm_lmidstr = (char *)MSG_ORIG(MSG_LMID_LDSO); /* * Determine whether we have a secure executable. */ security(uid, euid, gid, egid, auxflags); /* * Initialize a hardware capability descriptor for use in comparing * each loaded object. */ #ifdef AT_SUN_AUXFLAGS if (auxflags & AF_SUN_HWCAPVERIFY) { rtld_flags2 |= RT_FL2_HWCAP; hwcap = (ulong_t)hwcap_1; } #endif /* * Look for environment strings (allows things like LD_NOAUDIT to be * established, although debugging isn't enabled until later). */ if ((readenv_user((const char **)envp, &(lml_main.lm_flags), &(lml_main.lm_tflags), (aoutdyn != 0))) == 1) return (0); /* * Create a mapping descriptor for ld.so.1. We can determine our * two segments information from known symbols. */ if ((mmaps = calloc(3, sizeof (Mmap))) == 0) return (0); mmaps[0].m_vaddr = (caddr_t)M_PTRUNC(ld_base); mmaps[0].m_msize = (size_t)((caddr_t)&_etext - mmaps[0].m_vaddr); mmaps[0].m_fsize = mmaps[0].m_msize; mmaps[0].m_perm = (PROT_READ | PROT_EXEC); mmaps[1].m_vaddr = (caddr_t)M_PTRUNC((ulong_t)&r_debug); mmaps[1].m_msize = (size_t)((caddr_t)&_end - mmaps[1].m_vaddr); mmaps[1].m_fsize = (size_t)((caddr_t)&_edata - mmaps[1].m_vaddr); mmaps[1].m_perm = (PROT_READ | PROT_WRITE | PROT_EXEC); /* * Create a link map structure for ld.so.1. */ if ((rlmp = elf_new_lm(&lml_rtld, _rtldname, rtldname, dyn_ptr, ld_base, (ulong_t)&_etext, ALIST_OFF_DATA, (ulong_t)(eaddr - ld_base), 0, ld_base, (ulong_t)(eaddr - ld_base), mmaps, 2, NULL)) == 0) { return (0); } MODE(rlmp) |= (RTLD_LAZY | RTLD_NODELETE | RTLD_GLOBAL | RTLD_WORLD); FLAGS(rlmp) |= (FLG_RT_ANALYZED | FLG_RT_RELOCED | FLG_RT_INITDONE | FLG_RT_INITCLCT | FLG_RT_FINICLCT | FLG_RT_MODESET); /* * Initialize the runtime linkers information. */ interp = &_interp; interp->i_name = NAME(rlmp); interp->i_faddr = (caddr_t)ADDR(rlmp); ldso_plt_init(rlmp); /* * If ld.so.1 has been invoked directly, process its arguments. */ if (ldsoexec) { /* * Process any arguments that are specific to ld.so.1, and * reorganize the process stack to effectively remove ld.so.1 * from it. Reinitialize the environment pointer, as this may * have been shifted after skipping ld.so.1's arguments. */ if (rtld_getopt(argv, &envp, &auxv, &(lml_main.lm_flags), &(lml_main.lm_tflags), (aoutdyn != 0)) == 1) { eprintf(&lml_main, ERR_NONE, MSG_INTL(MSG_USG_BADOPT)); return (0); } _environ = envp; /* * Open the object that ld.so.1 is to execute. */ argvname = execname = argv[0]; if ((fd = open(argvname, O_RDONLY)) == -1) { int err = errno; eprintf(&lml_main, ERR_FATAL, MSG_INTL(MSG_SYS_OPEN), argvname, strerror(err)); return (0); } } /* * Map in the file, if exec has not already done so. If it has, * simply create a new link map structure for the executable. */ if (fd != -1) { Rej_desc rej; Fct *ftp; /* * Find out what type of object we have. */ (void) rtld_fstat(fd, &status); if ((ftp = are_u_this(&rej, fd, &status, argvname)) == 0) { Conv_reject_desc_buf_t rej_buf; eprintf(&lml_main, ERR_FATAL, MSG_INTL(err_reject[rej.rej_type]), argvname, conv_reject_desc(&rej, &rej_buf, M_MACH)); return (0); } /* * Map in object. */ if ((mlmp = (ftp->fct_map_so)(&lml_main, ALIST_OFF_DATA, execname, argvname, fd, NULL)) == 0) return (0); /* * We now have a process name for error diagnostics. */ if ((str = strrchr(argvname, '/')) != 0) procname = ++str; else procname = argvname; if (ldsoexec) { Addr brkbase = 0; /* * Since ld.so.1 was the primary executed object - the * brk() base has not yet been initialized, we need to * initialize it. For an executable, initialize it to * the end of the object. For a shared object (ET_DYN) * initialize it to the first page in memory. */ ehdr = (Ehdr *)ADDR(mlmp); if ((FCT(mlmp) == &elf_fct) && (ehdr->e_type == ET_EXEC)) { int i; Phdr *_phdr = (Phdr *)((uintptr_t)ADDR(mlmp) + ehdr->e_phoff); /* * We scan the program headers to find the tail * of the memory image. We can't use MSIZE() * since that's already been page aligned. */ for (i = 0; i < ehdr->e_phnum; i++, _phdr++) { if (_phdr->p_type == PT_LOAD) brkbase = _phdr->p_vaddr + _phdr->p_memsz; } } if (!brkbase) brkbase = syspagsz; if (_brk_unlocked((void *)brkbase) == -1) { int err = errno; eprintf(&lml_main, ERR_FATAL, MSG_INTL(MSG_SYS_BRK), argvname, strerror(err)); } } /* * The object has now been mmaped, we no longer need the file * descriptor. */ (void) close(fd); } else { /* * Set up function ptr and arguments according to the type * of file class the executable is. (Currently only supported * types are ELF and a.out format.) Then create a link map * for the executable. */ if (aoutdyn) { #ifdef A_OUT if ((mlmp = aout_new_lm(&lml_main, execname, argvname, aoutdyn, 0, 0, ALIST_OFF_DATA)) == 0) return (0); /* * Set the memory size. Note, we only know the end of * text, and although we could find the _end by looking * up the symbol, this may not be present. We should * set ADDR to MAIN_BASE, but presently all the a.out * relocation code assumes ADDR is 0 for the dynamic * executable. (these data items are only used for * dladdr(3x), and there aren't many a.out dladdr(3x) * users to warrant spending much time on this :-). */ MSIZE(mlmp) = MAIN_BASE + ETEXT(mlmp); /* * Disable any object configuration cache (BCP apps * bring in sbcp which can benefit from any object * cache, but both the app and sbcp can't use the same * objects). */ rtld_flags |= RT_FL_NOOBJALT; /* * Make sure no-direct bindings are in effect. */ lml_main.lm_tflags |= LML_TFLG_NODIRECT; #else eprintf(&lml_main, ERR_FATAL, MSG_INTL(MSG_ERR_REJ_UNKFILE), argvname); return (0); #endif } else if (phdr) { Phdr *pptr, *firstptr = 0, *lastptr; Phdr *tlsphdr = 0, *unwindphdr = 0; Dyn *dyn = 0; Cap *cap = 0; Off i_offset = 0; Addr base = 0; ulong_t memsize, phsize, entry, etext; uint_t mmapcnt = 0; int i; /* * Using the executables phdr address determine the base * address of the input file. NOTE, this assumes the * program headers and elf header are part of the same * mapped segment. Although this has held for many * years now, it might be more flexible if the kernel * gave use the ELF headers start address, rather than * the Program headers. * * Determine from the ELF header if we're been called * from a shared object or dynamic executable. If the * latter, then any addresses within the object are used * as is. Addresses within shared objects must be added * to the process's base address. */ ehdr = (Ehdr *)((Addr)phdr - phdr->p_offset); phsize = ehdr->e_phentsize; if (ehdr->e_type == ET_DYN) base = (Addr)ehdr; /* * Allocate a mapping array to retain mapped segment * information. */ if ((mmaps = calloc(ehdr->e_phnum, sizeof (Mmap))) == 0) return (0); /* * Extract the needed information from the segment * headers. */ for (i = 0, pptr = phdr; i < ehdr->e_phnum; i++) { if (pptr->p_type == PT_INTERP) { i_offset = pptr->p_offset; interp->i_faddr = (caddr_t)interp_base; } if ((pptr->p_type == PT_LOAD) && (pptr->p_filesz || pptr->p_memsz)) { int perm = (PROT_READ | PROT_EXEC); size_t off; if (!firstptr) firstptr = pptr; lastptr = pptr; if (i_offset && pptr->p_filesz && (i_offset >= pptr->p_offset) && (i_offset <= (pptr->p_memsz + pptr->p_offset))) { interp->i_name = (char *) pptr->p_vaddr + i_offset - pptr->p_offset + base; i_offset = 0; } if ((pptr->p_flags & (PF_R | PF_W)) == PF_R) etext = pptr->p_vaddr + pptr->p_memsz + base; else perm |= PROT_WRITE; /* * Retain segments mapping info. Round * each segment to a page boundary, as * this insures addresses are suitable * for mprotect() if required. */ off = pptr->p_vaddr + base; mmaps[mmapcnt].m_vaddr = (caddr_t)M_PTRUNC(off); off -= (size_t)mmaps[mmapcnt].m_vaddr; mmaps[mmapcnt].m_msize = pptr->p_memsz + off; mmaps[mmapcnt].m_fsize = pptr->p_filesz + off; mmaps[mmapcnt].m_perm = perm; mmapcnt++; } else if (pptr->p_type == PT_DYNAMIC) { dyn = (Dyn *)(pptr->p_vaddr + base); } else if ((pptr->p_type == PT_TLS) && pptr->p_memsz) { tlsphdr = pptr; } else if (pptr->p_type == PT_SUNW_UNWIND) { unwindphdr = pptr; } else if (pptr->p_type == PT_SUNWCAP) { cap = (Cap *)(pptr->p_vaddr + base); } pptr = (Phdr *)((ulong_t)pptr + phsize); } memsize = (lastptr->p_vaddr + lastptr->p_memsz) - S_ALIGN(firstptr->p_vaddr, syspagsz); entry = ehdr->e_entry; if (ehdr->e_type == ET_DYN) entry += (ulong_t)ehdr; if ((mlmp = elf_new_lm(&lml_main, execname, argvname, dyn, (Addr)ehdr, etext, ALIST_OFF_DATA, memsize, entry, (ulong_t)ehdr, memsize, mmaps, mmapcnt, NULL)) == 0) { return (0); } if (tlsphdr && (tls_assign(&lml_main, mlmp, tlsphdr) == 0)) return (0); if (unwindphdr) PTUNWIND(mlmp) = unwindphdr; if (cap) cap_assign(cap, mlmp); } } /* * Establish the interpretors name as that defined within the initial * object (executable). This provides for ORIGIN processing of ld.so.1 * dependencies. */ if (ldsoexec == 0) { size_t len = strlen(interp->i_name); (void) expand(&interp->i_name, &len, 0, 0, (PN_TKN_ISALIST | PN_TKN_HWCAP), rlmp); } PATHNAME(rlmp) = interp->i_name; if (FLAGS1(rlmp) & FL1_RT_RELATIVE) (void) fullpath(rlmp, 0); else ORIGNAME(rlmp) = PATHNAME(rlmp) = NAME(rlmp); /* * Having established the true runtime linkers name, simplify the name * for error diagnostics. */ if ((str = strrchr(PATHNAME(rlmp), '/')) != 0) rtldname = ++str; else rtldname = PATHNAME(rlmp); /* * Expand the fullpath name of the application. This typically occurs * as a part of loading an object, but as the kernel probably mapped * it in, complete this processing now. */ if (FLAGS1(mlmp) & FL1_RT_RELATIVE) (void) fullpath(mlmp, 0); /* * Some troublesome programs will change the value of argv[0]. Dupping * the process string protects us, and insures the string is left in * any core files. */ if ((str = (char *)strdup(procname)) == 0) return (0); procname = str; /* * If the kernel has provided hardware capabilities information, and * the executable contains hardware capabilities information, make * sure it's a valid object. */ if ((rtld_flags2 & RT_FL2_HWCAP) && HWCAP(mlmp)) { ulong_t mhwcap; if ((mhwcap = (HWCAP(mlmp) & ~hwcap)) != 0) { Conv_cap_val_hw1_buf_t cap_val_hw1_buf; const char *str = conv_cap_val_hw1(mhwcap, M_MACH, 0, &cap_val_hw1_buf); if (lml_main.lm_flags & LML_FLG_TRC_ENABLE) { (void) printf(MSG_INTL(MSG_LDD_GEN_HWCAP_1), NAME(mlmp), str); } else { eprintf(&lml_main, ERR_FATAL, MSG_INTL(MSG_GEN_BADHWCAP_1), str); return (0); } } } #if defined(_ELF64) /* * If this is a 64-bit process, determine whether this process has * restricted the process address space to 32-bits. Any dependencies * that are restricted to a 32-bit address space can only be loaded if * the executable has established this requirement. */ if (SFCAP(mlmp) & SF1_SUNW_ADDR32) rtld_flags2 |= RT_FL2_ADDR32; #endif FLAGS(mlmp) |= (FLG_RT_ISMAIN | FLG_RT_MODESET); FLAGS1(mlmp) |= FL1_RT_USED; /* * It's the responsibility of MAIN(crt0) to call it's _init and _fini * section, therefore null out any INIT/FINI so that this object isn't * collected during tsort processing. And, if the application has no * initarray or finiarray we can economize on establishing bindings. */ INIT(mlmp) = FINI(mlmp) = 0; if ((INITARRAY(mlmp) == 0) && (FINIARRAY(mlmp) == 0)) FLAGS1(mlmp) |= FL1_RT_NOINIFIN; /* * Identify lddstub if necessary. */ if (lml_main.lm_flags & LML_FLG_TRC_LDDSTUB) FLAGS1(mlmp) |= FL1_RT_LDDSTUB; /* * Retain our argument information for use in dlinfo. */ argsinfo.dla_argv = argv--; argsinfo.dla_argc = (long)*argv; argsinfo.dla_envp = envp; argsinfo.dla_auxv = auxv; (void) enter(0); /* * Add our two main link-maps to the dynlm_list */ if (list_append(&dynlm_list, &lml_main) == 0) return (0); if (list_append(&dynlm_list, &lml_rtld) == 0) return (0); /* * Reset the link-map counts for both lists. The init count is used to * track how many objects have pending init sections, this gets incre- * mented each time an object is relocated. Since ld.so.1 relocates * itself, it's init count will remain zero. * The object count is used to track how many objects have pending fini * sections, as ld.so.1 handles its own fini we can zero its count. */ lml_main.lm_obj = 1; lml_rtld.lm_obj = 0; /* * Initialize debugger information structure. Some parts of this * structure were initialized statically. */ r_debug.rtd_rdebug.r_map = (Link_map *)lml_main.lm_head; r_debug.rtd_rdebug.r_ldsomap = (Link_map *)lml_rtld.lm_head; r_debug.rtd_rdebug.r_ldbase = r_debug.rtd_rdebug.r_ldsomap->l_addr; r_debug.rtd_dynlmlst = &dynlm_list; if (platform) platform_sz = strlen(platform); /* * Determine the dev/inode information for the executable to complete * load_so() checking for those who might dlopen(a.out). */ if ((FLAGS1(mlmp) & FL1_RT_RELATIVE) && (rtld_stat(PATHNAME(mlmp), &status) == 0)) { STDEV(mlmp) = status.st_dev; STINO(mlmp) = status.st_ino; } /* * Initialize any configuration information. */ if (!(rtld_flags & RT_FL_NOCFG)) { if ((features = elf_config(mlmp, (aoutdyn != 0))) == -1) return (0); } /* * Establish the modes of the initial object. These modes are * propagated to any preloaded objects and explicit shared library * dependencies. * * If we're generating a configuration file using crle(1), remove * any RTLD_NOW use, as we don't want to trigger any relocation proc- * essing during crle(1)'s first past (this would just be unnecessary * overhead). Any filters are explicitly loaded, and thus RTLD_NOW is * not required to trigger filter loading. * * Note, RTLD_NOW may have been established during analysis of the * application had the application been built -z now. */ MODE(mlmp) |= (RTLD_NODELETE | RTLD_GLOBAL | RTLD_WORLD); if (rtld_flags & RT_FL_CONFGEN) { MODE(mlmp) |= RTLD_CONFGEN; MODE(mlmp) &= ~RTLD_NOW; rtld_flags2 &= ~RT_FL2_BINDNOW; } if ((MODE(mlmp) & RTLD_NOW) == 0) { if (rtld_flags2 & RT_FL2_BINDNOW) MODE(mlmp) |= RTLD_NOW; else MODE(mlmp) |= RTLD_LAZY; } /* * If debugging was requested initialize things now that any cache has * been established. A user can specify LD_DEBUG=help to discover the * list of debugging tokens available without running the application. * However, don't allow this setting from a configuration file. * * Note, to prevent recursion issues caused by loading and binding the * debugging libraries themselves, a local debugging descriptor is * initialized. Once the debugging setup has completed, this local * descriptor is copied to the global descriptor which effectively * enables diagnostic output. */ if (rpl_debug || prm_debug) { Dbg_desc _dbg_desc = {0, 0, 0}; if (rpl_debug) { uintptr_t ret; if ((ret = dbg_setup(rpl_debug, &_dbg_desc)) == S_ERROR) return (0); if (ret == 0) rtldexit(&lml_main, 0); } if (prm_debug) (void) dbg_setup(prm_debug, &_dbg_desc); *dbg_desc = _dbg_desc; } /* * Now that debugging is enabled generate any diagnostics from any * previous events. */ if (hwcap) DBG_CALL(Dbg_cap_val_hw1(&lml_main, hwcap, M_MACH)); if (features) DBG_CALL(Dbg_file_config_dis(&lml_main, config->c_name, features)); if (DBG_ENABLED) { DBG_CALL(Dbg_file_ldso(rlmp, envp, auxv, LIST(rlmp)->lm_lmidstr, ALIST_OFF_DATA)); if (FCT(mlmp) == &elf_fct) { DBG_CALL(Dbg_file_elf(&lml_main, PATHNAME(mlmp), (ulong_t)DYN(mlmp), ADDR(mlmp), MSIZE(mlmp), ENTRY(mlmp), LIST(mlmp)->lm_lmidstr, ALIST_OFF_DATA)); } else { DBG_CALL(Dbg_file_aout(&lml_main, PATHNAME(mlmp), (ulong_t)AOUTDYN(mlmp), (ulong_t)ADDR(mlmp), (ulong_t)MSIZE(mlmp), LIST(mlmp)->lm_lmidstr, ALIST_OFF_DATA)); } } /* * Enable auditing. */ if (rpl_audit || prm_audit || profile_lib) { int ndx; const char *aud[3]; aud[0] = rpl_audit; aud[1] = prm_audit; aud[2] = profile_lib; /* * Any global auditing (set using LD_AUDIT or LD_PROFILE) that * can't be established is non-fatal. */ if ((auditors = calloc(1, sizeof (Audit_desc))) == 0) return (0); for (ndx = 0; ndx < 3; ndx++) { if (aud[ndx]) { if ((auditors->ad_name = strdup(aud[ndx])) == 0) return (0); rtld_flags2 |= RT_FL2_FTL2WARN; (void) audit_setup(mlmp, auditors, PN_FLG_EXTLOAD, NULL); rtld_flags2 &= ~RT_FL2_FTL2WARN; } } lml_main.lm_tflags |= auditors->ad_flags; } if (AUDITORS(mlmp)) { /* * Any object required auditing (set with a DT_DEPAUDIT dynamic * entry) that can't be established is fatal. */ if (FLAGS1(mlmp) & FL1_RT_GLOBAUD) { /* * If this object requires global auditing, use the * local auditing information to set the global * auditing descriptor. The effect is that a * DT_DEPAUDIT act as an LD_AUDIT. */ if ((auditors == 0) && ((auditors = calloc(1, sizeof (Audit_desc))) == 0)) return (0); auditors->ad_name = AUDITORS(mlmp)->ad_name; if (audit_setup(mlmp, auditors, 0, NULL) == 0) return (0); lml_main.lm_tflags |= auditors->ad_flags; /* * Clear the local auditor information. */ free((void *) AUDITORS(mlmp)); AUDITORS(mlmp) = 0; } else { /* * Establish any local auditing. */ if (audit_setup(mlmp, AUDITORS(mlmp), 0, NULL) == 0) return (0); FLAGS1(mlmp) |= AUDITORS(mlmp)->ad_flags; lml_main.lm_flags |= LML_FLG_LOCAUDIT; } } /* * Explicitly add the initial object and ld.so.1 to those objects being * audited. Note, although the ld.so.1 link-map isn't auditable, * establish a cookie for ld.so.1 as this may be bound to via the * dl*() family. */ if ((lml_main.lm_tflags | FLAGS1(mlmp)) & LML_TFLG_AUD_MASK) { if (((audit_objopen(mlmp, mlmp) == 0) || (audit_objopen(mlmp, rlmp) == 0)) && (FLAGS1(mlmp) & LML_TFLG_AUD_MASK)) return (0); } /* * Map in any preloadable shared objects. Note, it is valid to preload * a 4.x shared object with a 5.0 executable (or visa-versa), as this * functionality is required by ldd(1). */ if (rpl_preload && (preload(rpl_preload, mlmp) == 0)) return (0); if (prm_preload && (preload(prm_preload, mlmp) == 0)) return (0); /* * Load all dependent (needed) objects. */ if (analyze_lmc(&lml_main, ALIST_OFF_DATA, mlmp, NULL) == 0) return (0); /* * Relocate all the dependencies we've just added. * * If this process has been established via crle(1), the environment * variable LD_CONFGEN will have been set. crle(1) may create this * process twice. The first time crle only needs to gather dependency * information. The second time, is to dldump() the images. * * If we're only gathering dependencies, relocation is unnecessary. * As crle(1) may be building an arbitrary family of objects, they may * not fully relocate either. Hence the relocation phase is not carried * out now, but will be called by crle(1) once all objects have been * loaded. */ if ((rtld_flags & RT_FL_CONFGEN) == 0) { DBG_CALL(Dbg_util_nl(&lml_main, DBG_NL_STD)); if (relocate_lmc(&lml_main, ALIST_OFF_DATA, mlmp, mlmp, NULL) == 0) return (0); /* * Inform the debuggers we're here and stable. Newer debuggers * can indicate their presence by setting the DT_DEBUG entry in * the dynamic executable (see elf_new_lm()). In this case call * getpid() so the debugger can catch the system call. This * handshake allows the debugger to initialize, and consequently * allows the user to set break points in .init code. */ rd_event(&lml_main, RD_DLACTIVITY, RT_CONSISTENT); rd_event(&lml_rtld, RD_DLACTIVITY, RT_CONSISTENT); if (rtld_flags & RT_FL_DEBUGGER) { r_debug.rtd_rdebug.r_flags |= RD_FL_ODBG; (void) getpid(); } } /* * Indicate preinit activity, and call any auditing routines. These * routines are called before initializing any threads via libc, or * before collecting the complete set of .inits on the primary link-map. * Although most libc interfaces are encapsulated in local routines * within libc, they have been known to escape (ie. call a .plt). As * the appcert auditor uses preinit as a trigger to establish some * external interfaces to the main link-maps libc, we need to activate * this trigger before exercising any code within libc. Additionally, * I wouldn't put it past an auditor to add additional objects to the * primary link-map. Hence, we collect .inits after the audit call. */ rd_event(&lml_main, RD_PREINIT, 0); if ((lml_main.lm_tflags | FLAGS1(mlmp)) & LML_TFLG_AUD_ACTIVITY) audit_activity(mlmp, LA_ACT_CONSISTENT); if ((lml_main.lm_tflags | FLAGS1(mlmp)) & LML_TFLG_AUD_PREINIT) audit_preinit(mlmp); /* * If we're creating initial configuration information, we're done * now that the auditing step has been called. */ if (rtld_flags & RT_FL_CONFGEN) { leave(LIST(mlmp), 0); return (mlmp); } /* * Sort the .init sections of all objects we've added. If we're * tracing we only need to execute this under ldd(1) with the -i or -u * options. */ lmflags = lml_main.lm_flags; if (((lmflags & LML_FLG_TRC_ENABLE) == 0) || (lmflags & (LML_FLG_TRC_INIT | LML_FLG_TRC_UNREF))) { if ((tobj = tsort(mlmp, LIST(mlmp)->lm_init, RT_SORT_REV)) == (Rt_map **)S_ERROR) return (0); } /* * If we are tracing we're done. This is the one legitimate use of a * direct call to rtldexit() rather than return, as we don't want to * return and jump to the application. */ if (lmflags & LML_FLG_TRC_ENABLE) { unused(&lml_main); rtldexit(&lml_main, 0); } #ifdef AT_SUN_AUXFLAGS /* * Check if this instance of the linker should have a primary link * map. This flag allows multiple copies of the -same- -version- * of the linker (and libc) to run in the same address space. * * Without this flag we only support one copy of the linker in a * process because by default the linker will always try to * initialize at one primary link map The copy of libc which is * initialized on a primary link map will initalize global TLS * data which can be shared with other copies of libc in the * process. The problem is that if there is more than one copy * of the linker, only one copy should link libc onto a primary * link map, otherwise libc will attempt to re-initialize global * TLS data. So when a copy of the linker is loaded with this * flag set, it will not initialize any primary link maps since * persumably another copy of the linker will do this. * * Note that this flag only allows multiple copies of the -same- * -version- of the linker (and libc) to coexist. This approach * will not work if we are trying to load different versions of * the linker and libc into the same process. The reason for * this is that the format of the global TLS data may not be * the same for different versions of libc. In this case each * different version of libc must have it's own primary link map * and be able to maintain it's own TLS data. The only way this * can be done is by carefully managing TLS pointers on transitions * between code associated with each of the different linkers. * Note that this is actually what is done for processes in lx * branded zones. Although in the lx branded zone case, the * other linker and libc are actually gld and glibc. But the * same general TLS management mechanism used by the lx brand * would apply to any attempts to run multiple versions of the * solaris linker and libc in a single process. */ if (auxflags & AF_SUN_NOPLM) rtld_flags2 |= RT_FL2_NOPLM; #endif /* * Establish any static TLS for this primary link-map. Note, regardless * of whether TLS is available, an initial handshake occurs with libc to * indicate we're processing the primary link-map. Having identified * the primary link-map, initialize threads. */ if (rt_get_extern(&lml_main, mlmp) == 0) return (0); if ((rtld_flags2 & RT_FL2_NOPLM) == 0) { if (tls_statmod(&lml_main, mlmp) == 0) return (0); rt_thr_init(&lml_main); rtld_flags2 |= RT_FL2_PLMSETUP; } else { rt_thr_init(&lml_main); } rtld_flags |= RT_FL_APPLIC; /* * Fire all dependencies .init sections. Identify any unused * dependencies, and leave the runtime linker - effectively calling * the dynamic executables entry point. */ call_array(PREINITARRAY(mlmp), (uint_t)PREINITARRAYSZ(mlmp), mlmp, SHT_PREINIT_ARRAY); if (tobj) call_init(tobj, DBG_INIT_SORT); rd_event(&lml_main, RD_POSTINIT, 0); unused(&lml_main); DBG_CALL(Dbg_util_call_main(mlmp)); rtld_flags |= RT_FL_OPERATION; leave(LIST(mlmp), 0); return (mlmp); }