/* * 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 (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved. * Copyright 2014 Nexenta Systems, Inc. */ #include #include #include #include #include #include <_libld.h> #include #include /* * A EH_FRAME_HDR consists of the following: * * Encoding Field * -------------------------------- * unsigned byte version * unsigned byte eh_frame_ptr_enc * unsigned byte fde_count_enc * unsigned byte table_enc * encoded eh_frame_ptr * encoded fde_count * [ binary search table ] * * The binary search table entries each consists of: * * encoded initial_func_loc * encoded FDE_address * * The entries in the binary search table are sorted * in a increasing order by the initial location. * * * version * * Version of the .eh_frame_hdr format. This value shall be 1. * * eh_frame_ptr_enc * * The encoding format of the eh_frame_ptr field. For shared * libraries the encoding must be * DW_EH_PE_sdata4|DW_EH_PE_pcrel or * DW_EH_PE_sdata4|DW_EH_PE_datarel. * * * fde_count_enc * * The encoding format of the fde_count field. A value of * DW_EH_PE_omit indicates the binary search table is not * present. * * table_enc * * The encoding format of the entries in the binary search * table. A value of DW_EH_PE_omit indicates the binary search * table is not present. For shared libraries the encoding * must be DW_EH_PE_sdata4|DW_EH_PE_pcrel or * DW_EH_PE_sdata4|DW_EH_PE_datarel. * * * eh_frame_ptr * * The encoded value of the pointer to the start of the * .eh_frame section. * * fde_count * * The encoded value of the count of entries in the binary * search table. * * binary search table * * A binary search table containing fde_count entries. Each * entry of the table consist of two encoded values, the * initial location of the function to which an FDE applies, * and the address of the FDE. The entries are sorted in an * increasing order by the initial location value. * */ /* * EH_FRAME sections * ================= * * The call frame information needed for unwinding the stack is output in * an ELF section(s) of type SHT_AMD64_UNWIND (amd64) or SHT_PROGBITS (other). * In the simplest case there will be one such section per object file and it * will be named ".eh_frame". An .eh_frame section consists of one or more * subsections. Each subsection contains a CIE (Common Information Entry) * followed by varying number of FDEs (Frame Descriptor Entry). A FDE * corresponds to an explicit or compiler generated function in a * compilation unit, all FDEs can access the CIE that begins their * subsection for data. * * If an object file contains C++ template instantiations, there shall be * a separate CIE immediately preceding each FDE corresponding to an * instantiation. * * Using the preferred encoding specified below, the .eh_frame section can * be entirely resolved at link time and thus can become part of the * text segment. * * .eh_frame Section Layout * ------------------------ * * EH_PE encoding below refers to the pointer encoding as specified in the * enhanced LSB Chapter 7 for Eh_Frame_Hdr. * * Common Information Entry (CIE) * ------------------------------ * CIE has the following format: * * Length * in * Field Byte Description * ----- ------ ----------- * 1. Length 4 Length of CIE (not including * this 4-byte field). * * 2. CIE id 4 Value Zero (0) for .eh_frame * (used to distinguish CIEs and * FDEs when scanning the section) * * 3. Version 1 Value One (1) * * 4. CIE Augmentation string Null-terminated string with legal * values being "" or 'z' optionally * followed by single occurrences of * 'P', 'L', or 'R' in any order. * String The presence of character(s) in the * string dictates the content of * field 8, the Augmentation Section. * Each character has one or two * associated operands in the AS. * Operand order depends on * position in the string ('z' must * be first). * * 5. Code Align Factor uleb128 To be multiplied with the * "Advance Location" instructions in * the Call Frame Instructions * * 6. Data Align Factor sleb128 To be multiplied with all offset * in the Call Frame Instructions * * 7. Ret Address Reg 1 A "virtual" register representation * of the return address. In Dwarf V2, * this is a byte, otherwise it is * uleb128. It is a byte in gcc 3.3.x * * 8. Optional CIE varying Present if Augmentation String in * Augmentation Section field 4 is not 0. * * z: * size uleb128 Length of the remainder of the * Augmentation Section * * P: * personality_enc 1 Encoding specifier - preferred * value is a pc-relative, signed * 4-byte * * * personality routine (encoded) Encoded pointer to personality * routine (actually to the PLT * entry for the personality * routine) * R: * code_enc 1 Non-default encoding for the * code-pointers (FDE members * "initial_location" and "address_range" * and the operand for DW_CFA_set_loc) * - preferred value is pc-relative, * signed 4-byte. * L: * lsda_enc 1 FDE augmentation bodies may contain * LSDA pointers. If so they are * encoded as specified here - * preferred value is pc-relative, * signed 4-byte possibly indirect * thru a GOT entry. * * * 9. Optional Call Frame varying * Instructions * * The size of the optional call frame instruction area must be computed * based on the overall size and the offset reached while scanning the * preceding fields of the CIE. * * * Frame Descriptor Entry (FDE) * ---------------------------- * FDE has the following format: * * Length * in * Field Byte Description * ----- ------ ----------- * 1. Length 4 Length of remainder of this FDE * * 2. CIE Pointer 4 Distance from this field to the * nearest preceding CIE * (uthe value is subtracted from the * current address). This value * can never be zero and thus can * be used to distinguish CIE's and * FDE's when scanning the * .eh_frame section * * 3. Initial Location varying Reference to the function code * corresponding to this FDE. * If 'R' is missing from the CIE * Augmentation String, the field is an * 8-byte absolute pointer. Otherwise, * the corresponding EH_PE encoding in the * CIE Augmentation Section is used to * interpret the reference. * * 4. Address Range varying Size of the function code corresponding * to this FDE. * If 'R' is missing from the CIE * Augmentation String, the field is an * 8-byte unsigned number. Otherwise, * the size is determined by the * corresponding EH_PE encoding in the * CIE Augmentation Section (the * value is always absolute). * * 5. Optional FDE varying present if CIE augmentation * Augmentation Section string is non-empty. * * * 'z': * length uleb128 length of the remainder of the * FDE augmentation section * * * 'L' (and length > 0): * LSDA varying LSDA pointer, encoded in the * format specified by the * corresponding operand in the CIE's * augmentation body. * * 6. Optional Call varying * Frame Instructions * * The size of the optional call frame instruction area must be computed * based on the overall size and the offset reached while scanning the * preceding fields of the FDE. * * The overall size of a .eh_frame section is given in the ELF section * header. The only way to determine the number of entries is to scan * the section till the end and count. * */ static uint_t extract_uint(const uchar_t *data, uint64_t *ndx, int do_swap) { uint_t r; uchar_t *p = (uchar_t *)&r; data += *ndx; if (do_swap) UL_ASSIGN_BSWAP_WORD(p, data); else UL_ASSIGN_WORD(p, data); (*ndx) += 4; return (r); } /* * Create an unwind header (.eh_frame_hdr) output section. * The section is created and space reserved, but the data * is not copied into place. That is done by a later call * to ld_unwind_populate(), after active relocations have been * processed. * * When GNU linkonce processing is in effect, we can end up in a situation * where the FDEs related to discarded sections remain in the eh_frame * section. Ideally, we would remove these dead entries from eh_frame. * However, that optimization has not yet been implemented. In the current * implementation, the number of dead FDEs cannot be determined until * active relocations are processed, and that processing follows the * call to this function. This means that we are unable to detect dead FDEs * here, and the section created by this routine is sized for maximum case * where all FDEs are valid. */ uintptr_t ld_unwind_make_hdr(Ofl_desc *ofl) { int bswap = (ofl->ofl_flags1 & FLG_OF1_ENCDIFF) != 0; Shdr *shdr; Elf_Data *elfdata; Is_desc *isp; size_t size; Xword fde_cnt; Aliste idx1; Os_desc *osp; /* * we only build a unwind header if we have * some unwind information in the file. */ if (aplist_nitems(ofl->ofl_unwind) == 0) return (1); /* * Allocate and initialize the Elf_Data structure. */ if ((elfdata = libld_calloc(1, sizeof (Elf_Data))) == NULL) return (S_ERROR); elfdata->d_type = ELF_T_BYTE; elfdata->d_align = ld_targ.t_m.m_word_align; elfdata->d_version = ofl->ofl_dehdr->e_version; /* * Allocate and initialize the Shdr structure. */ if ((shdr = libld_calloc(1, sizeof (Shdr))) == NULL) return (S_ERROR); shdr->sh_type = ld_targ.t_m.m_sht_unwind; shdr->sh_flags = SHF_ALLOC; shdr->sh_addralign = ld_targ.t_m.m_word_align; shdr->sh_entsize = 0; /* * Allocate and initialize the Is_desc structure. */ if ((isp = libld_calloc(1, sizeof (Is_desc))) == NULL) return (S_ERROR); isp->is_name = MSG_ORIG(MSG_SCN_UNWINDHDR); isp->is_shdr = shdr; isp->is_indata = elfdata; if ((ofl->ofl_unwindhdr = ld_place_section(ofl, isp, NULL, ld_targ.t_id.id_unwindhdr, NULL)) == (Os_desc *)S_ERROR) return (S_ERROR); /* * Scan through all of the input Frame information, counting each FDE * that requires an index. Each fde_entry gets a corresponding entry * in the binary search table. */ fde_cnt = 0; for (APLIST_TRAVERSE(ofl->ofl_unwind, idx1, osp)) { Aliste idx2; int os_isdescs_idx; OS_ISDESCS_TRAVERSE(os_isdescs_idx, osp, idx2, isp) { uchar_t *data; uint64_t off = 0; data = isp->is_indata->d_buf; size = isp->is_indata->d_size; while (off < size) { uint_t length, id; uint64_t ndx = 0; /* * Extract length in lsb format. A zero length * indicates that this CIE is a terminator and * that processing for unwind information is * complete. */ length = extract_uint(data + off, &ndx, bswap); if (length == 0) break; /* * Extract CIE id in lsb format. */ id = extract_uint(data + off, &ndx, bswap); /* * A CIE record has a id of '0', otherwise * this is a FDE entry and the 'id' is the * CIE pointer. */ if (id == 0) { uint_t cieversion; cieversion = data[off + ndx]; ndx += 1; /* BEGIN CSTYLED */ if (cieversion != 1 && cieversion != 3) { ld_eprintf(ofl, ERR_FATAL, MSG_INTL(MSG_UNW_BADCIEVERS), isp->is_file->ifl_name, isp->is_name, off); return (S_ERROR); } /* END CSTYLED */ } else { fde_cnt++; } off += length + 4; } } } /* * section size: * byte version +1 * byte eh_frame_ptr_enc +1 * byte fde_count_enc +1 * byte table_enc +1 * 4 bytes eh_frame_ptr +4 * 4 bytes fde_count +4 * [4 bytes] [4bytes] * fde_count ... */ size = 12 + (8 * fde_cnt); if ((elfdata->d_buf = libld_calloc(size, 1)) == NULL) return (S_ERROR); elfdata->d_size = size; shdr->sh_size = (Xword)size; return (1); } /* * the comparator function needs to calculate * the actual 'initloc' of a bintab entry - to * do this we initialize the following global to point * to it. */ static Addr framehdr_addr; static int bintabcompare(const void *p1, const void *p2) { uint_t *bintab1, *bintab2; uint_t ent1, ent2; bintab1 = (uint_t *)p1; bintab2 = (uint_t *)p2; assert(bintab1 != 0); assert(bintab2 != 0); ent1 = *bintab1 + framehdr_addr; ent2 = *bintab2 + framehdr_addr; if (ent1 > ent2) return (1); if (ent1 < ent2) return (-1); return (0); } uintptr_t ld_unwind_populate_hdr(Ofl_desc *ofl) { uchar_t *hdrdata; uint_t *binarytable; uint_t hdroff; Aliste idx; Addr hdraddr; Os_desc *hdrosp; Os_desc *osp; Os_desc *first_unwind; uint_t fde_count; uint_t *uint_ptr; int bswap = (ofl->ofl_flags1 & FLG_OF1_ENCDIFF) != 0; /* * Are we building the unwind hdr? */ if ((hdrosp = ofl->ofl_unwindhdr) == 0) return (1); hdrdata = hdrosp->os_outdata->d_buf; hdraddr = hdrosp->os_shdr->sh_addr; hdroff = 0; /* * version == 1 */ hdrdata[hdroff++] = 1; /* * The encodings are: * * eh_frameptr_enc sdata4 | pcrel * fde_count_enc udata4 * table_enc sdata4 | datarel */ hdrdata[hdroff++] = DW_EH_PE_sdata4 | DW_EH_PE_pcrel; hdrdata[hdroff++] = DW_EH_PE_udata4; hdrdata[hdroff++] = DW_EH_PE_sdata4 | DW_EH_PE_datarel; /* * Header Offsets * ----------------------------------- * byte version +1 * byte eh_frame_ptr_enc +1 * byte fde_count_enc +1 * byte table_enc +1 * 4 bytes eh_frame_ptr +4 * 4 bytes fde_count +4 */ /* LINTED */ binarytable = (uint_t *)(hdrdata + 12); first_unwind = 0; fde_count = 0; for (APLIST_TRAVERSE(ofl->ofl_unwind, idx, osp)) { uchar_t *data; size_t size; uint64_t off = 0, ujunk; int64_t sjunk; uint_t cieRflag = 0, ciePflag = 0; Shdr *shdr; /* * remember first UNWIND section to * point to in the frame_ptr entry. */ if (first_unwind == 0) first_unwind = osp; data = osp->os_outdata->d_buf; shdr = osp->os_shdr; size = shdr->sh_size; while (off < size) { uint_t length, id; uint64_t ndx = 0; /* * Extract length in lsb format. A zero length * indicates that this CIE is a terminator and that * processing of unwind information is complete. */ length = extract_uint(data + off, &ndx, bswap); if (length == 0) goto done; /* * Extract CIE id in lsb format. */ id = extract_uint(data + off, &ndx, bswap); /* * A CIE record has a id of '0'; otherwise * this is a FDE entry and the 'id' is the * CIE pointer. */ if (id == 0) { char *cieaugstr; uint_t cieaugndx; uint_t cieversion; ciePflag = 0; cieRflag = 0; /* * We need to drill through the CIE * to find the Rflag. It's the Rflag * which describes how the FDE code-pointers * are encoded. */ cieversion = data[off + ndx]; ndx += 1; /* * augstr */ cieaugstr = (char *)(&data[off + ndx]); ndx += strlen(cieaugstr) + 1; /* * calign & dalign */ if (uleb_extract(&data[off], &ndx, size - off, &ujunk) == DW_OVERFLOW) { ld_eprintf(ofl, ERR_FATAL, MSG_INTL(MSG_SCN_DWFOVRFLW), ofl->ofl_name, osp->os_name); return (S_ERROR); } if (sleb_extract(&data[off], &ndx, size - off, &sjunk) == DW_OVERFLOW) { ld_eprintf(ofl, ERR_FATAL, MSG_INTL(MSG_SCN_DWFOVRFLW), ofl->ofl_name, osp->os_name); return (S_ERROR); } /* * retreg */ if (cieversion == 1) { ndx++; } else { if (uleb_extract(&data[off], &ndx, size - off, &ujunk) == DW_OVERFLOW) { ld_eprintf(ofl, ERR_FATAL, MSG_INTL(MSG_SCN_DWFOVRFLW), ofl->ofl_name, osp->os_name); return (S_ERROR); } } /* * we walk through the augmentation * section now looking for the Rflag */ for (cieaugndx = 0; cieaugstr[cieaugndx]; cieaugndx++) { /* BEGIN CSTYLED */ switch (cieaugstr[cieaugndx]) { case 'z': /* size */ if (uleb_extract(&data[off], &ndx, size - off, &ujunk) == DW_OVERFLOW) { ld_eprintf(ofl, ERR_FATAL, MSG_INTL(MSG_SCN_DWFOVRFLW), ofl->ofl_name, osp->os_name); return (S_ERROR); } break; case 'P': /* personality */ ciePflag = data[off + ndx]; ndx++; /* * Just need to extract the * value to move on to the next * field. */ switch (dwarf_ehe_extract( &data[off], size - off, &ndx, &ujunk, ciePflag, ofl->ofl_dehdr->e_ident, B_FALSE, shdr->sh_addr, off + ndx, 0)) { case DW_OVERFLOW: ld_eprintf(ofl, ERR_FATAL, MSG_INTL(MSG_SCN_DWFOVRFLW), ofl->ofl_name, osp->os_name); return (S_ERROR); case DW_BAD_ENCODING: ld_eprintf(ofl, ERR_FATAL, MSG_INTL(MSG_SCN_DWFBADENC), ofl->ofl_name, osp->os_name, ciePflag); return (S_ERROR); case DW_SUCCESS: break; } break; case 'R': /* code encoding */ cieRflag = data[off + ndx]; ndx++; break; case 'L': /* lsda encoding */ ndx++; break; } /* END CSTYLED */ } } else { uint_t bintabndx; uint64_t initloc; uint64_t fdeaddr; uint64_t gotaddr = 0; if (ofl->ofl_osgot != NULL) gotaddr = ofl->ofl_osgot->os_shdr->sh_addr; switch (dwarf_ehe_extract(&data[off], size - off, &ndx, &initloc, cieRflag, ofl->ofl_dehdr->e_ident, B_FALSE, shdr->sh_addr, off + ndx, gotaddr)) { case DW_OVERFLOW: ld_eprintf(ofl, ERR_FATAL, MSG_INTL(MSG_SCN_DWFOVRFLW), ofl->ofl_name, osp->os_name); return (S_ERROR); case DW_BAD_ENCODING: ld_eprintf(ofl, ERR_FATAL, MSG_INTL(MSG_SCN_DWFBADENC), ofl->ofl_name, osp->os_name, cieRflag); return (S_ERROR); case DW_SUCCESS: break; } /* * Ignore FDEs with initloc set to 0. * initloc will not be 0 unless this FDE was * abandoned due to GNU linkonce processing. * The 0 value occurs because we don't resolve * sloppy relocations for unwind header target * sections. */ if (initloc != 0) { bintabndx = fde_count * 2; fde_count++; /* * FDEaddr is adjusted * to account for the length & id which * have already been consumed. */ fdeaddr = shdr->sh_addr + off; binarytable[bintabndx] = (uint_t)(initloc - hdraddr); binarytable[bintabndx + 1] = (uint_t)(fdeaddr - hdraddr); } } /* * the length does not include the length * itself - so account for that too. */ off += length + 4; } } done: /* * Do a quicksort on the binary table. If this is a cross * link from a system with the opposite byte order, xlate * the resulting values into LSB order. */ framehdr_addr = hdraddr; qsort((void *)binarytable, (size_t)fde_count, (size_t)(sizeof (uint_t) * 2), bintabcompare); if (bswap) { uint_t *btable = binarytable; uint_t cnt; for (cnt = fde_count * 2; cnt-- > 0; btable++) *btable = ld_bswap_Word(*btable); } /* * Fill in: * first_frame_ptr * fde_count */ hdroff = 4; /* LINTED */ uint_ptr = (uint_t *)(&hdrdata[hdroff]); *uint_ptr = first_unwind->os_shdr->sh_addr - (hdrosp->os_shdr->sh_addr + hdroff); if (bswap) *uint_ptr = ld_bswap_Word(*uint_ptr); hdroff += 4; /* LINTED */ uint_ptr = (uint_t *)&hdrdata[hdroff]; *uint_ptr = fde_count; if (bswap) *uint_ptr = ld_bswap_Word(*uint_ptr); /* * If relaxed relocations are active, then there is a chance * that we didn't use all the space reserved for this section. * For details, see the note at head of ld_unwind_make_hdr() above. * * Find the PT_SUNW_UNWIND program header, and change the size values * to the size of the subset of the section that was actually used. */ if (ofl->ofl_flags1 & FLG_OF1_RLXREL) { Word phnum = ofl->ofl_nehdr->e_phnum; Phdr *phdr = ofl->ofl_phdr; for (; phnum-- > 0; phdr++) { if (phdr->p_type == PT_SUNW_UNWIND) { phdr->p_memsz = 12 + (8 * fde_count); phdr->p_filesz = phdr->p_memsz; break; } } } return (1); } /* * Append an .eh_frame section to our output list if not already present. * * Usually, there is a single .eh_frame output section. However, there can * be more if there are incompatible section flags on incoming sections. * If this does happen, the frame_ptr field of the eh_frame_hdr section * will point at the base of the first output section, and the other * sections will not be accessible via frame_ptr. However, the .eh_frame_hdr * will be able to access all the data in the different .eh_frame sections, * because the entries in sorted table are all encoded as DW_EH_PE_datarel. */ uintptr_t ld_unwind_register(Os_desc *osp, Ofl_desc * ofl) { Aliste idx; Os_desc *_osp; /* * Check to see if this output section is already * on the list. */ for (APLIST_TRAVERSE(ofl->ofl_unwind, idx, _osp)) if (osp == _osp) return (1); /* * Append output section to unwind list */ if (aplist_append(&ofl->ofl_unwind, osp, AL_CNT_OFL_UNWIND) == NULL) return (S_ERROR); return (1); }