Improved Dwarf v5 support

CMakeLists.txt

@@ -0,0 +1,42 @@
+# Distributed under the OSI-approved BSD 3-Clause License.  See accompanying
+# file LICENSE.rst or https://cmake.org/licensing for details.
+
+cmake_minimum_required(VERSION 4.2.0)
+
+# Reject any attempt to use a toolchain file. We must not use one because
+# we could be downloading it here. If the CMAKE_TOOLCHAIN_FILE environment
+# variable is set, the cache variable will have been initialized from it.
+unset(CMAKE_TOOLCHAIN_FILE CACHE)
+unset(ENV{CMAKE_TOOLCHAIN_FILE})
+
+# We name the project and the target for the ExternalProject_Add() call
+# to something that will highlight to the user what we are working on if
+# something goes wrong and an error message is produced.
+
+project(libelfin-populate NONE)
+
+
+# Pass through things we've already detected in the main project to avoid
+# paying the cost of redetecting them again in ExternalProject_Add()
+set(GIT_EXECUTABLE [==[/opt/homebrew/bin/git]==])
+set(Git_VERSION [==[2.52.0]==])
+set_property(GLOBAL PROPERTY _CMAKE_FindGit_GIT_EXECUTABLE_VERSION
+  [==[/opt/homebrew/bin/git;2.52.0]==]
+)
+
+
+include(ExternalProject)
+ExternalProject_Add(libelfin-populate
+                     "UPDATE_DISCONNECTED" "False" "GIT_REPOSITORY" "https://github.com/plasma-umass/libelfin.git" "EXTERNALPROJECT_INTERNAL_ARGUMENT_SEPARATOR" "GIT_TAG" "8bd19d1e4bb19ec1046d44e1ac1b3bb72b91d0c5" "GIT_SHALLOW" "TRUE"
+                    SOURCE_DIR          "/Users/emery/git/coz-portage/_deps/libelfin-src"
+                    BINARY_DIR          "/Users/emery/git/coz-portage/_deps/libelfin-build"
+                    CONFIGURE_COMMAND   ""
+                    BUILD_COMMAND       ""
+                    INSTALL_COMMAND     ""
+                    TEST_COMMAND        ""
+                    USES_TERMINAL_DOWNLOAD  YES
+                    USES_TERMINAL_UPDATE    YES
+                    USES_TERMINAL_PATCH     YES
+)
+
+

dwarf/.gitignore

@@ -1,5 +1,4 @@
 *.o
-to_string.cc
 libdwarf++.a
 libdwarf++.so
 libdwarf++.so.*

dwarf/abbrev.cc

@@ -98,12 +98,37 @@ resolve_type(DW_AT name, DW_FORM form)
         case DW_FORM::ref_addr:
         case DW_FORM::ref_sig8:
         case DW_FORM::ref_udata:
+        case DW_FORM::ref_sup4:
+        case DW_FORM::ref_sup8:
                 return value::type::reference;
 
         case DW_FORM::string:
         case DW_FORM::strp:
+        case DW_FORM::line_strp:
+        case DW_FORM::strp_sup:
+        case DW_FORM::strx:
+        case DW_FORM::strx1:
+        case DW_FORM::strx2:
+        case DW_FORM::strx3:
+        case DW_FORM::strx4:
                 return value::type::string;
 
+        case DW_FORM::addrx:
+        case DW_FORM::addrx1:
+        case DW_FORM::addrx2:
+        case DW_FORM::addrx3:
+        case DW_FORM::addrx4:
+                return value::type::address;
+
+        case DW_FORM::implicit_const:
+                return value::type::constant;
+
+        case DW_FORM::loclistx:
+                return value::type::loclist;
+
+        case DW_FORM::rnglistx:
+                return value::type::rangelist;
+
         case DW_FORM::indirect:
                 // There's nothing meaningful we can do
                 return value::type::invalid;
@@ -138,15 +163,18 @@ resolve_type(DW_AT name, DW_FORM form)
                         return value::type::invalid;
 
                 default:
-                        throw format_error("DW_FORM_sec_offset not expected for attribute " +
-                                           to_string(name));
+                        // DWARF 5 has many new attributes using sec_offset
+                        // (str_offsets_base, addr_base, rnglists_base, loclists_base, etc.)
+                        // Just treat them as invalid for now to allow skipping
+                        return value::type::invalid;
                 }
         }
         throw format_error("unknown attribute form " + to_string(form));
 }
 
-attribute_spec::attribute_spec(DW_AT name, DW_FORM form)
-        : name(name), form(form), type(resolve_type(name, form))
+attribute_spec::attribute_spec(DW_AT name, DW_FORM form, int64_t implicit_const)
+        : name(name), form(form), type(resolve_type(name, form)),
+          implicit_const(implicit_const)
 {
 }
 
@@ -167,7 +195,10 @@ abbrev_entry::read(cursor *cur)
                 DW_FORM form = (DW_FORM)cur->uleb128();
                 if (name == (DW_AT)0 && form == (DW_FORM)0)
                         break;
-                attributes.push_back(attribute_spec(name, form));
+                int64_t implicit_const = 0;
+                if (form == DW_FORM::implicit_const)
+                        implicit_const = cur->sleb128();
+                attributes.push_back(attribute_spec(name, form, implicit_const));
         }
         attributes.shrink_to_fit();
         return true;

dwarf/cursor.cc

@@ -92,7 +92,9 @@ cursor::string(std::string &out)
         size_t size;
         const char *p = this->cstr(&size);
         out.resize(size);
-        memmove(&out.front(), p, size);
+        if (size > 0) {
+                memmove(&out.front(), p, size);
+        }
 }
 
 const char *
@@ -123,6 +125,8 @@ cursor::skip_form(DW_FORM form)
         case DW_FORM::sec_offset:
         case DW_FORM::ref_addr:
         case DW_FORM::strp:
+        case DW_FORM::line_strp:
+        case DW_FORM::strp_sup:
                 switch (sec->fmt) {
                 case format::dwarf32:
                         pos += 4;
@@ -156,29 +160,49 @@ cursor::skip_form(DW_FORM form)
 
                 // fixed-length forms
         case DW_FORM::flag_present:
+        case DW_FORM::implicit_const:
                 break;
         case DW_FORM::flag:
         case DW_FORM::data1:
         case DW_FORM::ref1:
+        case DW_FORM::strx1:
+        case DW_FORM::addrx1:
                 pos += 1;
                 break;
         case DW_FORM::data2:
         case DW_FORM::ref2:
+        case DW_FORM::strx2:
+        case DW_FORM::addrx2:
                 pos += 2;
                 break;
+        case DW_FORM::strx3:
+        case DW_FORM::addrx3:
+                pos += 3;
+                break;
         case DW_FORM::data4:
         case DW_FORM::ref4:
+        case DW_FORM::ref_sup4:
+        case DW_FORM::strx4:
+        case DW_FORM::addrx4:
                 pos += 4;
                 break;
         case DW_FORM::data8:
         case DW_FORM::ref_sig8:
+        case DW_FORM::ref_sup8:
                 pos += 8;
                 break;
+        case DW_FORM::data16:
+                pos += 16;
+                break;
 
                 // variable-length forms
         case DW_FORM::sdata:
         case DW_FORM::udata:
         case DW_FORM::ref_udata:
+        case DW_FORM::strx:
+        case DW_FORM::addrx:
+        case DW_FORM::loclistx:
+        case DW_FORM::rnglistx:
                 while (pos < sec->end && (*(uint8_t*)pos & 0x80))
                         pos++;
                 pos++;

dwarf/data.hh

@@ -250,7 +250,27 @@ enum class DW_FORM
         sec_offset   = 0x17,    // lineptr, loclistptr, macptr, rangelistptr
         exprloc      = 0x18,    // exprloc
         flag_present = 0x19,    // flag
+
+        // DWARF 5
+        strx         = 0x1a,    // string index in .debug_str_offsets
+        addrx        = 0x1b,    // address index in .debug_addr
+        ref_sup4     = 0x1c,    // reference
+        strp_sup     = 0x1d,    // string
+        data16       = 0x1e,    // constant
+        line_strp    = 0x1f,    // string
         ref_sig8     = 0x20,    // reference
+        implicit_const = 0x21,  // constant encoded in abbrev
+        loclistx     = 0x22,    // location list index
+        rnglistx     = 0x23,    // range list index
+        ref_sup8     = 0x24,    // reference
+        strx1        = 0x25,    // 1-byte string index
+        strx2        = 0x26,    // 2-byte string index
+        strx3        = 0x27,    // 3-byte string index
+        strx4        = 0x28,    // 4-byte string index
+        addrx1       = 0x29,    // 1-byte address index
+        addrx2       = 0x2a,    // 2-byte address index
+        addrx3       = 0x2b,    // 3-byte address index
+        addrx4       = 0x2c,    // 4-byte address index
 };
 
 std::string
@@ -560,6 +580,37 @@ enum class DW_LNE
 std::string
 to_string(DW_LNE v);
 
+// Line number content types (DWARF5 section 7.22 table 7.30)
+enum class DW_LNCT
+{
+        path = 0x01,
+        directory_index = 0x02,
+        timestamp = 0x03,
+        size = 0x04,
+        MD5 = 0x05,
+        lo_user = 0x2000,
+        hi_user = 0x3fff,
+};
+
+std::string
+to_string(DW_LNCT v);
+
+// Range list entry encodings (DWARF5 section 7.25)
+enum class DW_RLE : ubyte
+{
+        end_of_list    = 0x00,
+        base_addressx  = 0x01,
+        startx_endx    = 0x02,
+        startx_length  = 0x03,
+        offset_pair    = 0x04,
+        base_address   = 0x05,
+        start_end      = 0x06,
+        start_length   = 0x07,
+};
+
+std::string
+to_string(DW_RLE v);
+
 DWARFPP_END_NAMESPACE
 
 #endif

dwarf/die.cc

@@ -74,7 +74,7 @@ die::operator[](DW_AT attr) const
                 int i = 0;
                 for (auto &a : abbrev->attributes) {
                         if (a.name == attr)
-                                return value(cu, a.name, a.form, a.type, attrs[i]);
+                                return value(cu, a, attrs[i]);
                         i++;
                 }
         }
@@ -174,7 +174,7 @@ die::attributes() const
         // custom iterator.
         int i = 0;
         for (auto &a : abbrev->attributes) {
-                res.push_back(make_pair(a.name, value(cu, a.name, a.form, a.type, attrs[i])));
+                res.push_back(make_pair(a.name, value(cu, a, attrs[i])));
                 i++;
         }
         return res;