Add debuginfod support (#3954)

* Add DEBUGINFOD_URLS support
* Add rz_bin_dwarf_from_debuginfod
* Fix rz_bin_dwarf_attr_addr
* Add test for debuginfod
* Filter empty RzBinDWARF

.github/workflows/ci.yml

@@ -200,7 +200,7 @@ jobs:
         if: matrix.run_tests && matrix.enabled && matrix.os != 'macos-12'
         run: |
           sudo apt-get update
-          sudo apt-get --assume-yes install libc6 libc6-i386 libc6-dev
+          sudo apt-get --assume-yes install libc6 libc6-i386 libc6-dev debuginfod
       - name: Install gdbserver dependency
         if: matrix.run_tests && matrix.enabled && matrix.os == 'ubuntu-22.04'
         run: sudo apt-get --assume-yes install gdbserver
@@ -451,7 +451,7 @@ jobs:
       - name: Install test dependencies
         run: |
           sudo pip3 install 'git+https://github.com/rizinorg/rz-pipe#egg=rzpipe&subdirectory=python'
-          sudo apt-get install --yes tzdata
+          sudo apt-get install --yes tzdata debuginfod
       - name: Checkout rizin
         run: |
           git clone https://github.com/${{ github.repository }}
@@ -520,7 +520,7 @@ jobs:
       - name: Run debug tests
         run: |
           cd test
-          rz-test -L -o results.json db/archos/linux-x64
+          rz-test -L -e db/archos/linux-x64/debuginfod -e db/archos/linux-x64/dbg_dmh -o results.json db/archos/linux-x64
         working-directory: rizin
 
   build-static:
@@ -608,7 +608,7 @@ jobs:
     name: Test source tarball
     if: contains(github.head_ref, 'dist') || contains(github.head_ref, 'extras') || contains(github.ref, 'release-') || github.ref == 'refs/heads/stable'
     runs-on: ubuntu-22.04
-    needs: [create-tarball]
+    needs: [ create-tarball ]
     steps:
       - name: Install python and other dependencies
         run: sudo apt-get --assume-yes install python3-wheel python3-setuptools python3-pip
@@ -646,7 +646,7 @@ jobs:
     name: Build OSX package
     runs-on: macos-12
     if: contains(github.head_ref, 'dist') || contains(github.head_ref, 'osx') || contains(github.head_ref, 'mac') || ((contains(github.ref, 'release-') || github.ref == 'refs/heads/stable') && github.event_name == 'push') || github.event_name == 'schedule'
-    needs: [build-and-test]
+    needs: [ build-and-test ]
     steps:
       - uses: actions/checkout@v3
       - name: Install pkg-config with Homebrew
@@ -669,11 +669,11 @@ jobs:
     name: Build Windows zip/installer ${{ matrix.name }}
     runs-on: windows-latest
     if: contains(github.head_ref, 'dist') || contains(github.head_ref, 'windows') || (github.event_name == 'push' && (contains(github.ref, 'release-') || github.ref == 'refs/heads/stable'))
-    needs: [build-and-test]
+    needs: [ build-and-test ]
     strategy:
       fail-fast: false
       matrix:
-        name: [vs2019_static, clang_cl, clang_cl_x86]
+        name: [ vs2019_static, clang_cl, clang_cl_x86 ]
         include:
           - name: vs2019_static
             compiler: cl
@@ -714,11 +714,11 @@ jobs:
     name: Build Android ${{ matrix.name }} package
     runs-on: ubuntu-22.04
     if: contains(github.head_ref, 'dist') || contains(github.head_ref, 'android') || ((contains(github.ref, 'release-') || github.ref == 'refs/heads/stable') && github.event_name == 'push') || github.event_name == 'schedule'
-    needs: [build-and-test]
+    needs: [ build-and-test ]
     strategy:
       fail-fast: false
       matrix:
-        name: [x86_64, arm, aarch64]
+        name: [ x86_64, arm, aarch64 ]
     steps:
       - uses: actions/checkout@v3
       - name: Install dependencies
@@ -743,7 +743,7 @@ jobs:
     name: Include Rizin headers from C++ program (Linux)
     runs-on: ubuntu-22.04
     if: contains(github.head_ref, 'dist') || contains(github.head_ref, 'cpp') || ((contains(github.ref, 'release-') || github.ref == 'refs/heads/stable') && github.event_name == 'push') || github.event_name == 'schedule'
-    needs: [build-and-test]
+    needs: [ build-and-test ]
     steps:
       - uses: actions/checkout@v3
       - name: Install dependencies
@@ -769,7 +769,7 @@ jobs:
     name: Test EXTRA_PREFIX works well
     runs-on: ubuntu-22.04
     if: contains(github.head_ref, 'dist') || ((contains(github.ref, 'release-') || github.ref == 'refs/heads/stable') && github.event_name == 'push') || github.event_name == 'schedule'
-    needs: [build-and-test]
+    needs: [ build-and-test ]
     steps:
       - uses: actions/checkout@v3
       - name: Install dependencies
@@ -792,7 +792,7 @@ jobs:
   build-rzpipe:
     name: Build rizin rzpipe
     if: contains(github.head_ref, 'dist') || contains(github.ref, 'release-') || github.ref == 'refs/heads/stable' || github.event_name == 'schedule'
-    needs: [test-tarball]
+    needs: [ test-tarball ]
     runs-on: ubuntu-22.04
     env:
       RZPIPE_TESTS: "rz-pipe-py rz-pipe-go"
@@ -821,7 +821,7 @@ jobs:
   build-bindgen:
     name: Build rz-bindgen
     if: contains(github.head_ref, 'dist') || contains(github.ref, 'release-') || github.ref == 'refs/heads/stable' || github.event_name == 'schedule'
-    needs: [test-tarball]
+    needs: [ test-tarball ]
     runs-on: ubuntu-22.04
     steps:
       - uses: actions/download-artifact@v3
@@ -867,7 +867,7 @@ jobs:
         system:
           - macos-12
     runs-on: ${{ matrix.system }}
-    needs: [build-osx-pkg]
+    needs: [ build-osx-pkg ]
     steps:
       - name: Install pkg-config with Homebrew
         run: brew install pkg-config
@@ -888,10 +888,10 @@ jobs:
   test-windows-clang_cl:
     name: Test Windows installer built with ${{ matrix.name }}
     runs-on: windows-latest
-    needs: [build-windows]
+    needs: [ build-windows ]
     strategy:
       matrix:
-        name: [clang_cl, clang_cl_x86]
+        name: [ clang_cl, clang_cl_x86 ]
         include:
           - name: clang_cl
             bits: 64
@@ -1000,7 +1000,7 @@ jobs:
 
   publish-docker-image:
     name: Publish Docker image on Docker Hub
-    needs: [build-and-test]
+    needs: [ build-and-test ]
     runs-on: ubuntu-22.04
     if: github.event_name == 'push' || (github.event_name == 'pull_request' && contains(github.head_ref, 'container'))
     steps:

.github/workflows/tcc.yml

@@ -92,7 +92,7 @@ jobs:
       run: |
           python3 -m pip install --user 'git+https://github.com/rizinorg/rz-pipe#egg=rzpipe&subdirectory=python'
           sudo apt-get update
-          sudo apt-get --assume-yes install libc6 libc6-i386
+          sudo apt-get --assume-yes install libc6 libc6-i386 debuginfod
 
     - name: Run tests
       env:

librz/analysis/dwarf_process.c

@@ -1527,13 +1527,16 @@ static bool function_from_die(
 			fcn->link_name = attr_string(attr, ctx);
 			break;
 		case DW_AT_low_pc:
-			fcn->low_pc = rz_bin_dwarf_attr_udata(attr);
+			fcn->low_pc = rz_bin_dwarf_attr_addr(
+				attr, ctx->dw, ctx->unit->hdr.encoding.address_size, ctx->unit->addr_base);
 			break;
 		case DW_AT_high_pc:
-			fcn->high_pc = rz_bin_dwarf_attr_udata(attr);
+			fcn->high_pc = rz_bin_dwarf_attr_addr(
+				attr, ctx->dw, ctx->unit->hdr.encoding.address_size, ctx->unit->addr_base);
 			break;
 		case DW_AT_entry_pc:
-			fcn->entry_pc = rz_bin_dwarf_attr_udata(attr);
+			fcn->entry_pc = rz_bin_dwarf_attr_addr(
+				attr, ctx->dw, ctx->unit->hdr.encoding.address_size, ctx->unit->addr_base);
 			break;
 		case DW_AT_specification: /* u64 to declaration DIE with more info */
 		{

librz/bin/dwarf/addr.c

@@ -4,7 +4,9 @@
 #include <rz_bin_dwarf.h>
 #include "dwarf_private.h"
 
-RZ_IPI bool DebugAddr_get_address(const RzBinDwarfAddr *self, ut64 *address,
+RZ_API bool rz_bin_dwarf_addr_get(
+	RZ_BORROW RZ_NONNULL const RzBinDwarfAddr *self,
+	RZ_BORROW RZ_NONNULL ut64 *address,
 	ut8 address_size, ut64 base, ut64 index) {
 	rz_return_val_if_fail(self && self->reader && address, false);
 	RzBinEndianReader *reader = self->reader;
@@ -14,25 +16,25 @@ RZ_IPI bool DebugAddr_get_address(const RzBinDwarfAddr *self, ut64 *address,
 	return true;
 }
 
-RZ_IPI void DebugAddr_free(RzBinDwarfAddr *self) {
+RZ_API void rz_bin_dwarf_addr_free(RzBinDwarfAddr *self) {
 	if (!self) {
 		return;
 	}
 	RzBinEndianReader_free(self->reader);
 	free(self);
 }
 
-RZ_IPI RzBinDwarfAddr *DebugAddr_new(RzBinEndianReader *reader) {
+RZ_API RZ_OWN RzBinDwarfAddr *rz_bin_dwarf_addr_new(RZ_OWN RZ_NONNULL RzBinEndianReader *reader) {
 	rz_return_val_if_fail(reader, NULL);
 	RzBinDwarfAddr *self = RZ_NEW0(RzBinDwarfAddr);
 	RET_NULL_IF_FAIL(self);
 	self->reader = reader;
 	return self;
 }
 
-RZ_IPI RzBinDwarfAddr *DebugAddr_from_file(RzBinFile *bf) {
+RZ_API RZ_OWN RzBinDwarfAddr *rz_bin_dwarf_addr_from_file(RZ_BORROW RZ_NONNULL RzBinFile *bf) {
 	rz_return_val_if_fail(bf, NULL);
 	RzBinEndianReader *r = RzBinEndianReader_from_file(bf, ".debug_addr", false);
 	RET_NULL_IF_FAIL(r);
-	return DebugAddr_new(r);
+	return rz_bin_dwarf_addr_new(r);
 }

librz/bin/dwarf/attr.c

@@ -20,7 +20,7 @@ RZ_IPI bool RzBinDwarfAttr_parse(
 	// http://www.dwarfstd.org/doc/DWARF4.pdf#page=161&zoom=100,0,560
 	switch (attr->form) {
 	case DW_FORM_addr:
-		value->kind = RzBinDwarfAttr_Address;
+		value->kind = RzBinDwarfAttr_Addr;
 		RET_FALSE_IF_FAIL(read_address(reader, &value->u64, address_size));
 		break;
 	case DW_FORM_data1:
@@ -166,25 +166,25 @@ RZ_IPI bool RzBinDwarfAttr_parse(
 		DW_AT_addr_base attribute of the associated compilation unit.
 	    index into an array of addresses in the .debug_addr section.*/
 	case DW_FORM_addrx:
-		value->kind = RzBinDwarfAttr_Address;
+		value->kind = RzBinDwarfAttr_AddrIndex;
 		ULE128_OR_RET_FALSE(value->u64);
 		break;
 	case DW_FORM_addrx1:
-		value->kind = RzBinDwarfAttr_Address;
+		value->kind = RzBinDwarfAttr_AddrIndex;
 		U8_OR_RET_FALSE(value->u64);
 		break;
 	case DW_FORM_addrx2:
-		value->kind = RzBinDwarfAttr_Address;
+		value->kind = RzBinDwarfAttr_AddrIndex;
 		U_OR_RET_FALSE(16, value->u64);
 		break;
 	case DW_FORM_addrx3:
 		// TODO: .DW_FORM_addrx3
-		value->kind = RzBinDwarfAttr_Address;
+		value->kind = RzBinDwarfAttr_AddrIndex;
 		rz_buf_seek(reader->buffer, 3, RZ_BUF_CUR);
 		RZ_LOG_ERROR("TODO: DW_FORM_addrx3\n");
 		break;
 	case DW_FORM_addrx4:
-		value->kind = RzBinDwarfAttr_Address;
+		value->kind = RzBinDwarfAttr_AddrIndex;
 		U_OR_RET_FALSE(32, value->u64);
 		break;
 	case DW_FORM_line_ptr: // offset in a section .debug_line_str
@@ -211,7 +211,7 @@ RZ_IPI bool RzBinDwarfAttr_parse(
 		break;
 		// An index into the .debug_rnglists
 	case DW_FORM_rnglistx:
-		value->kind = RzBinDwarfAttr_Address;
+		value->kind = RzBinDwarfAttr_RangelistPtr;
 		ULE128_OR_RET_FALSE(value->u64);
 		break;
 	default:

librz/bin/dwarf/dwarf.c

@@ -22,7 +22,7 @@ static inline RZ_OWN RzBinDWARF *dwarf_from_file(
 	RzBinDWARF *dw = RZ_NEW0(RzBinDWARF);
 	RET_NULL_IF_FAIL(dw);
 
-	dw->addr = DebugAddr_from_file(bf);
+	dw->addr = rz_bin_dwarf_addr_from_file(bf);
 	dw->line_str = rz_bin_dwarf_line_str_from_file(bf);
 	dw->aranges = rz_bin_dwarf_aranges_from_file(bf);
 
@@ -38,6 +38,10 @@ static inline RZ_OWN RzBinDWARF *dwarf_from_file(
 	if (dw->info) {
 		dw->line = rz_bin_dwarf_line_from_file(bf, dw, is_dwo);
 	}
+	if (!(dw->addr || dw->line_str || dw->aranges || dw->str || dw->str_offsets || dw->loclists || dw->rnglists || dw->abbrev)) {
+		rz_bin_dwarf_free(dw);
+		return NULL;
+	}
 	return dw;
 }
 
@@ -185,6 +189,33 @@ RZ_API RZ_OWN RzBinDWARF *rz_bin_dwarf_search_debug_file_directory(
 	return NULL;
 }
 
+RZ_API RZ_OWN RzBinDWARF *rz_bin_dwarf_from_debuginfod(
+	RZ_BORROW RZ_NONNULL RzBinFile *bf,
+	RZ_BORROW RZ_NONNULL RzList /*<const char *>*/ *debuginfod_urls) {
+	rz_return_val_if_fail(bf && debuginfod_urls, NULL);
+
+	RzBinDWARF *dw = NULL;
+	char *build_id = read_build_id(bf);
+	if (!build_id) {
+		return NULL;
+	}
+	RzListIter *it = NULL;
+	const char *debuginfod_url = NULL;
+	rz_list_foreach (debuginfod_urls, it, debuginfod_url) {
+		char *url = rz_str_newf("%s/buildid/%s/debuginfo", debuginfod_url, build_id);
+		if (!url) {
+			break;
+		}
+		dw = rz_bin_dwarf_from_path(url, false);
+		free(url);
+		if (dw) {
+			break;
+		}
+	}
+	free(build_id);
+	return dw;
+}
+
 /**
  * \brief Load DWARF from split DWARF file
  * \param filepath The file path
@@ -225,7 +256,7 @@ RZ_API void rz_bin_dwarf_free(RZ_OWN RZ_NULLABLE RzBinDWARF *dw) {
 	rz_bin_dwarf_free(dw->parent);
 
 	DebugRngLists_free(dw->rnglists);
-	DebugAddr_free(dw->addr);
+	rz_bin_dwarf_addr_free(dw->addr);
 	rz_bin_dwarf_str_free(dw->str);
 	rz_bin_dwarf_str_offsets_free(dw->str_offsets);
 

librz/bin/dwarf/dwarf_private.h

@@ -52,13 +52,6 @@ static inline bool bf_bigendian(RzBinFile *bf) {
 }
 
 RZ_IPI bool RzBinDwarfEncoding_from_file(RzBinDwarfEncoding *encoding, RzBinFile *bf);
-/// addr
-
-RZ_IPI bool DebugAddr_get_address(const RzBinDwarfAddr *self, ut64 *address,
-	ut8 address_size, ut64 base, ut64 index);
-RZ_IPI void DebugAddr_free(RzBinDwarfAddr *self);
-RZ_IPI RzBinDwarfAddr *DebugAddr_new(RzBinEndianReader *reader);
-RZ_IPI RzBinDwarfAddr *DebugAddr_from_file(RzBinFile *bf);
 
 /// range
 

librz/bin/dwarf/loclists.c

@@ -165,7 +165,7 @@ static bool convert_raw(
 			*entry = NULL;
 			return true;
 		case DW_LLE_base_addressx:
-			ERR_IF_FAIL(DebugAddr_get_address(
+			ERR_IF_FAIL(rz_bin_dwarf_addr_get(
 				addr, &self->base_address,
 				encoding->address_size, cu->addr_base, raw->base_addressx.addr));
 			return true;
@@ -175,17 +175,17 @@ static bool convert_raw(
 		case DW_LLE_startx_endx:
 			range = RZ_NEW0(RzBinDwarfRange);
 			ERR_IF_FAIL(range);
-			ERR_IF_FAIL(DebugAddr_get_address(
+			ERR_IF_FAIL(rz_bin_dwarf_addr_get(
 				addr, &range->begin,
 				encoding->address_size, cu->addr_base, raw->startx_endx.begin));
-			ERR_IF_FAIL(DebugAddr_get_address(
+			ERR_IF_FAIL(rz_bin_dwarf_addr_get(
 				addr, &range->end,
 				encoding->address_size, cu->addr_base, raw->startx_endx.end));
 			break;
 		case DW_LLE_startx_length:
 			range = RZ_NEW0(RzBinDwarfRange);
 			ERR_IF_FAIL(range);
-			ERR_IF_FAIL(DebugAddr_get_address(
+			ERR_IF_FAIL(rz_bin_dwarf_addr_get(
 				addr, &range->begin,
 				encoding->address_size, cu->addr_base, raw->startx_length.begin));
 			range->end = (raw->startx_length.length + raw->startx_length.begin) & mask;

librz/bin/dwarf/rnglists.c

@@ -154,22 +154,22 @@ static bool convert_raw(
 			*out = NULL;
 			return true;
 		case DW_RLE_base_addressx:
-			RET_FALSE_IF_FAIL(DebugAddr_get_address(addr, &self->base_address,
+			RET_FALSE_IF_FAIL(rz_bin_dwarf_addr_get(addr, &self->base_address,
 				cu->hdr.encoding.address_size, cu->addr_base, raw->base_addressx.addr));
 			*out = NULL;
 			return true;
 		case DW_RLE_startx_endx:
 			range = RZ_NEW0(RzBinDwarfRange);
 			RET_FALSE_IF_FAIL(range);
-			RET_FALSE_IF_FAIL(DebugAddr_get_address(addr, &range->begin,
+			RET_FALSE_IF_FAIL(rz_bin_dwarf_addr_get(addr, &range->begin,
 				cu->hdr.encoding.address_size, cu->addr_base, raw->startx_endx.begin));
-			RET_FALSE_IF_FAIL(DebugAddr_get_address(addr, &range->end,
+			RET_FALSE_IF_FAIL(rz_bin_dwarf_addr_get(addr, &range->end,
 				cu->hdr.encoding.address_size, cu->addr_base, raw->startx_endx.end));
 			break;
 		case DW_RLE_startx_length:
 			range = RZ_NEW0(RzBinDwarfRange);
 			RET_FALSE_IF_FAIL(range);
-			RET_FALSE_IF_FAIL(DebugAddr_get_address(addr, &range->begin,
+			RET_FALSE_IF_FAIL(rz_bin_dwarf_addr_get(addr, &range->begin,
 				cu->hdr.encoding.address_size, cu->addr_base, raw->startx_length.begin));
 			range->end = (raw->startx_length.length + raw->startx_length.begin) & mask;
 			break;