Convert gensyms to C++

Change-Id: I435996d29ca6c7ad4ce892302729089a2767e643
Reviewed-on: http://gfw160.austin.ibm.com:8080/gerrit/5549
Reviewed-by: Brian H. Horton <brianh@linux.ibm.com>
Tested-by: Jenkins Server
Reviewed-by: Douglas R. Gilbert <dgilbert@us.ibm.com>
Reviewed-by: A. Patrick Williams III <iawillia@us.ibm.com>

5 files changed, 372 insertions, 161 deletions

diff --git a/src/build/linker/.gitignore b/src/build/linker/.gitignore
index 6b243f5f3..992cfdb19 100644
--- a/src/build/linker/.gitignore
+++ b/src/build/linker/.gitignore
@@ -1 +1,2 @@
 linker
+gensyms
diff --git a/src/build/linker/gensyms.C b/src/build/linker/gensyms.C
new file mode 100644
index 000000000..48f018d17
--- /dev/null
+++ b/src/build/linker/gensyms.C
@@ -0,0 +1,364 @@
+/* IBM_PROLOG_BEGIN_TAG                                                   */
+/* This is an automatically generated prolog.                             */
+/*                                                                        */
+/* $Source: src/build/linker/gensyms.C $                                  */
+/*                                                                        */
+/* IBM CONFIDENTIAL                                                       */
+/*                                                                        */
+/* COPYRIGHT International Business Machines Corp. 2013                   */
+/*                                                                        */
+/* p1                                                                     */
+/*                                                                        */
+/* Object Code Only (OCO) source materials                                */
+/* Licensed Internal Code Source Materials                                */
+/* IBM HostBoot Licensed Internal Code                                    */
+/*                                                                        */
+/* The source code for this program is not published or otherwise         */
+/* divested of its trade secrets, irrespective of what has been           */
+/* deposited with the U.S. Copyright Office.                              */
+/*                                                                        */
+/* Origin: 30                                                             */
+/*                                                                        */
+/* IBM_PROLOG_END_TAG                                                     */
+#include <cstdio>
+#include <cstdlib>
+#include <vector>
+#include <map>
+#include <string>
+#include <stdint.h>
+#include <cstring>
+#include <endian.h>
+#include <assert.h>
+#include <limits.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <sys/mman.h>
+
+using namespace std;
+
+/** Print tool usage */
+void print_usage();
+
+/** Prepend to a path the img/ subdirectory.
+ *
+ *  @param[in,out] io_path - The path to modify / prepend to.
+ */
+void add_image_subdir(string& io_path);
+
+/** Parse the image.modinfo file.
+ *
+ *  @param[in] i_image - The path to the image to parse the corresponding
+ *                       modinfo.
+ */
+void parse_modinfo_file(const string& i_image);
+
+/** Read the symbols from a module.
+ *
+ *  @param[in] pair<string, uint64_t>* - Pair of <Path of module, offset >.
+ *
+ *  Parameters are passed as a (void*) to allow this function to be started
+ *  as a thread.
+ *
+ *  @return Unused.
+ */
+void* read_module_symbols(void*);
+
+    /** Module information parsed from modinfo.  <Module, Offset> */
+vector<pair<string, uint64_t> > g_modules;
+
+    /** Name / path of the base image. */
+string g_imageName;
+    /** Name / path of the extended image. */
+string g_extImageName;
+    /** Pointer to the mmap of the base image. */
+const char* g_imageFile;
+    /** Size of the base image file. */
+size_t g_imageFileSize;
+    /** Pointer to the mmap of the extended image. */
+const char* g_extImageFile;
+    /** Size of the extended image file. */
+size_t g_extImageFileSize;
+    /** Offset (in memory) that the extended image is to be loaded at. */
+uint64_t g_extImageOffset = ULONG_MAX;
+
+    /** Cached value of the CROSS_PREFIX environment variable, used to
+     *  call binutils tools. */
+char* g_crossPrefix = NULL;
+
+    /** Resulting symbol addresses and names.
+     *
+     *  This is a multimap because there are some symbol addresses with
+     *  multiple names.  Ex. the data_start_address often collides with
+     *  a global symbol in the data section.
+     */
+multimap<uint64_t, string> g_symbols;
+    /** Mutex to protect symbol map. */
+pthread_mutex_t g_symbolMutex = PTHREAD_MUTEX_INITIALIZER;
+
+int main(int argc, char** argv)
+{
+    // Allow one argument (base image) or three arguments (base, extend, offset)
+    if ((argc != 2) && (argc != 4))
+    {
+        print_usage();
+    }
+
+    // Get base image name.
+    g_imageName = argv[1];
+    add_image_subdir(g_imageName);
+
+    // Get extended image name.
+    if (argc > 3)
+    {
+        g_extImageName = argv[2];
+        add_image_subdir(g_extImageName);
+
+        // Read extended image offset from options.
+        if (1 != sscanf(argv[3], "%lx", &g_extImageOffset))
+        {
+            print_usage();
+        }
+    }
+
+    // Open base image.
+    int base_fd = open(g_imageName.c_str(), O_RDONLY);
+    if (-1 == base_fd)
+    {
+        printf("Failed to open image file: %s.\n", g_imageName.c_str());
+        exit(-1);
+    }
+    struct stat base_stat;
+    if (0 != fstat(base_fd, &base_stat))
+    {
+        printf("Failed to stat image file: %s.\n", g_imageName.c_str());
+        exit(-1);
+    }
+    g_imageFileSize = base_stat.st_size;
+    g_imageFile = (const char*) mmap(NULL, base_stat.st_size,
+                                     PROT_READ, MAP_PRIVATE,
+                                     base_fd, 0);
+
+    // Open extended image.
+    if (string() != g_extImageName.c_str())
+    {
+        int ext_fd = open(g_extImageName.c_str(), O_RDONLY);
+        if (-1 == ext_fd)
+        {
+            printf("Failed to open image file: %s.\n", g_extImageName.c_str());
+            exit(-1);
+        }
+        struct stat ext_stat;
+        if (0 != fstat(ext_fd, &ext_stat))
+        {
+            printf("Failed to stat image file: %s.\n", g_extImageName.c_str());
+            exit(-1);
+        }
+        g_extImageFileSize = ext_stat.st_size;
+        g_extImageFile = (const char*) mmap(NULL, ext_stat.st_size,
+                                            PROT_READ, MAP_PRIVATE,
+                                            ext_fd, 0);
+    }
+
+    // Read CROSS_PREFIX environment variable.
+    g_crossPrefix = getenv("CROSS_PREFIX");
+    if (NULL == g_crossPrefix)
+    {
+        printf("Environment variable CROSS_PREFIX not set.\n");
+        exit(-1);
+    }
+    g_crossPrefix = strdup(g_crossPrefix);
+
+    // Parse modinfo file for base image.
+    parse_modinfo_file(g_imageName);
+
+    // Create threads for each ELF object in the image(s) to get their symbol
+    // information.
+    vector<pthread_t*> threads;
+    for(vector<pair<string, uint64_t> >::const_iterator i = g_modules.begin();
+        i != g_modules.end(); ++i)
+    {
+        const string& m = i->first;
+        // Filter out non-ELF files by filename.
+        if (strstr(m.c_str(), ".o") || strstr(m.c_str(), ".elf") ||
+            strstr(m.c_str(), ".so"))
+        {
+            pthread_t* thread = new pthread_t;
+            pthread_create(thread, NULL, read_module_symbols,
+                           new pair<string,uint64_t>(*i));
+            threads.push_back(thread);
+        }
+    }
+
+    // Wait for all threads to finish.
+    for(vector<pthread_t*>::const_iterator i = threads.begin();
+        i != threads.end(); ++i)
+    {
+        pthread_join(*(*i), NULL);
+    }
+
+    // Output (in order) each symbol information.
+    for (multimap<uint64_t, string>::const_iterator i = g_symbols.begin();
+         i != g_symbols.end(); ++i)
+    {
+        printf("%s", i->second.c_str());
+    }
+
+    return 0;
+}
+
+void print_usage()
+{
+    printf("gensyms <image> [<extimage> <extoffset>]\n");
+    exit(-1);
+}
+
+void add_image_subdir(string& io_path)
+{
+    // Prepend ./img if the path to the image directory is not already part
+    // of the path.
+    if (string::npos == io_path.find("img"))
+    {
+        io_path.insert(0, "./img/");
+    }
+}
+
+void parse_modinfo_file(const string& i_image)
+{
+    // Open modinfo file.
+    string modinfo_name = i_image + ".modinfo";
+    FILE* modinfo_file = fopen(modinfo_name.c_str(), "r");
+    if (NULL == modinfo_file)
+    {
+        printf("Unable to open modinfo file.\n");
+        exit(-1);
+    }
+
+    // Parse one line at a time.
+    char line[1024];
+    do
+    {
+        // fgets returns NULL when no additional lines are present, break.
+        if (NULL == fgets(line, 1024, modinfo_file)) break;
+
+        // Lines should be formatted: "object,offset\n"
+
+        // Skip lines without a comma.
+        char* comma = strchr(line, ',');
+        if (NULL == comma) continue;
+
+        // Extract module name (everything before comma).
+        string mod_name(line, comma - line);
+
+        // Parse module offset (hex integer after comma).
+        uint64_t mod_addr;
+        if (1 != sscanf(comma+1, "0x%lx", &mod_addr)) continue;
+
+        // Add to the module list.
+        g_modules.push_back(make_pair(mod_name, mod_addr));
+
+    } while(1);
+}
+
+void* read_module_symbols(void* input)
+{
+    // Get module name and offset from input parameter.
+    pair<string, uint64_t>* mod_info =
+        reinterpret_cast<pair<string,uint64_t>*>(input);
+    const string& module = mod_info->first;
+    uint64_t addr = mod_info->second;
+
+    // Determine the full path to the module based on the base image path.
+    // Assumes they are in the same subdirectory.
+    string module_path = g_imageName.substr(0, g_imageName.rfind('/') + 1) +
+                         module;
+
+    // Create the 'objdump' command for finding all the symbols and start as
+    // a sub-process.
+    string command = string(g_crossPrefix) + string("objdump --syms -C ") +
+                     module_path;
+    FILE* pipe = popen(command.c_str(), "r");
+    if (NULL == pipe) return NULL;
+
+    // Local symbol map (to reduce contention on the global symbol map).
+    //   No need to use the overhead of a map because we don't care about
+    //   order at this point.
+    vector<pair<uint64_t, string> > l_symbols;
+
+    // Parse each line of the 'objdump' output.
+    char line[1024];
+    do
+    {
+        if (NULL == fgets(line, 1024, pipe)) break;
+
+        // Skip absolute values (ex. constants) and undefined symbols.
+        if (strstr(line, "*ABS*") || strstr(line, "*UND*")) continue;
+        // Skip section symbols (marked by 'd' in the 22nd column).
+        if ('d' == line[22]) continue;
+
+        // First part of an objdump line is the symbol address, parse that.
+        uint64_t line_address;
+        if (1 != sscanf(line, "%16lx", &line_address)) continue;
+        line_address += addr;
+
+        // Determine if the symbol is a function and if it is in the .rodata
+        // section.  Symbols in the .rodata section have a slightly longer
+        // line than those in the .text/.data sections (by 2 characters).
+        bool is_function = ('F' == line[23]);
+        size_t rodata = (NULL != strstr(line, ".rodata")) ? 2 : 0;
+
+        // Parse the symbol size.
+        uint64_t symbol_size;
+        if (1 != sscanf(&line[32+rodata], "%lx", &symbol_size)) continue;
+
+        // Parse the function name.
+        string function = &line[48+rodata];
+        function.resize(function.length() - 1); // remove the newline.
+
+        // Function have two addresses: TOC entry and code address.  Objdump
+        // gives the TOC entry, so we need to read the file itself to determine
+        // the code address.  The first part of the TOC entry is the code
+        // address.
+        uint64_t code_addr = 0;
+        if (is_function)
+        {
+            // Module is in the extended image, read from it.
+            if (line_address > g_extImageOffset)
+            {
+                // Read code address.
+                assert((line_address - g_extImageOffset) < g_extImageFileSize);
+                memcpy(&code_addr,
+                       &g_extImageFile[line_address - g_extImageOffset], 8);
+            }
+            // Module is in the base image.
+            else
+            {
+                // Read code address.
+                assert(line_address < g_imageFileSize);
+                memcpy(&code_addr, &g_imageFile[line_address], 8);
+            }
+            // Fix up the endianness.
+            code_addr = be64toh(code_addr);
+
+            std::swap(code_addr, line_address);
+        }
+
+        // Print all of this into a new line and add to the symbol map.
+        sprintf(line, "%c,%08lx,%08lx,%08lx,%s\n",
+            is_function ? 'F' : 'V',
+            line_address, code_addr, symbol_size,
+            function.c_str());
+
+        l_symbols.push_back(make_pair(line_address, line));
+
+    } while(1);
+
+    // Close subprocess (done).
+    pclose(pipe);
+
+    // Copy our local symbol list all at once into the global symbol list.
+    pthread_mutex_lock(&g_symbolMutex);
+    g_symbols.insert(l_symbols.begin(), l_symbols.end());
+    pthread_mutex_unlock(&g_symbolMutex);
+
+    return NULL;
+}
diff --git a/src/build/linker/makefile b/src/build/linker/makefile
index 3957cbf6a..1c21bb0cd 100644
--- a/src/build/linker/makefile
+++ b/src/build/linker/makefile
@@ -22,8 +22,8 @@
 # IBM_PROLOG_END_TAG
 ROOTPATH = ../../..
 
-CODE_PASS_POST += linker
-CLEAN_TARGETS += linker
+CODE_PASS_POST += linker gensyms
+CLEAN_TARGETS += linker gensyms
 
 include $(ROOTPATH)/config.mk
 
@@ -32,3 +32,7 @@ linker: linker.C
 	$(C1)$(CCACHE) $(HOST_PREFIX)g++ -O3 -g linker.C -o linker \
 				         -lbfd -liberty -lz
 
+gensyms: gensyms.C
+	$(C2) "    CXX        $(notdir $<)"
+	$(C1)$(CCACHE) $(HOST_PREFIX)g++ -O3 -g gensyms.C -o gensyms \
+					 -lpthread
diff --git a/src/build/mkrules/images.rules.mk b/src/build/mkrules/images.rules.mk
index 8c23ec23f..58bb468dd 100644
--- a/src/build/mkrules/images.rules.mk
+++ b/src/build/mkrules/images.rules.mk
@@ -58,7 +58,7 @@ $(IMGDIR)/%.bin: $(IMGDIR)/%.elf \
 $(IMGDIR)/%.list.bz2 $(IMGDIR)/%.syms: $(IMGDIR)/%.bin
 	$(C2) "    GENLIST    $(notdir $*)"
 	$(C1)(cd $(ROOTPATH); \
-              src/build/tools/gensyms $*.bin $*_extended.bin 0x40000000 \
+              src/build/linker/gensyms $*.bin $*_extended.bin 0x40000000 \
                   > ./img/$*.syms ; \
               src/build/tools/genlist $*.bin | bzip2 -zc > ./img/$*.list.bz2)
 
diff --git a/src/build/tools/gensyms b/src/build/tools/gensyms
deleted file mode 100755
index 993720045..000000000
--- a/src/build/tools/gensyms
+++ /dev/null
@@ -1,158 +0,0 @@
-#!/usr/bin/perl
-# IBM_PROLOG_BEGIN_TAG
-# This is an automatically generated prolog.
-#
-# $Source: src/build/tools/gensyms $
-#
-# IBM CONFIDENTIAL
-#
-# COPYRIGHT International Business Machines Corp. 2011,2013
-#
-# p1
-#
-# Object Code Only (OCO) source materials
-# Licensed Internal Code Source Materials
-# IBM HostBoot Licensed Internal Code
-#
-# The source code for this program is not published or otherwise
-# divested of its trade secrets, irrespective of what has been
-# deposited with the U.S. Copyright Office.
-#
-# Origin: 30
-#
-# IBM_PROLOG_END_TAG
-
-use strict;
-
-use IO::Seekable;
-
-my $littleendian = (unpack("L", pack("N", 0xabcd1234)) != 0xabcd1234);
-
-sub add_image_subdir
-{
-    my $image = shift;
-    if (!($image =~ m/\/img/)) { $image = "./img/".$image };
-    return $image;
-}
-
-my $image_offset = $ENV{"HAL_IMAGE_OFFSET"};
-if (not $image_offset) { $image_offset = "0x0"; };
-$image_offset = hex $image_offset;
-
-my $image;
-my $extimage;
-my $extoffset;
-my $all_modules = 0;
-my @modules = ();
-
-if ($#ARGV <= 1)
-{
-    die "gensyms <image> <extimage> <extoffset> [modules]\n";
-}
-
-if ($#ARGV == 2)
-{
-    $all_modules = 1;
-}
-else
-{
-    @modules = @ARGV[3..$#ARGV];
-}
-
-my $extoffset = hex $ARGV[2];
-
-$image = add_image_subdir($ARGV[0]);
-open IMAGE, "< $image";
-binmode(IMAGE);
-$extimage = add_image_subdir($ARGV[1]);
-open EXTIMAGE, "< $extimage";
-binmode(EXTIMAGE);
-
-my %module_offsets = ();
-open MODINFO, "< $image.modinfo";
-
-while (my $modline = <MODINFO>)
-{
-    chomp $modline;
-    my @splitline = split /,/, $modline;
-    $module_offsets{@splitline[0]} = (hex @splitline[1]) + $image_offset;
-    if ($all_modules)
-    {
-	push @modules, @splitline[0];
-    }
-}
-
-my @output = ();
-foreach my $module (@modules)
-{
-    # Only search modules that are likely to be ELF files.
-    if (not (($module =~ m/\.o/) or
-             ($module =~ m/\.elf/) or
-             ($module =~ m/\.so/)))
-    {
-        next;
-    }
-
-    my $PREFIX = $ENV{'CROSS_PREFIX'};
-    open OBJDUMP, ("${PREFIX}objdump --syms -C ".add_image_subdir($module)."|");
-    while (my $line = <OBJDUMP>)
-    {
-	if (($line =~ m/\*ABS\*/) || ($line =~ m/\*UND\*/))
-	{
-	    next;
-	}
-	if ("d" eq substr($line, 22, 1))
-	{
-	    next;
-	}
-	if (!($line =~ m/^[0-9a-f]{16}/))
-	{
-	    next;
-	}
-
-	$line =~ s/[\s]*$//;
-	
-	my $address = (hex substr($line, 0, 16)) + $module_offsets{$module};
-	my $is_function = ("F" eq substr($line, 23, 1));
-        my $size = (hex substr($line, 32, 16));
-	my $name = substr($line, 48);
-	my $code_loc = 0;
-	if ($is_function)
-	{
-            if (($address - $image_offset) > $extoffset)
-            {
-                seek EXTIMAGE,
-                     ($address - ($image_offset + $extoffset)), SEEK_SET;
-                read EXTIMAGE, $code_loc, 8;
-            }
-            else
-            {
-                seek IMAGE, ($address - $image_offset), SEEK_SET;
-                read IMAGE, $code_loc, 8;
-            }
-            if ($littleendian)
-            {
-                $code_loc = unpack("Q", reverse($code_loc)) + $image_offset;
-            }
-            else
-            {
-                $code_loc = unpack("Q", $code_loc) + $image_offset;
-            }
-
-            my $tmp = $code_loc; $code_loc = $address; $address = $tmp;
-	}
-	my $outstring = "";
-	$outstring = sprintf "%s,%08x,%08x,%08x,%s\n", ($is_function?"F":"V"),$address,$code_loc,$size,$name;
-
-	push @output, $outstring;
-    }
-    close OBJDUMP;
-}
-
-close IMAGE;
-close EXTIMAGE;
-
-foreach my $outstring (sort { substr($a,2) cmp substr($b,2) } @output)
-{
-    print $outstring;
-}