/*--------------------------------------------------------------------*/
/*--- Startup: the real stuff m_main.c ---*/
/*--------------------------------------------------------------------*/
/*
This file is part of Valgrind, a dynamic binary instrumentation
framework.
Copyright (C) 2000-2015 Julian Seward
jseward@acm.org
This program is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License as
published by the Free Software Foundation; either version 2 of the
License, or (at your option) any later version.
This program is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
02111-1307, USA.
The GNU General Public License is contained in the file COPYING.
*/
#include "pub_core_basics.h"
#include "pub_core_vki.h"
#include "pub_core_vkiscnums.h"
#include "pub_core_threadstate.h"
#include "pub_core_xarray.h"
#include "pub_core_clientstate.h"
#include "pub_core_aspacemgr.h"
#include "pub_core_aspacehl.h"
#include "pub_core_commandline.h"
#include "pub_core_debuglog.h"
#include "pub_core_errormgr.h"
#include "pub_core_execontext.h"
#include "pub_core_gdbserver.h"
#include "pub_core_initimg.h"
#include "pub_core_libcbase.h"
#include "pub_core_libcassert.h"
#include "pub_core_libcfile.h"
#include "pub_core_libcprint.h"
#include "pub_core_libcproc.h"
#include "pub_core_libcsignal.h"
#include "pub_core_sbprofile.h"
#include "pub_core_syscall.h" // VG_(strerror)
#include "pub_core_mach.h"
#include "pub_core_machine.h"
#include "pub_core_mallocfree.h"
#include "pub_core_options.h"
#include "pub_core_debuginfo.h"
#include "pub_core_redir.h"
#include "pub_core_scheduler.h"
#include "pub_core_seqmatch.h" // For VG_(string_match)
#include "pub_core_signals.h"
#include "pub_core_stacks.h" // For VG_(register_stack)
#include "pub_core_syswrap.h"
#include "pub_core_tooliface.h"
#include "pub_core_translate.h" // For VG_(translate)
#include "pub_core_trampoline.h"
#include "pub_core_transtab.h"
#include "pub_core_inner.h"
#if defined(ENABLE_INNER_CLIENT_REQUEST)
#include "pub_core_clreq.h"
#endif
/*====================================================================*/
/*=== Command-line: variables, processing, etc ===*/
/*====================================================================*/
// See pub_{core,tool}_options.h for explanations of all these.
static void usage_NORETURN ( Bool debug_help )
{
/* 'usage1' contains a %s
- for the name of the GDB executable
- for the name of vgdb's path prefix
which must be supplied when they are VG_(printf)'d. */
const HChar usage1[] =
"usage: valgrind [options] prog-and-args\n"
"\n"
" tool-selection option, with default in [ ]:\n"
" --tool=<name> use the Valgrind tool named <name> [memcheck]\n"
"\n"
" basic user options for all Valgrind tools, with defaults in [ ]:\n"
" -h --help show this message\n"
" --help-debug show this message, plus debugging options\n"
" --version show version\n"
" -q --quiet run silently; only print error msgs\n"
" -v --verbose be more verbose -- show misc extra info\n"
" --trace-children=no|yes Valgrind-ise child processes (follow execve)? [no]\n"
" --trace-children-skip=patt1,patt2,... specifies a list of executables\n"
" that --trace-children=yes should not trace into\n"
" --trace-children-skip-by-arg=patt1,patt2,... same as --trace-children-skip=\n"
" but check the argv[] entries for children, rather\n"
" than the exe name, to make a follow/no-follow decision\n"
" --child-silent-after-fork=no|yes omit child output between fork & exec? [no]\n"
" --vgdb=no|yes|full activate gdbserver? [yes]\n"
" full is slower but provides precise watchpoint/step\n"
" --vgdb-error=<number> invoke gdbserver after <number> errors [%d]\n"
" to get started quickly, use --vgdb-error=0\n"
" and follow the on-screen directions\n"
" --vgdb-stop-at=event1,event2,... invoke gdbserver for given events [none]\n"
" where event is one of:\n"
" startup exit valgrindabexit all none\n"
" --track-fds=no|yes track open file descriptors? [no]\n"
" --time-stamp=no|yes add timestamps to log messages? [no]\n"
" --log-fd=<number> log messages to file descriptor [2=stderr]\n"
" --log-file=<file> log messages to <file>\n"
" --log-socket=ipaddr:port log messages to socket ipaddr:port\n"
"\n"
" user options for Valgrind tools that report errors:\n"
" --xml=yes emit error output in XML (some tools only)\n"
" --xml-fd=<number> XML output to file descriptor\n"
" --xml-file=<file> XML output to <file>\n"
" --xml-socket=ipaddr:port XML output to socket ipaddr:port\n"
" --xml-user-comment=STR copy STR verbatim into XML output\n"
" --demangle=no|yes automatically demangle C++ names? [yes]\n"
" --num-callers=<number> show <number> callers in stack traces [12]\n"
" --error-limit=no|yes stop showing new errors if too many? [yes]\n"
" --error-exitcode=<number> exit code to return if errors found [0=disable]\n"
" --error-markers=<begin>,<end> add lines with begin/end markers before/after\n"
" each error output in plain text mode [none]\n"
" --show-below-main=no|yes continue stack traces below main() [no]\n"
" --default-suppressions=yes|no\n"
" load default suppressions [yes]\n"
" --suppressions=<filename> suppress errors described in <filename>\n"
" --gen-suppressions=no|yes|all print suppressions for errors? [no]\n"
" --input-fd=<number> file descriptor for input [0=stdin]\n"
" --dsymutil=no|yes run dsymutil on Mac OS X when helpful? [yes]\n"
" --max-stackframe=<number> assume stack switch for SP changes larger\n"
" than <number> bytes [2000000]\n"
" --main-stacksize=<number> set size of main thread's stack (in bytes)\n"
" [min(max(current 'ulimit' value,1MB),16MB)]\n"
"\n"
" user options for Valgrind tools that replace malloc:\n"
" --alignment=<number> set minimum alignment of heap allocations [%s]\n"
" --redzone-size=<number> set minimum size of redzones added before/after\n"
" heap blocks (in bytes). [%s]\n"
"\n"
" uncommon user options for all Valgrind tools:\n"
" --fullpath-after= (with nothing after the '=')\n"
" show full source paths in call stacks\n"
" --fullpath-after=string like --fullpath-after=, but only show the\n"
" part of the path after 'string'. Allows removal\n"
" of path prefixes. Use this flag multiple times\n"
" to specify a set of prefixes to remove.\n"
" --extra-debuginfo-path=path absolute path to search for additional\n"
" debug symbols, in addition to existing default\n"
" well known search paths.\n"
" --debuginfo-server=ipaddr:port also query this server\n"
" (valgrind-di-server) for debug symbols\n"
" --allow-mismatched-debuginfo=no|yes [no]\n"
" for the above two flags only, accept debuginfo\n"
" objects that don't \"match\" the main object\n"
" --smc-check=none|stack|all|all-non-file [all-non-file]\n"
" checks for self-modifying code: none, only for\n"
" code found in stacks, for all code, or for all\n"
" code except that from file-backed mappings\n"
" --read-inline-info=yes|no read debug info about inlined function calls\n"
" and use it to do better stack traces. [yes]\n"
" on Linux/Android/Solaris for Memcheck/Helgrind/DRD\n"
" only. [no] for all other tools and platforms.\n"
" --read-var-info=yes|no read debug info on stack and global variables\n"
" and use it to print better error messages in\n"
" tools that make use of it (Memcheck, Helgrind,\n"
" DRD) [no]\n"
" --vgdb-poll=<number> gdbserver poll max every <number> basic blocks [%d] \n"
" --vgdb-shadow-registers=no|yes let gdb see the shadow registers [no]\n"
" --vgdb-prefix=<prefix> prefix for vgdb FIFOs [%s]\n"
" --run-libc-freeres=no|yes free up glibc memory at exit on Linux? [yes]\n"
" --sim-hints=hint1,hint2,... activate unusual sim behaviours [none] \n"
" where hint is one of:\n"
" lax-ioctls lax-doors fuse-compatible enable-outer\n"
" no-inner-prefix no-nptl-pthread-stackcache none\n"
" --fair-sched=no|yes|try schedule threads fairly on multicore systems [no]\n"
" --kernel-variant=variant1,variant2,...\n"
" handle non-standard kernel variants [none]\n"
" where variant is one of:\n"
" bproc android-no-hw-tls\n"
" android-gpu-sgx5xx android-gpu-adreno3xx none\n"
" --merge-recursive-frames=<number> merge frames between identical\n"
" program counters in max <number> frames) [0]\n"
" --num-transtab-sectors=<number> size of translated code cache [%d]\n"
" more sectors may increase performance, but use more memory.\n"
" --avg-transtab-entry-size=<number> avg size in bytes of a translated\n"
" basic block [0, meaning use tool provided default]\n"
" --aspace-minaddr=0xPP avoid mapping memory below 0xPP [guessed]\n"
" --valgrind-stacksize=<number> size of valgrind (host) thread's stack\n"
" (in bytes) ["
VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)
"]\n"
" --show-emwarns=no|yes show warnings about emulation limits? [no]\n"
" --require-text-symbol=:sonamepattern:symbolpattern abort run if the\n"
" stated shared object doesn't have the stated\n"
" text symbol. Patterns can contain ? and *.\n"
" --soname-synonyms=syn1=pattern1,syn2=pattern2,... synonym soname\n"
" specify patterns for function wrapping or replacement.\n"
" To use a non-libc malloc library that is\n"
" in the main exe: --soname-synonyms=somalloc=NONE\n"
" in libxyzzy.so: --soname-synonyms=somalloc=libxyzzy.so\n"
" --sigill-diagnostics=yes|no warn about illegal instructions? [yes]\n"
" --unw-stack-scan-thresh=<number> Enable stack-scan unwind if fewer\n"
" than <number> good frames found [0, meaning \"disabled\"]\n"
" NOTE: stack scanning is only available on arm-linux.\n"
" --unw-stack-scan-frames=<number> Max number of frames that can be\n"
" recovered by stack scanning [5]\n"
" --resync-filter=no|yes|verbose [yes on MacOS, no on other OSes]\n"
" attempt to avoid expensive address-space-resync operations\n"
" --max-threads=<number> maximum number of threads that valgrind can\n"
" handle [%d]\n"
"\n";
const HChar usage2[] =
"\n"
" debugging options for all Valgrind tools:\n"
" -d show verbose debugging output\n"
" --stats=no|yes show tool and core statistics [no]\n"
" --sanity-level=<number> level of sanity checking to do [1]\n"
" --trace-flags=<XXXXXXXX> show generated code? (X = 0|1) [00000000]\n"
" --profile-flags=<XXXXXXXX> ditto, but for profiling (X = 0|1) [00000000]\n"
" --profile-interval=<number> show profile every <number> event checks\n"
" [0, meaning only at the end of the run]\n"
" --trace-notbelow=<number> only show BBs above <number> [999999999]\n"
" --trace-notabove=<number> only show BBs below <number> [0]\n"
" --trace-syscalls=no|yes show all system calls? [no]\n"
" --trace-signals=no|yes show signal handling details? [no]\n"
" --trace-symtab=no|yes show symbol table details? [no]\n"
" --trace-symtab-patt=<patt> limit debuginfo tracing to obj name <patt>\n"
" --trace-cfi=no|yes show call-frame-info details? [no]\n"
" --debug-dump=syms mimic /usr/bin/readelf --syms\n"
" --debug-dump=line mimic /usr/bin/readelf --debug-dump=line\n"
" --debug-dump=frames mimic /usr/bin/readelf --debug-dump=frames\n"
" --trace-redir=no|yes show redirection details? [no]\n"
" --trace-sched=no|yes show thread scheduler details? [no]\n"
" --profile-heap=no|yes profile Valgrind's own space use\n"
" --core-redzone-size=<number> set minimum size of redzones added before/after\n"
" heap blocks allocated for Valgrind internal use (in bytes) [4]\n"
" --wait-for-gdb=yes|no pause on startup to wait for gdb attach\n"
" --sym-offsets=yes|no show syms in form 'name+offset' ? [no]\n"
" --command-line-only=no|yes only use command line options [no]\n"
"\n"
" Vex options for all Valgrind tools:\n"
" --vex-iropt-verbosity=<0..9> [0]\n"
" --vex-iropt-level=<0..2> [2]\n"
" --vex-iropt-unroll-thresh=<0..400> [120]\n"
" --vex-guest-max-insns=<1..100> [50]\n"
" --vex-guest-chase-thresh=<0..99> [10]\n"
" --vex-guest-chase-cond=no|yes [no]\n"
" Precise exception control. Possible values for 'mode' are as follows\n"
" and specify the minimum set of registers guaranteed to be correct\n"
" immediately prior to memory access instructions:\n"
" sp-at-mem-access stack pointer only\n"
" unwindregs-at-mem-access registers needed for stack unwinding\n"
" allregs-at-mem-access all registers\n"
" allregs-at-each-insn all registers are always correct\n"
" Default value for all 3 following flags is [unwindregs-at-mem-access].\n"
" --vex-iropt-register-updates=mode setting to use by default\n"
" --px-default=mode synonym for --vex-iropt-register-updates\n"
" --px-file-backed=mode optional setting for file-backed (non-JIT) code\n"
" Tracing and profile control:\n"
" --trace-flags and --profile-flags values (omit the middle space):\n"
" 1000 0000 show conversion into IR\n"
" 0100 0000 show after initial opt\n"
" 0010 0000 show after instrumentation\n"
" 0001 0000 show after second opt\n"
" 0000 1000 show after tree building\n"
" 0000 0100 show selecting insns\n"
" 0000 0010 show after reg-alloc\n"
" 0000 0001 show final assembly\n"
" 0000 0000 show summary profile only\n"
" (Nb: you need --trace-notbelow and/or --trace-notabove\n"
" with --trace-flags for full details)\n"
"\n"
" debugging options for Valgrind tools that report errors\n"
" --dump-error=<number> show translation for basic block associated\n"
" with <number>'th error context [0=show none]\n"
"\n"
" debugging options for Valgrind tools that replace malloc:\n"
" --trace-malloc=no|yes show client malloc details? [no]\n"
"\n";
const HChar usage3[] =
"\n"
" Extra options read from ~/.valgrindrc, $VALGRIND_OPTS, ./.valgrindrc\n"
"\n"
" %s is %s\n"
" Valgrind is Copyright (C) 2000-2015, and GNU GPL'd, by Julian Seward et al.\n"
" LibVEX is Copyright (C) 2004-2015, and GNU GPL'd, by OpenWorks LLP et al.\n"
"\n"
" Bug reports, feedback, admiration, abuse, etc, to: %s.\n"
"\n";
HChar default_alignment[30]; // large enough
HChar default_redzone_size[30]; // large enough
// Ensure the message goes to stdout
VG_(log_output_sink).fd = 1;
VG_(log_output_sink).is_socket = False;
if (VG_(needs).malloc_replacement) {
VG_(sprintf)(default_alignment, "%d", VG_MIN_MALLOC_SZB);
VG_(sprintf)(default_redzone_size, "%lu", VG_(tdict).tool_client_redzone_szB);
} else {
VG_(strcpy)(default_alignment, "not used by this tool");
VG_(strcpy)(default_redzone_size, "not used by this tool");
}
/* 'usage1' a type as described after each arg. */
VG_(printf)(usage1,
VG_(clo_vgdb_error) /* int */,
default_alignment /* char* */,
default_redzone_size /* char* */,
VG_(clo_vgdb_poll) /* int */,
VG_(vgdb_prefix_default)() /* char* */,
N_SECTORS_DEFAULT /* int */,
MAX_THREADS_DEFAULT /* int */
);
if (VG_(details).name) {
VG_(printf)(" user options for %s:\n", VG_(details).name);
if (VG_(needs).command_line_options)
VG_TDICT_CALL(tool_print_usage);
else
VG_(printf)(" (none)\n");
}
if (debug_help) {
VG_(printf)("%s", usage2);
if (VG_(details).name) {
VG_(printf)(" debugging options for %s:\n", VG_(details).name);
if (VG_(needs).command_line_options)
VG_TDICT_CALL(tool_print_debug_usage);
else
VG_(printf)(" (none)\n");
}
}
VG_(printf)(usage3, VG_(details).name, VG_(details).copyright_author,
VG_BUGS_TO);
VG_(exit)(0);
}
/* Peer at previously set up VG_(args_for_valgrind) and do some
minimal command line processing that must happen early on:
- show the version string, if requested (-v)
- extract any request for help (--help, -h, --help-debug)
- get the toolname (--tool=)
- set VG_(clo_max_stackframe) (--max-stackframe=)
- set VG_(clo_main_stacksize) (--main-stacksize=)
- set VG_(clo_sim_hints) (--sim-hints=)
- set VG_(clo_max_threads) (--max-threads)
That's all it does. The main command line processing is done below
by main_process_cmd_line_options. Note that
main_process_cmd_line_options has to handle but ignore the ones we
have handled here.
*/
static void early_process_cmd_line_options ( /*OUT*/Int* need_help,
/*OUT*/const HChar** tool )
{
UInt i;
HChar* str;
vg_assert( VG_(args_for_valgrind) );
/* parse the options we have (only the options we care about now) */
for (i = 0; i < VG_(sizeXA)( VG_(args_for_valgrind) ); i++) {
str = * (HChar**) VG_(indexXA)( VG_(args_for_valgrind), i );
vg_assert(str);
// Nb: the version string goes to stdout.
if VG_XACT_CLO(str, "--version", VG_(log_output_sink).fd, 1) {
VG_(log_output_sink).is_socket = False;
VG_(printf)("valgrind-" VERSION "\n");
VG_(exit)(0);
}
else if VG_XACT_CLO(str, "--help", *need_help, *need_help+1) {}
else if VG_XACT_CLO(str, "-h", *need_help, *need_help+1) {}
else if VG_XACT_CLO(str, "--help-debug", *need_help, *need_help+2) {}
// The tool has already been determined, but we need to know the name
// here.
else if VG_STR_CLO(str, "--tool", *tool) {}
// Set up VG_(clo_max_stackframe) and VG_(clo_main_stacksize).
// These are needed by VG_(ii_create_image), which happens
// before main_process_cmd_line_options().
else if VG_INT_CLO(str, "--max-stackframe", VG_(clo_max_stackframe)) {}
else if VG_INT_CLO(str, "--main-stacksize", VG_(clo_main_stacksize)) {}
// Set up VG_(clo_max_threads); needed for VG_(tl_pre_clo_init)
else if VG_INT_CLO(str, "--max-threads", VG_(clo_max_threads)) {}
// Set up VG_(clo_sim_hints). This is needed a.o. for an inner
// running in an outer, to have "no-inner-prefix" enabled
// as early as possible.
else if VG_USETX_CLO (str, "--sim-hints",
"lax-ioctls,lax-doors,fuse-compatible,"
"enable-outer,no-inner-prefix,"
"no-nptl-pthread-stackcache",
VG_(clo_sim_hints)) {}
}
/* For convenience */
VG_N_THREADS = VG_(clo_max_threads);
}
/* The main processing for command line options. See comments above
on early_process_cmd_line_options.
Comments on how the logging options are handled:
User can specify:
--log-fd= for a fd to write to (default setting, fd = 2)
--log-file= for a file name to write to
--log-socket= for a socket to write to
As a result of examining these and doing relevant socket/file
opening, a final fd is established. This is stored in
VG_(log_output_sink) in m_libcprint. Also, if --log-file=STR was
specified, then STR, after expansion of %p and %q templates within
it, is stored in VG_(clo_log_fname_expanded), in m_options, just in
case anybody wants to know what it is.
When printing, VG_(log_output_sink) is consulted to find the
fd to send output to.
Exactly analogous actions are undertaken for the XML output
channel, with the one difference that the default fd is -1, meaning
the channel is disabled by default.
*/
static
void main_process_cmd_line_options ( /*OUT*/Bool* logging_to_fd,
/*OUT*/const HChar** xml_fname_unexpanded,
const HChar* toolname )
{
// VG_(clo_log_fd) is used by all the messaging. It starts as 2 (stderr)
// and we cannot change it until we know what we are changing it to is
// ok. So we have tmp_log_fd to hold the tmp fd prior to that point.
SysRes sres;
Int i, tmp_log_fd, tmp_xml_fd;
Int toolname_len = VG_(strlen)(toolname);
const HChar* tmp_str; // Used in a couple of places.
enum {
VgLogTo_Fd,
VgLogTo_File,
VgLogTo_Socket
} log_to = VgLogTo_Fd, // Where is logging output to be sent?
xml_to = VgLogTo_Fd; // Where is XML output to be sent?
/* Temporarily holds the string STR specified with
--{log,xml}-{name,socket}=STR. 'fs' stands for
file-or-socket. */
const HChar* log_fsname_unexpanded = NULL;
const HChar* xml_fsname_unexpanded = NULL;
/* Whether the user has explicitly provided --sigill-diagnostics.
If not explicitly given depends on general verbosity setting. */
Bool sigill_diag_set = False;
/* Log to stderr by default, but usage message goes to stdout. XML
output is initially disabled. */
tmp_log_fd = 2;
tmp_xml_fd = -1;
/* Check for sane path in ./configure --prefix=... */
if (VG_LIBDIR[0] != '/')
VG_(err_config_error)("Please use absolute paths in "
"./configure --prefix=... or --libdir=...\n");
vg_assert( VG_(args_for_valgrind) );
VG_(clo_suppressions) = VG_(newXA)(VG_(malloc), "main.mpclo.4",
VG_(free), sizeof(HChar *));
VG_(clo_fullpath_after) = VG_(newXA)(VG_(malloc), "main.mpclo.5",
VG_(free), sizeof(HChar *));
VG_(clo_req_tsyms) = VG_(newXA)(VG_(malloc), "main.mpclo.6",
VG_(free), sizeof(HChar *));
/* Constants for parsing PX control flags. */
const HChar* pxStrings[5]
= { "sp-at-mem-access", "unwindregs-at-mem-access",
"allregs-at-mem-access", "allregs-at-each-insn", NULL };
const VexRegisterUpdates pxVals[5]
= { VexRegUpdSpAtMemAccess, VexRegUpdUnwindregsAtMemAccess,
VexRegUpdAllregsAtMemAccess, VexRegUpdAllregsAtEachInsn, 0/*inval*/ };
/* BEGIN command-line processing loop */
for (i = 0; i < VG_(sizeXA)( VG_(args_for_valgrind) ); i++) {
HChar* arg = * (HChar**) VG_(indexXA)( VG_(args_for_valgrind), i );
HChar* colon = arg;
UInt ix = 0;
// Look for a colon in the option name.
while (*colon && *colon != ':' && *colon != '=')
colon++;
// Does it have the form "--toolname:foo"? We have to do it at the start
// in case someone has combined a prefix with a core-specific option,
// eg. "--memcheck:verbose".
if (*colon == ':') {
if (VG_STREQN(2, arg, "--") &&
VG_STREQN(toolname_len, arg+2, toolname) &&
VG_STREQN(1, arg+2+toolname_len, ":"))
{
// Prefix matches, convert "--toolname:foo" to "--foo".
// Two things to note:
// - We cannot modify the option in-place. If we did, and then
// a child was spawned with --trace-children=yes, the
// now-non-prefixed option would be passed and could screw up
// the child.
// - We create copies, and never free them. Why? Non-prefixed
// options hang around forever, so tools need not make copies
// of strings within them. We need to have the same behaviour
// for prefixed options. The pointer to the copy will be lost
// once we leave this function (although a tool may keep a
// pointer into it), but the space wasted is insignificant.
// (In bug #142197, the copies were being freed, which caused
// problems for tools that reasonably assumed that arguments
// wouldn't disappear on them.)
if (0)
VG_(printf)("tool-specific arg: %s\n", arg);
arg = VG_(strdup)("main.mpclo.1", arg + toolname_len + 1);
arg[0] = '-';
arg[1] = '-';
} else {
// prefix doesn't match, skip to next arg
continue;
}
}
/* Ignore these options - they've already been handled */
if VG_STREQN( 7, arg, "--tool=") {}
else if VG_STREQN(20, arg, "--command-line-only=") {}
else if VG_STREQ( arg, "--") {}
else if VG_STREQ( arg, "-d") {}
else if VG_STREQN(17, arg, "--max-stackframe=") {}
else if VG_STREQN(17, arg, "--main-stacksize=") {}
else if VG_STREQN(14, arg, "--max-threads=") {}
else if VG_STREQN(12, arg, "--sim-hints=") {}
else if VG_STREQN(15, arg, "--profile-heap=") {}
else if VG_STREQN(20, arg, "--core-redzone-size=") {}
else if VG_STREQN(15, arg, "--redzone-size=") {}
else if VG_STREQN(17, arg, "--aspace-minaddr=") {}
else if VG_BINT_CLO(arg, "--valgrind-stacksize",
VG_(clo_valgrind_stacksize),
2*VKI_PAGE_SIZE, 10*VG_DEFAULT_STACK_ACTIVE_SZB)
{VG_(clo_valgrind_stacksize)
= VG_PGROUNDUP(VG_(clo_valgrind_stacksize));}
/* Obsolete options. Report an error and exit */
else if VG_STREQN(34, arg, "--vex-iropt-precise-memory-exns=no") {
VG_(fmsg_bad_option)
(arg,
"--vex-iropt-precise-memory-exns is obsolete\n"
"Use --vex-iropt-register-updates=unwindregs-at-mem-access instead\n");
}
else if VG_STREQN(35, arg, "--vex-iropt-precise-memory-exns=yes") {
VG_(fmsg_bad_option)
(arg,
"--vex-iropt-precise-memory-exns is obsolete\n"
"Use --vex-iropt-register-updates=allregs-at-mem-access instead\n"
" (or --vex-iropt-register-updates=allregs-at-each-insn)\n");
}
// These options are new.
else if (VG_STREQ(arg, "-v") ||
VG_STREQ(arg, "--verbose"))
VG_(clo_verbosity)++;
else if (VG_STREQ(arg, "-q") ||
VG_STREQ(arg, "--quiet"))
VG_(clo_verbosity)--;
else if VG_BOOL_CLO(arg, "--sigill-diagnostics", VG_(clo_sigill_diag))
sigill_diag_set = True;
else if VG_BOOL_CLO(arg, "--stats", VG_(clo_stats)) {}
else if VG_BOOL_CLO(arg, "--xml", VG_(clo_xml))
VG_(debugLog_setXml)(VG_(clo_xml));
else if VG_XACT_CLO(arg, "--vgdb=no", VG_(clo_vgdb), Vg_VgdbNo) {}
else if VG_XACT_CLO(arg, "--vgdb=yes", VG_(clo_vgdb), Vg_VgdbYes) {}
else if VG_XACT_CLO(arg, "--vgdb=full", VG_(clo_vgdb), Vg_VgdbFull) {
/* automatically updates register values at each insn
with --vgdb=full */
VG_(clo_vex_control).iropt_register_updates_default
= VG_(clo_px_file_backed)
= VexRegUpdAllregsAtEachInsn;
}
else if VG_INT_CLO (arg, "--vgdb-poll", VG_(clo_vgdb_poll)) {}
else if VG_INT_CLO (arg, "--vgdb-error", VG_(clo_vgdb_error)) {}
else if VG_USET_CLO (arg, "--vgdb-stop-at",
"startup,exit,valgrindabexit",
VG_(clo_vgdb_stop_at)) {}
else if VG_STR_CLO (arg, "--vgdb-prefix", VG_(clo_vgdb_prefix)) {
VG_(arg_vgdb_prefix) = arg;
}
else if VG_BOOL_CLO(arg, "--vgdb-shadow-registers",
VG_(clo_vgdb_shadow_registers)) {}
else if VG_BOOL_CLO(arg, "--demangle", VG_(clo_demangle)) {}
else if VG_STR_CLO (arg, "--soname-synonyms",VG_(clo_soname_synonyms)) {}
else if VG_BOOL_CLO(arg, "--error-limit", VG_(clo_error_limit)) {}
else if VG_INT_CLO (arg, "--error-exitcode", VG_(clo_error_exitcode)) {}
else if VG_STR_CLO (arg, "--error-markers", tmp_str) {
Int m;
const HChar *startpos = tmp_str;
const HChar *nextpos;
for (m = 0;
m < sizeof(VG_(clo_error_markers))
/sizeof(VG_(clo_error_markers)[0]);
m++) {
/* Release previous value if clo given multiple times. */
VG_(free)(VG_(clo_error_markers)[m]);
VG_(clo_error_markers)[m] = NULL;
nextpos = VG_(strchr)(startpos, ',');
if (!nextpos)
nextpos = startpos + VG_(strlen)(startpos);
if (startpos != nextpos) {
VG_(clo_error_markers)[m]
= VG_(malloc)("main.mpclo.2", nextpos - startpos + 1);
VG_(memcpy)(VG_(clo_error_markers)[m], startpos,
nextpos - startpos);
VG_(clo_error_markers)[m][nextpos - startpos] = '\0';
}
startpos = *nextpos ? nextpos + 1 : nextpos;
}
}
else if VG_BOOL_CLO(arg, "--show-emwarns", VG_(clo_show_emwarns)) {}
else if VG_BOOL_CLO(arg, "--run-libc-freeres", VG_(clo_run_libc_freeres)) {}
else if VG_BOOL_CLO(arg, "--show-below-main", VG_(clo_show_below_main)) {}
else if VG_BOOL_CLO(arg, "--time-stamp", VG_(clo_time_stamp)) {}
else if VG_BOOL_CLO(arg, "--track-fds", VG_(clo_track_fds)) {}
else if VG_BOOL_CLO(arg, "--trace-children", VG_(clo_trace_children)) {}
else if VG_BOOL_CLO(arg, "--child-silent-after-fork",
VG_(clo_child_silent_after_fork)) {}
else if VG_STR_CLO(arg, "--fair-sched", tmp_str) {
if (VG_(strcmp)(tmp_str, "yes") == 0)
VG_(clo_fair_sched) = enable_fair_sched;
else if (VG_(strcmp)(tmp_str, "try") == 0)
VG_(clo_fair_sched) = try_fair_sched;
else if (VG_(strcmp)(tmp_str, "no") == 0)
VG_(clo_fair_sched) = disable_fair_sched;
else
VG_(fmsg_bad_option)(arg,
"Bad argument, should be 'yes', 'try' or 'no'\n");
}
else if VG_BOOL_CLO(arg, "--trace-sched", VG_(clo_trace_sched)) {}
else if VG_BOOL_CLO(arg, "--trace-signals", VG_(clo_trace_signals)) {}
else if VG_BOOL_CLO(arg, "--trace-symtab", VG_(clo_trace_symtab)) {}
else if VG_STR_CLO (arg, "--trace-symtab-patt", VG_(clo_trace_symtab_patt)) {}
else if VG_BOOL_CLO(arg, "--trace-cfi", VG_(clo_trace_cfi)) {}
else if VG_XACT_CLO(arg, "--debug-dump=syms", VG_(clo_debug_dump_syms),
True) {}
else if VG_XACT_CLO(arg, "--debug-dump=line", VG_(clo_debug_dump_line),
True) {}
else if VG_XACT_CLO(arg, "--debug-dump=frames",
VG_(clo_debug_dump_frames), True) {}
else if VG_BOOL_CLO(arg, "--trace-redir", VG_(clo_trace_redir)) {}
else if VG_BOOL_CLO(arg, "--trace-syscalls", VG_(clo_trace_syscalls)) {}
else if VG_BOOL_CLO(arg, "--wait-for-gdb", VG_(clo_wait_for_gdb)) {}
else if VG_BOOL_CLO(arg, "--sym-offsets", VG_(clo_sym_offsets)) {}
else if VG_BOOL_CLO(arg, "--read-inline-info", VG_(clo_read_inline_info)) {}
else if VG_BOOL_CLO(arg, "--read-var-info", VG_(clo_read_var_info)) {}
else if VG_INT_CLO (arg, "--dump-error", VG_(clo_dump_error)) {}
else if VG_INT_CLO (arg, "--input-fd", VG_(clo_input_fd)) {}
else if VG_INT_CLO (arg, "--sanity-level", VG_(clo_sanity_level)) {}
else if VG_BINT_CLO(arg, "--num-callers", VG_(clo_backtrace_size), 1,
VG_DEEPEST_BACKTRACE) {}
else if VG_BINT_CLO(arg, "--num-transtab-sectors",
VG_(clo_num_transtab_sectors),
MIN_N_SECTORS, MAX_N_SECTORS) {}
else if VG_BINT_CLO(arg, "--avg-transtab-entry-size",
VG_(clo_avg_transtab_entry_size),
50, 5000) {}
else if VG_BINT_CLO(arg, "--merge-recursive-frames",
VG_(clo_merge_recursive_frames), 0,
VG_DEEPEST_BACKTRACE) {}
else if VG_XACT_CLO(arg, "--smc-check=none",
VG_(clo_smc_check), Vg_SmcNone) {}
else if VG_XACT_CLO(arg, "--smc-check=stack",
VG_(clo_smc_check), Vg_SmcStack) {}
else if VG_XACT_CLO(arg, "--smc-check=all",
VG_(clo_smc_check), Vg_SmcAll) {}
else if VG_XACT_CLO(arg, "--smc-check=all-non-file",
VG_(clo_smc_check), Vg_SmcAllNonFile) {}
else if VG_USETX_CLO (arg, "--kernel-variant",
"bproc,"
"android-no-hw-tls,"
"android-gpu-sgx5xx,"
"android-gpu-adreno3xx",
VG_(clo_kernel_variant)) {}
else if VG_BOOL_CLO(arg, "--dsymutil", VG_(clo_dsymutil)) {}
else if VG_STR_CLO (arg, "--trace-children-skip",
VG_(clo_trace_children_skip)) {}
else if VG_STR_CLO (arg, "--trace-children-skip-by-arg",
VG_(clo_trace_children_skip_by_arg)) {}
else if VG_BINT_CLO(arg, "--vex-iropt-verbosity",
VG_(clo_vex_control).iropt_verbosity, 0, 10) {}
else if VG_BINT_CLO(arg, "--vex-iropt-level",
VG_(clo_vex_control).iropt_level, 0, 2) {}
else if VG_STRINDEX_CLO(arg, "--vex-iropt-register-updates",
pxStrings, ix) {
vg_assert(ix < 4);
vg_assert(pxVals[ix] >= VexRegUpdSpAtMemAccess);
vg_assert(pxVals[ix] <= VexRegUpdAllregsAtEachInsn);
VG_(clo_vex_control).iropt_register_updates_default = pxVals[ix];
}
else if VG_STRINDEX_CLO(arg, "--px-default", pxStrings, ix) {
// NB: --px-default is an alias for the hard-to-remember
// --vex-iropt-register-updates, hence the same logic.
vg_assert(ix < 4);
vg_assert(pxVals[ix] >= VexRegUpdSpAtMemAccess);
vg_assert(pxVals[ix] <= VexRegUpdAllregsAtEachInsn);
VG_(clo_vex_control).iropt_register_updates_default = pxVals[ix];
}
else if VG_STRINDEX_CLO(arg, "--px-file-backed", pxStrings, ix) {
// Whereas --px-file-backed isn't
// the same flag as --vex-iropt-register-updates.
vg_assert(ix < 4);
vg_assert(pxVals[ix] >= VexRegUpdSpAtMemAccess);
vg_assert(pxVals[ix] <= VexRegUpdAllregsAtEachInsn);
VG_(clo_px_file_backed) = pxVals[ix];
}
else if VG_BINT_CLO(arg, "--vex-iropt-unroll-thresh",
VG_(clo_vex_control).iropt_unroll_thresh, 0, 400) {}
else if VG_BINT_CLO(arg, "--vex-guest-max-insns",
VG_(clo_vex_control).guest_max_insns, 1, 100) {}
else if VG_BINT_CLO(arg, "--vex-guest-chase-thresh",
VG_(clo_vex_control).guest_chase_thresh, 0, 99) {}
else if VG_BOOL_CLO(arg, "--vex-guest-chase-cond",
VG_(clo_vex_control).guest_chase_cond) {}
else if VG_INT_CLO(arg, "--log-fd", tmp_log_fd) {
log_to = VgLogTo_Fd;
log_fsname_unexpanded = NULL;
}
else if VG_INT_CLO(arg, "--xml-fd", tmp_xml_fd) {
xml_to = VgLogTo_Fd;
xml_fsname_unexpanded = NULL;
}
else if VG_STR_CLO(arg, "--log-file", log_fsname_unexpanded) {
log_to = VgLogTo_File;
}
else if VG_STR_CLO(arg, "--xml-file", xml_fsname_unexpanded) {
xml_to = VgLogTo_File;
}
else if VG_STR_CLO(arg, "--log-socket", log_fsname_unexpanded) {
log_to = VgLogTo_Socket;
}
else if VG_STR_CLO(arg, "--xml-socket", xml_fsname_unexpanded) {
xml_to = VgLogTo_Socket;
}
else if VG_STR_CLO(arg, "--debuginfo-server",
VG_(clo_debuginfo_server)) {}
else if VG_BOOL_CLO(arg, "--allow-mismatched-debuginfo",
VG_(clo_allow_mismatched_debuginfo)) {}
else if VG_STR_CLO(arg, "--xml-user-comment",
VG_(clo_xml_user_comment)) {}
else if VG_BOOL_CLO(arg, "--default-suppressions",
VG_(clo_default_supp)) {}
else if VG_STR_CLO(arg, "--suppressions", tmp_str) {
VG_(addToXA)(VG_(clo_suppressions), &tmp_str);
}
else if VG_STR_CLO (arg, "--fullpath-after", tmp_str) {
VG_(addToXA)(VG_(clo_fullpath_after), &tmp_str);
}
else if VG_STR_CLO (arg, "--extra-debuginfo-path",
VG_(clo_extra_debuginfo_path)) {}
else if VG_STR_CLO(arg, "--require-text-symbol", tmp_str) {
/* String needs to be of the form C?*C?*, where C is any
character, but is the same both times. Having it in this
form facilitates finding the boundary between the sopatt
and the fnpatt just by looking for the second occurrence
of C, without hardwiring any assumption about what C
is. */
HChar patt[7];
Bool ok = True;
ok = tmp_str && VG_(strlen)(tmp_str) > 0;
if (ok) {
patt[0] = patt[3] = tmp_str[0];
patt[1] = patt[4] = '?';
patt[2] = patt[5] = '*';
patt[6] = 0;
ok = VG_(string_match)(patt, tmp_str);
}
if (!ok) {
VG_(fmsg_bad_option)(arg,
"Invalid --require-text-symbol= specification.\n");
}
VG_(addToXA)(VG_(clo_req_tsyms), &tmp_str);
}
/* "stuvwxyz" --> stuvwxyz (binary) */
else if VG_STR_CLO(arg, "--trace-flags", tmp_str) {
Int j;
if (8 != VG_(strlen)(tmp_str)) {
VG_(fmsg_bad_option)(arg,
"--trace-flags argument must have 8 digits\n");
}
for (j = 0; j < 8; j++) {
if ('0' == tmp_str[j]) { /* do nothing */ }
else if ('1' == tmp_str[j]) VG_(clo_trace_flags) |= (1 << (7-j));
else {
VG_(fmsg_bad_option)(arg,
"--trace-flags argument can only contain 0s and 1s\n");
}
}
}
else if VG_INT_CLO (arg, "--trace-notbelow", VG_(clo_trace_notbelow)) {}
else if VG_INT_CLO (arg, "--trace-notabove", VG_(clo_trace_notabove)) {}
/* "stuvwxyz" --> stuvwxyz (binary) */
else if VG_STR_CLO(arg, "--profile-flags", tmp_str) {
Int j;
if (8 != VG_(strlen)(tmp_str)) {
VG_(fmsg_bad_option)(arg,
"--profile-flags argument must have 8 digits\n");
}
for (j = 0; j < 8; j++) {
if ('0' == tmp_str[j]) { /* do nothing */ }
else if ('1' == tmp_str[j]) VG_(clo_profyle_flags) |= (1 << (7-j));
else {
VG_(fmsg_bad_option)(arg,
"--profile-flags argument can only contain 0s and 1s\n");
}
}
VG_(clo_profyle_sbs) = True;
}
else if VG_INT_CLO (arg, "--profile-interval",
VG_(clo_profyle_interval)) {}
else if VG_XACT_CLO(arg, "--gen-suppressions=no",
VG_(clo_gen_suppressions), 0) {}
else if VG_XACT_CLO(arg, "--gen-suppressions=yes",
VG_(clo_gen_suppressions), 1) {}
else if VG_XACT_CLO(arg, "--gen-suppressions=all",
VG_(clo_gen_suppressions), 2) {}
else if VG_BINT_CLO(arg, "--unw-stack-scan-thresh",
VG_(clo_unw_stack_scan_thresh), 0, 100) {}
else if VG_BINT_CLO(arg, "--unw-stack-scan-frames",
VG_(clo_unw_stack_scan_frames), 0, 32) {}
else if VG_XACT_CLO(arg, "--resync-filter=no",
VG_(clo_resync_filter), 0) {}
else if VG_XACT_CLO(arg, "--resync-filter=yes",
VG_(clo_resync_filter), 1) {}
else if VG_XACT_CLO(arg, "--resync-filter=verbose",
VG_(clo_resync_filter), 2) {}
else if ( ! VG_(needs).command_line_options
|| ! VG_TDICT_CALL(tool_process_cmd_line_option, arg) ) {
VG_(fmsg_unknown_option)(arg);
}
}
/* END command-line processing loop */
/* Notify about deprecated features here. */
/* Determine the path prefix for vgdb */
if (VG_(clo_vgdb_prefix) == NULL)
VG_(clo_vgdb_prefix) = VG_(vgdb_prefix_default)();
/* Make VEX control parameters sane */
if (VG_(clo_vex_control).guest_chase_thresh
>= VG_(clo_vex_control).guest_max_insns)
VG_(clo_vex_control).guest_chase_thresh
= VG_(clo_vex_control).guest_max_insns - 1;
if (VG_(clo_vex_control).guest_chase_thresh < 0)
VG_(clo_vex_control).guest_chase_thresh = 0;
/* Check various option values */
if (VG_(clo_verbosity) < 0)
VG_(clo_verbosity) = 0;
if (!sigill_diag_set)
VG_(clo_sigill_diag) = (VG_(clo_verbosity) > 0);
if (VG_(clo_trace_notbelow) == -1) {
if (VG_(clo_trace_notabove) == -1) {
/* [] */
VG_(clo_trace_notbelow) = 2147483647;
VG_(clo_trace_notabove) = 0;
} else {
/* [0 .. notabove] */
VG_(clo_trace_notbelow) = 0;
}
} else {
if (VG_(clo_trace_notabove) == -1) {
/* [notbelow .. ] */
VG_(clo_trace_notabove) = 2147483647;
} else {
/* [notbelow .. notabove] */
}
}
VG_(dyn_vgdb_error) = VG_(clo_vgdb_error);
if (VG_(clo_gen_suppressions) > 0 &&
!VG_(needs).core_errors && !VG_(needs).tool_errors) {
VG_(fmsg_bad_option)("--gen-suppressions=yes",
"Can't use --gen-suppressions= with %s\n"
"because it doesn't generate errors.\n", VG_(details).name);
}
# if !defined(VGO_darwin)
if (VG_(clo_resync_filter) != 0) {
VG_(fmsg_bad_option)("--resync-filter=yes or =verbose",
"--resync-filter= is only available on MacOS X.\n");
/*NOTREACHED*/
}
# endif
/* If XML output is requested, check that the tool actually
supports it. */
if (VG_(clo_xml) && !VG_(needs).xml_output) {
VG_(clo_xml) = False;
VG_(fmsg_bad_option)("--xml=yes",
"%s does not support XML output.\n", VG_(details).name);
/*NOTREACHED*/
}
vg_assert( VG_(clo_gen_suppressions) >= 0 );
vg_assert( VG_(clo_gen_suppressions) <= 2 );
/* If we've been asked to emit XML, mash around various other
options so as to constrain the output somewhat, and to remove
any need for user input during the run.
*/
if (VG_(clo_xml)) {
/* We can't allow --gen-suppressions=yes, since that requires us
to print the error and then ask the user if she wants a
suppression for it, but in XML mode we won't print it until
we know whether we also need to print a suppression. Hence a
circular dependency. So disallow this.
(--gen-suppressions=all is still OK since we don't need any
user interaction in this case.) */
if (VG_(clo_gen_suppressions) == 1) {
VG_(fmsg_bad_option)(
"--xml=yes together with --gen-suppressions=yes",
"When --xml=yes is specified, --gen-suppressions=no\n"
"or --gen-suppressions=all is allowed, but not "
"--gen-suppressions=yes.\n");
}
/* Disallow dump_error in XML mode; sounds like a recipe for
chaos. No big deal; dump_error is a flag for debugging V
itself. */
if (VG_(clo_dump_error) > 0) {
VG_(fmsg_bad_option)("--xml=yes",
"Cannot be used together with --dump-error");
}
/* Disable error limits (this might be a bad idea!) */
VG_(clo_error_limit) = False;
/* Disable emulation warnings */
/* Also, we want to set options for the leak checker, but that
will have to be done in Memcheck's flag-handling code, not
here. */
}
/* All non-logging-related options have been checked. If the logging
option specified is ok, we can switch to it, as we know we won't
have to generate any other command-line-related error messages.
(So far we should be still attached to stderr, so we can show on
the terminal any problems to do with processing command line
opts.)
So set up logging now. After this is done, VG_(log_output_sink)
and (if relevant) VG_(xml_output_sink) should be connected to
whatever sink has been selected, and we indiscriminately chuck
stuff into it without worrying what the nature of it is. Oh the
wonder of Unix streams. */
vg_assert(VG_(log_output_sink).fd == 2 /* stderr */);
vg_assert(VG_(log_output_sink).is_socket == False);
vg_assert(VG_(clo_log_fname_expanded) == NULL);
vg_assert(VG_(xml_output_sink).fd == -1 /* disabled */);
vg_assert(VG_(xml_output_sink).is_socket == False);
vg_assert(VG_(clo_xml_fname_expanded) == NULL);
/* --- set up the normal text output channel --- */
switch (log_to) {
case VgLogTo_Fd:
vg_assert(log_fsname_unexpanded == NULL);
break;
case VgLogTo_File: {
HChar* logfilename;
vg_assert(log_fsname_unexpanded != NULL);
vg_assert(VG_(strlen)(log_fsname_unexpanded) <= 900); /* paranoia */
// Nb: we overwrite an existing file of this name without asking
// any questions.
logfilename = VG_(expand_file_name)("--log-file",
log_fsname_unexpanded);
sres = VG_(open)(logfilename,
VKI_O_CREAT|VKI_O_WRONLY|VKI_O_TRUNC,
VKI_S_IRUSR|VKI_S_IWUSR|VKI_S_IRGRP|VKI_S_IROTH);
if (!sr_isError(sres)) {
tmp_log_fd = sr_Res(sres);
VG_(clo_log_fname_expanded) = logfilename;
} else {
VG_(fmsg)("can't create log file '%s': %s\n",
logfilename, VG_(strerror)(sr_Err(sres)));
VG_(exit)(1);
/*NOTREACHED*/
}
break;
}
case VgLogTo_Socket: {
vg_assert(log_fsname_unexpanded != NULL);
vg_assert(VG_(strlen)(log_fsname_unexpanded) <= 900); /* paranoia */
tmp_log_fd = VG_(connect_via_socket)( log_fsname_unexpanded );
if (tmp_log_fd == -1) {
VG_(fmsg)("Invalid --log-socket spec of '%s'\n",
log_fsname_unexpanded);
VG_(exit)(1);
/*NOTREACHED*/
}
if (tmp_log_fd == -2) {
VG_(umsg)("failed to connect to logging server '%s'.\n"
"Log messages will sent to stderr instead.\n",
log_fsname_unexpanded );
/* We don't change anything here. */
vg_assert(VG_(log_output_sink).fd == 2);
tmp_log_fd = 2;
} else {
vg_assert(tmp_log_fd > 0);
VG_(log_output_sink).is_socket = True;
}
break;
}
}
/* --- set up the XML output channel --- */
switch (xml_to) {
case VgLogTo_Fd:
vg_assert(xml_fsname_unexpanded == NULL);
break;
case VgLogTo_File: {
HChar* xmlfilename;
vg_assert(xml_fsname_unexpanded != NULL);
vg_assert(VG_(strlen)(xml_fsname_unexpanded) <= 900); /* paranoia */
// Nb: we overwrite an existing file of this name without asking
// any questions.
xmlfilename = VG_(expand_file_name)("--xml-file",
xml_fsname_unexpanded);
sres = VG_(open)(xmlfilename,
VKI_O_CREAT|VKI_O_WRONLY|VKI_O_TRUNC,
VKI_S_IRUSR|VKI_S_IWUSR|VKI_S_IRGRP|VKI_S_IROTH);
if (!sr_isError(sres)) {
tmp_xml_fd = sr_Res(sres);
VG_(clo_xml_fname_expanded) = xmlfilename;
*xml_fname_unexpanded = xml_fsname_unexpanded;
} else {
VG_(fmsg)("can't create XML file '%s': %s\n",
xmlfilename, VG_(strerror)(sr_Err(sres)));
VG_(exit)(1);
/*NOTREACHED*/
}
break;
}
case VgLogTo_Socket: {
vg_assert(xml_fsname_unexpanded != NULL);
vg_assert(VG_(strlen)(xml_fsname_unexpanded) <= 900); /* paranoia */
tmp_xml_fd = VG_(connect_via_socket)( xml_fsname_unexpanded );
if (tmp_xml_fd == -1) {
VG_(fmsg)("Invalid --xml-socket spec of '%s'\n",
xml_fsname_unexpanded );
VG_(exit)(1);
/*NOTREACHED*/
}
if (tmp_xml_fd == -2) {
VG_(umsg)("failed to connect to XML logging server '%s'.\n"
"XML output will sent to stderr instead.\n",
xml_fsname_unexpanded);
/* We don't change anything here. */
vg_assert(VG_(xml_output_sink).fd == 2);
tmp_xml_fd = 2;
} else {
vg_assert(tmp_xml_fd > 0);
VG_(xml_output_sink).is_socket = True;
}
break;
}
}
/* If we've got this far, and XML mode was requested, but no XML
output channel appears to have been specified, just stop. We
could continue, and XML output will simply vanish into nowhere,
but that is likely to confuse the hell out of users, which is
distinctly Ungood. */
if (VG_(clo_xml) && tmp_xml_fd == -1) {
VG_(fmsg_bad_option)(
"--xml=yes, but no XML destination specified",
"--xml=yes has been specified, but there is no XML output\n"
"destination. You must specify an XML output destination\n"
"using --xml-fd, --xml-file or --xml-socket.\n"
);
}
// Finalise the output fds: the log fd ..
if (tmp_log_fd >= 0) {
// Move log_fd into the safe range, so it doesn't conflict with
// any app fds.
tmp_log_fd = VG_(fcntl)(tmp_log_fd, VKI_F_DUPFD, VG_(fd_hard_limit));
if (tmp_log_fd < 0) {
VG_(message)(Vg_UserMsg, "valgrind: failed to move logfile fd "
"into safe range, using stderr\n");
VG_(log_output_sink).fd = 2; // stderr
VG_(log_output_sink).is_socket = False;
} else {
VG_(log_output_sink).fd = tmp_log_fd;
VG_(fcntl)(VG_(log_output_sink).fd, VKI_F_SETFD, VKI_FD_CLOEXEC);
}
} else {
// If they said --log-fd=-1, don't print anything. Plausible for use in
// regression testing suites that use client requests to count errors.
VG_(log_output_sink).fd = -1;
VG_(log_output_sink).is_socket = False;
}
// Finalise the output fds: and the XML fd ..
if (tmp_xml_fd >= 0) {
// Move xml_fd into the safe range, so it doesn't conflict with
// any app fds.
tmp_xml_fd = VG_(fcntl)(tmp_xml_fd, VKI_F_DUPFD, VG_(fd_hard_limit));
if (tmp_xml_fd < 0) {
VG_(message)(Vg_UserMsg, "valgrind: failed to move XML file fd "
"into safe range, using stderr\n");
VG_(xml_output_sink).fd = 2; // stderr
VG_(xml_output_sink).is_socket = False;
} else {
VG_(xml_output_sink).fd = tmp_xml_fd;
VG_(fcntl)(VG_(xml_output_sink).fd, VKI_F_SETFD, VKI_FD_CLOEXEC);
}
} else {
// If they said --xml-fd=-1, don't print anything. Plausible for use in
// regression testing suites that use client requests to count errors.
VG_(xml_output_sink).fd = -1;
VG_(xml_output_sink).is_socket = False;
}
// Suppressions related stuff
if (VG_(clo_default_supp) &&
(VG_(needs).core_errors || VG_(needs).tool_errors)) {
/* If we haven't reached the max number of suppressions, load
the default one. */
static const HChar default_supp[] = "default.supp";
Int len = VG_(strlen)(VG_(libdir)) + 1 + sizeof(default_supp);
HChar *buf = VG_(malloc)("main.mpclo.3", len);
VG_(sprintf)(buf, "%s/%s", VG_(libdir), default_supp);
VG_(addToXA)(VG_(clo_suppressions), &buf);
}
*logging_to_fd = log_to == VgLogTo_Fd || log_to == VgLogTo_Socket;
}
// Write the name and value of log file qualifiers to the xml file.
// We can safely assume here that the format string is well-formed.
// It has been checked earlier in VG_(expand_file_name) when processing
// command line options.
static void print_file_vars(const HChar* format)
{
Int i = 0;
while (format[i]) {
if (format[i] == '%') {
// We saw a '%'. What's next...
i++;
if ('q' == format[i]) {
i++;
if ('{' == format[i]) {
// Get the env var name, print its contents.
HChar* qual;
Int begin_qualname = ++i;
while (True) {
if ('}' == format[i]) {
Int qualname_len = i - begin_qualname;
HChar qualname[qualname_len + 1];
VG_(strncpy)(qualname, format + begin_qualname,
qualname_len);
qualname[qualname_len] = '\0';
qual = VG_(getenv)(qualname);
i++;
VG_(printf_xml)("<logfilequalifier> <var>%pS</var> "
"<value>%pS</value> </logfilequalifier>\n",
qualname, qual);
break;
}
i++;
}
}
}
} else {
i++;
}
}
}
/*====================================================================*/
/*=== Printing the preamble ===*/
/*====================================================================*/
// Print the argument, escaping any chars that require it.
static void umsg_arg(const HChar* arg)
{
SizeT len = VG_(strlen)(arg);
const HChar* special = " \\<>";
Int i;
for (i = 0; i < len; i++) {
if (VG_(strchr)(special, arg[i])) {
VG_(umsg)("\\"); // escape with a backslash if necessary
}
VG_(umsg)("%c", arg[i]);
}
}
// Send output to the XML-stream and escape any XML meta-characters.
static void xml_arg(const HChar* arg)
{
VG_(printf_xml)("%pS", arg);
}
/* Ok, the logging sink is running now. Print a suitable preamble.
If logging to file or a socket, write details of parent PID and
command line args, to help people trying to interpret the
results of a run which encompasses multiple processes. */
static void print_preamble ( Bool logging_to_fd,
const HChar* xml_fname_unexpanded,
const HChar* toolname )
{
Int i;
const HChar* xpre = VG_(clo_xml) ? " <line>" : "";
const HChar* xpost = VG_(clo_xml) ? "</line>" : "";
UInt (*umsg_or_xml)( const HChar*, ... )
= VG_(clo_xml) ? VG_(printf_xml) : VG_(umsg);
void (*umsg_or_xml_arg)( const HChar* )
= VG_(clo_xml) ? xml_arg : umsg_arg;
vg_assert( VG_(args_for_client) );
vg_assert( VG_(args_for_valgrind) );
vg_assert( toolname );
if (VG_(clo_xml)) {
VG_(printf_xml)("<?xml version=\"1.0\"?>\n");
VG_(printf_xml)("\n");
VG_(printf_xml)("<valgrindoutput>\n");
VG_(printf_xml)("\n");
VG_(printf_xml)("<protocolversion>4</protocolversion>\n");
VG_(printf_xml)("<protocoltool>%s</protocoltool>\n", toolname);
VG_(printf_xml)("\n");
}
if (VG_(clo_xml) || VG_(clo_verbosity > 0)) {
if (VG_(clo_xml))
VG_(printf_xml)("<preamble>\n");
/* Tool details */
umsg_or_xml( VG_(clo_xml) ? "%s%pS%pS%pS, %pS%s\n" : "%s%s%s%s, %s%s\n",
xpre,
VG_(details).name,
NULL == VG_(details).version ? "" : "-",
NULL == VG_(details).version
? "" : VG_(details).version,
VG_(details).description,
xpost );
if (VG_(strlen)(toolname) >= 4 && VG_STREQN(4, toolname, "exp-")) {
umsg_or_xml(
"%sNOTE: This is an Experimental-Class Valgrind Tool%s\n",
xpre, xpost
);
}
umsg_or_xml( VG_(clo_xml) ? "%s%pS%s\n" : "%s%s%s\n",
xpre, VG_(details).copyright_author, xpost );
/* Core details */
umsg_or_xml(
"%sUsing Valgrind-%s and LibVEX; rerun with -h for copyright info%s\n",
xpre, VERSION, xpost
);
// Print the command line. At one point we wrapped at 80 chars and
// printed a '\' as a line joiner, but that makes it hard to cut and
// paste the command line (because of the "==pid==" prefixes), so we now
// favour utility and simplicity over aesthetics.
umsg_or_xml("%sCommand: ", xpre);
umsg_or_xml_arg(VG_(args_the_exename));
for (i = 0; i < VG_(sizeXA)( VG_(args_for_client) ); i++) {
HChar* s = *(HChar**)VG_(indexXA)( VG_(args_for_client), i );
umsg_or_xml(" ");
umsg_or_xml_arg(s);
}
umsg_or_xml("%s\n", xpost);
if (VG_(clo_xml))
VG_(printf_xml)("</preamble>\n");
}
// Print the parent PID, and other stuff, if necessary.
if (!VG_(clo_xml) && VG_(clo_verbosity) > 0 && !logging_to_fd) {
VG_(umsg)("Parent PID: %d\n", VG_(getppid)());
}
else
if (VG_(clo_xml)) {
VG_(printf_xml)("\n");
VG_(printf_xml)("<pid>%d</pid>\n", VG_(getpid)());
VG_(printf_xml)("<ppid>%d</ppid>\n", VG_(getppid)());
VG_(printf_xml)("<tool>%pS</tool>\n", toolname);
if (xml_fname_unexpanded)
print_file_vars(xml_fname_unexpanded);
if (VG_(clo_xml_user_comment)) {
/* Note: the user comment itself is XML and is therefore to
be passed through verbatim (%s) rather than escaped
(%pS). */
VG_(printf_xml)("<usercomment>%s</usercomment>\n",
VG_(clo_xml_user_comment));
}
VG_(printf_xml)("\n");
VG_(printf_xml)("<args>\n");
VG_(printf_xml)(" <vargv>\n");
if (VG_(name_of_launcher))
VG_(printf_xml)(" <exe>%pS</exe>\n",
VG_(name_of_launcher));
else
VG_(printf_xml)(" <exe>%pS</exe>\n",
"(launcher name unknown)");
for (i = 0; i < VG_(sizeXA)( VG_(args_for_valgrind) ); i++) {
VG_(printf_xml)(
" <arg>%pS</arg>\n",
* (HChar**) VG_(indexXA)( VG_(args_for_valgrind), i )
);
}
VG_(printf_xml)(" </vargv>\n");
VG_(printf_xml)(" <argv>\n");
VG_(printf_xml)(" <exe>%pS</exe>\n",
VG_(args_the_exename));
for (i = 0; i < VG_(sizeXA)( VG_(args_for_client) ); i++) {
VG_(printf_xml)(
" <arg>%pS</arg>\n",
* (HChar**) VG_(indexXA)( VG_(args_for_client), i )
);
}
VG_(printf_xml)(" </argv>\n");
VG_(printf_xml)("</args>\n");
}
// Last thing in the preamble is a blank line.
if (VG_(clo_xml))
VG_(printf_xml)("\n");
else if (VG_(clo_verbosity) > 0)
VG_(umsg)("\n");
if (VG_(clo_verbosity) > 1) {
# if defined(VGO_linux)
SysRes fd;
# endif
VexArch vex_arch;
VexArchInfo vex_archinfo;
if (!logging_to_fd)
VG_(message)(Vg_DebugMsg, "\n");
VG_(message)(Vg_DebugMsg, "Valgrind options:\n");
for (i = 0; i < VG_(sizeXA)( VG_(args_for_valgrind) ); i++) {
VG_(message)(Vg_DebugMsg,
" %s\n",
* (HChar**) VG_(indexXA)( VG_(args_for_valgrind), i ));
}
# if defined(VGO_linux)
VG_(message)(Vg_DebugMsg, "Contents of /proc/version:\n");
fd = VG_(open) ( "/proc/version", VKI_O_RDONLY, 0 );
if (sr_isError(fd)) {
VG_(message)(Vg_DebugMsg, " can't open /proc/version\n");
} else {
const SizeT bufsiz = 255;
HChar version_buf[bufsiz+1];
VG_(message)(Vg_DebugMsg, " ");
Int n, fdno = sr_Res(fd);
do {
n = VG_(read)(fdno, version_buf, bufsiz);
if (n < 0) {
VG_(message)(Vg_DebugMsg, " error reading /proc/version\n");
break;
}
version_buf[n] = '\0';
VG_(message)(Vg_DebugMsg, "%s", version_buf);
} while (n == bufsiz);
VG_(message)(Vg_DebugMsg, "\n");
VG_(close)(fdno);
}
# elif defined(VGO_darwin)
VG_(message)(Vg_DebugMsg, "Output from sysctl({CTL_KERN,KERN_VERSION}):\n");
/* Note: preferable to use sysctlbyname("kern.version", kernelVersion, &len, NULL, 0)
however that syscall is OS X 10.10+ only. */
Int mib[] = {CTL_KERN, KERN_VERSION};
SizeT len;
VG_(sysctl)(mib, sizeof(mib)/sizeof(Int), NULL, &len, NULL, 0);
HChar *kernelVersion = VG_(malloc)("main.pp.1", len);
VG_(sysctl)(mib, sizeof(mib)/sizeof(Int), kernelVersion, &len, NULL, 0);
VG_(message)(Vg_DebugMsg, " %s\n", kernelVersion);
VG_(free)( kernelVersion );
# elif defined(VGO_solaris)
/* There is no /proc/version file on Solaris so we try to get some
system information using the uname(2) syscall. */
{
struct vki_utsname uts;
VG_(message)(Vg_DebugMsg, "System information:\n");
SysRes res = VG_(do_syscall1)(__NR_uname, (UWord)&uts);
if (sr_isError(res))
VG_(message)(Vg_DebugMsg, " uname() failed\n");
else
VG_(message)(Vg_DebugMsg, " %s %s %s %s\n",
uts.sysname, uts.release, uts.version, uts.machine);
}
# endif
VG_(machine_get_VexArchInfo)( &vex_arch, &vex_archinfo );
VG_(message)(
Vg_DebugMsg,
"Arch and hwcaps: %s, %s, %s\n",
LibVEX_ppVexArch ( vex_arch ),
LibVEX_ppVexEndness ( vex_archinfo.endness ),
LibVEX_ppVexHwCaps ( vex_arch, vex_archinfo.hwcaps )
);
VG_(message)(
Vg_DebugMsg,
"Page sizes: currently %d, max supported %d\n",
(Int)VKI_PAGE_SIZE, (Int)VKI_MAX_PAGE_SIZE
);
VG_(message)(Vg_DebugMsg,
"Valgrind library directory: %s\n", VG_(libdir));
}
}
/*====================================================================*/
/*=== File descriptor setup ===*/
/*====================================================================*/
/* Number of file descriptors that Valgrind tries to reserve for
it's own use - just a small constant. */
#define N_RESERVED_FDS (10)
static void setup_file_descriptors(void)
{
struct vki_rlimit rl;
Bool show = False;
/* Get the current file descriptor limits. */
if (VG_(getrlimit)(VKI_RLIMIT_NOFILE, &rl) < 0) {
rl.rlim_cur = 1024;
rl.rlim_max = 1024;
}
# if defined(VGO_darwin)
/* Darwin lies. It reports file max as RLIM_INFINITY but
silently disallows anything bigger than 10240. */
if (rl.rlim_cur >= 10240 && rl.rlim_max == 0x7fffffffffffffffULL) {
rl.rlim_max = 10240;
}
# endif
if (show)
VG_(printf)("fd limits: host, before: cur %llu max %llu\n",
(ULong)rl.rlim_cur, (ULong)rl.rlim_max);
/* Work out where to move the soft limit to. */
if (rl.rlim_cur + N_RESERVED_FDS <= rl.rlim_max) {
rl.rlim_cur = rl.rlim_cur + N_RESERVED_FDS;
} else {
rl.rlim_cur = rl.rlim_max;
}
/* Reserve some file descriptors for our use. */
VG_(fd_soft_limit) = rl.rlim_cur - N_RESERVED_FDS;
VG_(fd_hard_limit) = rl.rlim_cur - N_RESERVED_FDS;
/* Update the soft limit. */
VG_(setrlimit)(VKI_RLIMIT_NOFILE, &rl);
if (show) {
VG_(printf)("fd limits: host, after: cur %lu max %lu\n",
(UWord)rl.rlim_cur, (UWord)rl.rlim_max);
VG_(printf)("fd limits: guest : cur %d max %d\n",
VG_(fd_soft_limit), VG_(fd_hard_limit));
}
if (VG_(cl_exec_fd) != -1)
VG_(cl_exec_fd) = VG_(safe_fd)( VG_(cl_exec_fd) );
}
/*====================================================================*/
/*=== main() ===*/
/*====================================================================*/
/* When main() is entered, we should be on the following stack, not
the one the kernel gave us. We will run on this stack until
simulation of the root thread is started, at which point a transfer
is made to a dynamically allocated stack. This is for the sake of
uniform overflow detection for all Valgrind threads. This is
marked global even though it isn't, because assembly code below
needs to reference the name. */
/*static*/ struct {
HChar bytes [VG_STACK_GUARD_SZB + VG_DEFAULT_STACK_ACTIVE_SZB + VG_STACK_GUARD_SZB];
} VG_(interim_stack);
/* These are the structures used to hold info for creating the initial
client image.
'iicii' mostly holds important register state present at system
startup (_start_valgrind). valgrind_main() then fills in the rest
of it and passes it to VG_(ii_create_image)(). That produces
'iifii', which is later handed to VG_(ii_finalise_image). */
/* In all OS-instantiations, the_iicii has a field .sp_at_startup.
This should get some address inside the stack on which we gained
control (eg, it could be the SP at startup). It doesn't matter
exactly where in the stack it is. This value is passed to the
address space manager at startup. On Linux, aspacem then uses it
to identify the initial stack segment and hence the upper end of
the usable address space. */
static IICreateImageInfo the_iicii;
static IIFinaliseImageInfo the_iifii;
/* A simple pair structure, used for conveying debuginfo handles to
calls to VG_TRACK(new_mem_startup, ...). */
typedef struct { Addr a; ULong ull; } Addr_n_ULong;
/* --- Forwards decls to do with shutdown --- */
static void final_tidyup(ThreadId tid);
/* Do everything which needs doing when the last thread exits */
static
void shutdown_actions_NORETURN( ThreadId tid,
VgSchedReturnCode tids_schedretcode );
/* --- end of Forwards decls to do with shutdown --- */
/* By the time we get to valgrind_main, the_iicii should already have
been filled in with any important details as required by whatever
OS we have been built for.
*/
static
Int valgrind_main ( Int argc, HChar **argv, HChar **envp )
{
const HChar* toolname = "memcheck"; // default to Memcheck
Int need_help = 0; // 0 = no, 1 = --help, 2 = --help-debug
ThreadId tid_main = VG_INVALID_THREADID;
Bool logging_to_fd = False;
const HChar* xml_fname_unexpanded = NULL;
Int loglevel, i;
struct vki_rlimit zero = { 0, 0 };
XArray* addr2dihandle = NULL;
//============================================================
//
// Nb: startup is complex. Prerequisites are shown at every step.
// *** Be very careful when messing with the order ***
//
// The first order of business is to get debug logging, the address
// space manager and the dynamic memory manager up and running.
// Once that's done, we can relax a bit.
//
//============================================================
/* This is needed to make VG_(getenv) usable early. */
VG_(client_envp) = (HChar**)envp;
//--------------------------------------------------------------
// Start up Mach kernel interface, if any
// p: none
//--------------------------------------------------------------
# if defined(VGO_darwin)
VG_(mach_init)();
# endif
//--------------------------------------------------------------
// Start up the logging mechanism
// p: none
//--------------------------------------------------------------
/* Start the debugging-log system ASAP. First find out how many
"-d"s were specified. This is a pre-scan of the command line. Also
get --profile-heap=yes, --core-redzone-size, --redzone-size
--aspace-minaddr which are needed by the time we start up dynamic
memory management. */
loglevel = 0;
for (i = 1; i < argc; i++) {
const HChar* tmp_str;
if (argv[i][0] != '-') break;
if VG_STREQ(argv[i], "--") break;
if VG_STREQ(argv[i], "-d") loglevel++;
if VG_BOOL_CLO(argv[i], "--profile-heap", VG_(clo_profile_heap)) {}
if VG_BINT_CLO(argv[i], "--core-redzone-size", VG_(clo_core_redzone_size),
0, MAX_CLO_REDZONE_SZB) {}
if VG_BINT_CLO(argv[i], "--redzone-size", VG_(clo_redzone_size),
0, MAX_CLO_REDZONE_SZB) {}
if VG_STR_CLO(argv[i], "--aspace-minaddr", tmp_str) {
Bool ok = VG_(parse_Addr) (&tmp_str, &VG_(clo_aspacem_minAddr));
if (!ok)
VG_(fmsg_bad_option)(argv[i], "Invalid address\n");
const HChar *errmsg;
if (!VG_(am_is_valid_for_aspacem_minAddr)(VG_(clo_aspacem_minAddr),
&errmsg))
VG_(fmsg_bad_option)(argv[i], "%s\n", errmsg);
}
}
/* ... and start the debug logger. Now we can safely emit logging
messages all through startup. */
VG_(debugLog_startup)(loglevel, "Stage 2 (main)");
VG_(debugLog)(1, "main", "Welcome to Valgrind version "
VERSION " debug logging\n");
//--------------------------------------------------------------
// Ensure we're on a plausible stack.
// p: logging
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Checking current stack is plausible\n");
{ HChar* limLo = (HChar*)(&VG_(interim_stack).bytes[0]);
HChar* limHi = limLo + sizeof(VG_(interim_stack));
HChar* volatile
aLocal = (HChar*)&limLo; /* any auto local will do */
/* Re "volatile": Apple clang version 4.0
(tags/Apple/clang-421.0.57) (based on LLVM 3.1svn)" appeared
to miscompile the following check, causing run to abort at
this point (in 64-bit mode) even though aLocal is within limLo
.. limHi. But in fact clang is within its rights to do
strange things here. "The reason is that the comparisons
aLocal < limLo and aLocal >= limHi cause undefined behaviour
(according to c99 6.5.8) because they compare pointers that do
not point into the same aggregate." Adding "volatile" appears
to fix it because "The compiler would have to prove that there
is undefined behavior in order to exploit it. But as a
volatile variable can change its value in ways invisible to
the compiler, the compiler must make the conservative
assumption that it points into the same aggregate as the other
pointer its compared against. I.e. the behaviour is possibly
defined." (Analysis by Florian Krohm). */
if (aLocal < limLo || aLocal >= limHi) {
/* something's wrong. Stop. */
VG_(debugLog)(0, "main", "Root stack %p to %p, a local %p\n",
limLo, limHi, aLocal );
VG_(debugLog)(0, "main", "Valgrind: FATAL: "
"Initial stack switched failed.\n");
VG_(debugLog)(0, "main", " Cannot continue. Sorry.\n");
VG_(exit)(1);
}
}
//--------------------------------------------------------------
// Ensure we have a plausible pointer to the stack on which
// we gained control (not the current stack!)
// p: logging
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Checking initial stack was noted\n");
if (the_iicii.sp_at_startup == 0) {
VG_(debugLog)(0, "main", "Valgrind: FATAL: "
"Initial stack was not noted.\n");
VG_(debugLog)(0, "main", " Cannot continue. Sorry.\n");
VG_(exit)(1);
}
//--------------------------------------------------------------
// Start up the address space manager, and determine the
// approximate location of the client's stack
// p: logging, plausible-stack
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Starting the address space manager\n");
vg_assert(VKI_PAGE_SIZE == 4096 || VKI_PAGE_SIZE == 65536
|| VKI_PAGE_SIZE == 16384);
vg_assert(VKI_MAX_PAGE_SIZE == 4096 || VKI_MAX_PAGE_SIZE == 65536
|| VKI_MAX_PAGE_SIZE == 16384);
vg_assert(VKI_PAGE_SIZE <= VKI_MAX_PAGE_SIZE);
vg_assert(VKI_PAGE_SIZE == (1 << VKI_PAGE_SHIFT));
vg_assert(VKI_MAX_PAGE_SIZE == (1 << VKI_MAX_PAGE_SHIFT));
the_iicii.clstack_end = VG_(am_startup)( the_iicii.sp_at_startup );
VG_(debugLog)(1, "main", "Address space manager is running\n");
//--------------------------------------------------------------
// Start up the dynamic memory manager
// p: address space management
// p: getting --profile-heap,--core-redzone-size,--redzone-size
// In fact m_mallocfree is self-initialising, so there's no
// initialisation call to do. Instead, try a simple malloc/
// free pair right now to check that nothing is broken.
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Starting the dynamic memory manager\n");
{ void* p = VG_(malloc)( "main.vm.1", 12345 );
VG_(free)( p );
}
VG_(debugLog)(1, "main", "Dynamic memory manager is running\n");
//============================================================
//
// Dynamic memory management is now available.
//
//============================================================
//--------------------------------------------------------------
// Initialise m_debuginfo
// p: dynamic memory allocation
VG_(debugLog)(1, "main", "Initialise m_debuginfo\n");
VG_(di_initialise)();
//--------------------------------------------------------------
// Look for alternative libdir
{ HChar *cp = VG_(getenv)(VALGRIND_LIB);
if (cp != NULL)
VG_(libdir) = cp;
VG_(debugLog)(1, "main", "VG_(libdir) = %s\n", VG_(libdir));
}
//--------------------------------------------------------------
// Extract the launcher name from the environment.
VG_(debugLog)(1, "main", "Getting launcher's name ...\n");
VG_(name_of_launcher) = VG_(getenv)(VALGRIND_LAUNCHER);
if (VG_(name_of_launcher) == NULL) {
VG_(printf)("valgrind: You cannot run '%s' directly.\n", argv[0]);
VG_(printf)("valgrind: You should use $prefix/bin/valgrind.\n");
VG_(exit)(1);
}
VG_(debugLog)(1, "main", "... %s\n", VG_(name_of_launcher));
//--------------------------------------------------------------
// Get the current process datasize rlimit, and set it to zero.
// This prevents any internal uses of brk() from having any effect.
// We remember the old value so we can restore it on exec, so that
// child processes will have a reasonable brk value.
VG_(getrlimit)(VKI_RLIMIT_DATA, &VG_(client_rlimit_data));
zero.rlim_max = VG_(client_rlimit_data).rlim_max;
VG_(setrlimit)(VKI_RLIMIT_DATA, &zero);
// Get the current process stack rlimit.
VG_(getrlimit)(VKI_RLIMIT_STACK, &VG_(client_rlimit_stack));
//--------------------------------------------------------------
// Figure out what sort of CPU we're on, and whether it is
// able to run V.
/* The vex_archinfo structure is passed down later to the client
* to verify the HW info settings are consistent.
*/
VexArchInfo vex_archinfo;
VG_(debugLog)(1, "main", "Get hardware capabilities ...\n");
{ VexArch vex_arch;
Bool ok = VG_(machine_get_hwcaps)();
if (!ok) {
VG_(printf)("\n");
VG_(printf)("valgrind: fatal error: unsupported CPU.\n");
VG_(printf)(" Supported CPUs are:\n");
VG_(printf)(" * x86 (practically any; Pentium-I or above), "
"AMD Athlon or above)\n");
VG_(printf)(" * AMD Athlon64/Opteron\n");
VG_(printf)(" * ARM (armv7)\n");
VG_(printf)(" * PowerPC (most; ppc405 and above)\n");
VG_(printf)(" * System z (64bit only - s390x; z990 and above)\n");
VG_(printf)("\n");
VG_(exit)(1);
}
VG_(machine_get_VexArchInfo)( &vex_arch, &vex_archinfo );
VG_(debugLog)(
1, "main", "... arch = %s, hwcaps = %s\n",
LibVEX_ppVexArch ( vex_arch ),
LibVEX_ppVexHwCaps ( vex_arch, vex_archinfo.hwcaps )
);
}
//--------------------------------------------------------------
// Record the working directory at startup
// p: none
VG_(debugLog)(1, "main", "Getting the working directory at startup\n");
{ Bool ok = VG_(record_startup_wd)();
if (!ok)
VG_(err_config_error)( "Can't establish current working "
"directory at startup\n");
}
VG_(debugLog)(1, "main", "... %s\n", VG_(get_startup_wd)() );
//============================================================
// Command line argument handling order:
// * If --help/--help-debug are present, show usage message
// (including the tool-specific usage)
// * (If no --tool option given, default to Memcheck)
// * Then, if client is missing, abort with error msg
// * Then, if any cmdline args are bad, abort with error msg
//============================================================
//--------------------------------------------------------------
// Split up argv into: C args, V args, V extra args, and exename.
// p: dynamic memory allocation
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Split up command line\n");
VG_(split_up_argv)( argc, argv );
vg_assert( VG_(args_for_valgrind) );
vg_assert( VG_(args_for_client) );
if (0) {
for (i = 0; i < VG_(sizeXA)( VG_(args_for_valgrind) ); i++)
VG_(printf)(
"varg %s\n",
* (HChar**) VG_(indexXA)( VG_(args_for_valgrind), i )
);
VG_(printf)(" exe %s\n", VG_(args_the_exename));
for (i = 0; i < VG_(sizeXA)( VG_(args_for_client) ); i++)
VG_(printf)(
"carg %s\n",
* (HChar**) VG_(indexXA)( VG_(args_for_client), i )
);
}
//--------------------------------------------------------------
// Extract tool name and whether help has been requested.
// Note we can't print the help message yet, even if requested,
// because the tool has not been initialised.
// p: split_up_argv [for VG_(args_for_valgrind)]
//--------------------------------------------------------------
VG_(debugLog)(1, "main",
"(early_) Process Valgrind's command line options\n");
early_process_cmd_line_options(&need_help, &toolname);
// BEGIN HACK
vg_assert(toolname != NULL);
vg_assert(VG_(clo_read_inline_info) == False);
# if !defined(VGO_darwin)
if (0 == VG_(strcmp)(toolname, "memcheck")
|| 0 == VG_(strcmp)(toolname, "helgrind")
|| 0 == VG_(strcmp)(toolname, "drd")) {
/* Change the default setting. Later on (just below)
main_process_cmd_line_options should pick up any
user-supplied setting for it and will override the default
set here. */
VG_(clo_read_inline_info) = True;
}
# endif
// END HACK
// Set default vex control params.
LibVEX_default_VexControl(& VG_(clo_vex_control));
//--------------------------------------------------------------
// Load client executable, finding in $PATH if necessary
// p: early_process_cmd_line_options() [for 'exec', 'need_help',
// clo_max_stackframe,
// clo_main_stacksize]
// p: layout_remaining_space [so there's space]
//
// Set up client's environment
// p: set-libdir [for VG_(libdir)]
// p: early_process_cmd_line_options [for toolname]
//
// Setup client stack, eip, and VG_(client_arg[cv])
// p: load_client() [for 'info']
// p: fix_environment() [for 'env']
//
// Setup client data (brk) segment. Initially a 1-page segment
// which abuts a shrinkable reservation.
// p: load_client() [for 'info' and hence VG_(brk_base)]
//
// p: _start_in_C (for zeroing out the_iicii and putting some
// initial values into it)
//--------------------------------------------------------------
if (!need_help) {
VG_(debugLog)(1, "main", "Create initial image\n");
# if defined(VGO_linux) || defined(VGO_darwin) || defined(VGO_solaris)
the_iicii.argv = argv;
the_iicii.envp = envp;
the_iicii.toolname = toolname;
# else
# error "Unknown platform"
# endif
/* NOTE: this call reads VG_(clo_main_stacksize). */
the_iifii = VG_(ii_create_image)( the_iicii, &vex_archinfo );
}
//==============================================================
//
// Finished loading/setting up the client address space.
//
//==============================================================
//--------------------------------------------------------------
// setup file descriptors
// p: n/a
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Setup file descriptors\n");
setup_file_descriptors();
//--------------------------------------------------------------
// create fake /proc/<pid>/cmdline and /proc/<pid>/auxv files
// and then unlink them, but hold onto the fds, so we can handr
// them out to the client when it tries to open
// /proc/<pid>/cmdline or /proc/<pid>/auxv for itself.
// p: setup file descriptors
// p: ii_create_image for VG_(client_auxv) setup.
//--------------------------------------------------------------
VG_(cl_cmdline_fd) = -1;
VG_(cl_auxv_fd) = -1;
#if defined(VGO_solaris)
VG_(cl_psinfo_fd) = -1;
#endif
#if defined(VGO_linux) || defined(VGO_solaris)
if (!need_help) {
HChar buf[50]; // large enough
HChar buf2[VG_(mkstemp_fullname_bufsz)(sizeof buf - 1)];
Int fd, r;
#if defined(VGO_linux) || defined(SOLARIS_PROC_CMDLINE)
/* Fake /proc/<pid>/cmdline only on Linux and Solaris if supported. */
HChar nul[1];
const HChar* exename;
VG_(debugLog)(1, "main", "Create fake /proc/<pid>/cmdline\n");
VG_(sprintf)(buf, "proc_%d_cmdline", VG_(getpid)());
fd = VG_(mkstemp)( buf, buf2 );
if (fd == -1)
VG_(err_config_error)("Can't create client cmdline file in %s\n", buf2);
nul[0] = 0;
exename = VG_(args_the_exename);
VG_(write)(fd, exename, VG_(strlen)( exename ));
VG_(write)(fd, nul, 1);
for (i = 0; i < VG_(sizeXA)( VG_(args_for_client) ); i++) {
HChar* arg = * (HChar**) VG_(indexXA)( VG_(args_for_client), i );
VG_(write)(fd, arg, VG_(strlen)( arg ));
VG_(write)(fd, nul, 1);
}
/* Don't bother to seek the file back to the start; instead do
it every time a copy of it is given out (by PRE(sys_open) or
PRE(sys_openat)). That is probably more robust across fork() etc. */
/* Now delete it, but hang on to the fd. */
r = VG_(unlink)( buf2 );
if (r)
VG_(err_config_error)("Can't delete client cmdline file in %s\n", buf2);
VG_(cl_cmdline_fd) = fd;
#endif // defined(VGO_linux) || defined(SOLARIS_PROC_CMDLINE)
/* Fake /proc/<pid>/auxv on both Linux and Solaris. */
VG_(debugLog)(1, "main", "Create fake /proc/<pid>/auxv\n");
VG_(sprintf)(buf, "proc_%d_auxv", VG_(getpid)());
fd = VG_(mkstemp)( buf, buf2 );
if (fd == -1)
VG_(err_config_error)("Can't create client auxv file in %s\n", buf2);
UWord *client_auxv = VG_(client_auxv);
unsigned int client_auxv_len = 0;
while (*client_auxv != 0) {
client_auxv++;
client_auxv++;
client_auxv_len += 2 * sizeof(UWord);
}
client_auxv_len += 2 * sizeof(UWord);
VG_(write)(fd, VG_(client_auxv), client_auxv_len);
/* Don't bother to seek the file back to the start; instead do
it every time a copy of it is given out (by PRE(sys_open)).
That is probably more robust across fork() etc. */
/* Now delete it, but hang on to the fd. */
r = VG_(unlink)( buf2 );
if (r)
VG_(err_config_error)("Can't delete client auxv file in %s\n", buf2);
VG_(cl_auxv_fd) = fd;
#if defined(VGO_solaris)
/* Fake /proc/<pid>/psinfo on Solaris.
* Contents will be fetched and partially faked later on the fly. */
VG_(debugLog)(1, "main", "Create fake /proc/<pid>/psinfo\n");
VG_(sprintf)(buf, "proc_%d_psinfo", VG_(getpid)());
fd = VG_(mkstemp)( buf, buf2 );
if (fd == -1)
VG_(err_config_error)("Can't create client psinfo file in %s\n", buf2);
/* Now delete it, but hang on to the fd. */
r = VG_(unlink)( buf2 );
if (r)
VG_(err_config_error)("Can't delete client psinfo file in %s\n", buf2);
VG_(cl_psinfo_fd) = fd;
#endif /* VGO_solaris */
}
#endif
//--------------------------------------------------------------
// Init tool part 1: pre_clo_init
// p: setup_client_stack() [for 'VG_(client_arg[cv]']
// p: setup_file_descriptors() [for 'VG_(fd_xxx_limit)']
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Initialise the tool part 1 (pre_clo_init)\n");
VG_(tl_pre_clo_init)();
// Activate var info readers, if the tool asked for it:
if (VG_(needs).var_info)
VG_(clo_read_var_info) = True;
//--------------------------------------------------------------
// If --tool and --help/--help-debug was given, now give the core+tool
// help message
// p: early_process_cmd_line_options() [for 'need_help']
// p: tl_pre_clo_init [for 'VG_(tdict).usage']
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Print help and quit, if requested\n");
if (need_help) {
usage_NORETURN(/*--help-debug?*/need_help >= 2);
}
//--------------------------------------------------------------
// Process command line options to Valgrind + tool
// p: setup_client_stack() [for 'VG_(client_arg[cv]']
// p: setup_file_descriptors() [for 'VG_(fd_xxx_limit)']
//--------------------------------------------------------------
VG_(debugLog)(1, "main",
"(main_) Process Valgrind's command line options, "
"setup logging\n");
main_process_cmd_line_options ( &logging_to_fd, &xml_fname_unexpanded,
toolname );
//--------------------------------------------------------------
// Zeroise the millisecond counter by doing a first read of it.
// p: none
//--------------------------------------------------------------
(void) VG_(read_millisecond_timer)();
//--------------------------------------------------------------
// Print the preamble
// p: tl_pre_clo_init [for 'VG_(details).name' and friends]
// p: main_process_cmd_line_options()
// [for VG_(clo_verbosity), VG_(clo_xml),
// logging_to_fd, xml_fname_unexpanded]
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Print the preamble...\n");
print_preamble(logging_to_fd, xml_fname_unexpanded, toolname);
VG_(debugLog)(1, "main", "...finished the preamble\n");
//--------------------------------------------------------------
// Init tool part 2: post_clo_init
// p: setup_client_stack() [for 'VG_(client_arg[cv]']
// p: setup_file_descriptors() [for 'VG_(fd_xxx_limit)']
// p: print_preamble() [so any warnings printed in post_clo_init
// are shown after the preamble]
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Initialise the tool part 2 (post_clo_init)\n");
VG_TDICT_CALL(tool_post_clo_init);
{
/* The tool's "needs" will by now be finalised, since it has no
further opportunity to specify them. So now sanity check
them. */
const HChar* s;
Bool ok;
ok = VG_(sanity_check_needs)( &s );
if (!ok) {
VG_(core_panic)(s);
}
}
//--------------------------------------------------------------
// Initialise translation table and translation cache
// p: aspacem [??]
// p: tl_pre_clo_init [for 'VG_(details).avg_translation_sizeB']
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Initialise TT/TC\n");
VG_(init_tt_tc)();
//--------------------------------------------------------------
// Initialise the redirect table.
// p: init_tt_tc [so it can call VG_(search_transtab) safely]
// p: aspacem [so can change ownership of sysinfo pages]
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Initialise redirects\n");
VG_(redir_initialise)();
//--------------------------------------------------------------
// Allow GDB attach
// p: main_process_cmd_line_options() [for VG_(clo_wait_for_gdb)]
//--------------------------------------------------------------
/* Hook to delay things long enough so we can get the pid and
attach GDB in another shell. */
if (VG_(clo_wait_for_gdb)) {
ULong iters, q;
VG_(debugLog)(1, "main", "Wait for GDB\n");
VG_(printf)("pid=%d, entering delay loop\n", VG_(getpid)());
# if defined(VGP_x86_linux)
iters = 10;
# elif defined(VGP_amd64_linux) || defined(VGP_ppc64be_linux) \
|| defined(VGP_ppc64le_linux) || defined(VGP_tilegx_linux)
iters = 10;
# elif defined(VGP_ppc32_linux)
iters = 5;
# elif defined(VGP_arm_linux)
iters = 5;
# elif defined(VGP_arm64_linux)
iters = 5;
# elif defined(VGP_s390x_linux)
iters = 10;
# elif defined(VGP_mips32_linux) || defined(VGP_mips64_linux)
iters = 10;
# elif defined(VGO_darwin)
iters = 3;
# elif defined(VGO_solaris)
iters = 10;
# else
# error "Unknown plat"
# endif
iters *= 1000ULL * 1000 * 1000;
for (q = 0; q < iters; q++)
__asm__ __volatile__("" ::: "memory","cc");
}
//--------------------------------------------------------------
// Search for file descriptors that are inherited from our parent
// p: main_process_cmd_line_options [for VG_(clo_track_fds)]
//--------------------------------------------------------------
if (VG_(clo_track_fds)) {
VG_(debugLog)(1, "main", "Init preopened fds\n");
VG_(init_preopened_fds)();
}
#if defined(VGO_solaris)
VG_(syswrap_init)();
#endif
//--------------------------------------------------------------
// Load debug info for the existing segments.
// p: setup_code_redirect_table [so that redirs can be recorded]
// p: mallocfree
// p: probably: setup fds and process CLOs, so that logging works
// p: initialise m_debuginfo
//
// While doing this, make a note of the debuginfo-handles that
// come back from VG_(di_notify_mmap).
// Later, in "Tell the tool about the initial client memory permissions"
// (just below) we can then hand these handles off to the tool in
// calls to VG_TRACK(new_mem_startup, ...). This gives the tool the
// opportunity to make further queries to m_debuginfo before the
// client is started, if it wants. We put this information into an
// XArray, each handle along with the associated segment start address,
// and search the XArray for the handles later, when calling
// VG_TRACK(new_mem_startup, ...).
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Load initial debug info\n");
vg_assert(!addr2dihandle);
addr2dihandle = VG_(newXA)( VG_(malloc), "main.vm.2",
VG_(free), sizeof(Addr_n_ULong) );
# if defined(VGO_linux) || defined(VGO_solaris)
{ Addr* seg_starts;
Int n_seg_starts;
Addr_n_ULong anu;
seg_starts = VG_(get_segment_starts)( SkFileC | SkFileV, &n_seg_starts );
vg_assert(seg_starts && n_seg_starts >= 0);
/* show them all to the debug info reader. allow_SkFileV has to
be True here so that we read info from the valgrind executable
itself. */
for (i = 0; i < n_seg_starts; i++) {
anu.ull = VG_(di_notify_mmap)( seg_starts[i], True/*allow_SkFileV*/,
-1/*Don't use_fd*/);
/* anu.ull holds the debuginfo handle returned by di_notify_mmap,
if any. */
if (anu.ull > 0) {
anu.a = seg_starts[i];
VG_(addToXA)( addr2dihandle, &anu );
}
}
VG_(free)( seg_starts );
}
# elif defined(VGO_darwin)
{ Addr* seg_starts;
Int n_seg_starts;
seg_starts = VG_(get_segment_starts)( SkFileC, &n_seg_starts );
vg_assert(seg_starts && n_seg_starts >= 0);
/* show them all to the debug info reader.
Don't read from V segments (unlike Linux) */
// GrP fixme really?
for (i = 0; i < n_seg_starts; i++) {
VG_(di_notify_mmap)( seg_starts[i], False/*don't allow_SkFileV*/,
-1/*don't use_fd*/);
}
VG_(free)( seg_starts );
}
# else
# error Unknown OS
# endif
//--------------------------------------------------------------
// Tell aspacem of ownership change of the asm helpers, so that
// m_translate allows them to be translated. However, only do this
// after the initial debug info read, since making a hole in the
// address range for the stage2 binary confuses the debug info reader.
// p: aspacem
//--------------------------------------------------------------
{ Bool change_ownership_v_c_OK;
Addr co_start = VG_PGROUNDDN( (Addr)&VG_(trampoline_stuff_start) );
Addr co_endPlus = VG_PGROUNDUP( (Addr)&VG_(trampoline_stuff_end) );
VG_(debugLog)(1,"redir",
"transfer ownership V -> C of 0x%llx .. 0x%llx\n",
(ULong)co_start, (ULong)co_endPlus-1 );
change_ownership_v_c_OK
= VG_(am_change_ownership_v_to_c)( co_start, co_endPlus - co_start );
vg_assert(change_ownership_v_c_OK);
}
if (VG_(clo_xml)) {
HChar buf[50]; // large enough
VG_(elapsed_wallclock_time)(buf, sizeof buf);
VG_(printf_xml)( "<status>\n"
" <state>RUNNING</state>\n"
" <time>%pS</time>\n"
"</status>\n",
buf );
VG_(printf_xml)( "\n" );
}
VG_(init_Threads)();
//--------------------------------------------------------------
// Initialise the scheduler (phase 1) [generates tid_main]
// p: none, afaics
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Initialise scheduler (phase 1)\n");
tid_main = VG_(scheduler_init_phase1)();
vg_assert(tid_main >= 0 && tid_main < VG_N_THREADS
&& tid_main != VG_INVALID_THREADID);
/* Tell the tool about tid_main */
VG_TRACK( pre_thread_ll_create, VG_INVALID_THREADID, tid_main );
//--------------------------------------------------------------
// Tell the tool about the initial client memory permissions
// p: aspacem
// p: mallocfree
// p: setup_client_stack
// p: setup_client_dataseg
//
// For each segment we tell the client about, look up in
// addr2dihandle as created above, to see if there's a debuginfo
// handle associated with the segment, that we can hand along
// to the tool, to be helpful.
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Tell tool about initial permissions\n");
{ Addr* seg_starts;
Int n_seg_starts;
vg_assert(addr2dihandle);
/* Mark the main thread as running while we tell the tool about
the client memory so that the tool can associate that memory
with the main thread. */
vg_assert(VG_(running_tid) == VG_INVALID_THREADID);
VG_(running_tid) = tid_main;
seg_starts = VG_(get_segment_starts)( SkFileC | SkAnonC | SkShmC,
&n_seg_starts );
vg_assert(seg_starts && n_seg_starts >= 0);
/* Show client segments to the tool */
for (i = 0; i < n_seg_starts; i++) {
Word j, n;
NSegment const* seg
= VG_(am_find_nsegment)( seg_starts[i] );
vg_assert(seg);
vg_assert(seg->kind == SkFileC || seg->kind == SkAnonC ||
seg->kind == SkShmC);
vg_assert(seg->start == seg_starts[i]);
{
VG_(debugLog)(2, "main",
"tell tool about %010lx-%010lx %c%c%c\n",
seg->start, seg->end,
seg->hasR ? 'r' : '-',
seg->hasW ? 'w' : '-',
seg->hasX ? 'x' : '-' );
/* search addr2dihandle to see if we have an entry
matching seg->start. */
n = VG_(sizeXA)( addr2dihandle );
for (j = 0; j < n; j++) {
Addr_n_ULong* anl = VG_(indexXA)( addr2dihandle, j );
if (anl->a == seg->start) {
vg_assert(anl->ull > 0); /* check it's a valid handle */
break;
}
}
vg_assert(j >= 0 && j <= n);
VG_TRACK( new_mem_startup, seg->start, seg->end+1-seg->start,
seg->hasR, seg->hasW, seg->hasX,
/* and the retrieved debuginfo handle, if any */
j < n
? ((Addr_n_ULong*)VG_(indexXA)( addr2dihandle, j ))->ull
: 0 );
}
}
VG_(free)( seg_starts );
VG_(deleteXA)( addr2dihandle );
/* Also do the initial stack permissions. */
{
SSizeT inaccessible_len;
NSegment const* seg
= VG_(am_find_nsegment)( the_iifii.initial_client_SP );
vg_assert(seg);
vg_assert(seg->kind == SkAnonC);
vg_assert(the_iifii.initial_client_SP >= seg->start);
vg_assert(the_iifii.initial_client_SP <= seg->end);
/* Stuff below the initial SP is unaddressable. Take into
account any ABI-mandated space below the stack pointer that
is required (VG_STACK_REDZONE_SZB). setup_client_stack()
will have allocated an extra page if a red zone is required,
to be on the safe side. */
inaccessible_len = the_iifii.initial_client_SP - VG_STACK_REDZONE_SZB
- seg->start;
vg_assert(inaccessible_len >= 0);
if (inaccessible_len > 0)
VG_TRACK( die_mem_stack,
seg->start,
inaccessible_len );
VG_(debugLog)(2, "main", "mark stack inaccessible %010lx-%010lx\n",
seg->start,
the_iifii.initial_client_SP-1 - VG_STACK_REDZONE_SZB);
}
/* Also the assembly helpers. */
VG_TRACK( new_mem_startup,
(Addr)&VG_(trampoline_stuff_start),
(Addr)&VG_(trampoline_stuff_end)
- (Addr)&VG_(trampoline_stuff_start),
False, /* readable? */
False, /* writable? */
True /* executable? */,
0 /* di_handle: no associated debug info */ );
/* Clear the running thread indicator */
VG_(running_tid) = VG_INVALID_THREADID;
vg_assert(VG_(running_tid) == VG_INVALID_THREADID);
/* Darwin only: tell the tools where the client's kernel commpage
is. It would be better to do this by telling aspacemgr about
it -- see the now disused record_system_memory() in
initimg-darwin.c -- but that causes the sync checker to fail,
since the mapping doesn't appear in the kernel-supplied
process map. So do it here instead. */
# if defined(VGP_amd64_darwin)
VG_TRACK( new_mem_startup,
0x7fffffe00000, 0x7ffffffff000-0x7fffffe00000,
True, False, True, /* r-x */
0 /* di_handle: no associated debug info */ );
# elif defined(VGP_x86_darwin)
VG_TRACK( new_mem_startup,
0xfffec000, 0xfffff000-0xfffec000,
True, False, True, /* r-x */
0 /* di_handle: no associated debug info */ );
# endif
}
//--------------------------------------------------------------
// Initialise the scheduler (phase 2)
// p: Initialise the scheduler (phase 1) [for tid_main]
// p: setup_file_descriptors() [else VG_(safe_fd)() breaks]
// p: setup_client_stack
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Initialise scheduler (phase 2)\n");
{ NSegment const* seg
= VG_(am_find_nsegment)( the_iifii.initial_client_SP );
vg_assert(seg);
vg_assert(seg->kind == SkAnonC);
vg_assert(the_iifii.initial_client_SP >= seg->start);
vg_assert(the_iifii.initial_client_SP <= seg->end);
VG_(scheduler_init_phase2)( tid_main,
seg->end, the_iifii.clstack_max_size );
}
//--------------------------------------------------------------
// Set up state for the root thread
// p: ?
// setup_scheduler() [for sched-specific thread 1 stuff]
// VG_(ii_create_image) [for 'the_iicii' initial info]
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Finalise initial image\n");
{ /* Mark the main thread as running while we tell the tool about
the client memory which could be tracked during initial image
finalisation. So the tool can associate that memory with the
main thread. */
vg_assert(VG_(running_tid) == VG_INVALID_THREADID);
VG_(running_tid) = tid_main;
VG_(ii_finalise_image)( the_iifii );
/* Clear the running thread indicator */
VG_(running_tid) = VG_INVALID_THREADID;
vg_assert(VG_(running_tid) == VG_INVALID_THREADID);
}
//--------------------------------------------------------------
// Initialise the signal handling subsystem
// p: n/a
//--------------------------------------------------------------
// Nb: temporarily parks the saved blocking-mask in saved_sigmask.
VG_(debugLog)(1, "main", "Initialise signal management\n");
/* Check that the kernel-interface signal definitions look sane */
VG_(vki_do_initial_consistency_checks)();
/* .. and go on to use them. */
VG_(sigstartup_actions)();
//--------------------------------------------------------------
// Read suppression file
// p: main_process_cmd_line_options() [for VG_(clo_suppressions)]
//--------------------------------------------------------------
if (VG_(needs).core_errors || VG_(needs).tool_errors) {
VG_(debugLog)(1, "main", "Load suppressions\n");
VG_(load_suppressions)();
}
//--------------------------------------------------------------
// register client stack
//--------------------------------------------------------------
VG_(clstk_id) = VG_(register_stack)(VG_(clstk_start_base), VG_(clstk_end));
//--------------------------------------------------------------
// Show the address space state so far
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "\n");
VG_(debugLog)(1, "main", "\n");
VG_(am_show_nsegments)(1,"Memory layout at client startup");
VG_(debugLog)(1, "main", "\n");
VG_(debugLog)(1, "main", "\n");
//--------------------------------------------------------------
// Run!
//--------------------------------------------------------------
VG_(debugLog)(1, "main", "Running thread 1\n");
/* As a result of the following call, the last thread standing
eventually winds up running shutdown_actions_NORETURN
just below. Unfortunately, simply exporting said function
causes m_main to be part of a module cycle, which is pretty
nonsensical. So instead of doing that, the address of said
function is stored in a global variable 'owned' by m_syswrap,
and it uses that function pointer to get back here when it needs
to. */
/* Set continuation address. */
VG_(address_of_m_main_shutdown_actions_NORETURN)
= & shutdown_actions_NORETURN;
/* Run the first thread, eventually ending up at the continuation
address. */
VG_(main_thread_wrapper_NORETURN)(1);
/*NOTREACHED*/
vg_assert(0);
}
/* Do everything which needs doing when the last thread exits or when
a thread exits requesting a complete process exit.
We enter here holding The Lock. For the case VgSrc_ExitProcess we
must never release it, because to do so would allow other threads
to continue after the system is ostensibly shut down. So we must
go to our grave, so to speak, holding the lock.
In fact, there is never any point in releasing the lock at this
point - we have it, we're shutting down the entire system, and
for the case VgSrc_ExitProcess doing so positively causes trouble.
So don't.
The final_tidyup call makes a bit of a nonsense of the ExitProcess
case, since it will run the libc_freeres function, thus allowing
other lurking threads to run again. Hmm. */
static
void shutdown_actions_NORETURN( ThreadId tid,
VgSchedReturnCode tids_schedretcode )
{
VG_(debugLog)(1, "main", "entering VG_(shutdown_actions_NORETURN)\n");
VG_(am_show_nsegments)(1,"Memory layout at client shutdown");
vg_assert(VG_(is_running_thread)(tid));
vg_assert(tids_schedretcode == VgSrc_ExitThread
|| tids_schedretcode == VgSrc_ExitProcess
|| tids_schedretcode == VgSrc_FatalSig );
if (tids_schedretcode == VgSrc_ExitThread) {
// We are the last surviving thread. Right?
vg_assert( VG_(count_living_threads)() == 1 );
// Wait for all other threads to exit.
// jrs: Huh? but they surely are already gone
VG_(reap_threads)(tid);
// Clean the client up before the final report
// this causes the libc_freeres function to run
final_tidyup(tid);
/* be paranoid */
vg_assert(VG_(is_running_thread)(tid));
vg_assert(VG_(count_living_threads)() == 1);
} else {
// We may not be the last surviving thread. However, we
// want to shut down the entire process. We hold the lock
// and we need to keep hold of it all the way out, in order
// that none of the other threads ever run again.
vg_assert( VG_(count_living_threads)() >= 1 );
// Clean the client up before the final report
// this causes the libc_freeres function to run
// perhaps this is unsafe, as per comment above
final_tidyup(tid);
/* be paranoid */
vg_assert(VG_(is_running_thread)(tid));
vg_assert(VG_(count_living_threads)() >= 1);
}
/* Final call to gdbserver, if requested. */
if (VG_(gdbserver_stop_at) (VgdbStopAt_Exit)) {
VG_(umsg)("(action at exit) vgdb me ... \n");
VG_(gdbserver) (tid);
}
VG_(threads)[tid].status = VgTs_Empty;
//--------------------------------------------------------------
// Finalisation: cleanup, messages, etc. Order not so important, only
// affects what order the messages come.
//--------------------------------------------------------------
// First thing in the post-amble is a blank line.
if (VG_(clo_xml))
VG_(printf_xml)("\n");
else if (VG_(clo_verbosity) > 0)
VG_(message)(Vg_UserMsg, "\n");
if (VG_(clo_xml)) {
HChar buf[50]; // large enough
VG_(elapsed_wallclock_time)(buf, sizeof buf);
VG_(printf_xml)( "<status>\n"
" <state>FINISHED</state>\n"
" <time>%pS</time>\n"
"</status>\n"
"\n",
buf);
}
/* Print out file descriptor summary and stats. */
if (VG_(clo_track_fds))
VG_(show_open_fds)("at exit");
/* Call the tool's finalisation function. This makes Memcheck's
leak checker run, and possibly chuck a bunch of leak errors into
the error management machinery. */
VG_TDICT_CALL(tool_fini, 0/*exitcode*/);
/* Show the error counts. */
if (VG_(clo_xml)
&& (VG_(needs).core_errors || VG_(needs).tool_errors)) {
VG_(show_error_counts_as_XML)();
}
/* In XML mode, this merely prints the used suppressions. */
if (VG_(needs).core_errors || VG_(needs).tool_errors)
VG_(show_all_errors)(VG_(clo_verbosity), VG_(clo_xml));
if (VG_(clo_xml)) {
VG_(printf_xml)("\n");
VG_(printf_xml)("</valgrindoutput>\n");
VG_(printf_xml)("\n");
}
VG_(sanity_check_general)( True /*include expensive checks*/ );
if (VG_(clo_stats))
VG_(print_all_stats)(VG_(clo_verbosity) >= 1, /* Memory stats */
False /* tool prints stats in the tool fini */);
/* Show a profile of the heap(s) at shutdown. Optionally, first
throw away all the debug info, as that makes it easy to spot
leaks in the debuginfo reader. */
if (VG_(clo_profile_heap)) {
if (0) VG_(di_discard_ALL_debuginfo)();
VG_(print_arena_cc_analysis)();
}
/* If profiling has been requested, but with zero interval, it
means "profile at the end of the run only". In which case we
need to dump the profile now. */
if (VG_(clo_profyle_sbs) && VG_(clo_profyle_interval) == 0) {
VG_(get_and_show_SB_profile)(0/*denoting end-of-run*/);
}
/* Print Vex storage stats */
if (0)
LibVEX_ShowAllocStats();
/* Flush any output cached by previous calls to VG_(message). */
VG_(message_flush)();
/* Terminate gdbserver if ever it was started. We terminate it here
so that it get the output above if output was redirected to
gdb */
VG_(gdbserver_exit) (tid, tids_schedretcode);
/* Ok, finally exit in the os-specific way, according to the scheduler's
return code. In short, if the (last) thread exited by calling
sys_exit, do likewise; if the (last) thread stopped due to a fatal
signal, terminate the entire system with that same fatal signal. */
VG_(debugLog)(1, "core_os",
"VG_(terminate_NORETURN)(tid=%u)\n", tid);
switch (tids_schedretcode) {
case VgSrc_ExitThread: /* the normal way out (Linux, Solaris) */
case VgSrc_ExitProcess: /* the normal way out (Darwin) */
/* Change the application return code to user's return code,
if an error was found */
if (VG_(clo_error_exitcode) > 0
&& VG_(get_n_errs_found)() > 0) {
VG_(client_exit)( VG_(clo_error_exitcode) );
} else {
/* otherwise, return the client's exit code, in the normal
way. */
VG_(client_exit)( VG_(threads)[tid].os_state.exitcode );
}
/* NOT ALIVE HERE! */
VG_(core_panic)("entered the afterlife in main() -- ExitT/P");
break; /* what the hell :) */
case VgSrc_FatalSig:
/* We were killed by a fatal signal, so replicate the effect */
vg_assert(VG_(threads)[tid].os_state.fatalsig != 0);
VG_(kill_self)(VG_(threads)[tid].os_state.fatalsig);
/* we shouldn't be alive at this point. But VG_(kill_self)
sometimes fails with EPERM on Darwin, for unclear reasons. */
# if defined(VGO_darwin)
VG_(debugLog)(0, "main", "VG_(kill_self) failed. Exiting normally.\n");
VG_(exit)(0); /* bogus, but we really need to exit now */
/* fall through .. */
# endif
VG_(core_panic)("main(): signal was supposed to be fatal");
break;
default:
VG_(core_panic)("main(): unexpected scheduler return code");
}
}
/* -------------------- */
/* Final clean-up before terminating the process.
Clean up the client by calling __libc_freeres() (if requested)
This is Linux-specific?
GrP fixme glibc-specific, anyway
*/
static void final_tidyup(ThreadId tid)
{
#if !defined(VGO_darwin)
Addr __libc_freeres_wrapper = VG_(client___libc_freeres_wrapper);
vg_assert(VG_(is_running_thread)(tid));
if ( !VG_(needs).libc_freeres ||
!VG_(clo_run_libc_freeres) ||
0 == __libc_freeres_wrapper )
return; /* can't/won't do it */
# if defined(VGP_ppc64be_linux)
Addr r2 = VG_(get_tocptr)( __libc_freeres_wrapper );
if (r2 == 0) {
VG_(message)(Vg_UserMsg,
"Caught __NR_exit, but can't run __libc_freeres()\n");
VG_(message)(Vg_UserMsg,
" since cannot establish TOC pointer for it.\n");
return;
}
# endif
if (VG_(clo_verbosity) > 2 ||
VG_(clo_trace_syscalls) ||
VG_(clo_trace_sched))
VG_(message)(Vg_DebugMsg,
"Caught __NR_exit; running __libc_freeres()\n");
/* set thread context to point to libc_freeres_wrapper */
/* ppc64be-linux note: __libc_freeres_wrapper gives us the real
function entry point, not a fn descriptor, so can use it
directly. However, we need to set R2 (the toc pointer)
appropriately. */
VG_(set_IP)(tid, __libc_freeres_wrapper);
# if defined(VGP_ppc64be_linux)
VG_(threads)[tid].arch.vex.guest_GPR2 = r2;
# elif defined(VGP_ppc64le_linux)
/* setting GPR2 but not really needed, GPR12 is needed */
VG_(threads)[tid].arch.vex.guest_GPR2 = __libc_freeres_wrapper;
VG_(threads)[tid].arch.vex.guest_GPR12 = __libc_freeres_wrapper;
# endif
/* mips-linux note: we need to set t9 */
# if defined(VGP_mips32_linux) || defined(VGP_mips64_linux)
VG_(threads)[tid].arch.vex.guest_r25 = __libc_freeres_wrapper;
# endif
/* Block all blockable signals by copying the real block state into
the thread's block state*/
VG_(sigprocmask)(VKI_SIG_BLOCK, NULL, &VG_(threads)[tid].sig_mask);
VG_(threads)[tid].tmp_sig_mask = VG_(threads)[tid].sig_mask;
/* and restore handlers to default */
VG_(set_default_handler)(VKI_SIGSEGV);
VG_(set_default_handler)(VKI_SIGBUS);
VG_(set_default_handler)(VKI_SIGILL);
VG_(set_default_handler)(VKI_SIGFPE);
// We were exiting, so assert that...
vg_assert(VG_(is_exiting)(tid));
// ...but now we're not again
VG_(threads)[tid].exitreason = VgSrc_None;
// run until client thread exits - ideally with LIBC_FREERES_DONE,
// but exit/exitgroup/signal will do
VG_(scheduler)(tid);
vg_assert(VG_(is_exiting)(tid));
#endif
}
/*====================================================================*/
/*=== Getting to main() alive: LINUX ===*/
/*====================================================================*/
#if defined(VGO_linux)
/* If linking of the final executables is done with glibc present,
then Valgrind starts at main() above as usual, and all of the
following code is irrelevant.
However, this is not the intended mode of use. The plan is to
avoid linking against glibc, by giving gcc the flags
-nodefaultlibs -lgcc -nostartfiles at startup.
From this derive two requirements:
1. gcc may emit calls to memcpy, memmove and memset to deal with
structure assignments etc. Since we have chosen to ignore all the
"normal" supporting libraries, we have to provide our own
implementations of them. No problem.
2. We have to provide a symbol "_start", to which the kernel
hands control at startup. Hence the code below.
*/
/* ---------------- Requirement 1 ---------------- */
void* memcpy(void *dest, const void *src, SizeT n);
void* memcpy(void *dest, const void *src, SizeT n) {
return VG_(memcpy)(dest,src,n);
}
void* memmove(void *dest, const void *src, SizeT n);
void* memmove(void *dest, const void *src, SizeT n) {
return VG_(memmove)(dest,src,n);
}
void* memset(void *s, int c, SizeT n);
void* memset(void *s, int c, SizeT n) {
return VG_(memset)(s,c,n);
}
/* BVA: abort() for those platforms that need it (PPC and ARM). */
void abort(void);
void abort(void){
VG_(printf)("Something called raise().\n");
vg_assert(0);
}
/* EAZG: ARM's EABI will call floating point exception handlers in
libgcc which boil down to an abort or raise, that's usually defined
in libc. Instead, define them here. */
#if defined(VGP_arm_linux)
void raise(void);
void raise(void){
VG_(printf)("Something called raise().\n");
vg_assert(0);
}
void __aeabi_unwind_cpp_pr0(void);
void __aeabi_unwind_cpp_pr0(void){
VG_(printf)("Something called __aeabi_unwind_cpp_pr0()\n");
vg_assert(0);
}
void __aeabi_unwind_cpp_pr1(void);
void __aeabi_unwind_cpp_pr1(void){
VG_(printf)("Something called __aeabi_unwind_cpp_pr1()\n");
vg_assert(0);
}
#if defined(ANDROID) && defined(__clang__)
/* Replace __aeabi_memcpy* functions with vgPlain_memcpy. */
void* __aeabi_memcpy(void *dest, const void *src, SizeT n);
void* __aeabi_memcpy(void *dest, const void *src, SizeT n)
{
return VG_(memcpy)(dest, src, n);
}
void* __aeabi_memcpy4(void *dest, const void *src, SizeT n);
void* __aeabi_memcpy4(void *dest, const void *src, SizeT n)
{
return VG_(memcpy)(dest, src, n);
}
void* __aeabi_memcpy8(void *dest, const void *src, SizeT n);
void* __aeabi_memcpy8(void *dest, const void *src, SizeT n)
{
return VG_(memcpy)(dest, src, n);
}
/* Replace __aeabi_memclr* functions with vgPlain_memset. */
void* __aeabi_memclr(void *dest, SizeT n);
void* __aeabi_memclr(void *dest, SizeT n)
{
return VG_(memset)(dest, 0, n);
}
void* __aeabi_memclr4(void *dest, SizeT n);
void* __aeabi_memclr4(void *dest, SizeT n)
{
return VG_(memset)(dest, 0, n);
}
void* __aeabi_memclr8(void *dest, SizeT n);
void* __aeabi_memclr8(void *dest, SizeT n)
{
return VG_(memset)(dest, 0, n);
}
#endif /* ANDROID __clang__ */
#endif
/* ---------------- Requirement 2 ---------------- */
/* Glibc's sysdeps/i386/elf/start.S has the following gem of a
comment, which explains how the stack looks right at process start
(when _start is jumped to). Hence _start passes %esp to
_start_in_C_linux, which extracts argc/argv/envp and starts up
correctly. */
/* This is the canonical entry point, usually the first thing in the text
segment. The SVR4/i386 ABI (pages 3-31, 3-32) says that when the entry
point runs, most registers' values are unspecified, except for:
%edx Contains a function pointer to be registered with `atexit'.
This is how the dynamic linker arranges to have DT_FINI
functions called for shared libraries that have been loaded
before this code runs.
%esp The stack contains the arguments and environment:
0(%esp) argc
4(%esp) argv[0]
...
(4*argc)(%esp) NULL
(4*(argc+1))(%esp) envp[0]
...
NULL
*/
/* The kernel hands control to _start, which extracts the initial
stack pointer and calls onwards to _start_in_C_linux. This also switches
the new stack. */
#if defined(VGP_x86_linux)
asm("\n"
".text\n"
"\t.globl _start\n"
"\t.type _start,@function\n"
"_start:\n"
/* set up the new stack in %eax */
"\tmovl $vgPlain_interim_stack, %eax\n"
"\taddl $"VG_STRINGIFY(VG_STACK_GUARD_SZB)", %eax\n"
"\taddl $"VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)", %eax\n"
/* allocate at least 16 bytes on the new stack, and aligned */
"\tsubl $16, %eax\n"
"\tandl $~15, %eax\n"
/* install it, and collect the original one */
"\txchgl %eax, %esp\n"
/* call _start_in_C_linux, passing it the startup %esp */
"\tmovl %eax, (%esp)\n"
"\tcall _start_in_C_linux\n"
"\thlt\n"
".previous\n"
);
#elif defined(VGP_amd64_linux)
asm("\n"
".text\n"
"\t.globl _start\n"
"\t.type _start,@function\n"
"_start:\n"
/* set up the new stack in %rdi */
"\tmovq $vgPlain_interim_stack, %rdi\n"
"\taddq $"VG_STRINGIFY(VG_STACK_GUARD_SZB)", %rdi\n"
"\taddq $"VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)", %rdi\n"
"\tandq $~15, %rdi\n"
/* install it, and collect the original one */
"\txchgq %rdi, %rsp\n"
/* call _start_in_C_linux, passing it the startup %rsp */
"\tcall _start_in_C_linux\n"
"\thlt\n"
".previous\n"
);
#elif defined(VGP_ppc32_linux)
asm("\n"
".text\n"
"\t.globl _start\n"
"\t.type _start,@function\n"
"_start:\n"
/* set up the new stack in r16 */
"\tlis 16,vgPlain_interim_stack@ha\n"
"\tla 16,vgPlain_interim_stack@l(16)\n"
"\tlis 17,("VG_STRINGIFY(VG_STACK_GUARD_SZB)" >> 16)\n"
"\tori 17,17,("VG_STRINGIFY(VG_STACK_GUARD_SZB)" & 0xFFFF)\n"
"\tlis 18,("VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)" >> 16)\n"
"\tori 18,18,("VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)" & 0xFFFF)\n"
"\tadd 16,17,16\n"
"\tadd 16,18,16\n"
"\trlwinm 16,16,0,0,27\n"
/* now r16 = &vgPlain_interim_stack + VG_STACK_GUARD_SZB +
VG_DEFAULT_STACK_ACTIVE_SZB rounded down to the nearest 16-byte
boundary. And r1 is the original SP. Set the SP to r16 and
call _start_in_C_linux, passing it the initial SP. */
"\tmr 3,1\n"
"\tmr 1,16\n"
"\tbl _start_in_C_linux\n"
"\ttrap\n"
".previous\n"
);
#elif defined(VGP_ppc64be_linux)
asm("\n"
/* PPC64 ELF ABI says '_start' points to a function descriptor.
So we must have one, and that is what goes into the .opd section. */
"\t.align 2\n"
"\t.global _start\n"
"\t.section \".opd\",\"aw\"\n"
"\t.align 3\n"
"_start:\n"
"\t.quad ._start,.TOC.@tocbase,0\n"
"\t.previous\n"
"\t.type ._start,@function\n"
"\t.global ._start\n"
"._start:\n"
/* set up the new stack in r16 */
"\tlis 16, vgPlain_interim_stack@highest\n"
"\tori 16,16,vgPlain_interim_stack@higher\n"
"\tsldi 16,16,32\n"
"\toris 16,16,vgPlain_interim_stack@h\n"
"\tori 16,16,vgPlain_interim_stack@l\n"
"\txor 17,17,17\n"
"\tlis 17,("VG_STRINGIFY(VG_STACK_GUARD_SZB)" >> 16)\n"
"\tori 17,17,("VG_STRINGIFY(VG_STACK_GUARD_SZB)" & 0xFFFF)\n"
"\txor 18,18,18\n"
"\tlis 18,("VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)" >> 16)\n"
"\tori 18,18,("VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)" & 0xFFFF)\n"
"\tadd 16,17,16\n"
"\tadd 16,18,16\n"
"\trldicr 16,16,0,59\n"
/* now r16 = &vgPlain_interim_stack + VG_STACK_GUARD_SZB +
VG_DEFAULT_STACK_ACTIVE_SZB rounded down to the nearest 16-byte
boundary. And r1 is the original SP. Set the SP to r16 and
call _start_in_C_linux, passing it the initial SP. */
"\tmr 3,1\n"
"\tmr 1,16\n"
"\tlis 14, _start_in_C_linux@highest\n"
"\tori 14,14,_start_in_C_linux@higher\n"
"\tsldi 14,14,32\n"
"\toris 14,14,_start_in_C_linux@h\n"
"\tori 14,14,_start_in_C_linux@l\n"
"\tld 14,0(14)\n"
"\tmtctr 14\n"
"\tbctrl\n"
"\tnop\n"
"\ttrap\n"
);
#elif defined(VGP_ppc64le_linux)
/* Little Endian uses ELF version 2 but in the future may also
* support other ELF versions.
*/
asm("\n"
"\t.align 2\n"
"\t.global _start\n"
"\t.type _start,@function\n"
"_start:\n"
"#if _CALL_ELF == 2 \n"
"0: addis 2,12,.TOC.-0b@ha\n"
" addi 2,2,.TOC.-0b@l\n"
" .localentry _start, .-_start\n"
"#endif \n"
/* set up the new stack in r16 */
"\tlis 16, vgPlain_interim_stack@highest\n"
"\tori 16,16,vgPlain_interim_stack@higher\n"
"\tsldi 16,16,32\n"
"\toris 16,16,vgPlain_interim_stack@h\n"
"\tori 16,16,vgPlain_interim_stack@l\n"
"\txor 17,17,17\n"
"\tlis 17,("VG_STRINGIFY(VG_STACK_GUARD_SZB)" >> 16)\n"
"\tori 17,17,("VG_STRINGIFY(VG_STACK_GUARD_SZB)" & 0xFFFF)\n"
"\txor 18,18,18\n"
"\tlis 18,("VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)" >> 16)\n"
"\tori 18,18,("VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)" & 0xFFFF)\n"
"\tadd 16,17,16\n"
"\tadd 16,18,16\n"
"\trldicr 16,16,0,59\n"
/* now r16 = &vgPlain_interim_stack + VG_STACK_GUARD_SZB +
VG_DEFAULT_STACK_ACTIVE_SZB rounded down to the nearest 16-byte
boundary. And r1 is the original SP. Set the SP to r16 and
call _start_in_C_linux, passing it the initial SP. */
"\tmr 3,1\n"
"\tmr 1,16\n"
"\tlis 14, _start_in_C_linux@highest\n"
"\tori 14,14,_start_in_C_linux@higher\n"
"\tsldi 14,14,32\n"
"\toris 14,14,_start_in_C_linux@h\n"
"\tori 14,14,_start_in_C_linux@l\n"
"\tmtctr 14\n"
"\tbctrl\n"
"\tnop\n"
"\ttrap\n"
);
#elif defined(VGP_s390x_linux)
/*
This is the canonical entry point, usually the first thing in the text
segment. Most registers' values are unspecified, except for:
%r14 Contains a function pointer to be registered with `atexit'.
This is how the dynamic linker arranges to have DT_FINI
functions called for shared libraries that have been loaded
before this code runs.
%r15 The stack contains the arguments and environment:
0(%r15) argc
8(%r15) argv[0]
...
(8*argc)(%r15) NULL
(8*(argc+1))(%r15) envp[0]
...
NULL
*/
asm("\n\t"
".text\n\t"
".globl _start\n\t"
".type _start,@function\n\t"
"_start:\n\t"
/* set up the new stack in %r1 */
"larl %r1, vgPlain_interim_stack\n\t"
"larl %r5, 1f\n\t"
"ag %r1, 0(%r5)\n\t"
"ag %r1, 2f-1f(%r5)\n\t"
"nill %r1, 0xFFF0\n\t"
/* install it, and collect the original one */
"lgr %r2, %r15\n\t"
"lgr %r15, %r1\n\t"
/* call _start_in_C_linux, passing it the startup %r15 */
"brasl %r14, _start_in_C_linux\n\t"
/* trigger execution of an invalid opcode -> halt machine */
"j .+2\n\t"
"1: .quad "VG_STRINGIFY(VG_STACK_GUARD_SZB)"\n\t"
"2: .quad "VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)"\n\t"
".previous\n"
);
#elif defined(VGP_arm_linux)
asm("\n"
"\t.text\n"
"\t.align 4\n"
"\t.type _start,#function\n"
"\t.global _start\n"
"_start:\n"
"\tldr r0, [pc, #36]\n"
"\tldr r1, [pc, #36]\n"
"\tadd r0, r1, r0\n"
"\tldr r1, [pc, #32]\n"
"\tadd r0, r1, r0\n"
"\tmvn r1, #15\n"
"\tand r0, r0, r1\n"
"\tmov r1, sp\n"
"\tmov sp, r0\n"
"\tmov r0, r1\n"
"\tb _start_in_C_linux\n"
"\t.word vgPlain_interim_stack\n"
"\t.word "VG_STRINGIFY(VG_STACK_GUARD_SZB)"\n"
"\t.word "VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)"\n"
);
#elif defined(VGP_arm64_linux)
asm("\n"
"\t.text\n"
"\t.align 2\n"
"\t.type _start,#function\n"
"\t.global _start\n"
"_start:\n"
"\tadrp x0, vgPlain_interim_stack\n"
"\tadd x0, x0, :lo12:vgPlain_interim_stack\n"
// The next 2 assume that VG_STACK_GUARD_SZB fits in 32 bits
"\tmov x1, (("VG_STRINGIFY(VG_STACK_GUARD_SZB)") >> 0) & 0xFFFF\n"
"\tmovk x1, (("VG_STRINGIFY(VG_STACK_GUARD_SZB)") >> 16) & 0xFFFF,"
" lsl 16\n"
"\tadd x0, x0, x1\n"
// The next 2 assume that VG_DEFAULT_STACK_ACTIVE_SZB fits in 32 bits
"\tmov x1, (("VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)") >> 0) & 0xFFFF\n"
"\tmovk x1, (("VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)") >> 16) & 0xFFFF,"
" lsl 16\n"
"\tadd x0, x0, x1\n"
"\tand x0, x0, -16\n"
"\tmov x1, sp\n"
"\tmov sp, x0\n"
"\tmov x0, x1\n"
"\tb _start_in_C_linux\n"
);
#elif defined(VGP_mips32_linux)
asm("\n"
"\t.type _gp_disp,@object\n"
".text\n"
"\t.globl __start\n"
"\t.type __start,@function\n"
"__start:\n"
"\tbal 1f\n"
"\tnop\n"
"1:\n"
"\tlui $28, %hi(_gp_disp)\n"
"\taddiu $28, $28, %lo(_gp_disp)\n"
"\taddu $28, $28, $31\n"
/* t1/$9 <- Addr(interim_stack) */
"\tlui $9, %hi(vgPlain_interim_stack)\n"
/* t1/$9 <- Addr(interim_stack) */
"\taddiu $9, %lo(vgPlain_interim_stack)\n"
"\tli $10, "VG_STRINGIFY(VG_STACK_GUARD_SZB)"\n"
"\tli $11, "VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)"\n"
"\taddu $9, $9, $10\n"
"\taddu $9, $9, $11\n"
"\tli $12, 0xFFFFFFF0\n"
"\tand $9, $9, $12\n"
/* now t1/$9 = &vgPlain_interim_stack + VG_STACK_GUARD_SZB +
VG_DEFAULT_STACK_ACTIVE_SZB rounded down to the nearest 16-byte
boundary. And $29 is the original SP. Set the SP to t1 and
call _start_in_C, passing it the initial SP. */
"\tmove $4, $29\n" // a0 <- $sp (_start_in_C first arg)
"\tmove $29, $9\n" // $sp <- t1 (new sp)
"\tlui $25, %hi(_start_in_C_linux)\n"
"\taddiu $25, %lo(_start_in_C_linux)\n"
"\tbal _start_in_C_linux\n"
"\tbreak 0x7\n"
".previous\n"
);
#elif defined(VGP_mips64_linux)
asm(
".text\n"
".globl __start\n"
".type __start,@function\n"
"__start:\n"
"\t.set noreorder\n"
"\t.cpload $25\n"
"\t.set reorder\n"
"\t.cprestore 16\n"
"\tlui $9, %hi(vgPlain_interim_stack)\n"
/* t1/$9 <- Addr(interim_stack) */
"\tdaddiu $9, %lo(vgPlain_interim_stack)\n"
"\tli $10, "VG_STRINGIFY(VG_STACK_GUARD_SZB)"\n"
"\tli $11, "VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)"\n"
"\tdaddu $9, $9, $10\n"
"\tdaddu $9, $9, $11\n"
"\tli $12, 0xFFFFFF00\n"
"\tand $9, $9, $12\n"
/* now t1/$9 = &vgPlain_interim_stack + VG_STACK_GUARD_SZB +
VG_DEFAULT_STACK_ACTIVE_SZB rounded down to the nearest 16-byte
boundary. And $29 is the original SP. Set the SP to t1 and
call _start_in_C, passing it the initial SP. */
"\tmove $4, $29\n" // a0 <- $sp (_start_in_C first arg)
"\tmove $29, $9\n" // $sp <- t1 (new sp)
"\tlui $9, %highest(_start_in_C_linux)\n"
"\tori $9, %higher(_start_in_C_linux)\n"
"\tdsll32 $9, $9, 0x0\n"
"\tlui $10, %hi(_start_in_C_linux)\n"
"\tdaddiu $10, %lo(_start_in_C_linux)\n"
"\tdaddu $25, $9, $10\n"
"\tjalr $25\n"
"\tnop\n"
".previous\n"
);
#elif defined(VGP_tilegx_linux)
asm("\n"
".text\n"
"\t.align 8\n"
"\t.globl _start\n"
"\t.type _start,@function\n"
"_start:\n"
"\tjal 1f\n"
"1:\n"
/* --FIXME, bundle them :) */
/* r19 <- Addr(interim_stack) */
"\tmoveli r19, hw2_last(vgPlain_interim_stack)\n"
"\tshl16insli r19, r19, hw1(vgPlain_interim_stack)\n"
"\tshl16insli r19, r19, hw0(vgPlain_interim_stack)\n"
"\tmoveli r20, hw1("VG_STRINGIFY(VG_STACK_GUARD_SZB)")\n"
"\tshl16insli r20, r20, hw0("VG_STRINGIFY(VG_STACK_GUARD_SZB)")\n"
"\tmoveli r21, hw1("VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)")\n"
"\tshl16insli r21, r21, hw0("VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)")\n"
"\tadd r19, r19, r20\n"
"\tadd r19, r19, r21\n"
"\tmovei r12, 0x0F\n"
"\tnor r12, zero, r12\n"
"\tand r19, r19, r12\n"
/* now r19 = &vgPlain_interim_stack + VG_STACK_GUARD_SZB +
VG_STACK_ACTIVE_SZB rounded down to the nearest 16-byte
boundary. And $54 is the original SP. Set the SP to r0 and
call _start_in_C, passing it the initial SP. */
"\tmove r0, r54\n" // r0 <- $sp (_start_in_C first arg)
"\tmove r54, r19\n" // $sp <- r19 (new sp)
"\tjal _start_in_C_linux\n"
);
#else
# error "Unknown linux platform"
#endif
/* --- !!! --- EXTERNAL HEADERS start --- !!! --- */
#define _GNU_SOURCE
#define _FILE_OFFSET_BITS 64
/* This is in order to get AT_NULL and AT_PAGESIZE. */
#include <elf.h>
/* --- !!! --- EXTERNAL HEADERS end --- !!! --- */
/* Avoid compiler warnings: this fn _is_ used, but labelling it
'static' causes gcc to complain it isn't.
attribute 'used' also ensures the code is not eliminated at link
time */
__attribute__ ((used))
void _start_in_C_linux ( UWord* pArgc );
__attribute__ ((used))
void _start_in_C_linux ( UWord* pArgc )
{
Int r;
Word argc = pArgc[0];
HChar** argv = (HChar**)&pArgc[1];
HChar** envp = (HChar**)&pArgc[1+argc+1];
// For an inner Valgrind, register the interim stack asap.
// This is needed to allow the outer valgrind to do stacktraces during init.
// Note that this stack is not unregistered when the main thread
// is switching to the (real) stack. Unregistering this would imply
// to save the stack id in a global variable, and have a "if"
// in run_a_thread_NORETURN to do the unregistration only for the
// main thread. This unregistration is not worth this complexity.
INNER_REQUEST
((void) VALGRIND_STACK_REGISTER
(&VG_(interim_stack).bytes[0],
&VG_(interim_stack).bytes[0] + sizeof(VG_(interim_stack))));
VG_(memset)( &the_iicii, 0, sizeof(the_iicii) );
VG_(memset)( &the_iifii, 0, sizeof(the_iifii) );
the_iicii.sp_at_startup = (Addr)pArgc;
# if defined(VGP_ppc32_linux) || defined(VGP_ppc64be_linux) \
|| defined(VGP_ppc64le_linux) || defined(VGP_arm64_linux)
{
/* ppc32/ppc64 can be configured with different page sizes.
Determine this early. This is an ugly hack and really should
be moved into valgrind_main. */
UWord *sp = &pArgc[1+argc+1];
while (*sp++ != 0)
;
for (; *sp != AT_NULL && *sp != AT_PAGESZ; sp += 2);
if (*sp == AT_PAGESZ) {
VKI_PAGE_SIZE = sp[1];
for (VKI_PAGE_SHIFT = 12;
VKI_PAGE_SHIFT <= VKI_MAX_PAGE_SHIFT; VKI_PAGE_SHIFT++)
if (VKI_PAGE_SIZE == (1UL << VKI_PAGE_SHIFT))
break;
}
}
# endif
r = valgrind_main( (Int)argc, argv, envp );
/* NOTREACHED */
VG_(exit)(r);
}
/*====================================================================*/
/*=== Getting to main() alive: darwin ===*/
/*====================================================================*/
#elif defined(VGO_darwin)
/*
Memory layout established by kernel:
0(%esp) argc
4(%esp) argv[0]
...
argv[argc-1]
NULL
envp[0]
...
envp[n]
NULL
executable name (presumably, a pointer to it)
NULL
Ditto in the 64-bit case, except all offsets from SP are obviously
twice as large.
*/
/* The kernel hands control to _start, which extracts the initial
stack pointer and calls onwards to _start_in_C_darwin. This also
switches to the new stack. */
#if defined(VGP_x86_darwin)
asm("\n"
".text\n"
".align 2,0x90\n"
"\t.globl __start\n"
"__start:\n"
/* set up the new stack in %eax */
"\tmovl $_vgPlain_interim_stack, %eax\n"
"\taddl $"VG_STRINGIFY(VG_STACK_GUARD_SZB)", %eax\n"
"\taddl $"VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)", %eax\n"
"\tsubl $16, %eax\n"
"\tandl $~15, %eax\n"
/* install it, and collect the original one */
"\txchgl %eax, %esp\n"
"\tsubl $12, %esp\n" // keep stack 16 aligned; see #295428
/* call _start_in_C_darwin, passing it the startup %esp */
"\tpushl %eax\n"
"\tcall __start_in_C_darwin\n"
"\tint $3\n"
"\tint $3\n"
);
#elif defined(VGP_amd64_darwin)
asm("\n"
".text\n"
"\t.globl __start\n"
".align 3,0x90\n"
"__start:\n"
/* set up the new stack in %rdi */
"\tmovabsq $_vgPlain_interim_stack, %rdi\n"
"\taddq $"VG_STRINGIFY(VG_STACK_GUARD_SZB)", %rdi\n"
"\taddq $"VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)", %rdi\n"
"\tandq $~15, %rdi\n"
/* install it, and collect the original one */
"\txchgq %rdi, %rsp\n"
/* call _start_in_C_darwin, passing it the startup %rsp */
"\tcall __start_in_C_darwin\n"
"\tint $3\n"
"\tint $3\n"
);
#endif
void* __memcpy_chk(void *dest, const void *src, SizeT n, SizeT n2);
void* __memcpy_chk(void *dest, const void *src, SizeT n, SizeT n2) {
// skip check
return VG_(memcpy)(dest,src,n);
}
void* __memset_chk(void *s, int c, SizeT n, SizeT n2);
void* __memset_chk(void *s, int c, SizeT n, SizeT n2) {
// skip check
return VG_(memset)(s,c,n);
}
void bzero(void *s, SizeT n);
void bzero(void *s, SizeT n) {
VG_(memset)(s,0,n);
}
void* memcpy(void *dest, const void *src, SizeT n);
void* memcpy(void *dest, const void *src, SizeT n) {
return VG_(memcpy)(dest,src,n);
}
void* memset(void *s, int c, SizeT n);
void* memset(void *s, int c, SizeT n) {
return VG_(memset)(s,c,n);
}
/* Avoid compiler warnings: this fn _is_ used, but labelling it
'static' causes gcc to complain it isn't. */
void _start_in_C_darwin ( UWord* pArgc );
void _start_in_C_darwin ( UWord* pArgc )
{
Int r;
Int argc = *(Int *)pArgc; // not pArgc[0] on LP64
HChar** argv = (HChar**)&pArgc[1];
HChar** envp = (HChar**)&pArgc[1+argc+1];
// See _start_in_C_linux
INNER_REQUEST
((void) VALGRIND_STACK_REGISTER
(&VG_(interim_stack).bytes[0],
&VG_(interim_stack).bytes[0] + sizeof(VG_(interim_stack))));
VG_(memset)( &the_iicii, 0, sizeof(the_iicii) );
VG_(memset)( &the_iifii, 0, sizeof(the_iifii) );
the_iicii.sp_at_startup = (Addr)pArgc;
r = valgrind_main( (Int)argc, argv, envp );
/* NOTREACHED */
VG_(exit)(r);
}
/*====================================================================*/
/*=== Getting to main() alive: Solaris ===*/
/*====================================================================*/
#elif defined(VGO_solaris)
#if defined(VGP_x86_solaris)
/* The kernel hands control to _start, which extracts the initial stack
pointer and calls onwards to _start_in_C_solaris. This also switches to
the new stack. */
asm("\n"
"\t.text\n"
"\t.globl _start\n"
"\t.type _start, @function\n"
"_start:\n"
/* Set up the new stack in %eax. */
"\tmovl $vgPlain_interim_stack, %eax\n"
"\taddl $"VG_STRINGIFY(VG_STACK_GUARD_SZB)", %eax\n"
"\taddl $"VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)", %eax\n"
"\tandl $~15, %eax\n"
/* Install it, and collect the original one. */
"\txchgl %eax, %esp\n"
"\tsubl $12, %esp\n" /* Keep stack 16-byte aligned. */
/* Call _start_in_C_solaris, passing it the startup %esp. */
"\tpushl %eax\n"
"\tcall _start_in_C_solaris\n"
/* NOTREACHED */
"\thlt\n"
"\t.previous\n"
);
#elif defined(VGP_amd64_solaris)
asm("\n"
".text\n"
"\t.globl _start\n"
"\t.type _start, @function\n"
"_start:\n"
/* Set up the new stack in %rdi. */
"\tmovq $vgPlain_interim_stack, %rdi\n"
"\taddq $"VG_STRINGIFY(VG_STACK_GUARD_SZB)", %rdi\n"
"\taddq $"VG_STRINGIFY(VG_DEFAULT_STACK_ACTIVE_SZB)", %rdi\n"
"\tandq $~15, %rdi\n"
/* Install it, and collect the original one. */
"\txchgq %rdi, %rsp\n"
/* Call _start_in_C_solaris, passing it the startup %rsp. */
"\tcall _start_in_C_solaris\n"
/* NOTREACHED */
"\thlt\n"
".previous\n"
);
#else
# error "Unknown Solaris platform"
#endif
void *memcpy(void *dest, const void *src, size_t n);
void *memcpy(void *dest, const void *src, size_t n) {
return VG_(memcpy)(dest, src, n);
}
__attribute__ ((used))
void _start_in_C_solaris ( UWord* pArgc );
__attribute__ ((used))
void _start_in_C_solaris ( UWord* pArgc )
{
Int r;
Word argc = pArgc[0];
HChar** argv = (HChar**)&pArgc[1];
HChar** envp = (HChar**)&pArgc[1 + argc + 1];
VG_(memset)( &the_iicii, 0, sizeof(the_iicii) );
VG_(memset)( &the_iifii, 0, sizeof(the_iifii) );
the_iicii.sp_at_startup = (Addr)pArgc;
r = valgrind_main((Int)argc, argv, envp);
/* NOTREACHED */
VG_(exit)(r);
}
#else
# error "Unknown OS"
#endif
Addr VG_(get_initial_client_SP)( void )
{
return the_iifii.initial_client_SP;
}
/*====================================================================*/
/*=== {u,}{div,mod}di3 replacements ===*/
/*====================================================================*/
/* For static linking on x86-darwin, we need to supply our own 64-bit
integer division code, else the link dies thusly:
ld_classic: Undefined symbols:
___udivdi3
___umoddi3
*/
#if defined(VGP_x86_darwin)
/* Routines for doing signed/unsigned 64 x 64 ==> 64 div and mod
(udivdi3, umoddi3, divdi3, moddi3) using only 32 x 32 ==> 32
division. Cobbled together from
http://www.hackersdelight.org/HDcode/divlu.c
http://www.hackersdelight.org/HDcode/divls.c
http://www.hackersdelight.org/HDcode/newCode/divDouble.c
The code from those three files is covered by the following license,
as it appears at:
http://www.hackersdelight.org/permissions.htm
You are free to use, copy, and distribute any of the code on
this web site, whether modified by you or not. You need not give
attribution. This includes the algorithms (some of which appear
in Hacker's Delight), the Hacker's Assistant, and any code
submitted by readers. Submitters implicitly agree to this.
*/
/* Long division, unsigned (64/32 ==> 32).
This procedure performs unsigned "long division" i.e., division of a
64-bit unsigned dividend by a 32-bit unsigned divisor, producing a
32-bit quotient. In the overflow cases (divide by 0, or quotient
exceeds 32 bits), it returns a remainder of 0xFFFFFFFF (an impossible
value).
The dividend is u1 and u0, with u1 being the most significant word.
The divisor is parameter v. The value returned is the quotient.
Max line length is 57, to fit in hacker.book. */
static Int nlz32(UInt x)
{
Int n;
if (x == 0) return(32);
n = 0;
if (x <= 0x0000FFFF) {n = n +16; x = x <<16;}
if (x <= 0x00FFFFFF) {n = n + 8; x = x << 8;}
if (x <= 0x0FFFFFFF) {n = n + 4; x = x << 4;}
if (x <= 0x3FFFFFFF) {n = n + 2; x = x << 2;}
if (x <= 0x7FFFFFFF) {n = n + 1;}
return n;
}
/* 64 x 32 ==> 32 unsigned division, using only 32 x 32 ==> 32
division as a primitive. */
static UInt divlu2(UInt u1, UInt u0, UInt v, UInt *r)
{
const UInt b = 65536; // Number base (16 bits).
UInt un1, un0, // Norm. dividend LSD's.
vn1, vn0, // Norm. divisor digits.
q1, q0, // Quotient digits.
un32, un21, un10, // Dividend digit pairs.
rhat; // A remainder.
Int s; // Shift amount for norm.
if (u1 >= v) { // If overflow, set rem.
if (r != NULL) // to an impossible value,
*r = 0xFFFFFFFF; // and return the largest
return 0xFFFFFFFF;} // possible quotient.
s = nlz32(v); // 0 <= s <= 31.
v = v << s; // Normalize divisor.
vn1 = v >> 16; // Break divisor up into
vn0 = v & 0xFFFF; // two 16-bit digits.
un32 = (u1 << s) | ((u0 >> (32 - s)) & (-s >> 31));
un10 = u0 << s; // Shift dividend left.
un1 = un10 >> 16; // Break right half of
un0 = un10 & 0xFFFF; // dividend into two digits.
q1 = un32/vn1; // Compute the first
rhat = un32 - q1*vn1; // quotient digit, q1.
again1:
if (q1 >= b || q1*vn0 > b*rhat + un1) {
q1 = q1 - 1;
rhat = rhat + vn1;
if (rhat < b) goto again1;}
un21 = un32*b + un1 - q1*v; // Multiply and subtract.
q0 = un21/vn1; // Compute the second
rhat = un21 - q0*vn1; // quotient digit, q0.
again2:
if (q0 >= b || q0*vn0 > b*rhat + un0) {
q0 = q0 - 1;
rhat = rhat + vn1;
if (rhat < b) goto again2;}
if (r != NULL) // If remainder is wanted,
*r = (un21*b + un0 - q0*v) >> s; // return it.
return q1*b + q0;
}
/* 64 x 32 ==> 32 signed division, using only 32 x 32 ==> 32 division
as a primitive. */
static Int divls(Int u1, UInt u0, Int v, Int *r)
{
Int q, uneg, vneg, diff, borrow;
uneg = u1 >> 31; // -1 if u < 0.
if (uneg) { // Compute the absolute
u0 = -u0; // value of the dividend u.
borrow = (u0 != 0);
u1 = -u1 - borrow;}
vneg = v >> 31; // -1 if v < 0.
v = (v ^ vneg) - vneg; // Absolute value of v.
if ((UInt)u1 >= (UInt)v) goto overflow;
q = divlu2(u1, u0, v, (UInt *)r);
diff = uneg ^ vneg; // Negate q if signs of
q = (q ^ diff) - diff; // u and v differed.
if (uneg && r != NULL)
*r = -*r;
if ((diff ^ q) < 0 && q != 0) { // If overflow,
overflow: // set remainder
if (r != NULL) // to an impossible value,
*r = 0x80000000; // and return the largest
q = 0x80000000;} // possible neg. quotient.
return q;
}
/* This file contains a program for doing 64/64 ==> 64 division, on a
machine that does not have that instruction but that does have
instructions for "long division" (64/32 ==> 32). Code for unsigned
division is given first, followed by a simple program for doing the
signed version by using the unsigned version.
These programs are useful in implementing "long long" (64-bit)
arithmetic on a machine that has the long division instruction. It will
work on 64- and 32-bit machines, provided the compiler implements long
long's (64-bit integers). It is desirable that the machine have the
Count Leading Zeros instruction.
In the GNU world, these programs are known as __divdi3 and __udivdi3,
and similar names are used here.
This material is not in HD, but may be in a future edition.
Max line length is 57, to fit in hacker.book. */
static Int nlz64(ULong x)
{
Int n;
if (x == 0) return(64);
n = 0;
if (x <= 0x00000000FFFFFFFFULL) {n = n + 32; x = x << 32;}
if (x <= 0x0000FFFFFFFFFFFFULL) {n = n + 16; x = x << 16;}
if (x <= 0x00FFFFFFFFFFFFFFULL) {n = n + 8; x = x << 8;}
if (x <= 0x0FFFFFFFFFFFFFFFULL) {n = n + 4; x = x << 4;}
if (x <= 0x3FFFFFFFFFFFFFFFULL) {n = n + 2; x = x << 2;}
if (x <= 0x7FFFFFFFFFFFFFFFULL) {n = n + 1;}
return n;
}
// ---------------------------- udivdi3 --------------------------------
/* The variables u0, u1, etc. take on only 32-bit values, but they
are declared long long to avoid some compiler warning messages and to
avoid some unnecessary EXTRs that the compiler would put in, to
convert long longs to ints.
First the procedure takes care of the case in which the divisor is a
32-bit quantity. There are two subcases: (1) If the left half of the
dividend is less than the divisor, one execution of DIVU is all that
is required (overflow is not possible). (2) Otherwise it does two
divisions, using the grade school method, with variables used as
suggested below.
q1 q0
________
v) u1 u0
q1*v
____
k u0 */
/* These macros must be used with arguments of the appropriate type
(unsigned long long for DIVU and long long for DIVS. They are
simulations of the presumed machines ops. I.e., they look at only the
low-order 32 bits of the divisor, they return garbage if the division
overflows, and they return garbage in the high-order half of the
quotient doubleword.
In practice, these would be replaced with uses of the machine's DIVU
and DIVS instructions (e.g., by using the GNU "asm" facility). */
static UInt DIVU ( ULong u, UInt v )
{
UInt uHi = (UInt)(u >> 32);
UInt uLo = (UInt)u;
return divlu2(uHi, uLo, v, NULL);
}
static Int DIVS ( Long u, Int v )
{
Int uHi = (Int)(u >> 32);
UInt uLo = (UInt)u;
return divls(uHi, uLo, v, NULL);
}
/* 64 x 64 ==> 64 unsigned division, using only 32 x 32 ==> 32
division as a primitive. */
static ULong udivdi3(ULong u, ULong v)
{
ULong u0, u1, v1, q0, q1, k, n;
if (v >> 32 == 0) { // If v < 2**32:
if (u >> 32 < v) // If u/v cannot overflow,
return DIVU(u, v) // just do one division.
& 0xFFFFFFFF;
else { // If u/v would overflow:
u1 = u >> 32; // Break u up into two
u0 = u & 0xFFFFFFFF; // halves.
q1 = DIVU(u1, v) // First quotient digit.
& 0xFFFFFFFF;
k = u1 - q1*v; // First remainder, < v.
q0 = DIVU((k << 32) + u0, v) // 2nd quot. digit.
& 0xFFFFFFFF;
return (q1 << 32) + q0;
}
}
// Here v >= 2**32.
n = nlz64(v); // 0 <= n <= 31.
v1 = (v << n) >> 32; // Normalize the divisor
// so its MSB is 1.
u1 = u >> 1; // To ensure no overflow.
q1 = DIVU(u1, v1) // Get quotient from
& 0xFFFFFFFF; // divide unsigned insn.
q0 = (q1 << n) >> 31; // Undo normalization and
// division of u by 2.
if (q0 != 0) // Make q0 correct or
q0 = q0 - 1; // too small by 1.
if ((u - q0*v) >= v)
q0 = q0 + 1; // Now q0 is correct.
return q0;
}
// ----------------------------- divdi3 --------------------------------
/* This routine presumes that smallish cases (those which can be done in
one execution of DIVS) are common. If this is not the case, the test for
this case should be deleted.
Note that the test for when DIVS can be used is not entirely
accurate. For example, DIVS is not used if v = 0xFFFFFFFF8000000,
whereas if could be (if u is sufficiently small in magnitude). */
// ------------------------------ cut ----------------------------------
static ULong my_llabs ( Long x )
{
ULong t = x >> 63;
return (x ^ t) - t;
}
/* 64 x 64 ==> 64 signed division, using only 32 x 32 ==> 32 division
as a primitive. */
static Long divdi3(Long u, Long v)
{
ULong au, av;
Long q, t;
au = my_llabs(u);
av = my_llabs(v);
if (av >> 31 == 0) { // If |v| < 2**31 and
// if (v << 32 >> 32 == v) { // If v is in range and
if (au < av << 31) { // |u|/|v| cannot
q = DIVS(u, v); // overflow, use DIVS.
return (q << 32) >> 32;
}
}
q = udivdi3(au,av); // Invoke udivdi3.
t = (u ^ v) >> 63; // If u, v have different
return (q ^ t) - t; // signs, negate q.
}
// ---------------------------- end cut --------------------------------
ULong __udivdi3 (ULong u, ULong v);
ULong __udivdi3 (ULong u, ULong v)
{
return udivdi3(u,v);
}
Long __divdi3 (Long u, Long v);
Long __divdi3 (Long u, Long v)
{
return divdi3(u,v);
}
ULong __umoddi3 (ULong u, ULong v);
ULong __umoddi3 (ULong u, ULong v)
{
ULong q = __udivdi3(u, v);
ULong r = u - q * v;
return r;
}
Long __moddi3 (Long u, Long v);
Long __moddi3 (Long u, Long v)
{
Long q = __divdi3(u, v);
Long r = u - q * v;
return r;
}
/* ------------------------------------------------
ld_classic: Undefined symbols:
___fixunsdfdi
------------------------------------------------
*/
/* ===-- fixunsdfdi.c - Implement __fixunsdfdi -----------------------------===
*
* The LLVM Compiler Infrastructure
*
* This file is dual licensed under the MIT and the University of Illinois Open
* Source Licenses. See LICENSE.TXT for details.
*
* ===----------------------------------------------------------------------===
*
* This file implements __fixunsdfdi for the compiler_rt library.
*
* ===----------------------------------------------------------------------===
*/
/* As per http://www.gnu.org/licenses/license-list.html#GPLCompatibleLicenses,
the "NCSA/University of Illinois Open Source License" is compatible
with the GPL (both version 2 and 3). What is claimed to be
compatible is this
http://www.opensource.org/licenses/UoI-NCSA.php
and the LLVM documentation at
http://www.llvm.org/docs/DeveloperPolicy.html#license
says all the code in LLVM is available under the University of
Illinois/NCSA Open Source License, at this URL
http://www.opensource.org/licenses/UoI-NCSA.php
viz, the same one that the FSF pages claim is compatible. So I
think it's OK to include it.
*/
/* Returns: convert a to a unsigned long long, rounding toward zero.
* Negative values all become zero.
*/
/* Assumption: double is a IEEE 64 bit floating point type
* du_int is a 64 bit integral type
* value in double is representable in du_int or is negative
* (no range checking performed)
*/
/* seee eeee eeee mmmm mmmm mmmm mmmm mmmm | mmmm mmmm mmmm mmmm mmmm mmmm mmmm mmmm */
typedef unsigned long long du_int;
typedef unsigned su_int;
typedef union
{
du_int all;
struct
{
#if VG_LITTLEENDIAN
su_int low;
su_int high;
#else
su_int high;
su_int low;
#endif /* VG_LITTLEENDIAN */
}s;
} udwords;
typedef union
{
udwords u;
double f;
} double_bits;
du_int __fixunsdfdi(double a);
du_int
__fixunsdfdi(double a)
{
double_bits fb;
fb.f = a;
int e = ((fb.u.s.high & 0x7FF00000) >> 20) - 1023;
if (e < 0 || (fb.u.s.high & 0x80000000))
return 0;
udwords r;
r.s.high = (fb.u.s.high & 0x000FFFFF) | 0x00100000;
r.s.low = fb.u.s.low;
if (e > 52)
r.all <<= (e - 52);
else
r.all >>= (52 - e);
return r.all;
}
#endif
/*====================================================================*/
/*=== Dummy _voucher_mach_msg_set for OSX 10.10 ===*/
/*====================================================================*/
#if defined(VGO_darwin) && DARWIN_VERS >= DARWIN_10_10
/* Builds on MacOSX 10.10+ seem to need this for some reason. */
/* extern boolean_t voucher_mach_msg_set(mach_msg_header_t *msg)
__attribute__((weak_import));
I haven't a clue what the return value means, so just return 0.
Looks like none of the generated uses in the tree look at the
return value anyway.
*/
UWord voucher_mach_msg_set ( UWord arg1 );
UWord voucher_mach_msg_set ( UWord arg1 )
{
return 0;
}
#endif
/*--------------------------------------------------------------------*/
/*--- end ---*/
/*--------------------------------------------------------------------*/