Hush Full Node software. We were censored from Github, this is where all development happens now. https://hush.is
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 

615 lines
16 KiB

// Copyright (c) 2009-2010 Satoshi Nakamoto
// Distributed under the MIT/X11 software license, see the accompanying
// file license.txt or http://www.opensource.org/licenses/mit-license.php.
#include "headers.h"
map<string, string> mapArgs;
map<string, vector<string> > mapMultiArgs;
bool fDebug = false;
bool fPrintToConsole = false;
bool fPrintToDebugger = false;
char pszSetDataDir[MAX_PATH] = "";
bool fShutdown = false;
bool fDaemon = false;
// Init openssl library multithreading support
static wxMutex** ppmutexOpenSSL;
void locking_callback(int mode, int i, const char* file, int line)
{
if (mode & CRYPTO_LOCK)
ppmutexOpenSSL[i]->Lock();
else
ppmutexOpenSSL[i]->Unlock();
}
// Init
class CInit
{
public:
CInit()
{
// Init openssl library multithreading support
ppmutexOpenSSL = (wxMutex**)OPENSSL_malloc(CRYPTO_num_locks() * sizeof(wxMutex*));
for (int i = 0; i < CRYPTO_num_locks(); i++)
ppmutexOpenSSL[i] = new wxMutex();
CRYPTO_set_locking_callback(locking_callback);
#ifdef __WXMSW__
// Seed random number generator with screen scrape and other hardware sources
RAND_screen();
#endif
// Seed random number generator with performance counter
RandAddSeed();
}
~CInit()
{
// Shutdown openssl library multithreading support
CRYPTO_set_locking_callback(NULL);
for (int i = 0; i < CRYPTO_num_locks(); i++)
delete ppmutexOpenSSL[i];
OPENSSL_free(ppmutexOpenSSL);
}
}
instance_of_cinit;
void RandAddSeed()
{
// Seed with CPU performance counter
int64 nCounter = PerformanceCounter();
RAND_add(&nCounter, sizeof(nCounter), 1.5);
memset(&nCounter, 0, sizeof(nCounter));
}
void RandAddSeedPerfmon()
{
RandAddSeed();
// This can take up to 2 seconds, so only do it every 10 minutes
static int64 nLastPerfmon;
if (GetTime() < nLastPerfmon + 10 * 60)
return;
nLastPerfmon = GetTime();
#ifdef __WXMSW__
// Don't need this on Linux, OpenSSL automatically uses /dev/urandom
// Seed with the entire set of perfmon data
unsigned char pdata[250000];
memset(pdata, 0, sizeof(pdata));
unsigned long nSize = sizeof(pdata);
long ret = RegQueryValueEx(HKEY_PERFORMANCE_DATA, "Global", NULL, NULL, pdata, &nSize);
RegCloseKey(HKEY_PERFORMANCE_DATA);
if (ret == ERROR_SUCCESS)
{
uint256 hash;
SHA256(pdata, nSize, (unsigned char*)&hash);
RAND_add(&hash, sizeof(hash), min(nSize/500.0, (double)sizeof(hash)));
hash = 0;
memset(pdata, 0, nSize);
printf("%s RandAddSeed() %d bytes\n", DateTimeStrFormat("%x %H:%M:%S", GetTime()).c_str(), nSize);
}
#else
printf("%s RandAddSeed()\n", DateTimeStrFormat("%x %H:%M:%S", GetTime()).c_str());
#endif
}
uint64 GetRand(uint64 nMax)
{
if (nMax == 0)
return 0;
// The range of the random source must be a multiple of the modulus
// to give every possible output value an equal possibility
uint64 nRange = (UINT64_MAX / nMax) * nMax;
uint64 nRand = 0;
do
RAND_bytes((unsigned char*)&nRand, sizeof(nRand));
while (nRand >= nRange);
return (nRand % nMax);
}
inline int OutputDebugStringF(const char* pszFormat, ...)
{
int ret = 0;
if (fPrintToConsole || wxTheApp == NULL)
{
// print to console
va_list arg_ptr;
va_start(arg_ptr, pszFormat);
ret = vprintf(pszFormat, arg_ptr);
va_end(arg_ptr);
}
else
{
// print to debug.log
char pszFile[MAX_PATH+100];
GetDataDir(pszFile);
strlcat(pszFile, "/debug.log", sizeof(pszFile));
FILE* fileout = fopen(pszFile, "a");
if (fileout)
{
//// Debug print useful for profiling
//fprintf(fileout, " %"PRI64d" ", wxGetLocalTimeMillis().GetValue());
va_list arg_ptr;
va_start(arg_ptr, pszFormat);
ret = vfprintf(fileout, pszFormat, arg_ptr);
va_end(arg_ptr);
fclose(fileout);
}
}
#ifdef __WXMSW__
if (fPrintToDebugger)
{
// accumulate a line at a time
static CCriticalSection cs_OutputDebugStringF;
CRITICAL_BLOCK(cs_OutputDebugStringF)
{
static char pszBuffer[50000];
static char* pend;
if (pend == NULL)
pend = pszBuffer;
va_list arg_ptr;
va_start(arg_ptr, pszFormat);
int limit = END(pszBuffer) - pend - 2;
int ret = _vsnprintf(pend, limit, pszFormat, arg_ptr);
va_end(arg_ptr);
if (ret < 0 || ret >= limit)
{
pend = END(pszBuffer) - 2;
*pend++ = '\n';
}
else
pend += ret;
*pend = '\0';
char* p1 = pszBuffer;
char* p2;
while (p2 = strchr(p1, '\n'))
{
p2++;
char c = *p2;
*p2 = '\0';
OutputDebugString(p1);
*p2 = c;
p1 = p2;
}
if (p1 != pszBuffer)
memmove(pszBuffer, p1, pend - p1 + 1);
pend -= (p1 - pszBuffer);
}
}
#endif
return ret;
}
// Safer snprintf
// - prints up to limit-1 characters
// - output string is always null terminated even if limit reached
// - return value is the number of characters actually printed
int my_snprintf(char* buffer, size_t limit, const char* format, ...)
{
if (limit == 0)
return 0;
va_list arg_ptr;
va_start(arg_ptr, format);
int ret = _vsnprintf(buffer, limit, format, arg_ptr);
va_end(arg_ptr);
if (ret < 0 || ret >= limit)
{
ret = limit - 1;
buffer[limit-1] = 0;
}
return ret;
}
string strprintf(const char* format, ...)
{
char buffer[50000];
char* p = buffer;
int limit = sizeof(buffer);
int ret;
loop
{
va_list arg_ptr;
va_start(arg_ptr, format);
ret = _vsnprintf(p, limit, format, arg_ptr);
va_end(arg_ptr);
if (ret >= 0 && ret < limit)
break;
if (p != buffer)
delete p;
limit *= 2;
p = new char[limit];
if (p == NULL)
throw std::bad_alloc();
}
#ifdef _MSC_VER
// msvc optimisation
if (p == buffer)
return string(p, p+ret);
#endif
string str(p, p+ret);
if (p != buffer)
delete p;
return str;
}
bool error(const char* format, ...)
{
char buffer[50000];
int limit = sizeof(buffer);
va_list arg_ptr;
va_start(arg_ptr, format);
int ret = _vsnprintf(buffer, limit, format, arg_ptr);
va_end(arg_ptr);
if (ret < 0 || ret >= limit)
{
ret = limit - 1;
buffer[limit-1] = 0;
}
printf("ERROR: %s\n", buffer);
return false;
}
void ParseString(const string& str, char c, vector<string>& v)
{
unsigned int i1 = 0;
unsigned int i2;
do
{
i2 = str.find(c, i1);
v.push_back(str.substr(i1, i2-i1));
i1 = i2+1;
}
while (i2 != str.npos);
}
string FormatMoney(int64 n, bool fPlus)
{
n /= CENT;
string str = strprintf("%"PRI64d".%02"PRI64d, (n > 0 ? n : -n)/100, (n > 0 ? n : -n)%100);
for (int i = 6; i < str.size(); i += 4)
if (isdigit(str[str.size() - i - 1]))
str.insert(str.size() - i, 1, ',');
if (n < 0)
str.insert((unsigned int)0, 1, '-');
else if (fPlus && n > 0)
str.insert((unsigned int)0, 1, '+');
return str;
}
bool ParseMoney(const char* pszIn, int64& nRet)
{
string strWhole;
int64 nCents = 0;
const char* p = pszIn;
while (isspace(*p))
p++;
for (; *p; p++)
{
if (*p == ',' && p > pszIn && isdigit(p[-1]) && isdigit(p[1]) && isdigit(p[2]) && isdigit(p[3]) && !isdigit(p[4]))
continue;
if (*p == '.')
{
p++;
if (isdigit(*p))
{
nCents = 10 * (*p++ - '0');
if (isdigit(*p))
nCents += (*p++ - '0');
}
break;
}
if (isspace(*p))
break;
if (!isdigit(*p))
return false;
strWhole.insert(strWhole.end(), *p);
}
for (; *p; p++)
if (!isspace(*p))
return false;
if (strWhole.size() > 14)
return false;
if (nCents < 0 || nCents > 99)
return false;
int64 nWhole = atoi64(strWhole);
int64 nPreValue = nWhole * 100 + nCents;
int64 nValue = nPreValue * CENT;
if (nValue / CENT != nPreValue)
return false;
if (nValue / COIN != nWhole)
return false;
nRet = nValue;
return true;
}
vector<unsigned char> ParseHex(const char* psz)
{
vector<unsigned char> vch;
while (isspace(*psz))
psz++;
vch.reserve((strlen(psz)+1)/3);
static char phexdigit[256] =
{ -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
0,1,2,3,4,5,6,7,8,9,-1,-1,-1,-1,-1,-1,
-1,0xa,0xb,0xc,0xd,0xe,0xf,-1,-1,-1,-1,-1,-1,-1,-1,-1,
-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
-1,0xa,0xb,0xc,0xd,0xe,0xf,-1,-1,-1,-1,-1,-1,-1,-1,-1
-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1, };
while (*psz)
{
char c = phexdigit[(unsigned char)*psz++];
if (c == -1)
break;
unsigned char n = (c << 4);
if (*psz)
{
char c = phexdigit[(unsigned char)*psz++];
if (c == -1)
break;
n |= c;
vch.push_back(n);
}
while (isspace(*psz))
psz++;
}
return vch;
}
vector<unsigned char> ParseHex(const std::string& str)
{
return ParseHex(str.c_str());
}
void ParseParameters(int argc, char* argv[])
{
mapArgs.clear();
mapMultiArgs.clear();
for (int i = 0; i < argc; i++)
{
char psz[10000];
strlcpy(psz, argv[i], sizeof(psz));
char* pszValue = (char*)"";
if (strchr(psz, '='))
{
pszValue = strchr(psz, '=');
*pszValue++ = '\0';
}
#ifdef __WXMSW__
_strlwr(psz);
if (psz[0] == '/')
psz[0] = '-';
#endif
mapArgs[psz] = pszValue;
mapMultiArgs[psz].push_back(pszValue);
}
}
void FormatException(char* pszMessage, std::exception* pex, const char* pszThread)
{
#ifdef __WXMSW__
char pszModule[MAX_PATH];
pszModule[0] = '\0';
GetModuleFileName(NULL, pszModule, sizeof(pszModule));
#else
// might not be thread safe, uses wxString
//const char* pszModule = wxStandardPaths::Get().GetExecutablePath().mb_str();
const char* pszModule = "bitcoin";
#endif
if (pex)
snprintf(pszMessage, 1000,
"EXCEPTION: %s \n%s \n%s in %s \n", typeid(*pex).name(), pex->what(), pszModule, pszThread);
else
snprintf(pszMessage, 1000,
"UNKNOWN EXCEPTION \n%s in %s \n", pszModule, pszThread);
}
void LogException(std::exception* pex, const char* pszThread)
{
char pszMessage[1000];
FormatException(pszMessage, pex, pszThread);
printf("\n%s", pszMessage);
}
void PrintException(std::exception* pex, const char* pszThread)
{
char pszMessage[1000];
FormatException(pszMessage, pex, pszThread);
printf("\n\n************************\n%s\n", pszMessage);
if (wxTheApp)
wxMessageBox(pszMessage, "Error", wxOK | wxICON_ERROR);
throw;
//DebugBreak();
}
void GetDataDir(char* pszDir)
{
// pszDir must be at least MAX_PATH length.
if (pszSetDataDir[0] != 0)
{
strlcpy(pszDir, pszSetDataDir, MAX_PATH);
static bool fMkdirDone;
if (!fMkdirDone)
{
fMkdirDone = true;
_mkdir(pszDir);
}
}
else
{
// This can be called during exceptions by printf, so we cache the
// value so we don't have to do memory allocations after that.
// wxStandardPaths::GetUserDataDir
// Return the directory for the user-dependent application data files:
// Unix: ~/.appname
// Windows: C:\Documents and Settings\username\Application Data\appname
// Mac: ~/Library/Application Support/appname
static char pszCachedDir[MAX_PATH];
if (pszCachedDir[0] == 0)
{
strlcpy(pszCachedDir, wxStandardPaths::Get().GetUserDataDir().c_str(), sizeof(pszCachedDir));
_mkdir(pszCachedDir);
}
strlcpy(pszDir, pszCachedDir, MAX_PATH);
}
}
string GetDataDir()
{
char pszDir[MAX_PATH];
GetDataDir(pszDir);
return pszDir;
}
int GetFilesize(FILE* file)
{
int nSavePos = ftell(file);
int nFilesize = -1;
if (fseek(file, 0, SEEK_END) == 0)
nFilesize = ftell(file);
fseek(file, nSavePos, SEEK_SET);
return nFilesize;
}
void ShrinkDebugFile()
{
// Scroll debug.log if it's getting too big
string strFile = GetDataDir() + "/debug.log";
FILE* file = fopen(strFile.c_str(), "r");
if (file && GetFilesize(file) > 10 * 1000000)
{
// Restart the file with some of the end
char pch[200000];
fseek(file, -sizeof(pch), SEEK_END);
int nBytes = fread(pch, 1, sizeof(pch), file);
fclose(file);
if (file = fopen(strFile.c_str(), "w"))
{
fwrite(pch, 1, nBytes, file);
fclose(file);
}
}
}
//
// "Never go to sea with two chronometers; take one or three."
// Our three chronometers are:
// - System clock
// - Median of other server's clocks
// - NTP servers
//
// note: NTP isn't implemented yet, so until then we just use the median
// of other nodes clocks to correct ours.
//
int64 GetTime()
{
return time(NULL);
}
static int64 nTimeOffset = 0;
int64 GetAdjustedTime()
{
return GetTime() + nTimeOffset;
}
void AddTimeData(unsigned int ip, int64 nTime)
{
int64 nOffsetSample = nTime - GetTime();
// Ignore duplicates
static set<unsigned int> setKnown;
if (!setKnown.insert(ip).second)
return;
// Add data
static vector<int64> vTimeOffsets;
if (vTimeOffsets.empty())
vTimeOffsets.push_back(0);
vTimeOffsets.push_back(nOffsetSample);
printf("Added time data, samples %d, offset %+"PRI64d" (%+"PRI64d" minutes)\n", vTimeOffsets.size(), vTimeOffsets.back(), vTimeOffsets.back()/60);
if (vTimeOffsets.size() >= 5 && vTimeOffsets.size() % 2 == 1)
{
sort(vTimeOffsets.begin(), vTimeOffsets.end());
int64 nMedian = vTimeOffsets[vTimeOffsets.size()/2];
nTimeOffset = nMedian;
if ((nMedian > 0 ? nMedian : -nMedian) > 5 * 60)
{
// Only let other nodes change our clock so far before we
// go to the NTP servers
/// todo: Get time from NTP servers, then set a flag
/// to make sure it doesn't get changed again
}
foreach(int64 n, vTimeOffsets)
printf("%+"PRI64d" ", n);
printf("| nTimeOffset = %+"PRI64d" (%+"PRI64d" minutes)\n", nTimeOffset, nTimeOffset/60);
}
}