Make improvements

- Make rand64() thread safe
- Introduce lemur64 lcg prng
- Improve strace on New Technology
- Improve msync() on New Technology
This commit is contained in:
Justine Tunney 2022-04-07 00:15:35 -07:00
parent 43ba3009b2
commit 29bf8b1a30
73 changed files with 888 additions and 269 deletions

View file

@ -16,59 +16,80 @@
TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
PERFORMANCE OF THIS SOFTWARE.
*/
#include "libc/assert.h"
#include "libc/bits/bits.h"
#include "libc/bits/xadd.h"
#include "libc/bits/lockcmpxchg16b.internal.h"
#include "libc/bits/lockxadd.internal.h"
#include "libc/bits/weaken.h"
#include "libc/calls/calls.h"
#include "libc/intrin/kprintf.h"
#include "libc/nexgen32e/rdtsc.h"
#include "libc/nexgen32e/x86feature.h"
#include "libc/nt/thread.h"
#include "libc/rand/rand.h"
#include "libc/runtime/runtime.h"
#include "libc/sysv/consts/auxv.h"
#include "libc/thread/create.h"
static uint64_t thepool;
extern int __pid;
static int thepid;
static int thecount;
static uint128_t thepool;
/**
* Returns nondeterministic random data.
*
* This function automatically seeds itself on startup and reseeds
* itself after fork() and vfork(). It takes about nanosecond to run.
* That makes it much slower than vigna() and rand() but much faster
* than rdrand() and rdseed().
* This function is similar to lemur64() except it'thepool intended to
* be unpredictable. This PRNG automatically seeds itself on startup
* using a much stronger and faster random source than `srand(time(0))`.
* This function will automatically reseed itself when new processes and
* threads are spawned. This function is thread safe in the sense that a
* race condition can't happen where two threads return the same result.
*
* @see rdseed(), rdrand(), rand(), random(), rngset()
* @note based on vigna's algorithm
* @note this function is not intended for cryptography
* @note this function passes bigcrush and practrand
* @note this function takes at minimum 30 cycles
* @asyncsignalsafe
* @threadsafe
* @vforksafe
*/
uint64_t rand64(void) {
bool cf;
register uint64_t t;
if (X86_HAVE(RDSEED)) {
asm volatile(CFLAG_ASM("rdseed\t%1")
: CFLAG_CONSTRAINT(cf), "=r"(t)
: /* no inputs */
: "cc");
if (cf) {
thepool ^= t;
return t;
void *d;
int c1, p1, p2;
uint128_t s1, s2;
do {
p1 = __pid;
p2 = thepid;
c1 = thecount;
asm volatile("" ::: "memory");
s1 = thepool;
if (p1 == p2) {
// fast path
s2 = s1;
} else {
// slow path
if (!p2) {
// first call so get some cheap entropy
if ((d = (void *)getauxval(AT_RANDOM))) {
memcpy(&s2, d, 16); // kernel entropy
} else {
s2 = kStartTsc; // rdtsc() @ _start()
}
} else {
// process contention so blend a timestamp
s2 = s1 ^ rdtsc();
}
// toss the new pid in there
s2 ^= p1;
// ordering for thepid probably doesn't matter
thepid = p1;
}
}
t = _xadd(&thepool, 0x9e3779b97f4a7c15);
t ^= (getpid() * 0x1001111111110001ull + 0xdeaadead) >> 31;
t = (t ^ (t >> 30)) * 0xbf58476d1ce4e5b9;
t = (t ^ (t >> 27)) * 0x94d049bb133111eb;
return t ^ (t >> 31);
// lemur64 pseudorandom number generator
s2 *= 15750249268501108917ull;
// sadly 128-bit values aren't atomic on x86
_lockcmpxchg16b(&thepool, &s1, s2);
// do it again if there's thread contention
} while (_lockxadd(&thecount, 1) != c1);
// the most important step in the prng
return s2 >> 64;
}
static textstartup void rand64_init(int argc, char **argv, char **envp,
intptr_t *auxv) {
for (; auxv[0]; auxv += 2) {
if (auxv[0] == AT_RANDOM) {
thepool = READ64LE((const char *)auxv[1]);
return;
}
}
thepool = kStartTsc;
}
const void *const g_rand64_init[] initarray = {rand64_init};