random: get rid of secondary crngs

As the comment said, this is indeed a "hack". Since it was introduced,
it's been a constant state machine nightmare, with lots of subtle early
boot issues and a wildly complex set of machinery to keep everything in
sync. Rather than continuing to play whack-a-mole with this approach,
this commit simply removes it entirely. This commit is preparation for
"random: use simpler fast key erasure flow on per-cpu keys" in this
series, which introduces a simpler (and faster) mechanism to accomplish
the same thing.

Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This commit is contained in:
Jason A. Donenfeld 2022-02-06 23:51:41 +01:00
parent 28f425e573
commit a9412d510a

View File

@ -323,14 +323,11 @@ static struct crng_state primary_crng = {
* its value (from 0->1->2). * its value (from 0->1->2).
*/ */
static int crng_init = 0; static int crng_init = 0;
static bool crng_need_final_init = false;
#define crng_ready() (likely(crng_init > 1)) #define crng_ready() (likely(crng_init > 1))
static int crng_init_cnt = 0; static int crng_init_cnt = 0;
static unsigned long crng_global_init_time = 0;
#define CRNG_INIT_CNT_THRESH (2 * CHACHA_KEY_SIZE) #define CRNG_INIT_CNT_THRESH (2 * CHACHA_KEY_SIZE)
static void _extract_crng(struct crng_state *crng, u8 out[CHACHA_BLOCK_SIZE]); static void extract_crng(u8 out[CHACHA_BLOCK_SIZE]);
static void _crng_backtrack_protect(struct crng_state *crng, static void crng_backtrack_protect(u8 tmp[CHACHA_BLOCK_SIZE], int used);
u8 tmp[CHACHA_BLOCK_SIZE], int used);
static void process_random_ready_list(void); static void process_random_ready_list(void);
static void _get_random_bytes(void *buf, int nbytes); static void _get_random_bytes(void *buf, int nbytes);
@ -365,7 +362,7 @@ static struct {
static void extract_entropy(void *buf, size_t nbytes); static void extract_entropy(void *buf, size_t nbytes);
static void crng_reseed(struct crng_state *crng); static void crng_reseed(void);
/* /*
* This function adds bytes into the entropy "pool". It does not * This function adds bytes into the entropy "pool". It does not
@ -464,7 +461,7 @@ static void credit_entropy_bits(int nbits)
trace_credit_entropy_bits(nbits, entropy_count, _RET_IP_); trace_credit_entropy_bits(nbits, entropy_count, _RET_IP_);
if (crng_init < 2 && entropy_count >= POOL_MIN_BITS) if (crng_init < 2 && entropy_count >= POOL_MIN_BITS)
crng_reseed(&primary_crng); crng_reseed();
} }
/********************************************************************* /*********************************************************************
@ -477,16 +474,7 @@ static void credit_entropy_bits(int nbits)
static DECLARE_WAIT_QUEUE_HEAD(crng_init_wait); static DECLARE_WAIT_QUEUE_HEAD(crng_init_wait);
/*
* Hack to deal with crazy userspace progams when they are all trying
* to access /dev/urandom in parallel. The programs are almost
* certainly doing something terribly wrong, but we'll work around
* their brain damage.
*/
static struct crng_state **crng_node_pool __read_mostly;
static void invalidate_batched_entropy(void); static void invalidate_batched_entropy(void);
static void numa_crng_init(void);
static bool trust_cpu __ro_after_init = IS_ENABLED(CONFIG_RANDOM_TRUST_CPU); static bool trust_cpu __ro_after_init = IS_ENABLED(CONFIG_RANDOM_TRUST_CPU);
static int __init parse_trust_cpu(char *arg) static int __init parse_trust_cpu(char *arg)
@ -495,24 +483,6 @@ static int __init parse_trust_cpu(char *arg)
} }
early_param("random.trust_cpu", parse_trust_cpu); early_param("random.trust_cpu", parse_trust_cpu);
static bool crng_init_try_arch(struct crng_state *crng)
{
int i;
bool arch_init = true;
unsigned long rv;
for (i = 4; i < 16; i++) {
if (!arch_get_random_seed_long(&rv) &&
!arch_get_random_long(&rv)) {
rv = random_get_entropy();
arch_init = false;
}
crng->state[i] ^= rv;
}
return arch_init;
}
static bool __init crng_init_try_arch_early(void) static bool __init crng_init_try_arch_early(void)
{ {
int i; int i;
@ -531,100 +501,17 @@ static bool __init crng_init_try_arch_early(void)
return arch_init; return arch_init;
} }
static void crng_initialize_secondary(struct crng_state *crng) static void __init crng_initialize(void)
{
chacha_init_consts(crng->state);
_get_random_bytes(&crng->state[4], sizeof(u32) * 12);
crng_init_try_arch(crng);
crng->init_time = jiffies - CRNG_RESEED_INTERVAL - 1;
}
static void __init crng_initialize_primary(void)
{ {
extract_entropy(&primary_crng.state[4], sizeof(u32) * 12); extract_entropy(&primary_crng.state[4], sizeof(u32) * 12);
if (crng_init_try_arch_early() && trust_cpu && crng_init < 2) { if (crng_init_try_arch_early() && trust_cpu && crng_init < 2) {
invalidate_batched_entropy(); invalidate_batched_entropy();
numa_crng_init();
crng_init = 2; crng_init = 2;
pr_notice("crng init done (trusting CPU's manufacturer)\n"); pr_notice("crng init done (trusting CPU's manufacturer)\n");
} }
primary_crng.init_time = jiffies - CRNG_RESEED_INTERVAL - 1; primary_crng.init_time = jiffies - CRNG_RESEED_INTERVAL - 1;
} }
static void crng_finalize_init(void)
{
if (!system_wq) {
/* We can't call numa_crng_init until we have workqueues,
* so mark this for processing later. */
crng_need_final_init = true;
return;
}
invalidate_batched_entropy();
numa_crng_init();
crng_init = 2;
crng_need_final_init = false;
process_random_ready_list();
wake_up_interruptible(&crng_init_wait);
kill_fasync(&fasync, SIGIO, POLL_IN);
pr_notice("crng init done\n");
if (unseeded_warning.missed) {
pr_notice("%d get_random_xx warning(s) missed due to ratelimiting\n",
unseeded_warning.missed);
unseeded_warning.missed = 0;
}
if (urandom_warning.missed) {
pr_notice("%d urandom warning(s) missed due to ratelimiting\n",
urandom_warning.missed);
urandom_warning.missed = 0;
}
}
static void do_numa_crng_init(struct work_struct *work)
{
int i;
struct crng_state *crng;
struct crng_state **pool;
pool = kcalloc(nr_node_ids, sizeof(*pool), GFP_KERNEL | __GFP_NOFAIL);
for_each_online_node(i) {
crng = kmalloc_node(sizeof(struct crng_state),
GFP_KERNEL | __GFP_NOFAIL, i);
spin_lock_init(&crng->lock);
crng_initialize_secondary(crng);
pool[i] = crng;
}
/* pairs with READ_ONCE() in select_crng() */
if (cmpxchg_release(&crng_node_pool, NULL, pool) != NULL) {
for_each_node(i)
kfree(pool[i]);
kfree(pool);
}
}
static DECLARE_WORK(numa_crng_init_work, do_numa_crng_init);
static void numa_crng_init(void)
{
if (IS_ENABLED(CONFIG_NUMA))
schedule_work(&numa_crng_init_work);
}
static struct crng_state *select_crng(void)
{
if (IS_ENABLED(CONFIG_NUMA)) {
struct crng_state **pool;
int nid = numa_node_id();
/* pairs with cmpxchg_release() in do_numa_crng_init() */
pool = READ_ONCE(crng_node_pool);
if (pool && pool[nid])
return pool[nid];
}
return &primary_crng;
}
/* /*
* crng_fast_load() can be called by code in the interrupt service * crng_fast_load() can be called by code in the interrupt service
* path. So we can't afford to dilly-dally. Returns the number of * path. So we can't afford to dilly-dally. Returns the number of
@ -702,17 +589,15 @@ static int crng_slow_load(const u8 *cp, size_t len)
return 1; return 1;
} }
static void crng_reseed(struct crng_state *crng) static void crng_reseed(void)
{ {
unsigned long flags; unsigned long flags;
int i; int i, entropy_count;
union { union {
u8 block[CHACHA_BLOCK_SIZE]; u8 block[CHACHA_BLOCK_SIZE];
u32 key[8]; u32 key[8];
} buf; } buf;
if (crng == &primary_crng) {
int entropy_count;
do { do {
entropy_count = READ_ONCE(input_pool.entropy_count); entropy_count = READ_ONCE(input_pool.entropy_count);
if (entropy_count < POOL_MIN_BITS) if (entropy_count < POOL_MIN_BITS)
@ -721,49 +606,54 @@ static void crng_reseed(struct crng_state *crng)
extract_entropy(buf.key, sizeof(buf.key)); extract_entropy(buf.key, sizeof(buf.key));
wake_up_interruptible(&random_write_wait); wake_up_interruptible(&random_write_wait);
kill_fasync(&fasync, SIGIO, POLL_OUT); kill_fasync(&fasync, SIGIO, POLL_OUT);
} else {
_extract_crng(&primary_crng, buf.block); spin_lock_irqsave(&primary_crng.lock, flags);
_crng_backtrack_protect(&primary_crng, buf.block,
CHACHA_KEY_SIZE);
}
spin_lock_irqsave(&crng->lock, flags);
for (i = 0; i < 8; i++) for (i = 0; i < 8; i++)
crng->state[i + 4] ^= buf.key[i]; primary_crng.state[i + 4] ^= buf.key[i];
memzero_explicit(&buf, sizeof(buf)); memzero_explicit(&buf, sizeof(buf));
WRITE_ONCE(crng->init_time, jiffies); WRITE_ONCE(primary_crng.init_time, jiffies);
spin_unlock_irqrestore(&crng->lock, flags); spin_unlock_irqrestore(&primary_crng.lock, flags);
if (crng == &primary_crng && crng_init < 2) if (crng_init < 2) {
crng_finalize_init(); invalidate_batched_entropy();
} crng_init = 2;
process_random_ready_list();
static void _extract_crng(struct crng_state *crng, u8 out[CHACHA_BLOCK_SIZE]) wake_up_interruptible(&crng_init_wait);
{ kill_fasync(&fasync, SIGIO, POLL_IN);
unsigned long flags, init_time; pr_notice("crng init done\n");
if (unseeded_warning.missed) {
if (crng_ready()) { pr_notice("%d get_random_xx warning(s) missed due to ratelimiting\n",
init_time = READ_ONCE(crng->init_time); unseeded_warning.missed);
if (time_after(READ_ONCE(crng_global_init_time), init_time) || unseeded_warning.missed = 0;
time_after(jiffies, init_time + CRNG_RESEED_INTERVAL)) }
crng_reseed(crng); if (urandom_warning.missed) {
pr_notice("%d urandom warning(s) missed due to ratelimiting\n",
urandom_warning.missed);
urandom_warning.missed = 0;
}
} }
spin_lock_irqsave(&crng->lock, flags);
chacha20_block(&crng->state[0], out);
if (crng->state[12] == 0)
crng->state[13]++;
spin_unlock_irqrestore(&crng->lock, flags);
} }
static void extract_crng(u8 out[CHACHA_BLOCK_SIZE]) static void extract_crng(u8 out[CHACHA_BLOCK_SIZE])
{ {
_extract_crng(select_crng(), out); unsigned long flags, init_time;
if (crng_ready()) {
init_time = READ_ONCE(primary_crng.init_time);
if (time_after(jiffies, init_time + CRNG_RESEED_INTERVAL))
crng_reseed();
}
spin_lock_irqsave(&primary_crng.lock, flags);
chacha20_block(&primary_crng.state[0], out);
if (primary_crng.state[12] == 0)
primary_crng.state[13]++;
spin_unlock_irqrestore(&primary_crng.lock, flags);
} }
/* /*
* Use the leftover bytes from the CRNG block output (if there is * Use the leftover bytes from the CRNG block output (if there is
* enough) to mutate the CRNG key to provide backtracking protection. * enough) to mutate the CRNG key to provide backtracking protection.
*/ */
static void _crng_backtrack_protect(struct crng_state *crng, static void crng_backtrack_protect(u8 tmp[CHACHA_BLOCK_SIZE], int used)
u8 tmp[CHACHA_BLOCK_SIZE], int used)
{ {
unsigned long flags; unsigned long flags;
u32 *s, *d; u32 *s, *d;
@ -774,17 +664,12 @@ static void _crng_backtrack_protect(struct crng_state *crng,
extract_crng(tmp); extract_crng(tmp);
used = 0; used = 0;
} }
spin_lock_irqsave(&crng->lock, flags); spin_lock_irqsave(&primary_crng.lock, flags);
s = (u32 *)&tmp[used]; s = (u32 *)&tmp[used];
d = &crng->state[4]; d = &primary_crng.state[4];
for (i = 0; i < 8; i++) for (i = 0; i < 8; i++)
*d++ ^= *s++; *d++ ^= *s++;
spin_unlock_irqrestore(&crng->lock, flags); spin_unlock_irqrestore(&primary_crng.lock, flags);
}
static void crng_backtrack_protect(u8 tmp[CHACHA_BLOCK_SIZE], int used)
{
_crng_backtrack_protect(select_crng(), tmp, used);
} }
static ssize_t extract_crng_user(void __user *buf, size_t nbytes) static ssize_t extract_crng_user(void __user *buf, size_t nbytes)
@ -1371,10 +1256,7 @@ static void __init init_std_data(void)
int __init rand_initialize(void) int __init rand_initialize(void)
{ {
init_std_data(); init_std_data();
if (crng_need_final_init) crng_initialize();
crng_finalize_init();
crng_initialize_primary();
crng_global_init_time = jiffies;
if (ratelimit_disable) { if (ratelimit_disable) {
urandom_warning.interval = 0; urandom_warning.interval = 0;
unseeded_warning.interval = 0; unseeded_warning.interval = 0;
@ -1544,8 +1426,7 @@ static long random_ioctl(struct file *f, unsigned int cmd, unsigned long arg)
return -EPERM; return -EPERM;
if (crng_init < 2) if (crng_init < 2)
return -ENODATA; return -ENODATA;
crng_reseed(&primary_crng); crng_reseed();
WRITE_ONCE(crng_global_init_time, jiffies - 1);
return 0; return 0;
default: default:
return -EINVAL; return -EINVAL;