The register_random_ready_notifier() notifier is somewhat complicated,
and was already recently rewritten to use notifier blocks. It is only
used now by one consumer in the kernel, vsprintf.c, for which the async
mechanism is really overly complex for what it actually needs. This
commit removes register_random_ready_notifier() and unregister_random_
ready_notifier(), because it just adds complication with little utility,
and changes vsprintf.c to just check on `!rng_is_initialized() &&
!rng_has_arch_random()`, which will eventually be true. Performance-
wise, that code was already using a static branch, so there's basically
no overhead at all to this change.
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Petr Mladek <pmladek@suse.com> # for vsprintf.c
Reviewed-by: Petr Mladek <pmladek@suse.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The RNG incorporates RDRAND into its state at boot and every time it
reseeds, so there's no reason for callers to use it directly. The
hashing that the RNG does on it is preferable to using the bytes raw.
The only current use case of get_random_bytes_arch() is vsprintf's
siphash key for pointer hashing, which uses it to initialize the pointer
secret earlier than usual if RDRAND is available. In order to replace
this narrow use case, just expose whether RDRAND is mixed into the RNG,
with a new function called rng_has_arch_random(). With that taken care
of, there are no users of get_random_bytes_arch() left, so it can be
removed.
Later, if trust_cpu gets turned on by default (as most distros are
doing), this one use of rng_has_arch_random() can probably go away as
well.
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Sergey Senozhatsky <senozhatsky@chromium.org>
Acked-by: Petr Mladek <pmladek@suse.com> # for vsprintf.c
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Much of random.c is devoted to initializing the rng and accounting for
when a sufficient amount of entropy has been added. In a perfect world,
this would all happen during init, and so we could mark these functions
as __init. But in reality, this isn't the case: sometimes the rng only
finishes initializing some seconds after system init is finished.
For this reason, at the moment, a whole host of functions that are only
used relatively close to system init and then never again are intermixed
with functions that are used in hot code all the time. This creates more
cache misses than necessary.
In order to pack the hot code closer together, this commit moves the
initialization functions that can't be marked as __init into
.text.unlikely by way of the __cold attribute.
Of particular note is moving credit_init_bits() into a macro wrapper
that inlines the crng_ready() static branch check. This avoids a
function call to a nop+ret, and most notably prevents extra entropy
arithmetic from being computed in mix_interrupt_randomness().
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The current code was a mix of "nbytes", "count", "size", "buffer", "in",
and so forth. Instead, let's clean this up by naming input parameters
"buf" (or "ubuf") and "len", so that you always understand that you're
reading this variety of function argument.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Since crng_ready() is only false briefly during initialization and then
forever after becomes true, we don't need to evaluate it after, making
it a prime candidate for a static branch.
One complication, however, is that it changes state in a particular call
to credit_init_bits(), which might be made from atomic context, which
means we must kick off a workqueue to change the static key. Further
complicating things, credit_init_bits() may be called sufficiently early
on in system initialization such that system_wq is NULL.
Fortunately, there exists the nice function execute_in_process_context(),
which will immediately execute the function if !in_interrupt(), and
otherwise defer it to a workqueue. During early init, before workqueues
are available, in_interrupt() is always false, because interrupts
haven't even been enabled yet, which means the function in that case
executes immediately. Later on, after workqueues are available,
in_interrupt() might be true, but in that case, the work is queued in
system_wq and all goes well.
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Sultan Alsawaf <sultan@kerneltoast.com>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
RDRAND and RDSEED can fail sometimes, which is fine. We currently
initialize the RNG with 512 bits of RDRAND/RDSEED. We only need 256 bits
of those to succeed in order to initialize the RNG. Instead of the
current "all or nothing" approach, actually credit these contributions
the amount that is actually contributed.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Currently, start_kernel() adds latent entropy and the command line to
the entropy bool *after* the RNG has been initialized, deferring when
it's actually used by things like stack canaries until the next time
the pool is seeded. This surely is not intended.
Rather than splitting up which entropy gets added where and when between
start_kernel() and random_init(), just do everything in random_init(),
which should eliminate these kinds of bugs in the future.
While we're at it, rename the awkwardly titled "rand_initialize()" to
the more standard "random_init()" nomenclature.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This expands to exactly the same code that it replaces, but makes things
consistent by using the same macro for jiffy comparisons throughout.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The CONFIG_WARN_ALL_UNSEEDED_RANDOM debug option controls whether the
kernel warns about all unseeded randomness or just the first instance.
There's some complicated rate limiting and comparison to the previous
caller, such that even with CONFIG_WARN_ALL_UNSEEDED_RANDOM enabled,
developers still don't see all the messages or even an accurate count of
how many were missed. This is the result of basically parallel
mechanisms aimed at accomplishing more or less the same thing, added at
different points in random.c history, which sort of compete with the
first-instance-only limiting we have now.
It turns out, however, that nobody cares about the first unseeded
randomness instance of in-kernel users. The same first user has been
there for ages now, and nobody is doing anything about it. It isn't even
clear that anybody _can_ do anything about it. Most places that can do
something about it have switched over to using get_random_bytes_wait()
or wait_for_random_bytes(), which is the right thing to do, but there is
still much code that needs randomness sometimes during init, and as a
geeneral rule, if you're not using one of the _wait functions or the
readiness notifier callback, you're bound to be doing it wrong just
based on that fact alone.
So warning about this same first user that can't easily change is simply
not an effective mechanism for anything at all. Users can't do anything
about it, as the Kconfig text points out -- the problem isn't in
userspace code -- and kernel developers don't or more often can't react
to it.
Instead, show the warning for all instances when CONFIG_WARN_ALL_UNSEEDED_RANDOM
is set, so that developers can debug things need be, or if it isn't set,
don't show a warning at all.
At the same time, CONFIG_WARN_ALL_UNSEEDED_RANDOM now implies setting
random.ratelimit_disable=1 on by default, since if you care about one
you probably care about the other too. And we can clean up usage around
the related urandom_warning ratelimiter as well (whose behavior isn't
changing), so that it properly counts missed messages after the 10
message threshold is reached.
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Initialization happens once -- by way of credit_init_bits() -- and then
it never happens again. Therefore, it doesn't need to be in
crng_reseed(), which is a hot path that is called multiple times. It
also doesn't make sense to have there, as initialization activity is
better associated with initialization routines.
After the prior commit, crng_reseed() now won't be called by multiple
concurrent callers, which means that we can safely move the
"finialize_init" logic into crng_init_bits() unconditionally.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Since all changes of crng_init now go through credit_init_bits(), we can
fix a long standing race in which two concurrent callers of
credit_init_bits() have the new bit count >= some threshold, but are
doing so with crng_init as a lower threshold, checked outside of a lock,
resulting in crng_reseed() or similar being called twice.
In order to fix this, we can use the original cmpxchg value of the bit
count, and only change crng_init when the bit count transitions from
below a threshold to meeting the threshold.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
crng_init represents a state machine, with three states, and various
rules for transitions. For the longest time, we've been managing these
with "0", "1", and "2", and expecting people to figure it out. To make
the code more obvious, replace these with proper enum values
representing the transition, and then redocument what each of these
states mean.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: Joe Perches <joe@perches.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The SipHash family of permutations is currently used in three places:
- siphash.c itself, used in the ordinary way it was intended.
- random32.c, in a construction from an anonymous contributor.
- random.c, as part of its fast_mix function.
Each one of these places reinvents the wheel with the same C code, same
rotation constants, and same symmetry-breaking constants.
This commit tidies things up a bit by placing macros for the
permutations and constants into siphash.h, where each of the three .c
users can access them. It also leaves a note dissuading more users of
them from emerging.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Now that fast_mix() has more than one caller, gcc no longer inlines it.
That's fine. But it also doesn't handle the compound literal argument we
pass it very efficiently, nor does it handle the loop as well as it
could. So just expand the code to spell out this function so that it
generates the same code as it did before. Performance-wise, this now
behaves as it did before the last commit. The difference in actual code
size on x86 is 45 bytes, which is less than a cache line.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Years ago, a separate fast pool was added for interrupts, so that the
cost associated with taking the input pool spinlocks and mixing into it
would be avoided in places where latency is critical. However, one
oversight was that add_input_randomness() and add_disk_randomness()
still sometimes are called directly from the interrupt handler, rather
than being deferred to a thread. This means that some unlucky interrupts
will be caught doing a blake2s_compress() call and potentially spinning
on input_pool.lock, which can also be taken by unprivileged users by
writing into /dev/urandom.
In order to fix this, add_timer_randomness() now checks whether it is
being called from a hard IRQ and if so, just mixes into the per-cpu IRQ
fast pool using fast_mix(), which is much faster and can be done
lock-free. A nice consequence of this, as well, is that it means hard
IRQ context FPU support is likely no longer useful.
The entropy estimation algorithm used by add_timer_randomness() is also
somewhat different than the one used for add_interrupt_randomness(). The
former looks at deltas of deltas of deltas, while the latter just waits
for 64 interrupts for one bit or for one second since the last bit. In
order to bridge these, and since add_interrupt_randomness() runs after
an add_timer_randomness() that's called from hard IRQ, we add to the
fast pool credit the related amount, and then subtract one to account
for add_interrupt_randomness()'s contribution.
A downside of this, however, is that the num argument is potentially
attacker controlled, which puts a bit more pressure on the fast_mix()
sponge to do more than it's really intended to do. As a mitigating
factor, the first 96 bits of input aren't attacker controlled (a cycle
counter followed by zeros), which means it's essentially two rounds of
siphash rather than one, which is somewhat better. It's also not that
much different from add_interrupt_randomness()'s use of the irq stack
instruction pointer register.
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Filipe Manana <fdmanana@suse.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
There are no code changes here; this is just a reordering of functions,
so that in subsequent commits, the timer entropy functions can call into
the interrupt ones.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Per the thread linked below, "premature next" is not considered to be a
realistic threat model, and leads to more serious security problems.
"Premature next" is the scenario in which:
- Attacker compromises the current state of a fully initialized RNG via
some kind of infoleak.
- New bits of entropy are added directly to the key used to generate the
/dev/urandom stream, without any buffering or pooling.
- Attacker then, somehow having read access to /dev/urandom, samples RNG
output and brute forces the individual new bits that were added.
- Result: the RNG never "recovers" from the initial compromise, a
so-called violation of what academics term "post-compromise security".
The usual solutions to this involve some form of delaying when entropy
gets mixed into the crng. With Fortuna, this involves multiple input
buckets. With what the Linux RNG was trying to do prior, this involves
entropy estimation.
However, by delaying when entropy gets mixed in, it also means that RNG
compromises are extremely dangerous during the window of time before
the RNG has gathered enough entropy, during which time nonces may become
predictable (or repeated), ephemeral keys may not be secret, and so
forth. Moreover, it's unclear how realistic "premature next" is from an
attack perspective, if these attacks even make sense in practice.
Put together -- and discussed in more detail in the thread below --
these constitute grounds for just doing away with the current code that
pretends to handle premature next. I say "pretends" because it wasn't
doing an especially great job at it either; should we change our mind
about this direction, we would probably implement Fortuna to "fix" the
"problem", in which case, removing the pretend solution still makes
sense.
This also reduces the crng reseed period from 5 minutes down to 1
minute. The rationale from the thread might lead us toward reducing that
even further in the future (or even eliminating it), but that remains a
topic of a future commit.
At a high level, this patch changes semantics from:
Before: Seed for the first time after 256 "bits" of estimated
entropy have been accumulated since the system booted. Thereafter,
reseed once every five minutes, but only if 256 new "bits" have been
accumulated since the last reseeding.
After: Seed for the first time after 256 "bits" of estimated entropy
have been accumulated since the system booted. Thereafter, reseed
once every minute.
Most of this patch is renaming and removing: POOL_MIN_BITS becomes
POOL_INIT_BITS, credit_entropy_bits() becomes credit_init_bits(),
crng_reseed() loses its "force" parameter since it's now always true,
the drain_entropy() function no longer has any use so it's removed,
entropy estimation is skipped if we've already init'd, the various
notifiers for "low on entropy" are now only active prior to init, and
finally, some documentation comments are cleaned up here and there.
Link: https://lore.kernel.org/lkml/YmlMGx6+uigkGiZ0@zx2c4.com/
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Nadia Heninger <nadiah@cs.ucsd.edu>
Cc: Tom Ristenpart <ristenpart@cornell.edu>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Before, the first 64 bytes of input, regardless of how entropic it was,
would be used to mutate the crng base key directly, and none of those
bytes would be credited as having entropy. Then 256 bits of credited
input would be accumulated, and only then would the rng transition from
the earlier "fast init" phase into being actually initialized.
The thinking was that by mixing and matching fast init and real init, an
attacker who compromised the fast init state, considered easy to do
given how little entropy might be in those first 64 bytes, would then be
able to bruteforce bits from the actual initialization. By keeping these
separate, bruteforcing became impossible.
However, by not crediting potentially creditable bits from those first 64
bytes of input, we delay initialization, and actually make the problem
worse, because it means the user is drawing worse random numbers for a
longer period of time.
Instead, we can take the first 128 bits as fast init, and allow them to
be credited, and then hold off on the next 128 bits until they've
accumulated. This is still a wide enough margin to prevent bruteforcing
the rng state, while still initializing much faster.
Then, rather than trying to piecemeal inject into the base crng key at
various points, instead just extract from the pool when we need it, for
the crng_init==0 phase. Performance may even be better for the various
inputs here, since there are likely more calls to mix_pool_bytes() then
there are to get_random_bytes() during this phase of system execution.
Since the preinit injection code is gone, bootloader randomness can then
do something significantly more straight forward, removing the weird
system_wq hack in hwgenerator randomness.
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
It's too hard to keep the batches synchronized, and pointless anyway,
since in !crng_ready(), we're updating the base_crng key really often,
where batching only hurts. So instead, if the crng isn't ready, just
call into get_random_bytes(). At this stage nothing is performance
critical anyhow.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Since the RNG loses freshness with system suspend/hibernation, when we
resume, immediately reseed using whatever data we can, which for this
particular case is the various timestamps regarding system suspend time,
in addition to more generally the RDSEED/RDRAND/RDTSC values that happen
whenever the crng reseeds.
On systems that suspend and resume automatically all the time -- such as
Android -- we skip the reseeding on suspend resumption, since that could
wind up being far too busy. This is the same trade-off made in
WireGuard.
In addition to reseeding upon resumption always mix into the pool these
various stamps on every power notification event.
Cc: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Currently, we do the jitter dance if two consecutive reads to the cycle
counter return different values. If they do, then we consider the cycle
counter to be fast enough that one trip through the scheduler will yield
one "bit" of credited entropy. If those two reads return the same value,
then we assume the cycle counter is too slow to show meaningful
differences.
This methodology is flawed for a variety of reasons, one of which Eric
posted a patch to fix in [1]. The issue that patch solves is that on a
system with a slow counter, you might be [un]lucky and read the counter
_just_ before it changes, so that the second cycle counter you read
differs from the first, even though there's usually quite a large period
of time in between the two. For example:
| real time | cycle counter |
| --------- | ------------- |
| 3 | 5 |
| 4 | 5 |
| 5 | 5 |
| 6 | 5 |
| 7 | 5 | <--- a
| 8 | 6 | <--- b
| 9 | 6 | <--- c
If we read the counter at (a) and compare it to (b), we might be fooled
into thinking that it's a fast counter, when in reality it is not. The
solution in [1] is to also compare counter (b) to counter (c), on the
theory that if the counter is _actually_ slow, and (a)!=(b), then
certainly (b)==(c).
This helps solve this particular issue, in one sense, but in another
sense, it mostly functions to disallow jitter entropy on these systems,
rather than simply taking more samples in that case.
Instead, this patch takes a different approach. Right now we assume that
a difference in one set of consecutive samples means one "bit" of
credited entropy per scheduler trip. We can extend this so that a
difference in two sets of consecutive samples means one "bit" of
credited entropy per /two/ scheduler trips, and three for three, and
four for four. In other words, we can increase the amount of jitter
"work" we require for each "bit", depending on how slow the cycle
counter is.
So this patch takes whole bunch of samples, sees how many of them are
different, and divides to find the amount of work required per "bit",
and also requires that at least some minimum of them are different in
order to attempt any jitter entropy.
Note that this approach is still far from perfect. It's not a real
statistical estimate on how much these samples vary; it's not a
real-time analysis of the relevant input data. That remains a project
for another time. However, it makes the same (partly flawed) assumptions
as the code that's there now, so it's probably not worse than the status
quo, and it handles the issue Eric mentioned in [1]. But, again, it's
probably a far cry from whatever a really robust version of this would
be.
[1] https://lore.kernel.org/lkml/20220421233152.58522-1-ebiggers@kernel.org/https://lore.kernel.org/lkml/20220421192939.250680-1-ebiggers@kernel.org/
Cc: Eric Biggers <ebiggers@google.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
All platforms are now guaranteed to provide some value for
random_get_entropy(). In case some bug leads to this not being so, we
print a warning, because that indicates that something is really very
wrong (and likely other things are impacted too). This should never be
hit, but it's a good and cheap way of finding out if something ever is
problematic.
Since we now have viable fallback code for random_get_entropy() on all
platforms, which is, in the worst case, not worse than jiffies, we can
count on getting the best possible value out of it. That means there's
no longer a use for using jiffies as entropy input. It also means we no
longer have a reason for doing the round-robin register flow in the IRQ
handler, which was always of fairly dubious value.
Instead we can greatly simplify the IRQ handler inputs and also unify
the construction between 64-bits and 32-bits. We now collect the cycle
counter and the return address, since those are the two things that
matter. Because the return address and the irq number are likely
related, to the extent we mix in the irq number, we can just xor it into
the top unchanging bytes of the return address, rather than the bottom
changing bytes of the cycle counter as before. Then, we can do a fixed 2
rounds of SipHash/HSipHash. Finally, we use the same construction of
hashing only half of the [H]SipHash state on 32-bit and 64-bit. We're
not actually discarding any entropy, since that entropy is carried
through until the next time. And more importantly, it lets us do the
same sponge-like construction everywhere.
Cc: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This reverts 35a33ff380 ("random: use memmove instead of memcpy for
remaining 32 bytes"), which was made on a totally bogus basis. The thing
it was worried about overlapping came from the stack, not from one of
its arguments, as Eric pointed out.
But the fact that this confusion even happened draws attention to the
fact that it's a bit non-obvious that the random_data parameter can
alias chacha_state, and in fact should do so when the caller can't rely
on the stack being cleared in a timely manner. So this commit documents
that.
Reported-by: Eric Biggers <ebiggers@kernel.org>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
In order to immediately overwrite the old key on the stack, before
servicing a userspace request for bytes, we use the remaining 32 bytes
of block 0 as the key. This means moving indices 8,9,a,b,c,d,e,f ->
4,5,6,7,8,9,a,b. Since 4 < 8, for the kernel implementations of
memcpy(), this doesn't actually appear to be a problem in practice. But
relying on that characteristic seems a bit brittle. So let's change that
to a proper memmove(), which is the by-the-books way of handling
overlapping memory copies.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Some implementations were returning type `unsigned long`, while others
that fell back to get_cycles() were implicitly returning a `cycles_t` or
an untyped constant int literal. That makes for weird and confusing
code, and basically all code in the kernel already handled it like it
was an `unsigned long`. I recently tried to handle it as the largest
type it could be, a `cycles_t`, but doing so doesn't really help with
much.
Instead let's just make random_get_entropy() return an unsigned long all
the time. This also matches the commonly used `arch_get_random_long()`
function, so now RDRAND and RDTSC return the same sized integer, which
means one can fallback to the other more gracefully.
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: Theodore Ts'o <tytso@mit.edu>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Rather than failing entirely if a copy_to_user() fails at some point,
instead we should return a partial read for the amount that succeeded
prior, unless none succeeded at all, in which case we return -EFAULT as
before.
This makes it consistent with other reader interfaces. For example, the
following snippet for /dev/zero outputs "4" followed by "1":
int fd;
void *x = mmap(NULL, 4096, PROT_WRITE, MAP_ANONYMOUS | MAP_PRIVATE, -1, 0);
assert(x != MAP_FAILED);
fd = open("/dev/zero", O_RDONLY);
assert(fd >= 0);
printf("%zd\n", read(fd, x, 4));
printf("%zd\n", read(fd, x + 4095, 4));
close(fd);
This brings that same standard behavior to the various RNG reader
interfaces.
While we're at it, we can streamline the loop logic a little bit.
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Jann Horn <jannh@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
In 1448769c9c ("random: check for signal_pending() outside of
need_resched() check"), Jann pointed out that we previously were only
checking the TIF_NOTIFY_SIGNAL and TIF_SIGPENDING flags if the process
had TIF_NEED_RESCHED set, which meant in practice, super long reads to
/dev/[u]random would delay signal handling by a long time. I tried this
using the below program, and indeed I wasn't able to interrupt a
/dev/urandom read until after several megabytes had been read. The bug
he fixed has always been there, and so code that reads from /dev/urandom
without checking the return value of read() has mostly worked for a long
time, for most sizes, not just for <= 256.
Maybe it makes sense to keep that code working. The reason it was so
small prior, ignoring the fact that it didn't work anyway, was likely
because /dev/random used to block, and that could happen for pretty
large lengths of time while entropy was gathered. But now, it's just a
chacha20 call, which is extremely fast and is just operating on pure
data, without having to wait for some external event. In that sense,
/dev/[u]random is a lot more like /dev/zero.
Taking a page out of /dev/zero's read_zero() function, it always returns
at least one chunk, and then checks for signals after each chunk. Chunk
sizes there are of length PAGE_SIZE. Let's just copy the same thing for
/dev/[u]random, and check for signals and cond_resched() for every
PAGE_SIZE amount of data. This makes the behavior more consistent with
expectations, and should mitigate the impact of Jann's fix for the
age-old signal check bug.
---- test program ----
#include <unistd.h>
#include <signal.h>
#include <stdio.h>
#include <sys/random.h>
static unsigned char x[~0U];
static void handle(int) { }
int main(int argc, char *argv[])
{
pid_t pid = getpid(), child;
signal(SIGUSR1, handle);
if (!(child = fork())) {
for (;;)
kill(pid, SIGUSR1);
}
pause();
printf("interrupted after reading %zd bytes\n", getrandom(x, sizeof(x), 0));
kill(child, SIGTERM);
return 0;
}
Cc: Jann Horn <jannh@google.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
signal_pending() checks TIF_NOTIFY_SIGNAL and TIF_SIGPENDING, which
signal that the task should bail out of the syscall when possible. This
is a separate concept from need_resched(), which checks
TIF_NEED_RESCHED, signaling that the task should preempt.
In particular, with the current code, the signal_pending() bailout
probably won't work reliably.
Change this to look like other functions that read lots of data, such as
read_zero().
Fixes: 1da177e4c3 ("Linux-2.6.12-rc2")
Signed-off-by: Jann Horn <jannh@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The fast key erasure RNG design relies on the key that's used to be used
and then discarded. We do this, making judicious use of
memzero_explicit(). However, reads to /dev/urandom and calls to
getrandom() involve a copy_to_user(), and userspace can use FUSE or
userfaultfd, or make a massive call, dynamically remap memory addresses
as it goes, and set the process priority to idle, in order to keep a
kernel stack alive indefinitely. By probing
/proc/sys/kernel/random/entropy_avail to learn when the crng key is
refreshed, a malicious userspace could mount this attack every 5 minutes
thereafter, breaking the crng's forward secrecy.
In order to fix this, we just overwrite the stack's key with the first
32 bytes of the "free" fast key erasure output. If we're returning <= 32
bytes to the user, then we can still return those bytes directly, so
that short reads don't become slower. And for long reads, the difference
is hopefully lost in the amortization, so it doesn't change much, with
that amortization helping variously for medium reads.
We don't need to do this for get_random_bytes() and the various
kernel-space callers, and later, if we ever switch to always batching,
this won't be necessary either, so there's no need to change the API of
these functions.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Jann Horn <jannh@google.com>
Fixes: c92e040d57 ("random: add backtracking protection to the CRNG")
Fixes: 186873c549 ("random: use simpler fast key erasure flow on per-cpu keys")
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
In 6f98a4bfee ("random: block in /dev/urandom"), we tried to make a
successful try_to_generate_entropy() call *required* if the RNG was not
already initialized. Unfortunately, weird architectures and old
userspaces combined in TCG test harnesses, making that change still not
realistic, so it was reverted in 0313bc278d ("Revert "random: block in
/dev/urandom"").
However, rather than making a successful try_to_generate_entropy() call
*required*, we can instead make it *best-effort*.
If try_to_generate_entropy() fails, it fails, and nothing changes from
the current behavior. If it succeeds, then /dev/urandom becomes safe to
use for free. This way, we don't risk the regression potential that led
to us reverting the required-try_to_generate_entropy() call before.
Practically speaking, this means that at least on x86, /dev/urandom
becomes safe. Probably other architectures with working cycle counters
will also become safe. And architectures with slow or broken cycle
counters at least won't be affected at all by this change.
So it may not be the glorious "all things are unified!" change we were
hoping for initially, but practically speaking, it makes a positive
impact.
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
add_hwgenerator_randomness() tries to only use the required amount of input
for fast init, but credits all the entropy, rather than a fraction of
it. Since it's hard to determine how much entropy is left over out of a
non-unformly random sample, either give it all to fast init or credit
it, but don't attempt to do both. In the process, we can clean up the
injection code to no longer need to return a value.
Signed-off-by: Jan Varho <jan.varho@gmail.com>
[Jason: expanded commit message]
Fixes: 73c7733f12 ("random: do not throw away excess input to crng_fast_load")
Cc: stable@vger.kernel.org # 5.17+, requires af704c856e
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Prior, the "input_pool_data" array needed no real initialization, and so
it was easy to mark it with __latent_entropy to populate it during
compile-time. In switching to using a hash function, this required us to
specifically initialize it to some specific state, which means we
dropped the __latent_entropy attribute. An unfortunate side effect was
this meant the pool was no longer seeded using compile-time random data.
In order to bring this back, we declare an array in rand_initialize()
with __latent_entropy and call mix_pool_bytes() on that at init, which
accomplishes the same thing as before. We make this __initconst, so that
it doesn't take up space at runtime after init.
Fixes: 6e8ec2552c ("random: use computational hash for entropy extraction")
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The comment about get_random_{u32,u64}() not invoking reseeding got
added in an unrelated commit, that then was recently reverted by
0313bc278d ("Revert "random: block in /dev/urandom""). So this adds
that little comment snippet back, and improves the wording a bit too.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
If CONFIG_RANDOM_TRUST_CPU is set, the RNG initializes using RDRAND.
But, the user can disable (or enable) this behavior by setting
`random.trust_cpu=0/1` on the kernel command line. This allows system
builders to do reasonable things while avoiding howls from tinfoil
hatters. (Or vice versa.)
CONFIG_RANDOM_TRUST_BOOTLOADER is basically the same thing, but regards
the seed passed via EFI or device tree, which might come from RDRAND or
a TPM or somewhere else. In order to allow distros to more easily enable
this while avoiding those same howls (or vice versa), this commit adds
the corresponding `random.trust_bootloader=0/1` toggle.
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Graham Christensen <graham@grahamc.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Link: https://github.com/NixOS/nixpkgs/pull/165355
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
At boot time, EFI calls add_bootloader_randomness(), which in turn calls
add_hwgenerator_randomness(). Currently add_hwgenerator_randomness()
feeds the first 64 bytes of randomness to the "fast init"
non-crypto-grade phase. But if add_hwgenerator_randomness() gets called
with more than POOL_MIN_BITS of entropy, there's no point in passing it
off to the "fast init" stage, since that's enough entropy to bootstrap
the real RNG. The "fast init" stage is just there to provide _something_
in the case where we don't have enough entropy to properly bootstrap the
RNG. But if we do have enough entropy to bootstrap the RNG, the current
logic doesn't serve a purpose. So, in the case where we're passed
greater than or equal to POOL_MIN_BITS of entropy, this commit makes us
skip the "fast init" phase.
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This reverts commit 6f98a4bfee.
It turns out we still can't do this. Way too many platforms that don't
have any real source of randomness at boot and no jitter entropy because
they don't even have a cycle counter.
As reported by Guenter Roeck:
"This causes a large number of qemu boot test failures for various
architectures (arm, m68k, microblaze, sparc32, xtensa are the ones I
observed).
Common denominator is that boot hangs at 'Saving random seed:'"
This isn't hugely unexpected - we tried it, it failed, so now we'll
revert it.
Link: https://lore.kernel.org/all/20220322155820.GA1745955@roeck-us.net/
Reported-and-bisected-by: Guenter Roeck <linux@roeck-us.net>
Cc: Jason Donenfeld <Jason@zx2c4.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Rather than waiting a full second in an interruptable waiter before
trying to generate entropy, try to generate entropy first and wait
second. While waiting one second might give an extra second for getting
entropy from elsewhere, we're already pretty late in the init process
here, and whatever else is generating entropy will still continue to
contribute. This has implications on signal handling: we call
try_to_generate_entropy() from wait_for_random_bytes(), and
wait_for_random_bytes() always uses wait_event_interruptible_timeout()
when waiting, since it's called by userspace code in restartable
contexts, where signals can pend. Since try_to_generate_entropy() now
runs first, if a signal is pending, it's necessary for
try_to_generate_entropy() to check for signals, since it won't hit the
wait until after try_to_generate_entropy() has returned. And even before
this change, when entering a busy loop in try_to_generate_entropy(), we
should have been checking to see if any signals are pending, so that a
process doesn't get stuck in that loop longer than expected.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
In order to chip away at the "premature first" problem, we augment our
existing entropy accounting with more frequent reseedings at boot.
The idea is that at boot, we're getting entropy from various places, and
we're not very sure which of early boot entropy is good and which isn't.
Even when we're crediting the entropy, we're still not totally certain
that it's any good. Since boot is the one time (aside from a compromise)
that we have zero entropy, it's important that we shepherd entropy into
the crng fairly often.
At the same time, we don't want a "premature next" problem, whereby an
attacker can brute force individual bits of added entropy. In lieu of
going full-on Fortuna (for now), we can pick a simpler strategy of just
reseeding more often during the first 5 minutes after boot. This is
still bounded by the 256-bit entropy credit requirement, so we'll skip a
reseeding if we haven't reached that, but in case entropy /is/ coming
in, this ensures that it makes its way into the crng rather rapidly
during these early stages.
Ordinarily we reseed if the previous reseeding is 300 seconds old. This
commit changes things so that for the first 600 seconds of boot time, we
reseed if the previous reseeding is uptime / 2 seconds old. That means
that we'll reseed at the very least double the uptime of the previous
reseeding.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Rather than sometimes checking `crng_init < 2`, we should always use the
crng_ready() macro, so that should we change anything later, it's
consistent. Additionally, that macro already has a likely() around it,
which means we don't need to open code our own likely() and unlikely()
annotations.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The current fast_mix() function is a piece of classic mailing list
crypto, where it just sort of sprung up by an anonymous author without a
lot of real analysis of what precisely it was accomplishing. As an ARX
permutation alone, there are some easily searchable differential trails
in it, and as a means of preventing malicious interrupts, it completely
fails, since it xors new data into the entire state every time. It can't
really be analyzed as a random permutation, because it clearly isn't,
and it can't be analyzed as an interesting linear algebraic structure
either, because it's also not that. There really is very little one can
say about it in terms of entropy accumulation. It might diffuse bits,
some of the time, maybe, we hope, I guess. But for the most part, it
fails to accomplish anything concrete.
As a reminder, the simple goal of add_interrupt_randomness() is to
simply accumulate entropy until ~64 interrupts have elapsed, and then
dump it into the main input pool, which uses a cryptographic hash.
It would be nice to have something cryptographically strong in the
interrupt handler itself, in case a malicious interrupt compromises a
per-cpu fast pool within the 64 interrupts / 1 second window, and then
inside of that same window somehow can control its return address and
cycle counter, even if that's a bit far fetched. However, with a very
CPU-limited budget, actually doing that remains an active research
project (and perhaps there'll be something useful for Linux to come out
of it). And while the abundance of caution would be nice, this isn't
*currently* the security model, and we don't yet have a fast enough
solution to make it our security model. Plus there's not exactly a
pressing need to do that. (And for the avoidance of doubt, the actual
cluster of 64 accumulated interrupts still gets dumped into our
cryptographically secure input pool.)
So, for now we are going to stick with the existing interrupt security
model, which assumes that each cluster of 64 interrupt data samples is
mostly non-malicious and not colluding with an infoleaker. With this as
our goal, we have a few more choices, simply aiming to accumulate
entropy, while discarding the least amount of it.
We know from <https://eprint.iacr.org/2019/198> that random oracles,
instantiated as computational hash functions, make good entropy
accumulators and extractors, which is the justification for using
BLAKE2s in the main input pool. As mentioned, we don't have that luxury
here, but we also don't have the same security model requirements,
because we're assuming that there aren't malicious inputs. A
pseudorandom function instance can approximately behave like a random
oracle, provided that the key is uniformly random. But since we're not
concerned with malicious inputs, we can pick a fixed key, which is not
secret, knowing that "nature" won't interact with a sufficiently chosen
fixed key by accident. So we pick a PRF with a fixed initial key, and
accumulate into it continuously, dumping the result every 64 interrupts
into our cryptographically secure input pool.
For this, we make use of SipHash-1-x on 64-bit and HalfSipHash-1-x on
32-bit, which are already in use in the kernel's hsiphash family of
functions and achieve the same performance as the function they replace.
It would be nice to do two rounds, but we don't exactly have the CPU
budget handy for that, and one round alone is already sufficient.
As mentioned, we start with a fixed initial key (zeros is fine), and
allow SipHash's symmetry breaking constants to turn that into a useful
starting point. Also, since we're dumping the result (or half of it on
64-bit so as to tax our hash function the same amount on all platforms)
into the cryptographically secure input pool, there's no point in
finalizing SipHash's output, since it'll wind up being finalized by
something much stronger. This means that all we need to do is use the
ordinary round function word-by-word, as normal SipHash does.
Simplified, the flow is as follows:
Initialize:
siphash_state_t state;
siphash_init(&state, key={0, 0, 0, 0});
Update (accumulate) on interrupt:
siphash_update(&state, interrupt_data_and_timing);
Dump into input pool after 64 interrupts:
blake2s_update(&input_pool, &state, sizeof(state) / 2);
The result of all of this is that the security model is unchanged from
before -- we assume non-malicious inputs -- yet we now implement that
model with a stronger argument. I would like to emphasize, again, that
the purpose of this commit is to improve the existing design, by making
it analyzable, without changing any fundamental assumptions. There may
well be value down the road in changing up the existing design, using
something cryptographically strong, or simply using a ring buffer of
samples rather than having a fast_mix() at all, or changing which and
how much data we collect each interrupt so that we can use something
linear, or a variety of other ideas. This commit does not invalidate the
potential for those in the future.
For example, in the future, if we're able to characterize the data we're
collecting on each interrupt, we may be able to inch toward information
theoretic accumulators. <https://eprint.iacr.org/2021/523> shows that `s
= ror32(s, 7) ^ x` and `s = ror64(s, 19) ^ x` make very good
accumulators for 2-monotone distributions, which would apply to
timestamp counters, like random_get_entropy() or jiffies, but would not
apply to our current combination of the two values, or to the various
function addresses and register values we mix in. Alternatively,
<https://eprint.iacr.org/2021/1002> shows that max-period linear
functions with no non-trivial invariant subspace make good extractors,
used in the form `s = f(s) ^ x`. However, this only works if the input
data is both identical and independent, and obviously a collection of
address values and counters fails; so it goes with theoretical papers.
Future directions here may involve trying to characterize more precisely
what we actually need to collect in the interrupt handler, and building
something specific around that.
However, as mentioned, the morass of data we're gathering at the
interrupt handler presently defies characterization, and so we use
SipHash for now, which works well and performs well.
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Jean-Philippe Aumasson <jeanphilippe.aumasson@gmail.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Drivers such as WireGuard need to learn when VMs fork in order to clear
sessions. This commit provides a simple notifier_block for that, with a
register and unregister function. When no VM fork detection is compiled
in, this turns into a no-op, similar to how the power notifier works.
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
We previously rolled our own randomness readiness notifier, which only
has two users in the whole kernel. Replace this with a more standard
atomic notifier block that serves the same purpose with less code. Also
unexport the symbols, because no modules use it, only unconditional
builtins. The only drawback is that it's possible for a notification
handler returning the "stop" code to prevent further processing, but
given that there are only two users, and that we're unexporting this
anyway, that doesn't seem like a significant drawback for the
simplification we receive here.
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Since add_vmfork_randomness() is only called from vmgenid.o, we can
guard it in CONFIG_VMGENID, similarly to how we do with
add_disk_randomness() and CONFIG_BLOCK. If we ever have multiple things
calling into add_vmfork_randomness(), we can add another shared Kconfig
symbol for that, but for now, this is good enough. Even though
add_vmfork_randomess() is a pretty small function, removing it means
that there are only calls to crng_reseed(false) and none to
crng_reseed(true), which means the compiler can constant propagate the
false, removing branches from crng_reseed() and its descendants.
Additionally, we don't even need the symbol to be exported if
CONFIG_VMGENID is not a module, so conditionalize that too.
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
When a VM forks, we must immediately mix in additional information to
the stream of random output so that two forks or a rollback don't
produce the same stream of random numbers, which could have catastrophic
cryptographic consequences. This commit adds a simple API, add_vmfork_
randomness(), for that, by force reseeding the crng.
This has the added benefit of also draining the entropy pool and setting
its timer back, so that any old entropy that was there prior -- which
could have already been used by a different fork, or generally gone
stale -- does not contribute to the accounting of the next 256 bits.
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Jann Horn <jannh@google.com>
Cc: Eric Biggers <ebiggers@google.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
We leave around these old sysctls for compatibility, and we keep them
"writable" for compatibility, but even after writing, we should keep
reporting the same value. This is consistent with how userspaces tend to
use sysctl_random_write_wakeup_bits, writing to it, and then later
reading from it and using the value.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This isn't used by anything or anywhere, but we can't delete it due to
compatibility. So at least give it the correct value of what it's
supposed to be instead of a garbage one.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This topic has come up countless times, and usually doesn't go anywhere.
This time I thought I'd bring it up with a slightly narrower focus,
updated for some developments over the last three years: we finally can
make /dev/urandom always secure, in light of the fact that our RNG is
now always seeded.
Ever since Linus' 50ee7529ec ("random: try to actively add entropy
rather than passively wait for it"), the RNG does a haveged-style jitter
dance around the scheduler, in order to produce entropy (and credit it)
for the case when we're stuck in wait_for_random_bytes(). How ever you
feel about the Linus Jitter Dance is beside the point: it's been there
for three years and usually gets the RNG initialized in a second or so.
As a matter of fact, this is what happens currently when people use
getrandom(). It's already there and working, and most people have been
using it for years without realizing.
So, given that the kernel has grown this mechanism for seeding itself
from nothing, and that this procedure happens pretty fast, maybe there's
no point any longer in having /dev/urandom give insecure bytes. In the
past we didn't want the boot process to deadlock, which was
understandable. But now, in the worst case, a second goes by, and the
problem is resolved. It seems like maybe we're finally at a point when
we can get rid of the infamous "urandom read hole".
The one slight drawback is that the Linus Jitter Dance relies on random_
get_entropy() being implemented. The first lines of try_to_generate_
entropy() are:
stack.now = random_get_entropy();
if (stack.now == random_get_entropy())
return;
On most platforms, random_get_entropy() is simply aliased to get_cycles().
The number of machines without a cycle counter or some other
implementation of random_get_entropy() in 2022, which can also run a
mainline kernel, and at the same time have a both broken and out of date
userspace that relies on /dev/urandom never blocking at boot is thought
to be exceedingly low. And to be clear: those museum pieces without
cycle counters will continue to run Linux just fine, and even
/dev/urandom will be operable just like before; the RNG just needs to be
seeded first through the usual means, which should already be the case
now.
On systems that really do want unseeded randomness, we already offer
getrandom(GRND_INSECURE), which is in use by, e.g., systemd for seeding
their hash tables at boot. Nothing in this commit would affect
GRND_INSECURE, and it remains the means of getting those types of random
numbers.
This patch goes a long way toward eliminating a long overdue userspace
crypto footgun. After several decades of endless user confusion, we will
finally be able to say, "use any single one of our random interfaces and
you'll be fine. They're all the same. It doesn't matter." And that, I
think, is really something. Finally all of those blog posts and
disagreeing forums and contradictory articles will all become correct
about whatever they happened to recommend, and along with it, a whole
class of vulnerabilities eliminated.
With very minimal downside, we're finally in a position where we can
make this change.
Cc: Dinh Nguyen <dinguyen@kernel.org>
Cc: Nick Hu <nickhu@andestech.com>
Cc: Max Filippov <jcmvbkbc@gmail.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Guo Ren <guoren@kernel.org>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Joshua Kinard <kumba@gentoo.org>
Cc: David Laight <David.Laight@aculab.com>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: Eric Biggers <ebiggers@google.com>
Cc: Ard Biesheuvel <ardb@kernel.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Lennart Poettering <mzxreary@0pointer.de>
Cc: Konstantin Ryabitsev <konstantin@linuxfoundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Taking spinlocks from IRQ context is generally problematic for
PREEMPT_RT. That is, in part, why we take trylocks instead. However, a
spin_try_lock() is also problematic since another spin_lock() invocation
can potentially PI-boost the wrong task, as the spin_try_lock() is
invoked from an IRQ-context, so the task on CPU (random task or idle) is
not the actual owner.
Additionally, by deferring the crng pre-init loading to the worker, we
can use the cryptographic hash function rather than xor, which is
perhaps a meaningful difference when considering this data has only been
through the relatively weak fast_mix() function.
The biggest downside of this approach is that the pre-init loading is
now deferred until later, which means things that need random numbers
after interrupts are enabled, but before workqueues are running -- or
before this particular worker manages to run -- are going to get into
trouble. Hopefully in the real world, this window is rather small,
especially since this code won't run until 64 interrupts had occurred.
Cc: Sultan Alsawaf <sultan@kerneltoast.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Eric Biggers <ebiggers@kernel.org>
Cc: Theodore Ts'o <tytso@mit.edu>
Acked-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
random_get_entropy() returns a cycles_t, not an unsigned long, which is
sometimes 64 bits on various 32-bit platforms, including x86.
Conversely, jiffies is always unsigned long. This commit fixes things to
use cycles_t for fields that use random_get_entropy(), named "cycles",
and unsigned long for fields that use jiffies, named "now". It's also
good to mix in a cycles_t and a jiffies in the same way for both
add_device_randomness and add_timer_randomness, rather than using xor in
one case. Finally, we unify the order of these volatile reads, always
reading the more precise cycles counter, and then jiffies, so that the
cycle counter is as close to the event as possible.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Rather than hard coding various lengths, we can use the right constants.
Strings should be `char *` while buffers should be `u8 *`. Rather than
have a nonsensical and unused maxlength, just remove it. Finally, use
snprintf instead of sprintf, just out of good hygiene.
As well, remove the old comment about returning a binary UUID via the
binary sysctl syscall. That syscall was removed from the kernel in 5.5,
and actually, the "uuid_strategy" function and related infrastructure
for even serving it via the binary sysctl syscall was removed with
894d249115 ("sysctl drivers: Remove dead binary sysctl support") back
in 2.6.33.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The only time that we need to wake up /dev/random writers on
RNDCLEARPOOL/RNDZAPPOOL is when we're changing from a value that is
greater than or equal to POOL_MIN_BITS to zero, because if we're
changing from below POOL_MIN_BITS to zero, the writers are already
unblocked.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
When the interrupt handler does not have a valid cycle counter, it calls
get_reg() to read a register from the irq stack, in round-robin.
Currently it does this assuming that registers are 32-bit. This is
_probably_ the case, and probably all platforms without cycle counters
are in fact 32-bit platforms. But maybe not, and either way, it's not
quite correct. This commit fixes that to deal with `unsigned long`
rather than `u32`.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
For the irq randomness fast pool, rather than having to use expensive
atomics, which were visibly the most expensive thing in the entire irq
handler, simply take care of the extreme edge case of resetting count to
zero in the cpuhp online handler, just after workqueues have been
reenabled. This simplifies the code a bit and lets us use vanilla
variables rather than atomics, and performance should be improved.
As well, very early on when the CPU comes up, while interrupts are still
disabled, we clear out the per-cpu crng and its batches, so that it
always starts with fresh randomness.
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Sultan Alsawaf <sultan@kerneltoast.com>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Acked-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This has no real functional change, as crng_pre_init_inject() (and
before that, crng_slow_init()) always checks for == 0, not >= 2. So
correct the outer unlocked change to reflect that. Before this used
crng_ready(), which was not correct.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
crng_fast_load() and crng_slow_load() have different semantics:
- crng_fast_load() xors and accounts with crng_init_cnt.
- crng_slow_load() hashes and doesn't account.
However add_hwgenerator_randomness() can afford to hash (it's called
from a kthread), and it should account. Additionally, ones that can
afford to hash don't need to take a trylock but can take a normal lock.
So, we combine these into one function, crng_pre_init_inject(), which
allows us to control these in a uniform way. This will make it simpler
later to simplify this all down when the time comes for that.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Since rand_initialize() is run while interrupts are still off and
nothing else is running, we don't need to repeatedly take and release
the pool spinlock, especially in the RDSEED loop.
Reviewed-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
On PREEMPT_RT, it's problematic to take spinlocks from hard irq
handlers. We can fix this by deferring to a workqueue the dumping of
the fast pool into the input pool.
We accomplish this with some careful rules on fast_pool->count:
- When it's incremented to >= 64, we schedule the work.
- If the top bit is set, we never schedule the work, even if >= 64.
- The worker is responsible for setting it back to 0 when it's done.
There are two small issues around using workqueues for this purpose that
we work around.
The first issue is that mix_interrupt_randomness() might be migrated to
another CPU during CPU hotplug. This issue is rectified by checking that
it hasn't been migrated (after disabling irqs). If it has been migrated,
then we set the count to zero, so that when the CPU comes online again,
it can requeue the work. As part of this, we switch to using an
atomic_t, so that the increment in the irq handler doesn't wipe out the
zeroing if the CPU comes back online while this worker is running.
The second issue is that, though relatively minor in effect, we probably
want to make sure we get a consistent view of the pool onto the stack,
in case it's interrupted by an irq while reading. To do this, we don't
reenable irqs until after the copy. There are only 18 instructions
between the cli and sti, so this is a pretty tiny window.
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
Acked-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Sultan Alsawaf <sultan@kerneltoast.com>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Now that we've re-documented the various sections, we can remove the
outdated text here and replace it with a high-level overview.
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 pulls all of the sysctl-focused functions into the sixth labeled
section.
No functional changes.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This pulls all of the userspace read/write-focused functions into the
fifth labeled section.
No functional changes.
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 pulls all of the entropy collection-focused functions into the
fourth labeled section.
No functional changes.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This pulls all of the entropy extraction-focused functions into the
third labeled section.
No functional changes.
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 pulls all of the crng-focused functions into the second labeled
section.
No functional changes.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This pulls all of the readiness waiting-focused functions into the first
labeled section.
No functional changes.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This is purely cosmetic. Future work involves figuring out which of
these headers we need and which we don't.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
In preparation for separating responsibilities, break out the entropy
count management part of crng_reseed() into its own function.
No functional changes.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
In the irq handler, we fill out 16 bytes differently on 32-bit and
64-bit platforms, and for 32-bit vs 64-bit cycle counters, which doesn't
always correspond with the bitness of the platform. Whether or not you
like this strangeness, it is a matter of fact. But it might not be a
fact you well realized until now, because the code that loaded the irq
info into 4 32-bit words was quite confusing. Instead, this commit
makes everything explicit by having separate (compile-time) branches for
32-bit and 64-bit types.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Convert the current license into the SPDX notation of "(GPL-2.0 OR
BSD-3-Clause)". This infers GPL-2.0 from the text "ALTERNATIVELY, this
product may be distributed under the terms of the GNU General Public
License, in which case the provisions of the GPL are required INSTEAD OF
the above restrictions" and it infers BSD-3-Clause from the verbatim
BSD 3 clause license in the file.
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
These explicit tracepoints aren't really used and show sign of aging.
It's work to keep these up to date, and before I attempted to keep them
up to date, they weren't up to date, which indicates that they're not
really used. These days there are better ways of introspecting anyway.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
With tools like kbench9000 giving more finegrained responses, and this
basically never having been used ever since it was initially added,
let's just get rid of this. There *is* still work to be done on the
interrupt handler, but this really isn't the way it's being developed.
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>
Now that we have an explicit base_crng generation counter, we don't need
a separate one for batched entropy. Rather, we can just move the
generation forward every time we change crng_init state or update the
base_crng key.
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>
crng_init is protected by primary_crng->lock. Therefore, we need
to hold this lock when increasing crng_init to 2. As we shouldn't
hold this lock for too long, only hold it for those parts which
require protection.
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This buffer may contain entropic data that shouldn't stick around longer
than needed, so zero out the temporary buffer at the end of write_pool().
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Jann Horn <jannh@google.com>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
In 79a8468747 ("random: check for increase of entropy_count because of
signed conversion"), a number of checks were added around what values
were passed to account(), because account() was doing fancy fixed point
fractional arithmetic, and a user had some ability to pass large values
directly into it. One of things in that commit was limiting those values
to INT_MAX >> 6. The first >> 3 was for bytes to bits, and the next >> 3
was for bits to 1/8 fractional bits.
However, for several years now, urandom reads no longer touch entropy
accounting, and so this check serves no purpose. The current flow is:
urandom_read_nowarn()-->get_random_bytes_user()-->chacha20_block()
Of course, we don't want that size_t to be truncated when adding it into
the ssize_t. But we arrive at urandom_read_nowarn() in the first place
either via ordinary fops, which limits reads to MAX_RW_COUNT, or via
getrandom() which limits reads to INT_MAX.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Jann Horn <jannh@google.com>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
We've been using a flurry of int, unsigned int, size_t, and ssize_t.
Let's unify all of this into size_t where it makes sense, as it does in
most places, and leave ssize_t for return values with possible errors.
In addition, keeping with the convention of other functions in this
file, functions that are dealing with raw bytes now take void *
consistently instead of a mix of that and u8 *, because much of the time
we're actually passing some other structure that is then interpreted as
bytes by the function.
We also take the opportunity to fix the outdated and incorrect comment
in get_random_bytes_arch().
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Jann Horn <jannh@google.com>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Since we have a hash function that's really fast, and the goal of
crng_slow_load() is reportedly to "touch all of the crng's state", we
can just hash the old state together with the new state and call it a
day. This way we dont need to reason about another LFSR or worry about
various attacks there. This code is only ever used at early boot and
then never again.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Rather than the clunky NUMA full ChaCha state system we had prior, this
commit is closer to the original "fast key erasure RNG" proposal from
<https://blog.cr.yp.to/20170723-random.html>, by simply treating ChaCha
keys on a per-cpu basis.
All entropy is extracted to a base crng key of 32 bytes. This base crng
has a birthdate and a generation counter. When we go to take bytes from
the crng, we first check if the birthdate is too old; if it is, we
reseed per usual. Then we start working on a per-cpu crng.
This per-cpu crng makes sure that it has the same generation counter as
the base crng. If it doesn't, it does fast key erasure with the base
crng key and uses the output as its new per-cpu key, and then updates
its local generation counter. Then, using this per-cpu state, we do
ordinary fast key erasure. Half of this first block is used to overwrite
the per-cpu crng key for the next call -- this is the fast key erasure
RNG idea -- and the other half, along with the ChaCha state, is returned
to the caller. If the caller desires more than this remaining half, it
can generate more ChaCha blocks, unlocked, using the now detached ChaCha
state that was just returned. Crypto-wise, this is more or less what we
were doing before, but this simply makes it more explicit and ensures
that we always have backtrack protection by not playing games with a
shared block counter.
The flow looks like this:
──extract()──► base_crng.key ◄──memcpy()───┐
│ │
└──chacha()──────┬─► new_base_key
└─► crngs[n].key ◄──memcpy()───┐
│ │
└──chacha()───┬─► new_key
└─► random_bytes
│
└────►
There are a few hairy details around early init. Just as was done
before, prior to having gathered enough entropy, crng_fast_load() and
crng_slow_load() dump bytes directly into the base crng, and when we go
to take bytes from the crng, in that case, we're doing fast key erasure
with the base crng rather than the fast unlocked per-cpu crngs. This is
fine as that's only the state of affairs during very early boot; once
the crng initializes we never use these paths again.
In the process of all this, the APIs into the crng become a bit simpler:
we have get_random_bytes(buf, len) and get_random_bytes_user(buf, len),
which both do what you'd expect. All of the details of fast key erasure
and per-cpu selection happen only in a very short critical section of
crng_make_state(), which selects the right per-cpu key, does the fast
key erasure, and returns a local state to the caller's stack. So, we no
longer have a need for a separate backtrack function, as this happens
all at once here. The API then allows us to extend backtrack protection
to batched entropy without really having to do much at all.
The result is a bit simpler than before and has fewer foot guns. The
init time state machine also gets a lot simpler as we don't need to wait
for workqueues to come online and do deferred work. And the multi-core
performance should be increased significantly, by virtue of having hardly
any locking on the fast path.
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Jann Horn <jannh@google.com>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
During crng_init == 0, we never credit entropy in add_interrupt_
randomness(), but instead dump it directly into the primary_crng. That's
fine, except for the fact that we then wind up throwing away that
entropy later when we switch to extracting from the input pool and
xoring into (and later in this series overwriting) the primary_crng key.
The two other early init sites -- add_hwgenerator_randomness()'s use
crng_fast_load() and add_device_ randomness()'s use of crng_slow_load()
-- always additionally give their inputs to the input pool. But not
add_interrupt_randomness().
This commit fixes that shortcoming by calling mix_pool_bytes() after
crng_fast_load() in add_interrupt_randomness(). That's partially
verboten on PREEMPT_RT, where it implies taking spinlock_t from an IRQ
handler. But this also only happens during early boot and then never
again after that. Plus it's a trylock so it has the same considerations
as calling crng_fast_load(), which we're already using.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Suggested-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Continuing the reasoning of "random: ensure early RDSEED goes through
mixer on init", we don't want RDRAND interacting with anything without
going through the mixer function, as a backdoored CPU could presumably
cancel out data during an xor, which it'd have a harder time doing when
being forced through a cryptographic hash function. There's actually no
need at all to be calling RDRAND in write_pool(), because before we
extract from the pool, we always do so with 32 bytes of RDSEED hashed in
at that stage. Xoring at this stage is needless and introduces a minor
liability.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Continuing the reasoning of "random: use RDSEED instead of RDRAND in
entropy extraction" from this series, at init time we also don't want to
be xoring RDSEED directly into the crng. Instead it's safer to put it
into our entropy collector and then re-extract it, so that it goes
through a hash function with preimage resistance. As a matter of hygiene,
we also order these now so that the RDSEED byte are hashed in first,
followed by the bytes that are likely more predictable (e.g. utsname()).
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This is a preparatory commit for the following one. We simply inline the
various functions that rand_initialize() calls that have no other
callers. The compiler was doing this anyway before. Doing this will
allow us to reorganize this after. We can then move the trust_cpu and
parse_trust_cpu definitions a bit closer to where they're actually used,
which makes the code easier to read.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
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>
When /dev/random was directly connected with entropy extraction, without
any expansion stage, extract_buf() was called for every 10 bytes of data
read from /dev/random. For that reason, RDRAND was used rather than
RDSEED. At the same time, crng_reseed() was still only called every 5
minutes, so there RDSEED made sense.
Those olden days were also a time when the entropy collector did not use
a cryptographic hash function, which meant most bets were off in terms
of real preimage resistance. For that reason too it didn't matter
_that_ much whether RDSEED was mixed in before or after entropy
extraction; both choices were sort of bad.
But now we have a cryptographic hash function at work, and with that we
get real preimage resistance. We also now only call extract_entropy()
every 5 minutes, rather than every 10 bytes. This allows us to do two
important things.
First, we can switch to using RDSEED in extract_entropy(), as Dominik
suggested. Second, we can ensure that RDSEED input always goes into the
cryptographic hash function with other things before being used
directly. This eliminates a category of attacks in which the CPU knows
the current state of the crng and knows that we're going to xor RDSEED
into it, and so it computes a malicious RDSEED. By going through our
hash function, it would require the CPU to compute a preimage on the
fly, which isn't going to happen.
Cc: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Suggested-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
crng_init is protected by primary_crng->lock, so keep holding that lock
when incrementing crng_init from 0 to 1 in crng_fast_load(). The call to
pr_notice() can wait until the lock is released; this code path cannot
be reached twice, as crng_fast_load() aborts early if crng_init > 0.
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Rather than use spinlocks to protect batched entropy, we can instead
disable interrupts locally, since we're dealing with per-cpu data, and
manage resets with a basic generation counter. At the same time, we
can't quite do this on PREEMPT_RT, where we still want spinlocks-as-
mutexes semantics. So we use a local_lock_t, which provides the right
behavior for each. Because this is a per-cpu lock, that generation
counter is still doing the necessary CPU-to-CPU communication.
This should improve performance a bit. It will also fix the linked splat
that Jonathan received with a PROVE_RAW_LOCK_NESTING=y.
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Suggested-by: Andy Lutomirski <luto@kernel.org>
Reported-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
Tested-by: Jonathan Neuschäfer <j.neuschaefer@gmx.net>
Link: https://lore.kernel.org/lkml/YfMa0QgsjCVdRAvJ@latitude/
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The primary_crng is always reseeded from the input_pool, while the NUMA
crngs are always reseeded from the primary_crng. Remove the redundant
'use_input_pool' parameter from crng_reseed() and just directly check
whether the crng is the primary_crng.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This is called from various hwgenerator drivers, so rather than having
one "safe" version for userspace and one "unsafe" version for the
kernel, just make everything safe; the checks are cheap and sensible to
have anyway.
Reported-by: Sultan Alsawaf <sultan@kerneltoast.com>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Now that POOL_BITS == POOL_MIN_BITS, we must unconditionally wake up
entropy writers after every extraction. Therefore there's no point of
write_wakeup_threshold, so we can move it to the dustbin of unused
compatibility sysctls. While we're at it, we can fix a small comparison
where we were waking up after <= min rather than < min.
Cc: Theodore Ts'o <tytso@mit.edu>
Suggested-by: Eric Biggers <ebiggers@kernel.org>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
30e37ec516 ("random: account for entropy loss due to overwrites")
assumed that adding new entropy to the LFSR pool probabilistically
cancelled out old entropy there, so entropy was credited asymptotically,
approximating Shannon entropy of independent sources (rather than a
stronger min-entropy notion) using 1/8th fractional bits and replacing
a constant 2-2/√𝑒 term (~0.786938) with 3/4 (0.75) to slightly
underestimate it. This wasn't superb, but it was perhaps better than
nothing, so that's what was done. Which entropy specifically was being
cancelled out and how much precisely each time is hard to tell, though
as I showed with the attack code in my previous commit, a motivated
adversary with sufficient information can actually cancel out
everything.
Since we're no longer using an LFSR for entropy accumulation, this
probabilistic cancellation is no longer relevant. Rather, we're now
using a computational hash function as the accumulator and we've
switched to working in the random oracle model, from which we can now
revisit the question of min-entropy accumulation, which is done in
detail in <https://eprint.iacr.org/2019/198>.
Consider a long input bit string that is built by concatenating various
smaller independent input bit strings. Each one of these inputs has a
designated min-entropy, which is what we're passing to
credit_entropy_bits(h). When we pass the concatenation of these to a
random oracle, it means that an adversary trying to receive back the
same reply as us would need to become certain about each part of the
concatenated bit string we passed in, which means becoming certain about
all of those h values. That means we can estimate the accumulation by
simply adding up the h values in calls to credit_entropy_bits(h);
there's no probabilistic cancellation at play like there was said to be
for the LFSR. Incidentally, this is also what other entropy accumulators
based on computational hash functions do as well.
So this commit replaces credit_entropy_bits(h) with essentially `total =
min(POOL_BITS, total + h)`, done with a cmpxchg loop as before.
What if we're wrong and the above is nonsense? It's not, but let's
assume we don't want the actual _behavior_ of the code to change much.
Currently that behavior is not extracting from the input pool until it
has 128 bits of entropy in it. With the old algorithm, we'd hit that
magic 128 number after roughly 256 calls to credit_entropy_bits(1). So,
we can retain more or less the old behavior by waiting to extract from
the input pool until it hits 256 bits of entropy using the new code. For
people concerned about this change, it means that there's not that much
practical behavioral change. And for folks actually trying to model
the behavior rigorously, it means that we have an even higher margin
against attacks.
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Jean-Philippe Aumasson <jeanphilippe.aumasson@gmail.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Our pool is 256 bits, and we only ever use all of it or don't use it at
all, which is decided by whether or not it has at least 128 bits in it.
So we can drastically simplify the accounting and cmpxchg loop to do
exactly this. While we're at it, we move the minimum bit size into a
constant so it can be shared between the two places where it matters.
The reason we want any of this is for the case in which an attacker has
compromised the current state, and then bruteforces small amounts of
entropy added to it. By demanding a particular minimum amount of entropy
be present before reseeding, we make that bruteforcing difficult.
Note that this rationale no longer includes anything about /dev/random
blocking at the right moment, since /dev/random no longer blocks (except
for at ~boot), but rather uses the crng. In a former life, /dev/random
was different and therefore required a more nuanced account(), but this
is no longer.
Behaviorally, nothing changes here. This is just a simplification of
the code.
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The current 4096-bit LFSR used for entropy collection had a few
desirable attributes for the context in which it was created. For
example, the state was huge, which meant that /dev/random would be able
to output quite a bit of accumulated entropy before blocking. It was
also, in its time, quite fast at accumulating entropy byte-by-byte,
which matters given the varying contexts in which mix_pool_bytes() is
called. And its diffusion was relatively high, which meant that changes
would ripple across several words of state rather quickly.
However, it also suffers from a few security vulnerabilities. In
particular, inputs learned by an attacker can be undone, but moreover,
if the state of the pool leaks, its contents can be controlled and
entirely zeroed out. I've demonstrated this attack with this SMT2
script, <https://xn--4db.cc/5o9xO8pb>, which Boolector/CaDiCal solves in
a matter of seconds on a single core of my laptop, resulting in little
proof of concept C demonstrators such as <https://xn--4db.cc/jCkvvIaH/c>.
For basically all recent formal models of RNGs, these attacks represent
a significant cryptographic flaw. But how does this manifest
practically? If an attacker has access to the system to such a degree
that he can learn the internal state of the RNG, arguably there are
other lower hanging vulnerabilities -- side-channel, infoleak, or
otherwise -- that might have higher priority. On the other hand, seed
files are frequently used on systems that have a hard time generating
much entropy on their own, and these seed files, being files, often leak
or are duplicated and distributed accidentally, or are even seeded over
the Internet intentionally, where their contents might be recorded or
tampered with. Seen this way, an otherwise quasi-implausible
vulnerability is a bit more practical than initially thought.
Another aspect of the current mix_pool_bytes() function is that, while
its performance was arguably competitive for the time in which it was
created, it's no longer considered so. This patch improves performance
significantly: on a high-end CPU, an i7-11850H, it improves performance
of mix_pool_bytes() by 225%, and on a low-end CPU, a Cortex-A7, it
improves performance by 103%.
This commit replaces the LFSR of mix_pool_bytes() with a straight-
forward cryptographic hash function, BLAKE2s, which is already in use
for pool extraction. Universal hashing with a secret seed was considered
too, something along the lines of <https://eprint.iacr.org/2013/338>,
but the requirement for a secret seed makes for a chicken & egg problem.
Instead we go with a formally proven scheme using a computational hash
function, described in sections 5.1, 6.4, and B.1.8 of
<https://eprint.iacr.org/2019/198>.
BLAKE2s outputs 256 bits, which should give us an appropriate amount of
min-entropy accumulation, and a wide enough margin of collision
resistance against active attacks. mix_pool_bytes() becomes a simple
call to blake2s_update(), for accumulation, while the extraction step
becomes a blake2s_final() to generate a seed, with which we can then do
a HKDF-like or BLAKE2X-like expansion, the first part of which we fold
back as an init key for subsequent blake2s_update()s, and the rest we
produce to the caller. This then is provided to our CRNG like usual. In
that expansion step, we make opportunistic use of 32 bytes of RDRAND
output, just as before. We also always reseed the crng with 32 bytes,
unconditionally, or not at all, rather than sometimes with 16 as before,
as we don't win anything by limiting beyond the 16 byte threshold.
Going for a hash function as an entropy collector is a conservative,
proven approach. The result of all this is a much simpler and much less
bespoke construction than what's there now, which not only plugs a
vulnerability but also improves performance considerably.
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Jean-Philippe Aumasson <jeanphilippe.aumasson@gmail.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
crng_finalize_init() returns instantly if it is called for another pool
than primary_crng. The test whether crng_finalize_init() is still required
can be moved to the relevant caller in crng_reseed(), and
crng_need_final_init can be reset to false if crng_finalize_init() is
called with workqueues ready. Then, no previous callsite will call
crng_finalize_init() unless it is needed, and we can get rid of the
superfluous function parameter.
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Both crng_initialize_primary() and crng_init_try_arch_early() are
only called for the primary_pool. Accessing it directly instead of
through a function parameter simplifies the code.
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
When account() is called, and the amount of entropy dips below
random_write_wakeup_bits, we wake up the random writers, so that they
can write some more in. However, the RNDZAPENTCNT/RNDCLEARPOOL ioctl
sets the entropy count to zero -- a potential reduction just like
account() -- but does not unblock writers. This commit adds the missing
logic to that ioctl to unblock waiting writers.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The rngd kernel thread may sleep indefinitely if the entropy count is
kept above random_write_wakeup_bits by other entropy sources. To make
best use of multiple sources of randomness, mix entropy from hardware
RNGs into the pool at least once within CRNG_RESEED_INTERVAL.
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
There is no good reason to keep genhd.h separate from the main blkdev.h
header that includes it. So fold the contents of genhd.h into blkdev.h
and remove genhd.h entirely.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20220124093913.742411-4-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
kernel/sysctl.c is a kitchen sink where everyone leaves their dirty
dishes, this makes it very difficult to maintain.
To help with this maintenance let's start by moving sysctls to places
where they actually belong. The proc sysctl maintainers do not want to
know what sysctl knobs you wish to add for your own piece of code, we
just care about the core logic.
So move the random sysctls to their own file and use
register_sysctl_init().
[mcgrof@kernel.org: commit log update to justify the move]
Link: https://lkml.kernel.org/r/20211124231435.1445213-3-mcgrof@kernel.org
Signed-off-by: Xiaoming Ni <nixiaoming@huawei.com>
Signed-off-by: Luis Chamberlain <mcgrof@kernel.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Amir Goldstein <amir73il@gmail.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Antti Palosaari <crope@iki.fi>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Benjamin LaHaise <bcrl@kvack.org>
Cc: Clemens Ladisch <clemens@ladisch.de>
Cc: David Airlie <airlied@linux.ie>
Cc: Douglas Gilbert <dgilbert@interlog.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Iurii Zaikin <yzaikin@google.com>
Cc: James E.J. Bottomley <jejb@linux.ibm.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: John Ogness <john.ogness@linutronix.de>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: Julia Lawall <julia.lawall@inria.fr>
Cc: Kees Cook <keescook@chromium.org>
Cc: Lukas Middendorf <kernel@tuxforce.de>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Martin K. Petersen <martin.petersen@oracle.com>
Cc: Paul Turner <pjt@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Phillip Potter <phil@philpotter.co.uk>
Cc: Qing Wang <wangqing@vivo.com>
Cc: "Rafael J. Wysocki" <rafael@kernel.org>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Sebastian Reichel <sre@kernel.org>
Cc: Sergey Senozhatsky <senozhatsky@chromium.org>
Cc: Stephen Kitt <steve@sk2.org>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Now that have_bytes is never modified, we can simplify this function.
First, we move the check for negative entropy_count to be first. That
ensures that subsequent reads of this will be non-negative. Then,
have_bytes and ibytes can be folded into their one use site in the
min_t() function.
Suggested-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This is an old driver that has seen a lot of different eras of kernel
coding style. In an effort to make it easier to code for, unify the
coding style around the current norm, by accepting some of -- but
certainly not all of -- the suggestions from clang-format. This should
remove ambiguity in coding style, especially with regards to spacing,
when code is being changed or amended. Consequently it also makes code
review easier on the eyes, following one uniform style rather than
several.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Jason A. Donenfeld