| Message ID | 20220623180555.1345684-1-Jason@zx2c4.com (mailing list archive) |
|---|---|
| State | Not Applicable |
| Delegated to: | Herbert Xu |
| Headers | show |
| Series | [v2] timekeeping: contribute wall clock to rng on time change | expand |
On Thu, Jun 23, 2022 at 08:05:55PM +0200, Jason A. Donenfeld wrote: > The rng's random_init() function contributes the real time to the rng at > boot time, so that events can at least start in relation to something > particular in the real world. But this clock might not yet be set that > point in boot, so nothing is contributed. In addition, the relation > between minor clock changes from, say, NTP, and the cycle counter is > potentially useful entropic data. > > This commit addresses this by mixing in a time stamp on calls to > settimeofday and adjtimex. No entropy is credited in doing so, so it > doesn't make initialization faster, but it is still useful input to > have. > > Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") > Cc: stable@vger.kernel.org > Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> > --- > kernel/time/timekeeping.c | 5 +++++ > 1 file changed, 5 insertions(+) > > diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c > index 8e4b3c32fcf9..89b894b3ede8 100644 > --- a/kernel/time/timekeeping.c > +++ b/kernel/time/timekeeping.c > @@ -23,6 +23,7 @@ > #include <linux/pvclock_gtod.h> > #include <linux/compiler.h> > #include <linux/audit.h> > +#include <linux/random.h> > > #include "tick-internal.h" > #include "ntp_internal.h" > @@ -1331,6 +1332,8 @@ int do_settimeofday64(const struct timespec64 *ts) > goto out; > } > > + add_device_randomness(&ts, sizeof(ts)); > + > tk_set_wall_to_mono(tk, timespec64_sub(tk->wall_to_monotonic, ts_delta)); This is now nested inside: raw_spin_lock_irqsave(&timekeeper_lock, flags); write_seqcount_begin(&tk_core.seq); Could there be a deadlock if random_get_entropy() in add_device_randomness() falls back to reading the monotonic clock? - Eric
On Thu, Jun 23, 2022 at 8:53 PM Eric Biggers <ebiggers@kernel.org> wrote: > > On Thu, Jun 23, 2022 at 08:05:55PM +0200, Jason A. Donenfeld wrote: > > The rng's random_init() function contributes the real time to the rng at > > boot time, so that events can at least start in relation to something > > particular in the real world. But this clock might not yet be set that > > point in boot, so nothing is contributed. In addition, the relation > > between minor clock changes from, say, NTP, and the cycle counter is > > potentially useful entropic data. > > > > This commit addresses this by mixing in a time stamp on calls to > > settimeofday and adjtimex. No entropy is credited in doing so, so it > > doesn't make initialization faster, but it is still useful input to > > have. > > > > Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") > > Cc: stable@vger.kernel.org > > Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> > > --- > > kernel/time/timekeeping.c | 5 +++++ > > 1 file changed, 5 insertions(+) > > > > diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c > > index 8e4b3c32fcf9..89b894b3ede8 100644 > > --- a/kernel/time/timekeeping.c > > +++ b/kernel/time/timekeeping.c > > @@ -23,6 +23,7 @@ > > #include <linux/pvclock_gtod.h> > > #include <linux/compiler.h> > > #include <linux/audit.h> > > +#include <linux/random.h> > > > > #include "tick-internal.h" > > #include "ntp_internal.h" > > @@ -1331,6 +1332,8 @@ int do_settimeofday64(const struct timespec64 *ts) > > goto out; > > } > > > > + add_device_randomness(&ts, sizeof(ts)); > > + > > tk_set_wall_to_mono(tk, timespec64_sub(tk->wall_to_monotonic, ts_delta)); > > This is now nested inside: > > raw_spin_lock_irqsave(&timekeeper_lock, flags); > write_seqcount_begin(&tk_core.seq); > > Could there be a deadlock if random_get_entropy() in add_device_randomness() > falls back to reading the monotonic clock? Also nice find as the raw_spin_lock itself is problematic on RT, because add_device_randomness can take a normal one. I'll do some hoisting. Jason
diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c index 8e4b3c32fcf9..89b894b3ede8 100644 --- a/kernel/time/timekeeping.c +++ b/kernel/time/timekeeping.c @@ -23,6 +23,7 @@ #include <linux/pvclock_gtod.h> #include <linux/compiler.h> #include <linux/audit.h> +#include <linux/random.h> #include "tick-internal.h" #include "ntp_internal.h" @@ -1331,6 +1332,8 @@ int do_settimeofday64(const struct timespec64 *ts) goto out; } + add_device_randomness(&ts, sizeof(ts)); + tk_set_wall_to_mono(tk, timespec64_sub(tk->wall_to_monotonic, ts_delta)); tk_set_xtime(tk, ts); @@ -2430,6 +2433,7 @@ int do_adjtimex(struct __kernel_timex *txc) ret = timekeeping_validate_timex(txc); if (ret) return ret; + add_device_randomness(txc, sizeof(*txc)); if (txc->modes & ADJ_SETOFFSET) { struct timespec64 delta; @@ -2447,6 +2451,7 @@ int do_adjtimex(struct __kernel_timex *txc) audit_ntp_init(&ad); ktime_get_real_ts64(&ts); + add_device_randomness(&ts, sizeof(ts)); raw_spin_lock_irqsave(&timekeeper_lock, flags); write_seqcount_begin(&tk_core.seq);
The rng's random_init() function contributes the real time to the rng at boot time, so that events can at least start in relation to something particular in the real world. But this clock might not yet be set that point in boot, so nothing is contributed. In addition, the relation between minor clock changes from, say, NTP, and the cycle counter is potentially useful entropic data. This commit addresses this by mixing in a time stamp on calls to settimeofday and adjtimex. No entropy is credited in doing so, so it doesn't make initialization faster, but it is still useful input to have. Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Cc: stable@vger.kernel.org Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> --- kernel/time/timekeeping.c | 5 +++++ 1 file changed, 5 insertions(+)