On Sat, 2018-12-15 at 18:11 +0100, Michael Niewöhner wrote: > On Thu, 2018-12-13 at 12:50 +0800, Louis Collard wrote: > > On Sun, Nov 18, 2018 at 4:15 AM Michael Niewöhner <linux@xxxxxxxxxxxxxx> > > wrote: > > > > > > Hi Louis, > > > > > > On Wed, 2018-09-26 at 11:24 +0800, Louis Collard wrote: > > > > Some systems, for example embedded systems, do not generate > > > > enough entropy on boot through interrupts, and boot may be blocked for > > > > several minutes waiting for a call to getrandom to complete. > > > > > > > > Currently, random data is read from a hwrng when it is registered, > > > > and is loaded into primary_crng. This data is treated in the same > > > > way as data that is device-specific but otherwise unchanging, and > > > > so primary_crng cannot become initialized with the data from the > > > > hwrng. > > > > > > > > This change causes the data initially read from the hwrng to be > > > > treated the same as subsequent data that is read from the hwrng if > > > > it's quality score is non-zero. > > > > > > > > The implications of this are: > > > > > > > > The data read from hwrng can cause primary_crng to become > > > > initialized, therefore avoiding problems of getrandom blocking > > > > on boot. > > > > > > > > Calls to getrandom (with GRND_RANDOM) may be using entropy > > > > exclusively (or in practise, almost exclusively) from the hwrng. > > > > > > > > Regarding the latter point; this behavior is the same as if a > > > > user specified a quality score of 1 (bit of entropy per 1024 bits) > > > > so hopefully this is not too scary a change to make. > > > > > > > > This change is the result of the discussion here: > > > > https://patchwork.kernel.org/patch/10453893/ > > > > > > > > Signed-off-by: Louis Collard <louiscollard@xxxxxxxxxxxx> > > > > Acked-by: Jarkko Sakkinen <jarkko.sakkinen@xxxxxxxxxxxxxxx> > > > > --- > > > > drivers/char/hw_random/core.c | 9 +++++++-- > > > > 1 file changed, 7 insertions(+), 2 deletions(-) > > > > > > > > diff --git a/drivers/char/hw_random/core.c > > > > b/drivers/char/hw_random/core.c > > > > index aaf9e5afaad4..47f358aa0c3d 100644 > > > > --- a/drivers/char/hw_random/core.c > > > > +++ b/drivers/char/hw_random/core.c > > > > @@ -24,6 +24,7 @@ > > > > #include <linux/sched.h> > > > > #include <linux/slab.h> > > > > #include <linux/uaccess.h> > > > > +#include <crypto/chacha20.h> > > > > > > > > #define RNG_MODULE_NAME "hw_random" > > > > > > > > @@ -64,13 +65,17 @@ static size_t rng_buffer_size(void) > > > > static void add_early_randomness(struct hwrng *rng) > > > > { > > > > int bytes_read; > > > > - size_t size = min_t(size_t, 16, rng_buffer_size()); > > > > + /* Read enough to initialize crng. */ > > > > + size_t size = 2*CHACHA20_KEY_SIZE; > > > > > > > > mutex_lock(&reading_mutex); > > > > bytes_read = rng_get_data(rng, rng_buffer, size, 1); > > > > mutex_unlock(&reading_mutex); > > > > if (bytes_read > 0) > > > > - add_device_randomness(rng_buffer, bytes_read); > > > > + /* Allow crng to become initialized, but do not add > > > > + * entropy to the pool. > > > > + */ > > > > + add_hwgenerator_randomness(rng_buffer, bytes_read, 0); > > > > } > > > > > > > > static inline void cleanup_rng(struct kref *kref) > > > > > > I found your patch by chance, searching for a solution for crng init delay > > > on my > > > headless machine. Unfortunately it hardly makes any difference for me. > > > With > > > the > > > patch the system hangs for about 80s instead of 120s until the "crng init > > > done" > > > message.In contrast, doing a `cat /dev/hwrng >/dev/random` or running rngd > > > initializes the crng instantly. > > > > > > Isn't that delay the problem this patch tries to fix? Any idea what is > > > wrong > > > here? > > > > > > Thanks! > > > > > > Best regards > > > Michael > > > > > > > > > > Yes that is the problem this is trying to address. My guess would be > > rng_get_data() is not returning as much data as requested, so the > > delay is reduced but not eliminated. Looking at implementation of > > rng_get_data() it appears this could be caused by device support for > > read() vs data_read(). I don't have a good feel for whether looping to > > retrieve more data here would be acceptable, it is certainly a bigger > > change than currently proposed. > > > > Thanks, > > Louis > > Hi Louis, > > that is what I thought first, too, but I was able to verify that 64 bytes are > read as expected. > > It seems this is exactly what David noticed in your discussion about the > quality > module parameter (https://patchwork.kernel.org/patch/10453893/#22130681): > > > Interesting. > > > > add_hwgenereator_randomness() will call crng_fast_load(), regardless > > of entropy estimate/quality, if crng_init is 0. So initializing > > crng_init from the hwrng, regardless of quality, is already the > > intent. > > > > But hw_random only calls add_hwgenerator_randomness() if > > current_quality > 0, via the hwrng_fillfn() kthread. > > > > All that to say, I agree. add_early_randomness() should (indirectly) > > call crng_fast_load(), like add_hwgenerator_randomness() does. > > When I set rng_quality=1024, the crng does get initialized more or less > instantly. > > > dmesg with default rng_quality=0: > > [ 0.003831] ACPI: TPM2 0x000000009E0B7F70 000034 (v03 LENOVO TC- > S06 00001260 AMI 00000000) > [ 0.161803] random: get_random_bytes called from start_kernel+0x8f/0x50e > with > crng_init=0 > [ 3.590433] tpm_tis MSFT0101:00: 2.0 TPM (device-id 0xFE, rev-id 2) > [ 3.644148] random: fast init done > [ 85.183006] random: crng init done > > > dmesg with rng_quality patch: > > [ 0.003837] ACPI: TPM2 0x000000009E0B7F70 000034 (v03 LENOVO TC- > S06 00001260 AMI 00000000) > [ 0.162136] random: get_random_bytes called from start_kernel+0x8f/0x50e > with > crng_init=0 > [ 3.582675] tpm_tis MSFT0101:00: 2.0 TPM (device-id 0xFE, rev-id 2) > [ 3.636408] random: fast init done > [ 3.650355] random: crng init done > > > Test patch: > > diff --git a/drivers/char/tpm/tpm_tis.c b/drivers/char/tpm/tpm_tis.c > index f08949a5f678..59e5a8753ba1 100644 > --- a/drivers/char/tpm/tpm_tis.c > +++ b/drivers/char/tpm/tpm_tis.c > @@ -214,6 +214,8 @@ static int tpm_tis_init(struct device *dev, struct > tpm_info > *tpm_info) > if (itpm || is_itpm(ACPI_COMPANION(dev))) > phy->priv.flags |= TPM_TIS_ITPM_WORKAROUND; > > + priv->rng_quality = 1; > + > return tpm_tis_core_init(dev, &phy->priv, irq, &tpm_tcg, > ACPI_HANDLE(dev)); > } > > > > Thanks > Michael Hi Louis, I think I might know what is causing the delay of crng init. Your patch is working as expected but only for fast init. add_hwgenerator_randomness calls crng_fast_load which does stage 1 initialization of the crng ("fast init done" -> crng_init=1). Currently no entropy comes from the TPM at all for 1) the second stage init / reseed 2) later operation because tpm_tis sets no quality while the default in tpm-chip is quality=0 and so khwrngd will not be started in hwrng_init. This is why setting the quality to 1024 instantly leads to setting crng_init=2 and reseeding the crng ("crng init done"). That means the current default value of rng_quality=0 renders the tpm hwrng completely useless.
