The way random data is read from hardware has changed from
Octeon CN10KA-B0 and later SoCs onwards. A new set of registers
have been added to read random data and to verify whether the
read data is valid or not. This patch extends and uses
RNM_PF_TRNG_DAT and RNM_PF_TRNG_STS CSRs to read random number
and status for the applicable silicon variants.
Signed-off-by: Bharat Bhushan <[email protected]>
---
drivers/char/hw_random/cn10k-rng.c | 64 ++++++++++++++++++++++++++++--
1 file changed, 60 insertions(+), 4 deletions(-)
diff --git a/drivers/char/hw_random/cn10k-rng.c b/drivers/char/hw_random/cn10k-rng.c
index c1193f85982c..42e44705320c 100644
--- a/drivers/char/hw_random/cn10k-rng.c
+++ b/drivers/char/hw_random/cn10k-rng.c
@@ -23,14 +23,49 @@
#define RNM_PF_RANDOM 0x400
#define RNM_TRNG_RESULT 0x408
+/* Extended TRNG Read and Status Registers */
+#define RNM_PF_TRNG_DAT 0x1000
+#define RNM_PF_TRNG_RES 0x1008
+
struct cn10k_rng {
void __iomem *reg_base;
struct hwrng ops;
struct pci_dev *pdev;
+ /* Octeon CN10K-A A0/A1, CNF10K-A A0/A1 and CNF10K-B A0/B0
+ * does not support extended TRNG registers
+ */
+ bool extended_trng_regs;
};
#define PLAT_OCTEONTX_RESET_RNG_EBG_HEALTH_STATE 0xc2000b0f
+#define PCI_SUBSYS_DEVID_CN10K_A_RNG 0xB900
+#define PCI_SUBSYS_DEVID_CNF10K_A_RNG 0xBA00
+#define PCI_SUBSYS_DEVID_CNF10K_B_RNG 0xBC00
+
+static bool cn10k_is_extended_trng_regs_supported(struct pci_dev *pdev)
+{
+ /* CN10K-A A0/A1 */
+ if ((pdev->subsystem_device == PCI_SUBSYS_DEVID_CN10K_A_RNG) &&
+ (!pdev->revision || (pdev->revision & 0xff) == 0x50 ||
+ (pdev->revision & 0xff) == 0x51))
+ return false;
+
+ /* CNF10K-A A0 */
+ if ((pdev->subsystem_device == PCI_SUBSYS_DEVID_CNF10K_A_RNG) &&
+ (!pdev->revision || (pdev->revision & 0xff) == 0x60 ||
+ (pdev->revision & 0xff) == 0x61))
+ return false;
+
+ /* CNF10K-B A0/B0 */
+ if ((pdev->subsystem_device == PCI_SUBSYS_DEVID_CNF10K_B_RNG) &&
+ (!pdev->revision || (pdev->revision & 0xff) == 0x70 ||
+ (pdev->revision & 0xff) == 0x74))
+ return false;
+
+ return true;
+}
+
static unsigned long reset_rng_health_state(struct cn10k_rng *rng)
{
struct arm_smccc_res res;
@@ -63,9 +98,22 @@ static int check_rng_health(struct cn10k_rng *rng)
return 0;
}
-static void cn10k_read_trng(struct cn10k_rng *rng, u64 *value)
+static size_t cn10k_read_trng(struct cn10k_rng *rng, u64 *value)
{
+ u16 retry_count = 0;
u64 upper, lower;
+ u64 status;
+
+ if (rng->extended_trng_regs) {
+ do {
+ *value = readq(rng->reg_base + RNM_PF_TRNG_DAT);
+ if (*value)
+ return 8;
+ status = readq(rng->reg_base + RNM_PF_TRNG_RES);
+ if (!status && (retry_count++ > 0x1000))
+ return 0;
+ } while (!status);
+ }
*value = readq(rng->reg_base + RNM_PF_RANDOM);
@@ -82,6 +130,7 @@ static void cn10k_read_trng(struct cn10k_rng *rng, u64 *value)
*value = (upper & 0xFFFFFFFF00000000) | (lower & 0xFFFFFFFF);
}
+ return 8;
}
static int cn10k_rng_read(struct hwrng *hwrng, void *data,
@@ -100,7 +149,9 @@ static int cn10k_rng_read(struct hwrng *hwrng, void *data,
size = max;
while (size >= 8) {
- cn10k_read_trng(rng, &value);
+ err = cn10k_read_trng(rng, &value);
+ if (!err)
+ goto out;
*((u64 *)pos) = value;
size -= 8;
@@ -108,7 +159,9 @@ static int cn10k_rng_read(struct hwrng *hwrng, void *data,
}
if (size > 0) {
- cn10k_read_trng(rng, &value);
+ err = cn10k_read_trng(rng, &value);
+ if (!err)
+ goto out;
while (size > 0) {
*pos = (u8)value;
@@ -118,6 +171,7 @@ static int cn10k_rng_read(struct hwrng *hwrng, void *data,
}
}
+out:
return max - size;
}
@@ -144,9 +198,11 @@ static int cn10k_rng_probe(struct pci_dev *pdev, const struct pci_device_id *id)
if (!rng->ops.name)
return -ENOMEM;
- rng->ops.read = cn10k_rng_read;
+ rng->ops.read = cn10k_rng_read;
rng->ops.priv = (unsigned long)rng;
+ rng->extended_trng_regs = cn10k_is_extended_trng_regs_supported(pdev);
+
reset_rng_health_state(rng);
err = devm_hwrng_register(&pdev->dev, &rng->ops);
--
2.17.1
On Thu, May 18, 2023 at 12:17:34PM +0530, Bharat Bhushan wrote:
>
> -static void cn10k_read_trng(struct cn10k_rng *rng, u64 *value)
> +static size_t cn10k_read_trng(struct cn10k_rng *rng, u64 *value)
Since the return value is either 0 or 8, why did you pick size_t
as the return value?
> {
> + u16 retry_count = 0;
> u64 upper, lower;
> + u64 status;
> +
> + if (rng->extended_trng_regs) {
> + do {
> + *value = readq(rng->reg_base + RNM_PF_TRNG_DAT);
> + if (*value)
> + return 8;
> + status = readq(rng->reg_base + RNM_PF_TRNG_RES);
> + if (!status && (retry_count++ > 0x1000))
> + return 0;
> + } while (!status);
> + }
>
> *value = readq(rng->reg_base + RNM_PF_RANDOM);
>
> @@ -82,6 +130,7 @@ static void cn10k_read_trng(struct cn10k_rng *rng, u64 *value)
>
> *value = (upper & 0xFFFFFFFF00000000) | (lower & 0xFFFFFFFF);
> }
> + return 8;
> }
>
> static int cn10k_rng_read(struct hwrng *hwrng, void *data,
> @@ -100,7 +149,9 @@ static int cn10k_rng_read(struct hwrng *hwrng, void *data,
> size = max;
>
> while (size >= 8) {
> - cn10k_read_trng(rng, &value);
> + err = cn10k_read_trng(rng, &value);
> + if (!err)
> + goto out;
It appears that err is either 8 or 0, so it's not really an error.
Could you plesae use a more meaningful name for this variable?
Cheers,
--
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
> -----Original Message-----
> From: Herbert Xu <[email protected]>
> Sent: Wednesday, May 24, 2023 3:26 PM
> To: Bharat Bhushan <[email protected]>
> Cc: [email protected]; [email protected]; [email protected]; linux-
> [email protected]; Sunil Kovvuri Goutham <[email protected]>
> Subject: [EXT] Re: [PATCH] hwrng: cn10k: Add extended trng register support
>
> External Email
>
> ----------------------------------------------------------------------
> On Thu, May 18, 2023 at 12:17:34PM +0530, Bharat Bhushan wrote:
> >
> > -static void cn10k_read_trng(struct cn10k_rng *rng, u64 *value)
> > +static size_t cn10k_read_trng(struct cn10k_rng *rng, u64 *value)
>
> Since the return value is either 0 or 8, why did you pick size_t as the return value?
This function returns number of valid bytes available in "value". 0 means no valid data. size_t is definitely wrong.
I think I will make this function to return bool, true when data is valid and false when data is not valid.
Thanks
-Bharat
>
> > {
> > + u16 retry_count = 0;
> > u64 upper, lower;
> > + u64 status;
> > +
> > + if (rng->extended_trng_regs) {
> > + do {
> > + *value = readq(rng->reg_base + RNM_PF_TRNG_DAT);
> > + if (*value)
> > + return 8;
> > + status = readq(rng->reg_base + RNM_PF_TRNG_RES);
> > + if (!status && (retry_count++ > 0x1000))
> > + return 0;
> > + } while (!status);
> > + }
> >
> > *value = readq(rng->reg_base + RNM_PF_RANDOM);
> >
> > @@ -82,6 +130,7 @@ static void cn10k_read_trng(struct cn10k_rng *rng,
> > u64 *value)
> >
> > *value = (upper & 0xFFFFFFFF00000000) | (lower & 0xFFFFFFFF);
> > }
> > + return 8;
> > }
> >
> > static int cn10k_rng_read(struct hwrng *hwrng, void *data, @@ -100,7
> > +149,9 @@ static int cn10k_rng_read(struct hwrng *hwrng, void *data,
> > size = max;
> >
> > while (size >= 8) {
> > - cn10k_read_trng(rng, &value);
> > + err = cn10k_read_trng(rng, &value);
> > + if (!err)
> > + goto out;
>
> It appears that err is either 8 or 0, so it's not really an error.
> Could you plesae use a more meaningful name for this variable?
>
> Cheers,
> --
> Email: Herbert Xu <[email protected]> Home Page:
> https://urldefense.proofpoint.com/v2/url?u=http-3A__gondor.apana.org.au_-
> 7Eherbert_&d=DwIBAg&c=nKjWec2b6R0mOyPaz7xtfQ&r=PAAlWswPe7d8gHlGb
> CLmy2YezyK7O3Hv_t2heGnouBw&m=Ya8SJ6OhypttG2itpa39dd7kXwEXGgbUOJ2
> Hqi95w7MetfxqRKKFX7aluHaLZpoc&s=pnRAp3oSzsNrbEAHZdLM6Eb9M4tDMxEd
> WLzANKejHwU&e=
> PGP Key: https://urldefense.proofpoint.com/v2/url?u=http-
> 3A__gondor.apana.org.au_-
> 7Eherbert_pubkey.txt&d=DwIBAg&c=nKjWec2b6R0mOyPaz7xtfQ&r=PAAlWswP
> e7d8gHlGbCLmy2YezyK7O3Hv_t2heGnouBw&m=Ya8SJ6OhypttG2itpa39dd7kXwE
> XGgbUOJ2Hqi95w7MetfxqRKKFX7aluHaLZpoc&s=pLpT-
> xdKvVvCI4sp5r8r5qYoKeEx537Gnlkq1dpTrYY&e=