Skip to contentSkip to conversation

hw_random: explicit ordering of initcalls

PATCH In review Dominik Brodowski 2 participants 8 messages Jan 24, 2022 – Feb 5, 2022

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/6 hw_random: explicit ordering of initcalls

hw-random device drivers depend on the hw-random core being initialized. Make this ordering explicit, also for the case these drivers are built-in. As the core itself depends on misc_register() which is set up at subsys_initcall time, advance the initialization of the core (only) to the fs_initcall() level.

Cc Matt Mackall <[email protected]>
Cc Herbert Xu <[email protected]>
Cc Jason A. Donenfeld <[email protected]>
---
 drivers/char/hw_random/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/char/hw_random/core.c +1 -1
index a3db27916256..e860e044b19e 100644
@@ -638,7 +638,7 @@ static void __exit hwrng_modexit(void)
 	unregister_miscdev();
 }
 
-module_init(hwrng_modinit);
+fs_initcall(hwrng_modinit); /* depends on misc_register() */
 module_exit(hwrng_modexit);
 
 MODULE_DESCRIPTION("H/W Random Number Generator (RNG) driver");
Signature · 1 line
2.34.1
#1 lore raw

PATCH 2/6 hw_random: read() callback must be called for size of 32 or more bytes

According to <linux/hw_random.h>, the @max parameter of the ->read callback "is a multiple of 4 and >= 32 bytes". That promise was not kept by add_early_randomness(), which only asked for 16 bytes. As rng_buffer_size() is at least 32, we can simply ask for 32 bytes.

Cc Matt Mackall <[email protected]>
Cc Herbert Xu <[email protected]>
Cc Jason A. Donenfeld <[email protected]>
---
 drivers/char/hw_random/core.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/char/hw_random/core.c +1 -2
index e860e044b19e..c2d260b5dd92 100644
@@ -64,10 +64,9 @@ 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());
 
 	mutex_lock(&reading_mutex);
-	bytes_read = rng_get_data(rng, rng_buffer, size, 0);
+	bytes_read = rng_get_data(rng, rng_buffer, 32, 0);
 	mutex_unlock(&reading_mutex);
 	if (bytes_read > 0)
 		add_device_randomness(rng_buffer, bytes_read);
Signature · 1 line
2.34.1
#2 lore raw

PATCH 3/6 hw_random: use rng_fillbuf in add_early_randomness()

In reply to Dominik Brodowski

Using rng_buffer in add_early_randomness() may race with rng_dev_read(). Use rng_fillbuf instead, as it is otherwise only used within the kernel by hwrng_fillfn() and therefore never exposed to userspace.

Cc Matt Mackall <[email protected]>
Cc Herbert Xu <[email protected]>
Cc Jason A. Donenfeld <[email protected]>
---
 drivers/char/hw_random/core.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/char/hw_random/core.c +2 -2
index c2d260b5dd92..89891ac87af0 100644
@@ -66,10 +66,10 @@ static void add_early_randomness(struct hwrng *rng)
 	int bytes_read;
 
 	mutex_lock(&reading_mutex);
-	bytes_read = rng_get_data(rng, rng_buffer, 32, 0);
+	bytes_read = rng_get_data(rng, rng_fillbuf, 32, 0);
 	mutex_unlock(&reading_mutex);
 	if (bytes_read > 0)
-		add_device_randomness(rng_buffer, bytes_read);
+		add_device_randomness(rng_fillbuf, bytes_read);
 }
 
 static inline void cleanup_rng(struct kref *kref)
Signature · 1 line
2.34.1
#3 lore raw

PATCH 4/6 hw_random: only set cur_rng_set_by_user if it is working

In reply to Dominik Brodowski

In case the user-specified rng device is not working, it is not used; therefore cur_rng_set_by_user must not be set to 1.

Cc Matt Mackall <[email protected]>
Cc Herbert Xu <[email protected]>
Cc Jason A. Donenfeld <[email protected]>
---
 drivers/char/hw_random/core.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/char/hw_random/core.c +2 -1
index 89891ac87af0..9405fcdace38 100644
@@ -335,8 +335,9 @@ static ssize_t rng_current_store(struct device *dev,
 	} else {
 		list_for_each_entry(rng, &rng_list, list) {
 			if (sysfs_streq(rng->name, buf)) {
-				cur_rng_set_by_user = 1;
 				err = set_current_rng(rng);
+				if (!err)
+					cur_rng_set_by_user = 1;
 				break;
 			}
 		}
Signature · 1 line
2.34.1
#4 lore raw

PATCH 5/6 hw_random: break out of hwrng_fillfn if current rng is not trusted

In reply to Dominik Brodowski

For two reasons, current_quality may become zero within the rngd kernel thread: (1) The user lowers current_quality to 0 by writing to the sysfs module parameter file (note that increasing the quality from zero is without effect at the moment), or (2) there are two or more hwrng devices registered, and those which provide quality>0 are unregistered, but one with quality==0 remains.

If current_quality is 0, the randomness is not trusted and cannot help to increase the entropy count. That will lead to continuous calls to the hwrngd thread and continuous stirring of the input pool with untrusted bits.

Cc Matt Mackall <[email protected]>
Cc Herbert Xu <[email protected]>
Cc Jason A. Donenfeld <[email protected]>
---
 drivers/char/hw_random/core.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/char/hw_random/core.c +3 -0
index 9405fcdace38..bc9f95cbac92 100644
@@ -429,6 +429,9 @@ static int hwrng_fillfn(void *unused)
 	while (!kthread_should_stop()) {
 		struct hwrng *rng;
 
+		if (!current_quality)
+			break;
+
 		rng = get_current_rng();
 		if (IS_ERR(rng) || !rng)
 			break;
Signature · 1 line
2.34.1
#5 lore raw

PATCH 6/6 hw_random: credit entropy for low quality sources of randomness

In reply to Dominik Brodowski

In case the entropy quality is low, there may be less than one bit to credit in the call to add_hwgenerator_randomness(): The number of bytes returned by rng_get_data() multiplied by the current quality (in entropy bits per 1024 bits of input) must be larger than 128 to credit at least one bit. However, imx-rngc.c sets the quality to 19, but may return less than 32 bytes; hid_u2fzero.c sets the quality to 1; and users may override the quality setting manually.

In case there is less than one bit to credit, keep track of it and add that credit to the next iteration.

Cc Matt Mackall <[email protected]>
Cc Herbert Xu <[email protected]>
Cc Jason A. Donenfeld <[email protected]>
---
 drivers/char/hw_random/core.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
drivers/char/hw_random/core.c +10 -1
index bc9f95cbac92..6d7f05641c7c 100644
@@ -427,6 +427,7 @@ static int hwrng_fillfn(void *unused)
 	long rc;
 
 	while (!kthread_should_stop()) {
+		size_t entropy, entropy_credit = 0; /* in 1/1024 of a bit */
 		struct hwrng *rng;
 
 		if (!current_quality)
@@ -445,9 +446,17 @@ static int hwrng_fillfn(void *unused)
 			msleep_interruptible(10000);
 			continue;
 		}
+
+		/* If we cannot credit at least one bit of entropy,
+		 * keep track of the remainder for the next iteration
+		 */
+		entropy = rc * current_quality * 8 + entropy_credit;
+		if ((entropy >> 10) == 0)
+			entropy_credit = entropy;
+
 		/* Outside lock, sure, but y'know: randomness. */
 		add_hwgenerator_randomness((void *)rng_fillbuf, rc,
-					   rc * current_quality * 8 >> 10);
+					   entropy >> 10);
 	}
 	hwrng_fill = NULL;
 	return 0;
Signature · 1 line
2.34.1
#6 lore raw

PATCH v2 6/6 hw_random: credit entropy for low quality sources of randomness

In case the entropy quality is low, there may be less than one bit to credit in the call to add_hwgenerator_randomness(): The number of bytes returned by rng_get_data() multiplied by the current quality (in entropy bits per 1024 bits of input) must be larger than 128 to credit at least one bit. However, imx-rngc.c sets the quality to 19, but may return less than 32 bytes; hid_u2fzero.c sets the quality to 1; and users may override the quality setting manually.

In case there is less than one bit to credit, keep track of it and add that credit to the next iteration.

Cc Herbert Xu <[email protected]>
Cc Jason A. Donenfeld <[email protected]>
---
 drivers/char/hw_random/core.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

This patch needed an update, as noted by the kernel test robot: the initialiation of entropy_credit = 0 must be outside the loop.

That's the only change between v1 and v2 of this patch. The other five patches sent earlier do not (yet?) need an update.

drivers/char/hw_random/core.c +10 -1
index bc9f95cbac92..f327f7493585 100644
@@ -424,6 +424,7 @@ static int __init register_miscdev(void)
 
 static int hwrng_fillfn(void *unused)
 {
+	size_t entropy, entropy_credit = 0; /* in 1/1024 of a bit */
 	long rc;
 
 	while (!kthread_should_stop()) {
@@ -445,9 +446,17 @@ static int hwrng_fillfn(void *unused)
 			msleep_interruptible(10000);
 			continue;
 		}
+
+		/* If we cannot credit at least one bit of entropy,
+		 * keep track of the remainder for the next iteration
+		 */
+		entropy = rc * current_quality * 8 + entropy_credit;
+		if ((entropy >> 10) == 0)
+			entropy_credit = entropy;
+
 		/* Outside lock, sure, but y'know: randomness. */
 		add_hwgenerator_randomness((void *)rng_fillbuf, rc,
-					   rc * current_quality * 8 >> 10);
+					   entropy >> 10);
 	}
 	hwrng_fill = NULL;
 	return 0;
Signature · 1 line
2.35.0
maintainer
#7 lore raw
In reply to Dominik Brodowski
Quoted 7 lines · Dominik Brodowski

hw-random device drivers depend on the hw-random core being initialized. Make this ordering explicit, also for the case these drivers are built-in. As the core itself depends on misc_register() which is set up at subsys_initcall time, advance the initialization of the core (only) to the fs_initcall() level.

Quoted 7 lines
Cc Matt Mackall <[email protected]>
Cc Herbert Xu <[email protected]>
Cc Jason A. Donenfeld <[email protected]>
---
 drivers/char/hw_random/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

All applied. Thanks.

Signature · 3 lines
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt