Skip to contentSkip to conversation

tpm: Actually fail on TPM errors during "get random"

PATCH Superseded Kees Cook 2 participants 2 messages Apr 1, 2019

Subsystems (2)
Keys-trusted TPM device driver

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Kees Cook

author
#0 lore raw

PATCH v2 1/1 tpm: Actually fail on TPM errors during "get random"

A "get random" may fail with a TPM error, but those codes were returned as-is to the caller, which assumed the result was the number of bytes that had been written to the target buffer, which could lead to a kernel heap memory exposure and over-read.

This fixes tpm1_get_random() and tpm2_get_random() to mask positive TPM errors into -EIO while still passing through any negative errors.

[   18.092103] tpm tpm0: A TPM error (379) occurred attempting get random
[   18.092106] usercopy: Kernel memory exposure attempt detected from SLUB object 'kmalloc-64' (offset 0, size 379)!
Link https://bugzilla.redhat.com/show_bug.cgi?id=1650989
Reported-by Phil Baker <[email protected]>
Reported-by Craig Robson <[email protected]>
Fixes 7aee9c52d7ac ("tpm: tpm1: rewrite tpm1_get_random() using tpm_buf structure")
Fixes ce63c05b664e ("tpm: migrate tpm2_get_random() to use struct tpm_buf")
Cc Laura Abbott <[email protected]>
Cc Tomas Winkler <[email protected]>
Cc Jarkko Sakkinen <[email protected]>
Cc [email protected]
Signed-off-by Kees Cook <[email protected]>

---
v2: - also fix tpm2 implementation (Jason Gunthorpe)

---
 drivers/char/tpm/tpm1-cmd.c | 7 +++++--
 drivers/char/tpm/tpm2-cmd.c | 7 +++++--
 2 files changed, 10 insertions(+), 4 deletions(-)
drivers/char/tpm/tpm1-cmd.c +5 -2
index 85dcf2654d11..8e17d4f30aeb 100644
@@ -510,7 +510,7 @@ struct tpm1_get_random_out {
  *
  * Return:
  * *  number of bytes read
- * * -errno or a TPM return code otherwise
+ * * -errno (positive TPM return codes are masked to -EIO)
  */
 int tpm1_get_random(struct tpm_chip *chip, u8 *dest, size_t max)
 {

@@ -524,7 +524,7 @@ int tpm1_get_random(struct tpm_chip *chip, u8 *dest, size_t max)
 
 	rc = tpm_buf_init(&buf, TPM_TAG_RQU_COMMAND, TPM_ORD_GET_RANDOM);
 	if (rc)
-		return rc;
+		goto fail;
 
 	do {
 		tpm_buf_append_u32(&buf, num_bytes);

@@ -559,6 +559,9 @@ int tpm1_get_random(struct tpm_chip *chip, u8 *dest, size_t max)
 	rc = total ? (int)total : -EIO;
 out:
 	tpm_buf_destroy(&buf);
+fail:
+	if (rc > 0)
+		rc = -EIO;
 	return rc;
 }
drivers/char/tpm/tpm2-cmd.c +5 -2
index e74c5b7b64bf..dd0677f50f9c 100644
@@ -301,7 +301,7 @@ struct tpm2_get_random_out {
  *
  * Return:
  *   size of the buffer on success,
- *   -errno otherwise
+ *   -errno otherwise ((positive TPM return codes are masked to -EIO)
  */
 int tpm2_get_random(struct tpm_chip *chip, u8 *dest, size_t max)
 {

@@ -319,7 +319,7 @@ int tpm2_get_random(struct tpm_chip *chip, u8 *dest, size_t max)
 
 	err = tpm_buf_init(&buf, 0, 0);
 	if (err)
-		return err;
+		goto fail;
 
 	do {
 		tpm_buf_reset(&buf, TPM2_ST_NO_SESSIONS, TPM2_CC_GET_RANDOM);

@@ -352,6 +352,9 @@ int tpm2_get_random(struct tpm_chip *chip, u8 *dest, size_t max)
 	return total ? total : -EIO;
 out:
 	tpm_buf_destroy(&buf);
+fail:
+	if (err > 0)
+		erro = -EIO;
 	return err;
 }
Signature · 5 lines
2.17.1


-- 
Kees Cook
maintainer
#1 lore raw

drivers/char/tpm/tpm1-cmd.c int tpm1_get_random(struct tpm_chip *chip, u8 · line 567

Quoted 10 lines · Kees Cook
drivers/char/tpm/tpm1-cmd.c +3 -0
@@ -559,6 +559,9 @@ int tpm1_get_random(struct tpm_chip *chip, u8
 *dest, size_t max)
 	rc = total ? (int)total : -EIO;
 out:
 	tpm_buf_destroy(&buf);
+fail:
+	if (rc > 0)
+		rc = -EIO;
 	return rc;
 }

No: same problem. If we're successful rc is set to total (a positive integer) so as it falls through to fail: it's converted to -EIO which means we never return success.

James