Skip to contentSkip to conversation

crypto: marvell/octeontx - prevent integer overflows

PATCH In review Dan Carpenter 2 participants 2 messages Sep 19, 2022 – Sep 30, 2022

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 crypto: marvell/octeontx - prevent integer overflows

The "code_length" value comes from the firmware file. If your firmware is untrusted realistically there is probably very little you can do to protect yourself. Still we try to limit the damage as much as possible. Also Smatch marks any data read from the filesystem as untrusted and prints warnings if it not capped correctly.

The "code_length * 2" can overflow. The round_up(ucode_size, 16) + sizeof() expression can overflow too. Prevent these overflows.

Fixes d9110b0b01ff ("crypto: marvell - add support for OCTEON TX CPT engine")

---
v2: In the first commit, I accidentally added some unnecessary checks.
Those were only drafts and I sent them uninitentionally. Sorry!

 .../crypto/marvell/octeontx/otx_cptpf_ucode.c  | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)
drivers/crypto/marvell/octeontx/otx_cptpf_ucode.c +16 -2
index 23c6edc70914..df9c2b8747e6 100644
@@ -286,6 +286,7 @@ static int process_tar_file(struct device *dev,
 	struct tar_ucode_info_t *tar_info;
 	struct otx_cpt_ucode_hdr *ucode_hdr;
 	int ucode_type, ucode_size;
+	unsigned int code_length;
 
 	/*
 	 * If size is less than microcode header size then don't report
@@ -303,7 +304,13 @@ static int process_tar_file(struct device *dev,
 	if (get_ucode_type(ucode_hdr, &ucode_type))
 		return 0;
 
-	ucode_size = ntohl(ucode_hdr->code_length) * 2;
+	code_length = ntohl(ucode_hdr->code_length);
+	if (code_length >= INT_MAX / 2) {
+		dev_err(dev, "Invalid code_length %u\n", code_length);
+		return -EINVAL;
+	}
+
+	ucode_size = code_length * 2;
 	if (!ucode_size || (size < round_up(ucode_size, 16) +
 	    sizeof(struct otx_cpt_ucode_hdr) + OTX_CPT_UCODE_SIGN_LEN)) {
 		dev_err(dev, "Ucode %s invalid size\n", filename);
@@ -886,6 +893,7 @@ static int ucode_load(struct device *dev, struct otx_cpt_ucode *ucode,
 {
 	struct otx_cpt_ucode_hdr *ucode_hdr;
 	const struct firmware *fw;
+	unsigned int code_length;
 	int ret;
 
 	set_ucode_filename(ucode, ucode_filename);
@@ -896,7 +904,13 @@ static int ucode_load(struct device *dev, struct otx_cpt_ucode *ucode,
 	ucode_hdr = (struct otx_cpt_ucode_hdr *) fw->data;
 	memcpy(ucode->ver_str, ucode_hdr->ver_str, OTX_CPT_UCODE_VER_STR_SZ);
 	ucode->ver_num = ucode_hdr->ver_num;
-	ucode->size = ntohl(ucode_hdr->code_length) * 2;
+	code_length = ntohl(ucode_hdr->code_length);
+	if (code_length >= INT_MAX / 2) {
+		dev_err(dev, "Ucode invalid code_length %u\n", code_length);
+		ret = -EINVAL;
+		goto release_fw;
+	}
+	ucode->size = code_length * 2;
 	if (!ucode->size || (fw->size < round_up(ucode->size, 16)
 	    + sizeof(struct otx_cpt_ucode_hdr) + OTX_CPT_UCODE_SIGN_LEN)) {
 		dev_err(dev, "Ucode %s invalid size\n", ucode_filename);
Signature · 1 line
2.35.1
maintainer
#1 lore raw
Quoted 9 lines · Dan Carpenter

The "code_length" value comes from the firmware file. If your firmware is untrusted realistically there is probably very little you can do to protect yourself. Still we try to limit the damage as much as possible. Also Smatch marks any data read from the filesystem as untrusted and prints warnings if it not capped correctly.

The "code_length * 2" can overflow. The round_up(ucode_size, 16) + sizeof() expression can overflow too. Prevent these overflows.

Quoted 8 lines
Fixes d9110b0b01ff ("crypto: marvell - add support for OCTEON TX CPT engine")

---
v2: In the first commit, I accidentally added some unnecessary checks.
Those were only drafts and I sent them uninitentionally. Sorry!

 .../crypto/marvell/octeontx/otx_cptpf_ucode.c  | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

Patch 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