20260610-linuxkm-fenrir by douzzer · Pull Request #10676 · wolfSSL/wolfssl

douzzer · 2026-06-12T23:19:11Z

Fixes for numerous "low" Fenrir issues in linuxkm/. See commit messages for the details.

tested with

wolfssl-multi-test.sh ...
check-source-text
quantum-safe-wolfssl-all-crypto-only-noasm-fips-dev-linuxkm-next-clang-tidy
quantum-safe-wolfssl-all-crypto-only-intelasm-sp-asm-sp-fips-dev-linuxkm-next-insmod

…oth RSA Paths

… Re-Init

…-Test Error Messages

F-1880: Unsigned word32 dec_len Silently Wraps Negative wc_RsaSSL_Verify Error Codes, Returning Wrong Errno on RSA Signature Verification Failure F-1881: word32 sig_len Silently Wraps Negative wc_RsaSSL_Sign Error in Self-Test Functions, Bypassing <= 0 Guard F-1882: word32 priv_len / pub_len Silently Wrap Negative wc_RsaKeyToDer / wc_RsaKeyToPublicDer Errors Across Three RSA Self-Test Functions

…ock encrypt path

also ForceZeros various other potentially sensitive allocations before freeing them, and uses unconditional free() per current libwolfssl best practice.

…or Code to Kernel LKCAPI

…hash_len

in AesGcmCrypt_1() and AesCcmCrypt_1(), check for overflow on assoclen+cryptlen in both encrypt and decrypt modes; in linuxkm_test_kpp_driver(), return MEMORY_E, not -ENOMEM; in km_direct_rsa_dec(), only update req->dst_len for -EOVERFLOW, not for -EINVAL.

F-1428: Missing ForceZero on sg_buf Containing Decrypted Plaintext in AES-GCM Non-Stream Path F-3293: AES-CCM Non-Contiguous SG Path Missing ForceZero on sg_buf Containing Decrypted Plaintext

…ating skcipher_walk Error

F-1433: AES CBC/CFB Self-Test Functions Silently Continue After enc2/dec2 Allocation Failure F-1434: linuxkm_test_aesgcm Silently Returns Success on Kernel Crypto Allocation Failure

… in km_hmac_init Error Path

…ith Plain Assignment Instead of wc_ForceZero

…uffer Overflow

wolfSSL-Fenrir-bot

Fenrir Automated Review — PR #10676

Scan targets checked: linuxkm-bugs, linuxkm-src

Findings: 1
1 finding(s) posted as inline comments (see file-level comments below)

This review was generated automatically by Fenrir. Findings are non-blocking.

wolfSSL-Fenrir-bot · 2026-06-12T23:31:29Z

        WC_SANITIZE_ENABLE();
        buf = 0;
    }
+    wc_ForceZero(&buf, 0, sizeof buf);


🔴 [High] wc_ForceZero called with 3 arguments instead of 2 · API contract violations

wc_ForceZero is prototyped as void wc_ForceZero(void *mem, size_t len) (2 params), but it is called as wc_ForceZero(&buf, 0, sizeof buf) with 3 arguments, which fails to compile when WC_LINUXKM_RDSEED_IN_GLUE_LAYER is defined (the guarding config).

Fix: Drop the spurious 0 argument: call wc_ForceZero(&buf, sizeof buf).

dgarske

Skoll Multi-Scan Review

Modes: review + review-securityOverall recommendation: REQUEST_CHANGES
Findings: 4 total — 4 posted, 0 skipped
4 finding(s) posted as inline comments (see file-level comments below)

Posted findings

[High] [review+review-security] wc_ForceZero() called with 3 arguments (compile error / broken entropy zeroization fix) — linuxkm/module_hooks.c:475
[Low] [review] Trailing comma instead of period in error message — linuxkm/lkcapi_aes_glue.c:3643
[Low] [review-security] km_direct_rsa_dec no longer reports required dst_len on -EINVAL path — linuxkm/lkcapi_rsa_glue.c:799
[Info] [review-security] x86_vector_register_glue.c allocation-failure log size argument changed — linuxkm/x86_vector_register_glue.c:72

Review generated by Skoll

dgarske · 2026-06-12T23:34:54Z

        WC_SANITIZE_ENABLE();
        buf = 0;
    }
+    wc_ForceZero(&buf, 0, sizeof buf);


🔴 [High] wc_ForceZero() called with 3 arguments (compile error / broken entropy zeroization fix) · Logic

Both the review and review-security modes independently flag this. The newly added stack-entropy scrub call invokes wc_ForceZero(&buf, 0, sizeof buf) with THREE arguments, but wc_ForceZero is declared as void wc_ForceZero(void *mem, size_t len) (wolfssl/wolfcrypt/memory.h:335; defined wolfcrypt/src/memory.c:1681) and takes only TWO. The stray middle 0 looks like a memset(&buf, 0, sizeof buf) signature carried over by mistake. This is a C constraint violation that GCC reports as error: too many arguments to function 'wc_ForceZero', and under the kernel build's -Werror it fails to compile. The same file already shows the correct 2-argument form at line 523: wc_ForceZero(&rndTmp, sizeof(rndTmp));. The function IntelRDseed64_r is compiled only under WC_LINUXKM_RDSEED_IN_GLUE_LAYER (guard at module_hooks.c:413-528) with HAVE_INTEL_RDSEED/HAVE_AMD_RDSEED (e.g. FIPS v5 RDSEED-in-glue builds), which is why the PR's tested configs (which took the WC_LINUXKM_WOLFENTROPY_IN_GLUE_LAYER branch at line 405) did not surface the break. Net effect: (1) the module will not build in the affected RDSEED-glue configs, and (2) the F-3294 entropy-zeroization hardening it implements is malformed. No 3-argument wc_ForceZero/ForceZero macro exists anywhere in the tree.

Fix: Drop the stray 0 argument so the call matches the 2-arg prototype: wc_ForceZero(&buf, sizeof buf);. Ideally add a CI config that defines WC_LINUXKM_RDSEED_IN_GLUE_LAYER (HAVE_INTEL_RDSEED / HAVE_AMD_RDSEED) so this path is compiled.

Caught locally too. Fixed, to commit&push in a minute.

dgarske · 2026-06-12T23:34:54Z

-        pr_err("error: malloc src failed: %ld\n",
-               PTR_ERR(src));
+        pr_err("error: malloc src failed,\n");
+        ret = MEMORY_E;


🔵 [Low] Trailing comma instead of period in error message · style

The reworked malloc-failure message reads pr_err("error: malloc src failed,\n"); with a trailing comma, whereas the sibling messages in the same function use a period (e.g. "error: malloc dst failed.\n"). Purely cosmetic; the removal of the bogus PTR_ERR(src) (which is meaningless for a NULL malloc return) and the added ret = MEMORY_E; are correct.

Fix: Change the trailing comma to a period for consistency: pr_err("error: malloc src failed.\n");. Not blocking.

Fixed. Typo obviously.

dgarske · 2026-06-12T23:34:54Z

@@ -793,11 +797,12 @@ static int km_direct_rsa_dec(struct akcipher_request *req)
    }

    if (req->dst_len != (unsigned int)ctx->key_len) {


🔵 [Low] km_direct_rsa_dec no longer reports required dst_len on -EINVAL path · Logic

The diff moved req->dst_len = ctx->key_len; inside the -EOVERFLOW branch. BEFORE, req->dst_len was set to ctx->key_len for both the -EOVERFLOW (dst_len too small) and -EINVAL (dst_len == 0 or dst_len > key_len) branches. AFTER, on the -EINVAL branch req->dst_len is left unchanged. This is a behavioral change but is benign and arguably more correct: the kernel akcipher convention only requires the driver to report the needed size on -EOVERFLOW so the caller can retry; -EINVAL is a hard error. No out-of-bounds or security impact -- flagged only as a behavior-delta note.

Fix: No change required if intentional. If callers were relying on the previous behavior of always receiving key_len in req->dst_len, restore the unconditional assignment; otherwise document the deliberate change.

This was indeed intentional for API compliance.

dgarske · 2026-06-12T23:34:54Z

@@ -72,7 +72,7 @@ WARN_UNUSED_RESULT int allocate_wolfcrypt_linuxkm_fpu_states(void)
    if (! wc_linuxkm_fpu_states) {


⚪ [Info] x86_vector_register_glue.c allocation-failure log size argument changed · Logic

The error message argument was changed from nr_cpu_ids * sizeof(struct fpu_state *) to nr_cpu_ids * sizeof(wc_linuxkm_fpu_states[0]), making the logged byte count track the actual element type of the allocation. This is a purely cosmetic logging accuracy improvement with no functional or security impact; noted for completeness.

Fix: No action needed; the change is correct.

github-actions · 2026-06-12T23:42:19Z

linuxkm/lkcapi_aes_glue.c: fix typo in linuxkm_test_aesgcm() error message.

douzzer added 25 commits June 12, 2026 18:14

linuxkm/lkcapi_aes_glue.c: clean up pr_err()s on allocation failures.

546ac0a

fix F-672: Copy-Paste Duplicate: pkcs1_sha384 Unregistered Twice in B…

1c9abac

…oth RSA Paths

fix F-673: Wrong sizeof Type in Allocation Failure Error Message

9c1d9c1

fix F-705: Missing wc_RsaSetRNG in sig_alg km_pkcs1_set_pub After Key…

f48efe5

… Re-Init

fix F-1424: Sysfs Node Resource Leak When Second Install Fails

682a09d

fix F-1425: PTR_ERR on malloc Result Always Yields 0 in Error Message

8c1013a

fix F-1426: PTR_ERR on malloc Results Always Yields 0 in AES-GCM Self…

b3cb06e

…-Test Error Messages

fix F-3948: Missing KAT mismatch check in aes_xts_256_test partial bl…

1df8d45

…ock encrypt path

fix F-676: Missing ForceZero on RSA Decrypted Plaintext Buffer

85091c7

also ForceZeros various other potentially sensitive allocations before freeing them, and uses unconditional free() per current libwolfssl best practice.

fix F-677: SHA3 Init Memory Leak on wc_InitSha3 Failure

38c2477

fix F-678: SHA3 Finup Memory Leak on Update Failure

3057314

fix F-679: HMAC Finup Memory Leak on Update Failure

cac0f23

fix F-680: ECDH Init Error Paths Leak RNG Resource

ec4146a

fix F-681: DH/FFDHE Init Error Paths Leak RNG Resource

591f52f

fix F-708: DH km_dh_set_secret Missing Private Key Cleanup on Error Path

8110d56

fix F-709: AES-XTS Stream Finalize Returns Untranslated wolfCrypt Err…

0272f0a

…or Code to Kernel LKCAPI

fix F-1238: Integer Overflow in ECDSA Verify Allocation of sig_len + …

c84c44f

…hash_len

fixes for

ab69af5

F-1428: Missing ForceZero on sg_buf Containing Decrypted Plaintext in AES-GCM Non-Stream Path F-3293: AES-CCM Non-Contiguous SG Path Missing ForceZero on sg_buf Containing Decrypted Plaintext

fix F-1430: AES-GCM Non-Stream Path Returns -EINVAL Instead of Propag…

5030846

…ating skcipher_walk Error

fixes for

d3279a8

F-1433: AES CBC/CFB Self-Test Functions Silently Continue After enc2/dec2 Allocation Failure F-1434: linuxkm_test_aesgcm Silently Returns Success on Kernel Crypto Allocation Failure

fix for F-3946: Missing ForceZero on Heap-Allocated HMAC State Buffer…

476cfd6

… in km_hmac_init Error Path

fix F-3294: IntelRDseed64_r Stack Buffer Containing Entropy Cleared w…

8624bca

…ith Plain Assignment Instead of wc_ForceZero

fix F-4409: Integer Overflow in PKCS1 Sign Length Check Allows Heap B…

1b22e4b

…uffer Overflow

douzzer requested review from philljj, wolfSSL-Bot and wolfSSL-Fenrir-bot June 12, 2026 23:19

douzzer assigned wolfSSL-Bot Jun 12, 2026

douzzer added the For This Release Release version 5.9.2 label Jun 12, 2026

wolfSSL-Fenrir-bot reviewed Jun 12, 2026

View reviewed changes

dgarske reviewed Jun 12, 2026

View reviewed changes

dgarske assigned douzzer Jun 12, 2026

linuxkm/module_hooks.c: fix flub in IntelRDseed64_r() ForceZero() call;

4d77baf

linuxkm/lkcapi_aes_glue.c: fix typo in linuxkm_test_aesgcm() error message.

douzzer removed their assignment Jun 13, 2026

douzzer requested a review from dgarske June 13, 2026 00:26

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

20260610-linuxkm-fenrir#10676

20260610-linuxkm-fenrir#10676
douzzer wants to merge 26 commits into
wolfSSL:masterfrom
douzzer:20260610-linuxkm-fenrir

douzzer commented Jun 12, 2026

Uh oh!

wolfSSL-Fenrir-bot left a comment

Uh oh!

wolfSSL-Fenrir-bot Jun 12, 2026

Uh oh!

douzzer Jun 13, 2026

Uh oh!

dgarske left a comment

Uh oh!

dgarske Jun 12, 2026

Uh oh!

douzzer Jun 13, 2026

Uh oh!

dgarske Jun 12, 2026

Uh oh!

douzzer Jun 13, 2026

Uh oh!

dgarske Jun 12, 2026

Uh oh!

douzzer Jun 13, 2026

Uh oh!

dgarske Jun 12, 2026

Uh oh!

github-actions Bot commented Jun 12, 2026 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

		@@ -793,11 +797,12 @@ static int km_direct_rsa_dec(struct akcipher_request *req)
		}

		if (req->dst_len != (unsigned int)ctx->key_len) {

		@@ -72,7 +72,7 @@ WARN_UNUSED_RESULT int allocate_wolfcrypt_linuxkm_fpu_states(void)
		if (! wc_linuxkm_fpu_states) {

Conversation

douzzer commented Jun 12, 2026

Uh oh!

wolfSSL-Fenrir-bot left a comment

Choose a reason for hiding this comment

Fenrir Automated Review — PR #10676

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dgarske left a comment

Choose a reason for hiding this comment

Skoll Multi-Scan Review

Posted findings

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented Jun 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

MemBrowse Memory Report

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

github-actions Bot commented Jun 12, 2026 •

edited

Loading