20260720-linuxkm-CMAC-native by douzzer · Pull Request #10980 · wolfSSL/wolfssl

douzzer · 2026-07-23T04:00:02Z

Implement glue logic for Linux kernel registration of native cmac(aes) implementation, and fix gaps in several other struct shash_alg glue implementations:

linuxkm/lkcapi_aes_glue.c, linuxkm/lkcapi_glue.c, configure.ac: implement linuxkm cmac(aes) glue.

linuxkm/lkcapi_sha_glue.c: fix sha.W lifecycle management to prevent leaks:

separate WC_LINUXKM_SHA_IMPLEMENT() into WC_LINUXKM_SHA1_IMPLEMENT() (no fixes
needed) and WC_LINUXKM_SHA2_IMPLEMENT() (with associated new helpers
WC_LINUXKM_SHA2_FREE_W(), WC_LINUXKM_SHA2_DECL_W(), WC_LINUXKM_SHA2_PUSH_W(),
and WC_LINUXKM_SHA2_POP_W(), that move .W to a stack buffer).
Reimplement SHA-2 one-shot digest callback to use only direct wolfCrypt calls
rather than proxy to other callbacks.

linuxkm/linuxkm_wc_port.h, linuxkm/module_hooks.c: implement magic checks on kernel struct wolfSSL_Mutex in WOLFSSL_LINUXKM_VERBOSE_DEBUG builds.

linuxkm/lkcapi_sha_glue.c: refactor SHA-3 state around struct km_Sha3TfmCtx,
adding km_sha3_init_tfm(), km_sha3_exit_tfm(), km_sha3_alloc_tstate(), and
km_sha3_free_tstate(), implementing garbage collection of abandoned shash_desc
objects at .exit_tfm.

linuxkm/lkcapi_sha_glue.c: add struct km_sha3_export_state, and implement km_sha3_export() and per-alg _import() methods, and km_sha3_test_export_import().

linuxkm/lkcapi_sha_glue.c: fix export/import for HMAC:

add to struct km_sha_hmac_pstate members desc_list_lock, desc_list, export_list, and export_list_len;
add struct km_sha_hmac_export_state;
add km_hmac_alloc_tstate(), km_hmac_export(), km_hmac_import(), km_hmac_test_export_import();
and update km_hmac_*() and WC_LINUXKM_HMAC_IMPLEMENT() to implement.

test/:

fixes for NO_DH;
fixes in test_wc_ed448_import_public() and test_wc_Ed448DecisionCoverage() for FIPS v6;
fixes in tests/api/test_sha3.c for KMAC keysize in FIPS builds.

tested with

wolfssl-multi-test.sh ...
check-source-text-fips-dev
quantum-safe-wolfssl-all-crypto-only-noasm-fips-dev-linuxkm-next-clang-tidy
'.*insmod.*'
all-crypto-openssl-extra-coexist-fips-dev

github-actions · 2026-07-23T04:10:52Z

MemBrowse Memory Report

linuxkm-pie

Data: __patchable_function_entries +200 B (+0.8%, 26,160 B)

linuxkm-standard

Data: __patchable_function_entries +176 B (+0.4%, 48,792 B)
No memory changes detected for:
gcc-arm-cortex-m0plus
gcc-arm-cortex-m3
gcc-arm-cortex-m4
gcc-arm-cortex-m4-baremetal
gcc-arm-cortex-m4-crypto-only
gcc-arm-cortex-m4-dtls13
gcc-arm-cortex-m4-min-ecc
gcc-arm-cortex-m4-openssl-compat
gcc-arm-cortex-m4-pkcs7
gcc-arm-cortex-m4-pq
gcc-arm-cortex-m4-rsa-only
gcc-arm-cortex-m4-sp-math
gcc-arm-cortex-m4-tls12
gcc-arm-cortex-m4-tls13
gcc-arm-cortex-m7
gcc-arm-cortex-m7-pq
gcc-arm-cortex-m7-tls13
stm32-sim-stm32h753

wolfSSL-Fenrir-bot

Fenrir Automated Review — PR #10980

Scan targets checked: linuxkm-bugs, linuxkm-src, wolfcrypt-bugs, wolfcrypt-rs-bugs, wolfcrypt-src, wolfssl-bugs, wolfssl-src

Findings: 5
5 finding(s) posted as inline comments (see file-level comments below)

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

philljj · 2026-07-23T17:23:43Z

Retest this please.

FATAL: [ssh-agent] Unable to start agent
java.io.IOException: Unexpected EOF

wolfSSL-Fenrir-bot

Fenrir Automated Review — PR #10980

Scan targets checked: linuxkm-bugs, linuxkm-src, wolfcrypt-bugs, wolfcrypt-rs-bugs, wolfcrypt-src, wolfssl-bugs, wolfssl-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.

Frauschi

🐺 Skoll Code Review

Overall recommendation: REQUEST_CHANGES
Findings: 4 total — 2 posted, 2 skipped

Posted findings

[Medium] enabled_fips declaration/use guard mismatch can break -Werror kernel builds — linuxkm/lkcapi_glue.c:227
[Medium] configure.ac lean-aesgcm-dev SHA512 logic inverted from disable to force-enable — configure.ac:7377-7379

Skipped findings

[Medium] km_hmac_import() calls allocating wc_HmacCopy() while holding spinlock (sleep-in-atomic)
[Low] cpuid_flag: reg[num] = 0 is dead code (overwritten by cpuid())

Review generated by Skoll via Claude/Codex

…ment linuxkm cmac(aes) glue.

…leaks: * separate WC_LINUXKM_SHA_IMPLEMENT() into WC_LINUXKM_SHA1_IMPLEMENT() (no fixes needed) and WC_LINUXKM_SHA2_IMPLEMENT() (with associated new helpers WC_LINUXKM_SHA2_FREE_W(), WC_LINUXKM_SHA2_DECL_W(), WC_LINUXKM_SHA2_PUSH_W(), and WC_LINUXKM_SHA2_POP_W(), that move .W to a stack buffer). * Reimplement SHA-2 one-shot digest callback to use only direct wolfCrypt calls rather than proxy to other callbacks.

…ecks on kernel struct wolfSSL_Mutex in WOLFSSL_LINUXKM_VERBOSE_DEBUG builds.

…TfmCtx, adding km_sha3_init_tfm(), km_sha3_exit_tfm(), km_sha3_alloc_tstate(), and km_sha3_free_tstate(), implementing garbage collection of abandoned shash_desc objects at .exit_tfm.

…ment km_sha3_export() and per-alg _import() methods, and km_sha3_test_export_import().

* add to struct km_sha_hmac_pstate members desc_list_lock, desc_list, export_list, and export_list_len; * add struct km_sha_hmac_export_state; * add km_hmac_alloc_tstate(), km_hmac_export(), km_hmac_import(), km_hmac_test_export_import(); * and update km_hmac_*() and WC_LINUXKM_HMAC_IMPLEMENT(() to implement.

* fixes for NO_DH; * fixes in test_wc_ed448_import_public() and test_wc_Ed448DecisionCoverage() for FIPS v6; * fixes in tests/api/test_sha3.c for KMAC keysize in FIPS builds.

… conjunctions to fix constant-folding-provoked compiler warnings around fips_enabled; for clarity, rename internal macros WC_LINUXKM_HAVE_SELFTEST and WC_LINUXKM_HAVE_SELFTEST_FULL to WC_LINUX_CONFIG_SELFTESTS and WC_LINUX_CONFIG_SELFTESTS_FULL respectively.

…hten up SHA-3 allocations, avoiding frivolous struct km_sha_state (with inline wc_Sha512) for SHA-3.

* Eliminate pointer from export blob, replacing it with a TFM-specific random cookie and per-desc serial ID. * Fixes kernel address disclosure risk, handle forgery risk, cross-TFM confusion risk, and heap pointer reuse risk.

…sl/wolfcrypt/settings.h: * Revert earlier changes adding WC_SHA256_W_SIZE and WC_SHA512_W_SIZE. * Add WC_SHA2_NO_SMALL_STACK to allow the W work buffer to move back onto the stack. * In wolfssl/wolfcrypt/settings.h, define WC_SHA2_NO_SMALL_STACK by default when WOLFSSL_KERNEL_MODE, and add a clause to #undef WOLFSSL_SMALL_STACK when building the SHA-2 implementations. linuxkm/lkcapi_sha_glue.c: * Refactor out superfluous struct km_sha_state. * Clean up some macro dynamics. * Add static asserts on HASH_MAX_STATESIZE to WC_LINUXKM_SHA1_IMPLEMENT() and WC_LINUXKM_SHA2_IMPLEMENT().

…ntegrity check and DRBG.

… SEGV from gcc misoptimization.

…_AesGcmInit().

… gating on its declaration (add dependency on defined(CONFIG_CRYPTO_FIPS)); use FIPS_NOT_ALLOWED_E rather than NOT_COMPILED_IN to signal FIPS-forbidden keysizes from linuxkm_test_ecdh_nist_driver() and linuxkm_test_ecdsa_nist_driver() to REGISTER_ALG_OPTIONAL().

…wc_FreeMutex().

douzzer · 2026-07-24T16:49:52Z

retest this please

PRB-aarch64-tests-v2

ERROR: Error cloning remote repo 'origin'
  hudson.plugins.git.GitException: Command "git fetch --tags --force --progress -- git@github.com:wolfssl/wolfssl +refs/heads/*:refs/remotes/origin/*" returned status code 128:
  stdout: 
  stderr: git@github.com: Permission denied (publickey).
  
PRB-makedist-check

ERROR: Error cloning remote repo 'origin'
  hudson.plugins.git.GitException: Command "git fetch --tags --force --progress -- git@github.com:wolfssl/wolfssl +refs/heads/*:refs/remotes/origin/*" returned status code 128:
  stdout: 
  stderr: git@github.com: Permission denied (publickey).

Frauschi

🐺 Skoll Code Review

Overall recommendation: REQUEST_CHANGES
Findings: 47 total — 6 posted, 41 skipped

Posted findings

[High] CMAC transient-copy design breaks under WC_DEBUG_CIPHER_LIFECYCLE: dangling/double-free of Aes.CipherLifecycleTag — linuxkm/lkcapi_aes_glue.c:5168
[High] configure.ac lean-aesgcm SHA-512 clause is logically inverted, force-ENABLING SHA-512/SHA-384 in the lean FIPS profile — configure.ac:7377-7379
[High] SHA-3 and CMAC shash_algs adopt .init_tfm/.exit_tfm without extending the >= 5.6 kernel-version guard — linuxkm/lkcapi_sha_glue.c:1177
[Medium] km_hmac_export reads snapshot->desc_id after releasing the list lock (use-after-free window) — linuxkm/lkcapi_sha_glue.c:1630-1648
[Medium] wc_UnLockMutex returns -1 without unlocking on magic mismatch, leaving the spinlock held with IRQs disabled — linuxkm/linuxkm_wc_port.h:2036-2045
[Medium] wc_AesGcmInit() WC_C_DYNAMIC_FALLBACK permanently clears aes->use_aesni instead of scoping the fallback to the call — wolfcrypt/src/aes.c:13993-14008

Skipped findings

[Medium] km_hmac_import and SHA-3 .init/.import orphan the desc's previous working node onto an uncapped list
[Medium] km_hmac_import performs wc_HmacCopy (which may allocate) while holding the IRQ-disabling spinlock
[Medium] WC_SHA256_W_SIZE / WC_SHA512_W_SIZE duplicate private allocation sizes from sha256.c / sha512.c
[Medium] WC_LINUXKM_SHA2_FREE_W frees an XMALLOC'd buffer with raw free()
[Medium] The entire SHA-2 stack-W path is compiled out in default kernel builds and therefore ships untested
[Medium] HMAC .export handle is not self-contained state and is valid only within one tfm and one eviction window
[Medium] km_hmac_export() reads the snapshot node after publishing it to export_list and dropping the lock (use-after-free read)
[Medium] New export/import self-tests allocate their tfm by cra_name and assert wolfSSL-private statesize, breaking registration under LINUXKM_LKCAPI_PRIORITY_ALLOW_MASKING
[Medium] km_hmac_import() orphans a working node per abandoned desc, giving unbounded kernel memory growth on the AF_ALG accept()/close() path
[Medium] AES-CMAC transient-copy design use-after-frees the Aes CipherLifecycleTag under WC_DEBUG_CIPHER_LIFECYCLE
[Medium] km_AesCmacSetKey() frees the tfm-owned pristine Cmac that concurrent descs dereference
[Medium] km_hmac_export() reads snapshot->desc_id after dropping desc_list_lock, racing eviction/free
[Medium] HMAC export-snapshot ring is sized by CPU count, not in-flight accepts, causing spurious -EINVAL under concurrent AF_ALG accept()
[Low] SHA-2 one-shot .digest still heap-allocates W, contradicting the stated no-heap-after-init design
[Low] types.h: undocumented !defined(GNUC) on the C23 static_assert branch changes which implementation GCC/clang select
[Low] configure.ac 'cmac(aes)' defines HAVE_AES_ECB (unused) instead of WOLFSSL_AES_DIRECT (required by the glue gate)
[Low] configure.ac lean-aesgcm: SHA-512 force-enable adds -DWOLFSSL_SHA384 while ENABLED_SHA384 stays "no"
[Low] km_AesCmacFinal/Finup bypass km_AesCmacDispose and bare-free the Cmac
[Low] SHA-3 one-shot digest places a ~480-byte struct on the kernel stack
[Low] km_AesCmacSetKey does not fail closed on the key-length rejection path
[Low] CMAC transient copy does not defensively NULL aes.streamData, unlike km_AesGet() in the same file
[Low] cpuid.c: replacing XMEMSET of the register array with a single-element zero is unrelated to this PR
[Low] Bare scope blocks introduced in km_AesCmacFinal / km_AesCmacFinup
[Low] KMAC-256 FIPS path drops the zero-length-key check instead of asserting KMAC_MIN_KEYLEN_E
[Low] SHA-3 .init and .import orphan any prior state node onto an unbounded per-TFM desc_list
[Low] Bounded HMAC export ring makes a previously valid state blob permanently unimportable, so concurrent AF_ALG accept() can fail spuriously
[Low] km_AesCmacSetKey() does not fail closed on the invalid-key-length path, contradicting its own rekey invariant
[Low] SHA-3 export blob carries no algorithm/variant tag, permitting cross-variant state import
[Low] cpuid.c XMEMSET removal not applied to cpuid_is_intel()/cpuid_is_amd(), which run first on the same path
[Low] km_hmac_import() performs the deep wc_HmacCopy() while holding the IRQ-disabling spinlock
[Low] wc_FreeMutex() now invalidates the lock, and wc_UnLockMutex() can bail out while still holding the spinlock
[Low] km_hmac_free_tstate()/km_sha3_free_tstate() silently skip cleanup when the list lock cannot be taken
[Info] linuxkm_test_aescmac ForceZeros an uninitialized stack struct on early-exit paths
[Info] wc_FreeMutex uses memset (not XMEMSET/ForceZero) and keeps a now-vestigial (void)m
[Info] WC_LINUXKM_SPINLOCK_MAGIC is an unexplained decimal literal
[Info] list_last_entry() without a list_empty() guard, protected only by an assert that can be silently skipped
[Info] Empty #if/#endif block left behind in the selftest-macro rename
[Info] settings.h #undefs WOLFSSL_SMALL_STACK before headers are parsed, changing struct layouts inside the sha256.c/sha512.c translation units
[Info] WC_LINUXKM_SHA2_FREE_W() releases an XMALLOC'd buffer with raw free()
[Info] WC_LINUXKM_SHA2_PUSH_W/POP_W generate bare brace-scope blocks in every SHA-2 shash entry point
[Info] SHA-3 .digest places a ~448-byte struct on the stack, contradicting the HMAC digest's stated rationale

Review generated by Skoll via Claude/Codex

Frauschi · 2026-07-24T19:40:54Z

+    if (! cmac)
+        return -ENOMEM;
+
+    XMEMCPY(cmac, t_ctx->pristine, sizeof(*cmac));


🟠 [High] CMAC transient-copy design breaks under WC_DEBUG_CIPHER_LIFECYCLE: dangling/double-free of Aes.CipherLifecycleTag
🚫 BLOCK bug

The new CMAC glue documents an invariant — "the embedded Aes owns no heap allocations in module configurations: wc_AesFree()'s only mainstream XFREE() target is the GCM-streaming streamData buffer" — and enforces it with a single #error for WOLF_CRYPTO_CB_FREE. That invariant is false under WC_DEBUG_CIPHER_LIFECYCLE: wolfssl/wolfcrypt/aes.h:444 adds void *CipherLifecycleTag, wolfcrypt/src/aes.c:15475 XMALLOCs it in wc_AesInit(), and wolfcrypt/src/aes.c:15553 XFREEs it (and NULLs it) in wc_AesFree(). km_AesCmacMaterialize() struct-copies the tfm-owned pristine Cmac (including that pointer), and every km_AesCmacUpdate()/Final()/Finup() then destroys the copy via wc_CmacFree()/wc_CmacFinal() -> wc_AesFree(), which frees the tag while t_ctx->pristine->aes.CipherLifecycleTag still points at it. The next materialize copies the dangling pointer and wc_debug_CipherLifecycleCheck() reads freed kernel memory. WOLF_CRYPTO_CB without _FREE is a second unguarded case: Cmac.devCtx is a per-instance pointer that would likewise ride every copy.

Suggestion:

Suggested change

XMEMCPY(cmac, t_ctx->pristine, sizeof(*cmac));

#if defined(WOLF_CRYPTO_CB) && defined(WOLF_CRYPTO_CB_FREE)

#error LINUXKM_LKCAPI_REGISTER_AESCMAC is incompatible with WOLF_CRYPTO_CB_FREE.

#endif

#ifdef WC_DEBUG_CIPHER_LIFECYCLE

#error LINUXKM_LKCAPI_REGISTER_AESCMAC is incompatible with WC_DEBUG_CIPHER_LIFECYCLE.

#endif

Frauschi · 2026-07-24T19:40:54Z

+                                                                           \
 static struct shash_alg name ## _alg =                                     \
 {                                                                          \
+    .init_tfm       =       km_sha3_init_tfm,                              \


🟠 [High] SHA-3 and CMAC shash_algs adopt .init_tfm/.exit_tfm without extending the >= 5.6 kernel-version guard
💡 SUGGEST

struct shash_alg's init_tfm/exit_tfm members were added in Linux 5.6. The file already encodes that fact in an explicit guard at lines 409-420 -- #error LINUXKM_LKCAPI_REGISTER for HMACs is supported only on Linux kernel versions >= 5.6.0. -- whose condition enumerates only the eight *_HMAC register macros, because HMAC was the only user of those callbacks. This PR adds .init_tfm = km_sha3_init_tfm and .exit_tfm = km_sha3_exit_tfm to the SHA-3 shash_alg (lines 1177/1188) and .exit_tfm = km_AesCmacExitTfm to cmacAesAlg (lkcapi_aes_glue.c:5350), but does not extend the guard's condition to cover LINUXKM_LKCAPI_REGISTER_SHA3_224/256/384/512 or LINUXKM_LKCAPI_REGISTER_AESCMAC. The module advertises support back to 5.4 (lkcapi_glue.c:30-32: #error LINUXKM_LKCAPI_REGISTER is supported only on Linux kernel versions >= 5.4.0.). Trigger: build against a 5.4 or 5.5 kernel with --enable-linuxkm-lkcapi-register=sha3 (or all, or cmac(aes)). Instead of the intended clear #error, the build fails with an opaque 'struct shash_alg has no member named init_tfm' from the macro expansion, or -- worse if a distro backport added only one of the two members -- compiles with the callback silently dropped, which for SHA-3 means desc_list_lock is never initialized and km_sha3_alloc_tstate() locks an uninitialized spinlock on the first digest.

Recommendation: Extend the guard's condition to every alg that now installs an init_tfm/exit_tfm callback, and retitle the message:

#if (LINUX_VERSION_CODE < KERNEL_VERSION(5, 6, 0)) &&
(defined(LINUXKM_LKCAPI_REGISTER_SHA1_HMAC) ||
... existing HMAC entries ...
defined(LINUXKM_LKCAPI_REGISTER_SHA3_224) ||
defined(LINUXKM_LKCAPI_REGISTER_SHA3_256) ||
defined(LINUXKM_LKCAPI_REGISTER_SHA3_384) ||
defined(LINUXKM_LKCAPI_REGISTER_SHA3_512))
#error LINUXKM_LKCAPI_REGISTER for HMACs and SHA-3 is supported only on Linux kernel versions >= 5.6.0.
#endif

and add the matching guard for LINUXKM_LKCAPI_REGISTER_AESCMAC in lkcapi_aes_glue.c.

Frauschi · 2026-07-24T19:40:55Z

+    snapshot->desc_id = p_ctx->cur_desc_id++;
+    /* list_add() prepends, so the tail from list_last_entry() is the oldest. */
+    list_add(&snapshot->desc_ent, &p_ctx->export_list);
+    p_ctx->export_list_len++;


🟡 [Medium] km_hmac_export reads snapshot->desc_id after releasing the list lock (use-after-free window)
🚫 BLOCK bug

The function's own design note says unlinking under the lock is what makes the ring safe: "the evicted node is now unlinked and unreachable ... so free it outside the lock." But after wc_UnLockMutex() the code dereferences snapshot again to fill in the blob (blob->desc_id = snapshot->desc_id;). Once the lock is dropped, a concurrent km_hmac_export() on the same tfm can evict this very node (it is the list head, becoming the tail after WC_LINUXKM_HMAC_EXPORT_LIST_MAX further exports) and free() it, so the read races a free. Every other post-unlock dereference in the function is on the deliberately-unlinked evicted node; this one is on a still-published node.

Suggestion:

Suggested change

p_ctx->export_list_len++;

word64 new_desc_id;

...

new_desc_id = snapshot->desc_id = p_ctx->cur_desc_id++;

list_add(&snapshot->desc_ent, &p_ctx->export_list);

p_ctx->export_list_len++;

(void)wc_UnLockMutex(&p_ctx->desc_list_lock);

...

blob->desc_id = new_desc_id;

Good catch. Fixed.

douzzer requested review from philljj, wolfSSL-Bot and wolfSSL-Fenrir-bot July 23, 2026 04:00

douzzer assigned wolfSSL-Bot Jul 23, 2026

wolfSSL-Fenrir-bot reviewed Jul 23, 2026

View reviewed changes

Comment thread linuxkm/lkcapi_sha_glue.c

Comment thread linuxkm/lkcapi_sha_glue.c Outdated

Comment thread wolfcrypt/src/sha256.c

Comment thread wolfcrypt/src/sha256.c

Comment thread linuxkm/lkcapi_sha_glue.c

philljj assigned douzzer and unassigned wolfSSL-Bot Jul 23, 2026

douzzer commented Jul 23, 2026

View reviewed changes

Comment thread linuxkm/lkcapi_sha_glue.c

douzzer force-pushed the 20260720-linuxkm-CMAC-native branch from 3e6a2f1 to d4c3410 Compare July 24, 2026 00:13

douzzer commented Jul 24, 2026

View reviewed changes

Comment thread wolfcrypt/src/sha256.c

douzzer commented Jul 24, 2026

View reviewed changes

Comment thread wolfcrypt/src/sha256.c

douzzer commented Jul 24, 2026

View reviewed changes

Comment thread linuxkm/lkcapi_sha_glue.c Outdated

douzzer assigned wolfSSL-Bot and unassigned douzzer Jul 24, 2026

douzzer requested a review from wolfSSL-Fenrir-bot July 24, 2026 00:22

douzzer commented Jul 24, 2026

View reviewed changes

Comment thread linuxkm/lkcapi_sha_glue.c

wolfSSL-Fenrir-bot reviewed Jul 24, 2026

View reviewed changes

Comment thread linuxkm/lkcapi_sha_glue.c

Frauschi requested changes Jul 24, 2026

View reviewed changes

Comment thread linuxkm/lkcapi_glue.c Outdated

Comment thread configure.ac

Frauschi assigned douzzer and unassigned wolfSSL-Bot Jul 24, 2026

douzzer added 8 commits July 24, 2026 10:04

linuxkm/lkcapi_aes_glue.c, linuxkm/lkcapi_glue.c, configure.ac: imple…

93da1d0

…ment linuxkm cmac(aes) glue.

linuxkm/linuxkm_wc_port.h, linuxkm/module_hooks.c: implement magic ch…

b345813

…ecks on kernel struct wolfSSL_Mutex in WOLFSSL_LINUXKM_VERBOSE_DEBUG builds.

linuxkm/lkcapi_sha_glue.c: refactor SHA-3 state around struct km_Sha3…

3cdf16a

…TfmCtx, adding km_sha3_init_tfm(), km_sha3_exit_tfm(), km_sha3_alloc_tstate(), and km_sha3_free_tstate(), implementing garbage collection of abandoned shash_desc objects at .exit_tfm.

linuxkm/lkcapi_sha_glue.c: add struct km_sha3_export_state, and imple…

9704ae6

…ment km_sha3_export() and per-alg _import() methods, and km_sha3_test_export_import().

test/:

6879f22

* fixes for NO_DH; * fixes in test_wc_ed448_import_public() and test_wc_Ed448DecisionCoverage() for FIPS v6; * fixes in tests/api/test_sha3.c for KMAC keysize in FIPS builds.

douzzer added 11 commits July 24, 2026 10:05

linuxkm/lkcapi_sha_glue.c: add struct km_sha3_state_by_pointer to tig…

a25d9c3

…hten up SHA-3 allocations, avoiding frivolous struct km_sha_state (with inline wc_Sha512) for SHA-3.

wolfssl/wolfcrypt/types.h: add wc_static_assert_if_const().

00a8b5c

configure.ac: enable SHA512 for fips=lean-aesgcm, to support SHA512 i…

1688935

…ntegrity check and DRBG.

tests/api/test_kdf.c: fix NO_SHA in test_wc_KdfFeatureCoverage().

e98b359

wolfcrypt/src/cpuid.c: in cpuid_flag(), refactor out XMEMSET() to fix…

2a0f06a

… SEGV from gcc misoptimization.

wolfcrypt/src/aes.c: implement WC_C_DYNAMIC_FALLBACK for AES-NI in wc…

03d1c81

…_AesGcmInit().

.github/workflows/codespell.yml: add *.bin to skip list.

4fcf62f

linuxkm/linuxkm_wc_port.h: fix raw_spinlock.magic poison strategy in …

5d377ee

…wc_FreeMutex().

douzzer force-pushed the 20260720-linuxkm-CMAC-native branch from d4c3410 to 5d377ee Compare July 24, 2026 15:49

douzzer assigned wolfSSL-Bot and unassigned douzzer Jul 24, 2026

Frauschi requested changes Jul 24, 2026

View reviewed changes

Frauschi assigned douzzer and unassigned wolfSSL-Bot Jul 24, 2026

-    XMEMCPY(cmac, t_ctx->pristine, sizeof(*cmac));
+#if defined(WOLF_CRYPTO_CB) && defined(WOLF_CRYPTO_CB_FREE)
+    #error LINUXKM_LKCAPI_REGISTER_AESCMAC is incompatible with WOLF_CRYPTO_CB_FREE.
+#endif
+#ifdef WC_DEBUG_CIPHER_LIFECYCLE
+    #error LINUXKM_LKCAPI_REGISTER_AESCMAC is incompatible with WC_DEBUG_CIPHER_LIFECYCLE.
+#endif

-    p_ctx->export_list_len++;
+word64 new_desc_id;
+    ...
+    new_desc_id = snapshot->desc_id = p_ctx->cur_desc_id++;
+    list_add(&snapshot->desc_ent, &p_ctx->export_list);
+    p_ctx->export_list_len++;
+    (void)wc_UnLockMutex(&p_ctx->desc_list_lock);
+    ...
+    blob->desc_id = new_desc_id;

Uh oh!

Conversation

douzzer commented Jul 23, 2026

Uh oh!

github-actions Bot commented Jul 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

MemBrowse Memory Report

linuxkm-pie

linuxkm-standard

Uh oh!

wolfSSL-Fenrir-bot left a comment

Choose a reason for hiding this comment

Fenrir Automated Review — PR #10980

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

philljj commented Jul 23, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

wolfSSL-Fenrir-bot left a comment

Choose a reason for hiding this comment

Fenrir Automated Review — PR #10980

Uh oh!

Uh oh!

Frauschi left a comment

Choose a reason for hiding this comment

🐺 Skoll Code Review

Posted findings

Uh oh!

Uh oh!

Uh oh!

douzzer commented Jul 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Frauschi left a comment

Choose a reason for hiding this comment

🐺 Skoll Code Review

Posted findings

Uh oh!

Frauschi Jul 24, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Frauschi Jul 24, 2026

Choose a reason for hiding this comment

Uh oh!

douzzer Jul 24, 2026

Choose a reason for hiding this comment

Uh oh!

Frauschi Jul 24, 2026

Choose a reason for hiding this comment

Uh oh!

douzzer Jul 24, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

github-actions Bot commented Jul 23, 2026 •

edited

Loading

douzzer commented Jul 24, 2026 •

edited

Loading