[Microchip TA-100] Fix port + update to cryptoauthlib v3.6.0 by danielinux · Pull Request #9702 · wolfSSL/wolfssl

danielinux · 2026-01-22T17:00:16Z

Description

Rework of TA-100 support to match cryptoauthlib v3.6.0's API. AES-GCM now working.

Fixes zd#20640

Testing

Used wolfDemo board (STM32U585) + TA100 mikrobus module.

Test log (with verbose TA-100 interactions) attached.

test_12a16e6cc.log

Copilot

Pull request overview

Copilot reviewed 19 out of 19 changed files in this pull request and generated 11 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

dgarske · 2026-02-12T05:49:47Z

Jenkins retest this please.

dgarske · 2026-02-19T21:16:56Z

Jenkins retest this please

dgarske · 2026-02-20T00:15:16Z

@danielinux Please rebase and squash this. Hopefully that will resolve things.

dgarske · 2026-03-04T19:57:59Z

@tmael can you please help get this PR rebased, merge conflicts resolved and ready for merge? This work has been lingering for way too long.

Copilot

Copilot was unable to review this pull request because the user who requested the review is ineligible. To be eligible to request a review, you need a paid Copilot license, or your organization must enable Copilot code review.

Copilot

Copilot was unable to review this pull request because the user who requested the review is ineligible. To be eligible to request a review, you need a paid Copilot license, or your organization must enable Copilot code review.

Copilot

Copilot was unable to review this pull request because the user who requested the review is ineligible. To be eligible to request a review, you need a paid Copilot license, or your organization must enable Copilot code review.

dgarske

Unrecognized macros used (new macros not registered in .wolfssl_known_macro_extras):

ATCA_HAL_I2C
ATCA_TFLEX_SUPPORT
ATECC_DEV_TYPE
TA100_ECC_TRACE
WOLFSSL_MANUALLY_SELECT_DEVICE_CONFIG
WOLFSSL_MICROCHIP_AESGCM
WOLFSSL_NO_DEL_HANDLE

Fix: Update .wolfssl_known_macro_extras in the PR branch:

- Using correct slot ID for AES keys - Adjust IV length - Fallback operations to software for unsupported ECC curves (all tests passing)

ECC384 should be supported in TA-100

…e AS_IF to conditionally run the manual search fallback

…kg.m4 is unavailable

…#ifndef WOLFSSL_MICROCHIP_TA100

Copilot

Pull request overview

Copilot reviewed 18 out of 18 changed files in this pull request and generated 12 comments.

Comments suppressed due to low confidence (2)

wolfcrypt/src/port/atmel/atmel.c:482

The slot availability check is incorrect: mSlotList[slotId] != ATECC_INVALID_SLOT && mSlotList[slotId] != slotId will treat an already-allocated slot (where mSlotList[slotId] == slotId) as available and re-allocate it. This can lead to multiple keys sharing the same hardware slot. The check should reject any slot where mSlotList[slotId] != ATECC_INVALID_SLOT (unless you’re using a different sentinel scheme).

        /* is slot available */
        if (mSlotList[slotId] != ATECC_INVALID_SLOT &&
            mSlotList[slotId] != slotId ) {
            slotId = ATECC_INVALID_SLOT;
        }
        else {
            mSlotList[slotId] = slotId;
        }

wolfcrypt/src/port/atmel/atmel.c:1724

ATCA_STATUS status is declared only inside #ifdef WOLFSSL_ATECC_TNGTLS, but it’s also used in the #ifdef WOLFSSL_ATECC_TFLXTLS block below. Building with TFLXTLS enabled (without TNGTLS) will fail to compile due to status being undeclared. Declare status at the function scope (or within each branch that uses it).

#ifdef WOLFSSL_ATECC_TFLXTLS
    /* MAKE SURE TO COPY YOUR CUSTOM CERTIFICATE FILES UNDER CAL/tng
     * Verify variable names, here below the code uses typical tflxtls
     *  proto example.
     *
     * g_cert_def_1_signer
     * g_cert_ca_public_key_1_signer
     * g_cert_def_3_device
     */

    status = atcacert_read_cert(&g_cert_def_1_signer,
                            (const uint8_t*)g_cert_ca_public_key_1_signer,
                            signerBuffer, &signerCertSize);
    if (status != ATCA_SUCCESS) {

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-23T13:51:30Z

+                /* not reserved in mSlotList, so return */
+                slotId = ATECC_SLOT_ECDHE_PRIV_ALICE;
+                goto exit;
+            case ATMEL_SLOT_ECDHE_BOB:
+                /* not reserved in mSlotList, so return */
+                slotId = ATECC_SLOT_ECDHE_PRIV_BOB;
+                goto exit;


ATMEL_SLOT_ECDHE_ALICE/ATMEL_SLOT_ECDHE_BOB return fixed slot IDs but are not marked as reserved/allocated in mSlotList. This allows ATMEL_SLOT_ANY allocations to potentially reuse the same slot IDs, causing slot collisions. Consider reserving these slots in mSlotList during init (or marking them as allocated in this switch case) so they can’t be returned by ATMEL_SLOT_ANY while in use.

Suggested change

/* not reserved in mSlotList, so return */

slotId = ATECC_SLOT_ECDHE_PRIV_ALICE;

goto exit;

case ATMEL_SLOT_ECDHE_BOB:

/* not reserved in mSlotList, so return */

slotId = ATECC_SLOT_ECDHE_PRIV_BOB;

goto exit;

/* use fixed slot and reserve it in mSlotList */

slotId = ATECC_SLOT_ECDHE_PRIV_ALICE;

break;

case ATMEL_SLOT_ECDHE_BOB:

/* use fixed slot and reserve it in mSlotList */

slotId = ATECC_SLOT_ECDHE_PRIV_BOB;

break;

Copilot · 2026-03-23T13:51:31Z


        /* generate new ephemeral key on device */
-        ret = atmel_ecc_create_key(slotId, peerKey);
+        ret = atmel_ecc_create_key(MAP_TO_HANDLE(slotId), ecc_curve, peerKey);


atmel_ecc_create_key() already applies MAP_TO_HANDLE() internally. Passing MAP_TO_HANDLE(slotId) here will double-map TA100 handles (e.g., 0x8006 + (0x8006 + slotId)), causing invalid handles and keygen failures on TA100. Pass the raw slotId instead and let the callee perform the mapping once.

Suggested change

ret = atmel_ecc_create_key(MAP_TO_HANDLE(slotId), ecc_curve, peerKey);

ret = atmel_ecc_create_key(slotId, ecc_curve, peerKey);

Copilot · 2026-03-23T13:51:31Z

+            ret = atmel_ecc_create_key(MAP_TO_HANDLE(slotId), otherKey->dp->id,
+                                                     peerKey);


Same double-mapping issue as in atcatls_create_key_cb: atmel_ecc_create_key() maps slot IDs internally, so passing MAP_TO_HANDLE(slotId) will produce an invalid TA100 handle. Pass slotId directly.

Suggested change

ret = atmel_ecc_create_key(MAP_TO_HANDLE(slotId), otherKey->dp->id,

peerKey);

ret = atmel_ecc_create_key(slotId, otherKey->dp->id, peerKey);

Copilot · 2026-03-23T13:51:31Z

    return ret;
 }
+#endif /* ATCA_TFLEX_SUPPORT */
+#endif /* ATCA_TFLEX_SUPPORT */


There are two consecutive #endif /* ATCA_TFLEX_SUPPORT */ lines closing the same #ifdef ATCA_TFLEX_SUPPORT block, which will break preprocessing/compilation (extra #endif). Remove the extra #endif or correct the guard structure.

Suggested change

#endif /* ATCA_TFLEX_SUPPORT */

Copilot · 2026-03-23T13:51:32Z

+    /* Test if data zone is locked */
+    status = talib_is_setup_locked(atcab_get_device(), &is_locked);
+    if (!is_locked) {
+        status = talib_lock_setup(atcab_get_device());
+        CHECK_STATUS(status);


Calling talib_lock_setup() here will permanently lock the device setup zone as a side effect of setting an AES key. That’s a high-risk, potentially irreversible operation that applications likely don’t expect from wc_AesSetKey. Consider removing this, gating it behind an explicit opt-in macro/API, or at least failing with a clear error telling users they must lock setup themselves as part of provisioning.

Suggested change

/* Test if data zone is locked */

status = talib_is_setup_locked(atcab_get_device(), &is_locked);

if (!is_locked) {

status = talib_lock_setup(atcab_get_device());

CHECK_STATUS(status);

/* Test if setup (data) zone is locked */

status = talib_is_setup_locked(atcab_get_device(), &is_locked);

CHECK_STATUS(status);

if (!is_locked) {

#ifdef WOLFSSL_TA100_AUTO_LOCK_SETUP

/* Optionally auto-lock setup zone if explicitly enabled */

status = talib_lock_setup(atcab_get_device());

CHECK_STATUS(status);

#else

/* Require caller/provisioning flow to lock setup explicitly */

return WC_HW_E;

#endif

Copilot · 2026-03-23T13:51:32Z

+#if defined(WOLFSSL_MICROCHIP_TA100)
+    uint16_t  rKeyH;        /* private key handle */
+    uint16_t  uKeyH;        /* public key handle */
+    byte uKey[TA_KEY_TYPE_RSA2048_SIZE]; /* public key */
+#endif


uKey is hard-coded to TA_KEY_TYPE_RSA2048_SIZE. If the build selects RSA-3072 (see the WOLFSSL_TA_KEY_TYPE_RSA* macros in atmel.h), this buffer will be too small and can lead to memory corruption. Use the selected size macro (e.g., WOLFSSL_TA_KEY_TYPE_RSA_SIZE) or otherwise enforce RSA-2048-only at compile time for TA100 RSA.

Copilot · 2026-03-23T13:51:33Z

+#if defined(WOLFSSL_MICROCHIP_TA100)
+                if (sig_len <= 0) {
+                    const RsaKey* r = (const RsaKey*)key;
+                    /* TA100 handles imply a 2048-bit RSA key. */
+                    if (r->rKeyH != 0 || r->uKeyH != 0) {
+                        sig_len = 256;
+                    }
+                }
+#endif


This assumes any TA100 handle implies a 2048-bit key (sig_len = 256). If TA100 RSA ever supports 3072 (and the build selects it), this will return the wrong size and can break callers that size output buffers based on this. Prefer deriving the size from the active TA100 RSA type/size macro (or from the key’s metadata) rather than hardcoding 256.

Suggested change

#if defined(WOLFSSL_MICROCHIP_TA100)

if (sig_len <= 0) {

const RsaKey* r = (const RsaKey*)key;

/* TA100 handles imply a 2048-bit RSA key. */

if (r->rKeyH != 0 || r->uKeyH != 0) {

sig_len = 256;

}

}

#endif

Copilot · 2026-03-23T13:51:33Z

+WOLFSSL_MICROCHIP_AESGCM can be used to enable AES-GCM but
+AESGCM support is not yet available for TA100 in both
+cryptauthlib-v3.3.3_397871.zip and cryptauthlib-v3.6.0_443271.zip.


This section says TA100 AES-GCM support is “not yet available” even for cryptoauthlib v3.6.0, but the PR description states AES-GCM is now working with v3.6.0. Please update this wording to reflect the current state (and clarify any remaining limitations like max AAD+data size, required IV/tag sizes, etc.).

Suggested change

WOLFSSL_MICROCHIP_AESGCM can be used to enable AES-GCM but

AESGCM support is not yet available for TA100 in both

cryptauthlib-v3.3.3_397871.zip and cryptauthlib-v3.6.0_443271.zip.

WOLFSSL_MICROCHIP_AESGCM can be used to enable AES-GCM for TA100 when

building against CryptoAuthLib version v3.6.0_443271 (or later). AES-GCM

is not supported for TA100 with cryptauthlib-v3.3.3_397871.zip. When using

TA100 AES-GCM, the current implementation requires a 12-byte IV and a

16-byte authentication tag, and the total size of AAD plus plaintext data

per operation is limited by the TA100 hardware interface (typically up to

4096 bytes).

Copilot · 2026-03-23T13:51:33Z

+    [with_cryptoauthlib=$withval],
+    [with_cryptoauthlib=no])
+
+AS_IF([test "x$with_cryptoauthlib" != "xno"], [


--with-cryptoauthlib only performs library detection and doesn’t enable any Microchip backend macros anymore. This means users can pass --with-cryptoauthlib and still end up without WOLFSSL_ATECC* / WOLFSSL_MICROCHIP_TA100 enabled (and no port code compiled). Consider erroring out when --with-cryptoauthlib is used without --enable-microchip=..., or explicitly document that both flags are required.

Suggested change

AS_IF([test "x$with_cryptoauthlib" != "xno"], [

AS_IF([test "x$with_cryptoauthlib" != "xno"], [

AS_IF([test "x$ENABLED_ATMEL" = "xno"], [

AC_MSG_ERROR([--with-cryptoauthlib requires --enable-microchip=<devices>. Please re-run configure with --enable-microchip=508,608,100 or appropriate devices.])

])

Copilot · 2026-03-23T13:51:33Z

+        PKG_CHECK_MODULES([CRYPTOAUTHLIB], [cryptoauthlib], [
+            CPPFLAGS="$CRYPTOAUTHLIB_CFLAGS $CPPFLAGS"
+            CFLAGS="$CRYPTOAUTHLIB_CFLAGS $CFLAGS"
+            LDFLAGS="$CRYPTOAUTHLIB_LIBS $LDFLAGS"


In the pkg-config path, CRYPTOAUTHLIB_LIBS is appended to both LDFLAGS and LIBS. Since *_LIBS commonly contains -l... entries, putting it into LDFLAGS is non-standard and can cause duplicated/incorrect linker flags downstream. Prefer appending CRYPTOAUTHLIB_LIBS only to LIBS and leaving LDFLAGS for -L.../linker flags.

Suggested change

LDFLAGS="$CRYPTOAUTHLIB_LIBS $LDFLAGS"

tmael · 2026-03-24T17:12:29Z

Jenkins retest this please.

Copilot AI review requested due to automatic review settings January 22, 2026 17:00

Copilot started reviewing on behalf of danielinux January 22, 2026 17:00 View session

This comment was marked as outdated.

Sign in to view

danielinux self-assigned this Jan 22, 2026

danielinux force-pushed the ta100_2025 branch 2 times, most recently from 03c4ab8 to 199c152 Compare February 10, 2026 12:54

danielinux requested a review from Copilot February 10, 2026 13:08

Copilot started reviewing on behalf of danielinux February 10, 2026 13:09 View session

Copilot AI reviewed Feb 10, 2026

View reviewed changes

danielinux assigned wolfSSL-Bot and unassigned danielinux Feb 10, 2026

SparkiDev assigned danielinux Feb 11, 2026

dgarske requested a review from tmael February 12, 2026 05:49

dgarske requested changes Feb 18, 2026

View reviewed changes

Comment thread configure.ac

danielinux force-pushed the ta100_2025 branch from f6c5f77 to 81086ff Compare February 19, 2026 08:24

danielinux assigned tmael Feb 19, 2026

danielinux force-pushed the ta100_2025 branch from 81086ff to 36b0e21 Compare February 20, 2026 19:30

dgarske unassigned danielinux Mar 4, 2026

Copilot AI review requested due to automatic review settings March 5, 2026 01:13

tmael force-pushed the ta100_2025 branch from 36b0e21 to 97a7359 Compare March 5, 2026 01:13

Copilot AI reviewed Mar 5, 2026

View reviewed changes

Copilot AI review requested due to automatic review settings March 5, 2026 23:30

tmael force-pushed the ta100_2025 branch from 97a7359 to 6289e72 Compare March 5, 2026 23:30

Copilot AI reviewed Mar 5, 2026

View reviewed changes

dgarske added the For This Release Release version 5.9.1 label Mar 6, 2026

dgarske previously approved these changes Mar 6, 2026

View reviewed changes

Comment thread wolfcrypt/test/test.c Outdated

Copilot AI review requested due to automatic review settings March 21, 2026 00:59

Copilot AI reviewed Mar 21, 2026

View reviewed changes

dgarske requested changes Mar 21, 2026

View reviewed changes

tmael and others added 14 commits March 23, 2026 14:34

TA100: Microchip Trust Anchor support RSA/ECC

e3ee55f

TA100: follow-up changes (squash after 4e64cb5)

492d388

[TA-100] Fixed GMAC, AES-GCM, AES, ECC

e400fec

- Using correct slot ID for AES keys - Adjust IV length - Fallback operations to software for unsupported ECC curves (all tests passing)

Reverted changes to fallback ECC

5887c9e

ECC384 should be supported in TA-100

[TA-100] Fixed ECC384. Adding RSA.

fdcf505

[TA-100] Fixed RSA keygen/sign/verify, tests

bde0142

Close PKG_CHECK_MODULES with a no-op failure action and use a separat…

9d8f643

…e AS_IF to conditionally run the manual search fallback

[TA-100] Fix build/test regressions in non-TA100 builds

927b732

wraps PKG_CHECK_MODULES with m4_ifdef so it's silently skipped when p…

90017ba

…kg.m4 is unavailable

Removed the dead #ifdef WOLFSSL_MICROCHIP_TA100 block (it was inside …

7d84796

…#ifndef WOLFSSL_MICROCHIP_TA100

Fix formatting

9cb5b7a

Fix comments and shellcheck warning

81c33bb

Fix typo

5074613

Correct Nested comment for Espressif build

6b9d658

danielinux assigned danielinux and unassigned tmael Mar 23, 2026

Addressed comment: added macros to .wolfssl_know_macro_extras

6eb4a30

Copilot AI review requested due to automatic review settings March 23, 2026 13:37

danielinux force-pushed the ta100_2025 branch from 3ed2aa2 to 6eb4a30 Compare March 23, 2026 13:37

danielinux assigned wolfSSL-Bot and unassigned danielinux and wolfSSL-Bot Mar 23, 2026

Copilot started reviewing on behalf of danielinux March 23, 2026 13:43 View session

Copilot AI reviewed Mar 23, 2026

View reviewed changes

danielinux assigned danielinux and unassigned wolfSSL-Bot Mar 23, 2026

	ret = atmel_ecc_create_key(MAP_TO_HANDLE(slotId), ecc_curve, peerKey);
	ret = atmel_ecc_create_key(slotId, ecc_curve, peerKey);

		ret = atmel_ecc_create_key(MAP_TO_HANDLE(slotId), otherKey->dp->id,
		peerKey);

	ret = atmel_ecc_create_key(MAP_TO_HANDLE(slotId), otherKey->dp->id,
	peerKey);
	ret = atmel_ecc_create_key(slotId, otherKey->dp->id, peerKey);

-WOLFSSL_MICROCHIP_AESGCM can be used to enable AES-GCM but
-AESGCM support is not yet available for TA100 in both
-cryptauthlib-v3.3.3_397871.zip and cryptauthlib-v3.6.0_443271.zip.
+WOLFSSL_MICROCHIP_AESGCM can be used to enable AES-GCM for TA100 when
+building against CryptoAuthLib version v3.6.0_443271 (or later). AES-GCM
+is not supported for TA100 with cryptauthlib-v3.3.3_397871.zip. When using
+TA100 AES-GCM, the current implementation requires a 12-byte IV and a
+-byte authentication tag, and the total size of AAD plus plaintext data
+per operation is limited by the TA100 hardware interface (typically up to
+bytes).

Conversation

danielinux commented Jan 22, 2026

Description

Testing

Uh oh!

This comment was marked as outdated.

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

dgarske commented Feb 12, 2026

Uh oh!

Uh oh!

dgarske commented Feb 19, 2026

Uh oh!

dgarske commented Feb 20, 2026

Uh oh!

dgarske commented Mar 4, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

dgarske left a comment

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Mar 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 23, 2026

Choose a reason for hiding this comment

Uh oh!

tmael commented Mar 24, 2026

Uh oh!

Reviewers