crypto: mbedtls_crypto_alt: replace legacy Mbed TLS crypto with PSA API and prepare for TF-PSA-Crypto 1.0 by valeriosetti · Pull Request #125 · zephyrproject-rtos/hostap

valeriosetti · 2026-02-24T13:57:18Z

Do all the changes which are required to make HostAP compatible with Mbed TLS 4.x/TF-PSA-Crypto 1.x.

It's a draft because tests are still in progress.

This will be integrated into Zephyr from zephyrproject-rtos/zephyr#104031

port/mbedtls/des.c

port/mbedtls/dhm.c

Since TF-PSA-Crypto the signature of mbedtls_nist_kw_wrap() and mbedtls_nist_kw_unwrap() have changed. Now they use a PSA API key instead of a legacy context. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

…tion Starting from TF-PSA-Crypto 1.0 PK functions won't require RNG function to be specified on input. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

HKDF legacy crypto module was removed from TF-PSA-Crypto and now only PSA API must be used for this scope. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

Remove inclusion of most legacy crypto headers, but add bignum that was missing. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

Remove usage of legacy crypto (ecp, ecdsa, ...) as much as possible and move to tf-psa-crypto allowed functions. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

MaochenWang1 · 2026-02-27T10:06:29Z

Hi @valeriosetti
based on your latest PR codebase, I add some changes in the hostap repo to fix the build error, and now the build is passed.
Should I push my commit on top of your commits in this same PR? What's your suggestion?

valeriosetti · 2026-02-27T10:08:38Z

Hi @valeriosetti based on your latest PR codebase, I add some changes in the hostap repo to fix the build error, and now the build is passed. Should I push my commit on top of your commits in this same PR? What's your suggestion?

Please wait, I'm working on that in parallel. Changes that I recently pushed here and zephyrproject-rtos/zephyr#104031 should make the build to work correctly.
Can you give this a try?

MaochenWang1 · 2026-02-27T10:22:58Z

Hi @valeriosetti based on your latest PR codebase, I add some changes in the hostap repo to fix the build error, and now the build is passed. Should I push my commit on top of your commits in this same PR? What's your suggestion?

Please wait, I'm working on that in parallel. Changes that I recently pushed here and zephyrproject-rtos/zephyr#104031 should make the build to work correctly. Can you give this a try?

I already rebased on your latest zephyr and hostap PR and I just check the local SHA values, same as yours.
There are still some errors I locally faced, as I use west build -p always -b rd_rw612_bga samples/net/wifi/shell -d wifi -DEXTRA_CONF_FILE="nxp/overlay_rw612.conf nxp/overlay_hostap_rw612.conf" to build.
I can wait and keep the change locally.
BTW, I find the WPA3 connection failure issue after the build passed, debugging this issue now.

src/crypto/crypto_mbedtls_alt.c

"psa_generate_random()" in Zephyr automatically uses either a real entropy driver or a fallback based on CTR-DRBG/HMAC-DRBG. There is no need to reimplement the mechanism in HostAP. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

When there is the possibility to chose between legacy crypto and PSA API always opt for the latter. This is equivalent to assume CONFIG_WIFI_NM_WPA_SUPPLICANT_CRYPTO_MBEDTLS_PSA always set and then removing dead code. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

DHE and RSA based key exchanges were removed from Mbed TLS 4.0. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

mbedtls_pk_get_type -> mbedtls_pk_get_key_type mbedtls_pk_type_t -> mbedtls_pk_sigalg_t Signed-off-by: Valerio Setti <vsetti@baylibre.com>

This is now an internal module in TF-PSA-Crypto and it cannot be used directly. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

DES and 3DES have been removed from cipher module. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

MaochenWang1 · 2026-02-28T06:43:02Z

Based on the latest codebase, there is new build error, which is not seen yesterday:

/home/nxf75317/zephyr-sdk-0.17.4/arm-zephyr-eabi/bin/../lib/gcc/arm-zephyr-eabi/12.2.0/../../../../arm-zephyr-eabi/bin/ld.bfd: zephyr/libzephyr.a(ms_funcs.c.obj): in function `challenge_response':
/home/nxf75317/code/zephyr_github/modules/lib/hostap/src/crypto/ms_funcs.c:150: undefined reference to `des_encrypt'
/home/nxf75317/zephyr-sdk-0.17.4/arm-zephyr-eabi/bin/../lib/gcc/arm-zephyr-eabi/12.2.0/../../../../arm-zephyr-eabi/bin/ld.bfd: /home/nxf75317/code/zephyr_github/modules/lib/hostap/src/crypto/ms_funcs.c:151: undefined reference to `des_encrypt'
/home/nxf75317/zephyr-sdk-0.17.4/arm-zephyr-eabi/bin/../lib/gcc/arm-zephyr-eabi/12.2.0/../../../../arm-zephyr-eabi/bin/ld.bfd: /home/nxf75317/code/zephyr_github/modules/lib/hostap/src/crypto/ms_funcs.c:156: undefined reference to `des_encrypt'
/home/nxf75317/zephyr-sdk-0.17.4/arm-zephyr-eabi/bin/../lib/gcc/arm-zephyr-eabi/12.2.0/../../../../arm-zephyr-eabi/bin/ld.bfd: zephyr/libzephyr.a(ms_funcs.c.obj): in function `nt_password_hash_encrypted_with_block':
/home/nxf75317/code/zephyr_github/modules/lib/hostap/src/crypto/ms_funcs.c:499: undefined reference to `des_encrypt'
/home/nxf75317/zephyr-sdk-0.17.4/arm-zephyr-eabi/bin/../lib/gcc/arm-zephyr-eabi/12.2.0/../../../../arm-zephyr-eabi/bin/ld.bfd: /home/nxf75317/code/zephyr_github/modules/lib/hostap/src/crypto/ms_funcs.c:500: undefined reference to `des_encrypt'
/home/nxf75317/zephyr-sdk-0.17.4/arm-zephyr-eabi/bin/../lib/gcc/arm-zephyr-eabi/12.2.0/../../../../arm-zephyr-eabi/bin/ld.bfd: zephyr/libzephyr.a(ms_funcs.c.obj):/home/nxf75317/code/zephyr_github/modules/lib/hostap/src/crypto/ms_funcs.c:499: more undefined references to `des_encrypt' follow
collect2: error: ld returned 1 exit status
ninja: build stopped: subcommand failed.

It looks like MBEDTLS_DES_C is not defined in hostap perspective.

MaochenWang1 · 2026-02-28T10:54:31Z

Just removing MBEDTLS_DES_C in the hostap can fix the build error.
With some local patches, the WPA3 connection can be successful (@valeriosetti if you agree, I can push my commit on top of your commits in this same PR, or I just hold the patch locally.).

For wifi enterprise test cases:
EAP-TLS with TLS 1.2, PEAPv0-MSCHAPv2, PEAPv1-GTC, EAP-PEAP-TLS are all successful.
EAP-TTLS-MSCHAPv2 is failed as CONFIG_FIPS is enabled by default, and @krish2718 is checking it.
EAP-TLS with TLS 1.3 is failed, return error in ssl_tls13_handle_hs_message_post_handshake.

the callstack is

Hi @valeriosetti I tried to disable MBEDTLS_SSL_SESSION_TICKETS, MBEDTLS_SSL_TLS1_3_KEY_EXCHANGE_MODE_PSK_ENABLED and MBEDTLS_SSL_TLS1_3_KEY_EXCHANGE_MODE_PSK_EPHEMERAL_ENABLED, then the connection is successful, but I am not sure if it's a correct way to fix this issue.
I also found that there is a commit in Mbedtls 3.6 https://github.com/Mbed-TLS/mbedtls/pull/9507/changes#diff-67127675c5612072a0cb4c12b6dd40b0cfdcc6cdb23839f48ce9e3a2257e77e7, but is missing in Mbedtls 4.x, which might affect this issue. Does Mbedtls4.x needs it？

Setting MBEDTLS_DES_C when building TF-PSA-Crypto is now prohibited because the symbols is both related to legacy crypto and its support has been removed from the library. Remove the guard for simplicity. If DES is not available the build will fail anyway. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

valeriosetti · 2026-02-28T14:02:54Z

Just removing MBEDTLS_DES_C in the hostap can fix the build error.

Added one more commit to implement this. Thanks for the notification.

I tried to disable MBEDTLS_SSL_SESSION_TICKETS, MBEDTLS_SSL_TLS1_3_KEY_EXCHANGE_MODE_PSK_ENABLED and MBEDTLS_SSL_TLS1_3_KEY_EXCHANGE_MODE_PSK_EPHEMERAL_ENABLED, then the connection is successful, but I am not sure if it's a correct way to fix this issue.

I don't know at all details of EAP-TLS with TLS 1.3. However the name suggests that it relies on TLS 1.3 and based on your comment you are disabling TLS 1.3 key exchanges, which makes me suspect the connection falls back to TLS 1.2? If that's the case then I don't know if this is fine or not...

I also found that there is a commit in Mbedtls 3.6 https://github.com/Mbed-TLS/mbedtls/pull/9507/changes#diff-67127675c5612072a0cb4c12b6dd40b0cfdcc6cdb23839f48ce9e3a2257e77e7, but is missing in Mbedtls 4.x, which might affect this issue. Does Mbedtls4.x needs it？

In Mbed TLS 4 I found the following:

build symbol MBEDTLS_SSL_TLS1_3_DEFAULT_NEW_SESSION_TICKETS
runtime configuration function mbedtls_ssl_conf_new_session_tickets

So perhaps in Mbed TLS 4 the commit you pointed to is not present because it's handled differently?

You can try to temporarly add the build symbol to modules/lib/openthread/third_party/mbedtls/mbedtls-config.h, set it to a value you think it's correct, and then re-run the test. If that works I will add the proper Kconfig for it (which I think it's missing now)

…euse Rework crypto_ecdh_init2() to clone the original PSA key using psa_copy_key() and wrap it into a new mbedtls_pk_context. The previous implementation copied the pk context directly, causing both contexts to share the same key_id. As a result, deinit of one context could destroy a key still in use. The new approach ensures a separate volatile key instance with its own key_id, and preserves key attributes and public key data. Signed-off-by: Maochen Wang <maochen.wang@nxp.com>

Update ECDH public key export and peer key processing to align with mbedtls 4.x PSA-based APIs. Uses psa_export_public_key() as the source of truth for EC points. Supports both compressed and uncompressed peer key formats. Handles X/Y reconstruction for compressed keys using mbedtls_ecp APIs. Rebuilds SubjectPublicKeyInfo DER with compressed ECPoint correctly. These updates restore DPP provisioning and authentication flows under mbedtls 4.x. Signed-off-by: Maochen Wang <maochen.wang@nxp.com>

Fix an issue in crypto_ec_key_set_pub() where pub_raw_len was left as zero after populating the X and Y coordinates. This caused crypto_ec_key_get_pubkey_point() to fail because the PK context reported no public key data. This fix updates pub_raw_len as bytes are written and sets pk_info to mbedtls_eckey_info, which is required by mbedtls 4.x PK wrapping logic when handling externally constructed EC keys. This restores correct DPP public key handling on mbedtls 4.x. Signed-off-by: Maochen Wang <maochen.wang@nxp.com>

Add NULL pointer checks to crypto_ecdh_deinit() and crypto_ec_key_deinit() to avoid dereferencing invalid contexts. Without these checks, callers that pass NULL could trigger crashes when freeing pk contexts or destroying PSA key IDs. Signed-off-by: Maochen Wang <maochen.wang@nxp.com>

… DPP Fix psa_copy_key() failures in DPP ECDH flow by adding PSA_KEY_USAGE_COPY to all ECC keypair and key import usage flags. Without this flag, mbedtls 4.x PSA enforces that a key cannot be cloned, causing crypto_ecdh_init2() to fail during key copy. Also set key lifetime to VOLATILE for imported private keys to match mbedtls 4.x expectations and ensure the copied key can be created. Signed-off-by: Maochen Wang <maochen.wang@nxp.com>

Replace calls of os_calloc(sizeof(type), 1) with the correct form os_calloc(1, sizeof(type)) to avoid argument order ambiguity and align with standard allocation patterns used elsewhere in hostap. Also fix buffer sizing in crypto_ec_key_get_ecprivate_key() by switching from MBEDTLS_PK_ECP_PUB_DER_MAX_BYTES to MBEDTLS_PK_ECP_PRV_DER_MAX_BYTES and replacing wpabuf_len() with wpabuf_size() to match the mbedtls_pk_write_key_der() API expectations. Signed-off-by: Maochen Wang <maochen.wang@nxp.com>

Fix crypto_ec_key_cmp() where the previous implementation exported private keys via PSA to compare them. However, the PK context held an invalid/zero key_id, causing psa_export_key() to fail and the comparison to be unreliable. Switch to comparing the public key raw buffer (pub_raw/pub_raw_len) only. This avoids dependence on key_id and PSA export, and provides a stable equality check. Signed-off-by: Maochen Wang <maochen.wang@nxp.com>

Fix a memory corruption in DPP auth flow caused by a type size mismatch. 'psa_key_bits_t key_bits' (smaller than size_t on this platform) was passed as '(size_t *)' to mbedtls_ecc_group_to_psa(), which wrote 4 bytes into a 2-byte field and overflowed adjacent memory. After dpp_set_pubkey_point(), auth->curve was corrupted (sometimes 0), making auth->curve->hash_len a huge value and leading to crashes (BUS fault in DPP authentication path). Use 'size_t key_bits' instead of 'psa_key_bits_t' in keypaths,this prevents out-of-bounds writes and restores stable DPP authentication. Signed-off-by: Maochen Wang <maochen.wang@nxp.com>

MaochenWang1 · 2026-03-12T09:12:54Z

Hi @valeriosetti please review the new commit

valeriosetti

Only one minor comment on the last commit

valeriosetti · 2026-03-12T09:24:48Z

src/crypto/crypto_mbedtls_alt.c

                            PSA_KEY_USAGE_SIGN_HASH | PSA_KEY_USAGE_VERIFY_HASH);
-    psa_set_key_algorithm(&key_attr, PSA_ALG_ECDH);
+    psa_set_key_algorithm(&key_attr, PSA_ALG_ECDSA(PSA_ALG_ANY_HASH));
+    psa_set_key_enrollment_algorithm(&key_attr, PSA_ALG_ECDH);


Warning: using psa_set_key_enrollment_algorithm is fine when PSA API are provided by Mbed TLS, but this is their own custom extension to the official PSA API. Indeed when this is used in Mbed TLS codebase it is usually guarded by MBEDTLS_PSA_CRYPTO_C. See this for example.

Long story short: this works with Mbed TLS, but I'm not sure it works when TF-M is the PSA API provider.

This is the reason why I didn't use psa_set_key_enrollment_algorithm in my initial implementation. However I understand that it's handy when you don't know ahead of time for which goal the key will be used, so I'm not against it.

Do I need to add MBEDTLS_PSA_CRYPTO_C guard here for psa_set_key_enrollment_algorithm?

You can and this would make the code being able to build also with TF-M as crypto backend, but of course the fix you just added (i.e. assigning 2 algorithms to the same key) wouldn't work in that case.
I don't think there is an elegant way to solve this a part from creating 2 different keys with different algorithms associated, which is a bit of a burden.

added the MBEDTLS_PSA_CRYPTO_C, currently I don't find a good way to fix this issue for TF-M case

I think this is fine for now. HostAP uses so much legacy crypto support from Mbed TLS that it's very far from using only PSA. This means that Mbed TLS is always present when we're building HostAP.

rlubos · 2026-03-12T10:17:12Z

I've retested the latest changes with nrf7002dk/nrf5340/cpuapp, WPA/WPA2/WPA3/EAP with TLS1.3, all worked just fine 👍

In mbedtls 4.x, EC keys were set with PSA_ALG_ECDH as the only algorithm, causing PSA_ERROR_NOT_PERMITTED when performing ECDSA signing (called by crypto_ec_key_sign_r_s). Set PSA_ALG_ECDSA(PSA_ALG_ANY_HASH) as primary and PSA_ALG_ECDH as enrollment algorithm, allowing keys to be used for both ECDSA signing and ECDH key agreement. Signed-off-by: Maochen Wang <maochen.wang@nxp.com>

MaochenWang1 · 2026-03-16T03:24:05Z

Hi @valeriosetti is this PR ready for review, could you change the mode from Draft to review mode?

valeriosetti · 2026-03-16T08:04:31Z

Hi @valeriosetti is this PR ready for review, could you change the mode from Draft to review mode?

So are all tests passing now?

MaochenWang1 · 2026-03-16T08:15:28Z

Hi @valeriosetti is this PR ready for review, could you change the mode from Draft to review mode?

So are all tests passing now?

basic tests pass

tomi-font · 2026-03-16T09:20:42Z

@krish2718 @rlubos please take a look when you have a moment

krish2718 · 2026-03-16T09:25:40Z

@krish2718 @rlubos please take a look when you have a moment

I am waiting on go-ahead from our QA @krga2022 , will update soon. I guess Robert already verified all security on nRF70 platform.

When CONFIG_MBEDTLS_PSA_CRYPTO_LEGACY_RNG=y, Hostap still enters the legacy CTR_DRBG code path, but MbedTLS 4.x with PSA no longer exposes mbedtls_ctr_drbg_context or the classic CTR_DRBG API. This causes build errors. Remove the unused CTR_DRBG declaration and initialization to align with PSA RNG usage. Signed-off-by: Maochen Wang <maochen.wang@nxp.com>

MaochenWang1 · 2026-03-18T10:22:26Z

add one commit to fix the build error when CONFIG_MBEDTLS_PSA_CRYPTO_LEGACY_RNG is enabled.

These modules were removed from TF-PSA-Crypto, but they are still used in some Zephyr module so this commit brings them back. Problematic modules are: - DES, DHM --> HostAP - ECDH --> ESP32 WiFi and BT drivers Here's the commits where they were taken from in TF-PSA-Crypto upstream repo: - DES: cb169403b0055edde4fdcfcfb22c3524c7d2f370 - DHM: 5b636e8c7ea96f2af57f876253a9d94e1958f962 - ECDH: 882a2cd42a6c52a31f99a2349361233afdcec83c Signed-off-by: Valerio Setti <vsetti@baylibre.com>

valeriosetti · 2026-03-18T16:44:39Z

I added 1 commit to move TF-PSA-Crypto "removed modules" from TF-PSA-Crypto fork to this one.
This was agreed with @carlescufi @tejlmand @rlubos @jukkar @frkv

valeriosetti changed the title ~~Prepare mbedtls4~~ crypto: mbedtls_crypto_alt: replace legacy Mbed TLS crypto with PSA API and prepare for TF-PSA-Crypto 1.0 Feb 24, 2026

github-actions bot mentioned this pull request Feb 24, 2026

modules: mbedtls: bump to 4.x zephyrproject-rtos/zephyr#104031

Open

valeriosetti force-pushed the prepare-mbedtls4 branch 3 times, most recently from 64a9b34 to 573a6a6 Compare February 26, 2026 23:03

MaochenWang1 reviewed Feb 27, 2026

View reviewed changes

port/mbedtls/des.c Show resolved Hide resolved

valeriosetti force-pushed the prepare-mbedtls4 branch from 573a6a6 to a0ab377 Compare February 27, 2026 08:17

MaochenWang1 reviewed Feb 27, 2026

View reviewed changes

port/mbedtls/dhm.c Show resolved Hide resolved

valeriosetti added 5 commits February 27, 2026 11:05

[noup] mbedtls: crypto: Adapt aes_[un]wrap() to TF-PSA-Crypto

be7aebb

Since TF-PSA-Crypto the signature of mbedtls_nist_kw_wrap() and mbedtls_nist_kw_unwrap() have changed. Now they use a PSA API key instead of a legacy context. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

[noup] mbedtls: tls|crypto: PK and TLS functions do not need RNG func…

30205de

…tion Starting from TF-PSA-Crypto 1.0 PK functions won't require RNG function to be specified on input. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

[noup] mbedtls: crypto: adapt hmac_kdf_expand() to TF-PSA-Crypto

942b96c

HKDF legacy crypto module was removed from TF-PSA-Crypto and now only PSA API must be used for this scope. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

[noup] mbedtls: crypto: fix headers inclusion

bd2755b

Remove inclusion of most legacy crypto headers, but add bignum that was missing. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

[noup] mbedtls: crypto: crypto: adapt EC management to tf-psa-crypto

5f04e17

Remove usage of legacy crypto (ecp, ecdsa, ...) as much as possible and move to tf-psa-crypto allowed functions. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

valeriosetti force-pushed the prepare-mbedtls4 branch from a0ab377 to a6e13d8 Compare February 27, 2026 10:05

MaochenWang1 reviewed Feb 28, 2026

View reviewed changes

src/crypto/crypto_mbedtls_alt.c Show resolved Hide resolved

valeriosetti added 6 commits February 28, 2026 06:21

[noup] mbedtls: tls: remove deprecated ciphersuites

704d08f

DHE and RSA based key exchanges were removed from Mbed TLS 4.0. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

[noup] mbedtls: tls: adapt to PK interface changes in Mbed TLS 4.0

08ae276

mbedtls_pk_get_type -> mbedtls_pk_get_key_type mbedtls_pk_type_t -> mbedtls_pk_sigalg_t Signed-off-by: Valerio Setti <vsetti@baylibre.com>

[noup] mbedtls: tls: remove CTR-DRBG usage

9d22ee6

This is now an internal module in TF-PSA-Crypto and it cannot be used directly. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

[noup] mbedtls: crypto: remove 3DES from crypto_cipher_init()

e255f01

DES and 3DES have been removed from cipher module. Signed-off-by: Valerio Setti <vsetti@baylibre.com>

valeriosetti force-pushed the prepare-mbedtls4 branch from 4bb0789 to e255f01 Compare February 28, 2026 05:23

MaochenWang1 closed this Mar 2, 2026

MaochenWang1 requested review from MaochenWang1 and rlubos March 12, 2026 02:35

MaochenWang1 added 8 commits March 12, 2026 10:42

MaochenWang1 force-pushed the prepare-mbedtls4 branch from cd2a66e to b0f081d Compare March 12, 2026 02:43

valeriosetti commented Mar 12, 2026

View reviewed changes

MaochenWang1 force-pushed the prepare-mbedtls4 branch from ebb5132 to fecb467 Compare March 12, 2026 11:06

valeriosetti marked this pull request as ready for review March 16, 2026 08:03

valeriosetti requested review from jukkar and krish2718 as code owners March 16, 2026 08:03

MaochenWang1 approved these changes Mar 16, 2026

View reviewed changes

valeriosetti mentioned this pull request Mar 18, 2026

West project readiness for Zephyr 4.4 zephyrproject-rtos/zephyr#105166

Open

82 tasks

Conversation

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

Uh oh!

Uh oh!

Uh oh!

Uh oh!

MaochenWang1 commented Feb 27, 2026

Uh oh!

valeriosetti commented Feb 27, 2026

Uh oh!

MaochenWang1 commented Feb 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

MaochenWang1 commented Feb 28, 2026

Uh oh!

MaochenWang1 commented Feb 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

valeriosetti commented Feb 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

MaochenWang1 commented Mar 12, 2026

Uh oh!

valeriosetti left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

valeriosetti Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

MaochenWang1 Mar 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

valeriosetti Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

MaochenWang1 Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

valeriosetti Mar 12, 2026

Choose a reason for hiding this comment

Uh oh!

rlubos commented Mar 12, 2026

Uh oh!

MaochenWang1 commented Mar 16, 2026

Uh oh!

valeriosetti commented Mar 16, 2026

Uh oh!

MaochenWang1 commented Mar 16, 2026

Uh oh!

tomi-font commented Mar 16, 2026

Uh oh!

krish2718 commented Mar 16, 2026

Uh oh!

MaochenWang1 commented Mar 18, 2026

Uh oh!

valeriosetti commented Mar 18, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

valeriosetti commented Feb 24, 2026 •

edited

Loading

MaochenWang1 commented Feb 27, 2026 •

edited

Loading

MaochenWang1 commented Feb 28, 2026 •

edited

Loading

valeriosetti commented Feb 28, 2026 •

edited

Loading

valeriosetti left a comment •

edited

Loading

MaochenWang1 Mar 12, 2026 •

edited

Loading