2378918_en-US

Yes

Suspend hang with dm-crypt on eMMC

We are seeing a suspend/resume issue on our target when using dm-crypt on an eMMC-backed partition.

The suspend failure starts as soon as the dm-crypt mapping is created with the following table:

0 crypt capi:cbc(aes)-plain :32:trusted:user_encrypted 0 1 allow_discards

With this table, suspend fails error logs attached for your reference.

We then changed the table as below:

0 crypt capi:cbc(aes-generic)-plain :32:trusted:user_encrypted 0 1 allow_discards

After this change, suspend works correctly.

So, from our testing:

plain partition mount - no suspend issues
dm-crypt mapping with capi:cbc(aes)-plain causes suspend issue
changing to capi:cbc(aes-generic)-plain resolves the issue

This makes us think the suspend problem is related to the AES crypto backend selected for aes on this platform, rather than to dm-crypt mapping logic in general.

We are using kernel version 6.12 and PREEMPT_RT is enabled and using iMx6ULL processor.

Could you please help check whether there is any known issue in the NXP BSP/kernel crypto driver path for capi:cbc(aes)-plain during suspend/resume?

We have even tried the commands which is shared in linux user guide

https://docs.nxp.com/bundle/UG10163/page/topics/dm-crypt_using_dcp_crypto_keys.html

We faced the same issue.

Can you please help us to resolve this issue?

Re: Suspend hang with dm-crypt on eMMC

Dear NXP team,

Can you please help me to resolve this issue?