RE: [ima] 9165b814d2: BUG:kernel_NULL_pointer_dereference,address

-----Original Message-----
From: kernel test robot [mailto:rong.a.chen@intel.com]
Sent: Monday, March 2, 2020 2:22 AM
To: Roberto Sassu <roberto.sassu@huawei.com>
Cc: zohar@linux.ibm.com; James.Bottomley@HansenPartnership.com;
jarkko.sakkinen@linux.intel.com; linux-integrity@vger.kernel.org; linux-
security-module@vger.kernel.org; linux-kernel@vger.kernel.org; Silviu
Vlasceanu [off-list ref]; Roberto Sassu
[off-list ref]; lkp@lists.01.org
Subject: [ima] 9165b814d2:
BUG:kernel_NULL_pointer_dereference,address

FYI, we noticed the following commit (built with gcc-7):

commit: 9165b814d2bea8cfeb557505bb206396331e8192 ("[PATCH v2 8/8]
ima: Use ima_hash_algo for collision detection in the measurement list")
url: https://github.com/0day-ci/linux/commits/Roberto-Sassu/ima-support-
stronger-algorithms-for-attestation/20200205-233901
base: https://git.kernel.org/cgit/linux/kernel/git/zohar/linux-integrity.git
next-integrity

Hi

thanks for the report.

Yes, version 2 had a bug:

---
		ima_algo_array[i++].algo = HASH_ALGO_SHA1;
	}

	if (ima_hash_algo_idx >= nr_allocated_banks) {
		ima_algo_array[i].tfm = ima_shash_tfm;
		ima_algo_array[i].algo = ima_hash_algo;
	}
---

The code allocated ima_algo_array with size 1 (TPM was not found and
the default algorithm is SHA1).

However, later it initializes ima_algo_array for SHA1 and increments the
i variable. Since the code does not check if the default algorithm is SHA1,
the last part is also executed and causes corruption, because ima_algo_array
has only one element.

I fixed already this bug in version 3 of the patch set.

Thanks

Roberto

HUAWEI TECHNOLOGIES Duesseldorf GmbH, HRB 56063
Managing Director: Li Peng, Li Jian, Shi Yanli