From 06cca5110774f7b59a1685431ac697865588f4ca Mon Sep 17 00:00:00 2001 From: Roberto Sassu Date: Thu, 15 Feb 2024 11:31:07 +0100 Subject: integrity: Move integrity_kernel_module_request() to IMA In preparation for removing the 'integrity' LSM, move integrity_kernel_module_request() to IMA, and rename it to ima_kernel_module_request(). Rewrite the function documentation, to explain better what the problem is. Compile it conditionally if CONFIG_INTEGRITY_ASYMMETRIC_KEYS is enabled, and call it from security.c (removed afterwards with the move of IMA to the LSM infrastructure). Adding this hook cannot be avoided, since IMA has no control on the flags passed to crypto_alloc_sig() in public_key_verify_signature(), and thus cannot pass CRYPTO_NOLOAD, which solved the problem for EVM hashing with commit e2861fa71641 ("evm: Don't deadlock if a crypto algorithm is unavailable"). EVM alone does not need to implement this hook, first because there is no mutex to deadlock, and second because even if it had it, there should be a recursive call. However, since verification from EVM can be initiated only by setting inode metadata, deadlock would occur if modprobe would do the same while loading a kernel module (which is unlikely). Signed-off-by: Roberto Sassu Reviewed-by: Stefan Berger Reviewed-by: Mimi Zohar Acked-by: Mimi Zohar Signed-off-by: Paul Moore --- security/integrity/ima/ima_main.c | 33 +++++++++++++++++++++++++++++++++ 1 file changed, 33 insertions(+) (limited to 'security/integrity/ima') diff --git a/security/integrity/ima/ima_main.c b/security/integrity/ima/ima_main.c index 02021ee467d3..3891b83efdb3 100644 --- a/security/integrity/ima/ima_main.c +++ b/security/integrity/ima/ima_main.c @@ -1091,6 +1091,39 @@ int ima_measure_critical_data(const char *event_label, } EXPORT_SYMBOL_GPL(ima_measure_critical_data); +#ifdef CONFIG_INTEGRITY_ASYMMETRIC_KEYS + +/** + * ima_kernel_module_request - Prevent crypto-pkcs1pad(rsa,*) requests + * @kmod_name: kernel module name + * + * Avoid a verification loop where verifying the signature of the modprobe + * binary requires executing modprobe itself. Since the modprobe iint->mutex + * is already held when the signature verification is performed, a deadlock + * occurs as soon as modprobe is executed within the critical region, since + * the same lock cannot be taken again. + * + * This happens when public_key_verify_signature(), in case of RSA algorithm, + * use alg_name to store internal information in order to construct an + * algorithm on the fly, but crypto_larval_lookup() will try to use alg_name + * in order to load a kernel module with same name. + * + * Since we don't have any real "crypto-pkcs1pad(rsa,*)" kernel modules, + * we are safe to fail such module request from crypto_larval_lookup(), and + * avoid the verification loop. + * + * Return: Zero if it is safe to load the kernel module, -EINVAL otherwise. + */ +int ima_kernel_module_request(char *kmod_name) +{ + if (strncmp(kmod_name, "crypto-pkcs1pad(rsa,", 20) == 0) + return -EINVAL; + + return 0; +} + +#endif /* CONFIG_INTEGRITY_ASYMMETRIC_KEYS */ + static int __init init_ima(void) { int error; -- cgit v1.2.3