Hello Mimi, On 7/27/21 7:25 PM, Mimi Zohar wrote: > Hi Simon, > > On Tue, 2021-07-27 at 16:33 +0000, THOBY Simon wrote: >> While users can restrict the accepted hash algorithms for the >> security.ima xattr file signature when appraising said file, users >> cannot restrict the algorithms that can be set on that attribute: >> any algorithm built in the kernel is accepted on a write. >> >> Define a new value for the ima policy option 'func' that restricts >> globally the hash algorithms accepted when writing the security.ima >> xattr. >> >> When a policy contains a rule of the form >> appraise func=SETXATTR_CHECK appraise_hash=sha256,sha384,sha512 >> only values corresponding to one of these three digest algorithms >> will be accepted for writing the security.ima xattr. >> Attempting to write the attribute using another algorithm (or "free-form" >> data) will be denied with an audit log message. >> In the absence of such a policy rule, the default is still to only >> accept hash algorithms built in the kernel (with all the limitations >> that entails). >> >> On policy update, the latest SETXATTR_CHECK rule is the only one >> that apply, and other SETXATTR_CHECK rules are deleted. >> >> Signed-off-by: Simon Thoby <simon.thoby@xxxxxxxxxx> > > Sorry, I was just getting to this patch, when you re-posted the patch > set. In the future, I'll make sure the responses are sent in quick > succession. > The mistake is mine, I will wait more between submission next time (and it would also be more reasonable of me to spam less the ML), sorry! > There are a number of assumptions related to the IMA policy: > - A builtin policy may be replaced by a custom policy. > - Depending on the Kconfig, the policy rules may not change be updated. > - Subsequent to replacing the builtin policy with a custom policy, > rules may only be appended, based on the Kconfig. > > The locking around the policy rules is dependent on these assumptions. > Removing policy rules is a major change that needs to be considered > carefully. Why should "func=SETXATTR_CHECK" be treated any > differently than any other policy rule? How would an attestation > server know which setxattr rule was enabled at the time the file was > appraised? > You're right, the "user convenience" gain is probably not much anyway, while the complexity burden arising from deleting nodes from a list in reverse order with potentially multiple concurrent readers is high, even with RCU doing most of the work. I will drop that in a future revision of the patchset. > thanks, > > Mimi > Many thanks for all the time you already spent on reviewing this work, Have a nice evening (or day, with timezones and all), Simon
