* Rename builtin/credential/aws-ec2 to aws
The aws-ec2 authentication backend is being expanded and will become the
generic aws backend. This is a small rename commit to keep the commit
history clean.
* Expand aws-ec2 backend to more generic aws
This adds the ability to authenticate arbitrary AWS IAM principals using
AWS's sts:GetCallerIdentity method. The AWS-EC2 auth backend is being to
just AWS with the expansion.
* Add missing aws auth handler to CLI
This was omitted from the previous commit
* aws auth backend general variable name cleanup
Also fixed a bug where allowed auth types weren't being checked upon
login, and added tests for it.
* Update docs for the aws auth backend
* Refactor aws bind validation
* Fix env var override in aws backend test
Intent is to override the AWS environment variables with the TEST_*
versions if they are set, but the reverse was happening.
* Update docs on use of IAM authentication profile
AWS now allows you to change the instance profile of a running instance,
so the use case of "a long-lived instance that's not in an instance
profile" no longer means you have to use the the EC2 auth method. You
can now just change the instance profile on the fly.
* Fix typo in aws auth cli help
* Respond to PR feedback
* More PR feedback
* Respond to additional PR feedback
* Address more feedback on aws auth PR
* Make aws auth_type immutable per role
* Address more aws auth PR feedback
* Address more iam auth PR feedback
* Rename aws-ec2.html.md to aws.html.md
Per PR feedback, to go along with new backend name.
* Add MountType to logical.Request
* Make default aws auth_type dependent upon MountType
When MountType is aws-ec2, default to ec2 auth_type for backwards
compatibility with legacy roles. Otherwise, default to iam.
* Pass MountPoint and MountType back up to the core
Previously the request router reset the MountPoint and MountType back to
the empty string before returning to the core. This ensures they get set
back to the correct values.
The rollback manager was using a saved MountTable rather than the
current table, causing it to attempt to rollback unmounted mounts, and
never rollback new mounts.
In fixing this, it became clear that bad things could happen to the
mount table...the table itself could be locked, but the table pointer
(which is what the rollback manager needs) could be modified at any time
without locking. This commit therefore also returns locking to a mutex
outside the table instead of inside, and plumbs RLock/RUnlock through to
the various places that are reading the table but not holding a write
lock.
Both unit tests and race detection pass.
Fixes#771
In order to implement this efficiently, I have introduced the concept of
"singleton" backends -- currently, 'sys' and 'cubbyhole'. There isn't
much reason to allow sys to be mounted at multiple places, and there
isn't much reason you'd need multiple per-token storage areas. By
restricting it to just one, I can store that particular mount instead of
iterating through them in order to call the appropriate revoke function.
Additionally, because revocation on the backend needs to be triggered by
the token store, the token store's salt is kept in the router and
client tokens going to the cubbyhole backend are double-salted by the
router. This allows the token store to drive when revocation happens
using its salted tokens.
Jeff Mitchell