Specifying --ipc=host --pid=host is broken #7109
Conversation
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: rhatdan The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
Fixes: #7100 |
| c.CGroupsNS: &s.CgroupNS, | ||
| } { | ||
| if k != "" { | ||
| *v, err = specgen.ParseNamespace(k) |
There was a problem hiding this comment.
This looks like a classic loop-variable overwrite/leak. Good catch!
There was a problem hiding this comment.
I think the issue was that there are collisions with the keys (both c.IPC and c.PID have the value "host")
|
LGTM |
|
Set lgtm/hold labels |
|
LGTM |
For some reason we were overwriting memory when handling both --pid=host and --ipc=host. Simplified the code to handle this correctly, and add test to make sure it does not happen again. Signed-off-by: Daniel J Walsh <dwalsh@redhat.com>
|
I believe that this will also fix containers/toolbox#501 - Toolbox uses Thanks for the fix, @rhatdan (and the tests)! |
|
/lgtm |
|
/hold cancel |
For some reason we were overwriting memory when handling both
--pid=host and --ipc=host. Simplified the code to handle this
correctly, and add test to make sure it does not happen again.
Signed-off-by: Daniel J Walsh dwalsh@redhat.com