Pete Dennis wrote:
It'd be good to have a writeup in the bug of why this happened. It's
a bad
interaction between my original code and Pete's fix for 9568. The
verify
code was unchanged in that changeset, but the way we pulled the bits off
disk did. But the specifics of the interaction aren't something I've
tracked down, so I can't say more than that.
Looking into this it would appear my testing for the 9568 was not
complete as it should have been (looks like I missed the pkg verify
testing).
Having said that I think that both lists need to be canonical as it
is possible for the device_policy to have the minor spec (ie the '*')
and the manifest to have it as well (the SUNWpcelx has this).
Yup. I found that last night. It's the only driver currently defined
that includes a "*" in it's "policy=" section of an "add driver" action.
I'll rework the changes for #14219 so that initial "*"'s are removed
from both lists. I'll also test against SUNWpcelx.
The underlying fix for 9568 was to put in a minor spec '*' so that
the calls to rem_drv would work when updating the image.
Does this help ?
Could you please add the comment to bug #12419 describing this.
Rich, please be sure that test_driver_policy_removal continues to
pass with
your fix.
And I'll make sure this test continues to pass.
Thanks.
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
