On Mon, Dec 14, 2020 at 2:04 AM David Howells <dhowells@xxxxxxxxxx> wrote:
>
> Here's a set of minor fixes/cleanups that I've collected from various
> people for the next merge window.
This doesn't even build.
And no, that's not because of some merge error on my part. Just to
verify, I tried to build the head of what you sent me (commit
1b91ea77dfeb: "certs: Replace K{U,G}IDT_INIT() with
GLOBAL_ROOT_{U,G}ID") and it fails the same way.
In file included from ./include/linux/cred.h:13,
from security/integrity/ima/ima_mok.c:12:
security/integrity/ima/ima_mok.c: In function ‘ima_mok_init’:
./include/linux/key.h:292:29: warning: passing argument 7 of
‘keyring_alloc’ makes pointer from integer without a cast
[-Wint-conversion]
.. ten more lines of warnings..
security/integrity/ima/ima_mok.c:36:26: error: too many arguments to
function ‘keyring_alloc’
36 | ima_blacklist_keyring = keyring_alloc(".ima_blacklist",
| ^~~~~~~~~~~~~
so these "fixes" have clearly had absolutely zero testing, haven't
been in linux-next, and are completely broken.
The bug was introduced by commit 33c36b2053de ("certs: Fix blacklist
flag type confusion"), which changed the IMA code without actually
testing it.
I suspect the fix is trivial (change the "," to "|"), but I will not
be pulling this - or anything else that hasn't been in linux-next -
from you this merge window.
The pain just isn't worth it, but more importantly, you simply need to
get your workflow in order, and not send me completely untested
garbage that hasn't even been compiled.
Linus
