[OpenAFS-devel] 1.4.2rc1 fails on RHEL4 2.6.9 kernel: missing keyring defines?

[chas williams - CONTRACTOR]

In message <sjmsljmkwra.fsf@cliodev.pgp.com>,Derek Atkins writes:
>> 	key_alloc(0,0,0,KEY_POS_VIEW|KEY_POS_SEARCH,0);
>
>Except that wont work properly, because the key_alloc test failure
>wont turn off keyring support.  The test for whether or not to
>use the keyring is based PURELY on the existence of keyctl() and
>request_key().

aok right.  i skipping testing for key_alloc() since it changed
during keyring development.  thus the other test for key_alloc().

>WHICH 2.6.9 rhel kernel?  There are a dozen of them, and I only have
>one, 2.6.9-5.  I can't test it on all of them, and I don't have the
>"state 2" kernel (see my previous email) which is causing this particular
>build issue.

i dont understand how the test passes when you dont have linux/key.h?
which apparently is the case with 2.6.9-5.  fixing the test for the
cases you listed in your other mail should fix the problem in
general.

>1) no linux/key.h
>2) linux/key.h but no KEY_POS_*
>3) linux/key.h with KEY_POS_* 
>
>2.6.9-5.EL is (1)
>2.6.9-22.EL is (2)
>2.6.9-34.EL is (3)
>
>The tests work for 1 and 3, but not 2.

can you get /afs/cmf/users/chas/conftest.dir.tar, unpack it where you
have the 2.6.9-5.EL kernel sources and run:

	make -C /where/is/my/bleeding/kernel M=`pwd`/conftest.dir modules