--Km1U/tdNT/EmXiR1
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Sun, Sep 17, 2006 at 09:08:19AM +0100, Iain Hibbert wrote:
>=20
> For the record, I think the answer is No.
>=20
> I tried without those persistent objects and still managed a uvm_fault,
> exactly the same:
>=20
> uvm_fault(0xcb166464, 0x35a000, 1) -> 0xe
> kernel: supervisor trap page fault, code=3D0
> Stopped in pid 6955.1 (rfcomm_sppd) at	0xcad0a545:	testl	%eax,netbsd:cpu_=
info_primary+0x188(,%ebx,4)
> db> bt
> _prop_dictionary_keysym32_pool(c0aa1000,c08b7800,cae17960,2000,c0102145) =
at 0xcad0a545
> pppintr(c0a6a400,f0b800f2,c0a6a400,246,0) at netbsd:pppintr+0x456

> I have DIAGNOSTIC enabled, is there anything else I can do that may give
> more information?

The two things that come to mind are: 1) make sure that 0xcad0a545 really=
=20
is the address of _prop_dictionary_keysym32_pool. 2) Add a static variable=
=20
and have the pppintr() routine shove different values into it at different=
=20
points. That way you don't change the overall timing much at all, and you=
=20
can track down what call went out into the weeds.

Take care,

Bill

--Km1U/tdNT/EmXiR1
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (NetBSD)

iD8DBQFFD1djWz+3JHUci9cRAghJAJ9/Sdxm8ZT05n0HnqUFUN6WI7jq3wCfT4kw
8Uu6Gvgxr8ci7enjELQzjWk=
=SNOP
-----END PGP SIGNATURE-----

--Km1U/tdNT/EmXiR1--