Subject: Re: Current build.sh release fail
To: Matthias Scheler <tron@zhadum.de>
From: Chavdar Ivanov <ci4ic4@gmail.com>
List: current-users
Date: 07/08/2005 22:00:10
On 07/07/05, Matthias Scheler <tron@zhadum.de> wrote:
> In article <bc8ff1fd0507070343170b4ea@mail.gmail.com>,
>         Chavdar Ivanov <ci4ic4@gmail.com> writes:
> > I wonder if this is to do with the gcc options in my /etc/mk.conf file
> > (if that is at all consulted during the build - I don't know) -
>=20
> Yes, it has, and yes they are used.
>=20
> > .sinclude "/usr/pkg/share/mk/cpuflags.mk"
>=20
> Enabling optimizations for a certain CPU variants is known to increase
> the code size and therefore break the build of the boot floppies. And
> the release will also not work on all the system types which
> normally would be supported.
>=20
> I recomment not to use such options for release builds but only for
> builds of kernel and binaries used on the local machine.
>=20
> > COPTS+=3D-pipe
> > CFLAGS+=3D-pipe

I took both the .sinclude statement and the -pipe flags. Indeed, the
floppies got created properly.

For some reason the process requires to find
{bld-dir}.../etc/pf.{os|conf}, which do not get created. I have copies
from an earlier version; I just copy them in the above place and
restart the process. I don't know, probably something did not get
updated properly.

The worst with 3.99.7 on my laptop was with the cardbus - it is anHP
Omnibook 6000, I have been using a Xircom Cardbus Ethernet
10/100+Modem56 on this machine for ages with no problems whatsoever (
in fact, the reason I jumped from OpenBSD to NetBSD was precisely that
- I could use the modem and the ethernet port on it; one if these did
not work under Open... at the time). Now all I get is:

cbb0 at pci0 dev 10 function 0: Texas Instruments PCI1420 PCI-CardBus
Bridge (rev. 0x00)
cbb1 at pci0 dev 10 function 1: Texas Instruments PCI1420 PCI-CardBus
Bridge (rev. 0x00)
cbb0: bad Vcc request. sock_ctrl 0x0, sock_status 0x30000326
cbb0: disabling socket
cbb0: interrupting at irq 10
cardslot0 at cbb0 slot 0 flags 0
cbb1: interrupting at irq 10
cardslot1 at cbb1 slot 1 flags 0
cbb0: wait took 0.009207s
cbb0: bad Vcc request. sock_ctrl 0x30, sock_status 0x30000b69
cbb0: disabling socket
cbb0: bad Vcc request. sock_ctrl 0x0, sock_status 0x30000326
cbb0: disabling socket
cbb0: wait took 0.009847s
cbb0: bad Vcc request. sock_ctrl 0x30, sock_status 0x30000b69
cbb0: disabling socket
cbb0: bad Vcc request. sock_ctrl 0x0, sock_status 0x30000326
cbb0: disabling socket

(b69 when connecting, 326 when disconnecting).=20

No big deal as I can use the ex0: ( 3Com 3c556 ) mini-pci i/f, but
still not nice. The same happens with other cardbus adapters ( I.E.
SMC EZConnect, this one is not supported by NetBSD yet as it is
Prism54, but the message from cbb0: is the same).

BTW I use GENERIC_LAPTOP. The weird thing is that booting 3.99.3 from
02/05/2005 did not change this ( using 3.99.7 userland, ofcourse) -
the cbb0 message is the same.

I will try to build an ACPI kernel to see if there will be any difference.=
=20

>=20
> IIRC this is known to work buggy.
>=20
>         Kind regards
>=20
> --
> Matthias Scheler                                  http://scheler.de/~matt=
hias/
>