On 01.07.2010 11:11, Arne Georg Gleditsch wrote:
On 07. juni 2010 16:59, Arne Georg Gleditsch wrote:
On 07. juni 2010 16:35, Carl-Daniel Hailfinger wrote:
On 26.05.2010 19:03, Arne Georg Gleditsch wrote:
On 26. mai 2010 16:18, Michael Karcher wrote:
Maybe because we preferrably choose chips with subsystem IDs. Maybe because someone thought that a new board revision very probably contains a more modern graphics chip. In this case, you are very right in suggesting to take the network chip as secondary PCI ID. You need to find out what subsystem IDs are used (if any) in the DL145 and the DL165.
If you send
- lspci -vvvxxxnn, superiotool -deV and flashrom -V output of your
DL165 (general requirement for having new boards added),
- an updated patch that uses the network instead of the
graphics chip for both boards, including a sign-off[1],
- preferably at least PCI ID info including subsystems for the
DL145,
According to http://merlin.ugent.be/~samuel/dl145g3/info/lspci-vnn.txt, the DL145 is equipped thus:
08:04.0 Ethernet controller [0200]: Broadcom Corporation NetXtreme BCM5715 Gigabit Ethernet [14e4:1678] (rev a3) Subsystem: Hewlett-Packard Company NC326i PCIe Dual Port Gigabit Server Adapter [103c:703e]
Full lspci listing and superio and flashrom output for the DL165 follows, as well as updated patch.
Signed-off-by: Arne Georg Gleditscharne.gleditsch@numascale.com
Mh. The lspci you linked to suggests that the network card is in a PCI-X slot and not on board. I diffed both PCI configurations against each other:
[..]
Unless we're totally sure the network controllers are onboard for both boards, I think we should
- use the LPC device subsystem on both boards or
- use the 1166:0103 PCI bridge on DL145 and the 1166:0238 WDTimer on
DL165 or
- use DMI strings and coreboot IDs.
The DL165 network controllers are definitely onboard. Re the DL145s, coreboot's src/mainboard/hp/dl145_g3/mptable.c contains the following
//onboard Broadcom GbE smp_write_intsrc(mc, mp_INT, MP_IRQ_TRIGGER_LEVEL|MP_IRQ_POLARITY_LOW,8, (4<<2)|0, m->apicid_bcm5785[2], 0x4); smp_write_intsrc(mc, mp_INT, MP_IRQ_TRIGGER_LEVEL|MP_IRQ_POLARITY_LOW,8, (4<<2)|1, m->apicid_bcm5785[2], 0x4);
which strongly indicates that this is the case for DL145 NICs too. (The two boxes are otherwise very similar, so anything else would be very surprising.)
Ok, I'm pinging this again. I don't mind maintaining this patch in my own tree for my personal use, but it would be nice to not have to. Do you want it upstream, or should I just drop it?
Acked-by: Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net
Updated to apply against latest svn and added a few comments.
Thanks for your patch, committed in r1065.
Regards, Carl-Daniel