Hi, Attached is a patch to board_enable.c for the HP DL165 board. Apparently, the PCI signature of the DL165 is identical to the DL145, but the enable pins are placed differently... :/ I'm not sure what the best way to handle that is. Signed-off-by: Arne Georg Gleditsch <arne.gleditsch@numascale.com> -- Arne.
Am Mittwoch, den 26.05.2010, 13:27 +0200 schrieb Arne Georg Gleditsch:
Attached is a patch to board_enable.c for the HP DL165 board. Apparently, the PCI signature of the DL165 is identical to the DL145, but the enable pins are placed differently... :/ I'm not sure what the best way to handle that is. Try to find an lspci dump of that DL145 and check whether some PCI signature is different. If not, check whether DMI contains some string that makes it possible to tell the systems apart. If nothing helps, PCI subsystem IDs (not the normal vendor/device IDs!) have to be ripped out for both systems, disabling auto-detection, and manual -m is needed in any case.
Regards, Michael Karcher
On 26. mai 2010 15:56, Michael Karcher wrote:
Try to find an lspci dump of that DL145 and check whether some PCI signature is different. If not, check whether DMI contains some string that makes it possible to tell the systems apart. If nothing helps, PCI subsystem IDs (not the normal vendor/device IDs!) have to be ripped out for both systems, disabling auto-detection, and manual -m is needed in any case.
As Google has it, it would appear that the onboard NIC is different: DL145: 08:04.0 Ethernet controller [0200]: Broadcom Corporation NetXtreme BCM5715 Gigabit Ethernet [14e4:1678] (rev a3) 08:04.1 Ethernet controller [0200]: Broadcom Corporation NetXtreme BCM5715 Gigabit Ethernet [14e4:1678] (rev a3) DL165: 02:02.0 Ethernet controller [0200]: Broadcom Corporation NetXtreme BCM5704 Gigabit Ethernet [14e4:1648] (rev 10) 02:02.1 Ethernet controller [0200]: Broadcom Corporation NetXtreme BCM5704 Gigabit Ethernet [14e4:1648] (rev 10) Perhaps that could be used for the secondary PCI device in the signature? I believe the device currently listed there is the on-board VGA chip; I'm not sure why that was chosen. -- Arne.
Am Mittwoch, den 26.05.2010, 16:08 +0200 schrieb Arne Georg Gleditsch:
On 26. mai 2010 15:56, Michael Karcher wrote:
Try to find an lspci dump of that DL145 and check whether some PCI signature is different. If not, check whether DMI contains some string
As Google has it, it would appear that the onboard NIC is different:
DL145: 08:04.0 Ethernet controller [0200]: Broadcom Corporation NetXtreme BCM5715 Gigabit Ethernet [14e4:1678] (rev a3) 08:04.1 Ethernet controller [0200]: Broadcom Corporation NetXtreme BCM5715 Gigabit Ethernet [14e4:1678] (rev a3)
DL165: 02:02.0 Ethernet controller [0200]: Broadcom Corporation NetXtreme BCM5704 Gigabit Ethernet [14e4:1648] (rev 10) 02:02.1 Ethernet controller [0200]: Broadcom Corporation NetXtreme BCM5704 Gigabit Ethernet [14e4:1648] (rev 10)
Perhaps that could be used for the secondary PCI device in the signature? I believe the device currently listed there is the on-board VGA chip; I'm not sure why that was chosen.
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, your patch is in my oppinion ready to be added to flashrom. Regards, Michael Karcher [1]: http://www.coreboot.org/Development_Guidelines#Sign-off_Procedure
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 Gleditsch <arne.gleditsch@numascale.com> -- Arne.
On 26. mai 2010 19:03, Arne Georg Gleditsch wrote:
Full lspci listing and superio and flashrom output for the DL165 follows, as well as updated patch.
Any takers? -- Arne.
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 Gleditsch <arne.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: --- dl145.txt 2010-06-07 16:25:52.000000000 +0200 +++ dl165.txt 2010-06-07 16:25:56.000000000 +0200 @@ -1,35 +1,38 @@ 00:01.0 PCI bridge [0604]: Broadcom BCM5785 [HT1000] PCI/PCI-X Bridge [1166:0036] - Capabilities: [d8] Subsystem: Gammagraphx, Inc. (or missing ID) Device [0000:0000] + Capabilities: [d8] Subsystem: Broadcom BCM5785 [HT1000] PCI/PCI-X Bridge [1166:0036] 00:02.0 Host bridge [0600]: Broadcom BCM5785 [HT1000] Legacy South Bridge [1166:0205] Subsystem: Broadcom Device [1166:0201] 00:02.1 IDE interface [0101]: Broadcom BCM5785 [HT1000] IDE [1166:0214] Subsystem: Hewlett-Packard Company Device [103c:320b] 00:02.2 ISA bridge [0601]: Broadcom BCM5785 [HT1000] LPC [1166:0234] - Subsystem: Broadcom Device [1166:0230] + Subsystem: Hewlett-Packard Company Device [103c:320d] +00:02.3 System peripheral [0880]: Broadcom BCM5785 [HT1000] WDTimer [1166:0238] (rev 01) + Subsystem: Broadcom BCM5785 [HT1000] WDTimer [1166:0238] 00:03.0 USB Controller [0c03]: Broadcom BCM5785 [HT1000] USB [1166:0223] (rev 01) Subsystem: Hewlett-Packard Company Device [103c:320c] 00:03.1 USB Controller [0c03]: Broadcom BCM5785 [HT1000] USB [1166:0223] (rev 01) Subsystem: Hewlett-Packard Company Device [103c:320c] 00:03.2 USB Controller [0c03]: Broadcom BCM5785 [HT1000] USB [1166:0223] (rev 01) Subsystem: Hewlett-Packard Company Device [103c:320d] -00:04.0 VGA compatible controller [0300]: Matrox Graphics, Inc. MGA G200e [Pilot] ServerEngines (SEP1) [102b:0522] (rev 02) - Subsystem: Hewlett-Packard Company Device [103c:31fa] 00:06.0 PCI bridge [0604]: Broadcom HT2100 PCI-Express Bridge [1166:0140] (rev a2) - Capabilities: [78] Subsystem: Gammagraphx, Inc. (or missing ID) Device [0000:0000] + Capabilities: [78] Subsystem: Hewlett-Packard Company Device [103c:0140] 00:07.0 PCI bridge [0604]: Broadcom HT2100 PCI-Express Bridge [1166:0142] (rev a2) - Capabilities: [78] Subsystem: Gammagraphx, Inc. (or missing ID) Device [0000:0000] + Capabilities: [78] Subsystem: Hewlett-Packard Company Device [103c:0142] 00:08.0 PCI bridge [0604]: Broadcom HT2100 PCI-Express Bridge [1166:0144] (rev a2) - Capabilities: [78] Subsystem: Gammagraphx, Inc. (or missing ID) Device [0000:0000] + Capabilities: [78] Subsystem: Hewlett-Packard Company Device [103c:0144] 00:09.0 PCI bridge [0604]: Broadcom HT2100 PCI-Express Bridge [1166:0142] (rev a2) - Capabilities: [78] Subsystem: Gammagraphx, Inc. (or missing ID) Device [0000:0000] + Capabilities: [78] Subsystem: Hewlett-Packard Company Device [103c:0142] 00:0a.0 PCI bridge [0604]: Broadcom HT2100 PCI-Express Bridge [1166:0144] (rev a2) - Capabilities: [78] Subsystem: Gammagraphx, Inc. (or missing ID) Device [0000:0000] + Capabilities: [78] Subsystem: Hewlett-Packard Company Device [103c:0144] 01:0d.0 PCI bridge [0604]: Broadcom BCM5785 [HT1000] PCI/PCI-X Bridge [1166:0104] (rev c0) - Capabilities: [88] Subsystem: Device [c133:a9d6] + Capabilities: [88] Subsystem: Hewlett-Packard Company Device [103c:0104] 01:0e.0 RAID bus controller [0104]: Broadcom BCM5785 [HT1000] SATA (Native SATA Mode) [1166:024a] Subsystem: Hewlett-Packard Company Device [103c:320b] -07:00.0 PCI bridge [0604]: Broadcom EPB PCI-Express to PCI-X Bridge [1166:0103] (rev b5) -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] -08:04.1 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] +02:02.0 Ethernet controller [0200]: Broadcom Corporation NetXtreme BCM5704 Gigabit Ethernet [14e4:1648] (rev 10) + Subsystem: Hewlett-Packard Company Device [103c:310f] +02:02.1 Ethernet controller [0200]: Broadcom Corporation NetXtreme BCM5704 Gigabit Ethernet [14e4:1648] (rev 10) + Subsystem: Hewlett-Packard Company Device [103c:310f] +06:00.0 RAID bus controller [0104]: Hewlett-Packard Company Smart Array G6 controllers [103c:323a] (rev 01) + Subsystem: Hewlett-Packard Company Device [103c:3241] +1e:00.0 VGA compatible controller [0300]: Matrox Graphics, Inc. MGA G200e [Pilot] ServerEngines (SEP1) [102b:0522] (rev 02) + Subsystem: Hewlett-Packard Company Device [103c:31fa] 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. Regards, Carl-Daniel -- http://www.hailfinger.org/
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 Gleditsch<arne.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.) -- Arne.
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 Gleditsch<arne.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? -- Arne.
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 Gleditsch<arne.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 -- http://www.hailfinger.org/
participants (3)
-
Arne Georg Gleditsch -
Carl-Daniel Hailfinger -
Michael Karcher