From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <003301c383ef$c9b17ae0$c50aa8c0@ksshin> From: =?ks_c_5601-1987?B?vcWw5rz2?= To: "Mark A. Greer" Cc: References: <009501c37dba$0f462470$c50aa8c0@ksshin> <3F723C33.9030003@mvista.com> <003b01c38321$81114d20$c50aa8c0@ksshin> <3F738F27.8000200@mvista.com> Subject: Re: PCI DMA in GT64260 Date: Fri, 26 Sep 2003 14:33:59 +0900 MIME-Version: 1.0 Content-Type: text/plain; charset="ks_c_5601-1987" Sender: owner-linuxppc-embedded@lists.linuxppc.org List-Id: Thank you. I check the GPP Pins and test that pins with osciloscope. At the time when CPU commands to the L2 S/W chip that 'transmit data', L2 S/W chip asserts PCI_REQ* signal(this is connected to GPP pin) and GT64260 alos asserts PCI_GNT* signal(also this pin is connected to GPP pin). and L2 S/W chip also asserts PCI_FRAME* signal. So I think GPP setting is no problem. I think that the value of BAR Regs in GT64260 Config Space is wrong, so I try to change the value with PCI_CONFIG_ADDR, PCI_CONFIG_DATA, when I scan PCI Bus, in bus no 0 or 1, dev no 0, I found GT64260 PCI Bridge, But I cannot change the value of BAR in PCI CONFIG Space. How can I change the value of BAR in PCI CONFIG Space. another is, you say that there are several working examples. But I can't found any examples related to PCI_DMA or changing the value of BAR in PCI CONFIG Space. please let me know how to solve the PCI DMA ----- Original Message ----- From: "Mark A. Greer" To: "신경수" Cc: Sent: Friday, September 26, 2003 9:58 AM Subject: Re: PCI DMA in GT64260 > > 신경수 wrote: > > >Thank you for your response. > > > >Well, Instead of giving my all code, I give the register lists and the value > >related to address decoding. please check the register and the value. > > > Well, there are several working examples that you can look at so I'm not > going to spend much time groping through registers. However, from what > you describe, I would guess that you have either a PCI arbitration > problem or your PCI slave window is horked. I did a very quick look at > the PCI slave windows & related regs and I didn't notice anything > wrong--that doesn't mean there isn't--so it may be an arbitration problem. > > Are you using the 64260's PCI arbiter or do you have an external one? If > 64260, then make sure that its configured correctly and that you have > the appropriate GPP pins configured correctly. Again, several working > examples so don't be shy about looking at them. > > Mark > > > ** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/