From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail.mock.com (gw.mock.com [209.157.146.194]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client CN "mail.mock.com", Issuer "CAcert Class 3 Root" (not verified)) by ozlabs.org (Postfix) with ESMTP id 3552F67B56 for ; Tue, 10 Oct 2006 01:44:10 +1000 (EST) Message-ID: <452A68FB.8010204@mock.com> Date: Mon, 09 Oct 2006 08:21:31 -0700 From: Jeff Mock MIME-Version: 1.0 To: powerpc440 Subject: Re: Problem with OPB access on 440GX and Linux 2.6.15 References: <452A52B4.1010303@googlemail.com> In-Reply-To: <452A52B4.1010303@googlemail.com> Content-Type: text/plain; charset=windows-1252 Cc: linuxppc-embedded@ozlabs.org List-Id: Linux on Embedded PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Hey, You're using the EBC, which is hooked up to the OPB (if I've got my IBM letter soup correct). The 440GX has 36-bit physical address. The EBC is mapped at address 0x100000000ULL-0x13fffffffULL. The phys_addr_t type in the PPC kernel (first parameter to ioremap) is an 8-byte thing to hold the 36-bit address. So, to map the physical address of the I/O device, your call should look something like: p = ioremap(0x100000000ULL, 1<<20); The actual address is set by the way you program the EBC registers for the chip select on your device. It's a good idea to look at all 8 pairs of registers for configuring chip selects, if they overlap or if one pair is strangely configured weird things can happen. EBC0_BxCR EBC0_BxCP It sounds like your EBC registers are getting setup in u-boot and not in the kernel since you can access the device in u-boot. I think this is a good way to do things. You probably just have the wrong physical address in your driver. jeff powerpc440 wrote: > Hi everybody, > > I'm having a PPC440GX based board, based on Ocotea. > On the board is a Xilinx CPLD chip, connected with OPB (On-Chip > Peripheral Bus). > Under U-Boot I can communicate – reading and writing, without any > problems, but when the Linux kernel ones is loaded, then I'm getting > error message: “Bus error”. This error message appears on reading or > writing. > How I can access the units, which are connected with OPB? > > There are the source code from the driver, that I use : > > static void io_addr = ioremap(0x48300000, 0x100000); /* allocate 1MB > adress space */ > long value = readl(io_addr + 0x7000); > > and the error message: > > VA: 0xe2180000 > Machine check in kernel mode. > Data Read PLB Error > PLB0: BEAR=0x0000000048307000 ACR= 0x9b000000 BESR= 0x0c000000 > POB0: BEAR=0x00000000040c0020 BESR0=0x00000000 BESR1=0x00000000 > OPB0: BEAR=0x0000000000000000 BSTAT=0x00000000 > Oops: machine check, sig: 7 [#4] > NIP: E107D078 LR: E107D068 CTR: 00000000 > REGS: c0335f50 TRAP: 0202 Not tainted (2.6.16.27-himadef-0) > MSR: 00029000 CR: 22004024 XER: 20000000 > TASK = dfa1cc10[412] 'cat' THREAD: dcc54000 > GPR00: E107D068 DCC55E80 DFA1CC10 00000010 00000010 E107D7AB D490F704 > 00000010 > GPR08: 00000000 E2180000 00000000 FFFFFFFF 22004048 1001C334 00000000 > 00000400 > GPR16: 00000001 100B0000 10014344 00000000 00000002 00000000 00000000 > DDAC8600 > GPR24: 00000000 DCC55EB8 DD3E5000 00000C00 E1080000 00000010 10014204 > 00001000 > NIP [E107D078] sensor_proc_read+0x58/0x94 [himaio] > LR [E107D068] sensor_proc_read+0x48/0x94 [himaio] > Call Trace: > [DCC55E80] [E107D068] sensor_proc_read+0x48/0x94 [himaio] (unreliable) > [DCC55EB0] [C008C6EC] proc_file_read+0x2a4/0x354 > [DCC55EF0] [C00564F0] vfs_read+0xcc/0x1a0 > [DCC55F10] [C0057228] sys_read+0x4c/0x90 > [DCC55F40] [C00019C4] ret_from_syscall+0x0/0x3c > Instruction dump: > 7cf93b78 4800048d 80dce160 3ca0e108 38a5d7a0 7f43d378 7f64db78 48000485 > 813ce160 7c7d1b78 38c97000 7cc0342c <0c060000> 4c00012c 3ca0e108 7c7a1a14 > > Thanks in advance! > _______________________________________________ > Linuxppc-embedded mailing list > Linuxppc-embedded@ozlabs.org > https://ozlabs.org/mailman/listinfo/linuxppc-embedded >