Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 1 | Intro: |
| 2 | ====== |
Joe Hamman | 9e3ed39 | 2007-12-13 06:45:14 -0600 | [diff] [blame] | 3 | |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 4 | The SBC8548 is a stand alone single board computer with a 1GHz |
| 5 | MPC8548 CPU, 8MB boot flash, 64MB user flash and, 256MB DDR2 400MHz |
| 6 | memory. It also has 128MB SDRAM 100MHz LBC memory, with both a PCI-e, |
| 7 | and a PCI-X slot, dual mini-DB9 for UART, and dual RJ-45 for eTSEC |
| 8 | ethernet connections. |
Joe Hamman | 9e3ed39 | 2007-12-13 06:45:14 -0600 | [diff] [blame] | 9 | |
Bin Meng | a187559 | 2016-02-05 19:30:11 -0800 | [diff] [blame] | 10 | U-Boot Configuration: |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 11 | ===================== |
Joe Hamman | 9e3ed39 | 2007-12-13 06:45:14 -0600 | [diff] [blame] | 12 | |
Bin Meng | a187559 | 2016-02-05 19:30:11 -0800 | [diff] [blame] | 13 | The following possible U-Boot configuration targets are available: |
Joe Hamman | 9e3ed39 | 2007-12-13 06:45:14 -0600 | [diff] [blame] | 14 | |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 15 | 1) sbc8548_config |
| 16 | 2) sbc8548_PCI_33_config |
| 17 | 3) sbc8548_PCI_66_config |
| 18 | 4) sbc8548_PCI_33_PCIE_config |
| 19 | 5) sbc8548_PCI_66_PCIE_config |
| 20 | |
| 21 | Generally speaking, most people should choose to use #5. Details |
| 22 | of each choice are listed below. |
| 23 | |
| 24 | Choice #1 does not enable CONFIG_PCI, and assumes that the PCI slot |
| 25 | will be left empty (M66EN high), and so the board will operate with |
Bin Meng | a187559 | 2016-02-05 19:30:11 -0800 | [diff] [blame] | 26 | a base clock of 66MHz. Note that you need both PCI enabled in U-Boot |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 27 | and linux in order to have functional PCI under linux. |
| 28 | |
| 29 | The second enables PCI support and builds for a 33MHz clock rate. Note |
| 30 | that if a 33MHz 32bit card is inserted in the slot, then the whole board |
| 31 | will clock down to a 33MHz base clock instead of the default 66MHz. This |
| 32 | will change the baud clocks and mess up your serial console output if you |
| 33 | were previously running at 66MHz. If you want to use a 33MHz PCI card, |
| 34 | then you should build a U-Boot with a _PCI_33_ config and store this |
| 35 | to flash prior to powering down the board and inserting the 33MHz PCI |
| 36 | card. [The above discussion assumes that the SW2[1-4] has not been changed |
| 37 | to reflect a different CCB:SYSCLK ratio] |
| 38 | |
| 39 | The third option builds PCI support in, and leaves the clocking at the |
Wolfgang Denk | f5cdc11 | 2012-04-16 23:13:51 +0200 | [diff] [blame] | 40 | default 66MHz. Options four and five are just repeats of option two |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 41 | and three, but with PCI-e support enabled as well. |
| 42 | |
| 43 | PCI output listing with an intel e1000 PCI-x and a Syskonnect SK-9Exx |
Wolfgang Denk | f5cdc11 | 2012-04-16 23:13:51 +0200 | [diff] [blame] | 44 | is shown below for sbc8548_PCI_66_PCIE_config. (Note that PCI-e with |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 45 | a 33MHz PCI configuration is currently untested.) |
| 46 | |
| 47 | => pci 0 |
| 48 | Scanning PCI devices on bus 0 |
Wolfgang Denk | f5cdc11 | 2012-04-16 23:13:51 +0200 | [diff] [blame] | 49 | BusDevFun VendorId DeviceId Device Class Sub-Class |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 50 | _____________________________________________________________ |
Wolfgang Denk | f5cdc11 | 2012-04-16 23:13:51 +0200 | [diff] [blame] | 51 | 00.00.00 0x1057 0x0012 Processor 0x20 |
| 52 | 00.01.00 0x8086 0x1026 Network controller 0x00 |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 53 | => pci 1 |
| 54 | Scanning PCI devices on bus 1 |
Wolfgang Denk | f5cdc11 | 2012-04-16 23:13:51 +0200 | [diff] [blame] | 55 | BusDevFun VendorId DeviceId Device Class Sub-Class |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 56 | _____________________________________________________________ |
Wolfgang Denk | f5cdc11 | 2012-04-16 23:13:51 +0200 | [diff] [blame] | 57 | 01.00.00 0x1957 0x0012 Processor 0x20 |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 58 | => pci 2 |
| 59 | Scanning PCI devices on bus 2 |
Wolfgang Denk | f5cdc11 | 2012-04-16 23:13:51 +0200 | [diff] [blame] | 60 | BusDevFun VendorId DeviceId Device Class Sub-Class |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 61 | _____________________________________________________________ |
Wolfgang Denk | f5cdc11 | 2012-04-16 23:13:51 +0200 | [diff] [blame] | 62 | 02.00.00 0x1148 0x9e00 Network controller 0x00 |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 63 | => |
Joe Hamman | 9e3ed39 | 2007-12-13 06:45:14 -0600 | [diff] [blame] | 64 | |
Paul Gortmaker | 7e44f2b | 2011-12-30 23:53:10 -0500 | [diff] [blame] | 65 | Memory Size and using SPD: |
| 66 | ========================== |
| 67 | |
| 68 | The default configuration uses hard coded memory configuration settings |
| 69 | for 256MB of DDR2 @400MHz. It does not by default use the DDR2 SPD |
| 70 | EEPROM data to read what memory is installed. |
| 71 | |
| 72 | There is a hardware errata, which causes the older local bus SDRAM |
| 73 | SPD EEPROM to land at the same address as the DDR2 SPD EEPROM, so |
Paul Gortmaker | 3e3262b | 2011-12-30 23:53:12 -0500 | [diff] [blame] | 74 | that the SPD data can not be read reliably. You can test if your |
| 75 | board has the errata fix by running "i2c probe". If you see 0x53 |
| 76 | as a valid device, it has been fixed. If you only see 0x50, 0x51 |
| 77 | then your board does not have the fix. |
| 78 | |
| 79 | You can also visually inspect the board to see if this hardware |
| 80 | fix has been applied: |
| 81 | |
| 82 | 1) Remove R314 (RES-R0174-033, 1K, 0603). R314 is located on |
Wolfgang Denk | f5cdc11 | 2012-04-16 23:13:51 +0200 | [diff] [blame] | 83 | the back of the PCB behind the DDR SDRAM SODIMM connector. |
Paul Gortmaker | 3e3262b | 2011-12-30 23:53:12 -0500 | [diff] [blame] | 84 | 2) Solder RES-R0174-033 (1K, 0603) resistor from R314 pin 2 pad |
Wolfgang Denk | f5cdc11 | 2012-04-16 23:13:51 +0200 | [diff] [blame] | 85 | to R313 pin 2. Pin 2 for each resistor is the end of the |
| 86 | resistor closest to the CPU. |
Paul Gortmaker | 3e3262b | 2011-12-30 23:53:12 -0500 | [diff] [blame] | 87 | |
| 88 | Boards without the mod will have R314 and R313 in parallel, like "||". |
| 89 | After the mod, they will be touching and form an "L" shape. |
Paul Gortmaker | 7e44f2b | 2011-12-30 23:53:10 -0500 | [diff] [blame] | 90 | |
| 91 | If you want to upgrade to larger RAM size, you can simply enable |
| 92 | #define CONFIG_SPD_EEPROM |
| 93 | #define CONFIG_DDR_SPD |
| 94 | in include/configs/sbc8548.h file. (The lines are already there |
| 95 | but listed as #undef). |
| 96 | |
Paul Gortmaker | 3e3262b | 2011-12-30 23:53:12 -0500 | [diff] [blame] | 97 | If you did the i2c test, and your board does not have the errata |
| 98 | fix, then you will have to physically remove the LBC 128MB DIMM |
Paul Gortmaker | 7e44f2b | 2011-12-30 23:53:10 -0500 | [diff] [blame] | 99 | from the board's socket to resolve the above i2c address overlap |
| 100 | issue and allow SPD autodetection of RAM to work. |
| 101 | |
Joe Hamman | 9e3ed39 | 2007-12-13 06:45:14 -0600 | [diff] [blame] | 102 | |
Bin Meng | a187559 | 2016-02-05 19:30:11 -0800 | [diff] [blame] | 103 | Updating U-Boot with U-Boot: |
Paul Gortmaker | dd9ca98 | 2009-09-25 11:14:11 -0400 | [diff] [blame] | 104 | ============================ |
| 105 | |
Bin Meng | a187559 | 2016-02-05 19:30:11 -0800 | [diff] [blame] | 106 | Note that versions of U-Boot up to and including 2009.08 had U-Boot stored |
Paul Gortmaker | dd9ca98 | 2009-09-25 11:14:11 -0400 | [diff] [blame] | 107 | at 0xfff8_0000 -> 0xffff_ffff (512k). Currently it is being stored from |
| 108 | 0xfffa_0000 -> 0xffff_ffff (384k). If you use an old macro/script to |
Bin Meng | a187559 | 2016-02-05 19:30:11 -0800 | [diff] [blame] | 109 | update U-Boot with U-Boot and it uses the old address, you will render |
Paul Gortmaker | dd9ca98 | 2009-09-25 11:14:11 -0400 | [diff] [blame] | 110 | your board inoperable, and you will require JTAG recovery. |
| 111 | |
| 112 | The following steps list how to update with the current address: |
| 113 | |
| 114 | tftp u-boot.bin |
| 115 | md 200000 10 |
| 116 | protect off all |
| 117 | erase fffa0000 ffffffff |
| 118 | cp.b 200000 fffa0000 60000 |
| 119 | md fffa0000 10 |
| 120 | protect on all |
| 121 | |
| 122 | The "md" steps in the above are just a precautionary step that allow |
Bin Meng | a187559 | 2016-02-05 19:30:11 -0800 | [diff] [blame] | 123 | you to confirm the U-Boot version that was downloaded, and then confirm |
Paul Gortmaker | dd9ca98 | 2009-09-25 11:14:11 -0400 | [diff] [blame] | 124 | that it was copied to flash. |
| 125 | |
Paul Gortmaker | f0aec4e | 2011-12-30 23:53:08 -0500 | [diff] [blame] | 126 | The above assumes that you are using the default board settings which |
Bin Meng | a187559 | 2016-02-05 19:30:11 -0800 | [diff] [blame] | 127 | have U-Boot in the 8MB flash, tied to /CS0. |
Paul Gortmaker | f0aec4e | 2011-12-30 23:53:08 -0500 | [diff] [blame] | 128 | |
| 129 | If you are running the default 8MB /CS0 settings but want to store an |
| 130 | image in the SODIMM that is built with CONFIG_SYS_ALT_BOOT enabled, |
| 131 | (as a backup, etc) then the steps will become: |
| 132 | |
| 133 | tftp u-boot.bin |
| 134 | md 200000 10 |
| 135 | protect off all |
| 136 | era eff00000 efffffff |
| 137 | cp.b 200000 eff00000 100000 |
| 138 | md eff00000 10 |
| 139 | protect on all |
| 140 | |
| 141 | Finally, if you are running the alternate 64MB /CS0 settings and want |
Bin Meng | a187559 | 2016-02-05 19:30:11 -0800 | [diff] [blame] | 142 | to update the in-use U-Boot image, then (again with CONFIG_SYS_ALT_BOOT |
Paul Gortmaker | f0aec4e | 2011-12-30 23:53:08 -0500 | [diff] [blame] | 143 | enabled) the steps will become: |
| 144 | |
| 145 | tftp u-boot.bin |
| 146 | md 200000 10 |
| 147 | protect off all |
| 148 | era fff00000 ffffffff |
| 149 | cp.b 200000 fff00000 100000 |
| 150 | md fff00000 10 |
| 151 | protect on all |
| 152 | |
Paul Gortmaker | dd9ca98 | 2009-09-25 11:14:11 -0400 | [diff] [blame] | 153 | |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 154 | Hardware Reference: |
| 155 | =================== |
Joe Hamman | 9e3ed39 | 2007-12-13 06:45:14 -0600 | [diff] [blame] | 156 | |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 157 | The following contains some summary information on hardware settings |
Bin Meng | a187559 | 2016-02-05 19:30:11 -0800 | [diff] [blame] | 158 | that are relevant to U-Boot, based on the board manual. For the |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 159 | most up to date and complete details of the board, please request the |
| 160 | reference manual ERG-00327-001.pdf from www.windriver.com |
| 161 | |
| 162 | Boot flash: |
| 163 | intel V28F640Jx, 8192x8 (one device) at 0xff80_0000 |
| 164 | |
| 165 | Sodimm flash: |
| 166 | intel V28F128Jx, 16384x8 (4 devices) at 0xfb80_0000 |
Paul Gortmaker | 3fd673c | 2011-12-30 23:53:07 -0500 | [diff] [blame] | 167 | Note that this address reflects the default setting for |
| 168 | the JTAG debugging tools, but since the alignment is |
Bin Meng | a187559 | 2016-02-05 19:30:11 -0800 | [diff] [blame] | 169 | rather inconvenient, U-Boot puts it at 0xec00_0000. |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 170 | |
| 171 | |
| 172 | Jumpers: |
| 173 | |
| 174 | Jumper Name ON OFF |
| 175 | ---------------------------------------------------------------- |
| 176 | JP12 CS0/CS6 swap see note[*] see note[*] |
| 177 | |
| 178 | JP13 SODIMM flash write OK writes disabled |
| 179 | write prot. |
| 180 | |
| 181 | JP14 HRESET/TRST joined isolated |
| 182 | |
| 183 | JP15 PWR ON when AC pwr use S1 for on/off |
| 184 | |
| 185 | JP16 Demo LEDs lit not lit |
| 186 | |
| 187 | JP19 PCI mode PCI PCI-X |
| 188 | |
| 189 | |
| 190 | [*]JP12, when jumpered parallel to the SODIMM, puts the boot flash |
| 191 | onto /CS0 and the SODIMM flash on /CS6 (default). When JP12 |
| 192 | is jumpered parallel to the LBC-SDRAM, then /CS0 is for the |
| 193 | SODIMM flash and /CS6 is for the boot flash. Note that in this |
Paul Gortmaker | f0aec4e | 2011-12-30 23:53:08 -0500 | [diff] [blame] | 194 | alternate setting, you also need to switch SW2.8 to ON. |
| 195 | See the setting CONFIG_SYS_ALT_BOOT if you want to use this setting |
Bin Meng | a187559 | 2016-02-05 19:30:11 -0800 | [diff] [blame] | 196 | and boot U-Boot from the 64MB SODIMM |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 197 | |
| 198 | |
| 199 | Switches: |
| 200 | |
| 201 | The defaults are marked with a * |
| 202 | |
| 203 | Name Desc. ON OFF |
| 204 | ------------------------------------------------------------------ |
| 205 | S1 Pwr toggle n/a n/a |
| 206 | |
| 207 | SW2.1 CFG_SYS_PLL0 1 0* |
| 208 | SW2.2 CFG_SYS_PLL1 1* 0 |
| 209 | SW2.3 CFG_SYS_PLL2 1* 0 |
| 210 | SW2.4 CFG_SYS_PLL3 1 0* |
| 211 | SW2.5 CFG_CORE_PLL0 1* 0 |
| 212 | SW2.6 CFG_CORE_PLL1 1 0* |
| 213 | SW2.7 CFG_CORE_PLL2 1* 0 |
| 214 | SW2.8 CFG_ROM_LOC1 1 0* |
| 215 | |
| 216 | SW3.1 CFG_HOST_AGT0 1* 0 |
| 217 | SW3.2 CFG_HOST_AGT1 1* 0 |
| 218 | SW3.3 CFG_HOST_AGT2 1* 0 |
| 219 | SW3.4 CFG_IO_PORTS0 1* 0 |
| 220 | SW3.5 CFG_IO_PORTS0 1 0* |
| 221 | SW3.6 CFG_IO_PORTS0 1 0* |
| 222 | |
| 223 | SerDes CLK(MHz) SW5.1 SW5.2 |
| 224 | ---------------------------------------------- |
| 225 | 25 0 0 |
| 226 | 100* 1 0 |
| 227 | 125 0 1 |
| 228 | 200 1 1 |
| 229 | |
| 230 | SerDes CLK spread SW5.3 SW5.4 |
| 231 | ---------------------------------------------- |
| 232 | +/- 0.25% 0 0 |
| 233 | -0.50% 1 0 |
| 234 | -0.75% 0 1 |
| 235 | No Spread* 1 1 |
| 236 | |
| 237 | SW4 settings are readable from the EPLD and are currently not used for |
| 238 | any hardware settings (i.e. user configuration switches). |
| 239 | |
| 240 | LEDs: |
| 241 | |
| 242 | Name Desc. ON OFF |
| 243 | ------------------------------------------------------------------ |
| 244 | D13 PCI/PCI-X PCI-X PCI |
| 245 | D14 3.3V PWR 3.3V no power |
| 246 | D15 SYSCLK 66MHz 33MHz |
| 247 | |
| 248 | |
| 249 | Default Memory Map: |
| 250 | |
| 251 | start end CS<n> width Desc. |
| 252 | ---------------------------------------------------------------------- |
| 253 | 0000_0000 0fff_ffff MCS0,1 64 DDR2 (256MB) |
| 254 | f000_0000 f7ff_ffff CS3,4 32 LB SDRAM (128MB) |
| 255 | f800_0000 f8b0_1fff CS5 - EPLD |
Paul Gortmaker | 3fd673c | 2011-12-30 23:53:07 -0500 | [diff] [blame] | 256 | fb80_0000 ff7f_ffff CS6 32 SODIMM flash (64MB) [*] |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 257 | ff80_0000 ffff_ffff CS0 8 Boot flash (8MB) |
| 258 | |
Paul Gortmaker | 3fd673c | 2011-12-30 23:53:07 -0500 | [diff] [blame] | 259 | [*] fb80 represents the default programmed by WR JTAG register files, |
Bin Meng | a187559 | 2016-02-05 19:30:11 -0800 | [diff] [blame] | 260 | but U-Boot places the flash at either ec00 or fc00 based on JP12. |
Paul Gortmaker | 3fd673c | 2011-12-30 23:53:07 -0500 | [diff] [blame] | 261 | |
Paul Gortmaker | bd42bbb | 2009-09-18 19:08:46 -0400 | [diff] [blame] | 262 | The EPLD on CS5 demuxes the following devices at the following offsets: |
| 263 | |
| 264 | offset size width device |
| 265 | -------------------------------------------------------- |
| 266 | 0 1fff 8 7 segment display LED |
| 267 | 10_0000 1fff 4 user switches |
| 268 | 30_0000 1fff 4 HW Rev. register |
| 269 | b0_0000 1fff 8 8kB EEPROM |