On Sat, Nov 12, 2011 at 12:16:05PM -0600, Kumar Gala wrote:
This entire file is identical to board/freescale/p1_p2_rdb/ddr.c. In fact, since this board only boots via the On-Chip ROM, the whole file is useless: fixed_sdram() should just return the RAM size. We're running from RAM when this function executes.
Is it ok with you if I replace the entire file with the following?
phys_size_t fixed_sdram(void) { return CONFIG_SYS_SDRAM_SIZE << 20; }
If the board has SO-DIMMs than I'd expect SPD support. Sounds like you're working on this w/Matt & York.
I don't need fixed_sdram() anymore. I figured out how to get L2 SRAM boot working this week. Also, I finally tracked down a bug in the Freescale DDR SPD code, exposed due to invalid data in the SPD. Darn hardware manufacturers!
diff --git a/board/freescale/p2020come/law.c b/board/freescale/p2020come/law.c new file mode 100644 index 0000000..56508db --- /dev/null +++ b/board/freescale/p2020come/law.c @@ -0,0 +1,36 @@ +/*
- Copyright 2009 Freescale Semiconductor, Inc.
- See file CREDITS for list of people who contributed to this
- project.
- This program is free software; you can redistribute it and/or
- modify it under the terms of the GNU General Public License as
- published by the Free Software Foundation; either version 2 of
- the License, or (at your option) any later version.
- This program is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.See the
- GNU General Public License for more details.
- You should have received a copy of the GNU General Public License
- along with this program; if not, write to the Free Software
- Foundation, Inc., 59 Temple Place, Suite 330, Boston,
- MA 02111-1307 USA
- */
+#include <common.h> +#include <asm/fsl_law.h> +#include <asm/mmu.h>
+struct law_entry law_table[] = {
- SET_LAW(CONFIG_SYS_PCIE1_MEM_PHYS, LAW_SIZE_512M, LAW_TRGT_IF_PCIE_1),
- SET_LAW(CONFIG_SYS_PCIE1_IO_PHYS, LAW_SIZE_64K, LAW_TRGT_IF_PCIE_1),
- SET_LAW(CONFIG_SYS_PCIE2_MEM_PHYS, LAW_SIZE_512M, LAW_TRGT_IF_PCIE_2),
- SET_LAW(CONFIG_SYS_PCIE2_IO_PHYS, LAW_SIZE_64K, LAW_TRGT_IF_PCIE_2),
- SET_LAW(CONFIG_SYS_PCIE3_MEM_PHYS, LAW_SIZE_512M, LAW_TRGT_IF_PCIE_3),
- SET_LAW(CONFIG_SYS_PCIE3_IO_PHYS, LAW_SIZE_64K, LAW_TRGT_IF_PCIE_3),
We normally set these up dynamically.
This is a modified version of the code from board/freescale/p1_p2_rdb/law.c. Can you suggest an in tree example of the way you'd like the code to look? I copied what I assume is a good example…
If you look at current board/freescale/p1_p2_rdb/law.c it doesn't have PCI LAWs anymore. I think for your example you just need an empty data structure:
struct law_entry law_table[] = { };
this should hopefully make num_law_entries = 0;
+};
+int num_law_entries = ARRAY_SIZE(law_table);
Unfortunately, having a law_table with no entries causes a bug. A nasty bug which was hard to track down.
When law_table is empty, and when num_law_entries = 0, both variables exist in BSS only.
Both the law_table and num_law_entries are used in init_laws(), called from cpu_init_early_f(). This happens before BSS is initialized. Also before DDR is initialized.
And now you see the bug. The U-Boot hangs due to an invalid memory access. This is before the console is initialized, making it hard to track down.
How do you suggest I work around this? A single redundant entry in the law_table works (the system boots).
Thanks, Ira