U-boot stalled shortly after Early IO Release Succeeded

Boot
1

I am trying to get my custom A10 board to boot into U-boot via QSPI flash. I loaded U-boot and the FPGA itb files using quartus_hps and it gets to the point of printing “FPGA: Early Release Succeeded.”, but stalls shortly after. I have debug defined in my socfpga_arria10_socdk.h file to see debug prints and here’s the tail end of what I receive before it stalls on multiple boards. How can I determine where it’s “stuck”?


FPGA: Additional 10 sync word needed
fdtdec_get_addr_size_fixed: reg: addr=00000000x, size=x
FPGA: Early Release Succeeded.
0 0
   - 0 0 'spi@ff809000'
   - found
spi_find_chip_select: plat=ffe29980, cs=0
spi_get_bus_and_cs: bus=ffe29880, slave=ffe2b158
0 -1
0 0
   - -1 -1 'sysreset'
   - not found
clk_set_defaults(rstmgr@ffd05000)
clk_set_default_parents: could not read assigned-clock-parents for ffe29a10
ofnode_read_prop: assigned-clock-rates: <not found>
fdtdec_get_addr_size_auto_parent: na=1, ns=1, fdtdec_get_addr_size_fixed: reg: addr=00000000x

I did not program any jffs2 filesystem to the QSPI flash - is it needed just to get to a U-boot prompt and NOT boot Linux?

Thank you…

2

Using debug prints, I have found that it gets stuck right at the warm reset reset_cpu(0); function call. I get to checkpoint 6, but not 7 …

/*
* Set this flag to scratch register, so that a proper
* boot progress before / after warm reset can be
* tracked by FSBL
*/
set_regular_boot(true);
WATCHDOG_RESET();
debug(“J: Checkpoint 6…\n”);
reset_cpu(0);
}
debug(“J: Checkpoint 7…\n”);

	/*
	 * Reset this flag to scratch register, so that a proper
	 * boot progress before / after warm reset can be
	 * tracked by FSBL
	 */
	set_regular_boot(false);
3

I found that if I comment out the reset_cpu(0); - then U-boot continues to program my FPGA core image and calibrate DDR, and ends up at the prompt.

Any ideas as to why reset_cpu(0) causes a hang, or, if that is really needed?

1 Like
4

I ran in exactly the same issue (did not try to comment out the reset_cpu() yet. Did you ever found the reason why this was happening ?