In the 3.12 kernel, I turned on earlyprintk so that I could troubleshoot another issue with kexec.
But whenever I enabled earlyprink with the kernel command line option, the boot would hang.
So after a long time of hacking around, inserting printk statements and other hacks, I had some partial success getting earlyprink output.
The first hang occurred when timer2 was disabled,
So I added the flags line in the following structure in omap_hwmod_33xx_data.c:
static struct omap_hwmod am33xx_timer2_hwmod = {
.name = "timer2",
.class = &am33xx_timer_hwmod_class,
.clkdm_name = "l4ls_clkdm",
.flags = (HWMOD_INIT_NO_IDLE | HWMOD_INIT_NO_RESET ),
.prcm = {
.omap4 = {
.modulemode = MODULEMODE_SWCTRL,
},
},
};
That got me much further in the bootup process.
The next hang occurred when the console uart was reset, so I did an incredibly evil hack in omap_hwmod.c: _ocp_softreset(...);
if (strcmp(oh->name, "uart2") == 0)
return -EINVAL;
That got me much further to the point where I get another error:
5.399159] 44e09000.serial: ttyO0 at MMIO 0x44e09000 (irq = 88, base_baud = 3000000) is a OMAP UART0
[ 5.409321] of_get_named_gpio_flags: can't parse gpios property of node '/ocp/serial@48022000[0]'
[ 5.418780] omap_hwmod: uart2: enabling
[ 5.422813] ------------[ cut here ]------------
[ 5.427670] WARNING: CPU: 0 PID: 1 at arch/arm/mach-omap2/omap_hwmod.c:2264 _enable+0x2a0/0x2c8()
[ 5.436879] omap_hwmod: uart2: enabled state can only be entered from initialized, idle, or disabled state
[ 5.446891] Modules linked in:
[ 5.450
Can someone shed some light on this and help me figure out what the right fix should be?
Thanks,
Brian
