| Message ID | CAMuHMdWrg+ShPBU3wS=+GKNGgE4-=NBGeK8EP4QoTCuaxbJYaw@mail.gmail.com (mailing list archive) |
|---|---|
| State | Rejected |
| Headers | show |
On 09/01/14 12:21, Geert Uytterhoeven wrote: > Hi Ben, > > On Thu, Jan 9, 2014 at 12:45 PM, Ben Dooks <ben.dooks@codethink.co.uk> wrote: >> I am using the devel branch from Simon Horms' git tree, SHA1 >> 3da593ea7d797d9f59bd6ed1ee629fab681d7a23 and not seeing a >> boot when using the multi arch defconfig and device-tree. >> >> The board does not produce any console output so I am not >> sure what is going on. Has anyone else had issues with using >> the fdt-only boot? > > I'm using Koelsch (not Lager), but usually the non-reference version as I > need Ethernet for NFS root. > The DT-version does boot for me until mounting root. > > Are you using an initramfs? I noticed the kernel crashes early when using > a non-trivial initramfs: No, we're just tftping and then nfs-root. > Error: unrecognized/unsupported machine ID (r1 = 0x41915083). > > Available machine support: > > ID (hex) NAME > ffffffff Generic R8A7791 (Flattened Device Tree) > ffffffff koelsch > > Please check your kernel config and/or bootloader. > > Apparently the device tree gets overwritten in memory. I tried loading the > DTB at a different address, but that didn't help. I didn't have more time > to investigate, though. I've managed to get it to boot with another boot method, however that has other limitations... > > FYI, I'm using the following patch to capture early kernel messages in RAM. > After a reset, look up the addres of big_buffer[] in System.map, and > inspect it from U-Boot to retrieve the messages (translate from 0xc0000000 > to 0x40000000). Ta, that's a useful thing to have i am considering it may be a dtc compiler bug or some other issue with the device-tree.
On 09/01/14 12:21, Geert Uytterhoeven wrote: > Hi Ben, > > On Thu, Jan 9, 2014 at 12:45 PM, Ben Dooks <ben.dooks@codethink.co.uk> wrote: >> I am using the devel branch from Simon Horms' git tree, SHA1 >> 3da593ea7d797d9f59bd6ed1ee629fab681d7a23 and not seeing a >> boot when using the multi arch defconfig and device-tree. >> >> The board does not produce any console output so I am not >> sure what is going on. Has anyone else had issues with using >> the fdt-only boot? Ok, having looked at this we have the following issues: sh_cmt sh _cmt.0: cannot get clock...4 sh_cmt: probe of sh_cmt.0 failed with error -2 sh-sci.4: can't get iclk. sh-sci: probe of sh-sci.4 failed with error -2
diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c index be7c86bae576..609bccc002ca 100644 --- a/kernel/printk/printk.c +++ b/kernel/printk/printk.c @@ -46,6 +46,7 @@ #include <linux/irq_work.h> #include <linux/utsname.h> +#include <asm/cacheflush.h> #include <asm/uaccess.h> #define CREATE_TRACE_POINTS @@ -1488,6 +1489,9 @@ static size_t cont_print_text(char *text, size_t size) return textlen; } +static char big_buffer[256*1024]; +static size_t big_buffer_index; + asmlinkage int vprintk_emit(int facility, int level, const char *dict, size_t dictlen, const char *fmt, va_list args) @@ -1546,6 +1550,16 @@ asmlinkage int vprintk_emit(int facility, int level, * prefix which might be passed-in as a parameter. */ text_len = vscnprintf(text, sizeof(textbuf), fmt, args); +if (big_buffer_index + text_len < sizeof(big_buffer)) { + memcpy(&big_buffer[big_buffer_index], text, text_len + 1); + big_buffer_index += text_len; +} else if (big_buffer_index < sizeof(big_buffer)) { + memset(&big_buffer[big_buffer_index], '*', + sizeof(big_buffer) - big_buffer_index); + big_buffer[sizeof(big_buffer) - 1] = 0; + big_buffer_index = sizeof(big_buffer); +} +flush_cache_all(); /* mark and strip a trailing newline */ if (text_len && text[text_len-1] == '\n') {