Hi,
On Mon, Dec 03, 2012 at 09:22:30AM +1100, Julian Calaby wrote:
> On Mon, Dec 3, 2012 at 3:50 AM, Aaro Koskinen <aaro.koskinen@xxxxxx> wrote:
> > Kernel (3.7-rc7) build is failing with binutils 2.23.1, because _end is
> > no longer absolute for some reason and piggyback can't find it:
> >
> > $ tail -3 System.map
> > 000000000086bcb8 b net_header
> > 000000000086bcc0 B __bss_stop
> > 000000000086bcc0 B _end
> >
> > [...]
> >
> > kernel: arch/sparc/boot/image is ready
> > HOSTCC arch/sparc/boot/piggyback
> > ELFTOAOUT arch/sparc/boot/tftpboot.img
> > PT 0 Entry: Loadable to 0x400000[0x46bcc0] from 0x0[0x40ed60] align 0x100000
> > PT 1 Entry: Note
> > PT 2 Entry: unknown
> > PIGGY arch/sparc/boot/tftpboot.img
> > Could not determine start and end from System.map
> > make[6]: *** [arch/sparc/boot/tftpboot.img] Error 1
> >
> > Is this a bug in binutils, or is the check in piggyback too strict?
>
> Given that the addresses below are the same as the addresses above,
> I'm suspecting that the test is too strict.
>
> I assume that you have a sparc64 system you can reliably boot with
> TFTP, could you try changing the code in the "end_line" function (line
> 91) to look for 'B' symbols and check that it still boots correctly?
Yes, I tried the hack below, and it produces a bootable TFTP image.
diff --git a/arch/sparc/boot/piggyback.c b/arch/sparc/boot/piggyback.c
index c0a798f..9eb1264 100644
--- a/arch/sparc/boot/piggyback.c
+++ b/arch/sparc/boot/piggyback.c
@@ -90,9 +90,9 @@ static int start_line(const char *line)
static int end_line(const char *line)
{
- if (strcmp(line + 8, " A _end\n") == 0)
+ if (strcmp(line + 10, " _end\n") == 0)
return 1;
- else if (strcmp (line + 16, " A _end\n") == 0)
+ else if (strcmp (line + 18, " _end\n") == 0)
return 1;
return 0;
}
A.
--
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
