[microblaze-uclinux] Re: Boot issues on XUPV5-LX110T

[Dana Goyette <dgoyette@xxxxxxxxxxx>]

John Williams wrote:
> Hi Dana,
>
> On Wed, Aug 26, 2009 at 4:32 AM, Dana 
Goyette<dgoyette-WTGyabjbekOVc3sceRu5cw@xxxxxxxxxxxxxxxx> wrote:
>
> Which toolchain are you using to build the petaLogix git tree?  And
> what about your userland - how are you building that?
>
I actually have two different sets of issues: in Petalinux, I'll use the 
packaged compiler and build scripts; my issues are mostly in getting 
userspace to build.  For example, implicit declaration of "exit" implies 
that the sources fail to #include stdlib.h.

With the upstream kernel, I am using the toolchain from here: 
http://www.monstr.eu/wiki/doku.php?id=fdt:fdt#toolchain
I couldn't find any other remotely usable toolkit... the Xilinx one 
failed to compile, and even made the host compiler segfault.  (Host is a 
64-bit Ubuntu Karmic or Jaunty (I've tried both; no change between the 
two.  A 32-bit host seemed to break in an entirely different way.)

I haven't even begun to build a userspace for the upstream kernel yet -- 
I haven't gotten the kernel to even reach the point where it panics with 
"no init found".

>
> This is an unfortunate side effect of how the kernel handles the
> log_buf - if it dies before initialising the serial console, the log
> messages stay in the buffer.
>
>
> This shouldn't be here.
>
Hmm, I wonder if there's something wrong with the device-tree generator 
itself.  I found one odd line: within the interrupt controller's entry: 
there's a line that says, merely, "interrupt-controller;".
>
> I've not seen this before - have you checked the referenced source
> line?  What is the BUG() condition?
>

>
> Keep things as simple as possible until you figure out the details -
> placing the kernel at a nice round base address would be a good idea.
>
Hmm, since simpleImage makes the bootloader unnecessary, I shall try 
again with the start address at an even 0x1000_0000....   -- I get the 
same error.  I just realized that I had been using the MMU toolchain to 
build a noMMU kernel; if I build an MMU kernel, I get an entirely 
different error that includes garbage characters at the beginning.

À
Gü����0.000000] Linux version 2.6.30-00025-g7368484 (dana@EliteBook) 
(gcc version 4.1.2) #8 Tue Aug 25 15:45:08 PDT 2009
<6>[    0.000000] setup_cpuinfo: initialising
<6>[    0.000000] setup_cpuinfo: Using full CPU PVR support
<6>[    0.000000] setup_memory: max_mapnr: 0x10000
<6>[    0.000000] setup_memory: min_low_pfn: 0x10000
<6>[    0.000000] setup_memory: max_low_pfn: 0x20000
<7>[    0.000000] On node 0 totalpages: 65536
<7>[    0.000000] free_area_init_node: node 0, pgdat c03d872c, 
node_mem_map c040f000
<7>[    0.000000]   Normal zone: 512 pages used for memmap
<7>[    0.000000]   Normal zone: 0 pages reserved
<7>[    0.000000]   Normal zone: 65024 pages, LIFO batch:15
<4>[    0.000000] Built 1 zonelists in Zone order, mobility grouping on. 
 Total pages: 65024
<5>[    0.000000] Kernel command line: console=ttyUL0
<6>[    0.000000] NR_IRQS:32
<4>[    0.000000] Oops: kernel access of bad area, sig: 11
<6>[    0.000000]  Registers dump: mode=1
<6>[    0.000000]  r1=C03DDFAC, r2=C03A28B4, r3=00000000, r4=00000000
<6>[    0.000000]  r5=C03D4B70, r6=C03483E4, r7=00000000, r8=00000FFC
<6>[    0.000000]  r9=00000600, r10=00000001, r11=0000006C, r12=C03D872C
<6>[    0.000000]  r13=C03DE000, r14=C0009518, r15=C03E1C04, r16=C027B818
<6>[    0.000000]  r17=305D2000, r18=00000000, r19=CFFFFDEC, r20=00000000
<6>[    0.000000]  r21=00000000, r22=00000004, r23=C0303358, r24=00000000
<6>[    0.000000]  r25=00000000, r26=00000000, r27=00000000, r28=00000000
<6>[    0.000000]  r29=00000000, r30=00000000, r31=C03A5464, rPC=C03E1C0C
<6>[    0.000000]  msr=800045AE, ear=00000000, esr=00000B12, fsr=00000000
<0>[    0.000000] Kernel panic - not syncing: Attempted to kill the idle 
task!
<0>[    0.000000] Rebooting in 120 seconds..


The next thing I'll try will be to use the noMMU toolchain to build a 
noMMU kernel.

>
> Are you referring to the kernel tree (2.6.2) within PetaLinux?  Or the
> distribution generally?
>
> I've not seen anything like this - we build these trees on a daily
> basis.  Something strange about your local setup perhaps?
>
For one example, the compilation error is: “mtd_info_t has no member 
named 'oobblock'”.  These errors all seem to be in the utilities that 
aren't enabled by default -- for example, dmesg, netcat, ssh, and such. 
 Another example: trying to build ethtool results in the system trying 
to run a Microblaze binary on the host, and then saying "can't build 
executables".  When I'm building Petalinux, I move my /usr/local 
(contains the newer toolchain versions) out of the way, to make the 
system use the Petalogix versions.

I also remember having an issue with the build system giving "ROOTDIR: 
command not found"  (in the kernel's .config).  ROOTDIR is supposed to 
be a variable, not a command.

Perhaps I should copy my base config files elsewhere, and re-clone the 
Petalinux tree, to undo the manual changes I had done to try to fix some 
compilation.

>
> I think it's a combination of unlucky and doing something wrong.  We
> build this stuff all the time, as do many others.
>
> Regards,
>
> John

Thanks for the advice.

-- Dana


___________________________
microblaze-uclinux mailing list
microblaze-uclinux@xxxxxxxxxxxxxx
Project Home Page : http://www.itee.uq.edu.au/~jwilliams/mblaze-uclinux
Mailing List Archive : http://www.itee.uq.edu.au/~listarch/microblaze-uclinux/