Syslinux - Jul 2009 - bug or user error - menu.c32 in syslinux-3.62-pre11

Im trying to use newer simple-menu features,
and have found that the following config-file
(closely based on wiki pxelinux example)
-   fails for 3.62-pre11,
-   works for 3.62-pre10

where 'working' is successive reloads such that re-edits of file are
seen on next reload
and broken is issuing this error:

Initial menu has no LABEL entries!
                                  boot:


#> more default

SERIAL  0 115200
CONSOLE 0
TIMEOUT 50
IPAPPEND 0

    default menu.c32
    prompt 0

    menu title PXE Wiki Boot Menu example
    menu INCLUDE pxelinux.cfg/graphics.conf
    MENU AUTOBOOT Starting Local System in # seconds

    label selfload
      menu label ^Main, self-reload
      menu default
      timeout 200
      KERNEL menu.c32
      APPEND pxelinux.cfg/default
      TOTALTIMEOUT 9000

    LABEL Tools Menu
      MENU LABEL ^Tools Men
      # timeout 50
      KERNEL menu.c32
      APPEND pxelinux.cfg/graphics.conf pxelinux.cfg/c32tools.menu

    LABEL Setup Menu
      MENU LABEL ^Setup Menu
      KERNEL menu.c32
      APPEND pxelinux.cfg/graphics.conf pxelinux.cfg/setup.menu

    LABEL Fixes Menu
      MENU LABEL ^Fixes Menu
      KERNEL menu.c32
      APPEND pxelinux.cfg/graphics.conf pxelinux.cfg/fixes.menu




And while Im writing, Ive also had some mixed success with the c32 sample tools:

MENU TITLE c32 Tools

    LABEL   A hdt.c32
	MENU LABEL	^a.  hdt.c32
	KERNEL        hdt.c32
    LABEL   B hello.c32
	MENU LABEL	^b.  hello.c32
	KERNEL        hello.c32
    LABEL   C localboot.c32
	MENU LABEL	^c.  localboot.c32
	KERNEL        localboot.c32
    LABEL   D meminfo.c32
	MENU LABEL	^d.  meminfo.c32
	KERNEL        meminfo.c32
    LABEL   E menu.c32
	MENU LABEL	^e.  menu.c32
	KERNEL        menu.c32
    LABEL   F pcitest.c32
	MENU LABEL	^f.  pcitest.c32
	KERNEL        pcitest.c32
    LABEL   G serialinfo.c32
	MENU LABEL	^g.  serialinfo.c32
	KERNEL        serialinfo.c32
    LABEL   H reboot.c32
	MENU LABEL	^h.  reboot.c32
	KERNEL        reboot.c32


1st, the working ones (output pasted from 3.62-pre1) :
 meminfo & serialinfo:

INT 15h = f000:50fe  DOS RAM: 582K (0x91800)  INT 12h: 582K (0x91800)
INT 15 88: 0xfc00 (64512K)  INT 15 E801: 0x3c00 (15360K) 0x0700 (114688K)
       0 0000000000000000 000000000009fc00 000000000009fc00 1 usable
       1 000000000009fc00 0000000000000400 00000000000a0000 2 reserved
       2 00000000000f0000 0000000000010000 0000000000100000 2 reserved
       3 0000000000100000 0000000007f00000 0000000008000000 1 usable
       4 00000000fff00000 0000000000100000 0000000100000000 2 reserved
boot:

Serial port base:    0x03f8
Serial port divisor:      1 (115200 baud)
Flow control bits:    0x8000
boot:

if you wait ~20 sec, or hit return again, then reload happens automatically.

hello.c32
pcitest.c32 - does nothing, just hangs

hdt.c32 - writes a broken table-looking-thing to screen, then blanks it.
I suspect that the "console 0" may be part of the problem; I need this
cuz my soekris does some console/vga mixing that doubles characters on-screen.


  Hardware Detection Tool 0.2.7 by Erwan Velu

+------ ----------------+
                                                       |  P
       *|#                                                      |
                *|#
  |                      *|#
           |  P                   *|#
                    |  -                   -|#
                             |  S                   *|#
                                      |  S                   *|#
                                               |  -
-|#                                                      |  S
          |#                                                      |
                  *|#
    |* E                    |#
             +-----------------------+#
                       #########################



by starting my picocom session inside a `script` session, I
got a better look at the output


  Hardware Detection
 Tool 0.2.7 by Erwan Velu
CPU: Detecting
DISKS: Detecting
  DISK 0x80: 0x80 : hdt. c32
	MEN: sectors=1936486255, s/t=63 head=16 : EDD=1.1
DMI: Detecting Table
DMI: ERROR ! Table not found !
DMI: Many hardware components will not be detected !
PCI: 10 devices detected
PCI: Resolving names
PCI: Resolving class names
PCI: Resolving module names
PCI: 10 Devices Found
VESA: Detecting
MENU: Processor menu done (14 items)
MENU: Disks menu done (1 items)
MENU: PCI menu done (10 items)
MENU: PXE menu done (8 items)
MENU: Kernel menu done (0 items)
MENU: Summary menu done (5 items)
MENU: Syslinux menu done (5 items)
MENU: VESA menu done (2 items)
MENU: About menu done (6 items)
Starting Menu (18 menus)


I cannot tell how to navigate the menus,
but Im happy to repeat this script-based capture,
and do some blind menu-browsing to stimulate
more output.  I'll look on your hdt website for clues (rummages..)







Id be inclined to refine the Wiki example (esp adding the
c32tools.menu file) once I can ascertain what exactly is wrong with
what Ive done.

Good day Jim,

It's been a while since you indicated that you suspected trouble
somewhere after Syslinux 3.62-pre10 in regards to config-file
reloading...  Did you make any progress there?  Also, could you be
interested in trying the latest Syslinux version out to see what fortune
you have with it?

Thanks,

- Shao Miller

On 07/15/2009 08:19 AM, Jim Cromie wrote:
> Im trying to use newer simple-menu features,
> and have found that the following config-file
> (closely based on wiki pxelinux example)
> -   fails for 3.62-pre11,
> -   works for 3.62-pre10
> 
> where 'working' is successive reloads such that re-edits of file
are
> seen on next reload
> and broken is issuing this error:
> 
> Initial menu has no LABEL entries!
>                                   boot:
Hmm... the config file is a bit sketchy, for one thing timeout and
totaltimeout are global options...

It's pretty impossible to figure out what would be without having all
the configuration files, including the included ones.

The other thing is that a LABEL statement should not have spaces in it.

Either way, looks like the new submenu code is what broke this for you,
which isn't really all that surprising.

	-hpa