platform/upstream/syslinux.git
11 years agomodule: Fix memory leak in spawn_load()
Matt Fleming [Tue, 27 Nov 2012 16:19:17 +0000 (16:19 +0000)]
module: Fix memory leak in spawn_load()

If for some reason we fail to load a module then we need to free the
memory allocated to that module.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agopxe: Don't leak inode on timeout
Matt Fleming [Tue, 27 Nov 2012 16:11:26 +0000 (16:11 +0000)]
pxe: Don't leak inode on timeout

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoscreensize: Dereference pointers when checking cols/rows
Matt Fleming [Tue, 27 Nov 2012 16:03:36 +0000 (16:03 +0000)]
screensize: Dereference pointers when checking cols/rows

Dereference 'rows' and 'cols' to check whether the data they point to
is zero, which would indicate the screen size is bogus, instead of
checking if they point to NULL.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agohdt-cli: Correct malloc() size argument
Matt Fleming [Fri, 16 Nov 2012 14:23:45 +0000 (14:23 +0000)]
hdt-cli: Correct malloc() size argument

We need to be allocating sizeof(char *) (4) not sizeof(char) (1) for
'new_argv'.

Cc: Erwan Velu <erwanaliasr1@gmail.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agochainboot: Delete extraneous free() in chainboot_file()
Matt Fleming [Thu, 15 Nov 2012 21:32:18 +0000 (21:32 +0000)]
chainboot: Delete extraneous free() in chainboot_file()

We don't need to call free(buf) if we're jumping to the 'bail' label
because 'buf' is always free'd there.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agovesa: Fix double close() bug in vesacon_load_background()
Matt Fleming [Thu, 15 Nov 2012 21:27:59 +0000 (21:27 +0000)]
vesa: Fix double close() bug in vesacon_load_background()

We always call fclose() on 'fp' if fopen() was successful, so delete
the extraneous fclose() call in read_jpeg_file().

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoasprintf: Ensure we always call va_end(ap)
Matt Fleming [Thu, 15 Nov 2012 20:56:09 +0000 (20:56 +0000)]
asprintf: Ensure we always call va_end(ap)

There's currently the potential for us to exit early from asprintf()
without calling va_end(ap). Rearrange things so that we always make
the call.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agobufprintf: Add va_end() for our va_copy()
Matt Fleming [Thu, 15 Nov 2012 20:39:30 +0000 (20:39 +0000)]
bufprintf: Add va_end() for our va_copy()

According to the stdarg(3) man page each invocation of va_copy()
should be paired with an invocation of va_end().

Cc: Erwan Velu <erwanaliasr1@gmail.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoldlinux: Delete dead code and useless printf syslinux-5.00-pre10
Matt Fleming [Tue, 13 Nov 2012 16:07:18 +0000 (16:07 +0000)]
ldlinux: Delete dead code and useless printf

Seeing the cmdline re-printed every time the user presses enter is
very annoying, so delete the offending printf(). Also delete some code
that has been commented out for ages.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoldlinux: Delete inaccurate 'insufficient memory' message
Matt Fleming [Tue, 13 Nov 2012 15:57:01 +0000 (15:57 +0000)]
ldlinux: Delete inaccurate 'insufficient memory' message

Instead, use strerror() to print a more useful message if we fail to
load a kernel.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agomk: Add Makefile variable for python
Matt Fleming [Mon, 12 Nov 2012 15:38:33 +0000 (15:38 +0000)]
mk: Add Makefile variable for python

Instead of hard-coding the python executable name use a make variable
like we do for all other executables used during the build. This has
the added bonus of allowing $(PYTHON) to be overridden on the command
line, e.g.

   make PYTHON=python2

which is particularly useful for distributions where /usr/bin/python
is actually python3.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agomenu: Inherit parent menu title
Matt Fleming [Thu, 25 Oct 2012 10:35:14 +0000 (11:35 +0100)]
menu: Inherit parent menu title

The documentation for the menu system clearly states that all
properties of the parent are inherited in the submenu. Unfortunately
there's a bug that means this isn't true for the parent's menu title,
so fix that.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoMerge branch 'strerror-for-mfleming' of git://github.com/geneC/syslinux into elflink
Matt Fleming [Wed, 7 Nov 2012 07:31:13 +0000 (07:31 +0000)]
Merge branch 'strerror-for-mfleming' of git://github.com/geneC/syslinux into elflink

Pull klibc strerror() patches from Gene Cumm,

  Update strerror() to give english messages

* 'strerror-for-mfleming' of git://github.com/geneC/syslinux:
  com32 strerror(): Remove macro WITH_ERRLIST use
  com32 strerror: add errlist.o to Make
  strerror: Use klibc version

11 years agocom32 strerror(): Remove macro WITH_ERRLIST use
Gene Cumm [Sat, 16 Jun 2012 18:26:11 +0000 (14:26 -0400)]
com32 strerror(): Remove macro WITH_ERRLIST use

Signed-off-by: Gene Cumm <gene.cumm@gmail.com>
11 years agocom32 strerror: add errlist.o to Make
Gene Cumm [Wed, 7 Nov 2012 04:32:51 +0000 (23:32 -0500)]
com32 strerror: add errlist.o to Make

Exclude in .gitignore

Signed-off-by: Gene Cumm <gene.cumm@gmail.com>
11 years agostrerror: Use klibc version
Gene Cumm [Sat, 16 Jun 2012 17:33:19 +0000 (13:33 -0400)]
strerror: Use klibc version

Remove old in favor of current klibc version

Signed-off-by: Gene Cumm <gene.cumm@gmail.com>
11 years agoMerge branch 'hexdump' of git://git.zytor.com/users/sha0/syslinux into elflink
Matt Fleming [Tue, 6 Nov 2012 19:56:01 +0000 (19:56 +0000)]
Merge branch 'hexdump' of git://git.zytor.com/users/sha0/syslinux into elflink

Pull new hexdump.c32 module from Shao Miller,

* 'hexdump' of git://git.zytor.com/users/sha0/syslinux:
  hexdump.c32: Simple file hex-dumper

11 years agoMerge branch 'multi_initrd2-5.00-pre9' of git://git.zytor.com/users/sha0/syslinux...
Matt Fleming [Mon, 5 Nov 2012 15:35:35 +0000 (15:35 +0000)]
Merge branch 'multi_initrd2-5.00-pre9' of git://git.zytor.com/users/sha0/syslinux into elflink

Pull multi-initrd patches from Shao Miller,

  These patches provide two new options for the linux.c32 module.

  The "initrd+=" option allows you to append initramfs-style blobs
  (files which can be produced with 'cpio -o -H newc') to whatever
  "initrd" was specified via the "initrd=" option (or was specified
  indirectly via the INITRD directive).

  The "initrdfile=" option allows you to load a file and encapsulate it
  as though you had used 'cpio -o -H newc', and pass the resulting blob
  alongside the other items that might have been specified with
  "initrd=" or "initrd+=".

* 'multi_initrd2-5.00-pre9' of git://git.zytor.com/users/sha0/syslinux:
  initramfs chain handling: Accounting fixes for padding, etc.
  linux.c32: Introduce initrdfile= option
  linux.c32: Add new initrd+= option for multiple initrds
  linux.c32: Move some initrd=x,y,z code out of main
  linux.c32: Add find_arguments function

11 years agocore: Fix realloc() code transcription error
Shao Miller [Mon, 5 Nov 2012 02:23:09 +0000 (21:23 -0500)]
core: Fix realloc() code transcription error

Commit 79459f631546eea83d4158f535c20ebd4ac18987 copied portions of
com32/lib/realloc.c into core/mem/malloc.c, with minor changes
for accessing allocation arena header attributes.  The previous
code used structure members and the current code uses bitmask
macros.

On a particular line in the original realloc(), there were two
assignments in a single expression that included a compound
assignment.  This fact was missed when the code was copied and
modified to use the bitmask macros.

Signed-off-by: Shao Miller <sha0.miller@gmail.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoextlinux: Avoid dereferencing a garbage pointer
Matt Fleming [Fri, 2 Nov 2012 17:02:36 +0000 (17:02 +0000)]
extlinux: Avoid dereferencing a garbage pointer

If opt.reset_adv is set the call to ext_read_adv() is skipped which
would have initialised 'filename'. This means that a pointer
containing random data from the stack is passed to ext_write_adv().

Just delete the opt.reset_adv logic since modify_adv() handles that
case anyway.

Reported-by: Frediano Ziglio <frediano.ziglio@citrix.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agohexdump.c32: Simple file hex-dumper
Shao Miller [Sun, 4 Nov 2012 04:26:32 +0000 (00:26 -0400)]
hexdump.c32: Simple file hex-dumper

Usage: %s [<option> [...]] <filename> [<option> [...]]

Options: -p
         --page . . . . . . . Pause output every 24 lines
         --no-buffer . . . .  Load the entire file before dumping
         --extended-ascii . . Use extended ASCII chars in dump
         -?
         -h
         --help  . . . . . .  Display this help

Signed-off-by: Shao Miller <sha0.miller@gmail.com>
11 years agoinitramfs chain handling: Accounting fixes for padding, etc.
Shao Miller [Fri, 26 Oct 2012 06:14:28 +0000 (02:14 -0400)]
initramfs chain handling: Accounting fixes for padding, etc.

Signed-off-by: Shao Miller <sha0.miller@gmail.com>
11 years agolinux.c32: Introduce initrdfile= option
Shao Miller [Fri, 26 Oct 2012 01:25:38 +0000 (21:25 -0400)]
linux.c32: Introduce initrdfile= option

It is useful to be able to load a file and pass it into a kernel's
rootfs via the initramfs scheme.  Given "initrdfile=foo", we will
load the file foo, encapsulate it with the initramfs cpio format,
then pass it alongside any initramfs files that were specified by
"initrd=" and "initrd+=" options.

One can specify the desired path/filename for the file to have
within the rootfs by using the at (@) sign, as in:

  initrdfile=foo@/goes/to/foo

One can also specify multiple files, separated by commas, such as:

  initrdfile=foo,bar@/somewhere/bar,baz

One can also use this option multiple times, as in:

  initrdfile=foo,bar initrdfile=baz@/somewhere/baz

Signed-off-by: Shao Miller <sha0.miller@gmail.com>
11 years agolinux.c32: Add new initrd+= option for multiple initrds
Shao Miller [Thu, 25 Oct 2012 07:58:07 +0000 (03:58 -0400)]
linux.c32: Add new initrd+= option for multiple initrds

linux.c32 now processes the kernel's command-line for "initrd+="
options.

Given "initrd+=foo", linux.c32 will concatenate the file "foo"
to the initrds that it has already loaded due to the "initrd="
option.

Given "initrd+=foo,bar", linux.c32 will concatenate both files
"foo" and "bar" to the initrds that it has already loaded due
to the "initrd=" option.  That is, multiple filenames can be
specified with comma separators.

Given "initrd+=foo initrd+=bar", linux.c32 will concatenate
both files "foo" and "bar" to the initrds that it has already
loaded due to the "initrd=" option.  That is, the "initrd+="
option can be specified multiple times.

The position of any "initrd=" option relative to any "initrd+="
option is irrelevant.  The "initrd=" option is always processed
before all "initrd+=" options.

PLEASE NOTE: It is important to note that there are NO SPACES
involved in using the "initrd+=" option.  "initrd += foo" will
not work.

Signed-off-by: Shao Miller <sha0.miller@gmail.com>
11 years agolinux.c32: Move some initrd=x,y,z code out of main
Shao Miller [Thu, 25 Oct 2012 06:17:36 +0000 (02:17 -0400)]
linux.c32: Move some initrd=x,y,z code out of main

In handling an "initrd=x,y,z" option, it seems reasonable to be able
to handle similar options in a similar fashion, so some of the code
has been moved out of 'main' and into a new 'process_initramfs_args'
function.

Signed-off-by: Shao Miller <sha0.miller@gmail.com>
11 years agolinux.c32: Add find_arguments function
Matt Fleming [Sat, 3 Nov 2012 01:52:52 +0000 (21:52 -0400)]
linux.c32: Add find_arguments function

The 'find_argument' function already finds the last instance of a
command-line option.  For symmetry, we introduce a 'find_arguments'
function which will help to iterate each instance of a command-line
option.

Also, this commit uses 'strncmp' in both, instead of 'memcmp'.

Modified-by: Shao Miller <sha0.miller@gmail.com>
Signed-off-by: Shao Miller <sha0.miller@gmail.com>
11 years agoCLI: Add Ctrl + V support for printing the Syslinux version
Matt Fleming [Thu, 1 Nov 2012 12:26:22 +0000 (12:26 +0000)]
CLI: Add Ctrl + V support for printing the Syslinux version

The cli code is missing support for printing the version string that
is available in Syslinux 4.06 and earlier.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agowin: Print error message if we fail to install to --directory
Matt Fleming [Thu, 1 Nov 2012 11:30:42 +0000 (11:30 +0000)]
win: Print error message if we fail to install to --directory

Instead of silently returning with no indication of error if we
couldn't install to the --directory argument, print an error message
that tells the user the files are not where they wanted.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agowin: Fix installing to a directory
Matt Fleming [Thu, 1 Nov 2012 11:12:03 +0000 (11:12 +0000)]
win: Fix installing to a directory

commit bda54cb68067 ("installers: Install ldlinux.c32 automatically)"
introduced a bug in the move_file() function. move_file() isn't
deleting the destination path, including the --directory argument,
that ldlinux.sys should be installed to ('new_name'), which means that
the MoveFile() calls fails. What it's actually doing is deleting
ldlinux.sys from the file destination system.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoCLI: Fix command history traversal
Matt Fleming [Thu, 1 Nov 2012 10:56:06 +0000 (10:56 +0000)]
CLI: Fix command history traversal

The up/down keys were broken with respect to traversing through the
command history because it was dereferencing a NULL pointer on the
first iteration.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agocom32: Include .init_array section in .ctors in linker script
Matt Fleming [Thu, 1 Nov 2012 09:50:46 +0000 (09:50 +0000)]
com32: Include .init_array section in .ctors in linker script

GCC 4.7 now places pointers to functions with the 'constructor' and
'destructor' function attributes in .init_array and .fini_array
sections, respectively, whereas previously they were in the .ctors and
.dtors sections. This change breaks the ctors/dtors code as it only
expects function to be in the .ctors and .dtors sections, meaning the
ctors and dtors functions are never executed.

While a COM32_INIT() macro exists that places functions in the
.init_array section, no function makes use of it, so there should be
no fallout from this change.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoldlinux: get_key() requires raw access to user input
Matt Fleming [Wed, 31 Oct 2012 15:55:02 +0000 (15:55 +0000)]
ldlinux: get_key() requires raw access to user input

commit 4503e6260c0 ("ldlinux: Use stdcon instead of rawcon for the
console") broke get_key() because it was no longer possible to read
raw data from stdin. Provide a new function raw_read() that will
return user input a character at a time.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoldlinux: Fix logic if no DEFAULT or UI directive is found
Matt Fleming [Wed, 31 Oct 2012 12:42:25 +0000 (12:42 +0000)]
ldlinux: Fix logic if no DEFAULT or UI directive is found

Somewhere along the way the code that prints,

    No DEFAULT or UI configuration directive found!

was broken, and now no longer prints at all. While we're fixing this
it's a good opportunity to rework the logic to be clearer. Now we only
print the message if a config file was found, since these directives
are obviously missing if there is no config file (a warning will be
printed about the lack of config file anyway).

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoldlinux: Print a warning if no config file is found
Matt Fleming [Wed, 31 Oct 2012 12:49:43 +0000 (12:49 +0000)]
ldlinux: Print a warning if no config file is found

This behaviour is taken from 4.06, where an error message is printed
if no config file is found. A warning is a more user-friendly choice
since Syslinux will still function without a config file.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agopxe: Don't call open_config() from the pxe core
Matt Fleming [Wed, 31 Oct 2012 14:17:49 +0000 (14:17 +0000)]
pxe: Don't call open_config() from the pxe core

Delete some stale comments about ldlinux.c32 requiring ConfigName to
be initialised before it's launched - this is no longer
true. ldlinux.c32 takes care of opening the config files.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agodoc: Document the 'PATH' directive syslinux-5.00-pre9
Matt Fleming [Wed, 31 Oct 2012 11:08:03 +0000 (11:08 +0000)]
doc: Document the 'PATH' directive

The 'PATH' directive allows the user to specify a list of directories
to search when loading modules.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoNEWS: Document the major changes in 5.00
Matt Fleming [Wed, 31 Oct 2012 10:32:42 +0000 (10:32 +0000)]
NEWS: Document the major changes in 5.00

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoMerge branch 'master' into elflink
Matt Fleming [Wed, 24 Oct 2012 13:41:00 +0000 (14:41 +0100)]
Merge branch 'master' into elflink

The following conflict is a result of the fact comapi_chainboot was
only stubbed out on the 'elflink' branch but was properly marked as
obsolete in 'master'.

Conflicts:
core/comboot.inc

11 years agodiag/mbr: Fix typo in README
Matt Fleming [Tue, 23 Oct 2012 15:49:23 +0000 (16:49 +0100)]
diag/mbr: Fix typo in README

Reported-by: Ady <ady-sf@hotmail.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoMerge branch 'elflink' of git://github.com/aericson/syslinux into elflink
Matt Fleming [Tue, 23 Oct 2012 14:02:50 +0000 (15:02 +0100)]
Merge branch 'elflink' of git://github.com/aericson/syslinux into elflink

Conflicts:
com32/modules/Makefile

11 years agomemdisk: Fix order of sectors/track and bytes/sector syslinux-4.06
Kenneth J. Davis [Mon, 22 Oct 2012 18:20:18 +0000 (19:20 +0100)]
memdisk: Fix order of sectors/track and bytes/sector

Swap order of sectors/track with bytes/sector to match diskette
parameter table order (as pointed to by INT 1Eh).

References: (additional ones available if necessary)
'System BIOS for IBM PC/XT/AT computers and compatibles' by Phoenix
Technologies Ltd. copyright 1987,88,89 3rd printing 1990 ISBN:
0-201-51806-6 pages 61-62
'PC Interrupts' 2nd edition by Ralf Brown & Jim Kyle copyright 1994
1st printing 1993 ISBN: 0-201-62485-0 page 211

Note: this fixes an issue with older FreeDOS kernels that expect this
structure to contain valid values.

Signed-off-by: Kenneth J. Davis <jeremyd@fdos.org>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoNEWS: Document the VFAT relative path fix
Matt Fleming [Mon, 22 Oct 2012 12:39:24 +0000 (13:39 +0100)]
NEWS: Document the VFAT relative path fix

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoREADME: Clarify the requirements for utils/isohybrid
Matt Fleming [Mon, 22 Oct 2012 12:02:19 +0000 (13:02 +0100)]
README: Clarify the requirements for utils/isohybrid

Be clear that utils/isohybrid requires the UUID library as well as
/usr/include/uuid/uuid.h.

Reported-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoNEWS: Fix some typos
Matt Fleming [Wed, 17 Oct 2012 12:07:07 +0000 (13:07 +0100)]
NEWS: Fix some typos

Reported-by: Ady <ady-sf@hotmail.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agodoc: minor documentation updates
Vanush "Misha" Paturyan [Wed, 17 Oct 2012 08:54:16 +0000 (09:54 +0100)]
doc: minor documentation updates

syslinux.txt: syslinux binary is in "linux" directory
menu.txt:     menu.c32 and vesamenu.c32 are in com32/menu directory

11 years agoRevert "win: Add -mno-ms-bitfields"
H. Peter Anvin [Mon, 22 Oct 2012 20:44:53 +0000 (13:44 -0700)]
Revert "win: Add -mno-ms-bitfields"

This reverts commit 7d9ee65cc9577eb69782a49a34895ae80820e9ff.

I am concerned that this may cause the Windows ABI header files to be
misinterpreted.

Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
11 years agoPacked definition: workaround for mingw + gcc 4.7
H. Peter Anvin [Mon, 22 Oct 2012 20:42:31 +0000 (13:42 -0700)]
Packed definition: workaround for mingw + gcc 4.7

__attribute__((packed)) does the wrong thing when in ms-struct
mode; this breaks compiling on mingw-w64.  Therefore force gcc struct
mode when we use packed.

This obsoletes checkin

  7d9ee65 win: Add -mno-ms-bitfields

Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
11 years agoldlinux: Use stdcon instead of rawcon for the console
Matt Fleming [Fri, 19 Oct 2012 13:41:12 +0000 (14:41 +0100)]
ldlinux: Use stdcon instead of rawcon for the console

Switch to the stdcon input device so things like fgets() work
correctly when reading from stdin.

I noticed this was broken when running dmitest.c32. fgets() was
returning NULL because there was no pending user input instead of
waiting for/buffering the input. Modules are expecting stdin to be a
line-oriented device, not a raw one where each module has to implement
its own line support.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoClean up $(GPLLIB) leak
Matt Fleming [Fri, 19 Oct 2012 09:55:13 +0000 (10:55 +0100)]
Clean up $(GPLLIB) leak

Modules were linking against com32/gpllib/libcom32gpl.c32 even when
NOGPL was set, so remove it from various modules Makefiles and use
$(C_LIBS) in mk/elf.mk which already has the conditional-magic for
$(GPLLIB).

Also, the object files in com32/gpllib/ are not distributed in the
release tarballs, which means that if a user does a 'make netinstall'
they hit the following install error,

install: cannot stat `com32/gpllib/*.c32': No such file or directory

Add a $(INSTALLABLE_MODULES) variable to separate the list of build
and install modules.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agocore: Print error message if we don't load ldlinux.c32
Matt Fleming [Fri, 19 Oct 2012 09:52:44 +0000 (10:52 +0100)]
core: Print error message if we don't load ldlinux.c32

If the user forgot to install ldlinux.c32 they're greeted with the
following unhelpful message,

  Boot failed: press a key to retry, or wait for reset...

We should explicitly tell the user that the reason for the boot
failing is that ldlinux.c32 could not be loaded.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agocomboot: remove comapi_chainboot syslinux-4.06-pre14
Sebastian Herbszt [Sun, 14 Oct 2012 13:38:58 +0000 (15:38 +0200)]
comboot: remove comapi_chainboot

comapi_chainboot was obsoleted in 3.80.

Signed-off-by: Sebastian Herbszt <herbszt@gmx.de>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agocomboot: fix documentation
Sebastian Herbszt [Sun, 14 Oct 2012 13:38:25 +0000 (15:38 +0200)]
comboot: fix documentation

comapi_shuffle was obsoleted in 3.80.

Signed-off-by: Sebastian Herbszt <herbszt@gmx.de>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoREADME: document the need for /usr/include/uuid/uuid.h
Geert Stappers [Fri, 12 Oct 2012 19:36:50 +0000 (21:36 +0200)]
README: document the need for /usr/include/uuid/uuid.h

utils/isohybrid build depends on /usr/include/uuid/uuid.h The package
name varies between distribution families.

11 years agomenu: Disallow navigation to a disabled entry via *any* key
Matt Fleming [Fri, 12 Oct 2012 11:59:32 +0000 (12:59 +0100)]
menu: Disallow navigation to a disabled entry via *any* key

Generalise the fix from commit c823574f53c1 ("menu: Don't highlight
disabled entries") as it only handled the case where we navigated to a
disabled last entry by pressing Ctrl + N or the DOWN arrow
key. Obviously, we can navigate with other keys such as END, PGDN, etc
so we need to handle all cases.

Reported-by: Ady <ady-sf@hotmail.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoprdhcp.c32: Turn off DEBUG syslinux-4.06-pre13
Matt Fleming [Thu, 11 Oct 2012 20:18:10 +0000 (21:18 +0100)]
prdhcp.c32: Turn off DEBUG

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoMerge branch 'pxechn-fix-news-for-hpa-2' of git://github.com/geneC/syslinux
Matt Fleming [Thu, 11 Oct 2012 20:16:37 +0000 (21:16 +0100)]
Merge branch 'pxechn-fix-news-for-hpa-2' of git://github.com/geneC/syslinux

Conflicts:
NEWS
com32/modules/Makefile

11 years agowin: Add -mno-ms-bitfields
Shao Miller [Thu, 11 Oct 2012 15:59:37 +0000 (11:59 -0400)]
win: Add -mno-ms-bitfields

Recent MinGW GCCs now have -mms-bitfields on by default, but it
appears to break our intentions with __attribute__((packed)).
See GCC bug 52991,

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52991

The undesired behaviour was noticed when the win32 installer
had troubles identifying the media type field of a boot sector.

Signed-off-by: Shao Miller <sha0.miller@gmail.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoisohybrid: fix isohybrid.c compile
Frediano Ziglio [Fri, 31 Aug 2012 11:00:16 +0000 (12:00 +0100)]
isohybrid: fix isohybrid.c compile

Make isohybrid.c compile on gcc 4.4 32 bit instead of producing the
following error,

isohybrid.c: In function â€˜lendian_64’:
isohybrid.c:437: error: integer constant is too large for â€˜long’ type

Cc: Matthew Garrett <mjg@redhat.com>
Cc: Michal Soltys <soltys@ziu.info>
Signed-off-by: Frediano Ziglio <frediano.ziglio@citrix.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoext2: add additional check to ext2 loader
Frediano Ziglio [Thu, 2 Aug 2012 08:32:36 +0000 (09:32 +0100)]
ext2: add additional check to ext2 loader

Check if a pointer is NULL due to specifying an invalid inode group
number.

Signed-off-by: Frediano Ziglio <frediano.ziglio@citrix.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agomboot: fix typo in comment
Frediano Ziglio [Thu, 2 Aug 2012 08:31:23 +0000 (09:31 +0100)]
mboot: fix typo in comment

Signed-off-by: Frediano Ziglio <frediano.ziglio@citrix.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agomake some functions static remove unused NADV define
Frediano Ziglio [Thu, 2 Aug 2012 08:30:04 +0000 (09:30 +0100)]
make some functions static remove unused NADV define

Just some minor cleanup

Signed-off-by: Frediano Ziglio <frediano.ziglio@citrix.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agomenu: Add MENU GOTO support to the MENU DEFAULT parsing code
Matt Fleming [Tue, 9 Oct 2012 12:45:00 +0000 (13:45 +0100)]
menu: Add MENU GOTO support to the MENU DEFAULT parsing code

Tomas reported that the 'test1' submenu in the following config file
snippet doesn't honour the MENU DEFAULT directive,

    MENU BEGIN test0
       MENU START
       LABEL -
       MENU DEFAULT
       MENU LABEL OptionA
       MENU GOTO test1
    MENU END

    MENU BEGIN test1
       LABEL -
       MENU LABEL OptionA1
       MENU GOTO test0
       LABEL -
       MENU DEFAULT
       MENU LABEL OptionB1
       MENU GOTO test0
    MENU END

The reason is that the menu config code doesn't know how to handle a
default item that is a MENU GOTO directive. A trivial addition to the
code in record() fixes this.

Reported-by: Tomas M <tomas@slax.org>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agomenu: Don't highlight disabled entries
Matt Fleming [Tue, 9 Oct 2012 12:33:51 +0000 (13:33 +0100)]
menu: Don't highlight disabled entries

There's an off-by-one bug in the menu input handling code which makes
it possible to highlight/select the last entry in a menu even if that
entry is disabled.

It should be noted that you can't actually *do* anything with the
disabled entry, i.e. edit or execute it. Still, allowing it to be
highlighted does look a little strange.

Reported-by: Matt Sephton <matt.sephton@gmail.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agocom32: Add ifmemdsk.c32 to Makefile
Matt Fleming [Mon, 8 Oct 2012 14:45:46 +0000 (15:45 +0100)]
com32: Add ifmemdsk.c32 to Makefile

... so that it actually gets built. This should have been added to the
Makefile when ifmemdsk.c was created.

Reported-by: Ady <ady-sf@hotmail.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoNEWS: Document changes between 4.05 and 4.06 syslinux-4.06-pre12
Matt Fleming [Fri, 13 Jul 2012 14:01:59 +0000 (15:01 +0100)]
NEWS: Document changes between 4.05 and 4.06

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoIn the current Lua.c32 DMI implementation, it is a flat table with dotted names in...
Hung-chi Lihn [Sat, 22 Sep 2012 16:25:49 +0000 (18:25 +0200)]
In the current Lua.c32 DMI implementation, it is a flat table with dotted names in table entries. It also misses a number DMI sub-tables.

This patch, cleans up the DMI table structure using Lua's nested table structure and adds all missing DMI sub-tables. If a DMI sub-table is not supported by the hardware (not filled), then the corresponding sub-table will not be generated. This helps to make the table structure cleaner and reflects the actual DMI information.

Signed-off-by: Hung-chi Lihn <hlihn@google.com>
Signed-off-by: Erwan Velu <erwanaliasr1@gmail.com>
11 years agoBug fix: Added the missing entry (boot_from_pcmcia) in struct s_characteristics.
Hung-chi Lihn [Sat, 22 Sep 2012 16:24:25 +0000 (18:24 +0200)]
Bug fix: Added the missing entry (boot_from_pcmcia) in struct s_characteristics.

In bios_charac_strings[] of /com32/gpllib/dmi/dmi_bios.c there are actually 29 entries. However, BIOS_CHAR_NB_ELEMENTS in /com32/gplinclude/dmi/dmi_bios.h was defined as 28 and struct s_characteristics in the same file also only had 28 entries. It turns out that there was a missing entry "boot_from_pcmcia" (the 16th entry representing bit 18) in s_characteristics.

Signed-off-by: Hung-chi Lihn <hlihn@google.com>
Signed-off-by: Erwan Velu <erwanaliasr1@gmail.com>
11 years agoAdded the native syslinux functions config_file(), ipappend_strs(), and reboot()...
Hung-chi Lihn [Sat, 22 Sep 2012 16:23:38 +0000 (18:23 +0200)]
Added the native syslinux functions config_file(), ipappend_strs(), and reboot() to Lua.c32. This allows the Lua script to query the config file name and the ipappend strings (pxelinux only), as well as to perform reboot (warm and cold) to the system.

In Lua.c32, the extension will be used as the following:
  1. syslinux.config_file() will return the config file string.
  2. syslinux.ipappend_strs() will return a table of IPAPPEND strings with numerical indices.
  3. syslinux.reboot() will perform cold reboot, while syslinux.reboot(1) will perform warm reboot.

Signed-off-by: Hung-chi Lihn <hlihn@google.com>
Signed-off-by: Erwan Velu <erwanaliasr1@gmail.com>
11 years agoThe current Lua.c32 does not enable io.read() due to some missing library functions...
Hung-chi Lihn [Sat, 22 Sep 2012 16:22:55 +0000 (18:22 +0200)]
The current Lua.c32 does not enable io.read() due to some missing library functions. However, this strongly limits the Lua script from getting user inputs and reading files (even in pxelinux via TFTP).

This patch enables io.read() in Lua.c32 with some restrictions:
  1. the io.read("*line") is fully supported.
  2. the io.read("*number") is not supported due to the missing buffering in underlying file I/Os. However, the user can read a line using io.read() and convert the string to numbers using the built in pattern matching and number conversion features.
  3. io.read(bytes) is supported. However, io.read(0) will not be a valid test for EOF due to the missing I/O buffering. io.read() will return nil if EOF is encountered. This offers an alternative way to handle EOF.

Signed-off-by: Hung-chi Lihn <hlihn@google.com>
Signed-off-by: Erwan Velu <erwanaliasr1@gmail.com>
11 years agoinstallers: Install ldlinux.c32 automatically syslinux-5.00-pre8
Matt Fleming [Fri, 14 Sep 2012 14:22:29 +0000 (15:22 +0100)]
installers: Install ldlinux.c32 automatically

Because ldlinux.c32 is required for Syslinux to function correctly, we
should be installing it automatically much like ldlinux.sys.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agodoc/pxechn.txt: update
Gene Cumm [Sun, 16 Sep 2012 15:19:41 +0000 (11:19 -0400)]
doc/pxechn.txt: update

Signed-off-by: Gene Cumm <gene.cumm@gmail.com>
11 years agoextlinux: Handle error case for find_mount()
Matt Fleming [Fri, 14 Sep 2012 13:54:40 +0000 (14:54 +0100)]
extlinux: Handle error case for find_mount()

find_mount() may return NULL, so we'd really better check for that
*before* dereferencing the pointer.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoNEWS: Adding 'silent' option from HDT
Erwan Velu [Mon, 10 Sep 2012 19:07:14 +0000 (21:07 +0200)]
NEWS: Adding 'silent' option from HDT

11 years agohdt: Release 0.5.2
Erwan Velu [Mon, 10 Sep 2012 19:03:50 +0000 (21:03 +0200)]
hdt: Release 0.5.2

New release with codename Manon

11 years agohdt: Adding Silent mode
Erwan Velu [Mon, 10 Sep 2012 19:02:02 +0000 (21:02 +0200)]
hdt: Adding Silent mode

By using the "silent" option at the cmdline, HDT tries not displaying
any boring message.

This is particulary interesting with the "display" option.

11 years agohdt: Removing remaining printf()
Erwan Velu [Mon, 10 Sep 2012 18:39:37 +0000 (20:39 +0200)]
hdt: Removing remaining printf()

Using more_printf allow to manage more easily several pages printing.

11 years agoNEWS: Updating for HDT & kontron_wdt
Erwan Velu [Mon, 10 Sep 2012 18:26:59 +0000 (20:26 +0200)]
NEWS: Updating for HDT & kontron_wdt

11 years agokontron_wdt: Fixing makefile
Erwan Velu [Mon, 10 Sep 2012 18:21:33 +0000 (20:21 +0200)]
kontron_wdt: Fixing makefile

Typo.

11 years agokontron_wdt: Enforce timeout value if already set
Erwan Velu [Mon, 10 Sep 2012 15:38:16 +0000 (17:38 +0200)]
kontron_wdt: Enforce timeout value if already set

If the watchdog was already engaged, let's rewrite the timeout to insure
the proper value is set.

11 years agokontron_wdt: Adjusting exit path
Erwan Velu [Mon, 10 Sep 2012 15:35:13 +0000 (17:35 +0200)]
kontron_wdt: Adjusting exit path

If we fail, at least let's boot the expected image

11 years agokontron_wdt: Adding watchdog com32 module
Erwan Velu [Mon, 10 Sep 2012 15:05:03 +0000 (17:05 +0200)]
kontron_wdt: Adding watchdog com32 module

When using a Kontron ETX board, it's possible to initialize and start
the watchdog during syslinux booting.

This allow protecting a boot sequence with a defined timeout.

Bootloader is starting, engage the watchdog and then start a default
entry (typically a Linux image).

If the loaded OS, feed or reinitalize the watchdog, nothing occurs
unless the system will reboot

Conflicts:

com32/modules/Makefile

11 years agodoc/pxechn.txt: Fix old example
Gene Cumm [Thu, 6 Sep 2012 22:27:30 +0000 (18:27 -0400)]
doc/pxechn.txt: Fix old example

Command line options changed.

Signed-off-by: Gene Cumm <gene.cumm@gmail.com>
11 years agoMerge remote-tracking branch 'hpa/master'
Erwan Velu [Wed, 5 Sep 2012 19:08:39 +0000 (21:08 +0200)]
Merge remote-tracking branch 'hpa/master'

11 years agomodule: Fixed the upper limit in symbol table walk through
Chandramouli Narayanan [Wed, 5 Sep 2012 06:49:49 +0000 (07:49 +0100)]
module: Fixed the upper limit in symbol table walk through

Number of entries in symbol table should be sized by the table entry
size. Without this, the loading of modules takes forever due to
walking all over the memory.

Signed-off-by: Chandramouli Narayanan <chandramouli.narayanan@intel.com>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoconsole: Close stdin, stdout, stderr on ldlinux.c32 unload
Matt Fleming [Mon, 3 Sep 2012 13:35:14 +0000 (14:35 +0100)]
console: Close stdin, stdout, stderr on ldlinux.c32 unload

The location of __file_info[] and ansicon_counter are split between
the core and ldlinux.c32. We can end up calling __ansicon_close (from
fp->oop->close) in the process of reloading ldlinux.c32, and while
ansicon_counter is set to its initialised data value of 0 when
reloading ldlinux.c32, the core is never reloaded, and so
__file_info[] retains its pointers.

What this amounts to is that when ldlinux.c32 tries to call
openconsole() after being reloaded, the core tries to close the
ansicon device (the core still has pointers to __ansicon_close)
despite the ansicon code having no knowledge of ever being opened
(because ansicon_counter was reinitialised).

The most logical way to deal with this is to close stdin, stdout and
stderr when unloading ldlinux.c32.

It also turns out that sys/openconsole.o was present in two Makefile
variables which lead to duplication of code, data, etc. We only need
it present in libcom32min.a and not in the core since the core never
handles opening consoles, at least not now that we've deleted
printf_init() which is safe because we always open a console in
ldlinux.c32 anyway.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agodoc/pxechn.txt: word wrap
Gene Cumm [Mon, 3 Sep 2012 18:13:11 +0000 (14:13 -0400)]
doc/pxechn.txt: word wrap

Signed-off-by: Gene Cumm <gene.cumm@gmail.com>
11 years agodoc/pxechn.txt: Updates for clarity/flow
Gene Cumm [Mon, 3 Sep 2012 17:48:40 +0000 (13:48 -0400)]
doc/pxechn.txt: Updates for clarity/flow

Signed-off-by: Gene Cumm <gene.cumm@gmail.com>
11 years agodoc/pxechn.txt: Change title for manpage use.
Gene Cumm [Mon, 3 Sep 2012 17:47:48 +0000 (13:47 -0400)]
doc/pxechn.txt: Change title for manpage use.

Signed-off-by: Gene Cumm <gene.cumm@gmail.com>
11 years agoldlinux: Use findpath() to lookup filenames syslinux-5.00-pre7
Matt Fleming [Tue, 7 Aug 2012 09:40:12 +0000 (10:40 +0100)]
ldlinux: Use findpath() to lookup filenames

We should use the same method of opening files as the module code when
searching for extensions. In particular, we should search all of PATH.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agorawcon_read: Fix reading high part of input keys
Matt Fleming [Thu, 2 Aug 2012 09:34:17 +0000 (10:34 +0100)]
rawcon_read: Fix reading high part of input keys

commit 8486142cf304 ("elflink: Replace __intcall() with direct
function calls") made the mistake of zero'ing the high part of the
input key on every invocation of __rawcon_read() instead of preserving
it across calls like the old code, which lead to function keys such as
KEY_UP, KEY_DOWN, etc not working.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
11 years agoldlinux: Return to command prompt after loading COM32
Matt Fleming [Thu, 2 Aug 2012 08:26:54 +0000 (09:26 +0100)]
ldlinux: Return to command prompt after loading COM32

The old COM32 loading code would drop the user at a command prompt
once execution returned from the COM32 amodule. We need to replicate
this because most callers of execute() don't expect it to return.

This bug was noticed when loading a COM32 module from
vesamenu.c32. Once execution returned from the COM32 module the
display became garbled because no code exists to reinitialise the
screen for VGA.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
12 years agocom32: Use --as-needed for LDFLAGS
Matt Fleming [Thu, 12 Jul 2012 19:35:14 +0000 (20:35 +0100)]
com32: Use --as-needed for LDFLAGS

This extends commit 4678fd92a21e ("elflink: fix dependency problem in
cmenu/Makefile") and basically reverts commit fb543aa635ff ("com32:
Per-object file LDFLAGS"). By using the --as-needed ld argument we can
specify all the required shared libraries for a directory, even if not
all of the libraries are needed by every .c32 file. --as-needed takes
care of only creating DT_NEEDED entries for the shared libraries that
are actually required to resolve undefined symbols in a given module.

Furthermore, by using $(LIBS) we create a dependency between .c32
files and their shared libraries, which allows multiple make jobs to
run in parallel.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
12 years agoUse pxe_call() instead of COMBOOT API
Matt Fleming [Thu, 12 Jul 2012 12:34:06 +0000 (13:34 +0100)]
Use pxe_call() instead of COMBOOT API

This involved some other changes, such as passing flags to
unload_pxe() now that we only call it from C code, and hence don't
need to update KeepPXE.

We also needed to move some of the pxe function prototypes to
com32/include/syslinux/pxe_api.h because they are now referenced
outside of the core.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
12 years agoelflink: Replace __intcall() with direct function calls
Matt Fleming [Tue, 3 Jul 2012 07:50:13 +0000 (08:50 +0100)]
elflink: Replace __intcall() with direct function calls

There's no reason to use the COMBOOT API at all now that we can have
any undefined symbols resolved at runtime - we can just access
functions directly.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
12 years agolib/sys/module: Add support for weak symbols
Matt Fleming [Thu, 12 Jul 2012 16:39:53 +0000 (17:39 +0100)]
lib/sys/module: Add support for weak symbols

We need a means of allowing Syslinux derivative-specific functions to
remain unresolved when there is no implementation for them. For
example, unload_pxe() is referenced by common code in com32/lib, but
is only implemented for PXELINUX. By marking unload_pxe() as __weak we
can substitute references with the undefined_symbol() function for
SYSLINUX, EXTLINUX and ISOLINUX. Provided that those references are
never executed everything works fine. However, if code calls actually
calls unload_pxe() from SYSLINUX, EXTLINUX or ISOLINUX an error
message is displayed and execution is halted.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
12 years agoldlinux: Stop using the internal KT_* image types
Matt Fleming [Wed, 11 Jul 2012 12:12:33 +0000 (13:12 +0100)]
ldlinux: Stop using the internal KT_* image types

The KT_* image types were never meant to be used outside of
com32/menu, so use the external image types from syslinux/boot.h

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
12 years agocore/elflink: Delete ADV code that's already in ldlinux
Matt Fleming [Wed, 11 Jul 2012 09:17:49 +0000 (10:17 +0100)]
core/elflink: Delete ADV code that's already in ldlinux

The exact same ADV code is in both com32/elflink/ldlinux and
core/elflink. Since there's no need to have the same code in two place
at once delete the copy from the core since there are no users in
core/.

Signed-off-by: Matt Fleming <matt.fleming@intel.com>
12 years agoOnly compile dprintf/vdprintf if DEBUG_PORT is defined
H. Peter Anvin [Thu, 19 Jul 2012 15:42:32 +0000 (08:42 -0700)]
Only compile dprintf/vdprintf if DEBUG_PORT is defined

We really, really don't want to accidentally spew output to a debug
port in production, so make it obligatory to define DEBUG_PORT or
DEBUG_STDIO in order for the debugging code to be compiled in.  Since
DEBUG_STDIO just turns the debugging code into stdio references, we
only need to compile the code for the DEBUG_PORT case.

Add a commented-out line to mk/devel.mk to make it easier for users.

Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Cc: Matt Fleming <matt.fleming@intel.com>
12 years agoMerge branch 'master' into elflink
Matt Fleming [Thu, 19 Jul 2012 15:03:19 +0000 (16:03 +0100)]
Merge branch 'master' into elflink

The reason for this merge is that we need the fix to core/fs/cache.c
without which we can potentially dereference an invalid pointer.

The below conflict is a result of the new setup_data blob code being
merged from the master branch.

Conflicts:
com32/lib/Makefile

12 years agoldlinux: Return immediately after we execute() a command
Matt Fleming [Fri, 6 Jul 2012 14:36:56 +0000 (15:36 +0100)]
ldlinux: Return immediately after we execute() a command

If we successfully lookup the filetype of a command we need to return
once we've called execute(), otherwise we'll also end up calling
new_linux_kernel().

Reported-by: Sebastian Herbszt <herbszt@gmx.de>
Signed-off-by: Matt Fleming <matt.fleming@intel.com>