H. Peter Anvin [Mon, 15 Oct 2007 23:20:32 +0000 (16:20 -0700)]
Use the ADV to support boot-once
Define ADV tag 1 as boot-once; overrides DEFAULT once only
H. Peter Anvin [Thu, 11 Oct 2007 23:03:30 +0000 (16:03 -0700)]
Initial support for auxilliary data vector
Initial support for the auxillary data vector, a storage mechanism
intended to support boot-once and similar features.
H. Peter Anvin [Thu, 13 Dec 2007 18:11:05 +0000 (10:11 -0800)]
Deprecate PXELINUX option 208 in accordance with RFC 5071
As per RFC 5071, deprecate the use of PXELINUX option 208 (magic
number); options 209-211 are now specifically assigned to PXELINUX.
H. Peter Anvin [Thu, 13 Dec 2007 18:01:43 +0000 (10:01 -0800)]
3.54 is gone, now working on 3.55
H. Peter Anvin [Wed, 12 Dec 2007 23:18:20 +0000 (15:18 -0800)]
BBS on Award, not Phoenix, is broken... don't know about Phoenix.
The systems on which BBS has been found broken is Award, not Phoenix.
H. Peter Anvin [Wed, 12 Dec 2007 21:32:28 +0000 (13:32 -0800)]
Stealth whitespace cleanup
H. Peter Anvin [Tue, 11 Dec 2007 18:32:57 +0000 (10:32 -0800)]
Document cfarcall thunking support
H. Peter Anvin [Tue, 11 Dec 2007 01:54:37 +0000 (17:54 -0800)]
cdecl helper is 3.54 material
H. Peter Anvin [Mon, 10 Dec 2007 22:49:21 +0000 (14:49 -0800)]
Separate attribute control for disabled menu items
"menu color disabled" for disabled menu items.
H. Peter Anvin [Mon, 10 Dec 2007 22:35:13 +0000 (14:35 -0800)]
Documentation update: NEWS, new MENU commands
Document menu separator/menu indent/menu disable, as well as the Qemu fix.
H. Peter Anvin [Mon, 10 Dec 2007 22:28:53 +0000 (14:28 -0800)]
Fix handling of VESA modes with noncontiguous memory buffers
Handle VESA modes with noncontiguous memory buffers. In particular,
this applies to Qemu/KVM in Cirrus emulation mode.
H. Peter Anvin [Wed, 21 Nov 2007 20:29:32 +0000 (12:29 -0800)]
Add "menu separator", "menu indent", "menu disabled"
A few more options for the menu: "menu separator", "menu indent",
"menu disabled".
H. Peter Anvin [Wed, 21 Nov 2007 20:26:44 +0000 (12:26 -0800)]
spec: Require NASM 0.98.39
NASM 0.98.38 probably doesn't work anymore.
Bernard Li [Tue, 20 Nov 2007 01:26:48 +0000 (17:26 -0800)]
Patch for syslinux.spec
H. Peter Anvin [Wed, 21 Nov 2007 05:17:40 +0000 (21:17 -0800)]
When setting up the cfarcall stack frame, copy the right amount
Forgot to convert to dwords by >> 2...
H. Peter Anvin [Wed, 21 Nov 2007 05:11:35 +0000 (21:11 -0800)]
Add missing thunk function for cfarcall
H. Peter Anvin [Wed, 21 Nov 2007 02:52:47 +0000 (18:52 -0800)]
Next version is presumably 3.54
H. Peter Anvin [Wed, 21 Nov 2007 02:51:54 +0000 (18:51 -0800)]
Add support for thunking cdecl/stdcall-style real-mode code
Add support for thunking cdecl/stdcall-style 16-bit functions which
pass parameters on the stack. This is necessary in order to be able
to call PnPBIOS functions without using the (rather scary) PM
interface.
H. Peter Anvin [Tue, 20 Nov 2007 17:54:21 +0000 (09:54 -0800)]
struct term_state: remove duplicate .xy and unused .attr
In struct term_state and its initialized instance default_state:
Remove duplicated initialization of .xy
Remove unused field .attr
H. Peter Anvin [Sun, 18 Nov 2007 06:54:57 +0000 (22:54 -0800)]
mbr: remove *.bin on "spotless", not "clean"
Remove mbr/*.bin on "spotless", not "clean"; otherwise we break "make
installer".
H. Peter Anvin [Sun, 18 Nov 2007 06:50:12 +0000 (22:50 -0800)]
Document changes for 3.53
H. Peter Anvin [Sun, 18 Nov 2007 00:02:49 +0000 (16:02 -0800)]
Fix invalid segment assumption in pxelinux
Fix an invalid segment assumption in pxelinux, which caused the
"naked ::" not to work correctly.
H. Peter Anvin [Thu, 15 Nov 2007 04:51:23 +0000 (20:51 -0800)]
Preserve ES:DI instead of probing for $PnP
The $PnP probe has been found to lock up at least one machine for
reasons unknown. Drop it; instead, burn the extra few bytes in the
bootsector to save away ES:DI for later restore.
H. Peter Anvin [Wed, 14 Nov 2007 23:01:28 +0000 (15:01 -0800)]
mbr.S: fix loading from logical partitions
Fix loading from logical partitions (a push/pop pair was not done in
reverse order.)
H. Peter Anvin [Wed, 14 Nov 2007 17:43:28 +0000 (09:43 -0800)]
ISOLINUX: Change DriveNo -> DriveNumber
Change DriveNo to DriveNumber in ISOLINUX, to match the other
derivatives.
H. Peter Anvin [Wed, 14 Nov 2007 05:51:27 +0000 (21:51 -0800)]
Fix boot sectors; cleanup_hardware must preserve registers
cleanup_hardware has registers live across it when called from the
boot sector code; it must preserve all registers.
H. Peter Anvin [Wed, 14 Nov 2007 00:37:18 +0000 (16:37 -0800)]
Remove unused variable BufSafeSec
Remove unused variable BufSafeSec; it was identical to BufSafe in
every way.
At the moment, BufSafe* are always constants; consider if they can be
swapped out for constants or if we want to retain them for flexibility.
H. Peter Anvin [Sat, 10 Nov 2007 02:18:31 +0000 (18:18 -0800)]
Add sha256crypt and sha512crypt; not yet enabled
Add support for sha256crypt and sha512crypt to the library. Not yet
included in the menu system, need to decide if it's worth 10K or if
the code can be realistically compacted.
H. Peter Anvin [Sat, 10 Nov 2007 00:56:02 +0000 (16:56 -0800)]
In 32-bit mode, "xor eax,eax" is shorter than "xor ax,ax"
For 32-bit code, 32-bit operations are shorter than 16-bit operations,
so use them.
H. Peter Anvin [Thu, 8 Nov 2007 01:29:28 +0000 (17:29 -0800)]
Typo cleanup
H. Peter Anvin [Wed, 7 Nov 2007 17:26:10 +0000 (09:26 -0800)]
Guard section to avoid future partial-sector-clobber issues
Create a guard section .uibss that explicitly is covering the
potential partial-sector-clobber zone, and contains items which are
safe to assume will only be used after full image load.
H. Peter Anvin [Wed, 7 Nov 2007 01:15:13 +0000 (17:15 -0800)]
Clean up the transition to PM for shuffle-and-boot-to-PM
Clean up the transition to PM for shuffle-and-boot-to-PM, and set LDTR
and TR to keep Intel VT happy.
H. Peter Anvin [Wed, 7 Nov 2007 01:14:03 +0000 (17:14 -0800)]
isolinux: protect bss variables clobbered due to final sector load
The image size loaded is bigger than the real image; this is due to
the final sector being fractional. Move variables needed during that
stage from .bss to .bss1.
XXX: Need to audit to make sure this doesn't affect other derivatives
and/or other variables, too.
H. Peter Anvin [Mon, 5 Nov 2007 00:34:38 +0000 (16:34 -0800)]
Setting cr0 requires a jmp to synchronize on 386/486
The 386 and 486 requires a jmp after any setting of cr0. It can be a
near jmp, but those processors do not correctly synchronize without
it.
We have jmps after setting cr0 in all spots except one, so make that
happen correctly.
H. Peter Anvin [Fri, 2 Nov 2007 22:55:23 +0000 (15:55 -0700)]
Use "nop nop" to clobber "mov cs,ax" even though it's longer
Clobber "mov cs,ax" with "nop nop" even though that adds a byte to the
source; it's a lot easier to comprehend while debugging.
H. Peter Anvin [Fri, 2 Nov 2007 22:46:28 +0000 (15:46 -0700)]
Optional debugging code for shuffle and boot and load_linux
Put (disabled) debugging code in shuffle.c and load_linux.c for future
needs.
H. Peter Anvin [Fri, 2 Nov 2007 22:42:54 +0000 (15:42 -0700)]
comboot: fix shuffle_and_boot_rm (used by linux.c32)
Fix shuffle_and_boot_rm, which generated an invalid trampoline; this
was used by linux.c32 among other things.
H. Peter Anvin [Fri, 2 Nov 2007 17:40:49 +0000 (10:40 -0700)]
Remove unused EndOfGetcBuf
H. Peter Anvin [Fri, 2 Nov 2007 01:25:10 +0000 (18:25 -0700)]
Code cleanup: get rid of hard-coded descriptor numbers
Get rid of hard-coded descriptor numbers, and instead use symbolic
names for the descriptors derived from the calculated offset in the
GDT.
H. Peter Anvin [Thu, 1 Nov 2007 22:53:25 +0000 (15:53 -0700)]
mkdiskimage: correct the partition type for > 1024 cylinders
Generate the correct partition type for > 1024 cylinders.
Gerald Britton [Mon, 29 Oct 2007 21:03:17 +0000 (17:03 -0400)]
syslinux $PnP search fix
I finally fixed the "chain.c32 hangs when run on modern qemu" problem.
Apparently it was hanging hitting the "Invalid $PnP header (too short)" branch
of the code which loops without ever incrementing bx, so it never terminates.
The patch below (taken against your current git) seems to fix the problem for
me, but having not explored the code much, it probably isn't the most elegant
of fixes.
-- Gerald
H. Peter Anvin [Thu, 1 Nov 2007 22:34:09 +0000 (15:34 -0700)]
Try to fix libfat/ulint.h for non-x86 platforms
libfat/ulint.h doesn't work as expected because of array devolution to
pointers. Try to fix.
H. Peter Anvin [Wed, 31 Oct 2007 04:46:58 +0000 (21:46 -0700)]
mkdiskimage: support more than 1024 cylinders
If the user wants to generate a disk image with more than 1024
cylinders, print a warning but do the right thing, including labelling
the partitions as LBA partitions.
H. Peter Anvin [Fri, 19 Oct 2007 22:20:53 +0000 (15:20 -0700)]
DHCP bootfile option is 67, not 64.
H. Peter Anvin [Fri, 19 Oct 2007 18:33:21 +0000 (11:33 -0700)]
Fix [BI]ESUBDIRS for make installer/tidy/clean/spotless
Fix the handling of [BI]ESUBDIRS for other targets than "all".
H. Peter Anvin [Thu, 11 Oct 2007 23:05:14 +0000 (16:05 -0700)]
Add .gitignore file so "git status" makes sense
H. Peter Anvin [Thu, 11 Oct 2007 22:01:32 +0000 (15:01 -0700)]
Mark TSS unbusy before protected mode entry
The LTR instruction marks the TSS descriptor busy, and any subsequent
attempt to load it will #GP unless maked unbusy first. Sigh.
H. Peter Anvin [Thu, 11 Oct 2007 21:47:23 +0000 (14:47 -0700)]
Set TR and LDTR when entering protected mode
Intel's VT daftly requires TR and LDTR to have valid protected-mode
values in order to kick in. Thus, give it at least a valid chunk of
memory for the TR and a null selector for the LDTR.
H. Peter Anvin [Wed, 10 Oct 2007 21:12:58 +0000 (14:12 -0700)]
Makefiles: create NASMOPT variable
Create NASMOPT variable, defaulting to -O9999. Mostly there to test
beta versions of NASM.
H. Peter Anvin [Wed, 3 Oct 2007 23:17:47 +0000 (16:17 -0700)]
Update the memory layout specification
Update the memory layout specification. Remove NASM 0.98.38 bug
workarounds, rebalance the bss sections, and introduce the new section
.adv which contain the auxilliary data vector.
H. Peter Anvin [Tue, 2 Oct 2007 20:42:06 +0000 (13:42 -0700)]
Next version is 3.53
H. Peter Anvin [Tue, 2 Oct 2007 20:41:55 +0000 (13:41 -0700)]
elf.c: better error message when memory is unavailable
Print something other than just "invalid ELF file or insufficient
memory" when the specific error is that a specific memory segment is
not available.
H. Peter Anvin [Tue, 2 Oct 2007 20:36:33 +0000 (13:36 -0700)]
memdump: remove debugging code never meant for release
H. Peter Anvin [Tue, 2 Oct 2007 18:17:36 +0000 (11:17 -0700)]
bootsect.inc: make sure the $PnP header can fit inside the segment
Terminate the search as soon as the $PnP header would fall outside the
segment, not just the $PnP signature itself.
H. Peter Anvin [Mon, 24 Sep 2007 21:18:18 +0000 (14:18 -0700)]
Fix building on a 64-bit system without a 32-bit system installed
A bunch of glibc header files were bogusly included. We should not
depend on having a 32-bit glibc installed, since we don't use it.
H. Peter Anvin [Fri, 21 Sep 2007 23:54:57 +0000 (16:54 -0700)]
localboot.c32: sample/test program for invoking syslinux_local_boot()
Simple test program which simply invokes syslinux_local_boot().
H. Peter Anvin [Fri, 21 Sep 2007 23:52:35 +0000 (16:52 -0700)]
memdump: a debugging utility to dump memory over a serial port
A memory-dumping utility which runs in real mode. Thus, it can be
used to get memory dumps from a relatively pristine system.
H. Peter Anvin [Fri, 21 Sep 2007 23:26:25 +0000 (16:26 -0700)]
Document memory overwrite fix
Document menu system memory overwrite fix
H. Peter Anvin [Fri, 21 Sep 2007 23:24:35 +0000 (16:24 -0700)]
memdisk: add "nopass" option to completely take over INT 13h
Add the "nopass" option to completely remove the system drives. This
is a pretty big hammer and is largely for debugging at this stage; it
probably needs more polishing before it can be made official.
H. Peter Anvin [Fri, 21 Sep 2007 23:22:58 +0000 (16:22 -0700)]
comboot: fix memory overwrite bug
Fix memory overwrite bug, specifically, the use of BP after it might
have gotten clobbered. Since at the point of use, BP is supposed to
be == SP, we can just add "mov bp,sp".
H. Peter Anvin [Thu, 20 Sep 2007 06:13:13 +0000 (23:13 -0700)]
ISOLINUX: save segment registers around INT 13h
There is at least one set of CD-ROM BIOSes out there which corrupt FS
on INT 13h, so save and restore all segment registers around INT 13h.
H. Peter Anvin [Tue, 28 Aug 2007 05:30:21 +0000 (22:30 -0700)]
NEWS: EDD now actually works in MEMDISK
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Duane Voth [Tue, 28 Aug 2007 02:04:19 +0000 (21:04 -0500)]
MEMDISK: Fix register messup in EDD
Hello all,
I ran across a couple bugs in memdisk concerning hd images the other
day. With the attached patch I've been able to successfully boot a
32MB gzipped hd image. (I'll put the image up for a short while at
ftp://ftp.io.com/pub/usr/duanev/fdoshd.img.gz)
The pxelinux.cfg entry I'm using is:
label fdos
kernel memdisk
append keeppxe initrd=fdoshd.img.gz
syslinux/memdisk is great, thanks Peter et. al.!
--
Duane Voth
duanev@io.com
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
H. Peter Anvin [Fri, 17 Aug 2007 01:27:48 +0000 (18:27 -0700)]
Clean up some of the new PCI code from Erwan Velu.
More needs to be done; the code has console I/O at the library level,
which isn't good.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Erwan Velu [Mon, 13 Aug 2007 15:16:03 +0000 (17:16 +0200)]
Improving PCI collected informations
This patch
- add a new pci_dev_info structure :
It contains additional informations about the pci devices like
the product/vendor name and the associated linux kernel module
- add a get_name_from_pci_ids() function in pci/scan.c
This function reads a pci.ids file from the boot device.
Then it assign for each pci device, its vendor/product name.
You just have to put this file in the root directory of your
isolinux/pxelinux (i.e the root directory of your tfptboot server
if you are using pxelinux).
- add a get_module_name_from_pci_ids() function in pci/scan.c
This function reads a modules.pcimap file from the boot device.
Then it assign for each pci_device its linux kernel module.
You just have to put this file in the root directory of your
isolinux/pxelinux (i.e the root directory of your tfptboot server
if you are using pxelinux).
- Add a call to get_name_from_pci_ids() into the pcitest COM32 module
- Add a call to get_module_name_from_pci_ids() into the pcitest COM32 module
- Fixing typedef struct { ... } s_pci...; by struct pci... {};
- Improving comments
- Fixing the memory allocation to prevent leaks
With this patch, pcitest.c32 act like lspci plus a bonus by displaying
the linux kernel module assiocated to each pci device.
Signed-off-by:Erwan Velu <erwan.velu@free.fr>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
H. Peter Anvin [Fri, 17 Aug 2007 01:07:52 +0000 (18:07 -0700)]
Use git push instead of cg-push
H. Peter Anvin [Tue, 14 Aug 2007 22:20:13 +0000 (15:20 -0700)]
Add reboot.c32 module
A very simple module to reboot the machine.
H. Peter Anvin [Tue, 14 Aug 2007 22:19:32 +0000 (15:19 -0700)]
com32: fix __farcall()
The address for the farcall thunk was inadvertently clobbered in entry.S.
H. Peter Anvin [Sun, 12 Aug 2007 22:37:07 +0000 (15:37 -0700)]
More Etherboot pointers
H. Peter Anvin [Sun, 12 Aug 2007 22:36:07 +0000 (15:36 -0700)]
Document gPXE chaining to work around PXE stack problems
H. Peter Anvin [Mon, 6 Aug 2007 23:24:23 +0000 (16:24 -0700)]
Fix section confusion...
H. Peter Anvin [Mon, 6 Aug 2007 23:24:14 +0000 (16:24 -0700)]
Give ourselves more room for code
H. Peter Anvin [Wed, 25 Jul 2007 21:17:58 +0000 (14:17 -0700)]
Handle plural correctly in countdown.
H. Peter Anvin [Wed, 25 Jul 2007 05:54:24 +0000 (22:54 -0700)]
Install sha1pass, md5pass
H. Peter Anvin [Wed, 25 Jul 2007 05:50:18 +0000 (22:50 -0700)]
Add header file missing from previous checkin (md5.h)
H. Peter Anvin [Wed, 25 Jul 2007 05:49:05 +0000 (22:49 -0700)]
Add support for Unix-standard MD5 password
H. Peter Anvin [Wed, 25 Jul 2007 00:26:39 +0000 (17:26 -0700)]
Document MENU HIDDEN
H. Peter Anvin [Wed, 25 Jul 2007 00:25:57 +0000 (17:25 -0700)]
Correctly handle switches between graphics and text mode
H. Peter Anvin [Wed, 25 Jul 2007 00:15:07 +0000 (17:15 -0700)]
Make the hidden timeout behaviour match the rest of the menu
H. Peter Anvin [Wed, 25 Jul 2007 00:12:51 +0000 (17:12 -0700)]
Move hidden_row to -2; don't re-timeout once a key is pressed
H. Peter Anvin [Tue, 24 Jul 2007 23:39:00 +0000 (16:39 -0700)]
Simple menu system: add support for "menu hidden"
H. Peter Anvin [Tue, 24 Jul 2007 00:26:49 +0000 (17:26 -0700)]
Document MEMDISK fixes.
H. Peter Anvin [Fri, 20 Jul 2007 23:00:45 +0000 (16:00 -0700)]
Add the actual zeroregs file; missing from a previous checkin.
H. Peter Anvin [Fri, 20 Jul 2007 22:31:09 +0000 (15:31 -0700)]
Add syslinux_reboot()
H. Peter Anvin [Fri, 20 Jul 2007 22:30:55 +0000 (15:30 -0700)]
com32: add a common zero-register structure
When we don't have any registers to pass, it's convenient to be able
to share a common all-zero structure.
H. Peter Anvin [Thu, 12 Jul 2007 00:51:27 +0000 (17:51 -0700)]
MBR: shave off another byte
Yet another byte bites the dust...
H. Peter Anvin [Thu, 12 Jul 2007 00:44:18 +0000 (17:44 -0700)]
MBR: shave off another byte
Shave off another byte by using the not-frequently-used jmpw *%sp
instruction!
H. Peter Anvin [Thu, 12 Jul 2007 00:39:55 +0000 (17:39 -0700)]
MBR: one more comment
Comments are good...
H. Peter Anvin [Thu, 12 Jul 2007 00:37:00 +0000 (17:37 -0700)]
MBR: shave another ~4 bytes through an ugly error message hack
Use the "return address points to data" trick to shave a few bytes off
the generation of error messages.
H. Peter Anvin [Thu, 12 Jul 2007 00:25:59 +0000 (17:25 -0700)]
MBR: add HLT to prevent system from melting on failure
In case we hit the death loop, put HLT in it. It's probably superfluous,
but it can't hurt, and we can always take it out to reclaim the space.
Arguable the death loop is superfluous at all, since int $0x18 should
never return.
H. Peter Anvin [Thu, 12 Jul 2007 00:23:11 +0000 (17:23 -0700)]
MBR: add CR LF to the end of messages
Six bytes down the drain, but people do expect it. If it becomes
an issue we can remove the period; it's better than the CR LF. There
might be some way to efficiently issue . \r \n in common code, too.
H. Peter Anvin [Thu, 12 Jul 2007 00:20:59 +0000 (17:20 -0700)]
MBR: shave 2 more bytes
We always re-read the last partition table on any kind of error, so
no need to prevent the code from clobbering it on failure.
H. Peter Anvin [Wed, 11 Jul 2007 23:51:46 +0000 (16:51 -0700)]
MBR: Don't clobber %eax before entering CBIOS code
H. Peter Anvin [Wed, 11 Jul 2007 23:45:27 +0000 (16:45 -0700)]
MBR: always do setup for EBIOS, makes stack handling easier and saves jmp
H. Peter Anvin [Wed, 11 Jul 2007 23:41:52 +0000 (16:41 -0700)]
MBR: Handle the stack in read_sector correctly...
H. Peter Anvin [Wed, 11 Jul 2007 23:39:25 +0000 (16:39 -0700)]
One more MBR byte
H. Peter Anvin [Wed, 11 Jul 2007 23:38:30 +0000 (16:38 -0700)]
Another 12 bytes shaved off the MBR...
H. Peter Anvin [Wed, 11 Jul 2007 23:22:46 +0000 (16:22 -0700)]
Another 2 bytes off the MBR.
H. Peter Anvin [Wed, 11 Jul 2007 23:14:07 +0000 (16:14 -0700)]
Add some comments to the MBR
H. Peter Anvin [Wed, 11 Jul 2007 23:11:36 +0000 (16:11 -0700)]
Another 3 bytes or so off the MBR...