cting 0x10 provided that the type
code is from 0x10 to 0x1f
Pavel Roskin
r"
So stage1_lba doesn't support floppy disks, does it?
Pavel Roskin
I use the first partition on /dev/hda
stage1 works properly.
Pavel Roskin
.
I agree that my rxvt may be not perfectly tuned, but other programs are
quite happy with it.
Full keyboard support in /sbin/grub still needs a lot of work.
Pavel Roskin
Index: stage2/char_io.c
===
RCS file: /gd/gnu/anoncvsroot
Hello!
It's ridiculous to go to the command line just to exit from GRUB.
The attached patch makes /sbin/grub exit on 'q' if the menu is active.
Pavel Roskin
--- stage2/stage2.c Thu Jun 24 04:03:29 1999
+++ stage2/stage2.c Mon Jul 19 20:12:46 1999
@@ -477,6 +477,13
chainload=(fd0)+1
It should restart the GRUB on the virtual floppy. 50% that it will fail.
Pavel Roskin
e CD.
I've even seen such CD.
bootable installation-CD. If anyone has a clue, please let me know...
If chainload=(fd0)+1 works, then chainload=(fd0)nn+1 should work too.
However even if it does, the question remains open.
Pavel Roskin
Quick fix for you might be to write a 2.88M image. It colud also hold more
cool stuff :-)
Pavel Roskin
are crucial for some machines.
I would even include minixfs support with some cautionary words.
You may want to wait until I write optional getopt.c inclusion (will be
ready in 24 hours), otherwise we can be overflooded with reports from *BSD
folks.
Pavel Roskin
optimization (-fomit-frame-pointer -mrtd etc) without fear to
break things.
2) No need to write everything in assembler. Just
if ((rc != 0) || (sectors == 0)) return 1;
Pavel Roskin
Index: stage2/asm.S
===
RCS file: /gd/gnu/anoncvsroot/grub
:
movb%ah, %bl/* save return value in %bl */
/* back to protected mode */
data32
===
Pavel Roskin
Selected item cannot fit into memory
And this is stage2:
grub kernel=/kernel
Invalid or unsupported executable format
Pavel Roskin
Index: stage2/boot.c
===
RCS file: /gd/gnu/anoncvsroot/grub/stage2/boot.c,v
retrieving revision 1.4
Hello, OKUJI!
Of course, I was wrong about a.out. Sorry for that.
Loading FreeBSD in the GRUB utility was already broken.
However, the real loading by stage2 is Ok.
By the way, I was very surprized to find that the kernel for FreeBSD
2.2.5 can boot FreeBSD 3.1 and run ELF-binaries!
Pavel
FreeBSD-2.2.5Ok Ok
NetBSD-1.1 Ok not tested
FreeBSD-3.1,3.2 Ok Ok
FreeBSD loader Ok Ok
Mach-a.out Loading belownot tested
1M not supported
Pavel Roskin
Index: stage2/boot.c
Ok
FreeBSD loader Ok Ok
Mach-a.out Loading belownot tested
1M not supported
Pavel Roskin
Index: stage2/boot.c
===
RCS file: /gd/gnu/anoncvsroot/grub/stage2/boot.c,v
Hello!
The attached patch fixes incorrect screen redrawing after inserting or
deleting lines in submenus.
Pavel Roskin
--- stage2/stage2.c Thu Jun 24 04:03:29 1999
+++ stage2/stage2.c Sun Aug 8 01:44:46 1999
@@ -354,7 +354,11
grows too big, we could introduce options for configure, such as
--enable-minixfs (or maybe --enable-fs=ext2,ffs,minix)
Pavel Roskin
for stage2 (because
of the item 4), make sure that they always have the needed prototypes.
Pavel Roskin
diff -urN grub/configure.in grub.new/configure.in
--- grub/configure.in Thu Aug 5 18:01:41 1999
+++ grub.new/configure.in Tue Aug 10 03:14:05 1999
@@ -117,5 +117,40 @@
# Check
be
overridden at the make time
Changelog for the previous patch (sorry, forgot it):
* stage2/fsys_fat.c (fat_create_blocklist): return 1 for
the root directory on FAT12 and FAT16
Pavel Roskin
Index: grub/Makefile.am
to 30 characters)
Pavel Roskin
Index: stage2/fsys_minix.c
===
RCS file: /gd/gnu/anoncvsroot/grub/stage2/fsys_minix.c,v
retrieving revision 1.1
diff -u -u -r1.1 fsys_minix.c
--- fsys_minix.c1999/08/26 09:42:57 1.1
nized as C: ?
Pavel Roskin
Hello!
The attached patch is absolutely trivial:
* Makefile.am: New target - "floppy" for creating a bootable
diskette
* stage2/cmdline.c [GRUB_UTIL]: space added in the LBA warning
It is usually a good idea to require less input from root.
Pavel Roskin
Oops, forgot the patch. Sorry.
Pavel Roskin
--- Makefile.am Fri Jun 25 11:28:15 1999
+++ Makefile.am Thu Sep 2 01:53:16 1999
@@ -3,3 +3,6 @@
# We get $(PACKAGE) and $(VERSION) from debian/changelog.
configure: debian/changelog
+
+floppy: all
+ cat stage1/stage1 stage2/stage2
Englishman checked the messages after me.
The help messages for "root", "rootnoverify" and "testload" are too
hard for me :-(
I have no idea what "fix up automatic determination" means.
Also the purpose of HDBIAS remains unclear to me.
Is it something li
0x80-0x87
set_device should check what safe_parse_maxint returns.
Pavel Roskin
Index: grub/asmstub.c
===
RCS file: /var/cvs/grub/grub/asmstub.c,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -r1.1.1.1 -r1.2
wgetch(), not getch() in -l[n]curses
* stage2/shared.h: define EXT_C. Include the best existing
header for [n]curses.
Please don't forget to run:
aclocal autoheader automake autoconf
cvs add acconfig.h
Pavel Roskin
diff -urN acconfig.h acconfig.h
--- acconfig.h Thu
called during the mount.
But it looks like you have fixed it.
Pavel Roskin
BLKFLSBUF if needed.
Please add doc/version.texi to CVS. Otherwise "./configure make"
fails.
Pavel Roskin
Index: grub/asmstub.c
===
RCS file: /usr/home/proski/cvs/grub/grub/asmstub.c,v
retrieving revision 1.1.1.1
diff -u
use of retval, because it may
apply to deeper matters.
Feel free to rename `retval'. I'm not a master of renamings.
ChangeLog:
* stage2/disk_io.c (set_device): variable `retval' made integer
and not used for return anymore.
Pavel Roskin
--- stage2/disk_io.cFri Sep 24 01:33:36
.
Pavel Roskin
--- stage2/defs.h Thu Jun 24 04:03:14 1999
+++ stage2/defs.h Tue Sep 28 00:58:13 1999
@@ -56,13 +56,13 @@
}
quad; /* an 8-byte item */
-typedef unsigned int time_t; /* an unsigned int */
-typedef unsigned int daddr_t
attributes in the directory)
But it will save us from touching the code for every filesystem.
Pavel Roskin
nough stable (except for the netboot
support), I'd like to ask Gordon to release the next beta release,
0.5.93.
Everything else seems to be Ok.
Pavel Roskin
to stop moving files and release 0.5.93 unless there
are critical bugreports within few days.
Cheers,
Pavel Roskin
--- stage2/disk_io.cThu Sep 30 11:14:39 1999
+++ stage2/disk_io.cThu Sep 30 21:24:02 1999
@@ -680,7 +680,7 @@
#ifdef STAGE1_5
/* In Stage 1.5, the first 4 bytes
), it would be a good idea to use 0x7F instead of 0x79
ChangeLog:
* stage2/disk_io.c (set_bootdev): use mask 0x7F instead of 0x79
for the device number
Pavel Roskin
--- stage2/disk_io.cThu Sep 30 21:24:02 1999
+++ stage2/disk_io.cThu Sep 30 23:03:53 1999
@@ -937,7
from CVS, I suppose).
Cheers,
Pavel Roskin
"Vortex/Boomerang 3Com PCI"
This means that the PCI probing should be slightly decentralized.
netboot/config.c should just ask every available driver if it can find a
supported board.
Pavel Roskin
ses "Error: 17" during
the boot.
Why may one wish not to use "p" anyway?
Maybe "p" should just be assumed (always or just for Stage 1.5) ?
I haven't found more serious problems so far.
Best regards,
Pavel Roskin
--- stage2/builtins.c Sat Oct 2 23:36:26 1
, I'll have to provide a workaround. Most likely, this
will mean passing bootinfo in the NetBSD format.
Best wishes,
Pavel Roskin
be linked with
-nostdlib switch on NetBSD-1.4.1
Another problem is that --without-curses must be specified on NetBSD. I
haven't yet decided how to fix it (either add more defines or disable
detecting _so_ old Curses).
Pavel Roskin
Index: acinclude.m4
Hello!
This message confirms that it's currently impossible to identify NetBSD
ELF kernels (at least easily ehough). This may change some day.
This justifies some assumptions in my patch for NetBSD ELF support.
Pavel Roskin
-- Forwarded message --
Date: Mon, 11 Oct 1999 11:35
as for
FreeBSD
Pavel Roskin
Index: grub/asmstub.c
===
RCS file: /gd/gnu/anoncvsroot/grub/grub/asmstub.c,v
retrieving revision 1.26
diff -u -r1.26 asmstub.c
--- asmstub.c 1999/10/13 23:02:02 1.26
+++ asmstub.c
S won't help if the drivers are already failed to detect
the card.
Example:
isapnp (hd0,4)/etc/isapnp.conf
7) bzip2 decompression
Pavel Roskin
://ftp.gnu.org/gnu/grub ? Anybody with write permissions in /pub/gnu ?
Pavel Roskin
kernel just "return" to the bootloader?
This would be the simplest solution.
Pavel Roskin
with the GRUB shell
Pavel Roskin
--- stage2/shared.h Thu Oct 14 00:17:13 1999
+++ stage2/shared.h Thu Oct 14 01:37:28 1999
@@ -279,16 +279,29 @@
#endif /* ! A_REVERSE */
/* Special graphics characters for IBM displays. */
-#define DISP_UL 218
-#define DISP_UR 191
.
Pavel Roskin
Index: stage2/shared.h
===
RCS file: /gd/gnu/anoncvsroot/grub/stage2/shared.h,v
retrieving revision 1.29
diff -u -r1.29 shared.h
--- shared.h1999/10/27 05:55:46 1.29
+++ shared.h1999/10/27 06:44:59
ave signed a contract with a company
based near Boston, MA. I'm arriving in Boston on the 4th of November.
My address @geocities remains valid, but maybe I'll have a better address
soon. Sorry for possible delays in answers.
/OFFTOPIC
Pavel Roskin
Index: st
On Fri, 12 Nov 1999, OKUJI Yoshinori wrote:
From: Pavel Roskin [EMAIL PROTECTED]
Subject: Old AmiBIOS support
Date: Thu, 11 Nov 1999 16:38:19 -0500 (EST)
AmiBIOS (C) 1994 has very limited LBA support.
Function 0x42 (LBA read) is not supported.
That's somewhat interesting... Which
function 0x42.
Well, then I would add this check to the Stage 1 unless 4 bytes are
considered too precious.
Pavel Roskin
Hello!
Probably caused by GRUB unable to grok partition id `f' as extended
- normal extended is `5', linux extended is `85'.
So, is that `f' an anomaly of my system, or a Grub bug?
I remember me fixing this bug in GRUB. Could you please test the CVS
version?
Pavel Roskin
please investigate why GRUB printed that message?
I was unable to reproduce it.
Pavel Roskin
(init_device_map): increase the number of
devices to be probed to 8 for IDE disks and 16 for SCSI
disks. Reported by Anton Anisimov [EMAIL PROTECTED]
Pavel Roskin
Index: grub/asmstub.c
--- grub/asmstub.c Sat Dec 11 00:06:19 1999
+++ grub/asmstub.c Sat Dec 11 00:14:09 1999
tloader reports an error. The commands are exactly
as above:
map (hd1) (hd0)
map (hd0) (hd1)
rootnoverify (hd0)
chainloader (hd1,0)+1
I don't have DR DOS to test. The only thing that works for me is
chainloading GRUB from another GRUB after remapping drives.
Pavel Roskin
e geometry cannot be determined
Pavel Roskin
Index: grub/asmstub.c
--- grub/asmstub.c Fri Nov 19 12:41:24 1999
+++ grub/asmstub.c Wed Dec 15 21:36:38 1999
@@ -491,10 +491,13 @@
check_device (const char *device)
{
char buf[512];
- FILE *fp;
+ int fd;
+#ifdef __linux__
.
If you are trying to boot a kernel that doesn't have a multiboot header,
i.e. Solaris, GRUB will not do it.
Pavel Roskin
triggered by using
nodist_noinst_DATA
Pavel Roskin
Hello!
Please note that CVS versions of both Autoconf and Automake are currently
quite unstable. I don't recommend using them with GRUB.
Hahaha, I don't think this is what you really think. Don't you
remember that you used the unreleased feature `per-executable flag' in
GRUB?
I mean
Hello, Tobias!
The menu.lst entry for booting FreeBSD is:
title FreeBSD
root (hd1,a)
Try
root (hd1,a) 0xFF
Pavel Roskin
of `bootp' (my and remote IP's)
testload (nd)/zImage
also keeps printing sleep and cannot be interrupted. However, the
network card seems to be recognized, and this is good.
Pavel Roskin
ux/hdreg.h /* HDIO_GETGEO */
# include linux/major.h /* FLOPPY_MAJOR */
+# include linux/kdev_t.h /* MAJOR */
# if (__GLIBC__ 2) || ((__GLIBC__ == 2) (__GLIBC_MINOR__ 1))
/* Maybe libc doesn't have large file support. */
# include linux/unistd.h/* _llseek */
=== cut here ===
Pavel Roskin
for tftp.
filemax is negative in grub_read(), filepos is set to filemax, and len
becomes 0. gunzip_read() refuses to decompress data of zero length.
Decompressing makes sence for network boot, so I hope that somebody will
fix it.
Pavel Roskin
) even if they refuse to put identifiers into their kernels.
3) Live in the world of separated bootloaders and bootselectors, where you
cannot choose between two kernels in a menu.
Regards,
Pavel Roskin
ll contain something like "ported from OpenBSD by ..." ?
Pavel Roskin
.
Regards,
Pavel Roskin
OS doesn't support it.
I'd prefer to be wrong though. In this case, patches will be welcome :-)
Regards,
Pavel Roskin
ean, you cannot load GRUB from the hard drive or floppy and then
instruct it to load an OS from the CD.
Regards,
Pavel Roskin
detection on bootable CD's
I would not actually implement this, but other people may want to :-)
Regards,
Pavel Roskin
east newer Phoenix BIOSes are known to put a
10k hole just before 64MB, but report a proper total. */
I propose to remove that comment and "if" immediately below.
Patch and ChangeLog available by request :-)
Regards,
Pavel Roskin
run "setup (fd0,-1)"
- We probably should warn if menu.lst doesn't exist in the target
directory.
Regards,
Pavel Roskin
cut here
--- util/grub-install.inSat May 13 21:38:27 2000
+++ util/grub-install.inWed May 24 00:33:44 2000
@@ -240,7 +240,7 @@
| se
he OS guys take over
their maintainership.
Regards,
Pavel Roskin
address space or have it
there already.
Regards,
Pavel Roskin
ore
likely) or it's called in a different way than it used to be.
The former case (bad loading) can be debugged in the grub shell, without
reboots and debug prints.
Regards,
Pavel Roskin
for creating this binary somewhere in
the GRUB sources, so that everybody can do it (and improve the script if
necessary)
Do you have that script? Is there anything that you have to do manually?
Maybe it should be part of the Makefile - something like "make bootimage"
Regards,
Pavel Roskin
re is "sed" for
Windows too :-)
Regards,
Pavel Roskin
pply usual
grammar rules to the the new of verbs.
Maybe "filedes" would be better? It's at least traditional.
Regards,
Pavel Roskin
);
extern char * strsep(char **,const char *);
extern __kernel_size_t strspn(const char *,const char *);
+#endif
/*
==
Regards,
Pavel Roskin
. It's a
bootloader, not an OS. It cannot do everything in the world.
Regards,
Pavel Roskin
r priority.
Regards,
Pavel Roskin
to work.
The first gives "/boot too old: upgrade" and the second gives "bad
magic".
It looks like that OpenBSD also expects some new interface to the
bootloader. The idea of a unified bootloader doesn't seem to be popular
among *BSD folks.
Regards,
Pavel Roskin
Hello, Daniel!
GNU GRUB, as it is released by FSF, should not (IMHO) ship with
potential security holes,
Can you identify the specific security holes you fear will be exposed by
the addition of encrypted image support to GRUB? I fully intend to
document the feature, including the
you!
Regards,
Pavel Roskin
will do it for us, and probably under closed-source license.
Well, I don't insist on this item.
Regards,
Pavel Roskin
should ban people who bring
us more code as long as new features remain optional and don't delay
releasing the first stable version of GNU GRUB.
Regards,
Pavel Roskin
t would make a huge difference for its acceptance by *BSD vendors.
Regards,
Pavel Roskin
?
Regards,
Pavel Roskin
diff -u -r1.18 tutorial.texi
--- tutorial.texi 2000/08/10 02:23:04 1.18
+++ tutorial.texi 2000/08/10 21:35:52
@@ -163,7 +163,7 @@
on the floppy.
@end quotation
-On a UNIX like operating system, that is done with the following
+On a
r
programs.
If there is any opposition to fixing that particular bug in Linux then
GRUB should pull the definitions from the Linux headers. We don't want
another holy war.
Regards,
Pavel Roskin
problems rebuilding the kernel.
Regards,
Pavel Roskin
rn how gcc
and binutils work on Cygwin.
Regards,
Pavel Roskin
ur
problem is that recent *BSD still use old Binutils it's not the right list
to discuss it (IMO).
Regards,
Pavel Roskin
"compile" to the directory with the
sources to be compiled, disregarding AC_CONFIG_AUX_DIR or its absence.
Regards,
Pavel Roskin
___
Bug-grub mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/bug-grub
Hello, Christoph!
as I already described before, my prefered solution to the problem
is the use of a kind of variables in GRUB like
$MEM, $IP, $SERVER_IP, etc, etc,...
I was unaware of your proposal. It's an excelent idea, much better than my
"--mem-option".
Regards,
Pa
@@
* command.
*/
-#define ENTRY(x) .globl EXT_C(x) ; EXT_C(x) ## :
+#define ENTRY(x) .globl EXT_C(x) ; EXT_C(x):
#define VARIABLE(x) ENTRY(x)
__
Regards,
Pavel Roskin
___
Bug-grub mailing list
[EMAIL PROTECTED]
http
The solution would be to put more checks to prevent users from using the
serial connection before the device is opened. This seems to be a problem
specific to the GRUB shell.
Are you ready to make a patch?
--
Regards,
Pavel Roskin
___
Bug-grub mailing
,
Pavel Roskin
___
Bug-grub mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/bug-grub
is proprietary software. Don't worry to much
about it - there are already people in charge of it.
I just hope that it helps.
--
Regards,
Pavel Roskin
___
Bug-grub mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/bug-grub
questions (where do I get your software?) from
non-subscribers will be rejected and answered by me. Moderation delay
will be 24 hours maximum except holidays. I promise to surrender
moderation permissions or find a co-moderator if I cannot provide quick
moderation.
--
Regards,
Pavel Roskin
.
It takes at most 10 minutes a day to reject or approve postings on hold,
so I'm surprized that GRUB doesn't use such filters yet.
--
Regards,
Pavel Roskin
___
Bug-grub mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/bug-grub
it adds even more
noice, so if anybody wants to reply and the issue is not of interest for
everybody, please omit the list address from the Cc: field.
--
Regards,
Pavel Roskin
___
Bug-grub mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman
the list address yet missed the
requirement to mention the version?
Both grub.conf and splashscreen are extensions not present in the official
GRUB sources. They exist in RedHat Linux 7.2 and possibly in other
distributions, but they are not supported by the GRUB developers.
--
Regards,
Pavel
devices are not accessible.
The best solution would be to recompile the kernel to support your SCSI
adaptor in the kernel. Besides, your kernel is old - consider applying
all RedHat updates.
--
Regards,
Pavel Roskin
___
Bug-grub mailing list
[EMAIL
1 - 100 of 126 matches
Mail list logo