Re: Wrong format of kernel.img if build on Cygwin (was: no subject)
Lode Leroy wrote: Did the core.img actually work as expected then? unfortunately no, it doesn't boot. booting the iso from grub-mkrescue in qemu I get the following message : Booting from CD-Rom... 1MB medium detected rub_env_seteub_exitub_fatal === (these letters in reverse video) This is possibly the problem that libbfd cannot not convert PE to ELF properly. Unfortunately the grub-pe2elf tool does support pe-i386 but not pei-i386. I will try to fix this next week. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: Wrong format of kernel.img if build on Cygwin (was: no subject)
lode leroy wrote: I tried compiling grub from bzr on cygwin. grub-mkimage failed with the following error: ~/grub/build/grub-core $ ../grub-mkimage -v -O i386-pc -d . -o core.img biosdisk part_msdos fat ntfs ../grub-mkimage: info: getting the size of ./biosdisk.mod. [...] ../grub-mkimage: info: getting the size of ./kernel.img. ../grub-mkimage: error: invalid ELF header. In the past (1.98), kernel.img was a raw binary. Now it is an ELF file which requires extra conversion if build on Cygwin. I will provide a patch to fix Cygwin build for 1.99. I found that the problem can be solved by replacing the following line in the Makefile strip -v -R .rel.dyn -R .reginfo -R .note -R .comment -o kernel.img kernel.exec.exe by strip -Felf32-i386 -v -R .rel.dyn -R .reginfo -R .note -R .comment -o kernel.img kernel.exec.exe Did the core.img actually work as expected then? Thanks for the problem report. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Build fails due to wrong placement of -llibrary options
I'm trying to fix the build on Cygwin after the recent move to the new autogen/automake build system. There is one Cygwin independent issue found so far: $ make [...] gcc [...] -lintl -o grub-bin2h.exe util/grub_bin2h-bin2h.o libgrub.a libgrub.a(...): undefined reference to `_libintl_gettext' [...] Workaround: $ make LIBS=-lintl [...] gcc [...] -lintl -o grub-bin2h.exe util/grub_bin2h-bin2h.o libgrub.a -lintl The placement of -llibrary options matter because the linker searches libraries and objects in the order specified. $ grep ^grub_bin2h_LD Makefile grub_bin2h_LDADD = libgrub.a grub_bin2h_LDFLAGS = $(AM_LDFLAGS) $(LDFLAGS_PROGRAM) $(LIBINTL) $(LIBDEVMAPPER) The above should be: grub_bin2h_LDADD = libgrub.a $(LIBINTL) $(LIBDEVMAPPER) grub_bin2h_LDFLAGS = $(AM_LDFLAGS) $(LDFLAGS_PROGRAM) Same for all other utils. This likely needs to be fixed in gentpl.py. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: Build fails due to wrong placement of -llibrary options
BVK Chaitanya wrote: On Sat, Aug 28, 2010 at 5:53 PM, Christian Franke ... wrote: $ grep ^grub_bin2h_LD Makefile grub_bin2h_LDADD = libgrub.a grub_bin2h_LDFLAGS = $(AM_LDFLAGS) $(LDFLAGS_PROGRAM) $(LIBINTL) $(LIBDEVMAPPER) The above should be: grub_bin2h_LDADD = libgrub.a $(LIBINTL) $(LIBDEVMAPPER) grub_bin2h_LDFLAGS = $(AM_LDFLAGS) $(LDFLAGS_PROGRAM) Same for all other utils. This likely needs to be fixed in gentpl.py. It is not necessary to modify gentpl.py, fixing in Makefile.util.def should be sufficient. Attached patch fixes all these AFAICS. With the patch it works, thanks! Please apply. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: grub2 build under windows environment.
BVK Chaitanya wrote: GRUB2 must build fine on cygwin. You need to have necessary development packages installed. There are also binary and source packages for GRUB 1.98 available in the Cygwin distro (http://cygwin.com/packages/). The file /usr/share/doc/Cygwin/grub.README file provides a list of necessary development packages. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Optimise memset on i386
richardvoigt wrote:
might I suggest:
unsigned long patternl = pattern8;
patternl |= patternl 8;
patternl |= patternl 16;
patternl |= patternl 32;
patternl |= patternl 64;
O(lg N) instead of O(N), no loop, no branches, and the compiler should
be smart enough to optimize away the last two lines on systems with
narrower long.
The latter is unfortunately not the case. At least gcc 4.5.0 prints a
warning but still produces code.
$ cat EOF f.c
unsigned long f(unsigned long x)
{
x |= x 32;
x |= x 64;
return x;
}
$ gcc -O3 -S f.c
x.c: In function ‘f’:
x.c:3: warning: left shift count = width of type
x.c:4: warning: left shift count = width of type
$ cat f.s
...
pushl %ebp
movl$32, %ecx
movl%esp, %ebp
movl8(%ebp), %eax
popl%ebp
movl%eax, %edx
sall%cl, %edx
movl$64, %ecx
orl %eax, %edx
movl%edx, %eax
sall%cl, %eax
orl %edx, %eax
ret
--
Regards,
Christian Franke
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Fix grub.d/10_windows for Cygwin 1.7
Vladimir 'φ-coder/phcoder' Serbinenko wrote:
Christian Franke wrote:
This change is already included in Cygwin package grub-1.98-2.
-
+ test -z $needmap || catEOF
+ drivemap -s (hd0) \$root
+EOF
You can't be sure on script runtime that C: is on hd0. Ironically especially
when drivemap is used.
But drivemap -s (hd0) (hd0) is a nop, so no harm done.
On the other hand Vista and 7 are incompatible with drivemap so you have to
skip drivemap on those systems. Look at 30_os-prober.in:
case ${LONGNAME} in
Windows\ Vista*|Windows\ 7*)
;;
*)
cat EOF
drivemap -s (hd0) \${root}
EOF
;;
esac
This is already considered in the script: $needmap is set if ntldr is
found but not if bootmgr is found.
--
Regards,
Christian Franke
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
[PATCH] Fix grub.d/10_windows for Cygwin 1.7
This change is already included in Cygwin package grub-1.98-2.
--
Regards,
Christian Franke
=== modified file 'ChangeLog'
--- ChangeLog 2010-05-05 19:19:55 +
+++ ChangeLog 2010-05-05 20:09:01 +
@@ -1,3 +1,9 @@
+2010-05-05 Christian Franke fra...@computer.org
+
+ * util/grub.d/10_windows.in: Use path names instead of
+ drive letters to prevent warning from Cygwin 1.7.
+ Add drivemap command to menuentry if needed.
+
2010-05-05 Vladimir Serbinenko phco...@gmail.com
* video/readers/jpeg.c: Indented.
=== modified file 'util/grub.d/10_windows.in'
--- util/grub.d/10_windows.in 2009-11-20 08:41:20 +
+++ util/grub.d/10_windows.in 2010-05-05 12:22:43 +
@@ -28,8 +28,8 @@
# Try C: even if current system is on other partition.
case $SYSTEMDRIVE in
- [Cc]:) dirlist=C: ;;
- [D-Zd-z]:) dirlist=C: $SYSTEMDRIVE ;;
+ [Cc]:) drives=C: ;;
+ [D-Zd-z]:) drives=C: $SYSTEMDRIVE ;;
*) exit 0 ;;
esac
@@ -51,7 +51,13 @@
}
-for dir in $dirlist ; do
+for drv in $drives ; do
+
+ # Convert to Cygwin path.
+ dir=`cygpath $drv`
+ test -n $dir || continue
+
+ needmap=
# Check for Vista bootmgr.
if [ -f $dir/bootmgr -a -f $dir/boot/bcd ] ; then
@@ -60,6 +66,7 @@
# Check for NTLDR.
elif [ -f $dir/ntldr -a -f $dir/ntdetect.com -a -f $dir/boot.ini ] ; then
OS=`get_os_name_from_boot_ini $dir/boot.ini` || OS=Windows NT/2000/XP loader
+needmap=t
else
continue
@@ -68,14 +75,16 @@
# Get boot /dev/ice.
dev=`${grub_probe} -t device $dir 2/dev/null` || continue
- echo Found $OS on $dir ($dev) 2
+ echo Found $OS on $drv ($dev) 2
cat EOF
menuentry $OS {
EOF
save_default_entry | sed -e 's,^,\t,'
prepare_grub_to_access_device $dev | sed 's,^,\t,'
-
+ test -z $needmap || cat EOF
+ drivemap -s (hd0) \$root
+EOF
cat EOF
chainloader +1
}
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Fix Cygwin path handling
Vladimir 'φ-coder/phcoder' Serbinenko wrote: Christian Franke wrote: + /* Include leading slash. */ + offset = 0; + break; Why do you make this change? It seems to make the behaviour of make_path_relative_to_its root inconsistent since E.g. /boot/grub - grub /etc - /etc Could not reproduce. Old and new code always include the leading slash (except / - ''). -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Fix Cygwin path handling
Vladimir 'φ-coder/phcoder' Serbinenko wrote:
I actually seen the '/' case and have incorrectly assumed it's always
so. However there is still one case when you patch changes the behaviour:
$ grub-mkrelpath /
$ ./grub-mkrelpath /
/
Although:
$ ./grub-mkrelpath /mnt/boot/
$ grub-mkrelpath /mnt/boot/
I actually did not test the / corner case, sorry. New patch attached.
--
Regards,
Christian Franke
=== modified file 'ChangeLog'
--- ChangeLog 2010-04-26 01:35:55 +
+++ ChangeLog 2010-04-26 09:44:43 +
@@ -1,5 +1,17 @@
2010-04-26 Christian Franke fra...@computer.org
+ * util/grub-mkconfig_lib.in (make_system_path_relative_to_its_root):
+ Remove broken Cygwin path conversion.
+ * util/misc.c: [__CYGWIN__] Add include and define.
+ [__CYGWIN__] (get_win32_path): Copy function from getroot.c, modify
+ for Cygwin 1.7.
+ (make_system_path_relative_to_its_root): Simplify loop, replace early
+ return by break.
+ [__CYGWIN__] Add conversion to win32 path.
+ Include / case in trailing slash removal.
+
+2010-04-26 Christian Franke fra...@computer.org
+
* include/grub/util/getroot.h (grub_get_prefix): Remove prototype.
* util/getroot.c [__CYGWIN__] (get_win32_path): Remove function.
(grub_get_prefix): Remove function.
=== modified file 'util/grub-mkconfig_lib.in'
--- util/grub-mkconfig_lib.in 2010-04-19 19:25:41 +
+++ util/grub-mkconfig_lib.in 2010-04-26 08:59:10 +
@@ -1,5 +1,5 @@
# Helper library for grub-mkconfig
-# Copyright (C) 2007,2008,2009 Free Software Foundation, Inc.
+# Copyright (C) 2007,2008,2009,2010 Free Software Foundation, Inc.
#
# GRUB is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
@@ -44,21 +44,7 @@
make_system_path_relative_to_its_root ()
{
- path=`${grub_mkrelpath} $1`
-
- case `uname 2/dev/null` in
-CYGWIN*)
- # Cygwin: Check if regular or emulated mount.
- if [ -z $dir ] || [ `stat -c %D $dir/..` != 62 ] ; then
-# Reached some mount point not below /cygdrive.
-# GRUB does not know Cygwin's emulated mounts,
-# convert to Win32 path and remove drive letter.
-path=`cygpath -m $path | sed -n 's,^[A-Za-z]:,,p'`
-test ! -z $path || return 1
- fi ;;
- esac
-
- echo $path
+ ${grub_mkrelpath} $1
}
is_path_readable_by_grub ()
=== modified file 'util/misc.c'
--- util/misc.c 2010-04-09 23:25:46 +
+++ util/misc.c 2010-04-26 10:08:04 +
@@ -53,6 +53,11 @@
# include malloc.h
#endif
+#ifdef __CYGWIN__
+# include sys/cygwin.h
+# define DEV_CYGDRIVE_MAJOR 98
+#endif
+
#ifdef __MINGW32__
#include windows.h
#include winioctl.h
@@ -456,6 +461,27 @@
return ret;
}
+#ifdef __CYGWIN__
+/* Convert POSIX path to Win32 path,
+ remove drive letter, replace backslashes. */
+static char *
+get_win32_path (const char *path)
+{
+ char winpath[PATH_MAX];
+ if (cygwin_conv_path (CCP_POSIX_TO_WIN_A, path, winpath, sizeof(winpath)))
+grub_util_error (cygwin_conv_path() failed);
+
+ int len = strlen (winpath);
+ int offs = (len 2 winpath[1] == ':' ? 2 : 0);
+
+ int i;
+ for (i = offs; i len; i++)
+if (winpath[i] == '\\')
+ winpath[i] = '/';
+ return xstrdup (winpath + offs);
+}
+#endif
+
/* This function never prints trailing slashes (so that its output
can be appended a slash unconditionally). */
char *
@@ -521,30 +547,31 @@
/* offset == 1 means root directory. */
if (offset == 1)
{
- free (buf);
- len = strlen (buf2);
- while (buf2[len - 1] == '/' len 1)
- {
- buf2[len - 1] = '\0';
- len--;
- }
- if (len 1)
- return buf2;
- else
- {
- /* This means path given is just a backslash. As above
- we have to return an empty string. */
- free (buf2);
- return xstrdup ();
- }
+ /* Include leading slash. */
+ offset = 0;
+ break;
}
}
free (buf);
buf3 = xstrdup (buf2 + offset);
free (buf2);
+#ifdef __CYGWIN__
+ if (st.st_dev != (DEV_CYGDRIVE_MAJOR 16))
+{
+ /* Reached some mount point not below /cygdrive.
+ GRUB does not know Cygwin's emulated mounts,
+ convert to Win32 path. */
+ grub_util_info (Cygwin path = %s\n, buf3);
+ char * temp = get_win32_path (buf3);
+ free (buf3);
+ buf3 = temp;
+}
+#endif
+
+ /* Remove trailing slashes, return empty string if root directory. */
len = strlen (buf3);
- while (buf3[len - 1] == '/' len 1)
+ while (len 0 buf3[len - 1] == '/')
{
buf3[len - 1] = '\0';
len--;
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: NEED_REGISTER_FRAME_INFO can be replaced by -static-libgcc
Vladimir 'φ-coder/phcoder' Serbinenko wrote: Christian Franke wrote: According to 'gcc -dumpspecs' of Cygwin gcc 4.3.4-3, options '-u ___[de]register_frame_info' are always passed to the linker if -static-libgcc is not specified. This is Cygwin and MinGW specific: In the exe startup code these symbols are loaded only if present, so the DLL must be forced to load first. As a consequence, the symbols are also set undefined when the GRUB pre-*.o files are generated with 'ld -r'. The attached patch works for me with the bzr revision preceding the NEED_REGISTER_FRAME_INFO fix. Doesn't it risk linking e.g. __bswapsi2 into every module which uses it during partial link? It would increase module size on RISC Partial linking with '-Wl,-r' does never pull any code from libraries. Specifying '-lgcc' has no effect in this case. Cannot test this on RISC, it is at least true on i686-linux, -freebsd and -cygwin. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
[PATCH] Remove grub_get_prefix()
This patch removes grub_get_prefix() and replaces the two remaining
calls with make_system_path_relative_to_its_root().
Already discussed in thread [PATCH] Fix Cygwin path handling.
Change in util/grub-emu.c was tested, but not the change in
util/sparc64/ieee1275/grub-setup.c.
--
Regards,
Christian Franke
=== modified file 'ChangeLog'
--- ChangeLog 2010-04-18 11:00:06 +
+++ ChangeLog 2010-04-18 14:39:41 +
@@ -1,3 +1,12 @@
+2010-04-18 Christian Franke fra...@computer.org
+
+ * include/grub/util/getroot.h (grub_get_prefix): Remove prototype.
+ * util/getroot.c [__CYGWIN__] (get_win32_path): Remove function.
+ (grub_get_prefix): Remove function.
+ * util/grub-emu.c (main): Replace grub_get_prefix () call by
+ make_system_path_relative_to_its_root ().
+ * util/sparc64/ieee1275/grub-setup.c (main): Likewise.
+
2010-04-18 Vladimir Serbinenko phco...@gmail.com
* po/POTFILES: Renamed multiboot_loader.c to multiboot.c
@@ -10,7 +19,7 @@
* disk/lvm.c (grub_lvm_memberlist): Issue an error if pv-disk = 0.
-2010-04-17 Christian Franke fra...@computer.org
+2010-04-17 Christian Franke fra...@computer.org
* Makefile.in: Add missing localedir setting.
=== modified file 'include/grub/util/getroot.h'
--- include/grub/util/getroot.h 2009-04-11 09:40:39 +
+++ include/grub/util/getroot.h 2010-04-18 14:19:18 +
@@ -26,7 +26,6 @@
};
char *grub_guess_root_device (const char *dir);
-char *grub_get_prefix (const char *dir);
int grub_util_get_dev_abstraction (const char *os_dev);
char *grub_util_get_grub_dev (const char *os_dev);
const char *grub_util_check_block_device (const char *blk_dev);
=== modified file 'util/getroot.c'
--- util/getroot.c 2010-04-10 15:07:33 +
+++ util/getroot.c 2010-04-18 14:21:03 +
@@ -79,103 +79,6 @@
return path;
}
-#ifdef __CYGWIN__
-/* Convert POSIX path to Win32 path,
- remove drive letter, replace backslashes. */
-static char *
-get_win32_path (const char *path)
-{
- char winpath[PATH_MAX];
- cygwin_conv_to_full_win32_path (path, winpath);
-
- int len = strlen (winpath);
- if (len 2 winpath[1] == ':')
-{
- len -= 2;
- memmove (winpath, winpath + 2, len + 1);
-}
-
- int i;
- for (i = 0; i len; i++)
-if (winpath[i] == '\\')
- winpath[i] = '/';
- return xstrdup (winpath);
-}
-#endif
-
-char *
-grub_get_prefix (const char *dir)
-{
- char *saved_cwd;
- char *abs_dir, *prev_dir;
- char *prefix;
- struct stat st, prev_st;
-
- /* Save the current directory. */
- saved_cwd = xgetcwd ();
-
- if (chdir (dir) 0)
-grub_util_error (cannot change directory to `%s', dir);
-
- abs_dir = xgetcwd ();
- strip_extra_slashes (abs_dir);
- prev_dir = xstrdup (abs_dir);
-
- if (stat (., prev_st) 0)
-grub_util_error (cannot stat `%s', dir);
-
- if (! S_ISDIR (prev_st.st_mode))
-grub_util_error (`%s' is not a directory, dir);
-
- while (1)
-{
- if (chdir (..) 0)
- grub_util_error (cannot change directory to the parent);
-
- if (stat (., st) 0)
- grub_util_error (cannot stat current directory);
-
- if (! S_ISDIR (st.st_mode))
- grub_util_error (current directory is not a directory???);
-
- if (prev_st.st_dev != st.st_dev || prev_st.st_ino == st.st_ino)
- break;
-
- free (prev_dir);
- prev_dir = xgetcwd ();
- prev_st = st;
-}
-
- strip_extra_slashes (prev_dir);
- prefix = xmalloc (strlen (abs_dir) - strlen (prev_dir) + 2);
- prefix[0] = '/';
- strcpy (prefix + 1, abs_dir + strlen (prev_dir));
- strip_extra_slashes (prefix);
-
- if (chdir (saved_cwd) 0)
-grub_util_error (cannot change directory to `%s', dir);
-
-#ifdef __CYGWIN__
- if (st.st_dev != (DEV_CYGDRIVE_MAJOR 16))
-{
- /* Reached some mount point not below /cygdrive.
- GRUB does not know Cygwin's emulated mounts,
- convert to Win32 path. */
- grub_util_info (Cygwin prefix = %s, prefix);
- char * wprefix = get_win32_path (prefix);
- free (prefix);
- prefix = wprefix;
-}
-#endif
-
- free (saved_cwd);
- free (abs_dir);
- free (prev_dir);
-
- grub_util_info (prefix = %s, prefix);
- return prefix;
-}
-
#ifdef __MINGW32__
static char *
=== modified file 'util/grub-emu.c'
--- util/grub-emu.c 2010-02-07 16:52:11 +
+++ util/grub-emu.c 2010-04-18 14:38:28 +
@@ -242,7 +242,7 @@
if (strcmp (root_dev, host) == 0)
dir = xstrdup (dir);
else
-dir = grub_get_prefix (dir);
+dir = make_system_path_relative_to_its_root (dir);
prefix = xmalloc (strlen (root_dev) + 2 + strlen (dir) + 1);
sprintf (prefix, (%s)%s, root_dev, dir);
free (dir);
=== modified file 'util/sparc64/ieee1275/grub-setup.c'
--- util/sparc64/ieee1275/grub-setup.c 2010-01-16 00:26:52 +
+++ util/sparc64/ieee1275/grub-setup.c 2010-04-18 14:38:02 +
@@ -635,7 +635,8 @@
find_dest_dev (ginfo, argv);
- ginfo.prefix = grub_get_prefix (ginfo.dir ? : DEFAULT_DIRECTORY);
+ ginfo.prefix = make_system_path_relative_to_its_root (ginfo.dir
Re: [PATCH] Fix Cygwin path handling
Vladimir 'φ-coder/phcoder' Serbinenko wrote: Christian Franke wrote: The Cywin path handling is broken since make_system_path_relative_to_its_root() functionality was moved from the lib script to misc.c. This patch should fix this. It reuses the Cygwin specific code from getroot.c:grub_get_prefix() which apparently is a different implementation of the same function. I would suggest to remove grub_get_prefix(), it is now only used in grub-emu.c and sparc64/ieee1275/grub-setup.c. Not included in the patch, should be done in a separate commit. 2010-04-14 Christian Frankefra...@computer.org * util/grub-mkconfig_lib.in (make_system_path_relative_to_its_root): Remove broken Cygwin path conversion. * util/misc.c: [__CYGWIN__] Add include and define. [__CYGWIN__] (get_win32_path): Copy function from getroot.c, modify for Cygwin 1.7. Please avoid duplicating code. Rather than that rename get_win32_path to grub_get_win32_path and remove static attribute Normally I would have done that but duplication was intentional in this case: The getroot.c:get_win32_path() can later be removed together with grub_get_prefix(), see my suggestion above. The patch takes this into account and adds new private misc.c:get_win32_path() and so avoids unnecessary temporary changes to misc.h and getroot.c. The actual code duplication happened when misc.c:make_system_path_relative_to_its_root() was added instead of moving and reusing getroot.c:grub_get_prefix() :-) BTW: My last commits to grub codebase were before the move to bzr. As far as I understand Bazaar workflow for GRUB (http://lists.gnu.org/archive/html/grub-devel/2010-01/msg00175.html) such changes should be 'bzr push'ed to e.g. '.../branches/feature-foo' (e.g. '.../branches/cygwin-path' in this case) after review has finished. Is this workflow still valid or is there a more current document? -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: NEED_REGISTER_FRAME_INFO can be replaced by -static-libgcc
Vladimir 'φ-coder/phcoder' Serbinenko wrote: Christian Franke wrote: The *_frame_info symbols are set undefined to force linkage of the libgcc_s shared library or dll. This can be prevented by TARGET_LDFLAGS=-static-libgcc. To build from grub-1.98 tarball on Cygwin, run configure with this argument. May also be necessary on other build platforms when -shared-libgcc is the default. It may be possible simply set -static-libgcc unconditionally. We already supply -lgcc --static-libgcc. Do you know why it still had issues? The above is only set when linking the kernel image. According to 'gcc -dumpspecs' of Cygwin gcc 4.3.4-3, options '-u ___[de]register_frame_info' are always passed to the linker if -static-libgcc is not specified. This is Cygwin and MinGW specific: In the exe startup code these symbols are loaded only if present, so the DLL must be forced to load first. As a consequence, the symbols are also set undefined when the GRUB pre-*.o files are generated with 'ld -r'. The attached patch works for me with the bzr revision preceding the NEED_REGISTER_FRAME_INFO fix. And I also doubt usefullness of pulling these functions since reference to them is purely dummy: no relocation uses it so it will only increase code size. Another concern is the behviour of these functions in grub environment if they ever get called. AFAICS from the gcc source, the functions are used in EH stack unwinding code in libgcc. The compiler itself does not generate calls to these functions. -- Regards, Christian Franke === modified file 'Makefile.in' --- Makefile.in 2010-04-10 23:14:31 + +++ Makefile.in 2010-04-14 08:04:34 + @@ -98,12 +98,12 @@ OBJCONV = @OBJCONV@ TARGET_CPPFLAGS = @TARGET_CPPFLAGS@ -I$(srcdir)/include -I$(builddir) -I$(builddir)/include \ -Wall -W -TARGET_LDFLAGS = -nostdlib @TARGET_LDFLAGS@ +TARGET_LDFLAGS = -nostdlib -static-libgcc @TARGET_LDFLAGS@ TARGET_IMG_LDSCRIPT = @TARGET_IMG_LDSCRIPT@ TARGET_IMG_LDFLAGS = -nostdlib @TARGET_IMG_LDFLAGS@ TARGET_IMG_CFLAGS = @TARGET_IMG_CFLAGS@ TARGET_OBJ2ELF = @TARGET_OBJ2ELF@ -kernel_img_LDFLAGS = -static-libgcc -lgcc +kernel_img_LDFLAGS = -lgcc EXEEXT = @EXEEXT@ OBJCOPY = @OBJCOPY@ STRIP = @STRIP@ ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
[PATCH] Fix Cygwin path handling
The Cywin path handling is broken since
make_system_path_relative_to_its_root() functionality was moved from the
lib script to misc.c.
This patch should fix this. It reuses the Cygwin specific code from
getroot.c:grub_get_prefix() which apparently is a different
implementation of the same function.
I would suggest to remove grub_get_prefix(), it is now only used in
grub-emu.c and sparc64/ieee1275/grub-setup.c. Not included in the patch,
should be done in a separate commit.
2010-04-14 Christian Frankefra...@computer.org
* util/grub-mkconfig_lib.in (make_system_path_relative_to_its_root):
Remove broken Cygwin path conversion.
* util/misc.c: [__CYGWIN__] Add include and define.
[__CYGWIN__] (get_win32_path): Copy function from getroot.c, modify
for Cygwin 1.7.
(make_system_path_relative_to_its_root): Simplify loop, replace early
return by break.
[__CYGWIN__] Add conversion to win32 path.
--
Regards,
Christian Franke
=== modified file 'util/grub-mkconfig_lib.in'
--- util/grub-mkconfig_lib.in 2010-04-13 12:57:56 +
+++ util/grub-mkconfig_lib.in 2010-04-14 16:48:44 +
@@ -1,5 +1,5 @@
# Helper library for grub-mkconfig
-# Copyright (C) 2007,2008,2009 Free Software Foundation, Inc.
+# Copyright (C) 2007,2008,2009,2010 Free Software Foundation, Inc.
#
# GRUB is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
@@ -38,21 +38,7 @@
make_system_path_relative_to_its_root ()
{
- path=`${grub_mkrelpath} $1`
-
- case `uname 2/dev/null` in
-CYGWIN*)
- # Cygwin: Check if regular or emulated mount.
- if [ -z $dir ] || [ `stat -c %D $dir/..` != 62 ] ; then
-# Reached some mount point not below /cygdrive.
-# GRUB does not know Cygwin's emulated mounts,
-# convert to Win32 path and remove drive letter.
-path=`cygpath -m $path | sed -n 's,^[A-Za-z]:,,p'`
-test ! -z $path || return 1
- fi ;;
- esac
-
- echo $path
+ ${grub_mkrelpath} $1
}
is_path_readable_by_grub ()
=== modified file 'util/misc.c'
--- util/misc.c 2010-04-09 23:25:46 +
+++ util/misc.c 2010-04-14 16:48:44 +
@@ -53,6 +53,11 @@
# include malloc.h
#endif
+#ifdef __CYGWIN__
+# include sys/cygwin.h
+# define DEV_CYGDRIVE_MAJOR 98
+#endif
+
#ifdef __MINGW32__
#include windows.h
#include winioctl.h
@@ -456,6 +461,31 @@
return ret;
}
+#ifdef __CYGWIN__
+/* Convert POSIX path to Win32 path,
+ remove drive letter, replace backslashes. */
+static char *
+get_win32_path (const char *path)
+{
+ char winpath[PATH_MAX];
+ if (cygwin_conv_path (CCP_POSIX_TO_WIN_A, path, winpath, sizeof(winpath)))
+grub_util_error (cygwin_conv_path() failed);
+
+ int len = strlen (winpath);
+ if (len 2 winpath[1] == ':')
+{
+ len -= 2;
+ memmove (winpath, winpath + 2, len + 1);
+}
+
+ int i;
+ for (i = 0; i len; i++)
+if (winpath[i] == '\\')
+ winpath[i] = '/';
+ return xstrdup (winpath);
+}
+#endif
+
/* This function never prints trailing slashes (so that its output
can be appended a slash unconditionally). */
char *
@@ -521,28 +551,28 @@
/* offset == 1 means root directory. */
if (offset == 1)
{
- free (buf);
- len = strlen (buf2);
- while (buf2[len - 1] == '/' len 1)
- {
- buf2[len - 1] = '\0';
- len--;
- }
- if (len 1)
- return buf2;
- else
- {
- /* This means path given is just a backslash. As above
- we have to return an empty string. */
- free (buf2);
- return xstrdup ();
- }
+ /* Include leading slash. */
+ offset = 0;
+ break;
}
}
free (buf);
buf3 = xstrdup (buf2 + offset);
free (buf2);
+#ifdef __CYGWIN__
+ if (st.st_dev != (DEV_CYGDRIVE_MAJOR 16))
+{
+ /* Reached some mount point not below /cygdrive.
+ GRUB does not know Cygwin's emulated mounts,
+ convert to Win32 path. */
+ grub_util_info (Cygwin path = %s\n, buf3);
+ char * temp = get_win32_path (buf3);
+ free (buf3);
+ buf3 = temp;
+}
+#endif
+
len = strlen (buf3);
while (buf3[len - 1] == '/' len 1)
{
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
NEED_REGISTER_FRAME_INFO can be replaced by -static-libgcc
Hi,
a note regarding this fix:
2010-04-11 Vladimir Serbinenko ...
Fix cygwin compilation.
* configure.ac: Define NEED_REGISTER_FRAME_INFO.
* include/grub/misc.h (__register_frame_info)
[NEED_REGISTER_FRAME_INFO !UTIL]: New export.
(__deregister_frame_info) [NEED_REGISTER_FRAME_INFO !UTIL]:
Likewise.
* kern/misc.c (__register_frame_info)
[NEED_REGISTER_FRAME_INFO !UTIL]: New empty function.
(__deregister_frame_info) [NEED_REGISTER_FRAME_INFO !UTIL]:
Likewise.
--- configure.ac2010-04-11 14:14:51 +
+++ configure.ac2010-04-12 14:58:44 +
@@ -376,7 +376,11 @@
# For platforms where ELF is not the default link format.
AC_MSG_CHECKING([for command to convert module to ELF format])
case ${host_os} in
- cygwin) TARGET_OBJ2ELF='grub-pe2elf' ;;
+ cygwin) TARGET_OBJ2ELF='grub-pe2elf';
+# FIXME: put proper test here
+ AC_DEFINE([NEED_REGISTER_FRAME_INFO], 1,
+[Define to 1 if GCC generates calls to __register_frame_info()])
+ ;;
*) ;;
The *_frame_info symbols are set undefined to force linkage of the
libgcc_s shared library or dll.
This can be prevented by TARGET_LDFLAGS=-static-libgcc. To build from
grub-1.98 tarball on Cygwin, run configure with this argument.
May also be necessary on other build platforms when -shared-libgcc is
the default. It may be possible simply set -static-libgcc unconditionally.
--
Regards,
Christian Franke
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
[PATCH] Add missing include to fix build on Cygwin
This fixes a regression in grub-pe2elf which breaks build on Cygwin.
2010-03-03 Christian Frankefra...@computer.org
* util/grub-pe2elf.c: Add missing include progname.h.
This fixes build on Cygwin.
--
Regards,
Christian Franke
=== modified file 'util/grub-pe2elf.c'
--- util/grub-pe2elf.c 2010-01-16 00:26:52 +
+++ util/grub-pe2elf.c 2010-03-03 19:54:44 +
@@ -29,6 +29,8 @@
#include stdlib.h
#include getopt.h
+#include progname.h
+
static struct option options[] = {
{help, no_argument, 0, 'h'},
{version, no_argument, 0, 'V'},
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: On gratuitous modularization
Vladimir 'φ-coder/phcoder' Serbinenko wrote: Christian Franke wrote: Hi Robert, Robert Millan wrote: Please be careful when adding modules. I see that too often new modules are added without any real need to host this code separately. ... kern/disk.c:struct grub_disk_ata_pass_through_parms *); this seems unnecessary. ata_pthru is very small. If it's only used by hdparm, why not just merge it? This also avoids the additional code in kernel. The module ata_pthru.mod exists only to keep ata.mod small, see: http://lists.gnu.org/archive/html/grub-devel/2009-02/msg00091.html Keeping the ata.mod specific pass-through function separate from hdparm.mod was intentional. Merging this function into hdparm.mod would only make sense if ata.mod will the only ATA access module with pass-through functionality in the future. Hdparm.mod would then depend on ata.mod. A 'hdparm -h' would load ata.mod and disable biosdisk access. I hope we will eventually have an ahci.mod :-) Are the parameters of current ata_pass_through ATA-specific or would they be the same on AHCI? Like e.g. the Linux HDIO_DRIVE_TASKFILE ioctl, the ata_pass_through() parameters are not controller or transport specific. It should work with any possible future implementation of ATA pass-through functionality, e.g. - native AHCI driver - other native drivers - SAT (ANSI INCITS 431-2007) ATA pass through command for: -- SATA drives behind USB-bridges with SAT support, or: -- SATA drives in SAS enclosures Merging ata_pthru.mod into hdparm.mod would make hdparm specific to one (outdated) class of controllers: the traditional IDE-controller (T13/1510D). BTW: I agree that using a global function pointer 'grub_disk_ata_pass_through' is a hack. A cleaner design would be possible with a grub_disk_dev.ioctl(.) call. Such an ioctl() call would IMO be the best way to handle such device class specific extra functionality. -- Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: On gratuitous modularization
Hi Robert, Robert Millan wrote: Please be careful when adding modules. I see that too often new modules are added without any real need to host this code separately. There are many examples where this happened. I just noticed: commands/hdparm.c: struct grub_disk_ata_pass_through_parms apt; commands/hdparm.c: if (grub_disk_ata_pass_through (disk,apt)) commands/hdparm.c: struct grub_disk_ata_pass_through_parms apt; commands/hdparm.c: if (grub_disk_ata_pass_through (disk,apt)) commands/hdparm.c: struct grub_disk_ata_pass_through_parms apt; commands/hdparm.c: if (grub_disk_ata_pass_through (disk,apt)) commands/hdparm.c: if (! grub_disk_ata_pass_through) disk/ata_pthru.c: struct grub_disk_ata_pass_through_parms *parms) disk/ata_pthru.c: grub_disk_ata_pass_through = grub_ata_pass_through; disk/ata_pthru.c: if (grub_disk_ata_pass_through == grub_ata_pass_through) disk/ata_pthru.c:grub_disk_ata_pass_through = NULL; include/grub/disk.h:struct grub_disk_ata_pass_through_parms include/grub/disk.h:extern grub_err_t (* EXPORT_VAR(grub_disk_ata_pass_through)) (grub_disk_t, include/grub/disk.h: struct grub_disk_ata_pass_through_parms *); kern/disk.c:grub_err_t (* grub_disk_ata_pass_through) (grub_disk_t, kern/disk.c:struct grub_disk_ata_pass_through_parms *); this seems unnecessary. ata_pthru is very small. If it's only used by hdparm, why not just merge it? This also avoids the additional code in kernel. The module ata_pthru.mod exists only to keep ata.mod small, see: http://lists.gnu.org/archive/html/grub-devel/2009-02/msg00091.html Keeping the ata.mod specific pass-through function separate from hdparm.mod was intentional. Merging this function into hdparm.mod would only make sense if ata.mod will the only ATA access module with pass-through functionality in the future. Hdparm.mod would then depend on ata.mod. A 'hdparm -h' would load ata.mod and disable biosdisk access. I hope we will eventually have an ahci.mod :-) BTW: I agree that using a global function pointer 'grub_disk_ata_pass_through' is a hack. A cleaner design would be possible with a grub_disk_dev.ioctl(.) call. -- Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: Request: hdparm module with -X transfer mode
Nando wrote: It's great to have hdparm output added. Another fixup that would be needed would be setting the ATAPI cable mode to be 80pin, rather than the slower 40pin. Users who add hard drives in place of PATA optical drives are liked affected. Eg: http://forum.notebookreview.com/showthread.php?p=5519419 (HP 2510P / optical drive caddy) http://www.gottabemobile.com/forum/forum_posts.asp?TID=8260PID=55271 http://www.gottabemobile.com/forum/forum_posts.asp?TID=8260PID=55271 (Dell XT / not sure why) Comments/suggestions? AFAICS this is not possible: Cable detection can be done either by host sensing the PDIAG:CBLID signal, by relying on sensing info from the device, or a combination of both. This is rather complex such that false results may be not unlikely. The device reports the state of PDIAG:CBLID in IDENTIFY word 93. The original Linux hdparm prints this info as: HW reset results: CBLID- above/below Vih. (I will add this to GRUB hdparm). There is no standard ATA command to change what the device reports in word 93. The host OS driver may or may not use this info from the device to detect cable type. Reference: T13/1532D Vol 2 Rev 4b (ATA-7 Parallel Transport) section 8.2.11 and annex D. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: gettext: commands/hdparm.c
Jordi Mallach wrote: On Sun, Dec 06, 2009 at 06:25:33PM +, Carles Pina i Estany wrote: grub_printf (Model: \%.40s\\n, le16_to_char (tmp, idw[27], 40)); grub_printf (Firmware: \%.8s\\n, le16_to_char (tmp, idw[23], 8)); grub_printf (Serial: \%.20s\\n, le16_to_char (tmp, idw[10], 20)); To have a proper alignment it needs to code something (dynamic tab). I could change it to: grub_printf (_(Model:\t\%.40s\\n), le16_to_char (tmp, idw[27], 40)); grub_printf (_(Firmware:\t\%.8s\\n), le16_to_char (tmp, idw[23], 8)); grub_printf (_(Serial:\t\%.20s\\n), le16_to_char (tmp, idw[10], 20)); It's not unusual to see stuff like this, but again, a small percent of translators will actually try to figure out if the string is correctly aligned or how many \t's they need to use (by manually calculating how the string will look, by reading the source code, or using qemu). If the alignment is just a cosmetic enhancement, dropping it could be a reasonable option. Yes, it is only cosmetic - feel free to drop it. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: Request: hdparm module with -X transfer mode
Nando wrote: I was pleasantly surprised when I saw a hdparm module as part of grub2. So I tried the -X parameter as is provided by the linux command line finding it isn't implemented. eg: # hdparm -X udma5 /dev/sda Can the development team consider adding this feature? A number of ppl's bios is incorrectly setting interfaces to UDMA2/MDMA2, particularly on pata optical drive interfaces which users are using a 2.5 HDDs instead via an optical bay caddy. The workaround being to wait for a (slow) Linux/Windows boot and then typing the above commandline. Doing this at the grub2 bootloader level being more preferable. OK, added to my local TODO list :-) Please note that GRUB hdparm does only work in conjunction with the ata+ata_pthru modules because a PC BIOS does not provide any ATA pass-through functionality. A 'hdparm -X' would only help if the actual boot works with ata.mod instead of biosdisk.mod. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: Easy grub2 installation for non Linux (Windows) systems?
Nando wrote: For non Linux users, it is possible to load the grub2 package and modules on an ntfs filesystem? ... Yes, GRUB 1.97 builds and runs out-of-the-box on Cygwin. (Note: requires previous binutils-20080624-2 due to a regression in ld from binutils-2.19.51-1) 1.96 is available as a Cygwin package, 1.97 will be provided when binutils regression is fixed. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: hdparm documentation
Pedro A ARANDA wrote: I'm trying to make the hdparm module run on my HP Mini 110. I have had no luck, neither in doing so or finding documentation which could help me doing so. If I get the right pointer and I make it run on my box, I'd be willing to post a HOWTO, to make things easier to other people. I'm currently running Ubuntu 2009.10 with grub2. When I set debug=ata insmod ata I actually get my drive listed as (ata0,0) but when I then try insmod hdparm hdparm -B 255 (ata0,0) grub keeps telling me that the drive (hd0) is not known... After 'insmod ata', loading of further modules typically fails because the drive in the 'prefix' variable is no longer valid. The following works for me: # Load needed modules first insmod hdparm # Switch to ATA driver and load ATA pass-through support # (this invalidates prefix) insmod ata_pthru # Turn APM off hdparm -B 255 (ata0) -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] ntldr support
Robert Millan wrote: It probably would make sense that the 'ntldr' command does simple signature checks and fail on unknown files unless '--force' is specified. You mean checking for the PE signature? Yes, this would be nice too. A check of the first byte (jmp, 0xe9) and some file size range check (e.g. 0x3...0x4) may be enough for a first ntldr command. May also work for bootmgr.exe. EXE (MZ) and PE headers appear at larger offsets: ntldr from XP SP2: size 251184, EXE header at 0x4d30, PE at 0x4e00 ntldr from XP SP3: size 251712, EXE header at 0x4d40, PE at 0x4e10 bootmgr.exe from Vista: ??? grub4dos checks for ntldr as follows: - file starts with 0xe9, 0x??, 0x01, - first sector does not end with bootsector signature 0x55,0xaa, - file size exceeds 0x3. -- Regards Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] ntldr support
Vladimir 'phcoder' Serbinenko wrote: grub4dos checks for ntldr as follows: - file starts with 0xe9, 0x??, 0x01, - first sector does not end with bootsector signature 0x55,0xaa, - file size exceeds 0x3. For me it sounds like a heuristic. I would prefer to trust user rather than introducing heuristics to check file type. I agree that such a heuristic is not needed. Grub2 uses different commands for different loaders (which is good) and therefore there is no need to auto-detect the loader type. But a simple health check (like in 'chainloader') would IMO make sense for any load command. For 'ntldr': check that the file size is reasonable and that code starts with a jmp instruction. Allow to override the check with '--force'. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] ntldr support
Michal Suchanek wrote: I tried putting the ntldr binary on an empty fat32 disk and loading with the grub ntldr command. The system just reboots. This also happens when ntldr is loaded by the regular boot code in PBR+boot area. It is apparently the normal(?-) behavior when ntdetect.com is missing. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] ntldr support
Vladimir 'phcoder' Serbinenko wrote: About the command, i think that it will be simpler for the user if we have only one command: chainloader (like in grub4dos) that will try to detect the type of the bootloader. This is only my personal opinion. I don't agree with this. chainloader and ntldr don't share the same syntax: chainloader expects a bootsector whereas ntldr expects an ntldr ot bootmgr file. GRUB2 is done to break with bad design decisions of GRUB1 one of them being kernel command. GRUB4DOS follows GRUB1 on this subject. I agree. It probably would make sense that the 'ntldr' command does simple signature checks and fail on unknown files unless '--force' is specified. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: Solving the grub-pe2elf problem
Bean wrote: You could also check out my new object format at http://github.com/bean123/grub, lib branch. It supports conversion of pe/elf/mach-o object file to a grub specific format, so that you can use native build tool in Linux/OSX/Windows, no need for linker and converter. It also reduce compile time by half and generate module file much smaller that the current format. Tested successfully on Cygwin - Nice work! -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: Solving the grub-pe2elf problem
Robert Millan wrote: I thought of a possible solution to the grub-pe2elf problem. It seems that it is burdensome to produce ELF binaries on Windows, but building PE binaries or even PE/win32 executables on GNU/Linux is not (thanks to Mingw32 toolchain which is available on most distributions). If building GRUB with ./configure --host=i586-mingw32msvc works, we could include win32 binaries in the upcoming 1.97 release. Other GNU projects do this as well. Then we could drop support for toolchains that lack ELF. Which binaries would you like to include? - kernel.img and modules, and/or - the grub-*.exe files (Note: mingw32 .exe != cygwin .exe).? The problem with the PE-ELF conversion affects build of the *.mod files, not the build of the *.exe. The linker script is necessary for kernel.img. Christian and Bean, are you interested in implementing this? No. Then building a Cygwin package would rely on the existence of (tested or untested?) binaries in the a release tarball. Building packages from SVN snapshots would no longer be possible. I really don't want to spent my open-source-time with such kind of feature removal :-) -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: Solving the grub-pe2elf problem
Vladimir 'phcoder' Serbinenko wrote: If you see pe2elf as being too much of a burden we may switch to objconv: http://www.agner.org/optimize/. It's already used in conjunction with Apple Mach-O toolchain. It's not a GNU project but is licensed under GPL. Thanks for the hint, I will try objconv it soon. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] ntldr support
Robert Millan wrote: This patch implements a loader for NTLDR boot semantics (which are the same in BootMGR, hence both are supported). It still needs some cleanup in chainloader.c before it can be merged [1], but I submit it so that others can test it and report if it works for them. Tested with WinXP in a VM: Before the test, original boot code behind partition boot sector was destroyed. (dd if=/dev/zero of=/dev/hda1 bs=512 seek=1 count=20) Then this fails: root=(hd0,1) chainloader (hd0,1)+1 boot But with the patch this still works: root=(hd0,1) ntldr /ntldr boot Thanks. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: grub-pe2elf
Robert Millan wrote: On Mon, Aug 03, 2009 at 04:29:18PM -0400, Pavel Roskin wrote: Committing code against the wish of maintainers should not be permitted. Those who do it should lose commit access. They are welcome to contribute, but their patches will need to be committed by others. I agree... I haven't seen any apology or explanation in this thread. ... but I will not apologize simply because I did not commit the grub-pe2elf patch. Yes, this is what worries me the most. I'm somewhat worried that these complaints are delayed by more than a year Commit 1726 is from 2008-07-24, start was here: http://lists.gnu.org/archive/html/grub-devel/2008-07/msg00452.html -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] ntldr support
Vladimir 'phcoder' Serbinenko wrote: On Tue, Aug 4, 2009 at 9:27 PM, Robert Millan... wrote: After thinking a bit about this, I don't think we want this command in its current form. The problem is it is misleading. It leads the user to think it can load ntldr as a standalone file, but in fact it is reading the PBR behind the scenes. I don't think it's of any problem since ntldr uses this PBR only as superblock to identify the partition. As such I would rather consider this loading as a special case of passing $root, just the form of it is a bit weird Agree. A 'ntldr' or 'chainloader --ntldr' command is not mandatory. But it is 'nice to have' because it allows to boot even if the boot code (6 sectors) in the area behind the PBR is not present for whatever reason. See my previsions mail with the test case. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: grub-pe2elf
Robert Millan wrote: On Sat, Aug 01, 2009 at 05:06:46PM +0200, Christian Franke wrote: Since ~r1584, util/grub-* could be build and run on Cygwin. Commit r1726 added support to build kernel and modules on Cygwin. There has to be more to it. I know you were using GRUB on Cygwin before that commit. Was this kludge being used since the beginning? I'm using GRUB on Cygwin since 'somewhat' before I announced the first patch 2007-10-16: http://lists.gnu.org/archive/html/grub-devel/2007-10/msg00071.html This version used the hack in grub's ELF-loader already discussed. It also contained some non Cygwin related fixes found during testing. The grub2 package in the Cygwin distro from 2008-03-30 was based on this private branch merged to 1.96+epsilon. What are the other options? I don't know. Waiting until someone fixes BFD or maintains a Linux cross-compiler package for Cygwin (which cannot be build OOTB, it requires build steps on Linux) may last until time_t wraps around and is IMO not an option :-) Even the isolation of grub-pe2elf in a separate package would not help much, because the linker script and some support in the build system would be still needed in the grub2 tree. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: grub-pe2elf
Robert Millan wrote: On Fri, Jul 31, 2009 at 08:27:58PM +0200, Christian Franke wrote: First of all, I'm worried that we have discussions to decide things and they are later forgotten or ignored. Bean, please can you provide some explanation on what happened? Also, I think this commit should be reverted (at least the part that adds grub-pe2elf). I would regret this. Then you could also revert all my Cygwin related additions (some also included in Bean's r1726) because these no longer make much sense unless someone fixes BFD or maintains a ELF toolchain for Cygwin. But we supported Cygwin before grub-pe2elf. What has changed? Since ~r1584, util/grub-* could be build and run on Cygwin. Commit r1726 added support to build kernel and modules on Cygwin. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: grub-pe2elf
Hi Robert, Robert Millan wrote: On a recent discussion on IRC, Bean pointed out that grub-pe2elf is essentially a workaround for a bug in objcopy. It is essentially a limitation of BFD: The object format specific backends are apparently not implemented such that conversion between different relocation formats is supported. In particular, a relocation 'addend' (-4 in this case) is lost during conversion if source or target relocation format do not support it as separate field. Some conversions may work. PE - ELF does not work, the problem is in both backends in this case. Bean explained that this bug is quite obvious. It seems Christian sent them a patch, but it hasn't been applied (I don't know if it was rejected, or just ignored, Christian could you explain?). The patch was actually an ad-hoc approach to fix it in objcopy itself for the PE-ELF case only: http://sourceware.org/ml/binutils/2007-10/msg00302.html A generic solution which would have more chances to get accepted would require substantial work in BFD itself. Conversion between relocation formats is apparently rarely needed. A related bugzilla entry was reopened in 2005 and is idle since then: http://sourceware.org/bugzilla/show_bug.cgi?id=970 Later in July 2008, Bean added grub-pe2elf to the repository. Yes, and I really appreciated this :-) First of all, I'm worried that we have discussions to decide things and they are later forgotten or ignored. Bean, please can you provide some explanation on what happened? Also, I think this commit should be reverted (at least the part that adds grub-pe2elf). I would regret this. Then you could also revert all my Cygwin related additions (some also included in Bean's r1726) because these no longer make much sense unless someone fixes BFD or maintains a ELF toolchain for Cygwin. Nevertheless, it's not an obligation for us to support win32. As long as we can do it sanely, I'm fine with it, but I don't want to see GRUB dragged into a situation in which we need to preserve win32 support at all costs. $ find grub2 -type f -name '*.[ch]' | xargs cat | wc -l 128175 $ cat grub2/util/grub-pe2elf.c | wc -l 521 0.4% of the total LoC is IMO less than 'all costs' :-) -- Regards Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [BUGFIX] Don't use DT_DIR: It doesn't work on non-ext* filesystems
Pavel Roskin wrote: On Thu, 2009-07-23 at 11:29 +0200, Vladimir 'phcoder' Serbinenko wrote: -#ifdef DT_DIR - info.dir = (de-d_type == DT_DIR); -#else info.dir = !! is_dir (path, de-d_name); -#endif Fine with me. Finally a patch that reduces the number of preprocessor directives :-) :-) Grub hostfs has no performance requirement like e.g. find. The optimization with d_type can simply be ignored - patch is OK. The idea behind d_type is that it should only be set != DT_UNKNOWN if the type info is available for free or at low cost. It is safe for a filesystem implementation to return DT_UNKNOWN always. A correct performance-aware solution would look like: #ifdef DT_DIR if (de-d_type == DT_DIR) info.dir = 1; else if (de-type == DT_FILE) info.dir = 0; else #endif info.dir = !! is_dir (path, de-d_name); This was actually wrong in my original patch (svn 1353) that was necessary because of missing d_type on Cygwin - Sorry! Meantime I added d_type support to Cygwin itself and therefore learned how this is supposed to work :-) -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [BUGFIX] Don't use DT_DIR: It doesn't work on non-ext* filesystems
Pavel Roskin wrote: On Fri, 2009-07-24 at 23:02 +0200, Christian Franke wrote: A correct performance-aware solution would look like: #ifdef DT_DIR if (de-d_type == DT_DIR) info.dir = 1; else if (de-type == DT_FILE) There in no DT_FILE in glibc, but there is DT_REG. Yes correct. DT_UNKNOWN is present. Perhaps the above line should be else if (de-type != DT_UNKNOWN) We only care if it's a directory or not. All other objects can be treated like files. It depends: If DT_UNKNOWN is used, you have to replace stat() by lstat() in is_dir(). Otherwise symlinked directories would be appear as dir only if d_type == DT_UNKNOWN and not if d_type == DT_LNK. I'm fine either way, whether we fix the high-performance code or remove it, as long as we don't have to add more checks. I would suggest to remove it. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
[PATCH] add a snapshot version string to GRUB title
Many GNU GRUB version 1.96 used in production are likely builds from
more recent SVN snapshots. It would (IMO) be useful to have some info
about the actual installed version.
This patch adds a script to generate some version info from ChangeLog
and (if available) from SVN.
The version syntax is PACKAGE_VERSION+DATE-N-rSVN where DATE is the
most recent date in ChangeLog and 'N' is the number of the entries for
this date. If the last ChangeLog entry suggests that this is an official
release, only PACKAGE_VERSION is used.
The current string is 1.96+20090702-1-r2391 or 1.96+20090702-1 if
'svn info' is not available.
The string is added to grub title only, utils are left out for now.
Regards,
Christian Franke
2009-07-02 Christian Franke fra...@computer.org
Auto-generate a snapshot version string for GRUB title.
* Makefile.in: Add targets for genversionstr.sh.
Add target and dependency for grub_version.h.
* configure.ac: Add genversionstr.sh to AC_CONFIG_FILES.
* genversionstr.sh.in: New script. Generates GRUB_VERSION
string from ChangeLog.
* normal/main.c: Add include grub_version.h.
(grub_normal_init_page): Replace PACKAGE_VERSION by
GRUB_VERSION in TITLE.
diff --git a/Makefile.in b/Makefile.in
index 3d208e7..6b56c83 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -419,6 +419,16 @@ gensymlist.sh: gensymlist.sh.in config.status
genkernsyms.sh: genkernsyms.sh.in config.status
$(SHELL) ./config.status
+genversionstr.sh: genversionstr.sh.in config.status
+ $(SHELL) ./config.status
+
+# Build version string
+grub_version.h: genversionstr.sh ChangeLog
+ $(SHELL) ./genversionstr.sh $@ || (rm -f $@ ; exit 1)
+
+normal_mod-normal_main.o: grub_version.h
+
+
.PHONY: all install install-strip uninstall clean mostlyclean distclean
.PHONY: maintainer-clean info dvi dist check
diff --git a/configure.ac b/configure.ac
index 8b12c58..890cf3d 100644
--- a/configure.ac
+++ b/configure.ac
@@ -529,6 +529,6 @@ else
rm -rf include/grub/machine
cp -rp $srcdir/include/grub/$target_cpu/$platform include/grub/machine 2/dev/null
fi
-AC_CONFIG_FILES([Makefile gensymlist.sh genkernsyms.sh])
+AC_CONFIG_FILES([Makefile gensymlist.sh genkernsyms.sh genversionstr.sh])
AC_CONFIG_FILES([stamp-h], [echo timestamp stamp-h])
AC_OUTPUT
diff --git a/genversionstr.sh.in b/genversionstr.sh.in
new file mode 100644
index 000..6e58fb3
--- /dev/null
+++ b/genversionstr.sh.in
@@ -0,0 +1,77 @@
+#! /bin/sh
+#
+# Copyright (C) 2009 Free Software Foundation, Inc.
+#
+# This genversionstr.sh.in is free software; the author
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY, to the extent permitted by law; without
+# even the implied warranty of MERCHANTABILITY or FITNESS FOR A
+# PARTICULAR PURPOSE.
+
+### The configure script will replace these variables.
+
+srcd...@srcdir@
+
+
+cat EOF
+/* This file is automatically generated by genversionstr.sh. DO NOT EDIT! */
+/*
+ * GRUB -- GRand Unified Bootloader
+ * Copyright (C) 2009 Free Software Foundation, Inc.
+ *
+ * GRUB is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * GRUB is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with GRUB. If not, see http://www.gnu.org/licenses/.
+ */
+
+EOF
+
+
+if sed -n '2,/^20/p' $srcdir/ChangeLog \
+ | grep 'configure.ac (AC_INIT): Bumped' /dev/null 2/dev/null
+then
+ # Last entry documents new PACKAGE_VERSION, assume official release
+
+ cat EOF
+#ifndef GRUB_VERSION
+ #define GRUB_VERSION PACKAGE_VERSION
+#endif
+EOF
+
+else
+ # Extract most recent date string from ChangeLog.
+ # Append number of log entries for this date.
+ datestr=`
+sed -n 's,^\(20[0-9][0-9]-[01][0-9]-[0-3][0-9]\) .*$,\1,p;1000q' $srcdir/ChangeLog \
+| uniq -c \
+| sed -n 's,^ *\([0-9][0-9]*\) \(20[0-9][0-9]\)-\([01][0-9]\)-\([0-3][0-9]\)$,+\2\3\4-\1,p;1q'
+ `
+
+ # get SVN revision if possible
+ svnstr=
+ if [ -d $srcdir/.svn ] ; then
+svnstr=`
+ ( cd $srcdir svn info ) 2 /dev/null \
+ | sed -n 's,^Revision: *\(.*\)$,-r\1,p'
+`
+ fi
+
+ cat EOF
+#ifndef GRUB_VERSION
+ #define GRUB_VERSION PACKAGE_VERSION ${datestr}${svnstr}
+#endif
+EOF
+fi
+
diff --git a/normal/main.c b/normal/main.c
index 7f6336e..a412827 100644
--- a/normal/main.c
+++ b/normal/main.c
@@ -29,6 +29,8 @@
#include grub
Re: [2342] 2009-06-18 Vladimir Serbinenko phco...@gmail.com
Pavel Roskin wrote: On Fri, 2009-06-19 at 22:05 +0200, Christian Franke wrote: This was the AC_MSG_RESULT(...) for the AC_MSG_CHECKING([for option to link raw image]) which is still present for the $grub_cv_apple_target_cc != yes case. I would suggest to move the AC_MSG_CHECKING outside the 'if $grub_cv_apple_target_cc = yes ... fi' and re-add the AC_MSG_RESULT. Thanks for pointing it out. Actually, the message is not needed at all. There is no checking that needs to be communicated to the user. It would be appropriate if configure was running a compiler or looked for some tool in the PATH. In this case, it's just shell variable assignments with on test for a file. I would prefer an output here even if the test done isn't a 'real' configure check. Would also make sense when we later apply your 'Use common linker script for all i386-pc systems' suggestion. But if you want to remove this messages it is also fine with me. Same would then possibly also apply to the AC_MSG_CHECKING([for command to convert module to ELF format]) a few lines below. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [2342] 2009-06-18 Vladimir Serbinenko phco...@gmail.com
Vladimir Serbinenko wrote: * configure.ac: remove a leftover AC_MSG_RESULT ... --- trunk/grub2/configure.ac2009-06-18 13:51:06 UTC (rev 2341) +++ trunk/grub2/configure.ac2009-06-18 14:02:23 UTC (rev 2342) @@ -308,7 +308,6 @@ AC_SUBST(TARGET_IMG_LDSCRIPT) AC_SUBST(TARGET_IMG_LDFLAGS) AC_SUBST(TARGET_IMG_CFLAGS) -AC_MSG_RESULT([$TARGET_IMG_LDFLAGS_AC]) This was the AC_MSG_RESULT(...) for the AC_MSG_CHECKING([for option to link raw image]) which is still present for the $grub_cv_apple_target_cc != yes case. I would suggest to move the AC_MSG_CHECKING outside the 'if $grub_cv_apple_target_cc = yes ... fi' and re-add the AC_MSG_RESULT. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [2290] 2009-06-10 Pavel Roskin pro...@gnu.org
Pavel Roskin wrote: Revision: 2290 http://svn.sv.gnu.org/viewvc/?view=revroot=grubrevision=2290 Author: proski Date: 2009-06-10 18:32:13 + (Wed, 10 Jun 2009) Log Message: --- 2009-06-10 Pavel Roskin pro...@gnu.org * configure.ac: Use -nostdlib when probing for the target. It should not be required to have libc for the target. ... # Set them to their new values for the tests below. CC=$TARGET_CC -CFLAGS=$TARGET_CFLAGS +CFLAGS=$TARGET_CFLAGS -nostdlib CPPFLAGS=$TARGET_CPPFLAGS LDFLAGS=$TARGET_LDFLAGS This change breaks build at least on Cygwin, ./configure fails with: checking whether target compiler is working... no configure: error: cannot compile for the target The link test in grub_PROG_TARGET_CC fails due to missing '__main' symbol from libgcc. According to gcc 4.3.2 man page, '-nostdlib' usually requires '-lgcc'. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [2290] 2009-06-10 Pavel Roskin pro...@gnu.org
Vladimir 'phcoder' Serbinenko wrote: The link test in grub_PROG_TARGET_CC fails due to missing '__main' symbol from libgcc. According to gcc 4.3.2 man page, '-nostdlib' usually requires '-lgcc'. Try compiling but not linking using -c option. If it works you can commit it since we never do complete linking for target Complete linking is not used during build but during configure. The following tests rely on it: grub_PROG_LD_BUILD_ID_NONE grub_CHECK_BSS_START_SYMBOL grub_CHECK_END_SYMBOL I would suggest to undo commit 2290 for now. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [2290] 2009-06-10 Pavel Roskin pro...@gnu.org
Pavel Roskin wrote: We can use -Wl,--defsym,___main=0x8100 as it's done elsewhere in configure.ac. Let me just check it doesn't break anything. This (r2310) works on Cygwin, thanks. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: status grub2 port of grub-legasy map command
Vladimir 'phcoder' Serbinenko wrote: Do not do this. Some BIOS functions (like ah=08h) return data in dl. Clients should not expect data in registers to be preserved across interrupt calls. I don't know if there is something like a 8086/PC-BIOS ABI document that we can find to confirm the non-preservation of dl, but the FreeDOS MBR should be fixed then. Thank you for pointing ah=8 function AFAIK there is no normative reference. The source which was mainly used for long years is helppc and it states following: - registers DS, BX, CX and DX are preserved (http://heim.ifi.uio.no/~stanisls/helppc/int_13.html) But now this reference is terribly outdated. SeaBIOS preserves %dl too T13 EDD provides a probably more up to date documentation of int13. There was no requirement to preserve registers is in EDD (2000) and EDD-2 (2002). It appeared in EDD-3 (2004) and is still in first EDD-4 draft (2009): The values in all registers that are not explicitly defined in the following sections shall be preserved at the completion of each function call From Section 8 of: BIOS Enhanced Disk Drive Services - 3 (T13/1572D Revision 3) Fortunately, T13 docs are (unlike T10 and SATA-IO) still publicly available: http://www.t13.org/Documents/MinutesDefault.aspx?DocumentType=4DocumentStage=2 -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] ata.mod under qemu
Pavel Roskin wrote: On Sat, 2009-05-23 at 23:04 +0200, Christian Franke wrote: New patch attached. Please test on the above hardware if possible. It's working fine. (ata6) and (ata7) appear, but (ata0) and (ata1) doesn't. When inserting ata.mod, I get: error: error reading ATA IDENTIFY data error: error reading ATA IDENTIFY data Looks like the (ata0/1) controller returns !BSY, DRQ, !ERR in the status register if no device is connected. This is now detected after the bogus IDENTIFY data is read. That must be the ghost devices. While it would be nice to get rid of such messages, I'm fine if your patch is committed as is. Modified version committed, messages should no longer appear. Thanks! -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] ata.mod under qemu
Christian Franke wrote:
Pavel Roskin wrote:
It helps detect CD-ROM under qemu. But on the real hardware, it
introduces ghost drives.
Without the patch, I have (ata7) for the SATA hard drive. With the
patch, I have (ata7) for the SATA hard drive, (ata6) for the SATA DVD-RW
and two bogus unreadable drives (ata0) and (ata1).
New patch attached. Please test on the above hardware if possible.
--
Regards,
Christian Franke
diff --git a/disk/ata.c b/disk/ata.c
index ea42d59..5fa0ef5 100644
--- a/disk/ata.c
+++ b/disk/ata.c
@@ -41,11 +41,14 @@ grub_ata_wait_not_busy (struct grub_ata_device *dev, int milliseconds)
grub_millisleep (1);
int i = 1;
- while (grub_ata_regget (dev, GRUB_ATA_REG_STATUS) GRUB_ATA_STATUS_BUSY)
+ grub_uint8_t sts;
+ while ((sts = grub_ata_regget (dev, GRUB_ATA_REG_STATUS))
+ GRUB_ATA_STATUS_BUSY)
{
if (i = milliseconds)
{
- grub_dprintf (ata, timeout: %dms\n, milliseconds);
+ grub_dprintf (ata, timeout: %dms, status=0x%x\n,
+ milliseconds, sts);
return grub_error (GRUB_ERR_TIMEOUT, ATA timeout);
}
@@ -151,6 +154,7 @@ grub_atapi_identify (struct grub_ata_device *dev)
return grub_errno;
grub_ata_regset (dev, GRUB_ATA_REG_DISK, 0xE0 | dev-device 4);
+ grub_ata_wait ();
if (grub_ata_check_ready (dev))
{
grub_free (info);
@@ -248,6 +252,7 @@ grub_ata_identify (struct grub_ata_device *dev)
info16 = (grub_uint16_t *) info;
grub_ata_regset (dev, GRUB_ATA_REG_DISK, 0xE0 | dev-device 4);
+ grub_ata_wait ();
if (grub_ata_check_ready (dev))
{
grub_free (info);
@@ -259,24 +264,38 @@ grub_ata_identify (struct grub_ata_device *dev)
if (grub_ata_wait_drq (dev, 0, GRUB_ATA_TOUT_STD))
{
- if (grub_ata_regget (dev, GRUB_ATA_REG_ERROR) 0x04) /* ABRT */
- {
- /* Device without ATA IDENTIFY, try ATAPI. */
- grub_free(info);
- grub_errno = GRUB_ERR_NONE;
- return grub_atapi_identify (dev);
- }
+ grub_free(info);
+ grub_errno = GRUB_ERR_NONE;
+ grub_uint8_t sts = grub_ata_regget (dev, GRUB_ATA_REG_STATUS);
+
+ if ((sts (GRUB_ATA_STATUS_BUSY | GRUB_ATA_STATUS_DRQ
+ | GRUB_ATA_STATUS_ERR)) == GRUB_ATA_STATUS_ERR
+ (grub_ata_regget (dev, GRUB_ATA_REG_ERROR) 0x04 /* ABRT */))
+ /* Device without ATA IDENTIFY, try ATAPI. */
+ return grub_atapi_identify (dev);
+
+ else if (sts == 0x00)
+ /* No device, return error but don't print message. */
+ return GRUB_ERR_UNKNOWN_DEVICE;
+
else
- {
- /* Error. */
- grub_free(info);
- return grub_error (GRUB_ERR_UNKNOWN_DEVICE,
- device can not be identified);
- }
+ /* Other Error. */
+ return grub_error (GRUB_ERR_UNKNOWN_DEVICE,
+ device can not be identified);
}
grub_ata_pio_read (dev, info, GRUB_DISK_SECTOR_SIZE);
+ /* Re-check status to avoid bogus identify data due to stuck DRQ. */
+ grub_uint8_t sts = grub_ata_regget (dev, GRUB_ATA_REG_STATUS);
+ if (sts (GRUB_ATA_STATUS_BUSY | GRUB_ATA_STATUS_DRQ | GRUB_ATA_STATUS_ERR))
+{
+ grub_dprintf(ata, bad status=0x%x, sts);
+ grub_free(info);
+ return grub_error (GRUB_ERR_UNKNOWN_DEVICE,
+ error reading ATA IDENTIFY data);
+}
+
/* Now it is certain that this is not an ATAPI device. */
dev-atapi = 0;
@@ -334,26 +353,12 @@ grub_ata_device_initialize (int port, int device, int addr, int addr2)
grub_ata_regset (dev, GRUB_ATA_REG_DISK, dev-device 4);
grub_ata_wait ();
- /* If status is 0x00, it is safe to assume that there
- is no device (or only a !READY) device connected. */
- grub_int8_t sts = grub_ata_regget (dev, GRUB_ATA_REG_STATUS);
- grub_dprintf (ata, status=0x%x\n, sts);
- if (sts == 0x00)
-{
- grub_free(dev);
- return 0;
-}
-
/* Try to detect if the port is in use by writing to it,
waiting for a while and reading it again. If the value
- was preserved, there is a device connected.
- But this tests often detects a second (slave) device
- connected to a SATA controller which supports only one
- (master) device. In this case, the status register
- check above usually works. */
+ was preserved, there is a device connected. */
grub_ata_regset (dev, GRUB_ATA_REG_SECTORS, 0x5A);
grub_ata_wait ();
- grub_int8_t sec = grub_ata_regget (dev, GRUB_ATA_REG_SECTORS);
+ grub_uint8_t sec = grub_ata_regget (dev, GRUB_ATA_REG_SECTORS);
grub_dprintf (ata, sectors=0x%x\n, sec);
if (sec != 0x5A)
{
@@ -361,6 +366,12 @@ grub_ata_device_initialize (int port, int device, int addr, int addr2)
return 0;
}
+ /* The above test may detect a second (slave) device
+ connected to a SATA controller which supports only one
+ (master) device. It is not safe to use the status register
+ READY bit to check for controller channel existence. Some
+ ATAPI commands (RESET, DIAGNOSTIC) may clear this bit. */
+
/* Use the IDENTIFY DEVICE command to query the device. */
if (grub_ata_identify
Re: [PATCH] ata.mod under qemu
Pavel Roskin wrote: On Wed, 2009-05-20 at 23:40 +0200, Christian Franke wrote: The attached patch should fix this. It should also prevents misleading error messages if a device does not exist. It helps detect CD-ROM under qemu. But on the real hardware, it introduces ghost drives. Without the patch, I have (ata7) for the SATA hard drive. With the patch, I have (ata7) for the SATA hard drive, (ata6) for the SATA DVD-RW and two bogus unreadable drives (ata0) and (ata1). Thanks for this test. I could not reproduce this yet. I presume that the controller behind (ata0) and (ata1) is the chipset's PATA controller. If no device is connected, these controllers often echo a write to (e.g. command) register address back to the next read of (e.g. status) register address if the time between is too short. (No VM emulates this behavior :-) Could you possibly repeat this test with 'debug=ata' enabled? If this removes the bogus drives, it is likely a problem of too short wait times. I will prepare a new patch with longer timeouts and more thorough register tests during identify. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] ata.mod under qemu
The attached patch should fix this. It should also prevents misleading
error messages if a device does not exist.
2009-05-20 Christian Franke fra...@computer.org
* disk/ata.c: (grub_ata_wait_not_busy): Add debug output of status
register.
(grub_ata_identify): Suppress error message if status register
is 0x00 after command failure.
(grub_device_initialize): Remove unsafe status register check.
Thanks to 'phcoder' for problem report and patch.
Prevent sign extension in debug message.
--
Regards,
Christian Franke
diff --git a/disk/ata.c b/disk/ata.c
index ea42d59..651937d 100644
--- a/disk/ata.c
+++ b/disk/ata.c
@@ -41,11 +41,14 @@ grub_ata_wait_not_busy (struct grub_ata_device *dev, int milliseconds)
grub_millisleep (1);
int i = 1;
- while (grub_ata_regget (dev, GRUB_ATA_REG_STATUS) GRUB_ATA_STATUS_BUSY)
+ grub_uint8_t sts;
+ while ((sts = grub_ata_regget (dev, GRUB_ATA_REG_STATUS))
+ GRUB_ATA_STATUS_BUSY)
{
if (i = milliseconds)
{
- grub_dprintf (ata, timeout: %dms\n, milliseconds);
+ grub_dprintf (ata, timeout: %dms, status=0x%x\n,
+ milliseconds, sts);
return grub_error (GRUB_ERR_TIMEOUT, ATA timeout);
}
@@ -259,20 +262,21 @@ grub_ata_identify (struct grub_ata_device *dev)
if (grub_ata_wait_drq (dev, 0, GRUB_ATA_TOUT_STD))
{
+ grub_free(info);
+ grub_errno = GRUB_ERR_NONE;
+
if (grub_ata_regget (dev, GRUB_ATA_REG_ERROR) 0x04) /* ABRT */
- {
- /* Device without ATA IDENTIFY, try ATAPI. */
- grub_free(info);
- grub_errno = GRUB_ERR_NONE;
- return grub_atapi_identify (dev);
- }
+ /* Device without ATA IDENTIFY, try ATAPI. */
+ return grub_atapi_identify (dev);
+
+ else if (grub_ata_regget (dev, GRUB_ATA_REG_STATUS) == 0x00)
+ /* No device, return error but don't print message. */
+ return GRUB_ERR_UNKNOWN_DEVICE;
+
else
- {
- /* Error. */
- grub_free(info);
- return grub_error (GRUB_ERR_UNKNOWN_DEVICE,
- device can not be identified);
- }
+ /* Other Error. */
+ return grub_error (GRUB_ERR_UNKNOWN_DEVICE,
+ device can not be identified);
}
grub_ata_pio_read (dev, info, GRUB_DISK_SECTOR_SIZE);
@@ -334,26 +338,12 @@ grub_ata_device_initialize (int port, int device, int addr, int addr2)
grub_ata_regset (dev, GRUB_ATA_REG_DISK, dev-device 4);
grub_ata_wait ();
- /* If status is 0x00, it is safe to assume that there
- is no device (or only a !READY) device connected. */
- grub_int8_t sts = grub_ata_regget (dev, GRUB_ATA_REG_STATUS);
- grub_dprintf (ata, status=0x%x\n, sts);
- if (sts == 0x00)
-{
- grub_free(dev);
- return 0;
-}
-
/* Try to detect if the port is in use by writing to it,
waiting for a while and reading it again. If the value
- was preserved, there is a device connected.
- But this tests often detects a second (slave) device
- connected to a SATA controller which supports only one
- (master) device. In this case, the status register
- check above usually works. */
+ was preserved, there is a device connected. */
grub_ata_regset (dev, GRUB_ATA_REG_SECTORS, 0x5A);
grub_ata_wait ();
- grub_int8_t sec = grub_ata_regget (dev, GRUB_ATA_REG_SECTORS);
+ grub_uint8_t sec = grub_ata_regget (dev, GRUB_ATA_REG_SECTORS);
grub_dprintf (ata, sectors=0x%x\n, sec);
if (sec != 0x5A)
{
@@ -361,6 +351,12 @@ grub_ata_device_initialize (int port, int device, int addr, int addr2)
return 0;
}
+ /* The above test may detect a second (slave) device
+ connected to a SATA controller which supports only one
+ (master) device. It is not safe to use the status register
+ READY bit to check for controller channel existence. Some
+ ATAPI commands (RESET, DIAGNOSTIC) may clear this bit. */
+
/* Use the IDENTIFY DEVICE command to query the device. */
if (grub_ata_identify (dev))
{
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] ata.mod under qemu
Pavel Roskin wrote: On Mon, 2009-05-18 at 19:36 -0400, Pavel Roskin wrote: I did a quick check, and your patch has no effect on my system. I'm using Fedora 11 with qemu 0.10. After inserting the ata module, ls shows no disks, with or without your patch. Never mind, my quick check was too quick. I forgot to change prefix, so it's just ls that was failing. Your patch makes the CD-ROM visible. Without your patch, only hard drivers are seen. Thanks for the problem report. According to 'qemu-0.10.4/hw/ide.c', the (READY bit of the) status register is actually cleared by some commands like RESET and DIAGNOSE if an ATAPI device is connected. This conforms to ATA parallel transport standard. When I added the '(sts == 0x00)' check, I made the false assumption that device detection through the READY bit is safe. It was added to speed-up the detection of the missing slave port of some SATA controllers working in IDE-mode. The check can be safely removed, the only drawback is an extra ~1s timeout for each SATA controller. I will commit a fix soon. -- Regards, Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Use common linker script for all i386-pc systems
Vladimir 'phcoder' Serbinenko wrote:
Hello.
On Sun, May 17, 2009 at 7:33 AM, Pavel Roskin ... wrote:
This allows us to remove checks for the linker symbols for the beginning
and the end of the .bss section. Instead, we use the names from the
linker script. Another benefit is a better unification of the build
system.
Good idea. Patch works on Cygwin. Does it work on Linux, FreeBSD, ... ?
The variable TARGET_IMG_LDFLAGS_AC can also be removed, see attached patch.
The linker script is actually identical to the script from my first
Cygwin patch. It can be further simplified if desired.
@@ -0,0 +1,53 @@
+/* Linker script to create grub .img files on Cygwin. */
This comment is out of sync
VARIABLE(grub_end_addr)
- .long END_SYMBOL
+ .long __bss_end__
This variable isn't used at all. In my Apple's CC patch I just remove it
'grub_start_addr' and '.globl start, _start' are also not used.
--
Christian Franke
diff --git a/configure.ac b/configure.ac
index a048828..e799a79 100644
--- a/configure.ac
+++ b/configure.ac
@@ -212,15 +212,13 @@ AC_MSG_CHECKING([for option to link raw image])
if test -f ${srcdir}/conf/${target_cpu}-${platform}.lds; then
TARGET_IMG_LDSCRIPT='$(top_srcdir)'/conf/${target_cpu}-${platform}.lds
TARGET_IMG_LDFLAGS=-Wl,-T${TARGET_IMG_LDSCRIPT}
- TARGET_IMG_LDFLAGS_AC=-Wl,-T${srcdir}/conf/${target_cpu}-${platform}.lds
else
TARGET_IMG_LDSCRIPT=
TARGET_IMG_LDFLAGS='-Wl,-N'
- TARGET_IMG_LDFLAGS_AC='-Wl,-N'
fi
AC_SUBST(TARGET_IMG_LDSCRIPT)
AC_SUBST(TARGET_IMG_LDFLAGS)
-AC_MSG_RESULT([$TARGET_IMG_LDFLAGS_AC])
+AC_MSG_RESULT([$TARGET_IMG_LDFLAGS])
# For platforms where ELF is not the default link format.
AC_MSG_CHECKING([for command to convert module to ELF format])
@@ -379,11 +377,6 @@ grub_PROG_OBJCOPY_ABSOLUTE
grub_PROG_LD_BUILD_ID_NONE
grub_ASM_USCORE
if test x$target_cpu = xi386; then
- if test ! -z $TARGET_IMG_LDSCRIPT; then
-# Check symbols provided by linker script.
-CFLAGS=$TARGET_CFLAGS -nostdlib $TARGET_IMG_LDFLAGS_AC
-Wl,-Ttext,8000,--defsym,___main=0x8100
- fi
- CFLAGS=$TARGET_CFLAGS
grub_I386_ASM_PREFIX_REQUIREMENT
grub_I386_ASM_ADDR32
grub_I386_ASM_ABSOLUTE_WITHOUT_ASTERISK
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [2216] Remove incorrect comment that the code must be position independent.
Pavel Roskin wrote: Revision: 2216 http://svn.sv.gnu.org/viewvc/?view=revroot=grubrevision=2216 Author: proski Date: 2009-05-15 14:47:44 + (Fri, 15 May 2009) Log Message: --- Remove incorrect comment that the code must be position independent. Modified Paths: -- trunk/grub2/commands/i386/pc/drivemap_int13h.S Function drivemap.c:install_int13_handler() copies the grub_drivemap_handler code to some low memory allocated by grub_mmap_malign_and_register(). No relocation is done. IMO the code must be position independent. -- Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: Eliminating nested functions
Pavel Roskin wrote: I suggest that we eliminate all nested functions. The reasons are: 1) They make the code less readable, as they make the parent functions longer. 2) They have problems with some popular compilers, as recent as gcc-4.0 when regparm(3) is used. 3) We failed to implement a reliable test for such problems. We are using regparm(1) for all compilers. 4) The existing test is one of the obstacles making it impossible to compile without having libc for the target (x86_64-i386 would be really nice), as we need to run the compiled test executable. 5) Non-i386 architectures define NESTED_FUNC_ATTR as an empty symbol, so developers on such architectures don't see if they use it correctly. 6) NESTED_FUNC_ATTR tends to proliferate to the file scope functions, as it happened with grub_pci_iterate(). It only takes one caller using a nested function to force NESTED_FUNC_ATTR on all functions used as an argument to the same function. 7) The grub utils may not run on some 'security hardened' platform which does not allow to enable stack execution. -- Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: precision formatting in grub_printf
Deepak Vankadaru wrote:
Hi,
There was a redundant piece of code in the diff file I sent earlier. I
have removed it and prepared another patch file. I am attaching the
same. Please use this one instead.
Following is the changelog.
Changelog:
2008-08-17 Deepak Vankadaru deepak.vankad...@gmail.com
mailto:deepak.vankad...@gmail.com
Support for precision formatting in grub_printf
* kern/misc.c: modified grub_vsprintf to parse and handle
precision formatting
Thanks.
Unfortunately, the patch can no longer be applied to current svn without
a conflict.
The precision parsing and '%s' precision (truncation) handling is
already implemented.
But the '%d' precision (zero fill) handling part of your patch is still
missing in the code.
Could you please repost an updated patch in diff -u format?
Note that the current code does not use a format2_default flag buts sets
format2 to ~0U ('infinite' :-) if no precision is specified.
Thanks
--
Christian Franke
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: precision formatting in grub_printf
Robert Millan wrote: Changelog: 2008-08-17 Deepak Vankadaru deepak.vankad...@gmail.com Support for precision formatting in grub_printf * kern/misc.c: modified grub_vsprintf to parse and handle precision formatting Thanks. Please could you provide this in unified diff format? (diff -u) I already added (fixed) the precision parsing and added handling of '%s' precision (truncation): http://svn.savannah.gnu.org/viewvc/trunk/grub2/kern/misc.c?root=grubr1=1936r2=1954 The parsing part of this new patch might conflict with this change. -- Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] implement grub_millisleep in util/misc.c for grub-emu
Robert Millan wrote: On Sun, Feb 15, 2009 at 07:49:38PM +0100, Felix Zielcke wrote: --- util/misc.c (revision 1996) +++ util/misc.c (working copy) @@ -27,6 +27,9 @@ #include sys/time.h #include unistd.h +#define _POSIX_C_SOURCE 199309L +#include time.h I'm not sure this is garanteed to work unless it's defined before any header is included. Did it compile without warnings? At least with glibc headers, the #define from the patch has no effect, because at least unistd.h has already included features.h which handles these settings. There is no warning, because none of the headers define _POSIX_C_SOURCE. From features.h: /* If none of the ANSI/POSIX macros are defined, use POSIX.1 and POSIX.2 (and IEEE Std 1003.1b-1993 unless _XOPEN_SOURCE is defined). */ #if (...!defined (_POSIX_C_SOURCE)...) # define _POSIX_SOURCE 1 # if defined _XOPEN_SOURCE ... # ... # else # define _POSIX_C_SOURCE 200112L # endif #endif I would suggest to remove the #define and commit the patch. BTW: compiles fine on Cygwin, with or without the #define. -- Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] implement grub_millisleep in util/misc.c for grub-emu
Christian Franke wrote: There is no warning, because none of the headers define _POSIX_C_SOURCE. This should read: There is a warning, because features.h defines _POSIX_C_SOURCE :-) -- Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: Building grub2 for windows
Alexandre Bique wrote: On Thu, Feb 26, 2009 at 2:35 PM, Christian Franke ... wrote: What is the error message you got on Cygwin ? The problem is that make install installs elf binaries and not win32 binaries. But win32 binaries are compiled. I moved win32 binaries by hand. I could not reproduce this. Is the Cygwin PATH correct, such that the Cygwin tools /usr/bin/make, /usr/bin/install ... are actually used during build? Where do those 'elf binaries' come from? The Cygwin gcc toolchain cannot produce elf binaries. That's why grub-pe2elf exists to convert the *.mod files to ELF format. There is also a problem if you specify a root directory with space, i think that some variables are not protected with $my_var in some shell scripts. Yes, that is correct and should IMO be fixed in the future. -- Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: Building grub2 for windows
Alexandre Bique wrote: I am trying to build grub2-svn for windows by using mingw32 or cygwin but it fails to generate valid windows executables (grub-setup, grub-mkimage, etc...). My configure line: ./configure --prefix $HOME/grub-install using mingw32, msys, gcc 3.4.5 (mingw-vista special r3), binutils 2.19.1 I also tried with cygwin, gcc 3.4.4, binutils 2.18.50.20080625. Grub2 should build out of the box on Cygwin. The following works for me with current grub2 svn (revision 2001) on current Cygwin 1.5.25-15: $ ./configure --enable-grub-fstest --prefix=/tmp/inst $ make $ make install $ /tmp/inst/bin/grub-mkrescue grub-rescue.iso ... boot grub-rescue.iso ... There is actually one issue in current grub2 svn: Build with --enable-grub-emu configure option does not work. It fails with undefined symbol 'grub_millisleep'. What is the error message you got on Cygwin ? -- Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] hdparm.mod - get/set ATA disk parameters
Robert Millan wrote:
On Sat, Feb 14, 2009 at 03:13:31PM +0100, Christian Franke wrote:
insmod ata_pthru
(note that module dependencies should make this unnecessary)
This insmod is necessary for now. hdparm.mod does not directly call
ata_pthru.mod, it uses kern/disk.c::grub_disk_ata_pass_through function
pointer instead. Otherwise, using other future ata_pass_through
functions (e.g. from some ahci.mod :-) would not be possible.
Another drawback of a hdparm-ata_pthru dependency would be that 'help'
or 'hdparm -h' would load ata.mod which disables e.g. biosdisk.mod.
insmod hdparm
# Make sure disks cannot be locked by an ATA password
hdparm --quiet --security-freeze (ata4)
hdparm --quiet --security-freeze (ata6)
menuentry Boot {
# Check health
if hdparm --quiet --health (ata4) ; then echo -n ; else
echo Warning: SMART status check failed
read
fi
# Set boot disk to fast, disable spin down
hdparm --quiet --aam=254 --standby-timeout=0 (ata4)
# Set other disk to quiet, spin down after 5min inactivity
hdparm --quiet --aam=128 --standby-timeout=60 (ata6)
# Boot ...
}
menuentry Memory Test {
# Spin down both disks after 10min
hdparm --quiet --standby-timeout=120 (ata4)
hdparm --quiet --standby-timeout=120 (ata6)
# Load memtest ...
}
Very interesting. Do you think any of these features could be useful as a
default option in grub-mkconfig?
At least the --health check and --security-freeze are IMO recommended
for each disk. Both is also done by a typical PC BIOS. AAM and standby
settings are more user specific. The problem is that this relies on
native ATA support which is now only available for controllers supported
by ata.mod.
At least for the boot disk, such hdparm calls could be added by e.g.
extending 'prepare_grub_to_access_device'.
Christian
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] hdparm.mod - get/set ATA disk parameters
Robert Millan wrote: On Sat, Feb 21, 2009 at 07:00:06PM +0100, Christian Franke wrote: Very interesting. Do you think any of these features could be useful as a default option in grub-mkconfig? At least the --health check and --security-freeze are IMO recommended for each disk. Both is also done by a typical PC BIOS. AAM and standby settings are more user specific. The problem is that this relies on native ATA support which is now only available for controllers supported by ata.mod. At least for the boot disk, such hdparm calls could be added by e.g. extending 'prepare_grub_to_access_device'. But if the firmware already did, wouldn't this be a waste of boot time? Yes, it would. Most BIOS perform both health check and security freeze, but some don't. For coreboot, it is not a waste of boot time. So if this is added to grub-mkconfig, it should be configurable by some variable. -- Christian Franke ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: Bugfix: ata pass-through broke compilation
phcoder wrote: Hello. Here is bugfix Thanks! Yes, I should rename all occurences... (gcc 3.4.4 did not detect this). Committed. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] hdparm.mod - get/set ATA disk parameters
Robert Millan wrote: On Sun, Feb 08, 2009 at 12:12:32AM +0100, Christian Franke wrote: Robert Millan wrote: I would suggest to move grub_ata_pass_through() to a new module (e.g. ata2.mod, atax.mod, ataex.mod, ...) How about ata_something.mod? (consistent with ntfs.mod ntfs_comp.mod) It is actually ntfscomp.mod. ata_pthru.mod ? Deal! I will prepare a patch - ata_pthru.mod, atapthru.mod or a more generic name to allow to add other ata_extension functions ? Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] hdparm.mod - get/set ATA disk parameters
Robert Millan wrote: On Sat, Jan 24, 2009 at 11:59:01PM +0100, Christian Franke wrote: This patch adds a command which allows to change a few (S)ATA drive settings. It relies on a new pass-through function in ata.mod. Is this function going to be used for normal use of ata? Note that ata.mod should be as small as possible, since coreboot users usually want to include it the GRUB that is installed to flash. Like 'hdparm.mod' itself, the ATA pass-through function is not needed for normal boot. Would it make sense to put the function elsewhere? I would suggest to move grub_ata_pass_through() to a new module (e.g. ata2.mod, atax.mod, ataex.mod, ...) The function should not be in hdparm.mod itself, because the hdparm command would also work with other ATA pass-through functions. For example with some future ahci.mod or even in conjunction with native USB support through a SAT tunnel. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] hdparm.mod - get/set ATA disk parameters
Robert Millan wrote: I would suggest to move grub_ata_pass_through() to a new module (e.g. ata2.mod, atax.mod, ataex.mod, ...) How about ata_something.mod? (consistent with ntfs.mod ntfs_comp.mod) ata_pthru.mod ? Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Skip identical lines in hexdump
Christian Franke wrote: This skips identical lines in hexdump output, like 'od' or 'xxd -a' also do. Committed. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Allow more than one line in option help text
Christian Franke wrote: This small patch allows '\n' in option help texts, for example: Patch simplified committed. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
[PATCH] Skip identical lines in hexdump
This skips identical lines in hexdump output, like 'od' or 'xxd -a' also do.
Christian
PS: I would suggest to change hexdump 'buf' parameter from 'char *' to
'const void *' to avoid unnecessary casts.
2009-01-30 Christian Franke fra...@computer.org
* lib/hexdump.c (hexdump): Print at most 3 lines if data is identical.
diff --git a/lib/hexdump.c b/lib/hexdump.c
index 9b79f45..7689edf 100644
--- a/lib/hexdump.c
+++ b/lib/hexdump.c
@@ -61,6 +61,22 @@ hexdump (unsigned long bse, char *buf, int len)
grub_printf (%s\n, line);
+ /* Print only first and last line of more than 3 lines are identical. */
+ if (len = 4 * 16
+ ! grub_memcmp (buf, buf + 1 * 16, 16)
+ ! grub_memcmp (buf, buf + 2 * 16, 16)
+ ! grub_memcmp (buf, buf + 3 * 16, 16))
+ {
+ grub_printf (*\n);
+ do
+ {
+ bse += 16;
+ buf += 16;
+ len -= 16;
+ }
+ while (len = 3 * 16 ! grub_memcmp (buf, buf + 2 * 16, 16));
+ }
+
bse += 16;
buf += 16;
len -= cnt;
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] remove target_os
Javier Martín wrote:
Unfortunately, gcc has no '-fno_os' option to specify the bare CPU as
target.
Might -ffreestanding be what you are looking for?
The option '-ffreestanding' is the same as '-fno-hosted'.
According to gcc (4.3.1) source, '-fno-hosted' clears variable
'flag_hosted' and sets '-fno-builtin'. The latter is already set within
GRUB build.
A cleared 'flag_hosted' apparently has only 2 effects:
- disable the special handling of 'main()'.
- #define __STDC_HOSTED__ to 0 instead of 1
There is no effect on the target_os dependent parts of the gcc code
generation. For example, on i386, __enable_execute_stack() calls are
generated for target_os netbsd, openbsd and cygwin, but not for linux.
The emit call is hard-coded in
gcc/configs/i386/i386.c:x86_initialize_trampoline().
Other workarounds are needed to support building GRUB with code
generators tailored for various target_os.
AC_MSG_CHECKING([for command to convert module to ELF format])
-case ${host_os}:${target_os} in
- cygwin:cygwin) TARGET_OBJ2ELF='grub-pe2elf' ;;
+case ${host_os} in
+ cygwin) TARGET_OBJ2ELF='grub-pe2elf' ;;
*) ;;
esac
This won't work for a Linux cross compiler hosted on Cygwin. It would
emit ELF format and does not need pe2elf.
A, say, AMD64 Linux cross compiler hosted on x86 Cygwin would have
$build=i686-pc-cygwin and $host=amd64-linux-gnu. Thus, no conflict ought
to arise even with cross compilation enabled.
But the opposite won't work: $host=i686-pc-cygwin would enable
grub-pe2elf, even if this gcc emits ELF for a linux target.
Christian
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] remove target_os
Javier Martín wrote:
El mar, 27-01-2009 a las 18:21 +0100, Christian Franke escribió:
Javier Martín wrote:
Unfortunately, gcc has no '-fno_os' option to specify the bare CPU as
target.
Might -ffreestanding be what you are looking for?
The option '-ffreestanding' is the same as '-fno-hosted'.
According to gcc (4.3.1) source, '-fno-hosted' clears variable
'flag_hosted' and sets '-fno-builtin'. The latter is already set within
GRUB build.
A cleared 'flag_hosted' apparently has only 2 effects:
- disable the special handling of 'main()'.
- #define __STDC_HOSTED__ to 0 instead of 1
There is no effect on the target_os dependent parts of the gcc code
generation. For example, on i386, __enable_execute_stack() calls are
generated for target_os netbsd, openbsd and cygwin, but not for linux.
The emit call is hard-coded in
gcc/configs/i386/i386.c:x86_initialize_trampoline().
In that case, we are dealing with a GCC bug. We might want to require
the user to create a bare no-OS cross compiler.
Is building a bare 'no-OS' compiler supported by the upstream GCC sources?
Probably a too strict prerequisite for building GRUB.
AC_MSG_CHECKING([for command to convert module to ELF format])
-case ${host_os}:${target_os} in
- cygwin:cygwin) TARGET_OBJ2ELF='grub-pe2elf' ;;
+case ${host_os} in
+ cygwin) TARGET_OBJ2ELF='grub-pe2elf' ;;
*) ;;
esac
This won't work for a Linux cross compiler hosted on Cygwin. It would
emit ELF format and does not need pe2elf.
A, say, AMD64 Linux cross compiler hosted on x86 Cygwin would have
$build=i686-pc-cygwin and $host=amd64-linux-gnu. Thus, no conflict ought
to arise even with cross compilation enabled.
But the opposite won't work: $host=i686-pc-cygwin would enable
grub-pe2elf, even if this gcc emits ELF for a linux target.
The opposite of a the described situation would be a Linux AMD64
machine cross compiling for a Cygwin x86 machine, that is
$build=amd64-linux-gnu and $host=i686-pc-cygwin. In that case,
i686-pc-cygwin-gcc would generate PE executables, and thus grub-pe2elf
_is_ required indeed.
Yes - I was wrong, sorry.
Robert: Patch is IMO GTG.
Christian
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] remove target_os
Robert Millan wrote:
Hi,
Based on the description of host/target triplets in configure.ac:
dnl build -- the environment for building GRUB
dnl host -- the environment for running utilities
dnl target -- the environment for running GRUB
it seems that target_os is an oxymoron. There's no OS in the environment
where GRUB will run (well, there's the firmware, but we already use
target_vendor for that, and _os has a well-defined meaning).
Attached patch fixes that by supressing all references to target_os, and
replacing them with host_os where suitable.
Christian: since this mostly affects Cygwin, could you verify that it doesn't
cause breakage before I commit it?
Hi Robert,
thanks for sending the patch first. Cygwin build looks good.
But even if GRUB itself is build for some $target_cpu-$target_vendor
(i386-pc), the target_os is not useless if cross-compilation support is
desired:
- GRUB tools like grub-setup are build for the target_os
- Package structure may depend on target_os.
- The gcc code generator may emit special code tailored for the assumed
target_os
Examples:
Stack frame checks
Nested function trampolines (__enable_execute_stack())
Unfortunately, gcc has no '-fno_os' option to specify the bare CPU as
target.
Fortunately, all current issues can be handled by autoconf checks.
AC_MSG_CHECKING([for command to convert module to ELF format])
-case ${host_os}:${target_os} in
- cygwin:cygwin) TARGET_OBJ2ELF='grub-pe2elf' ;;
+case ${host_os} in
+ cygwin) TARGET_OBJ2ELF='grub-pe2elf' ;;
*) ;;
esac
This won't work for a Linux cross compiler hosted on Cygwin. It would
emit ELF format and does not need pe2elf.
But all this is theoretical unless cross compilation is really needed.
Christian
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] (ata.mod) Fix read/write, improve status check
Christian Franke wrote: This patch fixes these open issues: - grub_pio_read/write() check the ERR bit without ensuring !BSY. - ata_read fails if (batch % size) == 0. - ata_write does not work at all, it uses the read cmd. Tested on several PC + Committed. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Add 'precision' to grub_printf %s format
Christian Franke wrote: This patch adds 'precision' support to grub_printf '%s' format, e.g.: Committed. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
[PATCH] Add 'precision' to grub_printf %s format
This patch adds 'precision' support to grub_printf '%s' format, e.g.:
grub_printf(data=%.20s...\n, data);
This feature of standard printf() is useful to limit output length or to
print strings which are not 0-terminated.
Christian
2009-01-21 Christian Franke fra...@computer.org
* kern/misc.c (grub_vsprintf): Fix size and termination of `format2'
(precision) digit string. Allow `.format2' without `format1' (width).
Limit input chars for `%s' output to `format2' if specified. This is
compatible with standard printf ().
diff --git a/kern/misc.c b/kern/misc.c
index 6fbc651..da64eb0 100644
--- a/kern/misc.c
+++ b/kern/misc.c
@@ -700,7 +700,7 @@ grub_vsprintf (char *str, const char *fmt, va_list args)
char tmp[32];
char *p;
unsigned int format1 = 0;
- unsigned int format2 = 3;
+ unsigned int format2 = ~ 0U;
char zerofill = ' ';
int rightfill = 0;
int n;
@@ -727,20 +727,22 @@ grub_vsprintf (char *str, const char *fmt, va_list args)
zerofill = '0';
format1 = grub_strtoul (s, 0, 10);
fmt = p;
- if (*p *p == '.')
+ }
+
+ if (*p *p == '.')
+ {
+ p++;
+ fmt++;
+ while (*p grub_isdigit (*p))
+ p++;
+
+ if (p fmt)
{
- p++;
- fmt++;
- while (*p grub_isdigit (*p))
- p++;
-
- if (p fmt)
- {
- char fstr[p - fmt];
- grub_strncpy (fstr, fmt, p - fmt);
- format2 = grub_strtoul (fstr, 0, 10);
- fmt = p;
- }
+ char fstr[p - fmt + 1];
+ grub_strncpy (fstr, fmt, p - fmt);
+ fstr[p - fmt] = 0;
+ format2 = grub_strtoul (fstr, 0, 10);
+ fmt = p;
}
}
@@ -847,13 +849,19 @@ grub_vsprintf (char *str, const char *fmt, va_list args)
p = va_arg (args, char *);
if (p)
{
- if (!rightfill grub_strlen (p) format1)
- write_fill (zerofill, format1 - grub_strlen (p));
-
- write_str (p);
-
- if (rightfill grub_strlen (p) format1)
- write_fill (zerofill, format1 - grub_strlen (p));
+ grub_size_t len = 0;
+ while (len format2 p[len])
+ len++;
+
+ if (!rightfill len format1)
+ write_fill (zerofill, format1 - len);
+
+ grub_size_t i;
+ for (i = 0; i len; i++)
+ write_char (*p++);
+
+ if (rightfill len format1)
+ write_fill (zerofill, format1 - len);
}
else
write_str ((null));
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] (ata.mod) Fix ATAPI protocol
Marco Gerards wrote:
Committed, plus an additional wait in grub_atapi_identify.
Great! :-)
Open issues:
- grub_pio_read/write() check the ERR bit without ensuring !BSY.
- ata_read fails if (batch % size) == 0.
- ata_write does not work at all, it uses the read cmd.
grub_ata_write does not use the read cmd, or do you mean indirectly?
In that case, where does it use the read command?
cmd_write is set, but not used:
grub_ata_readwrite (...)
{
...
cmd = GRUB_ATA_CMD_READ_SECTORS;
cmd_write = GRUB_ATA_CMD_WRITE_SECTORS;
if (rw == 0)
{
...
} else {
/* Write sectors. */
if (grub_ata_cmd (dev, cmd))
-^^^ GRUB_ATA_CMD_READ_SECTORS
...
}
...
}
I will post a patch for all 3 issues soon.
Christian
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
[PATCH] (ata.mod) Fix read/write, improve status check
This patch fixes these open issues:
- grub_pio_read/write() check the ERR bit without ensuring !BSY.
- ata_read fails if (batch % size) == 0.
- ata_write does not work at all, it uses the read cmd.
Christian
2009-01-20 Christian Franke fra...@computer.org
* disk/ata.c (grub_ata_wait_status): Replace by ...
(grub_ata_wait_not_busy): ... this function. Checks only BSY bit,
other status bits may be invalid while BSY is asserted.
(grub_ata_check_ready): New function.
(grub_ata_cmd): Removed.
(grub_ata_wait_drq): New function.
(grub_ata_strncpy): Remove inline.
(grub_ata_pio_read): Reduce to actual block transfer. BSY wait
and error check now done by grub_ata_wait_drq ().
(grub_ata_pio_write): Likewise.
(grub_atapi_identify): Set DEV before check for !BSY. Use
grub_ata_wait_drq () to wait for data.
(grub_atapi_wait_drq): Use grub_ata_wait_not_busy ().
(grub_atapi_packet): Set DEV before check for !BSY. Replace
transfer loop by grub_ata_pio_write ().
(grub_ata_identify): Set DEV before check for !BSY. Use
grub_ata_wait_drq () to wait for data.
(grub_ata_setaddress): Set DEV before check for !BSY.
(grub_ata_readwrite): Remove duplicate code, handle batch/rest and
read/write in one loop. Fix invalid command on write. Fix incomplete
command on (size % batch) == 0. Add missing error check after write of
last block. Add debug messages.
(grub_atapi_read): Replace transfer loop by grub_ata_pio_read ().
diff --git a/disk/ata.c b/disk/ata.c
index c0904fb..c710f13 100644
--- a/disk/ata.c
+++ b/disk/ata.c
@@ -152,25 +152,29 @@ grub_ata_regget2 (struct grub_ata_device *dev, int reg)
return grub_inb (dev-ioaddress2 + reg);
}
-static inline grub_err_t
-grub_ata_wait_status (struct grub_ata_device *dev,
- grub_uint8_t maskset, grub_uint8_t maskclear,
- int milliseconds)
+/* Wait for !BSY. */
+static grub_err_t
+grub_ata_wait_not_busy (struct grub_ata_device *dev, int milliseconds)
{
- int i;
+ /* ATA requires 400ns (after a write to CMD register) or
+ 1 PIO cycle (after a DRQ block transfer) before
+ first check of BSY. */
+ grub_millisleep (1);
- for (i = 0; i milliseconds; i++)
+ int i = 1;
+ while (grub_ata_regget (dev, GRUB_ATA_REG_STATUS) GRUB_ATA_STATUS_BUSY)
{
- grub_uint8_t reg;
-
- reg = grub_ata_regget (dev, GRUB_ATA_REG_STATUS);
- if ((reg maskset) == maskset (reg maskclear) == 0)
- return GRUB_ERR_NONE;
+ if (i = milliseconds)
+{
+ grub_dprintf (ata, timeout: %dms\n, milliseconds);
+ return grub_error (GRUB_ERR_TIMEOUT, ATA timeout);
+ }
grub_millisleep (1);
+ i++;
}
- return grub_error (GRUB_ERR_TIMEOUT, ata timeout);
+ return GRUB_ERR_NONE;
}
static inline void
@@ -179,19 +183,42 @@ grub_ata_wait (void)
grub_millisleep (50);
}
+/* Check for !BSY before issuing a new command. */
+static inline grub_err_t
+grub_ata_check_ready (struct grub_ata_device *dev)
+{
+ if (grub_ata_regget (dev, GRUB_ATA_REG_STATUS) GRUB_ATA_STATUS_BUSY)
+return grub_ata_wait_not_busy (dev, GRUB_ATA_TOUT_STD);
+
+ return GRUB_ERR_NONE;
+}
+
+/* Wait for !BSY, DRQ. */
static grub_err_t
-grub_ata_cmd (struct grub_ata_device *dev, int cmd)
+grub_ata_wait_drq (struct grub_ata_device *dev, int rw,
+ int milliseconds)
{
- if (grub_ata_wait_status (dev, 0, GRUB_ATA_STATUS_BUSY, GRUB_ATA_TOUT_STD))
+ if (grub_ata_wait_not_busy (dev, milliseconds))
return grub_errno;
- grub_ata_regset (dev, GRUB_ATA_REG_CMD, cmd);
+ /* !DRQ implies error condition. */
+ grub_uint8_t sts = grub_ata_regget (dev, GRUB_ATA_REG_STATUS);
+ if ((sts (GRUB_ATA_STATUS_DRQ | GRUB_ATA_STATUS_ERR))
+ != GRUB_ATA_STATUS_DRQ)
+{
+ grub_dprintf (ata, ata error: status=0x%x, error=0x%x\n,
+ sts, grub_ata_regget (dev, GRUB_ATA_REG_ERROR));
+ if (! rw)
+return grub_error (GRUB_ERR_READ_ERROR, ATA read error);
+ else
+return grub_error (GRUB_ERR_WRITE_ERROR, ATA write error);
+}
return GRUB_ERR_NONE;
}
/* Byteorder has to be changed before strings can be read. */
-static inline void
+static void
grub_ata_strncpy (char *dst, char *src, grub_size_t len)
{
grub_uint16_t *src16 = (grub_uint16_t *) src;
@@ -203,52 +230,26 @@ grub_ata_strncpy (char *dst, char *src, grub_size_t len)
dst[len] = '\0';
}
-static grub_err_t
-grub_ata_pio_read (struct grub_ata_device *dev, char *buf,
- grub_size_t size, int milliseconds)
+static void
+grub_ata_pio_read (struct grub_ata_device *dev, char *buf, grub_size_t size)
{
grub_uint16_t *buf16 = (grub_uint16_t *) buf;
unsigned int i;
- if (grub_ata_regget (dev, GRUB_ATA_REG_STATUS) GRUB_ATA_STATUS_ERR)
-return grub_error (GRUB_ERR_READ_ERROR, ATA read error);
-
- /* Wait until the data is available. */
- if (grub_ata_wait_status (dev
Re: [PATCH] (ata.mod) Fix ATAPI protocol
Marco Gerards wrote: Christian Franke ... writes: This patch fixes the data I/O protocol of the ATA PACKET command. The current implementation does not read the byte count registers. ATAPI read may not work if the drive sends the data in more than one block. In conjunction with my recent SCSI blocksize patch, ATAPI CD/DVD drives should now work. Sorry for not commenting on many things on this list. I am too busy to check all this. N.P. Meantime, I have tested the SCSI and ATAPI fixes on several PC. If there are no further comments, I will commit these soon. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] (scsi.mod) Fix SCSI read for blocksize != 512
Christian Franke wrote: This patch fixes SCSI blocksize handling, necessary for ATAPI CD/DVD. Committed. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] (ata.mod) Fix ATAPI protocol
Christian Franke wrote: Marco Gerards wrote: Christian Franke ... writes: This patch fixes the data I/O protocol of the ATA PACKET command. The current implementation does not read the byte count registers. ATAPI read may not work if the drive sends the data in more than one block. In conjunction with my recent SCSI blocksize patch, ATAPI CD/DVD drives should now work. Sorry for not commenting on many things on this list. I am too busy to check all this. N.P. Meantime, I have tested the SCSI and ATAPI fixes on several PC. If there are no further comments, I will commit these soon. Committed, plus an additional wait in grub_atapi_identify. Open issues: - grub_pio_read/write() check the ERR bit without ensuring !BSY. - ata_read fails if (batch % size) == 0. - ata_write does not work at all, it uses the read cmd. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
[PATCH] (ata.mod) Fix ATAPI protocol
This patch fixes the data I/O protocol of the ATA PACKET command.
The current implementation does not read the byte count registers. ATAPI
read may not work if the drive sends the data in more than one block.
In conjunction with my recent SCSI blocksize patch, ATAPI CD/DVD drives
should now work.
Christian
2008-01-18 Christian Franke fra...@computer.org
* disk/ata.c (GRUB_ATAPI_REG_*): New defines.
(GRUB_ATAPI_IREASON_*): Likewise.
(grub_ata_pio_write): Fix timeout error return.
(grub_atapi_wait_drq): New function.
(grub_atapi_packet): New parameter `size'.
Use grub_atapi_wait_drq () and direct write instead of
grub_ata_pio_write ().
(grub_atapi_read): Replace grub_ata_pio_read () by a loop which
reads the number of bytes requested by the device for each DRQ
assertion.
(grub_atapi_write): Remove old implementation, return not
implemented instead.
diff --git a/disk/ata.c b/disk/ata.c
index af7158c..bb891d4 100644
--- a/disk/ata.c
+++ b/disk/ata.c
@@ -44,12 +44,15 @@ static const int grub_ata_ioaddress2[] = { 0x3f6, 0x376 };
#define GRUB_ATA_REG_ERROR 1
#define GRUB_ATA_REG_FEATURES 1
#define GRUB_ATA_REG_SECTORS 2
+#define GRUB_ATAPI_REG_IREASON 2
#define GRUB_ATA_REG_SECTNUM 3
#define GRUB_ATA_REG_CYLLSB 4
#define GRUB_ATA_REG_CYLMSB 5
#define GRUB_ATA_REG_LBALOW 3
#define GRUB_ATA_REG_LBAMID 4
+#define GRUB_ATAPI_REG_CNTLOW 4
#define GRUB_ATA_REG_LBAHIGH 5
+#define GRUB_ATAPI_REG_CNTHIGH 5
#define GRUB_ATA_REG_DISK 6
#define GRUB_ATA_REG_CMD 7
#define GRUB_ATA_REG_STATUS 7
@@ -65,6 +68,13 @@ static const int grub_ata_ioaddress2[] = { 0x3f6, 0x376 };
#define GRUB_ATA_STATUS_READY 0x40
#define GRUB_ATA_STATUS_BUSY 0x80
+/* ATAPI interrupt reason values (I/O, D/C bits). */
+#define GRUB_ATAPI_IREASON_MASK 0x3
+#define GRUB_ATAPI_IREASON_DATA_OUT 0x0
+#define GRUB_ATAPI_IREASON_CMD_OUT 0x1
+#define GRUB_ATAPI_IREASON_DATA_IN 0x2
+#define GRUB_ATAPI_IREASON_ERROR0x3
+
enum grub_ata_commands
{
GRUB_ATA_CMD_READ_SECTORS = 0x20,
@@ -229,7 +239,7 @@ grub_ata_pio_write (struct grub_ata_device *dev, char *buf,
/* Wait until drive is ready to read data. */
if (grub_ata_wait_status (dev, GRUB_ATA_STATUS_DRQ, 0, milliseconds))
-return 0;
+return grub_errno;
/* Write the data, word by word. */
for (i = 0; i size / 2; i++)
@@ -300,21 +310,65 @@ grub_atapi_identify (struct grub_ata_device *dev)
}
static grub_err_t
-grub_atapi_packet (struct grub_ata_device *dev, char *packet)
+grub_atapi_wait_drq (struct grub_ata_device *dev,
+ grub_uint8_t ireason,
+ int milliseconds)
{
+ grub_millisleep (1); /* ATA allows 400ns to assert BSY. */
+
+ if (grub_ata_wait_status (dev, 0, GRUB_ATA_STATUS_BUSY, milliseconds))
+return grub_errno;
+
+ grub_uint8_t sts = grub_ata_regget (dev, GRUB_ATA_REG_STATUS);
+ grub_uint8_t irs = grub_ata_regget (dev, GRUB_ATAPI_REG_IREASON);
+
+ /* OK if DRQ is asserted and interrupt reason is as expected. */
+ if ((sts GRUB_ATA_STATUS_DRQ)
+ (irs GRUB_ATAPI_IREASON_MASK) == ireason)
+return GRUB_ERR_NONE;
+
+ /* No DRQ implies error condition. */
+ grub_dprintf(ata, atapi error: status=0x%x, ireason=0x%x, error=0x%x\n,
+ sts, irs, grub_ata_regget (dev, GRUB_ATA_REG_ERROR));
+
+ if (! (sts GRUB_ATA_STATUS_DRQ)
+ (irs GRUB_ATAPI_IREASON_MASK) == GRUB_ATAPI_IREASON_ERROR)
+{
+ if (ireason == GRUB_ATAPI_IREASON_CMD_OUT)
+ return grub_error (GRUB_ERR_READ_ERROR, ATA PACKET command error);
+ else
+ return grub_error (GRUB_ERR_READ_ERROR, ATAPI read error);
+}
+
+ return grub_error (GRUB_ERR_READ_ERROR, ATAPI protocol error);
+}
+
+static grub_err_t
+grub_atapi_packet (struct grub_ata_device *dev, char *packet,
+ grub_size_t size)
+{
+ if (grub_ata_wait_status(dev, 0, GRUB_ATA_STATUS_BUSY, GRUB_ATA_TOUT_STD))
+return grub_errno;
+
+ /* Send ATA PACKET command. */
grub_ata_regset (dev, GRUB_ATA_REG_DISK, dev-device 4);
grub_ata_regset (dev, GRUB_ATA_REG_FEATURES, 0);
- grub_ata_regset (dev, GRUB_ATA_REG_SECTORS, 0);
- grub_ata_regset (dev, GRUB_ATA_REG_LBAHIGH, 0xFF);
- grub_ata_regset (dev, GRUB_ATA_REG_LBAMID, 0xFF);
+ grub_ata_regset (dev, GRUB_ATAPI_REG_IREASON, 0);
+ grub_ata_regset (dev, GRUB_ATAPI_REG_CNTHIGH, size 8);
+ grub_ata_regset (dev, GRUB_ATAPI_REG_CNTLOW, size 0xFF);
grub_ata_regset (dev, GRUB_ATA_REG_CMD, GRUB_ATA_CMD_PACKET);
- grub_ata_wait ();
-
- if (grub_ata_pio_write (dev, packet, 12, GRUB_ATA_TOUT_STD))
+ /* Wait for !BSY, DRQ, !I/O, C/D. */
+ if (grub_atapi_wait_drq (dev, GRUB_ATAPI_IREASON_CMD_OUT, GRUB_ATA_TOUT_STD))
return grub_errno;
+ /* Write the packet. */
+ grub_uint16_t *buf16 = (grub_uint16_t *) packet;
+ unsigned i;
+ for (i = 0; i 12 / 2; i++)
+grub_outw (grub_cpu_to_le16 (buf16[i]), dev-ioaddress + GRUB_ATA_REG_DATA);
+
return GRUB_ERR_NONE;
}
@@ -847,27 +901,51
[PATCH] (scsi.mod) Fix SCSI read for blocksize != 512
This patch fixes SCSI blocksize handling, necessary for ATAPI CD/DVD.
OT: There might be a memory leak in scsi_open(): free(scsi) is missing,
at least on open error.
Christian
2009-01-17 Christian Franke fra...@computer.org
* disk/scsi.c (grub_scsi_read10): Use scsi-blocksize instead
of 512 to calculate data size.
(grub_scsi_read12): Likewise.
(grub_scsi_write10): Likewise.
(grub_scsi_write12): Likewise.
(grub_scsi_read): Adjust size according to blocksize.
Add checks for invalid blocksize and unaligned transfer.
diff --git a/disk/scsi.c b/disk/scsi.c
index b7b6834..d165309 100644
--- a/disk/scsi.c
+++ b/disk/scsi.c
@@ -119,7 +119,7 @@ grub_scsi_read10 (grub_disk_t disk, grub_disk_addr_t sector,
rd.reserved2 = 0;
rd.pad = 0;
- return scsi-dev-read (scsi, sizeof (rd), (char *) rd, size * 512, buf);
+ return scsi-dev-read (scsi, sizeof (rd), (char *) rd, size * scsi-blocksize, buf);
}
/* Send a SCSI request for DISK: read SIZE sectors starting with
@@ -140,7 +140,7 @@ grub_scsi_read12 (grub_disk_t disk, grub_disk_addr_t sector,
rd.reserved = 0;
rd.control = 0;
- return scsi-dev-read (scsi, sizeof (rd), (char *) rd, size * 512, buf);
+ return scsi-dev-read (scsi, sizeof (rd), (char *) rd, size * scsi-blocksize, buf);
}
#if 0
@@ -163,7 +163,7 @@ grub_scsi_write10 (grub_disk_t disk, grub_disk_addr_t sector,
wr.reserved2 = 0;
wr.pad = 0;
- return scsi-dev-write (scsi, sizeof (wr), (char *) wr, size * 512, buf);
+ return scsi-dev-write (scsi, sizeof (wr), (char *) wr, size * scsi-blocksize, buf);
}
/* Send a SCSI request for DISK: write the data stored in BUF to SIZE
@@ -184,7 +184,7 @@ grub_scsi_write12 (grub_disk_t disk, grub_disk_addr_t sector,
wr.reserved = 0;
wr.pad = 0;
- return scsi-dev-write (scsi, sizeof (wr), (char *) wr, size * 512, buf);
+ return scsi-dev-write (scsi, sizeof (wr), (char *) wr, size * scsi-blocksize, buf);
}
#endif
@@ -325,8 +325,22 @@ grub_scsi_read (grub_disk_t disk, grub_disk_addr_t sector,
/* SCSI sectors are variable in size. GRUB uses 512 byte
sectors. */
- sector = grub_divmod64 (sector, scsi-blocksize GRUB_DISK_SECTOR_BITS,
- NULL);
+ if (scsi-blocksize != GRUB_DISK_SECTOR_SIZE)
+{
+ unsigned spb = scsi-blocksize GRUB_DISK_SECTOR_BITS;
+ if (! (spb != 0 (scsi-blocksize GRUB_DISK_SECTOR_SIZE) == 0))
+ return grub_error (GRUB_ERR_NOT_IMPLEMENTED_YET,
+ Unsupported SCSI block size);
+
+ grub_int32_t sector_mod = 0;
+ sector = grub_divmod64 (sector, spb, sector_mod);
+
+ if (! (sector_mod == 0 size % spb == 0))
+ return grub_error (GRUB_ERR_NOT_IMPLEMENTED_YET,
+ Unaligned SCSI read not supported);
+
+ size /= spb;
+}
/* Depending on the type, select a read function. */
switch (scsi-devtype)
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] (ata.mod) Add variable timeouts, fix identify, fix device selection
Christian Franke wrote: This patch fixes the following issues in ata.mod I found during testing on several PC and VMs: ... Committed. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] (ata.mod) Add variable timeouts, fix identify, fix device selection
Christian Franke wrote: ... Other issues found, not addressed in this patch: ... - atapi_read does not always work (occasional timeouts, crash on VmWare) Root of this problem is in scsi.mod: It always assumes a sector size of 512 bytes which is not the case for ATAPI CD/DVD. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
[PATCH] (ata.mod) Add variable timeouts, fix identify, fix device selection
This patch fixes the following issues in ata.mod I found during testing
on several PC and VMs:
- The 1s timeout is too short for read, at least for CD/DVD. Now use
variable timeout depending on command.
- The DEV bit is not set before other registers are set. If both master
and slave are present, the registers of the wrong device may be set.
- DEVICE DIAGNOSTICS may not work for device detection for several
reasons, see changelog.
Now uses only IDENTIFY and then IDENTIFY PACKET, like (at least) the
traditional Linux IDE driver did.
Other issues found, not addressed in this patch:
- ata_read fails if (batch % size) == 0.
- ata_write does not work at all, it uses the read cmd.
- atapi_read does not always work (occasional timeouts, crash on VmWare)
I will provide patches for these later.
Christian
2009-01-14 Christian Franke fra...@computer.org
* disk/ata.c (enum grub_ata_commands): Remove EXEC_DEV_DIAGNOSTICS.
(enum grub_ata_timeout_milliseconds): New enum.
(grub_ata_wait_status): Add parameter milliseconds.
(grub_ata_cmd): Remove variable `err'. Remove wait for !DRQ to allow
recovery from timed-out commands.
(grub_ata_pio_read): Add parameter milliseconds. Fix error return,
return grub_errno instead of REG_ERROR.
(grub_ata_pio_write): Add parameter milliseconds.
(grub_atapi_identify): Fix size of ATAPI IDENTIFY sector.
Pass milliseconds to grub_ata_wait_status () and
grub_ata_pio_read ().
(grub_atapi_packet): Pass milliseconds to grub_ata_pio_write ().
(grub_ata_identify): Remove variable `ataerr'. Pass milliseconds to
grub_ata_wait_status (). Fix IDENTIFY timeout check.
(grub_ata_device_initialize): Remove EXECUTE DEVICE DIAGNOSTICS.
It is not suitable for device detection, because DEV bit is ignored,
the command may run too long, and not all devices set the signature
properly.
(grub_ata_pciinit): Clear grub_errno before grub_ata_device_initialize
().
(grub_ata_setaddress): Pass milliseconds to grub_ata_wait_status ().
Fix device selection, DEV bit must be set first to address the registers
of the correct device.
(grub_ata_readwrite): Pass milliseconds to grub_ata_wait_status () and
grub_ata_pio_read/write ().
(grub_atapi_read): Pass milliseconds to grub_ata_pio_read ().
(grub_atapi_write): Pass milliseconds to grub_ata_pio_write ().
diff --git a/disk/ata.c b/disk/ata.c
index 4ca63c2..af7158c 100644
--- a/disk/ata.c
+++ b/disk/ata.c
@@ -74,7 +74,12 @@ enum grub_ata_commands
GRUB_ATA_CMD_IDENTIFY_DEVICE = 0xEC,
GRUB_ATA_CMD_IDENTIFY_PACKET_DEVICE = 0xA1,
GRUB_ATA_CMD_PACKET = 0xA0,
-GRUB_ATA_CMD_EXEC_DEV_DIAGNOSTICS = 0x90
+ };
+
+enum grub_ata_timeout_milliseconds
+ {
+GRUB_ATA_TOUT_STD = 1000, /* 1s standard timeout. */
+GRUB_ATA_TOUT_DATA = 1 /* 10s DATA I/O timeout. */
};
struct grub_ata_device
@@ -139,11 +144,12 @@ grub_ata_regget2 (struct grub_ata_device *dev, int reg)
static inline grub_err_t
grub_ata_wait_status (struct grub_ata_device *dev,
- grub_uint8_t maskset, grub_uint8_t maskclear)
+ grub_uint8_t maskset, grub_uint8_t maskclear,
+ int milliseconds)
{
int i;
- for (i = 0; i 1000; i++)
+ for (i = 0; i milliseconds; i++)
{
grub_uint8_t reg;
@@ -166,12 +172,8 @@ grub_ata_wait (void)
static grub_err_t
grub_ata_cmd (struct grub_ata_device *dev, int cmd)
{
- grub_err_t err;
-
- err = grub_ata_wait_status (dev, 0,
- GRUB_ATA_STATUS_DRQ | GRUB_ATA_STATUS_BUSY);
- if (err)
-return err;
+ if (grub_ata_wait_status (dev, 0, GRUB_ATA_STATUS_BUSY, GRUB_ATA_TOUT_STD))
+return grub_errno;
grub_ata_regset (dev, GRUB_ATA_REG_CMD, cmd);
@@ -193,16 +195,16 @@ grub_ata_strncpy (char *dst, char *src, grub_size_t len)
static grub_err_t
grub_ata_pio_read (struct grub_ata_device *dev, char *buf,
- grub_size_t size)
+ grub_size_t size, int milliseconds)
{
grub_uint16_t *buf16 = (grub_uint16_t *) buf;
unsigned int i;
if (grub_ata_regget (dev, GRUB_ATA_REG_STATUS) GRUB_ATA_STATUS_ERR)
-return grub_ata_regget (dev, GRUB_ATA_REG_ERROR);
+return grub_error (GRUB_ERR_READ_ERROR, ATA read error);
/* Wait until the data is available. */
- if (grub_ata_wait_status (dev, GRUB_ATA_STATUS_DRQ, 0))
+ if (grub_ata_wait_status (dev, GRUB_ATA_STATUS_DRQ, 0, milliseconds))
return grub_errno;;
/* Read in the data, word by word. */
@@ -217,16 +219,16 @@ grub_ata_pio_read (struct grub_ata_device *dev, char *buf,
static grub_err_t
grub_ata_pio_write (struct grub_ata_device *dev, char *buf,
- grub_size_t size)
+ grub_size_t size, int milliseconds)
{
grub_uint16_t *buf16 = (grub_uint16_t *) buf;
unsigned int i;
if (grub_ata_regget (dev, GRUB_ATA_REG_STATUS) GRUB_ATA_STATUS_ERR)
-return grub_ata_regget (dev
[PATCH] Fix ATA compatibility mode check
For the compatibility mode check, the ATA driver uses the bits in the
Revision byte instead of the Programming Interface Byte of the class
value (See T13/1510D). The compat_use[] flags are never set.
This patch fixes both issues.
Christian
2009-01-11 Christian Franke fra...@computer.org
* disk/ata.c (grub_ata_pciinit): Fix bit numbers of compatibility
mode check. Fix setting of compat_use[].
diff --git a/disk/ata.c b/disk/ata.c
index 6e0b8b8..4ca63c2 100644
--- a/disk/ata.c
+++ b/disk/ata.c
@@ -504,7 +504,7 @@ grub_ata_pciinit (int bus, int device, int func,
for (i = 0; i 2; i++)
{
/* Set to 0 when the channel operated in compatibility mode. */
- int compat = (class (2 * i)) 1;
+ int compat = (class (8 + 2 * i)) 1;
rega = 0;
regb = 0;
@@ -515,7 +515,7 @@ grub_ata_pciinit (int bus, int device, int func,
{
rega = grub_ata_ioaddress[i];
regb = grub_ata_ioaddress2[i];
- compat_use[i] = 0;
+ compat_use[i] = 1;
}
else if (compat)
{
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: __FILE__ (Re: [PATCH] framework for building modules externally)
Robert Millan wrote: On Wed, Nov 05, 2008 at 10:41:20PM +0100, Christian Franke wrote: PS: The current use of __FILE__ may also add extra unexpected size: For packaging, configure is often run outside of $srcdir with a absolute path name. This may result in long __FILE__ strings, like /home/maintainer/packaging/grub/tmp/grub-1.96+20081105-1/src/grub-1.96/kern/disk.c This has annoyed me for a while. Do you know a proper fix? There is apparently no option to modify __FILE__ expansion. Therefore, the relative file name has to be specified in the module itself. See attached patch for a possible fix: Each module using grub_dprintf (here disk.c) may specify its name in 'this_file'. When all modules are changed, the '#define this_file' and all #undefs can be removed. Christian diff --git a/include/grub/misc.h b/include/grub/misc.h index 15c18f5..36e4bcc 100644 --- a/include/grub/misc.h +++ b/include/grub/misc.h @@ -27,7 +27,9 @@ #define ALIGN_UP(addr, align) (((grub_uint64_t)addr + align - 1) ~(align - 1)) -#define grub_dprintf(condition, fmt, args...) grub_real_dprintf(__FILE__, __LINE__, condition, fmt, ## args); +#define grub_dprintf(condition, fmt, args...) grub_real_dprintf(this_file, __LINE__, condition, fmt, ## args); +#define this_file __FILE__ + /* XXX: If grub_memmove is too slow, we must implement grub_memcpy. */ #define grub_memcpy(d,s,n) grub_memmove ((d), (s), (n)) diff --git a/kern/disk.c b/kern/disk.c index ed82506..039c011 100644 --- a/kern/disk.c +++ b/kern/disk.c @@ -25,6 +25,9 @@ #include grub/time.h #include grub/file.h +#undef this_file +static const char this_file[] = kern/disk.c; + #define GRUB_CACHE_TIMEOUT 2 /* The last time the disk was used. */ ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] framework for building modules externally
Robert Millan wrote:
...
How do you treat differences in ABI ?
Dis-allow loading of module with different value of ABI ? Or were you
planning that module itself adapts to different versions of GRUB 2 ABI's?
The module itself could:
GRUB_MOD_INIT(foo)
{
if (grub_abi != GRUB_ABI)
{
grub_printf (abi mismatch!\n);
return;
}
/* register our commands/terminals/disks/whatever */
}
Alternative: Export a symbol describing the ABI version in kernel
(grub_abi_3_14). Access this symbol in each module (this can be hidden
in GRUB_MOD_INIT).
Then loading a module with wrong ABI version fails due to undefined symbol.
--
Christian Franke
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Add __enable_execute_stack() if required
walt wrote: Robert Millan wrote: On Tue, Sep 16, 2008 at 09:42:50PM +0200, Christian Franke wrote: Some gcc versions generate a call to __enable_execute_stack() in trampolines for nested functions. This is the case for new Cygwin gcc-4.3.2. Other GRUB2 target platforms may be affected - the following files in 'gcc-4.3.2/gcc/config' source directory contains implementations of this function for libgcc: alpha/osf.h darwin.h netbsd.h I recall this was also an issue for NetBSD indeed. Could someone confirm it fixes the problem there? It does fix the problem on NetBSD, thanks Christian. You're welcome. Thanks for testing. Patch committed. If more such functions are necessary in the future, it is probably better to collect those in some libgcc.a replacement module. This would be the case if e.g. for the 64-bit integer division. On 32-bit architectures, gcc typically generates a call to __divdi3(). Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Add __enable_execute_stack() if required
Robert Millan wrote: On Tue, Sep 16, 2008 at 09:42:50PM +0200, Christian Franke wrote: Some gcc versions generate a call to __enable_execute_stack() in trampolines for nested functions. This is the case for new Cygwin gcc-4.3.2. Other GRUB2 target platforms may be affected - the following files in 'gcc-4.3.2/gcc/config' source directory contains implementations of this function for libgcc: alpha/osf.h darwin.h netbsd.h I recall this was also an issue for NetBSD indeed. Could someone confirm it fixes the problem there? There was also a related discussion for Mac OS X here: http://lists.gnu.org/archive/html/grub-devel/2005-12/msg00044.html If there are no complaints, I will commit the patch on friday. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] fix disk-id abuse
Robert Millan wrote:
On Tue, Sep 02, 2008 at 09:12:04PM +0200, Christian Franke wrote:
If disk-id is supposed to be a GUID ('Grub Unique Identifier' in this
case :-), then a pointer to the private data structure for the disk
should work. This id is unique until disk close.
For drivers without disk-data, simply use the address of e.g. the open
function itself.
This is fine for single-disk drivers, but for multi-disk ones we need it to
be unique among different disks provided by the same driver.
I apparently made the false assumption, that multi-disk drivers always
have disk-data as a real pointer.
Although, of course, I don't see why can't we just make it use a pointer to
itself:
disk-id = (unsigned long) disk-id;
but then what's the point of storing that in a variable anyway. We might as
well just remove this variable and whoever uses it can use a pointer to the
structure instead?
This works on the assumption that disk structures are never reallocated, but
I suppose that's a sane thing to assume...
Sounds good.
Probably add an inline function to improve readability:
typedef unsigned long grub_disk_id_t;
inline grub_disk_id_t disk_id (const grub_disk_t * disk)
{
return (grub_id_t)(disk);
}
...
data = grub_disk_cache_fetch (disk-dev-id, disk_id(disk), start_sector);
Christian
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Mingw support for grub2
Bean wrote:
Hi,
This patch add support for mingw, now you can create native executable
for windows.
Nice!
Does grub-setup work?
...
--- a/include/grub/util/misc.h
+++ b/include/grub/util/misc.h
...
+#ifdef __MINGW32__
+
+#include windows.h
+
+grub_int64_t fseeko (FILE *fp, grub_int64_t offset, int whence);
+grub_int64_t ftello (FILE *fp);
The mingw runtime provides fseeko64/ftello64(), see
/usr/include/mingw/stdio.h
So the following may work:
#ifdef __MINGW32__
#define fseeko fseeko64
#define ftello ftello64
#endif
or use inline functions.
+void sync (void);
+int asprintf (char **buf, const char *fmt, ...);
+
I would suggest to add AC_CHECK_FUNC(asprintf) to configure.
asprintf() is a GNU extension and not part of C99 or POSIX.
+grub_int64_t grub_util_get_disk_size (char *name);
+
+#define sleep Sleep
The Sleep() parameter specifies milliseconds.
#define sleep(s) Sleep((s)*1000)
or
inline void sleep(unsigned s) { Sleep(s * 1000); }
or
add sleep() to util/misc.c to avoid global inclusion of the namespace
polluter windows.h :-)
Christian
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Mingw support for grub2
Vesa Jääskeläinen wrote:
Bean wrote:
On Mon, Aug 25, 2008 at 12:04 AM, Vesa Jääskeläinen [EMAIL PROTECTED] wrote:
Bean wrote:
Yes, it works, although mingw can't use device names like /dev/sda,
but it can use windows special name //./PHYSICALDRIVE0.
That should be with back slashes ('\'). Or is there some hack somewhere
to support that?
Hi,
Mingw would translate / to \, so both are ok. But using / is more
clear, as \ is the escape character in cygwin shell, so we need to use
two of them, .\\PHYSICALDRIVE0.
In cygwin shell... yes... but mingw programs are not meant to be run
under that. Mingw programs are native windows programs.
Unlike Cygwin, Mingw does not provide an extra library layer. Most
standard library functions directly link to msvcrt.dll.
The Win32 API (like the DOS API) itself accepts both / and \, only the
user level tools usually don't. This is likely because in the early
(CP/M-DOS) days, someone decided to use '/' as the option char. Later,
when path names are 'borrowed' from Unix, it was too late :-)
In fact, all these work to open the first disk:
h = CreateFile(.\\PhysicalDrive0, ...)
h = CreateFile(./PhysicalDrive0, ...)
h = CreateFile(//.\\PhysicalDrive0, ...)
h = CreateFile(//./PhysicalDrive0, ...)
Christian
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Don't install Cygwin specific files elsewhere
Christian Franke wrote: This patch prevents the install of grub.d/10_windows on other OS. grub-pe2elf is only installed if requested by --enable-grub-pe2elf. Even on Cygwin, grub-pe2elf is only required during build. Committed. ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: bug: cygwin support breaks cross compiles on cygwin
Felix Zielcke wrote: Am Freitag, den 08.08.2008, 23:13 +0200 schrieb Christian Franke: I agree. I posted a patch which handles both grub.d/10_windows and grub-pe2elf. Thanks I even saw that mail before this. I even reverted now my rm -f 10_windows change to current Debian trunk, but to commit I wait untill your patch is commited upstream ;) Done. The quick fix for the cross-compiler problem is also included. Does not actually check the output format of the target compiler, but should work for now. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
Re: bug: cygwin support breaks cross compiles on cygwin
Patrick Georgi wrote:
Hi,
I used cygwin to build grub2/i386-coreboot using an i386-elf cross
compiler. unfortunately the test in configure only looks for the host
platform and enables pe2elf in that case, even though my compiler
emits ELF already.
my workaround is to just change that test (there's no OS cyg_win), but
of course that's no permanent solution.
Bean proposed on IRC to look if the target compiler emits PE, and
enable pe2elf based on that.
Thanks for the info.
A simple fix is probably a test for target_os also, please try attached
patch (and re-run autoconf).
Christian
diff --git a/configure.ac b/configure.ac
index 7a5938a..41e08f1 100644
--- a/configure.ac
+++ b/configure.ac
@@ -206,9 +206,10 @@ AC_MSG_RESULT([$TARGET_IMG_LDFLAGS_AC])
# For platforms where ELF is not the default link format.
AC_MSG_CHECKING([for command to convert module to ELF format])
-if test $host_os = cygwin; then
- TARGET_OBJ2ELF='grub-pe2elf.exe'
-fi
+case ${host_os}:${target_os} in
+ cygwin:cygwin) TARGET_OBJ2ELF='grub-pe2elf.exe' ;;
+ *) ;;
+esac
AC_SUBST(TARGET_OBJ2ELF)
AC_MSG_RESULT([$TARGET_OBJ2ELF])
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: bug: cygwin support breaks cross compiles on cygwin
Felix Zielcke wrote: Hello, Am Freitag, den 08.08.2008, 14:59 +0200 schrieb Christian Franke: No, grub-pe2elf is build and installed even if is not necessary for the build process. This should be no problem. Well it's the same as with that --enable-debug thingy ;) In our Debian trunk SVN I already commited a change to remove grub.d/10_windows because Debian/Ubuntu users just don't have a need for this. It would be better if Robert and me don't need to do such changes. We could just remove grub-pe2elf, too. But why should it be build for everybody if only Cygwin users have a use for it? Then clearly the whole build process needs a change so that grub-pe2elf is only build on Cygwin. I agree. I posted a patch which handles both grub.d/10_windows and grub-pe2elf. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
[PATCH] Don't install Cygwin specific files elsewhere
This patch prevents the install of grub.d/10_windows on other OS.
grub-pe2elf is only installed if requested by --enable-grub-pe2elf. Even
on Cygwin, grub-pe2elf is only required during build.
Christian
2008-08-08 Christian Franke [EMAIL PROTECTED]
* Makefile.in: Add `target_os' and `enable_grub_pe2elf'.
* conf/common.rmk: Install `grub-pe2elf' only if requested.
Install `grub.d/10_windows' only on Cygwin.
* configure.ac: Add subst of `target_os'.
Check `target_os' also before setting TARGET_OBJ2ELF.
Add `--enable-grub-pe2elf'.
diff --git a/Makefile.in b/Makefile.in
index 2121431..6137974 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -48,6 +48,7 @@ PACKAGE_STRING = @PACKAGE_STRING@
PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
target_cpu = @target_cpu@
+target_os = @target_os@
platform = @platform@
INSTALL = @INSTALL@
@@ -92,6 +93,7 @@ UNIFONT_HEX = @UNIFONT_HEX@
# Options.
enable_grub_emu = @enable_grub_emu@
enable_grub_fstest = @enable_grub_fstest@
+enable_grub_pe2elf = @enable_grub_pe2elf@
enable_lzo = @enable_lzo@
### General variables.
diff --git a/conf/common.rmk b/conf/common.rmk
index 3d674a6..95859f7 100644
--- a/conf/common.rmk
+++ b/conf/common.rmk
@@ -100,7 +100,10 @@ grub_editenv_SOURCES = util/grub-editenv.c lib/envblk.c util/misc.c kern/misc.c
CLEANFILES += grub-editenv
# for grub-pe2elf
+ifeq ($(enable_grub_pe2elf), yes)
bin_UTILITIES += grub-pe2elf
+endif
+
grub_pe2elf_SOURCES = util/grub-pe2elf.c util/misc.c
CLEANFILES += grub-pe2elf
@@ -120,7 +123,11 @@ CLEANFILES += update-grub_lib
%: util/grub.d/%.in config.status
./config.status --file=$@:$
chmod +x $@
-update-grub_SCRIPTS = 00_header 10_linux 10_hurd 10_windows 30_os-prober 40_custom
+update-grub_SCRIPTS = 00_header 10_linux 10_hurd 30_os-prober 40_custom
+ifeq ($(target_os), cygwin)
+update-grub_SCRIPTS += 10_windows
+endif
+
CLEANFILES += $(update-grub_SCRIPTS)
update-grub_DATA += util/grub.d/README
diff --git a/configure.ac b/configure.ac
index 7a5938a..f561b8b 100644
--- a/configure.ac
+++ b/configure.ac
@@ -100,6 +100,7 @@ case $target_cpu in
esac
AC_SUBST(target_cpu)
+AC_SUBST(target_os)
AC_SUBST(platform)
#
@@ -206,9 +207,10 @@ AC_MSG_RESULT([$TARGET_IMG_LDFLAGS_AC])
# For platforms where ELF is not the default link format.
AC_MSG_CHECKING([for command to convert module to ELF format])
-if test $host_os = cygwin; then
- TARGET_OBJ2ELF='grub-pe2elf.exe'
-fi
+case ${host_os}:${target_os} in
+ cygwin:cygwin) TARGET_OBJ2ELF='grub-pe2elf' ;;
+ *) ;;
+esac
AC_SUBST(TARGET_OBJ2ELF)
AC_MSG_RESULT([$TARGET_OBJ2ELF])
@@ -385,6 +387,11 @@ AC_ARG_ENABLE([grub-fstest],
[build and install the `grub-fstest' debugging utility])])
AC_SUBST([enable_grub_fstest])
+AC_ARG_ENABLE([grub-pe2elf],
+ [AS_HELP_STRING([--enable-grub-pe2elf],
+ [build and install the `grub-pe2elf' conversion utility])])
+AC_SUBST([enable_grub_pe2elf])
+
# Output files.
AC_CONFIG_LINKS([include/grub/cpu:include/grub/$target_cpu
include/grub/machine:include/grub/$target_cpu/$platform])
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
[PATCH] Fix infinite loop in grub_pit_wait()
grub2 from current SVN hangs if run in VirtualPC, the problem was
introduced with 'svn diff -r 1779:1780'.
Here is a proposed fix.
Christian
2008-08-07 Christian Franke [EMAIL PROTECTED]
* kern/i386/pit.c (TIMER2_SPEAKER): New define.
(TIMER2_GATE): Likewise.
(grub_pit_wait): Add enable/disable of the timer2 gate
bit of port 0x61. This fixes a possible infinite loop.
diff --git a/kern/i386/pit.c b/kern/i386/pit.c
index d0a6eda..a3fab26 100644
--- a/kern/i386/pit.c
+++ b/kern/i386/pit.c
@@ -28,13 +28,26 @@
#define TIMER_ENABLE_LSB 0x20
#define TIMER_ENABLE_MSB 0x10
#define TIMER2_LATCH 0x20
+#define TIMER2_SPEAKER 0x02
+#define TIMER2_GATE 0x01
void
grub_pit_wait (grub_uint16_t tics)
{
+ /* Disable timer2 gate and speaker. */
+ grub_outb (grub_inb (TIMER2_REG_LATCH) ~ (TIMER2_SPEAKER | TIMER2_GATE), TIMER2_REG_LATCH);
+
+ /* Set tics. */
grub_outb (TIMER2_SELECT | TIMER_ENABLE_LSB | TIMER_ENABLE_MSB, TIMER_REG_COMMAND);
grub_outb (tics 0xff, TIMER2_REG_CONTROL);
grub_outb (tics 8, TIMER2_REG_CONTROL);
+ /* Enable timer2 gate, keep speaker disabled. */
+ grub_outb ((grub_inb (TIMER2_REG_LATCH) ~ TIMER2_SPEAKER) | TIMER2_GATE, TIMER2_REG_LATCH);
+
+ /* Wait. */
while ((grub_inb (TIMER2_REG_LATCH) TIMER2_LATCH) == 0x00);
+
+ /* Disable timer2 gate and speaker. */
+ grub_outb (grub_inb (TIMER2_REG_LATCH) ~ (TIMER2_SPEAKER | TIMER2_GATE), TIMER2_REG_LATCH);
}
___
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel
Re: [PATCH] Fix infinite loop in grub_pit_wait()
Robert Millan wrote: On Thu, Aug 07, 2008 at 05:32:42PM +0200, Christian Franke wrote: grub2 from current SVN hangs if run in VirtualPC, the problem was introduced with 'svn diff -r 1779:1780'. I'm not familiar with this part of the PIT interface; why does it hang only on VirtualPC? Possibly different BIOS init of 0x61? Or different implementation of the legacy hardware, which IIRC looks like this: . +---+ /-0x61[5] ....--Clock--| Timer | | .0x61[0]---Gate--| 2 |-Out-+-+---+ . +---+| |--Speaker .0x61[1]--+---+ Resulting values of tsc_ticks_per_ms (AMD [EMAIL PROTECTED]): . origwith patch VMware: ~1500~200 VirtualBox: ~200 ~200 VirtualPC: hang ~200 ... + /* Disable timer2 gate and speaker. */ + grub_outb (grub_inb (TIMER2_REG_LATCH) ~ (TIMER2_SPEAKER | TIMER2_GATE), TIMER2_REG_LATCH); } This doesn't AFAICT preserve the existing value of the timer2 gate and speaker. I assume that is ok? IMO yes. The timer state itself cannot be preserved, so it should be safe to leave it disabled. To test, try If BIOS speaker output (echo -e \a) still works. Thanks! You're welcome. Patch committed. Christian ___ Grub-devel mailing list Grub-devel@gnu.org http://lists.gnu.org/mailman/listinfo/grub-devel
