OSDN Git Service
Peter Jones [Wed, 5 Feb 2014 04:26:47 +0000 (23:26 -0500)]
Get rid of the 1024-byte limit on ESP file path.
Not that I think 1024 was something we were going to hit there, but...
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Wed, 5 Feb 2014 04:02:25 +0000 (23:02 -0500)]
Handle some other device paths.
Add support for Vendor-Defined Messaging Device Paths as well as
FVFILE() and FVVOL()
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Wed, 5 Feb 2014 03:39:00 +0000 (22:39 -0500)]
Get rid of 1024-byte limit for displaying variables as well.
This basically makes all of unparse_path() and its derivatives into
snprintf()-like functions, so we don't have to have static buffers.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Tue, 4 Feb 2014 23:51:45 +0000 (18:51 -0500)]
Add a TODO here so I can keep track of what is yet to be done.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Tue, 10 Sep 2013 18:46:00 +0000 (14:46 -0400)]
Typo fix.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Tue, 10 Sep 2013 15:25:41 +0000 (11:25 -0400)]
Correctly fill out sa_family for MAC address device types.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 18:49:28 +0000 (14:49 -0400)]
Handle virtio network devices.
Just like virtio scsi devices, look up their bus/device/function by the
virtio device instead of the traditional way.
Signed-off-by: Peter Jones <pjones@redhat.com>
Tested-and-reviewed-by: Laszlo Ersek <lersek@redhat.com>
Laszlo Ersek [Mon, 9 Sep 2013 09:37:39 +0000 (11:37 +0200)]
'-h' is a valid option, recognize it
This completes commit
e562a948.
Signed-off-by: Laszlo Ersek <lersek@redhat.com>
Peter Jones [Fri, 6 Sep 2013 20:18:12 +0000 (16:18 -0400)]
Ignore patches.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 20:17:20 +0000 (16:17 -0400)]
Remove 40-character limit on file paths in /boot.
This is totally unnecessary now; the only real point before was
everything was fixed size to keep it within 1024.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 18:46:18 +0000 (14:46 -0400)]
Don't allow negative boot numbers.
In the case where you typo the command line and do:
efibootmgr -b -a blah blah blah
instead of
efibootmgr -b 0001 -a blah blah blah
sscanf(..., "%X", &num) happily produces a successful return code and a
value of -10. This later results in us creating a boot variable named
BootFFFFFFFF6 , which is clearly not okay. So if we've got a negative
number there, it's probably not what the user wanted.
Bounds check the whole thing.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 18:16:24 +0000 (14:16 -0400)]
Make make_linux_load_option() slightly more readable.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 15:31:03 +0000 (11:31 -0400)]
Make add_extra_args_file() work properly.
Two things:
1) make it actually return the data
2) _ascii() and _unicode() will get invoked in some cases, and they need
to handle the case where they do nothing.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 15:23:16 +0000 (11:23 -0400)]
Audit strlen() in append_extra_args_ascii() as well.
Similar code, similar bugs.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 15:13:39 +0000 (11:13 -0400)]
Restore old boot order behavior.
Previously we prepended rather than appending new elements to the boot
order.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 15:02:13 +0000 (11:02 -0400)]
Do better error checking in read_boot_u16(), and simplify it.
Why am I even doing this stupid bitshifting?
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 14:57:50 +0000 (10:57 -0400)]
Fix byte assignment in read_boot_u16().
It helps if you put the bytes in the right order, and the parenthesis
around the right things to keep them from being all zero.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 14:43:35 +0000 (10:43 -0400)]
More work on append_extra_args_unicode()
two problems:
1) strlen off-by-one
2) efichar_from_char() awesomely takes a length in bytes, returns in
bytes, and operates on efi_char16_t sizes. This can be somewhat
confusing.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 14:24:22 +0000 (10:24 -0400)]
You really can't free something you put through realloc().
This should make '-u' work again.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 14:10:18 +0000 (10:10 -0400)]
Fix typo in arg parsing.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 14:00:40 +0000 (10:00 -0400)]
Remove some dead code.
If anybody's looking for it, it's in git.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Fri, 6 Sep 2013 13:01:14 +0000 (09:01 -0400)]
Clean up .gitignores some.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Thu, 5 Sep 2013 19:53:06 +0000 (15:53 -0400)]
Make --help also -h, and don't have it exit with an error status.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Thu, 5 Sep 2013 19:52:50 +0000 (15:52 -0400)]
Make tilt_slashes() static.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Thu, 5 Sep 2013 18:29:35 +0000 (14:29 -0400)]
Fix uninitialized variable.
valgrind found a couple of errors, one of which fixes some odd behavior
with BootOrder I was seeing.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Thu, 5 Sep 2013 18:16:42 +0000 (14:16 -0400)]
Fix minor compiler warning.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Thu, 5 Sep 2013 18:15:53 +0000 (14:15 -0400)]
Fix completely wrong compiler warning.
gcc's data flow analysis still kind of blows.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Thu, 5 Sep 2013 17:46:43 +0000 (13:46 -0400)]
Make the error case from make_boot_var() cleaner.
bcl thinks it looks better at the end, and I agree.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Thu, 5 Sep 2013 17:41:31 +0000 (13:41 -0400)]
Don't do unnecessary math.
bcl found this one too - we know len % 5 is 4, because it's tested a few
lines up.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Thu, 5 Sep 2013 17:39:17 +0000 (13:39 -0400)]
Fix error message that says "delete" on the "activate" case.
bcl found this reviewing it.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Wed, 4 Sep 2013 22:48:05 +0000 (18:48 -0400)]
Use libefivar.
Using libefivar divorces us from the kernel api, which is a very good
thing. It also means we don't need to know about efi_status_t or efi
error types, or carry our own headers for describing variables.
This also makes device paths > 1024 bytes long work (on newer kernels)
Signed-off-by: Peter Jones <pjones@redhat.com>
Conflicts:
src/efibootmgr/efibootmgr.c
src/lib/efi.c
Peter Jones [Wed, 4 Sep 2013 22:46:56 +0000 (18:46 -0400)]
Ignore some files so "git status" is clearer.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Wed, 4 Sep 2013 22:46:34 +0000 (18:46 -0400)]
Silence some minor compiler warnings.
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Mon, 13 Jan 2014 21:08:16 +0000 (16:08 -0500)]
Release 0.6.1
Signed-off-by: Peter Jones <pjones@redhat.com>
Peter Jones [Thu, 5 Sep 2013 18:01:25 +0000 (14:01 -0400)]
Modernize "make tarball" for our glorious git future.
Signed-off-by: Peter Jones <pjones@redhat.com>
Daniel Jared Dominguez [Mon, 13 Jan 2014 19:20:56 +0000 (13:20 -0600)]
Prepping for move to GitHub. Updated documentation to refer to new maintainer and new git repository.
Daniel Jared Dominguez [Mon, 13 Jan 2014 18:56:25 +0000 (12:56 -0600)]
From: Colin Watson <cjwatson@ubuntu.com>
Fix -w option to leave an existing unique MBR signature intact. This makes the
implementation match the documentation.
Daniel Jared Dominguez [Mon, 13 Jan 2014 18:45:42 +0000 (12:45 -0600)]
From: Bdale Garbee <bdale@gag.com>
change computeroutput to programlisting to fix man page formatting issue,
update rules to use docbook source for man page, closes Debian bug #680716
Jordan Hargrave [Fri, 18 Oct 2013 18:49:59 +0000 (13:49 -0500)]
Fix unchecked malloc calls.
Resolves: rhbz#715216
From Peter Jones@Redhat
Jordan Hargrave [Fri, 18 Oct 2013 18:34:24 +0000 (13:34 -0500)]
produce virtio-scsi device paths under EDD 3.0
Patch from Laszlo Ersek@RedHat
Jordan Hargrave [Fri, 18 Oct 2013 18:29:40 +0000 (13:29 -0500)]
prevent out-of-bounds access with SCSI_IOCTL_GET_PCI
Patch from Laszlo Ersek@RedHat
Jordan Hargrave [Fri, 18 Oct 2013 18:15:46 +0000 (13:15 -0500)]
disk_get_virt_pci(): generalize to arbitrary major & minor device numbers
Patch from Laszlo Ersek@RedHat
Jordan Hargrave [Fri, 18 Oct 2013 18:04:26 +0000 (13:04 -0500)]
disk_info_from_fd(): group output parameters into struct disk_info
Patch from Laszlo Ersek@RedHat
Jordan Hargrave [Fri, 18 Oct 2013 17:39:58 +0000 (12:39 -0500)]
produce virtio-blk device paths under EDD 3.0
Patch from Laszlo Ersek@RedHat
Jordan Hargrave [Fri, 18 Oct 2013 17:26:18 +0000 (12:26 -0500)]
disk_get_pci(): expose detected interface type
Patch from Laszlo Ersek@RedHat
Jordan Hargrave [Fri, 18 Oct 2013 17:14:45 +0000 (12:14 -0500)]
disk_info_from_fd(): get major/minor from buf.st_rdev
Patch from Laszlo Ersek@RedHat
Jordan Hargrave [Tue, 2 Jul 2013 15:28:12 +0000 (10:28 -0500)]
Add support for UNIX-style path names (forward-slash)
Submitted by Keshav Padram [the.ridikulus.rat@gmail.com]
Jordan Hargrave [Tue, 29 Jan 2013 20:37:33 +0000 (14:37 -0600)]
Update manpage email
Jordan Hargrave [Wed, 23 Jan 2013 18:36:32 +0000 (12:36 -0600)]
Bump efibootmgr version to include latest changes
Jordan Hargrave [Thu, 29 Nov 2012 16:11:43 +0000 (10:11 -0600)]
Add patches from fedora release
* Wed Nov 28 2012 Matthew Garrett <mjg59@srcf.ucam.org> - 0.5.4-14
- efibootmgr-0.5.4-Work-around-broken-Apple-firmware.patch
Resolves: #873629
- efibootmgr-0.5.4-Remove-device-path-padding-on-non-Itanium.patch - improve
spec conformance
Jordan Hargrave [Tue, 24 Apr 2012 17:58:57 +0000 (12:58 -0500)]
From: Lane Winner<lane.winner@oracle.com>
make_boot_var does not check for failed status with create_variable.
This can result in a memory leak.
Additionally the user should be notified of the problem.
We encounter this issue on one system after filling up the UEFI boot list
with dummy devices.
The patch fix the problem. It was verified on a Mensa system using RHEL 6.0
Signed-off-by: Yinghai Lu<yinghai@kernel.org>
Peter Jones [Wed, 14 Jul 2010 18:26:49 +0000 (11:26 -0700)]
Handle sector_size != 512.
Disks can have 4kB sectors now, so don't just bail out when that's the
case.
Matt Domsch [Thu, 23 Jul 2009 19:20:19 +0000 (14:20 -0500)]
fix disk minor number discovery
Raymund Will noted disk_info_from_fd() incorrectly used logical &&
instead of bitwise & when obtaining the minor number.
Reported in https://bugzilla.novell.com/show_bug.cgi?id=524529#c1
Matt Domsch [Thu, 23 Jul 2009 19:18:11 +0000 (14:18 -0500)]
fix minor memory leak
David Binderman noted new_data was being allocated but not freed. Not
a big deal as the program exits soon thereafter (and is thus freed),
but worth fixing anyhow.
Fixes https://bugzilla.novell.com/show_bug.cgi?id=524529#c1
Matt Domsch [Fri, 11 Jan 2008 21:09:12 +0000 (15:09 -0600)]
update ChangeLog
Matt Domsch [Fri, 11 Jan 2008 21:08:12 +0000 (15:08 -0600)]
replacing elilo < 3.6-6, not -5
per conversation with clumens.
Matt Domsch [Thu, 3 Jan 2008 18:19:28 +0000 (12:19 -0600)]
RPM spec rewrite for Fedora inclusion
efibootmgr is being split out of the elilo SRPM in Fedora, into its
own SRPM.
Matt Domsch [Thu, 3 Jan 2008 18:18:31 +0000 (12:18 -0600)]
Makefile cleanups
bump version
take an EXTRA_CFLAGS argument so rpmbuild can give us it's CFLAGS
without overriding ours.
exclude .git and *~ files from the tarball.
Matt Domsch [Thu, 3 Jan 2008 18:16:35 +0000 (12:16 -0600)]
src/lib/efi.c: include linux/types.h
patch from Fedora.
Matt Domsch [Thu, 3 Jan 2008 18:15:18 +0000 (12:15 -0600)]
make sure GPT_HEADER_SIGNATURE is a 64-bit value
patch from Debian / Ubuntu 0.5.3-3ubuntu1.
Matt Domsch [Mon, 12 Nov 2007 19:31:32 +0000 (13:31 -0600)]
avoid more unaligned access warnings
Matt Domsch [Mon, 12 Nov 2007 18:25:42 +0000 (12:25 -0600)]
cleanup a few unaligned access warnings
Matt Domsch [Mon, 12 Nov 2007 18:12:37 +0000 (12:12 -0600)]
cleanup exit values a little
Doug Chapman [Mon, 12 Nov 2007 16:32:12 +0000 (11:32 -0500)]
patch to make efibootmgr have non-zero exit code on errors
We have some automated tools that use efibootmgr which are having a hard
time detecting when efibootmgr fails since it exits with 0 on many
failures. This patch catches (most) errors and exits with non-zero. I
also added several error messages for to make it more obvious what was
wrong with the command line arguments.
Signed-off-by: Matt Domsch <Matt_Domsch@dell.com>
Matt Domsch [Mon, 9 Jul 2007 16:47:50 +0000 (16:47 +0000)]
add -lz to libs, needed when libpci happens to need it
Matt Domsch [Mon, 9 Jul 2007 16:23:34 +0000 (16:23 +0000)]
apply patch from Dave Jiang <djiang@mvista.com> to enable cross-building
Matt Domsch [Thu, 11 Aug 2005 17:37:04 +0000 (17:37 +0000)]
v0.5.2.2
Matt Domsch [Thu, 11 Aug 2005 17:36:29 +0000 (17:36 +0000)]
document -@ option
Matt Domsch [Thu, 11 Aug 2005 16:55:20 +0000 (16:55 +0000)]
make efi_variable_t.DataSize be unsigned long, to match kernel implementation. This isn't ideal, I'd prefer to be able to have a 32-bit efibootmgr on a 64-bit kernel, but with efi_status_t being a long also, we're already stuck with needing a native copy.
Matt Domsch [Wed, 10 Aug 2005 15:59:33 +0000 (15:59 +0000)]
clean up extra_opts_file handling
Matt Domsch [Wed, 10 Aug 2005 15:28:30 +0000 (15:28 +0000)]
new option -@ grabs extra variable args from a file
Matt Domsch [Wed, 10 Aug 2005 05:34:53 +0000 (05:34 +0000)]
add -g flag, move _FILE_OFFSET_BITS=64 to Makefile
Matt Domsch [Wed, 10 Aug 2005 05:30:14 +0000 (05:30 +0000)]
v0.5.2
Matt Domsch [Wed, 10 Aug 2005 05:28:23 +0000 (05:28 +0000)]
move _FILE_OFFSET_BITS definition into Makefile
Matt Domsch [Tue, 26 Jul 2005 19:37:11 +0000 (19:37 +0000)]
parse PCI domain information in make_net_load_option() on 2.6 kernels
Matt Domsch [Tue, 26 Jul 2005 19:32:24 +0000 (19:32 +0000)]
properly free allocated memory (caught by valgrind)
Matt Domsch [Tue, 8 Mar 2005 22:54:53 +0000 (22:54 +0000)]
applied patch from Alex Williamson which causes BootXXXX values to be printed in uppercase hex rather than lowercase. This also now allows the full 64K boot entries, rather than requiring they start with a zero. Release 0.5.1
Matt Domsch [Thu, 26 Aug 2004 21:02:14 +0000 (21:02 +0000)]
put linux.dell.com in spec file
Matt Domsch [Thu, 26 Aug 2004 20:27:28 +0000 (20:27 +0000)]
version 0.5.0-test4 released as v0.5.0 with no code changes
Matt Domsch [Wed, 9 Jun 2004 21:51:19 +0000 (21:51 +0000)]
Add PCI parent devices to device path for netboot and EDD30 entries
Matt Domsch [Wed, 9 Jun 2004 21:50:22 +0000 (21:50 +0000)]
changelog for version 0.5.0-test4
Matt Domsch [Wed, 9 Jun 2004 21:49:58 +0000 (21:49 +0000)]
version 0.5.0-test4
Matt Domsch [Wed, 9 Jun 2004 21:49:33 +0000 (21:49 +0000)]
Don't let read_boot_order() return EFI_NOT_FOUND when it is about to create it.
Matt Domsch [Wed, 9 Jun 2004 21:48:33 +0000 (21:48 +0000)]
Add libpci to libraries linked in
Matt Domsch [Wed, 9 Jun 2004 21:47:53 +0000 (21:47 +0000)]
Add Joshua Giles entry
Matt Domsch [Sat, 24 Apr 2004 15:18:57 +0000 (15:18 +0000)]
fix reversed logic in create_or_edit_variable(), remove extraneous printks in sysfs_read/write()
Matt Domsch [Thu, 5 Feb 2004 23:01:53 +0000 (23:01 +0000)]
Add ability to add/change/delete Timeout, reflect this everywhere.
Matt Domsch [Fri, 19 Dec 2003 22:09:01 +0000 (22:09 +0000)]
Remove extra sizeof() in definition of BLKGETSIZE64. It's
unnecessary, and causes new gcc 3.3.x compilers to complain and fail.
Matt Domsch [Tue, 9 Sep 2003 19:49:27 +0000 (19:49 +0000)]
More procfs to sysfs cleanups
Release version 0.5.0-test1
Matt Domsch [Thu, 4 Sep 2003 19:36:18 +0000 (19:36 +0000)]
- Seperated access to variables through /proc into efivars_procfs.[ch]
- Added efivars_sysfs.h to access variables through sysfs.
- Moved around some functions, cleaned up some duplication.
Matt Domsch [Mon, 11 Aug 2003 20:28:05 +0000 (20:28 +0000)]
* Mon Aug 11 2003 Matt Domsch <Matt_Domsch@dell.com>
- fixed unaligned access errors
- removed extraneous printing of mac addr when creating netboot entries
- sync docbook to README
- whitespace cleanups
Matt Domsch [Thu, 31 Jul 2003 18:50:33 +0000 (18:50 +0000)]
Adding filelist-rpm
Matt Domsch [Thu, 31 Jul 2003 18:46:46 +0000 (18:46 +0000)]
* Thu Jul 31 2003 Matt Domsch <Matt_Domsch@dell.com>
- Applied patch from Dann Frazier to enable creating netboot entries.
- update AUTHORS with Dann's netboot contribution.
- Until we can get the ACPI HID and UID programatically, make the user
pass them in when creating a netboot entry.
- Add O_DIRECT support for reading the disk.
- Fix unparse_hardware_path() for the PCI case - the device and function
values were printed in reverse order.
- Fix the README file to reflect all the options that can be passed, and
add a new item for netboot entries.
- whitespace cleanups
Matt Domsch [Fri, 25 Oct 2002 15:22:44 +0000 (15:22 +0000)]
- Patch from Fabien Lebaillif - Delamare increases the length of the boot
option description from 40 to 64 bytes.
Matt Domsch [Fri, 25 Oct 2002 15:07:41 +0000 (15:07 +0000)]
Added efibootmgr.8 man page (generated from efibootmgr.8.docbook
with docbook2man)
Matt Domsch [Fri, 25 Oct 2002 15:04:17 +0000 (15:04 +0000)]
* Tue Oct 22 2002 Matt Domsch <Matt_Domsch@dell.com>
- ran docbook2man on the man page Dann Frazier created
- Added man page to spec file
- Added patches from Dann Frazier
- Removes the requirement to be root to run efibootmgr. This lets
mortal users run efibootmgr to check the version, and see the
--help output. It should also allow efibootmgr to work under
security systems without the strict root/non-root dichotomy.
- Checks to see if a boot option was specified when a boot option
deletion was requested.
- Released version 0.4.1
* Sun Jul 21 2002 Matt Domsch <Matt_Domsch@dell.com>
- Added kernel_has_blkgetsize64() test, now uses BLKGETSIZE64 if on a
known-good kernel. This is important when the Large Block Device (64-bit
block address) patch gets merged into the 2.5 kernel.
Matt Domsch [Mon, 22 Jul 2002 01:39:19 +0000 (01:39 +0000)]
moved uname() call into kernel_has_blkgetsize64, renamed that function
Matt Domsch [Sun, 21 Jul 2002 04:06:55 +0000 (04:06 +0000)]
added uname test for calling BLKGETSIZE64 ioctl
Matt Domsch [Wed, 1 May 2002 18:20:25 +0000 (18:20 +0000)]
moved executable to /usr/sbin/
Matt Domsch [Wed, 1 May 2002 16:06:27 +0000 (16:06 +0000)]
moved location of tarball on domsch.com