Mostly for cosmetic reasons, we add a "net_dhcp" command, which is (at the
moment) identical to the existing "net_bootp" command. Both actually trigger
a DHCP handshake now, and both should be able to deal with pure BOOTP servers.
We could think about dropping the DHCP options from the initial DISCOVER packet
when the user issues the net_bootp command, but it's unclear whether this is
really useful, as both protocols should be able to coexist.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Even though we were parsing some DHCP options sent by the server, so far
we are only using the BOOTP 2-way handshake, even when talking to a DHCP
server.
Change this by actually sending out DHCP DISCOVER packets instead of the
generic (mostly empty) BOOTP BOOTREQUEST packets.
A pure BOOTP server would ignore the extra DHCP options in the DISCOVER
packet and would just reply with a BOOTREPLY packet, which we also
handle in the code.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
In respone to a BOOTREQUEST packet a BOOTP server would answer with a BOOTREPLY
packet, which ends the conversation for good. DHCP uses a 4-way handshake,
where the initial server respone is an OFFER, which has to be answered with
REQUEST by the client again, only to be completed by an ACKNOWLEDGE packet
from the server.
Teach the grub_net_process_dhcp() function to deal with OFFER packets,
and treat ACK packets the same es BOOTREPLY packets.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
The BOOTP RFC describes the boot file name and the server name as being part
of the integral BOOTP data structure, with some limits on the size of them.
DHCP extends this by allowing them to be separate DHCP options, which is more
flexible.
Teach the code dealing with those fields to check for those DHCP options first
and use this information, if provided. We fall back to using the BOOTP
information if those options are not used.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Currently we have a global timeout for all network cards in the BOOTP/DHCP
discovery process.
Make this timeout a per-interface one, so better accommodate the upcoming
4-way DHCP handshake and to also cover the lease time limit a DHCP offer
will come with.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Change the interface of the function dealing with incoming BOOTP packets
to take an interface instead of a card, to allow more fine per-interface
state (timeout, handshake state) later on.
Use the opportunity to clean up the code a bit.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
In contrast to BOOTP, DHCP uses a 4-way handshake, so requires to send
packets more often.
Refactor the generation and sending of the BOOTREQUEST packet into
a separate function, so that future code can more easily reuse this.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
DHCP specifies a special dummy option OVERLOAD, to allow DHCP options to
spill over into the (legacy) BOOTFILE and SNAME fields.
Parse and handle this option properly.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
For proper DHCP support we will need to parse DHCP options from a packet
more often and at various places.
Refactor the option parsing into a new function, which will scan a packet to
find *a particular* option field. Use that new function in places where we
were dealing with DHCP options before.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
The comment is right, the "giaddr" fields holds the IP address of the BOOTP
relay, not a general purpose router address. Just remove the commented code,
archeologists can find it in the git history.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
In util/getroot and efidisk slightly modify exitsing comment to mostly
retain it but still make GCC7 compliant with respect to fall through
annotation.
In grub-core/lib/xzembed/xz_dec_lzma2.c it adds same comments as
upstream.
In grub-core/tests/setjmp_tets.c declare functions as "noreturn" to
suppress GCC7 warning.
In grub-core/gnulib/regexec.c use new __attribute__, because existing
annotation is not recognized by GCC7 parser (which requires that comment
immediately precedes case statement).
Otherwise add FALLTHROUGH comment.
Closes: 50598
Fixed loading of ACPI tables on EFI (side effect was apparent memory
corruption ranging from unpredictable behavior to system reset).
Reported by Nando Eva <nando4eva@ymail.com>
iPXE adds Simple File System Protocol to loaded image handle, as side
effect it also adds Block IO protocol (according to comments, to work
around some bugs in EDK2). GRUB assumes that every device with Block IO
is disk and skips network initialization entirely. But iPXE Block IO
implementation is just a stub which always fails for every operation
so cannot be used. Attempt to detect and skip such devices.
We are using media ID which iPXE sets to "iPXE" and block IO size in
hope that no real device would announce 1B block ...
Closes: 50518
1. Make sure files are not multiple of block size. This will ensure tail packing
for squash4 and may also trigger more codes paths in other filesystems.
2. Call mksquashfs with -always-use-fragments to force tail packing.
UEFI 2.6 9.3.6.4 File Path Media Device Path says that Path Name is
"A NULL-terminated Path string including directory and file names".
Strip final NULL from Path Name in each File Path node when constructing
full path. To be on safe side, strip all of them.
Fixes failure chainloading grub from grub, when loaded grub truncates
image path and does not find its grub.cfg.
https://bugzilla.opensuse.org/show_bug.cgi?id=1026344
This was triggered by commit ce95549cc54b5d6f494608a7c390dba3aab4fba7;
before it we built Path Name without trailing NULL, and apparently all
other bootloaders use single File Path node, thus not exposing this bug.
1. Do not assume block list and fragment are mutually exclusive. Squash
can pack file tail as fragment (unless -no-fragments is specified); so
check read offset and read either from block list or from fragments as
appropriate.
2. Support sparse files with zero blocks.
3. Fix fragment read - frag.offset is absolute fragment position,
not offset relative to ino.chunk.
Reported and tested by Carlo Caione <carlo@endlessm.com>
These entries have placeholder for device name and so are useless for our
purpose. grub failed with something like
grub-install: error: failed to get canonical path of `systemd-1'.
When we see autofs entry, record it (to keep parent-child relationship) but
continue to look for real mount. If it is found, we process it as usual. If
only autofs entry exists, attempt to trigger mount by opening mount point
and retry. Mount point itself is then kept open to avoid timeout.
Recent systemd is by default using automount for /boot/efi so this should
become more popular problem on EFI systems.
Closes: 49942
glibc 2.25 still includes sys/sysmacros.h in sys/types.h but also emits
deprecation warning. So test for sys/types.h succeeds in configure but later
compilation fails because we use -Werror by default.
While this is fixed in current autoconf GIT, we really cannot force everyone
to use bleeding edge (that is not even released right now). So run test under
-Werror as well to force proper detection.
This should have no impact on autoconf 2.70+ as AC_HEADER_MAJOR in this version
simply checks for header existence.
Reported and tested by Khem Raj <raj.khem@gmail.com>
EFI File Path Media Device Path is defined as NULL terminated string;
but chainloader built file paths without final NULL. This caused error
with Secure Boot and Linux Foundation PreLoader on Acer with InsydeH20 BIOS.
Apparently firmware failed verification with EFI_INVALID_PARAMETER which is
considered fatal error by PreLoader.
Reported and tested by Giovanni Santini <itachi.sama.amaterasu@gmail.com>
Network boot autoconfiguration sets default server to next server IP
(siaddr) from BOOTP/DHCP reply, but manual configuration using net_bootp
exports only server name. Unfortunately semantic of server name is not
clearly defined. BOOTP RFC 951 defines it only for client request, and
DHCP RFC 1541 only mentions it, without any implied usage. It looks like
this field is mostly empty in server replies.
Export next server IP as net_<interface>_next_server variable. This allows
grub configuration script to set $root/$prefix based on information obtained
by net_bootp.
Reported and tested by: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
Cc: nikunj@linux.vnet.ibm.com
v2: change variable name to net_<interface>_next_server as discussed on the list
LLVM 3.9 now emits short form of jump instructions, but it is still using
32 bit addresses for some movl instructions. Fortunately it was caught early:
clang ... boot/i386/pc/boot.S
clang -cc1as: fatal error: error in backend: invalid .org offset '440' (at offset '441')
Add additional check to catch it during configure run and force -no-integrated-as.
Closes: 49200
More details in
https://lists.gnu.org/archive/html/grub-devel/2015-02/msg00099.htmlhttps://llvm.org/bugs/show_bug.cgi?id=22662
We may get more than one response before exiting out of loop in
grub_net_dns_lookup, but buffer was allocated for the first response only,
so storing answers from subsequent replies wrote past allocated size.
We never really use more than the very first address during lookup so there
is little point in collecting all of them. Just quit early if we already have
some reply.
Code needs serious redesign to actually collect multiple answers
and select the best fit according to requested type (IPv4 or IPv6).
Reported and tested by Michael Chang <mchang@suse.com>
XFS V5 stores UUID in metadata and compares them with superblock UUID.
To allow changing of user-visible UUID it stores original value in new
superblock field (meta_uuid) and sets incompatible flag to indicate that
new field must be used to verify metadata. Our driver currently does not
check metadata UUID so simply accept such filesystem.
Reported-By: Marcos Mello <marcosfrm@outlook.com>
Reviewd by Jan Kara <jack@suse.cz>
If we open new connection, we need to reset stall indication, otherwise
nothing will ever be polled (low level code rely on this field being
zero when establishing connection).
OpenBSD 5.9 apparently defaults to -fpie. We use -fno-PIE when appropriate
already, but that is not enough - it does not turn off -fpie.
Actually check for -fPIE is not precise enough. __PIE__ is set for both
-fpie and -fPIE but with different values. As far as I can tell, both
options were introduced at the same time, so both should always be supported.
This fixes compilation on OpenBSD 5.9 which otherwise created insanely big
lzma_decompress.img.
Reported, suggested and tested by: Jiri B <jirib@devio.us>
Map EFI_NO_MEDIA to GRUB_ERR_OUT_OF_RANGE that is ignored by diskfilter. This
actually matches pretty close (we obviously attempt to read outside of media)
and avoids adding more error codes.
This affects only internally initiated scans. If read/write from removable is
explicitly requested, we still return an error and text explanation is more
clear for user than generic error.
Reported and tested by Andreas Loew <Andreas.Loew@gmx.net>
all_video module does not have any code or data and exists solely for
.moddeps section to pull in dependencies. This makes all symbols unneeded.
While in current binutils (last released version as of this commit is 2.26)
``strip --strip-unneeded'' unintentionally adds section symbols for each
existing section, this behavior was considered a bug and changed in commit
14f2c699ddca1e2f706342dffc59a6c7e23e844c to completely strip symbol table
in this case.
Older binutils (verified with 2.17) and some other toolchains (at least
elftoolchain r3223M), both used in FreeBSD, remove symbol table in all_video
as well.
Relax run-time check and do not return error for modules without symbol table.
Add additional checks to module verifier to make sure such modules
a) have non-empty .moddeps section. Without either externally visible symbols
or .moddeps modules are completely useless and should not be built.
b) do not have any relocations.
Closes: 46986
v2: add run-time check for empty symbol table if relocations are present as
suggested by Vladimir.
If root filesystem is multidev btrfs, do not attempt to pass all devices as
kernel root= argument. This results in splitting command line in GRUB due to
embedded newline and even if we managed to quote it, kernel does not know how
to interpret it anyway. Multidev btrfs requires user space device scanning,
so passing single device would not work too.
This still respects user settings GRUB_DISABLE_LINUX_UUID. Not sure what we
should do in this case.
Closes: 45709
server cannot be NULL at this point (we return error earlier if it is).
Also structure is zalloc'ed, so no need to explicitly initialize
members to 0.
Found by: Coverity scan.
CID: 73837
Condition was apparently reversed so GRUB assumed all devices were
files. This later made it skip BLKFLSBUF ioctl on Linux which caused
various page cache coherency issues. Observed were
- failure to validate blocklist install (read content did not match
just written)
- failure to detect Linux MD on disk after online hot addition
(GRUB got stale superblock)
Closes: 46691
grub_xfs_iterate_dir did not restore first character after inline
name when match was found. Dependning on XFS format this character
could be inode number and we could return to the same node later in
find_file if processing cycled symlinks.
CID: 86724
If line contains single word, line and argv[0] are aliases, so
no NULL dereference is possible, but Coverity does not know it.
Change code to avoid ambiguity and also remove redundant call to
grub_strchr.
CID: 86725
Currently only Windows gets distinguished icons, everything else is displayed
using the same generic one. Add additional --class based on os-prober returned
label, which usually is expected to match primary distribution name.
Also use it for Windows as well - chainloader prober may actually return
different strings (Windows, MS-DOS, Windows9xME).
This was lost when code was refactored. Patch restores previous behavior.
It is still not clear whether this is the right one. Due to the way we
detect DM abstraction, partitions on DM are skipped, we fall through to
generic detection which ends up in assuming parent device is BIOS disk.
It is useful to install GRUB on VM disk from the host. But it also means
that GRUB will mistakenly allow install on real system as well.
For now let's fix regression; future behavior needs to be discussed.
Closes: 45163
GRUB keeps partition offset and size in units of 512B sectors. Media paths
are defined in terms of LBA which are presumed to match HDD sector size.
This is probably cosmetic (EFI requires that partition is searched by GUID)
and still incorrect if GPT was created using different logical block size.
But current code is obviously wrong and new has better chances to be correct.
Otherwise it causes subsequent file open to fail, because grub_file_open
misinterprets set grub_errno for grub_file_get_device_name failure.
Closes: 46540
gnulib files are already handled by recursive make distdir invocation.
Including all generated headers (after make completed) causes build
failure if target system is different (different compile version etc).
Macports add extra information after version itself:
$flex --version
flex 2.5.35 Apple(flex-31)
We require at least felx 2.5.35 so do not need to care about prehistoric
"flex version n.n.n"; just use second field always.
Reported by Peter Cheung <mcheung63@hotmail.com>
9be4c45dbe added switch case between
fall through cases, causing all memory regions of unknown type to be
marked as available.
Move default case into its own block and add explicit FALLTHROUGH
annotation.
Reported by Elliott, Robert (Persistent Memory) <elliott@hpe.com>
This makes it impossible to read from stdin without controlling tty:
10:/mnt # echo -e passwd\\npasswd | setsid ./grub-mkpasswd-pbkdf2
Enter password:
Reenter password: ./grub-mkpasswd-pbkdf2: error: failure to read password.
10:/mnt
Andrei Borzenkov