A netbuff shouldn't be too huge. It's bounded by MTU and TCP segment
reassembly.
This helps avoid some bugs (and provides a spot to instrument to catch
them at their source).
Signed-off-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
In some cases attempting to display arbitrary binary strings leads
to ASAN splats reading the widthspec array out of bounds.
Check the index. If it would be out of bounds, return a width of 1.
I don't know if that's strictly correct, but we're not really expecting
great display of arbitrary binary data, and it's certainly not worse than
an OOB read.
Signed-off-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Certain 1 px wide images caused a wild pointer write in
grub_jpeg_ycrcb_to_rgb(). This was caused because in grub_jpeg_decode_data(),
we have the following loop:
for (; data->r1 < nr1 && (!data->dri || rst);
data->r1++, data->bitmap_ptr += (vb * data->image_width - hb * nc1) * 3)
We did not check if vb * width >= hb * nc1.
On a 64-bit platform, if that turns out to be negative, it will underflow,
be interpreted as unsigned 64-bit, then be added to the 64-bit pointer, so
we see data->bitmap_ptr jump, e.g.:
0x6180_0000_0480 to
0x6181_0000_0498
^
~--- carry has occurred and this pointer is now far away from
any object.
On a 32-bit platform, it will decrement the pointer, creating a pointer
that won't crash but will overwrite random data.
Catch the underflow and error out.
Fixes: CVE-2021-3697
Signed-off-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
An invalid file could contain multiple start of stream blocks, which
would cause us to reallocate and leak our bitmap. Refuse to handle
multiple start of streams.
Additionally, fix a grub_error() call formatting.
Signed-off-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Fix a memory leak where an invalid file could cause us to reallocate
memory for a huffman table we had already allocated memory for.
Signed-off-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Fuzzing revealed some inputs that were taking a long time, potentially
forever, because they did not bail quickly upon encountering an I/O error.
Try to catch I/O errors sooner and bail out.
Signed-off-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
ASAN picked up two OOB global reads: we weren't checking if some code
values fit within the cplens or cpdext arrays. Check and throw an error
if not.
Signed-off-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
In fuzzing we observed crashes where a code would attempt to be inserted
into a huffman table before the start, leading to a set of heap OOB reads
and writes as table entries with negative indices were shifted around and
the new code written in.
Catch the case where we would underflow the array and bail.
Fixes: CVE-2021-3696
Signed-off-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
A 16-bit greyscale PNG without alpha is processed in the following loop:
for (i = 0; i < (data->image_width * data->image_height);
i++, d1 += 4, d2 += 2)
{
d1[R3] = d2[1];
d1[G3] = d2[1];
d1[B3] = d2[1];
}
The increment of d1 is wrong. d1 is incremented by 4 bytes per iteration,
but there are only 3 bytes allocated for storage. This means that image
data will overwrite somewhat-attacker-controlled parts of memory - 3 bytes
out of every 4 following the end of the image.
This has existed since greyscale support was added in 2013 in commit
3ccf16dff9 (grub-core/video/readers/png.c: Support grayscale).
Saving starfield.png as a 16-bit greyscale image without alpha in the gimp
and attempting to load it causes grub-emu to crash - I don't think this code
has ever worked.
Delete all PNG greyscale support.
Fixes: CVE-2021-3695
Signed-off-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
This causes the bitmap to be leaked. Do not permit multiple image headers.
Signed-off-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Fuzzing revealed some inputs that were taking a long time, potentially
forever, because they did not bail quickly upon encountering an I/O error.
Try to catch I/O errors sooner and bail out.
Signed-off-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
If we have an error in grub_file_open() before we free device_name, we
will leak it.
Free device_name in the error path and null out the pointer in the good
path once we free it there.
Signed-off-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
We must not allow other verifiers to pass things like the GRUB modules.
Instead of maintaining a blocklist, maintain an allowlist of things
that we do not care about.
This allowlist really should be made reusable, and shared by the
lockdown verifier, but this is the minimal patch addressing
security concerns where the TPM verifier was able to mark modules
as verified (or the OpenPGP verifier for that matter), when it
should not do so on shim-powered secure boot systems.
Fixes: CVE-2022-28735
Signed-off-by: Julian Andres Klode <julian.klode@canonical.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
This ports the EFI chainloader to use grub_loader_set_ex() in order to fix
a use-after-free bug that occurs when grub_cmd_chainloader() is executed
more than once before a boot attempt is performed.
Fixes: CVE-2022-28736
Signed-off-by: Chris Coulson <chris.coulson@canonical.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Loaders rely on global variables for saving context which is consumed
in the boot hook and freed in the unload hook. In the case where a loader
command is executed twice, calling grub_loader_set a second time executes
the unload hook, but in some cases this runs when the loader's global
context has already been updated, resulting in the updated context being
freed and potential use-after-free bugs when the boot hook is subsequently
called.
This adds a new API (grub_loader_set_ex) which allows a loader to specify
context that is passed to its boot and unload hooks. This is an alternative
to requiring that loaders call grub_loader_unset before mutating their
global context.
Signed-off-by: Chris Coulson <chris.coulson@canonical.com>
(cherry picked from commit 4322a64dde7e8fedb58e50b79408667129d45dd3)
The chainloader command retains the source buffer and device path passed
to LoadImage(), requiring the unload hook passed to grub_loader_set() to
free them. It isn't required to retain this state though - they aren't
required by StartImage() or anything else in the boot hook, so clean them
up before grub_cmd_chainloader() finishes.
Signed-off-by: Chris Coulson <chris.coulson@canonical.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
minilzo fails to build on a number of Debian release architectures
(armel, mips64el, mipsel, ppc64el) with errors such as:
../../grub-core/lib/minilzo/minilzo.c: In function 'lzo_memops_get_le16':
../../grub-core/lib/minilzo/minilzo.c:3479:11: error: dereferencing type-punned pointer will break strict-aliasing rules [-Werror=strict-aliasing]
3479 | * (lzo_memops_TU2p) (lzo_memops_TU0p) (dd) = * (const lzo_memops_TU2p) (const lzo_memops_TU0p) (ss); \
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../../grub-core/lib/minilzo/minilzo.c:3530:5: note: in expansion of macro 'LZO_MEMOPS_COPY2'
3530 | LZO_MEMOPS_COPY2(&v, ss);
| ^~~~~~~~~~~~~~~~
The latest upstream version is 2.10, so updating to it seems like a good
idea on general principles, and it fixes builds on all the above
architectures.
The update procedure documented in the GRUB Developers Manual worked; I
just updated the version numbers to make it clear that it's been
executed recently.
Signed-off-by: Colin Watson <cjwatson@debian.org>
Forwarded: https://lists.gnu.org/archive/html/grub-devel/2021-11/msg00095.html
Last-Update: 2021-11-29
Patch-Name: minilzo-2.10.patch
The "ide-drive" device was removed in QEMU 6.0. The "ide-hd" has been
available for more than 10 years now in QEMU. Thus there shouldn't be
any need for backwards compatible names.
Signed-off-by: Marius Bakke <marius@gnu.org>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Origin: upstream, https://git.savannah.gnu.org/cgit/grub.git/commit/?id=aaea244a6ddd1e35aed60a5c7a08ddc41f51805b
Last-Update: 2021-09-24
Patch-Name: tests-ahci-update-qemu-device-name.patch
The commit 8b1e5d193 (fs/xfs: Add bigtime incompat feature support)
introduced the bigtime support by adding some features in v3 inodes.
This change extended grub_xfs_inode struct by 76 bytes but also changed
the computation of XFS_V2_INODE_SIZE and XFS_V3_INODE_SIZE. Prior this
commit, XFS_V2_INODE_SIZE was 100 bytes. After the commit it's 84 bytes
XFS_V2_INODE_SIZE becomes 16 bytes too small.
As a result, the data structures aren't properly aligned and the GRUB
generates "attempt to read or write outside of partition" errors when
trying to read the XFS filesystem:
GNU GRUB version 2.11
....
grub> set debug=efi,gpt,xfs
grub> insmod part_gpt
grub> ls (hd0,gpt1)/
partmap/gpt.c:93: Read a valid GPT header
partmap/gpt.c:115: GPT entry 0: start=4096, length=1953125
fs/xfs.c:931: Reading sb
fs/xfs.c:270: Validating superblock
fs/xfs.c:295: XFS v4 superblock detected
fs/xfs.c:962: Reading root ino 128
fs/xfs.c:515: Reading inode (128) - 64, 0
fs/xfs.c:515: Reading inode (739521961424144223) - 344365866970255880, 3840
error: attempt to read or write outside of partition.
This commit change the XFS_V2_INODE_SIZE computation by subtracting 76
bytes instead of 92 bytes from the actual size of grub_xfs_inode struct.
This 76 bytes value comes from added members:
20 grub_uint8_t unused5
1 grub_uint64_t flags2
48 grub_uint8_t unused6
This patch explicitly splits the v2 and v3 parts of the structure.
The unused4 is still ending of the v2 structures and the v3 starts
at unused5. Thanks to this we will avoid future corruptions of v2
or v3 inodes.
The XFS_V2_INODE_SIZE is returning to its expected size and the
filesystem is back to a readable state:
GNU GRUB version 2.11
....
grub> set debug=efi,gpt,xfs
grub> insmod part_gpt
grub> ls (hd0,gpt1)/
partmap/gpt.c:93: Read a valid GPT header
partmap/gpt.c:115: GPT entry 0: start=4096, length=1953125
fs/xfs.c:931: Reading sb
fs/xfs.c:270: Validating superblock
fs/xfs.c:295: XFS v4 superblock detected
fs/xfs.c:962: Reading root ino 128
fs/xfs.c:515: Reading inode (128) - 64, 0
fs/xfs.c:515: Reading inode (128) - 64, 0
fs/xfs.c:931: Reading sb
fs/xfs.c:270: Validating superblock
fs/xfs.c:295: XFS v4 superblock detected
fs/xfs.c:962: Reading root ino 128
fs/xfs.c:515: Reading inode (128) - 64, 0
fs/xfs.c:515: Reading inode (128) - 64, 0
fs/xfs.c:515: Reading inode (128) - 64, 0
fs/xfs.c:515: Reading inode (131) - 64, 768
efi/ fs/xfs.c:515: Reading inode (3145856) - 1464904, 0
grub2/ fs/xfs.c:515: Reading inode (132) - 64, 1024
grub/ fs/xfs.c:515: Reading inode (139) - 64, 2816
grub>
Fixes: 8b1e5d193 (fs/xfs: Add bigtime incompat feature support)
Signed-off-by: Erwan Velu <e.velu@criteo.com>
Tested-by: Carlos Maiolino <cmaiolino@redhat.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Origin: upstream, https://git.savannah.gnu.org/cgit/grub.git/commit/?id=a4b495520e4dc41a896a8b916a64eda9970c50ea
Last-Update: 2021-09-24
Patch-Name: xfs-fix-v4-superblock.patch
Commit f60ba9e594 (util/mkimage: Refactor section setup to use a helper)
added a helper function to setup PE sections, but it caused regressions
in some arches where the natural alignment lead to wrong section sizes.
This patch fixes a few things that were caused the section sizes to be
calculated wrongly. These fixes are:
* Only align the virtual memory addresses but not the raw data offsets.
* Use aligned sizes for virtual memory sizes but not for raw data sizes.
* Always align the sizes to set the virtual memory sizes.
These seems to not cause problems for x64 and aa64 EFI platforms but was
a problem for ia64. Because the size of the ".data" and "mods" sections
were wrong and didn't have the correct content. Which lead to GRUB not
being able to load any built-in module.
Reported-by: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Bug-Debian: https://bugs.debian.org/987103
Patch-Name: mkimage-fix-section-sizes.patch
Given no core functions on i386-pc would require verifiers to work and
the only consumer of the verifier API is the pgp module, it looks good
to me that we can move the verifiers out of the kernel image and let
moddep.lst to auto-load it when pgp is loaded on i386-pc platform.
This helps to reduce the size of core image and thus can relax the
tension of exploding on some i386-pc system with very short MBR gap
size. See also a very comprehensive summary from Colin [1] about the
details.
[1] https://lists.gnu.org/archive/html/grub-devel/2021-03/msg00240.html
V2:
Drop COND_NOT_i386_pc and use !COND_i386_pc.
Add comment in kern/verifiers.c to help understanding what's going on
without digging into the commit history.
Reported-by: Colin Watson <cjwatson@debian.org>
Reviewed-by: Colin Watson <cjwatson@debian.org>
Signed-off-by: Michael Chang <mchang@suse.com>
Origin: other, https://lists.gnu.org/archive/html/grub-devel/2021-03/msg00251.html
Bug-Debian: https://bugs.debian.org/984488
Bug-Debian: https://bugs.debian.org/985374
Last-Update: 2021-09-24
Patch-Name: pc-verifiers-module.patch
For complicated reasons, even if you have XSM/FLASK disabled (as is
the default) the Xen build system still builds a policy file and puts
it in /boot.
Even so, we shouldn't be loading this in the usual non-"XSM enabled"
entries. It doesn't do any particular harm but it is quite confusing.
Signed-off-by: Ian Jackson <ian.jackson@eu.citrix.com>
Bug-Debian: https://bugs.debian.org/961673
Last-Update: 2020-05-29
Patch-Name: xen-no-xsm-policy-in-non-xsm-options.patch
In this case, we need both the signed shim as /EFI/BOOT/BOOTXXX.EFI
and signed Grub as /EFI/BOOT/grubXXX.efi.
Also install the BOOTXXX.CSV into /EFI/debian, and FBXXX.EFI into
/EFI/BOOT/ so that it can work when needed (*iff* we're updating the
NVRAM).
[cjwatson: Refactored also_install_removable somewhat for brevity and so
that we're using consistent case-insensitive logic.]
Bug-Debian: https://bugs.debian.org/930531
Last-Update: 2021-09-24
Patch-Name: grub-install-removable-shim.patch
Some UEFI firmware is easily provoked into running out of space in its
variable storage. This is usually due to certain kernel drivers (e.g.
pstore), but regardless of the cause it can cause grub-install to fail
because it currently asks efibootmgr to delete and re-add entries, and
the deletion often doesn't result in an immediate garbage collection.
Writing variables frequently also increases wear on the NVRAM which may
have limited write cycles. For these reasons, it's desirable to find a
way to minimise writes while still allowing grub-install to ensure that
a suitable boot entry exists.
Unfortunately, efibootmgr doesn't offer an interface that would let
grub-install do this. It doesn't in general make very much effort to
minimise writes; it doesn't allow modifying an existing Boot* variable
entry, except in certain limited ways; and current versions don't have a
way to export the expected variable data so that grub-install can
compare it to the current data. While it would be possible (and perhaps
desirable?) to add at least some of this to efibootmgr, that would still
leave the problem that there isn't a good upstreamable way for
grub-install to guarantee that it has a new enough version of
efibootmgr. In any case, it's cumbersome and slow for grub-install to
have to fork efibootmgr to get things done.
Fortunately, a few years ago Peter Jones helpfully factored out a
substantial part of efibootmgr to the efivar and efiboot libraries, and
so it's now possible to have grub-install use those directly. We still
have to use some code from efibootmgr, but much less than would
previously have been necessary.
grub-install now reuses existing boot entries where possible, and avoids
writing to variables when the new contents are the same as the old
contents. In the common upgrade case where nothing needs to change, it
no longer writes to NVRAM at all. It's also now slightly faster, since
using libefivar is faster than forking efibootmgr.
Fixes Debian bug #891434.
Signed-off-by: Colin Watson <cjwatson@ubuntu.com>
Bug-Debian: https://bugs.debian.org/891434
Forwarded: https://lists.gnu.org/archive/html/grub-devel/2019-03/msg00119.html
Last-Update: 2019-03-23
Patch-Name: efi-variable-storage-minimise-writes.patch
On full-encrypted systems, including /boot, the current code omits
cryptodisk commands needed to open the drives if Secure Boot is enabled.
This prevents grub2 from reading any further configuration residing on
the encrypted disk.
This patch fixes this issue by adding the needed "cryptomount" commands in
the load.cfg file that is then copied in the EFI partition.
Bug-Debian: https://bugs.debian.org/917117
Last-Update: 2019-02-10
Patch-Name: uefi-secure-boot-cryptomount.patch
