kumquat-buildroot/boot/grub2/0105-fs-hfsplus-Don-t-use-uninitialized-data-on-corrupt-f.patch

109 lines
4.4 KiB
Diff
Raw Normal View History

boot/grub2: Backport 2021/03/02 securify fixes Details: https://lists.gnu.org/archive/html/grub-devel/2021-03/msg00007.html As detailed in commit 7e64a050fbd9add07ed84d48054ffee1b659d079, it is difficult to utilize the upstream patches directly, so a number of patches include changes to generated files so that we don't need invoke the gentpl.py script. In addition to the security fixes, these required patches has been backported: f76a27996 efi: Make shim_lock GUID and protocol type public 04ae030d0 efi: Return grub_efi_status_t from grub_efi_get_variable() ac5c93675 efi: Add a function to read EFI variables with attributes d7e54b2e5 efi: Add secure boot detection The following security issues are fixed: CVE-2020-14372 grub2: The acpi command allows privileged user to load crafted ACPI tables when Secure Boot is enabled CWE-184 7.5/CVSS:3.1/AV:L/AC:H/PR:H/UI:N/S:C/C:H/I:H/A:H GRUB2 enables the use of the command acpi even when Secure Boot is signaled by the firmware. An attacker with local root privileges to can drop a small SSDT in /boot/efi and modify grub.cfg to instruct grub to load said SSDT. The SSDT then gets run by the kernel and it overwrites the kernel lock down configuration enabling the attacker to load unsigned kernel modules and kexec unsigned code. Reported-by: Máté Kukri ******************************************************************************* CVE-2020-25632 grub2: Use-after-free in rmmod command CWE-416 7.5/CVSS:3.1/AV:L/AC:H/PR:H/UI:N/S:C/C:H/I:H/A:H The rmmod implementation for GRUB2 is flawed, allowing an attacker to unload a module used as dependency without checking if any other dependent module is still loaded. This leads to an use-after-free scenario possibly allowing an attacker to execute arbitrary code and by-pass Secure Boot protections. Reported-by: Chris Coulson (Canonical) ******************************************************************************* CVE-2020-25647 grub2: Out-of-bound write in grub_usb_device_initialize() CWE-787 6.9/CVSS:3.1/AV:P/AC:H/PR:H/UI:N/S:C/C:H/I:H/A:H grub_usb_device_initialize() is called to handle USB device initialization. It reads out the descriptors it needs from the USB device and uses that data to fill in some USB data structures. grub_usb_device_initialize() performs very little bounds checking and simply assumes the USB device provides sane values. This behavior can trigger memory corruption. If properly exploited, this would lead to arbitrary code execution allowing the attacker to by-pass Secure Boot mechanism. Reported-by: Joseph Tartaro (IOActive) and Ilja van Sprundel (IOActive) ******************************************************************************* CVE-2020-27749 grub2: Stack buffer overflow in grub_parser_split_cmdline CWE-121 7.5/CVSS:3.1/AV:L/AC:H/PR:H/UI:N/S:C/C:H/I:H/A:H grub_parser_split_cmdline() expands variable names present in the supplied command line in to their corresponding variable contents and uses a 1kB stack buffer for temporary storage without sufficient bounds checking. If the function is called with a command line that references a variable with a sufficiently large payload, it is possible to overflow the stack buffer, corrupt the stack frame and control execution. An attacker may use this to circumvent Secure Boot protections. Reported-by: Chris Coulson (Canonical) ******************************************************************************* CVE-2020-27779 grub2: The cutmem command allows privileged user to remove memory regions when Secure Boot is enabled CWE-285 7.5/CVSS:3.1/AV:L/AC:H/PR:H/UI:N/S:C/C:H/I:H/A:H The GRUB2's cutmem command does not honor Secure Boot locking. This allows an privileged attacker to remove address ranges from memory creating an opportunity to circumvent Secure Boot protections after proper triage about grub's memory layout. Reported-by: Teddy Reed ******************************************************************************* CVE-2021-3418 - grub2: GRUB 2.05 reintroduced CVE-2020-15705 CWE-281 6.4/CVSS:3.1/AV:L/AC:H/PR:H/UI:N/S:U/C:H/I:H/A:H The GRUB2 upstream reintroduced the CVE-2020-15705. This refers to a distro specific flaw which made upstream in the mentioned version. If certificates that signed GRUB2 are installed into db, GRUB2 can be booted directly. It will then boot any kernel without signature validation. The booted kernel will think it was booted in Secure Boot mode and will implement lock down, yet it could have been tampered. This flaw only affects upstream and distributions using the shim_lock verifier. Reported-by: Dimitri John Ledkov (Canonical) ******************************************************************************* CVE-2021-20225 grub2: Heap out-of-bounds write in short form option parser CWE-787 7.5/CVSS:3.1/AV:L/AC:H/PR:H/UI:N/S:C/C:H/I:H/A:H The option parser in GRUB2 allows an attacker to write past the end of a heap-allocated buffer by calling certain commands with a large number of specific short forms of options. Reported-by: Daniel Axtens (IBM) ******************************************************************************* CVE-2021-20233 grub2: Heap out-of-bound write due to mis-calculation of space required for quoting CWE-787 7.5/CVSS:3.1/AV:L/AC:H/PR:H/UI:N/S:C/C:H/I:H/A:H There's a flaw on GRUB2 menu rendering code setparam_prefix() in the menu rendering code performs a length calculation on the assumption that expressing a quoted single quote will require 3 characters, while it actually requires 4 characters. This allow an attacker to corrupt memory by one byte for each quote in the input. Reported-by: Daniel Axtens (IBM) ******************************************************************************* Signed-off-by: Stefan Sørensen <stefan.sorensen@spectralink.com> Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
2021-03-11 10:53:06 +01:00
From 2ca0e5dbcdcb6fc93ccae39a0f39d0dba4a7ff20 Mon Sep 17 00:00:00 2001
From: Daniel Axtens <dja@axtens.net>
Date: Tue, 2 Feb 2021 16:59:35 +1100
Subject: [PATCH] fs/hfsplus: Don't use uninitialized data on corrupt
filesystems
Valgrind identified the following use of uninitialized data:
==2782220== Conditional jump or move depends on uninitialised value(s)
==2782220== at 0x42B364: grub_hfsplus_btree_search (hfsplus.c:566)
==2782220== by 0x42B21D: grub_hfsplus_read_block (hfsplus.c:185)
==2782220== by 0x42A693: grub_fshelp_read_file (fshelp.c:386)
==2782220== by 0x42C598: grub_hfsplus_read_file (hfsplus.c:219)
==2782220== by 0x42C598: grub_hfsplus_mount (hfsplus.c:330)
==2782220== by 0x42B8C5: grub_hfsplus_dir (hfsplus.c:958)
==2782220== by 0x4C1AE6: grub_fs_probe (fs.c:73)
==2782220== by 0x407C94: grub_ls_list_files (ls.c:186)
==2782220== by 0x407C94: grub_cmd_ls (ls.c:284)
==2782220== by 0x4D7130: grub_extcmd_dispatcher (extcmd.c:55)
==2782220== by 0x4045A6: execute_command (grub-fstest.c:59)
==2782220== by 0x4045A6: fstest (grub-fstest.c:433)
==2782220== by 0x4045A6: main (grub-fstest.c:772)
==2782220== Uninitialised value was created by a heap allocation
==2782220== at 0x483C7F3: malloc (in /usr/lib/x86_64-linux-gnu/valgrind/vgpreload_memcheck-amd64-linux.so)
==2782220== by 0x4C0305: grub_malloc (mm.c:42)
==2782220== by 0x42C21D: grub_hfsplus_mount (hfsplus.c:239)
==2782220== by 0x42B8C5: grub_hfsplus_dir (hfsplus.c:958)
==2782220== by 0x4C1AE6: grub_fs_probe (fs.c:73)
==2782220== by 0x407C94: grub_ls_list_files (ls.c:186)
==2782220== by 0x407C94: grub_cmd_ls (ls.c:284)
==2782220== by 0x4D7130: grub_extcmd_dispatcher (extcmd.c:55)
==2782220== by 0x4045A6: execute_command (grub-fstest.c:59)
==2782220== by 0x4045A6: fstest (grub-fstest.c:433)
==2782220== by 0x4045A6: main (grub-fstest.c:772)
This happens when the process of reading the catalog file goes sufficiently
wrong that there's an attempt to read the extent overflow file, which has
not yet been loaded. Keep track of when the extent overflow file is
fully loaded and refuse to use it before then.
The load valgrind doesn't like is btree->nodesize, and that's then used
to allocate a data structure. It looks like there are subsequently a lot
of reads based on that pointer so OOB reads are likely, and indeed crashes
(albeit difficult-to-replicate ones) have been observed in fuzzing.
Signed-off-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Signed-off-by: Stefan Sørensen <stefan.sorensen@spectralink.com>
---
grub-core/fs/hfsplus.c | 14 ++++++++++++++
include/grub/hfsplus.h | 2 ++
2 files changed, 16 insertions(+)
diff --git a/grub-core/fs/hfsplus.c b/grub-core/fs/hfsplus.c
index 1c7791b..361e5be 100644
--- a/grub-core/fs/hfsplus.c
+++ b/grub-core/fs/hfsplus.c
@@ -177,6 +177,17 @@ grub_hfsplus_read_block (grub_fshelp_node_t node, grub_disk_addr_t fileblock)
break;
}
+ /*
+ * If the extent overflow tree isn't ready yet, we can't look
+ * in it. This can happen where the catalog file is corrupted.
+ */
+ if (!node->data->extoverflow_tree_ready)
+ {
+ grub_error (GRUB_ERR_BAD_FS,
+ "attempted to read extent overflow tree before loading");
+ break;
+ }
+
/* Set up the key to look for in the extent overflow file. */
extoverflow.extkey.fileid = node->fileid;
extoverflow.extkey.type = 0;
@@ -241,6 +252,7 @@ grub_hfsplus_mount (grub_disk_t disk)
return 0;
data->disk = disk;
+ data->extoverflow_tree_ready = 0;
/* Read the bootblock. */
grub_disk_read (disk, GRUB_HFSPLUS_SBLOCK, 0, sizeof (volheader),
@@ -357,6 +369,8 @@ grub_hfsplus_mount (grub_disk_t disk)
if (data->extoverflow_tree.nodesize < 2)
goto fail;
+ data->extoverflow_tree_ready = 1;
+
if (grub_hfsplus_read_file (&data->attr_tree.file, 0, 0,
sizeof (struct grub_hfsplus_btnode),
sizeof (header), (char *) &header) <= 0)
diff --git a/include/grub/hfsplus.h b/include/grub/hfsplus.h
index 117740a..e14dd31 100644
--- a/include/grub/hfsplus.h
+++ b/include/grub/hfsplus.h
@@ -113,6 +113,8 @@ struct grub_hfsplus_data
struct grub_hfsplus_btree extoverflow_tree;
struct grub_hfsplus_btree attr_tree;
+ int extoverflow_tree_ready;
+
struct grub_hfsplus_file dirroot;
struct grub_hfsplus_file opened_file;
--
2.14.2