summaryrefslogtreecommitdiff
path: root/util
diff options
context:
space:
mode:
authorNicola Corna <nicola@corna.info>2017-01-23 15:28:24 +0100
committerMartin Roth <martinroth@google.com>2017-01-25 18:19:58 +0100
commit9bcc002f1e7958c7b4234ea804b5b2b4f675ea3a (patch)
treef1c8749c75b647287ef293da7427f6b46858f586 /util
parentb144a34c6031ef988d32946129920672d29d38a9 (diff)
downloadcoreboot-9bcc002f1e7958c7b4234ea804b5b2b4f675ea3a.tar.xz
util: Add me_cleaner
me_cleaner is a tool to strip down Intel ME/TXE images by removing all the non-fundamental code, while keeping the ME/TXE image valid and suitable for booting the system. The remaining code (ROMP and BUP modules) is the one responsible for the very basic initialization of the ME/TXE subsystem and can't be removed. This tool exploits the fact that: * Each ME/TXE partition is signed individually and it is possible to remove both the partition and the signature. * The ME/TXE modules are not signed directly, instead they are hashed and the list of their hashes is hashed again and signed: this means that modifying a module doesn't invalidate the signature, but only the hash of that single module. * The modules hashes are checked only when the corresponding module needs to be executed. * The system can boot after the execution of the first module (BUP, inside the FTPR partition), even if the subsequent stages fail. Currently me_cleaner works on every Intel platform with Intel ME or Intel TXE with the following limitations: * Doesn't work when Intel Boot Guard is set in Verified Boot mode. * Doesn't fully work on Nehalem yet. * On Skylake and later generations, since the partitions' internal structure has changed, me_cleaner leaves intact the FTPR partition, removing all the the other partitions. This tool has been tested on multiple platforms and architectures by different users, and seems to be stable. The reports are available here: https://github.com/corna/me_cleaner/issues/3 A more in-depth description of me_cleaner is available here: https://github.com/corna/me_cleaner/wiki/How-does-it-work%3F Change-Id: I9013799e9adea0dea0775b9afe718de5fc4ca748 Signed-off-by: Nicola Corna <nicola@corna.info> Reviewed-on: https://review.coreboot.org/18203 Tested-by: build bot (Jenkins) Reviewed-by: Philipp Deppenwiese <zaolin.daisuki@gmail.com> Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Diffstat (limited to 'util')
-rw-r--r--util/me_cleaner/README.md25
-rwxr-xr-xutil/me_cleaner/me_cleaner.py313
2 files changed, 338 insertions, 0 deletions
diff --git a/util/me_cleaner/README.md b/util/me_cleaner/README.md
new file mode 100644
index 0000000000..9a09b550eb
--- /dev/null
+++ b/util/me_cleaner/README.md
@@ -0,0 +1,25 @@
+# ME cleaner
+
+A cleaner for Intel ME/TXE images.
+
+This tools removes any unnecessary partition from an Intel ME/TXE firmware, reducing
+its size and its ability to interact with the system.
+It should work both with coreboot and with the factory firmware.
+
+Currently this tool:
+ * Scans the FPT (partition table) and checks that everything is correct
+ * Removes any partition entry (except for FTPR) from FPT
+ * Removes any partition except for the fundamental one (FTPR)
+ * Removes the EFFS presence flag
+ * Corrects the FPT checksum
+ * Removes any non-essential LZMA or Huffman compressed module (pre-Skylake only)
+ * Checks the validity of the RSA signature of the FTPR partition
+
+Don't forget to power cycle your PC after flashing the modified ME/TXE image
+(power off and power on, not just reboot).
+
+See the [current status](https://github.com/corna/me_cleaner/wiki/me_cleaner-status)
+or [a more detailed description](https://github.com/corna/me_cleaner/wiki/How-does-it-work%3F)
+of me_cleaner.
+
+Special thanks to Federico Amedeo Izzo for his help during the study of Intel ME.
diff --git a/util/me_cleaner/me_cleaner.py b/util/me_cleaner/me_cleaner.py
new file mode 100755
index 0000000000..2ad5898213
--- /dev/null
+++ b/util/me_cleaner/me_cleaner.py
@@ -0,0 +1,313 @@
+#!/usr/bin/python
+
+# me_cleaner - Tool for partial deblobbing of Intel ME/TXE firmware images
+# Copyright (C) 2016, 2017 Nicola Corna <nicola@corna.info>
+#
+# This program 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.
+#
+# This program 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.
+#
+
+import sys
+import itertools
+import binascii
+import hashlib
+from struct import pack, unpack
+
+
+unremovable_modules = ("BUP", "ROMP")
+
+
+def get_chunks_offsets(llut, me_start):
+ chunk_count = unpack("<I", llut[0x04:0x08])[0]
+ huffman_stream_end = sum(unpack("<II", llut[0x10:0x18])) + me_start
+ nonzero_offsets = [huffman_stream_end]
+ offsets = []
+
+ for i in range(0, chunk_count):
+ chunk = llut[0x40 + i * 4:0x44 + i * 4]
+ offset = 0
+
+ if chunk[3] != 0x80:
+ offset = unpack("<I", chunk[0:3] + b"\x00")[0] + me_start
+
+ offsets.append([offset, 0])
+ if offset != 0:
+ nonzero_offsets.append(offset)
+
+ nonzero_offsets.sort()
+
+ for i in offsets:
+ if i[0] != 0:
+ i[1] = nonzero_offsets[nonzero_offsets.index(i[0]) + 1]
+
+ return offsets
+
+
+def fill_range(f, start, end, fill):
+ block = fill * 4096
+ f.seek(start)
+ f.writelines(itertools.repeat(block, (end - start) // 4096))
+ f.write(block[:(end - start) % 4096])
+
+
+def remove_modules(f, mod_headers, ftpr_offset):
+ comp_str = ("Uncomp.", "Huffman", "LZMA")
+ unremovable_huff_chunks = []
+ chunks_offsets = []
+ base = 0
+ chunk_size = 0
+
+ for mod_header in mod_headers:
+ name = mod_header[0x04:0x14].rstrip(b"\x00").decode("ascii")
+ offset = unpack("<I", mod_header[0x38:0x3C])[0] + ftpr_offset
+ size = unpack("<I", mod_header[0x40:0x44])[0]
+ flags = unpack("<I", mod_header[0x50:0x54])[0]
+ comp_type = (flags >> 4) & 7
+
+ sys.stdout.write(" {:<16} ({:<7}, ".format(name, comp_str[comp_type]))
+
+ if comp_type == 0x00 or comp_type == 0x02:
+ sys.stdout.write("0x{:06x} - 0x{:06x}): "
+ .format(offset, offset + size))
+
+ if name in unremovable_modules:
+ print("NOT removed, essential")
+ else:
+ fill_range(f, offset, offset + size, b"\xff")
+ print("removed")
+
+ elif comp_type == 0x01:
+ sys.stdout.write("fragmented data ): ")
+ if not chunks_offsets:
+ f.seek(offset)
+ llut = f.read(4)
+ if llut == b"LLUT":
+ llut += f.read(0x3c)
+
+ chunk_count = unpack("<I", llut[0x4:0x8])[0]
+ base = unpack("<I", llut[0x8:0xc])[0] + 0x10000000
+ huff_data_len = unpack("<I", llut[0x10:0x14])[0]
+ chunk_size = unpack("<I", llut[0x30:0x34])[0]
+
+ llut += f.read(chunk_count * 4 + huff_data_len)
+ chunks_offsets = get_chunks_offsets(llut, me_start)
+ else:
+ sys.exit("Huffman modules found, but LLUT is not present")
+
+ if name in unremovable_modules:
+ print("NOT removed, essential")
+ module_base = unpack("<I", mod_header[0x34:0x38])[0]
+ module_size = unpack("<I", mod_header[0x3c:0x40])[0]
+ first_chunk_num = (module_base - base) // chunk_size
+ last_chunk_num = first_chunk_num + module_size // chunk_size
+
+ unremovable_huff_chunks += \
+ [x for x in chunks_offsets[first_chunk_num:
+ last_chunk_num + 1] if x[0] != 0]
+ else:
+ print("removed")
+
+ else:
+ sys.stdout.write("0x{:06x} - 0x{:06x}): unknown compression, "
+ "skipping".format(offset, offset + size))
+
+ if chunks_offsets:
+ removable_huff_chunks = []
+
+ for chunk in chunks_offsets:
+ if all(not(unremovable_chk[0] <= chunk[0] < unremovable_chk[1] or
+ unremovable_chk[0] < chunk[1] <= unremovable_chk[1])
+ for unremovable_chk in unremovable_huff_chunks):
+ removable_huff_chunks.append(chunk)
+
+ for removable_chunk in removable_huff_chunks:
+ if removable_chunk[1] > removable_chunk[0]:
+ fill_range(f, removable_chunk[0], removable_chunk[1], b"\xff")
+
+
+def check_partition_signature(f, offset):
+ f.seek(offset)
+ header = f.read(0x80)
+ modulus = int(binascii.hexlify(f.read(0x100)[::-1]), 16)
+ public_exponent = int(binascii.hexlify(f.read(0x4)[::-1]), 16)
+ signature = int(binascii.hexlify(f.read(0x100)[::-1]), 16)
+
+ header_len = unpack("<I", header[0x4:0x8])[0] * 4
+ manifest_len = unpack("<I", header[0x18:0x1c])[0] * 4
+ f.seek(offset + header_len)
+
+ sha256 = hashlib.sha256()
+ sha256.update(header)
+ sha256.update(f.read(manifest_len - header_len))
+
+ decrypted_sig = pow(signature, public_exponent, modulus)
+
+ return "{:#x}".format(decrypted_sig).endswith(sha256.hexdigest()) # FIXME
+
+
+if len(sys.argv) != 2 or sys.argv[1] == "-h" or sys.argv[1] == "--help":
+ print("Usage: \n"
+ " me_cleaner.py me_or_txe_image.bin\n"
+ "or\n"
+ " me_cleaner.py full_dump.bin")
+else:
+ with open(sys.argv[1], "r+b") as f:
+ f.seek(0x10)
+ magic = f.read(4)
+
+ if magic == b"$FPT":
+ print("ME/TXE image detected")
+ me_start = 0
+ f.seek(0, 2)
+ me_end = f.tell()
+
+ elif magic == b"\x5a\xa5\xf0\x0f":
+ print("Full image detected")
+ f.seek(0x14)
+ flmap0 = unpack("<I", f.read(4))[0]
+ nr = flmap0 >> 24 & 0x7
+ frba = flmap0 >> 12 & 0xff0
+ if nr >= 2:
+ f.seek(frba + 0x8)
+ flreg2 = unpack("<I", f.read(4))[0]
+ me_start = (flreg2 & 0x1fff) << 12
+ me_end = flreg2 >> 4 & 0x1fff000 | 0xfff
+
+ if me_start >= me_end:
+ sys.exit("The ME/TXE region in this image has been "
+ "disabled")
+
+ f.seek(me_start + 0x10)
+ if f.read(4) != b"$FPT":
+ sys.exit("The ME/TXE region is corrupted or missing")
+
+ print("The ME/TXE region goes from {:#x} to {:#x}"
+ .format(me_start, me_end))
+ else:
+ sys.exit("This image does not contains a ME/TXE firmware "
+ "(NR = {})".format(nr))
+ else:
+ sys.exit("Unknown image")
+
+ print("Found FPT header at {:#x}".format(me_start + 0x10))
+
+ f.seek(me_start + 0x14)
+ entries = unpack("<I", f.read(4))[0]
+ print("Found {} partition(s)".format(entries))
+
+ f.seek(me_start + 0x14)
+ header_len = unpack("B", f.read(1))[0]
+
+ f.seek(me_start + 0x30)
+ partitions = f.read(entries * 0x20)
+
+ ftpr_header = b""
+
+ for i in range(entries):
+ if partitions[i * 0x20:(i * 0x20) + 4] == b"FTPR":
+ ftpr_header = partitions[i * 0x20:(i + 1) * 0x20]
+ break
+
+ if ftpr_header == b"":
+ sys.exit("FTPR header not found, this image doesn't seem to be "
+ "valid")
+
+ ftpr_offset, ftpr_lenght = unpack("<II", ftpr_header[0x08:0x10])
+ ftpr_offset += me_start
+ print("Found FTPR header: FTPR partition spans from {:#x} to {:#x}"
+ .format(ftpr_offset, ftpr_offset + ftpr_lenght))
+ print("Removing extra partitions...")
+
+ fill_range(f, me_start + 0x30, ftpr_offset, b"\xff")
+ fill_range(f, ftpr_offset + ftpr_lenght, me_end, b"\xff")
+
+ print("Removing extra partition entries in FPT...")
+ f.seek(me_start + 0x30)
+ f.write(ftpr_header)
+ f.seek(me_start + 0x14)
+ f.write(pack("<I", 1))
+
+ print("Removing EFFS presence flag...")
+ f.seek(me_start + 0x24)
+ flags = unpack("<I", f.read(4))[0]
+ flags &= ~(0x00000001)
+ f.seek(me_start + 0x24)
+ f.write(pack("<I", flags))
+
+ f.seek(me_start, 0)
+ header = bytearray(f.read(0x30))
+ checksum = (0x100 - (sum(header) - header[0x1b]) & 0xff) & 0xff
+
+ print("Correcting checksum (0x{:02x})...".format(checksum))
+ # The checksum is just the two's complement of the sum of the first
+ # 0x30 bytes (except for 0x1b, the checksum itself). In other words,
+ # the sum of the first 0x30 bytes must be always 0x00.
+ f.seek(me_start + 0x1b)
+ f.write(pack("B", checksum))
+
+ f.seek(ftpr_offset)
+ if f.read(4) == b"$CPD":
+ me11 = True
+ num_entries = unpack("<I", f.read(4))[0]
+ f.seek(ftpr_offset + 0x10 + num_entries * 0x18 + 0x24)
+ else:
+ me11 = False
+ f.seek(ftpr_offset + 0x24)
+
+ version = unpack("<HHHH", f.read(0x08))
+ print("ME/TXE firmware version {}"
+ .format('.'.join(str(i) for i in version)))
+
+ if not me11:
+ print("Reading FTPR modules list...")
+ f.seek(ftpr_offset + 0x1c)
+ tag = f.read(4)
+
+ if tag == b"$MN2":
+ f.seek(ftpr_offset + 0x20)
+ num_modules = unpack("<I", f.read(4))[0]
+ f.seek(ftpr_offset + 0x290)
+ data = f.read(0x84)
+
+ module_header_size = 0
+ if data[0x0:0x4] == b"$MME":
+ if data[0x60:0x64] == b"$MME":
+ module_header_size = 0x60
+ elif data[0x80:0x84] == b"$MME":
+ module_header_size = 0x80
+
+ if module_header_size != 0:
+ f.seek(ftpr_offset + 0x290)
+ mod_headers = [f.read(module_header_size)
+ for i in range(0, num_modules)]
+
+ if all(mod_h.startswith(b"$MME") for mod_h in mod_headers):
+ remove_modules(f, mod_headers, ftpr_offset)
+ else:
+ print("Found less modules than expected in the FTPR "
+ "partition; skipping modules removal")
+ else:
+ print("Can't find the module header size; skipping "
+ "modules removal")
+ else:
+ print("Wrong FTPR partition tag ({}); skipping modules removal"
+ .format(tag))
+ else:
+ print("Modules removal in ME v11 or greater is not yet supported")
+
+ sys.stdout.write("Checking FTPR RSA signature... ")
+ if check_partition_signature(f, ftpr_offset):
+ print("VALID")
+ else:
+ print("INVALID!!")
+ sys.exit("The FTPR partition signature is not valid. Is the input "
+ "ME/TXE image valid?")
+
+ print("Done! Good luck!")