Files
reverse-engineering-anything/docs/guides/native/kernel-build-config.md
T
f21fda0fc0 feat(elf): report a Linux kernel image's version banner and embedded build config (#9573)
* feat(elf): report a Linux kernel image's version banner and embedded build config

inspect_kernel_build_config and inspect-kernel-build-config read one
local vmlinux, raw arm64 Image, gzip bzImage or zImage, or configs.ko
and report its Linux version banner and the IKCONFIG .config embedded
between IKCFG_ST and IKCFG_ED, without executing it. A config is present
only when a gzip member ends exactly at IKCFG_ED, passes its CRC-32 and
ISIZE checks and parses as .config text; framing that fails is reported
malformed, never absent. bzImage gzip payloads are inflated within a
budget and other codecs are named as unsupported, using codec detection
now shared with the kernel module parser. The config search, inflate
sizes and option counts are bounded, scans honor cancellation, and
target-derived text is marked untrusted.

* fix(elf): reject a gzip member whose header checksum fails

When FHCRC is set, compare the low 16 bits of the header CRC-32 before
inflating, so a damaged header no longer yields a present config or an
inflated bzImage payload.

* fix(elf): detect the bzImage codec inside the declared payload

Codec detection now reads the same payload_offset/payload_length slice that
is inflated, so bytes past a short payload cannot name its codec.

* fix(elf): report a framed config that does not decode as malformed

A damaged or cut-off deflate stream between IKCFG_ST and IKCFG_ED was
skipped and could end as an absent config with complete coverage. It is
now malformed with config_gzip_invalid, and a later valid config still wins.

* fix(elf): scan config lines in place and accept long string values

Config text is read one line at a time instead of split into an array, and
a quoted value is validated before it is decoded once. The 4096-character
string cap is removed: Kconfig has no such limit and max_config_bytes
already bounds every line, so a long CONFIG_EXTRA_FIRMWARE list is no
longer reported as malformed.

* fix(elf): let a live cancellation reach the kernel config parser

The parser ran synchronously, so an abort raised during parsing could not
change the signal before it returned. Inflation now runs on the zlib thread
pool and each scan yields to the event loop once per MiB, so the existing
cancellation checks observe a real abort.

* refactor(elf): carry the refused kernel config budget as a typed fact

Budget failures are now KernelBuildConfigBudgetFailure with a knob field,
and the provider builds its recovery advice from that field instead of
searching the diagnostic message for an option name.

* docs(elf): move the kernel build config guide under docs/guides/native

The documentation guide assigns analyst workflows to docs/guides/<domain>/.
Move the guide there, place its navigation and directory entries in the
Native binaries group, and update links from the docs index, CLI guide,
spec and authored skill. The public /kernel-build-config route is unchanged.

* fix(elf): yield while checking one long config line

A config made of one long quoted value never reached the per-line yield, so
the control-byte check, value validation and SHA-256 ran in one synchronous
block. Each now works one MiB window per event-loop turn, and the line
patterns match only a bounded name prefix instead of the whole value.

* fix(elf): return paged config entries that do not pin the decoded text

Entry names and values were V8 slices of the whole decoded config, so a
one-entry page kept the full text alive. Entries in the returned page are
now standalone copies, so retained memory follows the page size.

* fix(elf): yield while kernel marker searches find nothing

IKCFG_ST, IKCFG_ED and banner searches used one Buffer.indexOf over the
rest of the image, so a large image with no marker never yielded. They now
search overlapping one MiB windows and yield after each window without a
match.

* fix(elf): preserve long kernel config names

* fix(elf): frame kernel configs by gzip member boundaries

---------

Co-authored-by: Matt Van Horn <455140+mvanhorn@users.noreply.github.com>
Co-authored-by: morluto <76467478+morluto@users.noreply.github.com>
2026-11-09 21:41:56 +08:00

8.3 KiB

Linux kernel build config

inspect_kernel_build_config / inspect-kernel-build-config reads the Linux version banner and the embedded .config from one selected local kernel image without executing or booting it. Accepted inputs are a vmlinux ELF, a raw arm64 Image, an x86 bzImage or zImage whose payload is gzip, and a configs.ko module. It does not require a Linux host, extract-ikconfig, or /proc/config.gz.

rea inspect-kernel-build-config /absolute/path/to/vmlinux --json
rea inspect-kernel-build-config /absolute/path/to/bzImage --name CONFIG_SECURITY_SELINUX --json

The CLI resolves relative paths against the caller's working directory. MCP requires an absolute filesystem path and preserves that exact selection:

{
  "name": "inspect_kernel_build_config",
  "arguments": { "path": "/artifacts/vmlinux", "name": "CONFIG_MODULE_SIG" }
}

Observations

The inline Evidence result contains the artifact path, SHA-256, and byte length. format is elf when the file starts with the ELF magic and file otherwise.

banner is the first printable string that starts with Linux version followed by a version token beginning with a digit and ends at a NUL or newline within 512 bytes. version is that token. builder is the first parenthesized group (user@host). compiler is the second group up to the last ) #, so nested parentheses in compiler strings are kept. The banner is absent when no such string was found in the searched bytes and unknown when the bytes were not searched.

config is present only when all of the following hold:

  • IKCFG_ST is immediately followed by a gzip member (1f 8b 08).
  • The member inflates within max_config_bytes and its CRC-32 and ISIZE trailer match.
  • The member trailer ends exactly at the first IKCFG_ED after the gzip header. Inflation reads only the bytes up to that marker.
  • Every non-blank line of the inflated text is a # comment, # CONFIG_X is not set, or CONFIG_X=value, and the text is valid UTF-8 with no control bytes other than tab, CR, and LF.

An IKCFG_ST string that is not followed by gzip magic, or whose complete member is not followed by IKCFG_ED, is skipped, so an ordinary string is never reported as a config. When IKCFG_ST, gzip magic, and a later IKCFG_ED frame a member that does not decode, a deflate stream cut off by that marker, a member that fails its CRC-32 or ISIZE check, or text that fails the grammar, the config is malformed with coverage incomplete, and marker, gzip, and end_marker locate the blob. A malformed blob is never reported as an absent config. A valid config later in the searched bytes still wins. marker, gzip, and end_marker give the three ranges. sha256 hashes the inflated text. option_count counts every CONFIG_* line. matched_count counts the options that match the optional exact name filter, and entries is the offset/limit page of those matches in file order.

Each entry has a name, a kind, and a value:

kind Source line value
y, m CONFIG_X=y, CONFIG_X=m y or m
number decimal or 0x hex the text as written
string CONFIG_X="..." decoded text (\", \\)
not_set # CONFIG_X is not set null

Banner text and configuration values are target-derived and untrusted.

bzImage payloads

A file with HdrS at offset 0x202 is an x86 setup header. From boot protocol 2.08, payload_offset and payload_length locate the compressed kernel after the (setup_sects + 1) * 512 setup code. payload reports the container, codec, file range, and inflated length. A gzip payload is inflated within max_payload_bytes and the inflated bytes are searched. Locations in those bytes have space: "inflated_payload" and are offsets in the inflated payload, not file offsets. Only the leading gzip member is inflated, and its own ISIZE trailer gives the inflated length. Kbuild appends a separate 4-byte size only for the non-gzip codecs. Any bytes after the gzip member inside the payload are not read.

xz, lz4, lzma, zstd, bzip2, and lzo payloads are unsupported. The result names the codec and external_decompression as the next probe. This is an external step, not a REA operation: decompress the payload with the matching utility into a separate file, then call inspect_kernel_build_config on that file. A file that is itself a compressed stream, including a whole-file gzip such as Image.gz, is reported as unsupported with its codec and is not inflated. Only a gzip payload inside a bzImage is inflated.

Limits and failures

The selected file is limited to 64 MiB by default (512 MiB maximum). An IKCFG member may inflate to 4 MiB by default (32 MiB maximum) and a bzImage payload to 128 MiB by default (512 MiB maximum). At most 16384 CONFIG_* options are examined by default (65536 maximum). A page holds 256 entries by default (4096 maximum). Exceeding any budget fails with no partial success and names the option to raise. An IKCFG member past max_config_bytes fails that way only when no later candidate is accepted, so an oversized decoy cannot hide a valid config.

One search inflates at most 16 IKCFG_ST gzip candidates and at most four times max_config_bytes of output across them. A candidate whose stream fails partway through is charged its full per-candidate allowance, since its partial output is not measured. When that search budget runs out before a config is accepted, the config is incomplete with reason config_search_exhausted instead of failing or reporting an absent config. Inflation runs off the event loop. Every scan, including marker searches that find nothing, config text checks, validation of one long value, and hashing, yields to it at least once per MiB, so a cancellation that arrives while the image is being parsed is honored at the next check. One inflate call runs to its output bound once started.

Results that could not search or read the kernel bytes stay incomplete, malformed, or unsupported and are never reported as an absent config:

config.reason Meaning
config_gzip_invalid a framed IKCFG_ST gzip member does not decode or fails its CRC-32 or ISIZE
config_text_invalid a framed IKCFG_ST gzip member is not .config text
config_gzip_truncated an IKCFG_ST gzip member runs past the searched bytes
config_search_exhausted decoy candidates used up the config search budget
payload_truncated the declared bzImage payload or its gzip member ends early
payload_gzip_invalid the bzImage gzip payload fails to inflate or its trailer
payload_codec_unsupported a compressed payload or whole-file stream was not inflated
unrecognized_payload_codec the bzImage payload starts with no known codec magic
setup_header_payload_unavailable the setup header predates 2.08 or has no payload length

A kernel built without CONFIG_IKCONFIG carries no embedded config, so an absent config describes only the bytes that were searched. A kernel built with CONFIG_IKCONFIG=m keeps the config in configs.ko; inspect that module instead of the image. This workflow does not execute the image, compare the config with a source tree, or evaluate Kconfig dependencies.

Format references

The markers follow Linux v6.12's kernel/configs.c and scripts/extract-ikconfig. The setup header fields follow the x86 boot protocol. The banner format follows init/version-timestamp.c. These references establish the decoded layout, not runtime behavior.