ext: Sync `ext-libfdt` with dtc v1.7.2 by BobbyRBruce · Pull Request #3037 · gem5/gem5

BobbyRBruce · 2026-03-26T17:47:21Z

Clang 21 introduces -Wtautological-compare, which causes our 2013-era libfdt snapshot to fail because it tries to spot overflow with if (p + len < p) pointer arithmetic (see here:

gem5/ext/libfdt/fdt.c

Line 67 in 7a2b0e4

if (p + len < p)

). The warning/error returned is:

 [ SHCC] ext/libfdt/fdt_ro.c -> ALL/ext/libfdt/fdt_ro.os
ext/libfdt/fdt.c:67:21: error: pointer comparison always evaluates to false
 [-Werror,-Wtautological-compare]

This change refreshes ext/libfdt to the version shipped with the lastest releast of dtc - v1.7.2. The newer code performs its bounds checking in integer space, so the new clang warning goes away while keeping the same behaviour.

Copilot

Pull request overview

This PR refreshes the vendored ext/libfdt snapshot to match dtc v1.7.2, primarily to resolve Clang 21’s -Wtautological-compare failures by moving bounds checks into integer space and updating the library’s internal probing/checking logic.

Changes:

Update libfdt headers and core implementation to the dtc v1.7.2 API/behavior (new probing helpers, new error codes, updated helpers/macros).
Add new dtc v1.7.2 libfdt modules (fdt_overlay.c, fdt_check.c, fdt_addresses.c) and update documentation to point at the upstream source/release.
Adjust SCons build metadata/comments for the ext library.

Reviewed changes

Copilot reviewed 16 out of 16 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
ext/libfdt/libfdt_internal.h	Updates internal helper API, adds assumption-masking helpers used by the refreshed libfdt.
ext/libfdt/libfdt_env.h	Updates environment/typing helpers and adds portability glue (e.g., Apple `strnlen` fallback).
ext/libfdt/libfdt.h	Large API refresh: new helpers (unaligned loads/stores), new traversal helpers, new overlay/address APIs, new error codes.
ext/libfdt/fdt_wip.c	Updates write-in-place functions; adds `fdt_setprop_inplace_namelen_partial()`.
ext/libfdt/fdt_sw.c	Refreshes sequential-write implementation with probe/state handling and new creation flags/APIs.
ext/libfdt/fdt_strerror.c	Expands strerror table for new error codes and updates formatting.
ext/libfdt/fdt_rw.c	Refreshes RW editing logic and introduces new splice/probe helpers and placeholder APIs.
ext/libfdt/fdt_ro.c	Refreshes RO helpers; adds string/phandle utilities and strengthens validation paths.
ext/libfdt/fdt_overlay.c	New overlay application support from dtc v1.7.2.
ext/libfdt/fdt_empty_tree.c	Minor updates to match refreshed API/headers.
ext/libfdt/fdt_check.c	New “full DTB validation” helper (`fdt_check_full()`).
ext/libfdt/fdt_addresses.c	New address/size-cells helpers and `fdt_appendprop_addrrange()`.
ext/libfdt/fdt.h	Refreshes core structs/macros (flexible array members, updated headers/guards).
ext/libfdt/fdt.c	Implements new probe/header-size logic and replaces pointer-overflow checks with integer-space checks.
ext/libfdt/SConscript	Updates metadata and (currently) leaves new modules commented out from the build list.
ext/libfdt/README	Updates provenance text to reference upstream dtc repo and v1.7.2 release tag.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Clang 21 introduces `-Wtautological-compare`, which causes our 2013-era `libfdt` snapshot to fail because it tries to spot overflow with `if (p + len < p)` pointer arithmetic (see here: https://github.com/gem5/gem5/blob/7a2b0e413d06c5ce7097104abef3b1d9eaabca91/ext/libfdt/fdt.c#L67). The warning/error returned is: ``` [ SHCC] ext/libfdt/fdt_ro.c -> ALL/ext/libfdt/fdt_ro.os ext/libfdt/fdt.c:67:21: error: pointer comparison always evaluates to false [-Werror,-Wtautological-compare] ``` This change refreshes `ext/libfdt` to the version shipped with the lastest releast of dtc - v1.7.2. The newer code performs its bounds checking in integer space, so the new clang warning goes away while keeping the same behaviour.

Copilot

Pull request overview

Copilot reviewed 17 out of 17 changed files in this pull request and generated 7 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-26T23:00:28Z

+		for (i = 0; i < fixup_len; i++) {
+			fdt32_t *refp;
+
+			refp = (fdt32_t *)(tree_val + fdt32_ld_(fixup_val + i));
+
+			/*
+			 * phandles to fixup can be unaligned, so use
+			 * fdt32_{ld,st}() to read/write them.
+			 */
+			fdt32_st(refp, fdt32_ld(refp) + delta);


The fixup offset read from __local_fixups__ (fdt32_ld_(fixup_val + i)) is used to form refp without validating it against tree_len. A malformed overlay can make this read/write outside the property buffer. Validate that each offset is within [0, tree_len - sizeof(fdt32_t)] (and consider overflow/unaligned cases) before calling fdt32_ld/st.

Copilot · 2026-03-26T23:00:28Z

+		for (i = 0; i < fixup_len; i++) {
+			fdt32_t *refp;
+			uint32_t valp;
+
+			refp = (fdt32_t *)(tree_val + fdt32_ld_(fixup_val + i));
+			valp = fdt32_ld(refp);
+
+			if (valp == fdto_phandle)
+				fdt32_st(refp, fdt_phandle);


Same as above: the fixup-provided offset is used to compute refp without checking it is within the bounds of tree_val/tree_len, which can lead to out-of-bounds reads/writes when resolving conflicting references. Add bounds/overflow checks for each fixup offset before dereferencing.

Copilot · 2026-03-26T23:00:29Z

+		poffset = strtoul(sep + 1, &endptr, 10);
+		if ((*endptr != '\0') || (endptr <= (sep + 1)))
+			return -FDT_ERR_BADOVERLAY;


poffset is parsed with strtoul() but stored in an int and then passed on as an offset into a property value. This can overflow/truncate on large inputs and, combined with the lack of robust overflow checking in the inplace setter, can lead to incorrect bounds checks. Parse into an unsigned type and validate the range (e.g., <= UINT32_MAX and within the target property's length).

Copilot · 2026-03-26T23:00:29Z

+ * Internal helpers to access tructural elements of the device tree
+ * blob (rather than for exaple reading integers from within property


Typo in comment: "tructural" → "structural".

Suggested change

* Internal helpers to access tructural elements of the device tree

* blob (rather than for exaple reading integers from within property

* Internal helpers to access structural elements of the device tree

* blob (rather than for example reading integers from within property

Copilot · 2026-03-26T23:00:29Z

-        return rsv_table + n;
+/*
+ * Internal helpers to access tructural elements of the device tree
+ * blob (rather than for exaple reading integers from within property


Typo in comment: "exaple" → "example".

Suggested change

* blob (rather than for exaple reading integers from within property

* blob (rather than for example reading integers from within property

Copilot · 2026-03-26T23:00:30Z

+ * signature or hash check before using libfdt.
+ *
+ * For situations where security is not a concern it may be safe to enable
+ * ASSUME_SANE.


The comment suggests enabling ASSUME_SANE, but there is no ASSUME_SANE value in the ASSUME_... enum below. This is confusing for anyone configuring FDT_ASSUME_MASK; please update the comment to reference the actual assumption(s) intended (e.g., ASSUME_VALID_DTB / ASSUME_VALID_INPUT) or remove the mention.

Suggested change

* ASSUME_SANE.

* less-strict assumptions.

Copilot · 2026-03-26T23:00:30Z

+#include "libfdt_env.h"
+
+#include <fdt.h>
+#include <libfdt.h>
+
+#include "libfdt_internal.h"


This source file defines public APIs declared in libfdt.h (e.g., fdt_overlay_apply() / fdt_overlay_target_offset()), but ext/libfdt/SConscript currently only builds a fixed list of .c files and does not include fdt_overlay.c (nor the other newly added sources like fdt_check.c / fdt_addresses.c). Unless the build script is updated, these symbols won't be part of libfdt and may cause link errors when used.

Clang 21 introduces `-Wtautological-compare`, which causes our 2013-era `libfdt` snapshot to fail because it tries to spot overflow with `if (p + len < p)` pointer arithmetic (see here: https://github.com/gem5/gem5/blob/7a2b0e413d06c5ce7097104abef3b1d9eaabca91/ext/libfdt/fdt.c#L67). The warning/error returned is: ``` [ SHCC] ext/libfdt/fdt_ro.c -> ALL/ext/libfdt/fdt_ro.os ext/libfdt/fdt.c:67:21: error: pointer comparison always evaluates to false [-Werror,-Wtautological-compare] ``` This change refreshes `ext/libfdt` to the version shipped with the lastest releast of dtc - v1.7.2. The newer code performs its bounds checking in integer space, so the new clang warning goes away while keeping the same behaviour.

Copilot AI review requested due to automatic review settings March 26, 2026 17:47

BobbyRBruce added the ext Components external to gem5 such as third-party code. Typically found in "ext" label Mar 26, 2026

github-project-automation bot added this to Tester Roadmap for gem5 project management Mar 26, 2026

github-project-automation bot moved this to Ready in Tester Roadmap for gem5 project management Mar 26, 2026

Copilot started reviewing on behalf of BobbyRBruce March 26, 2026 17:47 View session

Copilot AI reviewed Mar 26, 2026

View reviewed changes

Comment thread ext/libfdt/fdt_check.c

Comment thread ext/libfdt/SConscript Outdated

Comment thread ext/libfdt/fdt_wip.c

Comment thread ext/libfdt/fdt.c

powerjg requested changes Mar 26, 2026

View reviewed changes

Comment thread ext/libfdt/fdt.c

github-project-automation bot moved this from Ready to Waiting on developer in Tester Roadmap for gem5 project management Mar 26, 2026

BobbyRBruce force-pushed the update-libfdt-to-1.7.2 branch from 018cf02 to 53584ab Compare March 26, 2026 22:51

Copilot AI review requested due to automatic review settings March 26, 2026 22:54

BobbyRBruce force-pushed the update-libfdt-to-1.7.2 branch from 53584ab to 91fa71a Compare March 26, 2026 22:54

powerjg approved these changes Mar 26, 2026

View reviewed changes

Copilot started reviewing on behalf of BobbyRBruce March 26, 2026 22:55 View session

github-project-automation bot moved this from Waiting on developer to Is approved in Tester Roadmap for gem5 project management Mar 26, 2026

Copilot AI reviewed Mar 26, 2026

View reviewed changes

github-actions bot mentioned this pull request Mar 28, 2026

[repo-status] Weekly Status Report – Week of March 28, 2026 erin-le/gem5#19

Closed

13 tasks

erin-le merged commit bdc8c2b into gem5:develop Mar 31, 2026
42 of 43 checks passed

github-project-automation bot moved this from Is approved to Done in Tester Roadmap for gem5 project management Mar 31, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ext: Sync `ext-libfdt` with dtc v1.7.2#3037

ext: Sync `ext-libfdt` with dtc v1.7.2#3037
erin-le merged 1 commit intogem5:developfrom
BobbyRBruce:update-libfdt-to-1.7.2

BobbyRBruce commented Mar 26, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 26, 2026

Uh oh!

Copilot AI Mar 26, 2026

Uh oh!

Copilot AI Mar 26, 2026

Uh oh!

Copilot AI Mar 26, 2026

Uh oh!

Copilot AI Mar 26, 2026

Uh oh!

Copilot AI Mar 26, 2026

Uh oh!

Copilot AI Mar 26, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

		* Internal helpers to access tructural elements of the device tree
		* blob (rather than for exaple reading integers from within property

	* blob (rather than for exaple reading integers from within property
	* blob (rather than for example reading integers from within property

Conversation

BobbyRBruce commented Mar 26, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Mar 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 26, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 26, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants