All of lore.kernel.org
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Arnaldo Carvalho de Melo" <acme@kernel.org>
Cc: linux-perf-users@vger.kernel.org
Subject: Re: [PATCH v7 02/13] perf scripts: Add a coverage harness for perf
Date: Wed, 18 Nov 2026 22:08:34 +0000	[thread overview]
Message-ID: <sashiko-outbox-222085@kernel.org> (raw)
In-Reply-To: <20261118215716.114868-3-acme@kernel.org>

Thank you for your contribution! Sashiko AI review found 3 potential
issues to consider:
- [Medium] perf scripts: abspath() symlink resolution and non-root
  bypass in coverage.sh
- [Low] perf scripts: check_root_path() leaf sticky exemption and
  unchecked state read
- [Low] perf scripts: fix_ownership() chowns existing root BUILD_DIR
  without mindepth
--

commit f925befacbe87aac1cdf421770ce4909c03a1946
Author: Arnaldo Carvalho de Melo <acme@redhat.com>

perf scripts: Add a coverage harness for perf

This commit adds tools/perf/scripts/coverage.sh and corresponding make
targets to build perf with clang source-based coverage, run the perf test
suite, and report line and function coverage.

> diff --git a/tools/perf/scripts/coverage.sh b/tools/perf/scripts/coverage.sh
> new file mode 100755
> index 0000000000000..9c77644e77f91
> --- /dev/null
> +++ b/tools/perf/scripts/coverage.sh
> @@ -0,0 +1,1114 @@

[ ... ]

> +# The tests run from the kernel tree root, so keep these absolute.  The
> +# Makefile passes FULL_O resolved with readlink -f; resolve here too so an
> +# in-tree build directory compares equal to the physical PERF_DIR below.
> +abspath()
> +{
> +	realpath -m "$1"
> +}
> +BUILD_DIR=$(abspath "$BUILD_DIR")
> +
> +#
> +# Root executes the perf in the build directory and deletes and rewrites
> +# its cov/ directory, and that directory comes from a file the invoking
> +# user wrote or is the default under /tmp, so refuse one owned by somebody
> +# else or writable by another group or by others.  Every parent is checked
> +# too: a directory another user can write lets them swap an entry between
> +# this check and the run.  The trusted user is the one behind sudo, or the
> +# owner of the kernel tree when root runs the script directly, as in a
> +# container with the tree mounted from the host.
> +#
> +TRUST_UID=0
> +TRUST_GID=0
> +if [ "$ROOT" = 1 ]; then
> +	if [ -n "${SUDO_UID:-}" ]; then
> +		TRUST_UID=$SUDO_UID
> +		TRUST_GID=${SUDO_GID:-$SUDO_UID}
> +	else
> +		read -r TRUST_UID TRUST_GID < \
> +			<(stat -c '%u %g' "$PERF_DIR" 2>/dev/null) ||
> +			die "could not stat $PERF_DIR"
> +	fi
> +fi
> +
> +check_root_path()
> +{
> +	local path=$1 uid gid mode owners=root
> +	[ "$ROOT" = 1 ] || return 0
> +	[ "$TRUST_UID" = 0 ] || owners="root or uid $TRUST_UID"
> +	# abspath() resolves links such as /home -> /var/home on Silverblue
> +	# and CoreOS, so the walk below checks their targets.  A path not
> +	# resolved that way, like $BUILD_DIR/perf, must not be a link.
> +	[ -L "$path" ] && die "$path is a symbolic link, use its target"

[Severity: Medium]
Can another local user hijack the default /tmp/build/perf-coverage path
because abspath() resolves symlinks with realpath -m before
check_root_path() runs, and check_root_path() returns immediately when
ROOT=0?

During a root run, if another user creates /tmp/build and places a
symlink at /tmp/build/perf-coverage pointing to a root- or
developer-owned directory, realpath -m resolves the symlink to its
target before check_root_path() walks the path, bypassing validation of
the untrusted /tmp/build directory and redirecting rm -rf "$COV_DIR/raw"
or fix_ownership()'s chown -Rh.  During a non-root run (such as make
coverage-prep), check_root_path() is skipped and mkdir_checked() accepts
any pre-existing directory, so a pre-created mode 0777
/tmp/build/perf-coverage lets another user replace $BUILD_DIR/perf
before tests or a subsequent root run execute it.

> +	while :; do
> +		if [ -e "$path" ]; then
> +			if ! read -r uid gid mode < \
> +				<(stat -c '%u %g %a' "$path" 2>/dev/null); then
> +				die "could not stat $path"
> +			fi
> +			# A sticky directory owned by root, like /tmp, lets
> +			# anybody create entries but not replace one that
> +			# belongs to somebody else.  Anyone can set the sticky
> +			# bit on what they own, so nothing else is exempt.
> +			if ! { [ -d "$path" ] && [ "$uid" = 0 ] &&
> +			       [ $(( 8#$mode & 8#1000 )) != 0 ]; }; then
> +				[ "$uid" = 0 ] || [ "$uid" = "$TRUST_UID" ] ||
> +					die "$path is owned by uid $uid, not by $owners"
> +				# With user private groups, which most
> +				# distributions set up, umask 002 gives
> +				# 775 and only the developer's own group
> +				# may write here.
> +				[ $(( 8#$mode & 8#002 )) = 0 ] ||
> +					die "$path is writable by others"
> +				if [ $(( 8#$mode & 8#020 )) != 0 ] &&
> +				   [ "$gid" != "$TRUST_GID" ]; then
> +					die "$path is writable by group $gid"
> +				fi
> +			fi
> +		fi
> +		[ "$path" = / ] && return 0
> +		path=$(dirname "$path")
> +	done
> +}
> +
> +# The state file is read below, so check BUILD_DIR before opening it: a
> +# directory another user can write lets them replace state with a fifo
> +# and hang root.
> +check_root_path "$BUILD_DIR"
> +
> +if [ "$COV_DIR_SET" = 0 ] && [ -f "$BUILD_DIR/cov/state" ]; then
> +	COV_DIR=$(awk -F= '$1 == "COV_DIR" { print $2; exit }' \
> +		"$BUILD_DIR/cov/state")
> +fi

[Severity: Low]
Does check_root_path() exempt a sticky directory even when it is the
target directory $1 itself rather than an ancestor, and should
check_root_path() also validate "$BUILD_DIR/cov/state" before awk reads
it?

Because the sticky-bit exemption applies on the first iteration of the
loop, passing /tmp or /var/tmp as BUILD_DIR or COV_DIR passes
check_root_path() and lets root read, write, and chown entries directly
inside a world-writable sticky directory.  In addition, only $BUILD_DIR
is checked before reading $BUILD_DIR/cov/state, so an untrusted or
symlinked cov/state is opened before check_root_path "$COV_DIR" runs.

[ ... ]

> +	if [ -n "$BUILD_STAMP" ] && [ -f "$BUILD_STAMP" ]; then
> +		find "$BUILD_DIR" -user 0 -newer "$BUILD_STAMP" \
> +			-exec chown -h "$uid:$gid" {} + ||
> +			warn "could not give the files root built in $BUILD_DIR back to uid $uid"
> +		rm -f "$BUILD_STAMP" || true
> +	fi

[Severity: Low]
Without -mindepth 1, doesn't find "$BUILD_DIR" -user 0 -newer
"$BUILD_STAMP" also match and chown $BUILD_DIR itself when root builds
in a pre-existing root-owned directory?

Creating build outputs such as $BUILD_DIR/perf updates $BUILD_DIR's
mtime after $BUILD_STAMP is created, so find includes the top-level
directory and hands ownership of $BUILD_DIR to $uid:$gid despite
BUILD_CREATED=0.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261118215716.114868-1-acme@kernel.org?part=2

  reply	other threads:[~2026-11-18 22:08 UTC|newest]

Thread overview: 27+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-18 21:57 [PATCH v7 0/13] perf coverage: A harness for perf, and the fixes it found Arnaldo Carvalho de Melo
2026-11-18 21:57 ` [PATCH v7 01/13] perf build: Link the jvmti agent and the GTK UI with EXTRA_CFLAGS Arnaldo Carvalho de Melo
2026-11-18 22:01   ` sashiko-bot
2026-11-18 21:57 ` [PATCH v7 02/13] perf scripts: Add a coverage harness for perf Arnaldo Carvalho de Melo
2026-11-18 22:08   ` sashiko-bot [this message]
2026-11-18 21:57 ` [PATCH v7 03/13] perf annotate: Handle absolute addressing in data type profiling Arnaldo Carvalho de Melo
2026-11-18 22:14   ` sashiko-bot
2026-11-18 21:57 ` [PATCH v7 04/13] perf tests: Exercise DW_OP_addrx global variables Arnaldo Carvalho de Melo
2026-11-18 22:08   ` sashiko-bot
2026-11-18 21:57 ` [PATCH v7 05/13] perf tests: Exercise the flexible array member path Arnaldo Carvalho de Melo
2026-11-18 22:01   ` sashiko-bot
2026-11-18 21:57 ` [PATCH v7 06/13] perf annotate-data: Count flexible array accesses for the whole struct Arnaldo Carvalho de Melo
2026-11-18 22:04   ` sashiko-bot
2026-11-18 21:57 ` [PATCH v7 07/13] perf tests: Exercise data type resolution from a register Arnaldo Carvalho de Melo
2026-11-18 22:03   ` sashiko-bot
2026-11-18 21:57 ` [PATCH v7 08/13] perf annotate: Print the stats after the annotations Arnaldo Carvalho de Melo
2026-11-18 22:03   ` sashiko-bot
2026-11-18 21:57 ` [PATCH v7 09/13] perf tests: Silence the libbfd probe in build-test Arnaldo Carvalho de Melo
2026-11-18 22:00   ` sashiko-bot
2026-11-18 21:57 ` [PATCH v7 10/13] perf build: Record the features checked outside FEATURE_TESTS Arnaldo Carvalho de Melo
2026-11-18 22:04   ` sashiko-bot
2026-11-18 21:57 ` [PATCH v7 11/13] perf tests: Fix the Intel PT virtual LBR check Arnaldo Carvalho de Melo
2026-11-18 22:03   ` sashiko-bot
2026-11-18 21:57 ` [PATCH v7 12/13] perf test: Drop samples without an IP in the dlfilter test Arnaldo Carvalho de Melo
2026-11-18 22:03   ` sashiko-bot
2026-11-18 21:57 ` [PATCH v7 13/13] perf test: Skip the LBR inline unwinding check when unsupported Arnaldo Carvalho de Melo
2026-11-18 22:04   ` sashiko-bot

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=sashiko-outbox-222085@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=acme@kernel.org \
    --cc=linux-perf-users@vger.kernel.org \
    --cc=sashiko-reviews@lists.linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.