From mboxrd@z Thu Jan 1 00:00:00 1970 Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.remote-ip=100.103.45.18 ARC-Seal: i=1; d=subspace.kernel.org; s=arc20261015; a=rsa-sha256; cv=none; t=1792961139; b=WUmfKhKeLUqMi65sQaHm0AW5bmQtcFzPbKBi+W/795EIMMZbZT/iyvnPeikKAyLuHrCf Tah7slzDV35CxPRDd2r2N5wwzau1+VbmcwvLzRat9QSPyahDIY3rDzm5NZq0G+mhR5hSX yvVVEdsefwLN8TlDC19G1j+1f1r741Ub3I6ZUYanunyrKVLjp15Uae17Z+ZqbPza5wf/E bjpJ+79JjctAPNZpNDwcNJBy5PDGMk+FstZMIidYpra1apx+4eTe/JnE7ATpSpTykXmWA mXbCyEiJk4RHKEUSSVYDYA9qjjlq0yPaKekrTEfb9//gaFwoxWy8UOcRPlffN3NqSmw== ARC-Message-Signature: i=1; d=subspace.kernel.org; s=arc20261015; a=rsa-sha256; c=relaxed/relaxed; t=1792961139; h=Received:DKIM-Signature:From:To:Cc:Subject:Date:Message-ID:X-Mailer: MIME-Version:Content-Transfer-Encoding; bh=9ynOepXHL/8dv5MR32+GlROxxwXybPjSYd9AdFHCLkY=; b=zhjjT5QprdwioEl8nNyRZqqsY+/wX8tUSFte1sHQqcqD3kTljWhq/XZ4JpqJs/rht9v9 iotJUEM619ShUAPBl65NVTPoDbzHafXF3WJM/T7nbUcHBg5wUYW+7edX8jSiIc5pRr37m OlAj1OWQ0UuOIeR38uJju8Rdz9sOFb6M293wP/Bo8GDn4+6Esz6DTNZuv4aBw2FntVMLQ C+bW8QSGKDDE8UVU+fbEUyVLysasJHiKTDACjoYyzMPjyh2e4LMHe2v2yhWHWjsmObSVj a9ET6htB3sUVAnrD1EwmffyqO1mErkLgkkGRS6Gb4be+UQNLOPOi9j1Aew1pcbrhonA== ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dkim=pass header.d=kernel.org header.i=@kernel.org header.a=rsa-sha256 header.s=k20260515 header.b=S5mNVWqk; arc=none smtp.remote-ip=100.103.45.18 Received: from smtp.kernel.org (quasi.space.kernel.org [100.103.45.18]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature ECDSA (prime256v1)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 96A7E606A31A; Sun, 25 Oct 2026 20:45:38 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key, unprotected) header.d=kernel.org header.i=@kernel.org header.a=rsa-sha256 header.s=k20260515 header.b=S5mNVWqk Received: by smtp.kernel.org (Postfix) with ESMTPSA id E3E6D1F01825; Sun, 25 Oct 2026 20:45:37 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=kernel.org; s=k20260515; t=1792961138; bh=9ynOepXHL/8dv5MR32+GlROxxwXybPjSYd9AdFHCLkY=; h=From:To:Cc:Subject:Date; b=S5mNVWqkAcdLDRboNQIV8cdqH5qxiVE/j6YCovkSFgkjqLF7FG2qkoJo0nxcgL0Vf J90UaBaHEBx2j9V0ofiA1HiJhNs2xv+7fKxIt3xGEE3VoEstz7Oo+m8O0w+lpvwTUi UNRkvCZ5Gv2UYEplVGLtKokgokNyoMT7ROekLVIMp4DBBwf3c9Szq7AZE2aLI7P9TO mWDyStaaktQOalV9AcUyjNFpPM2qszfG1iH17zuhnaKlQgI/HW8qb9Z5WxDsqsqLzo SAtURp56TtPE8Uc4Ucnho8Mzns6/CaFYcySAUxiif7Prn+v7OnNyGYZFms6Xhbk01p S23BQCLg9zQQQ== From: Eric Biggers To: fsverity@lists.linux.dev Cc: linux-security-module@vger.kernel.org, Fan Wu , Paul Moore , James Morris , "Serge E . Hallyn" , Eric Biggers , stable@vger.kernel.org Subject: [PATCH] fsverity: Tell LSMs about signature only after verity is enabled Date: Sun, 25 Oct 2026 13:45:34 -0700 Message-ID: <20261025204534.122506-1-ebiggers@kernel.org> X-Mailer: git-send-email 2.56.0 X-Mailing-List: linux-security-module@vger.kernel.org List-Id: List-Post: List-Help: List-Subscribe: List-Unsubscribe: List-Owner: Precedence: list MIME-Version: 1.0 Content-Transfer-Encoding: 8bit FS_IOC_ENABLE_VERITY calls security_inode_setintegrity() before verity has finished being enabled. That's broken because enabling verity can still fail at that point, e.g. due to insufficient xattr space. Afterwards, arbitrary contents can be written to the file, which doesn't actually have verity enabled. The in-core inode stays the same, so the LSM considers it to have verity enabled with a valid signature. (Or rather, in the case of IPE, which is the only LSM that implements security_inode_setintegrity(), it also checks IS_VERITY(), which can be bypassed by enabling verity with no signature after the file writes.) Fix this by moving the call to security_inode_setintegrity() out of fsverity_verify_signature() and into a new function that is called from the appropriate places. Fixes: 7c373e4f1445 ("fsverity: expose verified fsverity built-in signatures to LSMs") Cc: stable@vger.kernel.org Signed-off-by: Eric Biggers --- fs/verity/enable.c | 8 ++++++ fs/verity/fsverity_private.h | 8 ++++++ fs/verity/open.c | 9 ++++++ fs/verity/signature.c | 55 ++++++++++++++++++++++++++---------- 4 files changed, 65 insertions(+), 15 deletions(-) diff --git a/fs/verity/enable.c b/fs/verity/enable.c index ad4ff71d7dd95..2949fb8afb2fa 100644 --- a/fs/verity/enable.c +++ b/fs/verity/enable.c @@ -302,6 +302,14 @@ static int enable_verity(struct file *filp, } else if (WARN_ON_ONCE(!IS_VERITY(inode))) { fsverity_remove_info(vi); err = -EINVAL; + } else { + /* + * Now that verity is enabled, report the signature (if any) to + * LSMs. We tolerate that a process racing with this ioctl can + * briefly see verity enabled with no signature reported yet. + */ + fsverity_report_valid_sig(inode, desc->signature, + le32_to_cpu(desc->sig_size)); } out: kfree(params.hashstate); diff --git a/fs/verity/fsverity_private.h b/fs/verity/fsverity_private.h index 6f85f63861aaa..2c41428d484b3 100644 --- a/fs/verity/fsverity_private.h +++ b/fs/verity/fsverity_private.h @@ -160,6 +160,8 @@ void __init fsverity_init_info_cache(void); extern int fsverity_require_signatures; int fsverity_verify_signature(const struct fsverity_info *vi, const u8 *signature, size_t sig_size); +void fsverity_report_valid_sig(const struct inode *inode, const u8 *signature, + size_t sig_size); void __init fsverity_init_signature(void); #else /* !CONFIG_FS_VERITY_BUILTIN_SIGNATURES */ @@ -170,6 +172,12 @@ fsverity_verify_signature(const struct fsverity_info *vi, return 0; } +static inline void fsverity_report_valid_sig(const struct inode *inode, + const u8 *signature, + size_t sig_size) +{ +} + static inline void fsverity_init_signature(void) { } diff --git a/fs/verity/open.c b/fs/verity/open.c index 6646290430380..13fa69d7e4be4 100644 --- a/fs/verity/open.c +++ b/fs/verity/open.c @@ -387,6 +387,15 @@ int fsverity_ensure_verity_info(struct inode *inode) goto out_free_desc; } + /* + * Now that the signature (if any) has been verified, report it to LSMs. + * Do this before inserting the fsverity_info into the hashtable, so + * that anyone who concurrently retrieves the fsverity_info for the same + * inode is guaranteed to see the LSM's signature status populated. + */ + fsverity_report_valid_sig(inode, desc->signature, + le32_to_cpu(desc->sig_size)); + /* * Multiple tasks may race to set the inode's verity info, in which case * we might find an existing fsverity_info in the hash table. diff --git a/fs/verity/signature.c b/fs/verity/signature.c index 0302a4e506ec4..212e873c84998 100644 --- a/fs/verity/signature.c +++ b/fs/verity/signature.c @@ -42,13 +42,16 @@ static struct key *fsverity_keyring; * @sig_size: size of signature in bytes, or 0 if no signature * * If the file includes a signature of its fs-verity file digest, verify it - * against the certificates in the fs-verity keyring. Note that signatures - * are verified regardless of the state of the 'fsverity_require_signatures' - * variable and the LSM subsystem relies on this behavior to help enforce - * file integrity policies. Please discuss changes with the LSM list - * (thank you!). + * against the certificates in the fs-verity keyring. * - * Return: 0 on success (signature valid or not required); -errno on failure + * Note that signatures are verified regardless of the state of the + * 'fsverity_require_signatures' variable, and the LSM subsystem relies on this + * behavior to help enforce file integrity policies. (See + * fsverity_report_valid_sig(), which assumes that the signature, if present, + * was verified here.) Please discuss changes with the LSM list (thank you!). + * + * Return: 0 if the signature is present and valid, or absent and not required; + * -errno on failure */ int fsverity_verify_signature(const struct fsverity_info *vi, const u8 *signature, size_t sig_size) @@ -111,18 +114,40 @@ int fsverity_verify_signature(const struct fsverity_info *vi, return err; } - err = security_inode_setintegrity(inode, - LSM_INT_FSVERITY_BUILTINSIG_VALID, - signature, - sig_size); + return 0; +} - if (err) { +/* + * If the signature exists, this tells the LSM that it was verified. + * + * fsverity_verify_signature() must have already been called with the same + * signature, *and* the inode must have verity enabled with the signature + * matching it. If called from FS_IOC_ENABLE_VERITY, the latter means that we + * must be past the point of no return, i.e. verity won't fail to be enabled at + * a later point. This is critical, since LSMs must be told about the signature + * only when the verity status is permanent in the in-core inode. + */ +void fsverity_report_valid_sig(const struct inode *inode, const u8 *signature, + size_t sig_size) +{ + int err; + + if (WARN_ON_ONCE(!IS_VERITY(inode))) + return; + + if (sig_size == 0) + return; + + err = security_inode_setintegrity( + inode, LSM_INT_FSVERITY_BUILTINSIG_VALID, signature, sig_size); + if (err) fsverity_err(inode, "Error %d exposing file signature to LSMs", err); - return err; - } - - return 0; + /* + * Besides the logging above, ignore errors here. The LSM hook just + * records the signature and doesn't perform access control. And in the + * case of FS_IOC_ENABLE_VERITY verity was already enabled. + */ } void __init fsverity_init_signature(void) base-commit: 21a7f44fd0765fcc5194130b28767786eba1d1e7 -- 2.56.0