linux-kernel - [PATCH 08/10] x86/pkeys: Simplify PKRU update in signal frame

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20250416021720.12305-9-chang.seok.bae@intel.com>
Date: Tue, 15 Apr 2025 19:16:58 -0700
From: "Chang S. Bae" <chang.seok.bae@...el.com>
To: mingo@...hat.com
Cc: linux-kernel@...r.kernel.org,
	x86@...nel.org,
	tglx@...utronix.de,
	bp@...en8.de,
	dave.hansen@...ux.intel.com,
	chang.seok.bae@...el.com,
	Aruna Ramakrishna <aruna.ramakrishna@...cle.com>,
	Tony W Wang-oc <TonyWWang-oc@...oxin.com>
Subject: [PATCH 08/10] x86/pkeys: Simplify PKRU update in signal frame

The signal delivery logic was modified to always set the PKRU bit in
xregs_state->header->xfeatures by this commit:

    ae6012d72fa6 ("x86/pkeys: Ensure updated PKRU value is XRSTOR'd")

However, the change derives the bitmask value using XGETBV(1), rather
than simply updating the buffer that already holds the value. Thus, this
approach induces an unnecessary dependency on XGETBV1 for PKRU handling.

Eliminate the dependency by using the established helper function.
Subsequently, remove the now-unused 'mask' argument.

Signed-off-by: Chang S. Bae <chang.seok.bae@...el.com>
Cc: Aruna Ramakrishna <aruna.ramakrishna@...cle.com>
Cc: Tony W Wang-oc <TonyWWang-oc@...oxin.com>
---
Changes from the last posting:
https://lore.kernel.org/lkml/20250214010607.7067-3-chang.seok.bae@intel.com/
* Massage the changelog

Additional Context:
  Previously, the concern was raised about environments where XGETBV1 is
  unavailable — such as in some virtual machines:

    https://lore.kernel.org/lkml/20250102075419.2559-1-TonyWWang-oc@zhaoxin.com

  That patch tried to sidestep the problem by skipping PKRU updates
  entirely when XGETBV1 is inaccessible. However, this assumed the
  dependency was necessary, which isn’t the case.
---
 arch/x86/kernel/fpu/xstate.h | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

diff --git a/arch/x86/kernel/fpu/xstate.h b/arch/x86/kernel/fpu/xstate.h
index aadf02aed071..a6d987c16293 100644
--- a/arch/x86/kernel/fpu/xstate.h
+++ b/arch/x86/kernel/fpu/xstate.h
@@ -84,18 +84,15 @@ static inline int set_xfeature_in_sigframe(struct xregs_state __user *xbuf, u64
 /*
  * Update the value of PKRU register that was already pushed onto the signal frame.
  */
-static inline int update_pkru_in_sigframe(struct xregs_state __user *buf, u64 mask, u32 pkru)
+static inline int update_pkru_in_sigframe(struct xregs_state __user *buf, u32 pkru)
 {
-	u64 xstate_bv;
 	int err;
 
 	if (unlikely(!cpu_feature_enabled(X86_FEATURE_OSPKE)))
 		return 0;
 
 	/* Mark PKRU as in-use so that it is restored correctly. */
-	xstate_bv = (mask & xfeatures_in_use()) | XFEATURE_MASK_PKRU;
-
-	err =  __put_user(xstate_bv, &buf->header.xfeatures);
+	err = set_xfeature_in_sigframe(buf, XFEATURE_MASK_PKRU);
 	if (err)
 		return err;
 
@@ -319,7 +316,7 @@ static inline int xsave_to_user_sigframe(struct xregs_state __user *buf, u32 pkr
 	clac();
 
 	if (!err)
-		err = update_pkru_in_sigframe(buf, mask, pkru);
+		err = update_pkru_in_sigframe(buf, pkru);
 
 	return err;
 }
-- 
2.45.2