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: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20230818011256.211078-5-peter.hilber@opensynergy.com>
Date: Fri, 18 Aug 2023 03:12:55 +0200
From: Peter Hilber <peter.hilber@...nsynergy.com>
To: linux-kernel@...r.kernel.org,
	kvm@...r.kernel.org,
	kvmarm@...ts.linux.dev
Cc: Peter Hilber <peter.hilber@...nsynergy.com>,
	Thomas Gleixner <tglx@...utronix.de>,
	Ingo Molnar <mingo@...hat.com>,
	Borislav Petkov <bp@...en8.de>,
	Dave Hansen <dave.hansen@...ux.intel.com>,
	x86@...nel.org,
	"H. Peter Anvin" <hpa@...or.com>,
	Richard Cochran <richardcochran@...il.com>,
	John Stultz <jstultz@...gle.com>,
	Stephen Boyd <sboyd@...nel.org>,
	netdev@...r.kernel.org,
	Marc Zyngier <maz@...nel.org>,
	Paolo Bonzini <pbonzini@...hat.com>,
	Oliver Upton <oliver.upton@...ux.dev>,
	James Morse <james.morse@....com>,
	Suzuki K Poulose <suzuki.poulose@....com>,
	Zenghui Yu <yuzenghui@...wei.com>,
	Sean Christopherson <seanjc@...gle.com>
Subject: [RFC PATCH 4/4] treewide: Use clocksource id for struct system_counterval_t

Use enum clocksource_ids, rather than struct clocksource *, in struct
system_counterval_t. This allows drivers to use
get_device_system_crosststamp() without referring to clocksource structs,
which are clocksource driver implementation details.

As a "treewide" change, this modifies code in timekeeping, ptp/kvm, and
x86/tsc.

So far, get_device_system_crosststamp() requires the caller to provide a
clocksource * as part of the cross-timestamp (through the
system_counterval_t * parameter in the get_time_fn() callback). But
clocksource * are not easily accessible to most code. As a workaround,
custom functions returning the clocksource * may be put into the
clocksource driver code (cf. kvm_arch_ptp_get_crosststamp() in
arm_arch_timer.c).

To avoid more interference with clocksource driver implementation details
from future code, change system_counterval_t to use enum clocksource_ids,
which serves as a unique identifier for relevant clocksources.

This change will allow the virtio_rtc driver, available as an RFC patch
series, to work without modifying the arm_arch_timer implementation.

This change should not alter any behavior. For the x86 ART related code,
.cs == NULL corresponds to system_counterval_t.cs_id == CSID_GENERIC (0).

The one drawback of this change is that clocksources now need a custom id
to work with get_device_system_crosststamp(), since they cannot be
distinguished otherwise.

Signed-off-by: Peter Hilber <peter.hilber@...nsynergy.com>
---
 arch/x86/kernel/tsc.c        | 6 +++---
 drivers/ptp/ptp_kvm_common.c | 3 ++-
 include/linux/timekeeping.h  | 4 ++--
 kernel/time/timekeeping.c    | 3 ++-
 4 files changed, 9 insertions(+), 7 deletions(-)

diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
index e56cc4e97d0d..eadfb819a0b5 100644
--- a/arch/x86/kernel/tsc.c
+++ b/arch/x86/kernel/tsc.c
@@ -1313,7 +1313,7 @@ struct system_counterval_t convert_art_to_tsc(u64 art)
 	res += tmp + art_to_tsc_offset;
 
 	return (struct system_counterval_t) {
-		.cs = have_art ? &clocksource_tsc : NULL,
+		.cs_id = have_art ? CSID_TSC : CSID_GENERIC,
 		.cycles = res
 	};
 }
@@ -1335,7 +1335,7 @@ EXPORT_SYMBOL(convert_art_to_tsc);
  * struct system_counterval_t - system counter value with the pointer to the
  *	corresponding clocksource
  *	@cycles:	System counter value
- *	@cs:		Clocksource corresponding to system counter value. Used
+ *	@cs_id:		Clocksource corresponding to system counter value. Used
  *			by timekeeping code to verify comparability of two cycle
  *			values.
  */
@@ -1353,7 +1353,7 @@ struct system_counterval_t convert_art_ns_to_tsc(u64 art_ns)
 	res += tmp;
 
 	return (struct system_counterval_t) {
-		.cs = have_art ? &clocksource_tsc : NULL,
+		.cs_id = have_art ? CSID_TSC : CSID_GENERIC,
 		.cycles = res
 	};
 }
diff --git a/drivers/ptp/ptp_kvm_common.c b/drivers/ptp/ptp_kvm_common.c
index 2418977989be..d72e6ed71b7a 100644
--- a/drivers/ptp/ptp_kvm_common.c
+++ b/drivers/ptp/ptp_kvm_common.c
@@ -4,6 +4,7 @@
  *
  * Copyright (C) 2017 Red Hat Inc.
  */
+#include <linux/clocksource.h>
 #include <linux/device.h>
 #include <linux/err.h>
 #include <linux/init.h>
@@ -46,7 +47,7 @@ static int ptp_kvm_get_time_fn(ktime_t *device_time,
 	preempt_enable_notrace();
 
 	system_counter->cycles = cycle;
-	system_counter->cs = cs;
+	system_counter->cs_id = cs->id;
 
 	*device_time = timespec64_to_ktime(tspec);
 
diff --git a/include/linux/timekeeping.h b/include/linux/timekeeping.h
index fe1e467ba046..2e7ae0f7f19e 100644
--- a/include/linux/timekeeping.h
+++ b/include/linux/timekeeping.h
@@ -270,12 +270,12 @@ struct system_device_crosststamp {
  * struct system_counterval_t - system counter value with the pointer to the
  *				corresponding clocksource
  * @cycles:	System counter value
- * @cs:		Clocksource corresponding to system counter value. Used by
+ * @cs_id:	Clocksource corresponding to system counter value. Used by
  *		timekeeping code to verify comparibility of two cycle values
  */
 struct system_counterval_t {
 	u64			cycles;
-	struct clocksource	*cs;
+	enum clocksource_ids	cs_id;
 };
 
 /*
diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 266d02809dbb..56428eadf4c1 100644
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -1236,7 +1236,8 @@ int get_device_system_crosststamp(int (*get_time_fn)
 		 * system counter value is the same as the currently installed
 		 * timekeeper clocksource
 		 */
-		if (tk->tkr_mono.clock != system_counterval.cs)
+		if (system_counterval.cs_id == CSID_GENERIC ||
+		    tk->tkr_mono.clock->id != system_counterval.cs_id)
 			return -ENODEV;
 		cycles = system_counterval.cycles;
 
-- 
2.39.2


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ