diff mbox

xen/x86: Do not call xen_init_time_ops() until shared_info is initialized

Message ID 1493842851-10859-1-git-send-email-boris.ostrovsky@oracle.com (mailing list archive)
State New, archived
Headers show

Commit Message

Boris Ostrovsky May 3, 2017, 8:20 p.m. UTC
Routines that are set by xen_init_time_ops() use shared_info's
pvclock_vcpu_time_info area. This area is not properly available until
shared_info is mapped in xen_setup_shared_info().

This became especially problematic due to commit dd759d93f4dd ("x86/timers:
Add simple udelay calibration") where we end up reading tsc_to_system_mul
from xen_dummy_shared_info (i.e. getting zero value) and then trying
to divide by it in pvclock_tsc_khz().

Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
---
 arch/x86/xen/enlighten_pv.c | 9 +++++++--
 arch/x86/xen/time.c         | 2 +-
 2 files changed, 8 insertions(+), 3 deletions(-)

Comments

Jürgen Groß May 4, 2017, 6:37 a.m. UTC | #1
On 03/05/17 22:20, Boris Ostrovsky wrote:
> Routines that are set by xen_init_time_ops() use shared_info's
> pvclock_vcpu_time_info area. This area is not properly available until
> shared_info is mapped in xen_setup_shared_info().
> 
> This became especially problematic due to commit dd759d93f4dd ("x86/timers:
> Add simple udelay calibration") where we end up reading tsc_to_system_mul
> from xen_dummy_shared_info (i.e. getting zero value) and then trying
> to divide by it in pvclock_tsc_khz().
> 
> Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>

Reviewed-by: Juergen Gross <jgross@suse.com>


Thanks,

Juergen
Jürgen Groß May 5, 2017, 8:55 a.m. UTC | #2
On 03/05/17 22:20, Boris Ostrovsky wrote:
> Routines that are set by xen_init_time_ops() use shared_info's
> pvclock_vcpu_time_info area. This area is not properly available until
> shared_info is mapped in xen_setup_shared_info().
> 
> This became especially problematic due to commit dd759d93f4dd ("x86/timers:
> Add simple udelay calibration") where we end up reading tsc_to_system_mul
> from xen_dummy_shared_info (i.e. getting zero value) and then trying
> to divide by it in pvclock_tsc_khz().
> 
> Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>

Pushed to xen/tip for-linus-4.12b


Thanks,

Juergen
diff mbox

Patch

diff --git a/arch/x86/xen/enlighten_pv.c b/arch/x86/xen/enlighten_pv.c
index a732bc2..6f80a01 100644
--- a/arch/x86/xen/enlighten_pv.c
+++ b/arch/x86/xen/enlighten_pv.c
@@ -988,6 +988,13 @@  void xen_setup_shared_info(void)
 #endif
 
 	xen_setup_mfn_list_list();
+
+	/*
+	 * Now that shared info is set up we can start using routines that
+	 * point to pvclock area.
+	 */
+	if (system_state == SYSTEM_BOOTING)
+		xen_init_time_ops();
 }
 
 /* This is called once we have the cpu_possible_mask */
@@ -1286,8 +1293,6 @@  asmlinkage __visible void __init xen_start_kernel(void)
 	x86_init.oem.arch_setup = xen_arch_setup;
 	x86_init.oem.banner = xen_banner;
 
-	xen_init_time_ops();
-
 	/*
 	 * Set up some pagetable state before starting to set any ptes.
 	 */
diff --git a/arch/x86/xen/time.c b/arch/x86/xen/time.c
index 090c7eb..a1895a8 100644
--- a/arch/x86/xen/time.c
+++ b/arch/x86/xen/time.c
@@ -406,7 +406,7 @@  static void __init xen_time_init(void)
 		pvclock_gtod_register_notifier(&xen_pvclock_gtod_notifier);
 }
 
-void __init xen_init_time_ops(void)
+void __ref xen_init_time_ops(void)
 {
 	pv_time_ops = xen_time_ops;