Skip to contentSkip to conversation

KVM: kvmclock: Fix vCPUs > 64 can't be online/hotpluged

PATCH Superseded Wanpeng Li 2 participants 5 messages Jan 18, 2021 – Jan 28, 2021

Subsystems (3)

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

Wanpeng Li

author
#0 lore raw

PATCH v2 1/1 KVM: kvmclock: Fix vCPUs > 64 can't be online/hotpluged

From: Wanpeng Li <[email protected]>

The per-cpu vsyscall pvclock data pointer assigns either an element of the static array hv_clock_boot (#vCPU <= 64) or dynamically allocated memory hvclock_mem (vCPU > 64), the dynamically memory will not be allocated if kvmclock vsyscall is disabled, this can result in cpu hotpluged fails in kvmclock_setup_percpu() which returns -ENOMEM. This patch fixes it by not assigning vsyscall pvclock data pointer if kvmclock vdso_clock_mode is not VDSO_CLOCKMODE_PVCLOCK.

Fixes 6a1cac56f4 ("x86/kvm: Use __bss_decrypted attribute in shared variables")
Cc Brijesh Singh <[email protected]>
Cc [email protected]#v4.19-rc5+

---
v1 -> v2:
* add code comments

 arch/x86/kernel/kvmclock.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
arch/x86/kernel/kvmclock.c +4 -2
index aa59374..01d4e55c 100644
@@ -294,9 +294,11 @@ static int kvmclock_setup_percpu(unsigned int cpu)
 	/*
 	 * The per cpu area setup replicates CPU0 data to all cpu
 	 * pointers. So carefully check. CPU0 has been set up in init
-	 * already.
+	 * already. Assign vsyscall pvclock data pointer iff kvmclock
+	 * vsyscall is enabled.
 	 */
-	if (!cpu || (p && p != per_cpu(hv_clock_per_cpu, 0)))
+	if (!cpu || (p && p != per_cpu(hv_clock_per_cpu, 0)) ||
+	    (kvm_clock.vdso_clock_mode != VDSO_CLOCKMODE_PVCLOCK))
 		return 0;
 
 	/* Use the static page for the first CPUs, allocate otherwise */
Signature · 1 line
2.7.4
author
#1 lore raw

ping,

Quoted 45 lines · Wanpeng Li

From: Wanpeng Li <[email protected]>

The per-cpu vsyscall pvclock data pointer assigns either an element of the static array hv_clock_boot (#vCPU <= 64) or dynamically allocated memory hvclock_mem (vCPU > 64), the dynamically memory will not be allocated if kvmclock vsyscall is disabled, this can result in cpu hotpluged fails in kvmclock_setup_percpu() which returns -ENOMEM. This patch fixes it by not assigning vsyscall pvclock data pointer if kvmclock vdso_clock_mode is not VDSO_CLOCKMODE_PVCLOCK.

Fixes 6a1cac56f4 ("x86/kvm: Use __bss_decrypted attribute in shared variables")
Cc Brijesh Singh <[email protected]>
Cc [email protected]#v4.19-rc5+

---
v1 -> v2:
* add code comments

 arch/x86/kernel/kvmclock.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
arch/x86/kernel/kvmclock.c +4 -3
index aa59374..01d4e55c 100644
@@ -294,9 +294,11 @@ static int kvmclock_setup_percpu(unsigned int cpu)
        /*
         * The per cpu area setup replicates CPU0 data to all cpu
         * pointers. So carefully check. CPU0 has been set up in init
-        * already.
+        * already. Assign vsyscall pvclock data pointer iff kvmclock
+        * vsyscall is enabled.
         */
-       if (!cpu || (p && p != per_cpu(hv_clock_per_cpu, 0)))
+       if (!cpu || (p && p != per_cpu(hv_clock_per_cpu, 0)) ||
+           (kvm_clock.vdso_clock_mode != VDSO_CLOCKMODE_PVCLOCK))
                return 0;
 
        /* Use the static page for the first CPUs, allocate otherwise */
--

2.7.4

maintainer
#2 lore raw
Quoted 12 lines · Wanpeng Li

ping,

Quoted 10 lines · Wanpeng Li

From: Wanpeng Li <[email protected]>

The per-cpu vsyscall pvclock data pointer assigns either an element of the static array hv_clock_boot (#vCPU <= 64) or dynamically allocated memory hvclock_mem (vCPU > 64), the dynamically memory will not be allocated if kvmclock vsyscall is disabled, this can result in cpu hotpluged fails in kvmclock_setup_percpu() which returns -ENOMEM. This patch fixes it by not assigning vsyscall pvclock data pointer if kvmclock vdso_clock_mode is not VDSO_CLOCKMODE_PVCLOCK.

I am sorry, I still cannot figure out this patch.

Is hotplug still broken if kvm vsyscall is enabled?

Paolo

Quoted 35 lines
Quoted 34 lines
Fixes 6a1cac56f4 ("x86/kvm: Use __bss_decrypted attribute in shared variables")
Cc Brijesh Singh <[email protected]>
Cc [email protected]#v4.19-rc5+

---
v1 -> v2:
* add code comments

  arch/x86/kernel/kvmclock.c | 6 ++++--
  1 file changed, 4 insertions(+), 2 deletions(-)
arch/x86/kernel/kvmclock.c +4 -3
index aa59374..01d4e55c 100644
@@ -294,9 +294,11 @@ static int kvmclock_setup_percpu(unsigned int cpu)
         /*
          * The per cpu area setup replicates CPU0 data to all cpu
          * pointers. So carefully check. CPU0 has been set up in init
-        * already.
+        * already. Assign vsyscall pvclock data pointer iff kvmclock
+        * vsyscall is enabled.
          */
-       if (!cpu || (p && p != per_cpu(hv_clock_per_cpu, 0)))
+       if (!cpu || (p && p != per_cpu(hv_clock_per_cpu, 0)) ||
+           (kvm_clock.vdso_clock_mode != VDSO_CLOCKMODE_PVCLOCK))
                 return 0;
 
         /* Use the static page for the first CPUs, allocate otherwise */
--

2.7.4

author
#3 lore raw
Quoted 15 lines · Paolo Bonzini
Quoted 12 lines · Wanpeng Li

ping,

Quoted 10 lines · Wanpeng Li

From: Wanpeng Li <[email protected]>

The per-cpu vsyscall pvclock data pointer assigns either an element of the static array hv_clock_boot (#vCPU <= 64) or dynamically allocated memory hvclock_mem (vCPU > 64), the dynamically memory will not be allocated if kvmclock vsyscall is disabled, this can result in cpu hotpluged fails in kvmclock_setup_percpu() which returns -ENOMEM. This patch fixes it by not assigning vsyscall pvclock data pointer if kvmclock vdso_clock_mode is not VDSO_CLOCKMODE_PVCLOCK.

Quoted 3 lines

I am sorry, I still cannot figure out this patch.

Is hotplug still broken if kvm vsyscall is enabled?

Just when kvm vsyscall is disabled. :)

# lscpu
Architecture:           x86_64
CPU op-mode(s):    32-bit, 64-bit
Byte Order:             Little Endian
CPU(s):                   88
On-line CPU(s) list:   0-63
Off-line CPU(s) list:  64-87
# cat /proc/cmdline
BOOT_IMAGE=/vmlinuz-5.10.0-rc3-tlinux2-0050+ root=/dev/mapper/cl-root
ro rd.lvm.lv=cl/root rhgb quiet console=ttyS0 LANG=en_US
.UTF-8 no-kvmclock-vsyscall
# echo 1 > /sys/devices/system/cpu/cpu76/online
-bash: echo: write error: Cannot allocate memory

Wanpeng

author
#4 lore raw
Quoted 31 lines · Wanpeng Li
Quoted 15 lines · Paolo Bonzini
Quoted 12 lines · Wanpeng Li

ping,

Quoted 10 lines · Wanpeng Li

From: Wanpeng Li <[email protected]>

The per-cpu vsyscall pvclock data pointer assigns either an element of the static array hv_clock_boot (#vCPU <= 64) or dynamically allocated memory hvclock_mem (vCPU > 64), the dynamically memory will not be allocated if kvmclock vsyscall is disabled, this can result in cpu hotpluged fails in kvmclock_setup_percpu() which returns -ENOMEM. This patch fixes it by not assigning vsyscall pvclock data pointer if kvmclock vdso_clock_mode is not VDSO_CLOCKMODE_PVCLOCK.

Quoted 3 lines

I am sorry, I still cannot figure out this patch.

Is hotplug still broken if kvm vsyscall is enabled?

Just when kvm vsyscall is disabled. :)

# lscpu
Architecture:           x86_64
CPU op-mode(s):    32-bit, 64-bit
Byte Order:             Little Endian
CPU(s):                   88
On-line CPU(s) list:   0-63
Off-line CPU(s) list:  64-87
Quoted 7 lines
# cat /proc/cmdline
BOOT_IMAGE=/vmlinuz-5.10.0-rc3-tlinux2-0050+ root=/dev/mapper/cl-root
ro rd.lvm.lv=cl/root rhgb quiet console=ttyS0 LANG=en_US
.UTF-8 no-kvmclock-vsyscall
# echo 1 > /sys/devices/system/cpu/cpu76/online
-bash: echo: write error: Cannot allocate memory

The original bug report is here.
https://bugzilla.kernel.org/show_bug.cgi?id=210213

Wanpeng