Skip to contentSkip to conversation

x86/virt/tdx: Fix lockdep assertion failure in cache flush for kexec

PATCH Superseded Kai Huang 4 participants 14 messages Mar 2 – Mar 10

Subsystems (3)

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Kai Huang

author
#0 lore raw

PATCH 1/1 x86/virt/tdx: Fix lockdep assertion failure in cache flush for kexec

TDX can leave the cache in an incoherent state for the memory it uses. During kexec the kernel does a WBINVD for each CPU before memory gets reused in the second kernel.

There were two considerations for where this WBINVD should happen. In order to handle cases where the cache might get into an incoherent state while the kexec is in the initial stages, it is needed to do this later in the kexec path, when the kexecing CPU stops all remote CPUs. However, the later kexec process is sensitive to existing races. So to avoid perturbing that operation, it is better to do it earlier.

The existing solution is to track the need for the kexec time WBINVD generically (i.e., not just for TDX) in a per-cpu var. The late invocation only happens if the earlier TDX specific logic in tdx_cpu_flush_cache_for_kexec() didn’t take care of the work. This earlier WBINVD logic was built into KVM’s existing syscore ops shutdown() handler, which is called earlier in the kexec path.

However, this accidentally added it to KVM’s unload path as well (also the "error path" when bringing up TDX during KVM module load), which uses the same internal functions. This makes some sense too, though, because if KVM is getting unloaded, TDX cache affecting operations will likely cease. So it is a good point to do the work before KVM is unloaded and won't have a chance to handle the shutdown operation in the future.

Unfortunately this KVM unload invocation triggers a lockdep warning in tdx_cpu_flush_cache_for_kexec(). Since tdx_cpu_flush_cache_for_kexec() is doing WBINVD on a specific CPU, it has an assert for preemption being disabled. This works fine for the kexec time invocation, but the KVM unload path calls this as part of a CPUHP callback for which, despite always executing on the target CPU, preemption is not disabled.

It might be better to add the earlier invocation logic to a dedicated arch/x86 TDX syscore shutdown() handler, but to make the fix more backport friendly just adjust the lockdep assert in the tdx_cpu_flush_cache_for_kexec().

The real requirement is tdx_cpu_flush_cache_for_kexec() must be done on the same CPU. It's OK that it can be preempted in the middle as long as it won't be rescheduled to another CPU.

Remove the too strong lockdep_assert_preemption_disabled(), and change this_cpu_{read|write}() to __this_cpu_{read|write}() which provide the more proper check (when CONFIG_DEBUG_PREEMPT is true), which checks all conditions that the context cannot be moved to another CPU to run in the middle.

Fixes 61221d07e815 ("KVM/TDX: Explicitly do WBINVD when no more TDX SEAMCALLs")

---

Hi Dave, Paolo, Sean,

/facepalm.

This was recently reported by Vishal. Sorry that I forgot to test the module unloading (but too focused on kexecing path, which doesn't have this issue). This wasn't caught by our CI because there's no such test case in CI. Right now we are adding this to the CI so it will be covered in the future.

---
 arch/x86/virt/vmx/tdx/tdx.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
arch/x86/virt/vmx/tdx/tdx.c +2 -4
index 8b8e165a2001..6f6be1df4b78 100644
@@ -1872,9 +1872,7 @@ EXPORT_SYMBOL_FOR_KVM(tdh_phymem_page_wbinvd_hkid);
 #ifdef CONFIG_KEXEC_CORE
 void tdx_cpu_flush_cache_for_kexec(void)
 {
-	lockdep_assert_preemption_disabled();
-
-	if (!this_cpu_read(cache_state_incoherent))
+	if (!__this_cpu_read(cache_state_incoherent))
 		return;
 
 	/*
@@ -1883,7 +1881,7 @@ void tdx_cpu_flush_cache_for_kexec(void)
 	 * there should be no more SEAMCALLs on this CPU.
 	 */
 	wbinvd();
-	this_cpu_write(cache_state_incoherent, false);
+	__this_cpu_write(cache_state_incoherent, false);
 }
 EXPORT_SYMBOL_FOR_KVM(tdx_cpu_flush_cache_for_kexec);
 #endif

base-commit: b5425f5406ee1b4bd84720f68020ef18ce380bab

Signature · 1 line
2.53.0
author
#1 lore raw

PATCH v2 1/1 x86/virt/tdx: Fix lockdep assertion failure in cache flush for kexec

TDX can leave the cache in an incoherent state for the memory it uses. During kexec the kernel does a WBINVD for each CPU before memory gets reused in the second kernel.

There were two considerations for where this WBINVD should happen. In order to handle cases where the cache might get into an incoherent state while the kexec is in the initial stages, it is needed to do this later in the kexec path, when the kexecing CPU stops all remote CPUs. However, the later kexec process is sensitive to existing races. So to avoid perturbing that operation, it is better to do it earlier.

The existing solution is to track the need for the kexec time WBINVD generically (i.e., not just for TDX) in a per-cpu var. The late invocation only happens if the earlier TDX specific logic in tdx_cpu_flush_cache_for_kexec() didn’t take care of the work. This earlier WBINVD logic was built into KVM’s existing syscore ops shutdown() handler, which is called earlier in the kexec path.

However, this accidentally added it to KVM’s unload path as well (also the "error path" when bringing up TDX during KVM module load), which uses the same internal functions. This makes some sense too, though, because if KVM is getting unloaded, TDX cache affecting operations will likely cease. So it is a good point to do the work before KVM is unloaded and won't have a chance to handle the shutdown operation in the future.

Unfortunately this KVM unload invocation triggers a lockdep warning in tdx_cpu_flush_cache_for_kexec(). Since tdx_cpu_flush_cache_for_kexec() is doing WBINVD on a specific CPU, it has an assert for preemption being disabled. This works fine for the kexec time invocation, but the KVM unload path calls this as part of a CPUHP callback for which, despite always executing on the target CPU, preemption is not disabled.

It might be better to add the earlier invocation logic to a dedicated arch/x86 TDX syscore shutdown() handler, but to make the fix more backport friendly just adjust the lockdep assert in the tdx_cpu_flush_cache_for_kexec().

The real requirement is tdx_cpu_flush_cache_for_kexec() must be done on the same CPU. It's OK that it can be preempted in the middle as long as it won't be rescheduled to another CPU.

Remove the too strong lockdep_assert_preemption_disabled(), and change this_cpu_{read|write}() to __this_cpu_{read|write}() which provide the more proper check (when CONFIG_DEBUG_PREEMPT is true), which checks all conditions that the context cannot be moved to another CPU to run in the middle.

Fixes 61221d07e815 ("KVM/TDX: Explicitly do WBINVD when no more TDX SEAMCALLs")

---

v1 -> v2:
- Improve changelog as discussed in v1.
- Also mention this can also be triggered in "error path" in changelog.

Hi Rick,

Are you OK with sending this patch out to public, or do you have more comments?

-- below is for public --

Hi Dave, Paolo, Sean,

/facepalm.

This was recently reported by Vishal. Sorry that I forgot to test the module unloading (but too focused on kexecing path, which doesn't have this issue). This wasn't caught by our CI because there's no such test case in CI. Right now we are adding this to the CI so it will be covered in the future.

---
 arch/x86/virt/vmx/tdx/tdx.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
arch/x86/virt/vmx/tdx/tdx.c +2 -4
index 8b8e165a2001..6f6be1df4b78 100644
@@ -1872,9 +1872,7 @@ EXPORT_SYMBOL_FOR_KVM(tdh_phymem_page_wbinvd_hkid);
 #ifdef CONFIG_KEXEC_CORE
 void tdx_cpu_flush_cache_for_kexec(void)
 {
-	lockdep_assert_preemption_disabled();
-
-	if (!this_cpu_read(cache_state_incoherent))
+	if (!__this_cpu_read(cache_state_incoherent))
 		return;
 
 	/*
@@ -1883,7 +1881,7 @@ void tdx_cpu_flush_cache_for_kexec(void)
 	 * there should be no more SEAMCALLs on this CPU.
 	 */
 	wbinvd();
-	this_cpu_write(cache_state_incoherent, false);
+	__this_cpu_write(cache_state_incoherent, false);
 }
 EXPORT_SYMBOL_FOR_KVM(tdx_cpu_flush_cache_for_kexec);
 #endif

base-commit: 7dff99b354601dd01829e1511711846e04340a69

Signature · 1 line
2.53.0
author
#2 lore raw
Quoted 4 lines

v1 -> v2:
- Improve changelog as discussed in v1.

Apologize, this is the internal version that I forgot to delete from my tree, and was sent out by accident.

Please ignore this.

#3 lore raw
In reply to Kai Huang
Quoted 33 lines · Kai Huang

TDX can leave the cache in an incoherent state for the memory it uses. During kexec the kernel does a WBINVD for each CPU before memory gets reused in the second kernel.

There were two considerations for where this WBINVD should happen. In order to handle cases where the cache might get into an incoherent state while the kexec is in the initial stages, it is needed to do this later in the kexec path, when the kexecing CPU stops all remote CPUs. However, the later kexec process is sensitive to existing races. So to avoid perturbing that operation, it is better to do it earlier.

The existing solution is to track the need for the kexec time WBINVD generically (i.e., not just for TDX) in a per-cpu var. The late invocation only happens if the earlier TDX specific logic in tdx_cpu_flush_cache_for_kexec() didn’t take care of the work. This earlier WBINVD logic was built into KVM’s existing syscore ops shutdown() handler, which is called earlier in the kexec path.

However, this accidentally added it to KVM’s unload path as well (also the "error path" when bringing up TDX during KVM module load), which uses the same internal functions. This makes some sense too, though, because if KVM is getting unloaded, TDX cache affecting operations will likely cease. So it is a good point to do the work before KVM is unloaded and won't have a chance to handle the shutdown operation in the future.

Unfortunately this KVM unload invocation triggers a lockdep warning in tdx_cpu_flush_cache_for_kexec(). Since tdx_cpu_flush_cache_for_kexec() is doing WBINVD on a specific CPU, it has an assert for preemption being disabled. This works fine for the kexec time invocation, but the KVM unload path calls this as part of a CPUHP callback for which, despite always executing on the target CPU, preemption is not disabled.

Quoted 8 lines

It might be better to add the earlier invocation logic to a dedicated arch/x86 TDX syscore shutdown() handler, but to make the fix more backport friendly just adjust the lockdep assert in the tdx_cpu_flush_cache_for_kexec().

The real requirement is tdx_cpu_flush_cache_for_kexec() must be done on the same CPU. It's OK that it can be preempted in the middle as long as it won't be rescheduled to another CPU.

TLDR: It wants migration disabled.

Quoted 7 lines

Remove the too strong lockdep_assert_preemption_disabled(), and change this_cpu_{read|write}() to __this_cpu_{read|write}() which provide the more proper check (when CONFIG_DEBUG_PREEMPT is true), which checks all conditions that the context cannot be moved to another CPU to run in the middle.

Quoted 5 lines
Fixes 61221d07e815 ("KVM/TDX: Explicitly do WBINVD when no more TDX SEAMCALLs")

So how exactly does this patch prevent the BUG: printk in check_preemption_disabled from triggering, if the lockdep assert was triggering?

author
#4 lore raw
Quoted 7 lines
Quoted 5 lines

The real requirement is tdx_cpu_flush_cache_for_kexec() must be done on the same CPU. It's OK that it can be preempted in the middle as long as it won't be rescheduled to another CPU.

TLDR: It wants migration disabled.

Basically yes.

Quoted 15 lines
Quoted 8 lines

Remove the too strong lockdep_assert_preemption_disabled(), and change this_cpu_{read|write}() to __this_cpu_{read|write}() which provide the more proper check (when CONFIG_DEBUG_PREEMPT is true), which checks all conditions that the context cannot be moved to another CPU to run in the middle.

Quoted 5 lines
Fixes 61221d07e815 ("KVM/TDX: Explicitly do WBINVD when no more TDX SEAMCALLs")
Quoted 3 lines

So how exactly does this patch prevent the BUG: printk in check_preemption_disabled from triggering, if the lockdep assert was triggering?

There's no real BUG here. It's just the
lockdep_assert_preemption_disabled() is misused.

Reviewed-by
#5 lore raw
Quoted 25 lines · Kai
Quoted 7 lines
Quoted 5 lines

The real requirement is tdx_cpu_flush_cache_for_kexec() must be done on the same CPU. It's OK that it can be preempted in the middle as long as it won't be rescheduled to another CPU.

TLDR: It wants migration disabled.

Basically yes.

Quoted 15 lines
Quoted 13 lines

Remove the too strong lockdep_assert_preemption_disabled(), and change this_cpu_{read|write}() to __this_cpu_{read|write}() which provide the more proper check (when CONFIG_DEBUG_PREEMPT is true), which checks all conditions that the context cannot be moved to another CPU to run in the middle.

Fixes 61221d07e815 ("KVM/TDX: Explicitly do WBINVD when no more TDX SEAMCALLs")
Quoted 6 lines
Quoted 3 lines

So how exactly does this patch prevent the BUG: printk in check_preemption_disabled from triggering, if the lockdep assert was triggering?

There's no real BUG here. It's just the
lockdep_assert_preemption_disabled() is misused.

Essentially in check_preemption_disabled() the check is considered passed IF ANY of the preempt disable conditions is met, i.e it's more laxed. So yeah, makes sense!

author
#6 lore raw
Quoted 5 lines
Quoted 5 lines
Quoted 4 lines

So how exactly does this patch prevent the BUG: printk in check_preemption_disabled from triggering, if the lockdep assert was triggering?

Quoted 8 lines
Quoted 2 lines

There's no real BUG here. It's just the
lockdep_assert_preemption_disabled() is misused.

Essentially in check_preemption_disabled() the check is considered passed IF ANY of the preempt disable conditions is met, i.e it's more laxed. So yeah, makes sense!

Thanks!

Acked-by on v2 1/1
#7 lore raw
In reply to Kai Huang
Quoted 48 lines · Kai Huang

TDX can leave the cache in an incoherent state for the memory it uses. During kexec the kernel does a WBINVD for each CPU before memory gets reused in the second kernel.

There were two considerations for where this WBINVD should happen. In order to handle cases where the cache might get into an incoherent state while the kexec is in the initial stages, it is needed to do this later in the kexec path, when the kexecing CPU stops all remote CPUs. However, the later kexec process is sensitive to existing races. So to avoid perturbing that operation, it is better to do it earlier.

The existing solution is to track the need for the kexec time WBINVD generically (i.e., not just for TDX) in a per-cpu var. The late invocation only happens if the earlier TDX specific logic in tdx_cpu_flush_cache_for_kexec() didn’t take care of the work. This earlier WBINVD logic was built into KVM’s existing syscore ops shutdown() handler, which is called earlier in the kexec path.

However, this accidentally added it to KVM’s unload path as well (also the "error path" when bringing up TDX during KVM module load), which uses the same internal functions. This makes some sense too, though, because if KVM is getting unloaded, TDX cache affecting operations will likely cease. So it is a good point to do the work before KVM is unloaded and won't have a chance to handle the shutdown operation in the future.

Unfortunately this KVM unload invocation triggers a lockdep warning in tdx_cpu_flush_cache_for_kexec(). Since tdx_cpu_flush_cache_for_kexec() is doing WBINVD on a specific CPU, it has an assert for preemption being disabled. This works fine for the kexec time invocation, but the KVM unload path calls this as part of a CPUHP callback for which, despite always executing on the target CPU, preemption is not disabled.

It might be better to add the earlier invocation logic to a dedicated arch/x86 TDX syscore shutdown() handler, but to make the fix more backport friendly just adjust the lockdep assert in the tdx_cpu_flush_cache_for_kexec().

The real requirement is tdx_cpu_flush_cache_for_kexec() must be done on the same CPU. It's OK that it can be preempted in the middle as long as it won't be rescheduled to another CPU.

Remove the too strong lockdep_assert_preemption_disabled(), and change this_cpu_{read|write}() to __this_cpu_{read|write}() which provide the more proper check (when CONFIG_DEBUG_PREEMPT is true), which checks all conditions that the context cannot be moved to another CPU to run in the middle.

Quoted 6 lines
Fixes 61221d07e815 ("KVM/TDX: Explicitly do WBINVD when no more TDX SEAMCALLs")

---

maintainer Reviewed-by on v1 1/1
#8 lore raw
In reply to Kai Huang
Quoted 50 lines · Kai Huang

TDX can leave the cache in an incoherent state for the memory it uses. During kexec the kernel does a WBINVD for each CPU before memory gets reused in the second kernel.

There were two considerations for where this WBINVD should happen. In order to handle cases where the cache might get into an incoherent state while the kexec is in the initial stages, it is needed to do this later in the kexec path, when the kexecing CPU stops all remote CPUs.  However, the later kexec process is sensitive to existing races.  So to avoid perturbing that operation, it is better to do it earlier.

The existing solution is to track the need for the kexec time WBINVD generically (i.e., not just for TDX) in a per-cpu var.  The late invocation only happens if the earlier TDX specific logic in tdx_cpu_flush_cache_for_kexec() didn’t take care of the work.  This earlier WBINVD logic was built into KVM’s existing syscore ops shutdown() handler, which is called earlier in the kexec path.

However, this accidentally added it to KVM’s unload path as well (also the "error path" when bringing up TDX during KVM module load), which uses the same internal functions.  This makes some sense too, though, because if KVM is getting unloaded, TDX cache affecting operations will likely cease.  So it is a good point to do the work before KVM is unloaded and won't have a chance to handle the shutdown operation in the future.

Unfortunately this KVM unload invocation triggers a lockdep warning in tdx_cpu_flush_cache_for_kexec().  Since tdx_cpu_flush_cache_for_kexec() is doing WBINVD on a specific CPU, it has an assert for preemption being disabled.  This works fine for the kexec time invocation, but the KVM unload path calls this as part of a CPUHP callback for which, despite always executing on the target CPU, preemption is not disabled.

It might be better to add the earlier invocation logic to a dedicated arch/x86 TDX syscore shutdown() handler, but to make the fix more backport friendly just adjust the lockdep assert in the tdx_cpu_flush_cache_for_kexec().

The real requirement is tdx_cpu_flush_cache_for_kexec() must be done on the same CPU.  It's OK that it can be preempted in the middle as long as it won't be rescheduled to another CPU.

Remove the too strong lockdep_assert_preemption_disabled(), and change this_cpu_{read|write}() to __this_cpu_{read|write}() which provide the more proper check (when CONFIG_DEBUG_PREEMPT is true), which checks all conditions that the context cannot be moved to another CPU to run in the middle.

Quoted 6 lines
Fixes 61221d07e815 ("KVM/TDX: Explicitly do WBINVD when no more TDX

SEAMCALLs")

But this issue is also solved by:
https://lore.kernel.org/kvm/[email protected]/

I guess that these changes are correct in either case. There is no need for the stricter asserts. But depending on the order the log would be confusing in the history when it talks about lockdep warnings. So we'll have to keep an eye on things. If this goes first, then it's fine.

You know, it might have helped to include the splat if you end up with a v2.

author
#9 lore raw
Quoted 58 lines · Rick P
Quoted 50 lines · Kai Huang

TDX can leave the cache in an incoherent state for the memory it uses. During kexec the kernel does a WBINVD for each CPU before memory gets reused in the second kernel.

There were two considerations for where this WBINVD should happen. In order to handle cases where the cache might get into an incoherent state while the kexec is in the initial stages, it is needed to do this later in the kexec path, when the kexecing CPU stops all remote CPUs.  However, the later kexec process is sensitive to existing races.  So to avoid perturbing that operation, it is better to do it earlier.

The existing solution is to track the need for the kexec time WBINVD generically (i.e., not just for TDX) in a per-cpu var.  The late invocation only happens if the earlier TDX specific logic in tdx_cpu_flush_cache_for_kexec() didn’t take care of the work.  This earlier WBINVD logic was built into KVM’s existing syscore ops shutdown() handler, which is called earlier in the kexec path.

However, this accidentally added it to KVM’s unload path as well (also the "error path" when bringing up TDX during KVM module load), which uses the same internal functions.  This makes some sense too, though, because if KVM is getting unloaded, TDX cache affecting operations will likely cease.  So it is a good point to do the work before KVM is unloaded and won't have a chance to handle the shutdown operation in the future.

Unfortunately this KVM unload invocation triggers a lockdep warning in tdx_cpu_flush_cache_for_kexec().  Since tdx_cpu_flush_cache_for_kexec() is doing WBINVD on a specific CPU, it has an assert for preemption being disabled.  This works fine for the kexec time invocation, but the KVM unload path calls this as part of a CPUHP callback for which, despite always executing on the target CPU, preemption is not disabled.

It might be better to add the earlier invocation logic to a dedicated arch/x86 TDX syscore shutdown() handler, but to make the fix more backport friendly just adjust the lockdep assert in the tdx_cpu_flush_cache_for_kexec().

The real requirement is tdx_cpu_flush_cache_for_kexec() must be done on the same CPU.  It's OK that it can be preempted in the middle as long as it won't be rescheduled to another CPU.

Remove the too strong lockdep_assert_preemption_disabled(), and change this_cpu_{read|write}() to __this_cpu_{read|write}() which provide the more proper check (when CONFIG_DEBUG_PREEMPT is true), which checks all conditions that the context cannot be moved to another CPU to run in the middle.

Quoted 6 lines
Fixes 61221d07e815 ("KVM/TDX: Explicitly do WBINVD when no more TDX

SEAMCALLs")

Quoted 4 lines

But this issue is also solved by:
https://lore.kernel.org/kvm/[email protected]/

This depends on Sean's series to move VMXON to x86 core, so it's not stable friendly.

Quoted 5 lines

I guess that these changes are correct in either case. There is no need for the stricter asserts. But depending on the order the log would be confusing in the history when it talks about lockdep warnings. So we'll have to keep an eye on things. If this goes first, then it's fine.

I see. Will keep this in mind.

Quoted 3 lines

You know, it might have helped to include the splat if you end up with a v2.

I thought lockdep warn should be obvious even w/o the actual splat, but fine I can include the splat if v2 is needed.

Hi Sean, Paolo, Kirill,

It would be good to merge this upstream and backport to stable. Appreciate if you can ack if it looks good to you? Thanks.

#10 lore raw
Quoted 15 lines · Kai Huang
Quoted 14 lines · Rick P
Quoted 12 lines · Kai Huang

Remove the too strong lockdep_assert_preemption_disabled(), and change this_cpu_{read|write}() to __this_cpu_{read|write}() which provide the more proper check (when CONFIG_DEBUG_PREEMPT is true), which checks all conditions that the context cannot be moved to another CPU to run in the middle.

Fixes 61221d07e815 ("KVM/TDX: Explicitly do WBINVD when no more TDX

SEAMCALLs")

Quoted 4 lines
Quoted 4 lines

But this issue is also solved by:
https://lore.kernel.org/kvm/[email protected]/

Even when that series comes along, I would rather have __this_cpu_{read|write}() instead of the explicit lockdep_assert_preemption_disabled(). Similar to the WARN about IRQs being disabled that got removed, explicitly requiring that preemption be disabled feels like a description of the current code, not an actual requirement.

Asserting that preemption is disabled gives the false impression that the current task must not be scheduled out, between reading and writing cache_state_incoherent. Which then raises the question of why scheduling out the current task is bad".

Quoted 4 lines

This depends on Sean's series to move VMXON to x86 core, so it's not stable friendly.

Quoted 1 line
Quoted 10 lines
Quoted 4 lines

I guess that these changes are correct in either case. There is no need for the stricter asserts. But depending on the order the log would be confusing in the history when it talks about lockdep warnings. So we'll have to keep an eye on things. If this goes first, then it's fine.

I see. Will keep this in mind.

Quoted 3 lines

You know, it might have helped to include the splat if you end up with a v2.

+1. I can read a backtrace about 10x faster than a full sentence describing the backtrace.

#11 lore raw
Quoted 4 lines · Sean Christopherson

Even when that series comes along, I would rather have __this_cpu_{read|write}() instead of the explicit lockdep_assert_preemption_disabled().  Similar to the WARN about IRQs being disabled that got removed, explicitly requiring that preemption be disabled feels like a description of the current code, not an actual requirement.

Agreed. It's just going to confusing if the log talks about a KVM unload path that is not there. The simpler answer is just merge this first.

author
#12 lore raw
In reply to Sean Christopherson
Quoted 21 lines · Sean Christopherson
Quoted 15 lines · Kai Huang
Quoted 14 lines · Rick P
Quoted 12 lines · Kai Huang

Remove the too strong lockdep_assert_preemption_disabled(), and change this_cpu_{read|write}() to __this_cpu_{read|write}() which provide the more proper check (when CONFIG_DEBUG_PREEMPT is true), which checks all conditions that the context cannot be moved to another CPU to run in the middle.

Fixes 61221d07e815 ("KVM/TDX: Explicitly do WBINVD when no more TDX

SEAMCALLs")

Quoted 4 lines
Quoted 4 lines

But this issue is also solved by:
https://lore.kernel.org/kvm/[email protected]/

Quoted 8 lines

Even when that series comes along, I would rather have __this_cpu_{read|write}() instead of the explicit lockdep_assert_preemption_disabled(). Similar to the WARN about IRQs being disabled that got removed, explicitly requiring that preemption be disabled feels like a description of the current code, not an actual requirement.

Asserting that preemption is disabled gives the false impression that the current task must not be scheduled out, between reading and writing cache_state_incoherent. Which then raises the question of why scheduling out the current task is bad".

Agreed.

Quoted 10 lines
Quoted 10 lines

This depends on Sean's series to move VMXON to x86 core, so it's not stable friendly.

Quoted 5 lines

I guess that these changes are correct in either case. There is no need for the stricter asserts. But depending on the order the log would be confusing in the history when it talks about lockdep warnings. So we'll have to keep an eye on things. If this goes first, then it's fine.

Quoted 8 lines
Quoted 5 lines

I see. Will keep this in mind.

Quoted 3 lines

You know, it might have helped to include the splat if you end up with a v2.

+1. I can read a backtrace about 10x faster than a full sentence describing the backtrace.

I'll include the actual WARN splat in v2.

Thanks for the ack!