Skip to contentSkip to conversation

x86/sgx: Synchronize encl->srcu in sgx_encl_release().

PATCH In review Jarkko Sakkinen 3 participants 8 messages Dec 11, 2020 – Dec 16, 2020

Subsystems (2)
lore
Share & tools
lore

Jarkko Sakkinen

author maintainer
#0 lore raw

PATCH 1/1 x86/sgx: Synchronize encl->srcu in sgx_encl_release().

Each sgx_mmun_notifier_release() starts a grace period, which means that one extra synchronize_rcu() in sgx_encl_release(). Add it there.

sgx_release() has the loop that drains the list but with bad luck the entry is already gone from the list before that loop processes it.

Fixes 1728ab54b4be ("x86/sgx: Add a page reclaimer")
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>
---
 arch/x86/kernel/cpu/sgx/encl.c | 7 +++++++
 1 file changed, 7 insertions(+)
arch/x86/kernel/cpu/sgx/encl.c +7 -0
index ee50a5010277..48539a6ee315 100644
@@ -438,6 +438,13 @@ void sgx_encl_release(struct kref *ref)
 	if (encl->backing)
 		fput(encl->backing);
 
+	/*
+	 * Each sgx_mmun_notifier_release() starts a grace period. Thus one
+	 * "extra" synchronize_rcu() is required here. This can go undetected by
+	 * sgx_release() when it drains the mm list.
+	 */
+	synchronize_srcu(&encl->srcu);
+
 	cleanup_srcu_struct(&encl->srcu);
 
 	WARN_ON_ONCE(!list_empty(&encl->mm_list));
Signature · 1 line
2.27.0
#1 lore raw
Quoted 1 line · Jarkko Sakkinen

Each sgx_mmun_notifier_release() starts a grace period, which means that

Should be sgx_mmu_notifier_release(), here and in the comment.

Quoted 4 lines

one extra synchronize_rcu() in sgx_encl_release(). Add it there.

sgx_release() has the loop that drains the list but with bad luck the entry is already gone from the list before that loop processes it.

Why not include the actual analysis that "proves" the bug? The splat that Haitao reported would also be useful info.

Quoted 4 lines
Fixes 1728ab54b4be ("x86/sgx: Add a page reclaimer")
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>

Haitao reported the bug, and for all intents and purposes provided the fix. I just did the analysis to verify that there was a legitimate bug and that the synchronization in sgx_encl_release() was indeed necessary.

Quoted 26 lines
---
 arch/x86/kernel/cpu/sgx/encl.c | 7 +++++++
 1 file changed, 7 insertions(+)
arch/x86/kernel/cpu/sgx/encl.c +7 -0
index ee50a5010277..48539a6ee315 100644
@@ -438,6 +438,13 @@ void sgx_encl_release(struct kref *ref)
 	if (encl->backing)
 		fput(encl->backing);
 
+	/*
+	 * Each sgx_mmun_notifier_release() starts a grace period. Thus one
+	 * "extra" synchronize_rcu() is required here. This can go undetected by
+	 * sgx_release() when it drains the mm list.
+	 */
+	synchronize_srcu(&encl->srcu);
+
 	cleanup_srcu_struct(&encl->srcu);
 
 	WARN_ON_ONCE(!list_empty(&encl->mm_list));
Signature · 1 line
2.27.0
author maintainer
#2 lore raw
Quoted 4 lines · Sean Christopherson
Quoted 1 line · Jarkko Sakkinen

Each sgx_mmun_notifier_release() starts a grace period, which means that

Should be sgx_mmu_notifier_release(), here and in the comment.

Thanks.

Quoted 7 lines
Quoted 4 lines

one extra synchronize_rcu() in sgx_encl_release(). Add it there.

sgx_release() has the loop that drains the list but with bad luck the entry is already gone from the list before that loop processes it.

Why not include the actual analysis that "proves" the bug? The splat that Haitao reported would also be useful info.

True. I can include a snippet of dmesg to the commit message.

Quoted 8 lines
Quoted 4 lines
Fixes 1728ab54b4be ("x86/sgx: Add a page reclaimer")
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>

Haitao reported the bug, and for all intents and purposes provided the fix. I just did the analysis to verify that there was a legitimate bug and that the synchronization in sgx_encl_release() was indeed necessary.

Good and valid point. The way I see it, the tags should be:

Haitao pointed out the bug but from your analysis I could resolve that this is the fix to implement, and was able to write the long description for the commit.

Does this make sense to you?

/Jarkko

author maintainer
#3 lore raw
Quoted 29 lines · Jarkko Sakkinen
Quoted 4 lines · Sean Christopherson
Quoted 1 line · Jarkko Sakkinen

Each sgx_mmun_notifier_release() starts a grace period, which means that

Should be sgx_mmu_notifier_release(), here and in the comment.

Thanks.

Quoted 7 lines
Quoted 4 lines

one extra synchronize_rcu() in sgx_encl_release(). Add it there.

sgx_release() has the loop that drains the list but with bad luck the entry is already gone from the list before that loop processes it.

Why not include the actual analysis that "proves" the bug? The splat that Haitao reported would also be useful info.

True. I can include a snippet of dmesg to the commit message.

Quoted 8 lines
Quoted 4 lines
Fixes 1728ab54b4be ("x86/sgx: Add a page reclaimer")
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>

Haitao reported the bug, and for all intents and purposes provided the fix. I just did the analysis to verify that there was a legitimate bug and that the synchronization in sgx_encl_release() was indeed necessary.

Good and valid point. The way I see it, the tags should be:

Quoted 8 lines

Haitao pointed out the bug but from your analysis I could resolve that this is the fix to implement, and was able to write the long description for the commit.

Does this make sense to you?

I'm sending v2 next week (this week on vacation).

/Jarkko

#4 lore raw
Quoted 42 lines · Jarkko Sakkinen
Quoted 41 lines · Jarkko Sakkinen
Quoted 2 lines · Sean Christopherson
Quoted 1 line · Jarkko Sakkinen

Each sgx_mmun_notifier_release() starts a grace period, which means

that

Quoted 2 lines

Should be sgx_mmu_notifier_release(), here and in the comment.

Thanks.

Quoted 3 lines
Quoted 3 lines

one extra synchronize_rcu() in sgx_encl_release(). Add it there.

sgx_release() has the loop that drains the list but with bad luck

the

Quoted 3 lines
Quoted 1 line

entry is already gone from the list before that loop processes it.

Why not include the actual analysis that "proves" the bug? The splat

that

Quoted 1 line

Haitao reported would also be useful info.

True. I can include a snippet of dmesg to the commit message.

Quoted 6 lines
Quoted 4 lines
Fixes 1728ab54b4be ("x86/sgx: Add a page reclaimer")
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>

Haitao reported the bug, and for all intents and purposes provided

the fix. I

Quoted 1 line

just did the analysis to verify that there was a legitimate bug and

that the

Quoted 1 line

synchronization in sgx_encl_release() was indeed necessary.

Good and valid point. The way I see it, the tags should be:

Haitao pointed out the bug but from your analysis I could resolve that this is the fix to implement, and was able to write the long description for the commit.

Quoted 5 lines
Quoted 1 line

Does this make sense to you?

I'm sending v2 next week (this week on vacation).

/Jarkko

I don't mind either how tags are assigned. But our testing reveals significant latency introduced in scenarios of heavy loading/unloading enclaves. synchronize_srcu_expedited fixed the issue. Please analyze and confirm if that's more appropriate than synchronize_srcu here.

author maintainer
#5 lore raw
Quoted 47 lines · Haitao Huang
Quoted 44 lines · Jarkko Sakkinen
Quoted 42 lines · Jarkko Sakkinen
Quoted 2 lines · Sean Christopherson
Quoted 1 line · Jarkko Sakkinen

Each sgx_mmun_notifier_release() starts a grace period, which

means that

Quoted 2 lines

Should be sgx_mmu_notifier_release(), here and in the comment.

Thanks.

Quoted 3 lines
Quoted 3 lines

one extra synchronize_rcu() in sgx_encl_release(). Add it there.

sgx_release() has the loop that drains the list but with bad

luck the

Quoted 3 lines
Quoted 1 line

entry is already gone from the list before that loop processes it.

Why not include the actual analysis that "proves" the bug? The

splat that

Quoted 1 line

Haitao reported would also be useful info.

True. I can include a snippet of dmesg to the commit message.

Quoted 6 lines
Quoted 4 lines
Fixes 1728ab54b4be ("x86/sgx: Add a page reclaimer")
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>

Haitao reported the bug, and for all intents and purposes provided

the fix. I

Quoted 1 line

just did the analysis to verify that there was a legitimate bug

and that the

Quoted 1 line

synchronization in sgx_encl_release() was indeed necessary.

Good and valid point. The way I see it, the tags should be:

Haitao pointed out the bug but from your analysis I could resolve that this is the fix to implement, and was able to write the long description for the commit.

Does this make sense to you?

Quoted 8 lines
Quoted 3 lines

I'm sending v2 next week (this week on vacation).

/Jarkko

I don't mind either how tags are assigned. But our testing reveals significant latency introduced in scenarios of heavy loading/unloading enclaves. synchronize_srcu_expedited fixed the issue. Please analyze and confirm if that's more appropriate than synchronize_srcu here.

I don't see any obvious reason why *_expedited could not be used here, as most of the time sync's are taken care of sgx_release() loop, and the final sync is with sgx_mmu_notifier_release(). More aggressive spinning should not do any harm here.

About the tags. I just try to get them right, and it is sometimes not straight-forward. So I guess, with all things considered, I'll put suggested-by from you. Once I get a refined patch out, try it out with your workloads and provide me tested-by, if it is working for you.

/Jarkko

#6 lore raw
In reply to Jarkko Sakkinen
Quoted 7 lines · Jarkko Sakkinen
Quoted 3 lines · Sean Christopherson

Haitao reported the bug, and for all intents and purposes provided the fix. I just did the analysis to verify that there was a legitimate bug and that the synchronization in sgx_encl_release() was indeed necessary.

Good and valid point. The way I see it, the tags should be:

Quoted 8 lines

Haitao pointed out the bug but from your analysis I could resolve that this is the fix to implement, and was able to write the long description for the commit.

Does this make sense to you?

Yep, works for me.

author maintainer
#7 lore raw
Quoted 11 lines · Sean Christopherson
Quoted 10 lines · Jarkko Sakkinen
Quoted 3 lines · Sean Christopherson

Haitao reported the bug, and for all intents and purposes provided the fix. I just did the analysis to verify that there was a legitimate bug and that the synchronization in sgx_encl_release() was indeed necessary.

Good and valid point. The way I see it, the tags should be:

Quoted 7 lines
Quoted 5 lines

Haitao pointed out the bug but from your analysis I could resolve that this is the fix to implement, and was able to write the long description for the commit.

Does this make sense to you?

Yep, works for me.

I'll just add two suggested-by's. Process guide does not forbid that and it best describes matters.

/Jarkko