[PATCH v5 0/7] Basic recovery for machine checks inside SGX

[PATCH v5 0/7] Basic recovery for machine checks inside SGX

[Tony Luck]

Now version 5.

Changes since v4:

Jarkko Sakkinen:
	+ Add linux-sgx@vger.kernel.org to Cc: list
	+ Remove explicit struct sgx_va_page *va_page type
	from argument and use in sgx_alloc_va_page(). Just
	use "void *" as this code doesn't do anything with the
	internals of struct sgx_va_page.
	+ Drop the union of all possible types for the "owner"
	field in struct sgx_epc_page (sorry Dave Hansen, this
	went in last time from your comment, but it doesn't
	seem to add much value). Back to "void *owner;"
	+ rename the xarray that tracks which addresses are
	EPC pages from "epc_page_ranges" to "sgx_epc_address_space".

Dave Hansen:
	+ Use more generic names for the globally visible
	functions that are needed in generic code:
		sgx_memory_failure -> arch_memory_failure
		sgx_is_epc_page -> arch_is_platform_page

Tony Luck:
	+ Found that ghes code spits warnings for memory addresses
	that it thinks are bad. Add a check for SGX pages.

Tony Luck (7):
  x86/sgx: Provide indication of life-cycle of EPC pages
  x86/sgx: Add infrastructure to identify SGX EPC pages
  x86/sgx: Initial poison handling for dirty and free pages
  x86/sgx: Add SGX infrastructure to recover from poison
  x86/sgx: Hook arch_memory_failure() into mainline code
  x86/sgx: Add hook to error injection address validation
  x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

 .../firmware-guide/acpi/apei/einj.rst         |  19 +++
 arch/x86/include/asm/processor.h              |   8 +
 arch/x86/include/asm/set_memory.h             |   4 +
 arch/x86/kernel/cpu/sgx/encl.c                |   5 +-
 arch/x86/kernel/cpu/sgx/encl.h                |   2 +-
 arch/x86/kernel/cpu/sgx/ioctl.c               |   2 +-
 arch/x86/kernel/cpu/sgx/main.c                | 140 ++++++++++++++++--
 arch/x86/kernel/cpu/sgx/sgx.h                 |  14 +-
 drivers/acpi/apei/einj.c                      |   3 +-
 drivers/acpi/apei/ghes.c                      |   2 +-
 include/linux/mm.h                            |  13 ++
 mm/memory-failure.c                           |  19 ++-
 12 files changed, 203 insertions(+), 28 deletions(-)


base-commit: 6880fa6c56601bb8ed59df6c30fd390cc5f6dd8f
-- 
2.31.1

[PATCH v5 1/7] x86/sgx: Provide indication of life-cycle of EPC pages

[Tony Luck]

SGX EPC pages go through the following life cycle:

        DIRTY ---> FREE ---> IN-USE --\
                    ^                 |
                    \-----------------/

Recovery action for poison for a DIRTY or FREE page is simple. Just
make sure never to allocate the page. IN-USE pages need some extra
handling.

It would be good to use the sgx_epc_page->owner field as an indicator
of where an EPC page is currently in that cycle (owner != NULL means
the EPC page is IN-USE). But there is one caller, sgx_alloc_va_page(),
that calls with NULL.

Since there are multiple uses of the "owner" field with different types
change the sgx_epc_page structure to define an anonymous union with
each of the uses explicitly called out.

Start epc_pages out with a non-NULL owner while they are in DIRTY state.

Fix up the one holdout to provide a non-NULL owner.

Refactor the allocation sequence so that changes to/from NULL
value happen together with adding/removing the epc_page from
a free list while the node->lock is held.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/encl.c  |  5 +++--
 arch/x86/kernel/cpu/sgx/encl.h  |  2 +-
 arch/x86/kernel/cpu/sgx/ioctl.c |  2 +-
 arch/x86/kernel/cpu/sgx/main.c  | 23 ++++++++++++-----------
 arch/x86/kernel/cpu/sgx/sgx.h   | 12 ++++++++----
 5 files changed, 25 insertions(+), 19 deletions(-)

diff --git a/arch/x86/kernel/cpu/sgx/encl.c b/arch/x86/kernel/cpu/sgx/encl.c
index 001808e3901c..ad8c61933b0a 100644
--- a/arch/x86/kernel/cpu/sgx/encl.c
+++ b/arch/x86/kernel/cpu/sgx/encl.c
@@ -667,6 +667,7 @@ int sgx_encl_test_and_clear_young(struct mm_struct *mm,
 
 /**
  * sgx_alloc_va_page() - Allocate a Version Array (VA) page
+ * @va_page:	struct sgx_va_page connected to this VA page
  *
  * Allocate a free EPC page and convert it to a Version Array (VA) page.
  *
@@ -674,12 +675,12 @@ int sgx_encl_test_and_clear_young(struct mm_struct *mm,
  *   a VA page,
  *   -errno otherwise
  */
-struct sgx_epc_page *sgx_alloc_va_page(void)
+struct sgx_epc_page *sgx_alloc_va_page(struct sgx_va_page *va_page)
 {
 	struct sgx_epc_page *epc_page;
 	int ret;
 
-	epc_page = sgx_alloc_epc_page(NULL, true);
+	epc_page = sgx_alloc_epc_page(va_page, true);
 	if (IS_ERR(epc_page))
 		return ERR_CAST(epc_page);
 
diff --git a/arch/x86/kernel/cpu/sgx/encl.h b/arch/x86/kernel/cpu/sgx/encl.h
index fec43ca65065..3d12dbeae14a 100644
--- a/arch/x86/kernel/cpu/sgx/encl.h
+++ b/arch/x86/kernel/cpu/sgx/encl.h
@@ -111,7 +111,7 @@ void sgx_encl_put_backing(struct sgx_backing *backing, bool do_write);
 int sgx_encl_test_and_clear_young(struct mm_struct *mm,
 				  struct sgx_encl_page *page);
 
-struct sgx_epc_page *sgx_alloc_va_page(void);
+struct sgx_epc_page *sgx_alloc_va_page(struct sgx_va_page *va_page);
 unsigned int sgx_alloc_va_slot(struct sgx_va_page *va_page);
 void sgx_free_va_slot(struct sgx_va_page *va_page, unsigned int offset);
 bool sgx_va_page_full(struct sgx_va_page *va_page);
diff --git a/arch/x86/kernel/cpu/sgx/ioctl.c b/arch/x86/kernel/cpu/sgx/ioctl.c
index 83df20e3e633..655ce0bb069d 100644
--- a/arch/x86/kernel/cpu/sgx/ioctl.c
+++ b/arch/x86/kernel/cpu/sgx/ioctl.c
@@ -30,7 +30,7 @@ static struct sgx_va_page *sgx_encl_grow(struct sgx_encl *encl)
 		if (!va_page)
 			return ERR_PTR(-ENOMEM);
 
-		va_page->epc_page = sgx_alloc_va_page();
+		va_page->epc_page = sgx_alloc_va_page(va_page);
 		if (IS_ERR(va_page->epc_page)) {
 			err = ERR_CAST(va_page->epc_page);
 			kfree(va_page);
diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 63d3de02bbcc..4a5b51d16133 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -457,7 +457,7 @@ static bool __init sgx_page_reclaimer_init(void)
 	return true;
 }
 
-static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
+static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(void *private, int nid)
 {
 	struct sgx_numa_node *node = &sgx_numa_nodes[nid];
 	struct sgx_epc_page *page = NULL;
@@ -471,6 +471,7 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
 
 	page = list_first_entry(&node->free_page_list, struct sgx_epc_page, list);
 	list_del_init(&page->list);
+	page->private = private;
 	sgx_nr_free_pages--;
 
 	spin_unlock(&node->lock);
@@ -480,6 +481,7 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
 
 /**
  * __sgx_alloc_epc_page() - Allocate an EPC page
+ * @owner:	the owner of the EPC page
  *
  * Iterate through NUMA nodes and reserve ia free EPC page to the caller. Start
  * from the NUMA node, where the caller is executing.
@@ -488,14 +490,14 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
  * - an EPC page:	A borrowed EPC pages were available.
  * - NULL:		Out of EPC pages.
  */
-struct sgx_epc_page *__sgx_alloc_epc_page(void)
+struct sgx_epc_page *__sgx_alloc_epc_page(void *private)
 {
 	struct sgx_epc_page *page;
 	int nid_of_current = numa_node_id();
 	int nid = nid_of_current;
 
 	if (node_isset(nid_of_current, sgx_numa_mask)) {
-		page = __sgx_alloc_epc_page_from_node(nid_of_current);
+		page = __sgx_alloc_epc_page_from_node(private, nid_of_current);
 		if (page)
 			return page;
 	}
@@ -506,7 +508,7 @@ struct sgx_epc_page *__sgx_alloc_epc_page(void)
 		if (nid == nid_of_current)
 			break;
 
-		page = __sgx_alloc_epc_page_from_node(nid);
+		page = __sgx_alloc_epc_page_from_node(private, nid);
 		if (page)
 			return page;
 	}
@@ -559,7 +561,7 @@ int sgx_unmark_page_reclaimable(struct sgx_epc_page *page)
 
 /**
  * sgx_alloc_epc_page() - Allocate an EPC page
- * @owner:	the owner of the EPC page
+ * @private:	per-caller private data
  * @reclaim:	reclaim pages if necessary
  *
  * Iterate through EPC sections and borrow a free EPC page to the caller. When a
@@ -574,16 +576,14 @@ int sgx_unmark_page_reclaimable(struct sgx_epc_page *page)
  *   an EPC page,
  *   -errno on error
  */
-struct sgx_epc_page *sgx_alloc_epc_page(void *owner, bool reclaim)
+struct sgx_epc_page *sgx_alloc_epc_page(void *private, bool reclaim)
 {
 	struct sgx_epc_page *page;
 
 	for ( ; ; ) {
-		page = __sgx_alloc_epc_page();
-		if (!IS_ERR(page)) {
-			page->owner = owner;
+		page = __sgx_alloc_epc_page(private);
+		if (!IS_ERR(page))
 			break;
-		}
 
 		if (list_empty(&sgx_active_page_list))
 			return ERR_PTR(-ENOMEM);
@@ -624,6 +624,7 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 
 	spin_lock(&node->lock);
 
+	page->private = NULL;
 	list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
 
@@ -652,7 +653,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
 		section->pages[i].flags = 0;
-		section->pages[i].owner = NULL;
+		section->pages[i].private = "dirty";
 		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
 	}
 
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index 4628acec0009..8b1be10a46f6 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -28,8 +28,12 @@
 
 struct sgx_epc_page {
 	unsigned int section;
-	unsigned int flags;
-	struct sgx_encl_page *owner;
+	int flags;
+	union {
+		void *private;
+		struct sgx_encl_page *owner;
+		struct sgx_encl_page *vepc;
+	};
 	struct list_head list;
 };
 
@@ -77,12 +81,12 @@ static inline void *sgx_get_epc_virt_addr(struct sgx_epc_page *page)
 	return section->virt_addr + index * PAGE_SIZE;
 }
 
-struct sgx_epc_page *__sgx_alloc_epc_page(void);
+struct sgx_epc_page *__sgx_alloc_epc_page(void *private);
 void sgx_free_epc_page(struct sgx_epc_page *page);
 
 void sgx_mark_page_reclaimable(struct sgx_epc_page *page);
 int sgx_unmark_page_reclaimable(struct sgx_epc_page *page);
-struct sgx_epc_page *sgx_alloc_epc_page(void *owner, bool reclaim);
+struct sgx_epc_page *sgx_alloc_epc_page(void *private, bool reclaim);
 
 #ifdef CONFIG_X86_SGX_KVM
 int __init sgx_vepc_init(void);
-- 
2.31.1

Re: [PATCH v5 1/7] x86/sgx: Provide indication of life-cycle of EPC pages

[Jarkko Sakkinen]

On Fri, 2021-09-17 at 14:38 -0700, Tony Luck wrote:
> SGX EPC pages go through the following life cycle:
> 
>         DIRTY ---> FREE ---> IN-USE --\
>                     ^                 |
>                     \-----------------/
> 
> Recovery action for poison for a DIRTY or FREE page is simple. Just
> make sure never to allocate the page. IN-USE pages need some extra
> handling.
> 
> It would be good to use the sgx_epc_page->owner field as an indicator
> of where an EPC page is currently in that cycle (owner != NULL means
> the EPC page is IN-USE). But there is one caller, sgx_alloc_va_page(),
> that calls with NULL.
> 
> Since there are multiple uses of the "owner" field with different types
> change the sgx_epc_page structure to define an anonymous union with
> each of the uses explicitly called out.

But it's still always a pointer.

And not only that, but two alternative fields in that union have *exactly* the
same type, so it's kind of artifically representing the problem more complex
than it really is.

I'm not just getting, why all this complexity, and not a few casts instead?

I neither get the rename of "owner" to "private". It serves very little value.
I'm not saying that "owner" is best name ever but it's not *that* confusing
either. That I'm sure that it is definitely not very productive to rename it.

Also there was still this "dirty". We could use ((void *)-1), which was also
suggested for earlier revisions.

/Jarkko

RE: [PATCH v5 1/7] x86/sgx: Provide indication of life-cycle of EPC pages

[Luck, Tony]

>> Since there are multiple uses of the "owner" field with different types
>> change the sgx_epc_page structure to define an anonymous union with
>> each of the uses explicitly called out.
>
> But it's still always a pointer.
>
> And not only that, but two alternative fields in that union have *exactly* the
> same type, so it's kind of artifically representing the problem more complex
> than it really is.

Bother! I seem to have jumbled some old bits of v4 into this series.

I agree that we just want "void *owner; here.  I even made the changes.
Then managed to lose them while updating.

I'll find the bits I lost and re-merge them in.

-Tony

Re: [PATCH v5 1/7] x86/sgx: Provide indication of life-cycle of EPC pages

[Jarkko Sakkinen]

On Tue, 2021-09-21 at 21:34 +0000, Luck, Tony wrote:
> > > Since there are multiple uses of the "owner" field with different types
> > > change the sgx_epc_page structure to define an anonymous union with
> > > each of the uses explicitly called out.
> > 
> > But it's still always a pointer.
> > 
> > And not only that, but two alternative fields in that union have *exactly* the
> > same type, so it's kind of artifically representing the problem more complex
> > than it really is.
> 
> Bother! I seem to have jumbled some old bits of v4 into this series.
> 
> I agree that we just want "void *owner; here.  I even made the changes.
> Then managed to lose them while updating.
> 
> I'll find the bits I lost and re-merge them in.
> 
> -Tony

Yeah, ok, cool, thank you. Just reporting what I was observing :-)

/Jarkko

Re: [PATCH v5 1/7] x86/sgx: Provide indication of life-cycle of EPC pages

[Dave Hansen]

On 9/21/21 2:28 PM, Jarkko Sakkinen wrote:
>> Since there are multiple uses of the "owner" field with different types
>> change the sgx_epc_page structure to define an anonymous union with
>> each of the uses explicitly called out.
> But it's still always a pointer.
> 
> And not only that, but two alternative fields in that union have *exactly* the
> same type, so it's kind of artifically representing the problem more complex
> than it really is.
> 
> I'm not just getting, why all this complexity, and not a few casts instead?

I suggested this.  It makes the structure more self-describing because
it explicitly lists the possibles uses of the space in the structure.

Maybe I stare at 'struct page' and its 4 unions too much and I'm
enamored by their shininess.  But, in the end, I prefer unions to casting.

Re: [PATCH v5 1/7] x86/sgx: Provide indication of life-cycle of EPC pages

[Jarkko Sakkinen]

On Tue, 2021-09-21 at 15:15 -0700, Dave Hansen wrote:
> On 9/21/21 2:28 PM, Jarkko Sakkinen wrote:
> > > Since there are multiple uses of the "owner" field with different types
> > > change the sgx_epc_page structure to define an anonymous union with
> > > each of the uses explicitly called out.
> > But it's still always a pointer.
> > 
> > And not only that, but two alternative fields in that union have *exactly* the
> > same type, so it's kind of artifically representing the problem more complex
> > than it really is.
> > 
> > I'm not just getting, why all this complexity, and not a few casts instead?
> 
> I suggested this.  It makes the structure more self-describing because
> it explicitly lists the possibles uses of the space in the structure.
> 
> Maybe I stare at 'struct page' and its 4 unions too much and I'm
> enamored by their shininess.  But, in the end, I prefer unions to casting.

Yeah, packing data into constrained space (as in the case of struct page) is
the only application for, where you can speak of a quantitative decision, when
you pick union.

/Jarkko

[PATCH v5 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[Tony Luck]

X86 machine check architecture reports a physical address when there
is a memory error. Handling that error requires a method to determine
whether the physical address reported is in any of the areas reserved
for EPC pages by BIOS.

SGX EPC pages do not have Linux "struct page" associated with them.

Keep track of the mapping from ranges of EPC pages to the sections
that contain them using an xarray.

Create a function arch_is_platform_page() that simply reports whether an address
is an EPC page for use elsewhere in the kernel. The ACPI error injection
code needs this function and is typically built as a module, so export it.

Note that arch_is_platform_page() will be slower than other similar "what type
is this page" functions that can simply check bits in the "struct page".
If there is some future performance critical user of this function it
may need to be implemented in a more efficient way.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 10 ++++++++++
 arch/x86/kernel/cpu/sgx/sgx.h  |  1 +
 2 files changed, 11 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 4a5b51d16133..10892513212d 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -20,6 +20,7 @@ struct sgx_epc_section sgx_epc_sections[SGX_MAX_EPC_SECTIONS];
 static int sgx_nr_epc_sections;
 static struct task_struct *ksgxd_tsk;
 static DECLARE_WAIT_QUEUE_HEAD(ksgxd_waitq);
+static DEFINE_XARRAY(epc_page_ranges);
 
 /*
  * These variables are part of the state of the reclaimer, and must be accessed
@@ -649,6 +650,9 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	}
 
 	section->phys_addr = phys_addr;
+	section->end_phys_addr = phys_addr + size - 1;
+	xa_store_range(&epc_page_ranges, section->phys_addr,
+		       section->end_phys_addr, section, GFP_KERNEL);
 
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
@@ -660,6 +664,12 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	return true;
 }
 
+bool arch_is_platform_page(u64 paddr)
+{
+	return !!xa_load(&epc_page_ranges, paddr);
+}
+EXPORT_SYMBOL_GPL(arch_is_platform_page);
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index 8b1be10a46f6..6a55b1971956 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -54,6 +54,7 @@ struct sgx_numa_node {
  */
 struct sgx_epc_section {
 	unsigned long phys_addr;
+	unsigned long end_phys_addr;
 	void *virt_addr;
 	struct sgx_epc_page *pages;
 	struct sgx_numa_node *node;
-- 
2.31.1

Re: [PATCH v5 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[Dave Hansen]

On 9/17/21 2:38 PM, Tony Luck wrote:
>  /*
>   * These variables are part of the state of the reclaimer, and must be accessed
> @@ -649,6 +650,9 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
>  	}
>  
>  	section->phys_addr = phys_addr;
> +	section->end_phys_addr = phys_addr + size - 1;
> +	xa_store_range(&epc_page_ranges, section->phys_addr,
> +		       section->end_phys_addr, section, GFP_KERNEL);

Did we ever figure out how much space storing really big ranges in the
xarray consumes?

RE: [PATCH v5 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[Luck, Tony]

>>  	section->phys_addr = phys_addr;
>> +	section->end_phys_addr = phys_addr + size - 1;
>> +	xa_store_range(&epc_page_ranges, section->phys_addr,
>> +		       section->end_phys_addr, section, GFP_KERNEL);
>
> Did we ever figure out how much space storing really big ranges in the
> xarray consumes?

No. Willy said the existing xarray code would be less than optimal with
this usage, but that things would be much better when he applied some
maple tree updates to the internals of xarray.

If there is some easy way to measure the memory backing an xarray I'm
happy to get the data. Or if someone else can synthesize it ... the two
ranges on my system that are added to the xarray are:

$ dmesg | grep -i sgx
[    8.496844] sgx: EPC section 0x8000c00000-0x807f7fffff
[    8.505118] sgx: EPC section 0x10000c00000-0x1007fffffff

I.e. two ranges of a bit under 2GB each.

But I don't think the overhead can be too hideous:

$ grep MemFree /proc/meminfo
MemFree:        1048682016 kB

I still have ~ 1TB free. Which is much greater that the 640 KB which should
be "enough for anybody" :-).

-Tony

Re: [PATCH v5 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[Dave Hansen]

On 9/21/21 1:50 PM, Luck, Tony wrote:
>> Did we ever figure out how much space storing really big ranges in the
>> xarray consumes?
> No. Willy said the existing xarray code would be less than optimal with
> this usage, but that things would be much better when he applied some
> maple tree updates to the internals of xarray.
> 
> If there is some easy way to measure the memory backing an xarray I'm
> happy to get the data. Or if someone else can synthesize it ... the two
> ranges on my system that are added to the xarray are:
> 
> $ dmesg | grep -i sgx
> [    8.496844] sgx: EPC section 0x8000c00000-0x807f7fffff
> [    8.505118] sgx: EPC section 0x10000c00000-0x1007fffffff
> 
> I.e. two ranges of a bit under 2GB each.
> 
> But I don't think the overhead can be too hideous:
> 
> $ grep MemFree /proc/meminfo
> MemFree:        1048682016 kB
> 
> I still have ~ 1TB free. Which is much greater that the 640 KB which should
> be "enough for anybody" :-).

There is a kmem_cache_create() for the xarray nodes.  So, you should be
able to see the difference in /proc/meminfo's "Slab" field.  Maybe boot
with init=/bin/sh to reduce the noise and look at meminfo both with and
without SGX your patch applied, or just with the xarray bits commented out.

I don't quite know how the data structures are munged, but xas_alloc()
makes it look like 'struct xa_node' is allocated from
radix_tree_node_cachep.  If that's the case, you should also be able to
see this in even more detail in:

# grep radix /proc/slabinfo
radix_tree_node   432305 482412    584   28    4 : tunables    0    0
 0 : slabdata  17229  17229      0

again, on a system with and without your new code enabled.

Re: [PATCH v5 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[Luck, Tony]

On Tue, Sep 21, 2021 at 03:32:14PM -0700, Dave Hansen wrote:
> On 9/21/21 1:50 PM, Luck, Tony wrote:
> >> Did we ever figure out how much space storing really big ranges in the
> >> xarray consumes?
> > No. Willy said the existing xarray code would be less than optimal with
> > this usage, but that things would be much better when he applied some
> > maple tree updates to the internals of xarray.
> > 
> > If there is some easy way to measure the memory backing an xarray I'm
> > happy to get the data. Or if someone else can synthesize it ... the two
> > ranges on my system that are added to the xarray are:
> > 
> > $ dmesg | grep -i sgx
> > [    8.496844] sgx: EPC section 0x8000c00000-0x807f7fffff
> > [    8.505118] sgx: EPC section 0x10000c00000-0x1007fffffff
> > 
> > I.e. two ranges of a bit under 2GB each.
> > 
> > But I don't think the overhead can be too hideous:
> > 
> > $ grep MemFree /proc/meminfo
> > MemFree:        1048682016 kB
> > 
> > I still have ~ 1TB free. Which is much greater that the 640 KB which should
> > be "enough for anybody" :-).
> 
> There is a kmem_cache_create() for the xarray nodes.  So, you should be
> able to see the difference in /proc/meminfo's "Slab" field.  Maybe boot
> with init=/bin/sh to reduce the noise and look at meminfo both with and
> without SGX your patch applied, or just with the xarray bits commented out.
> 
> I don't quite know how the data structures are munged, but xas_alloc()
> makes it look like 'struct xa_node' is allocated from
> radix_tree_node_cachep.  If that's the case, you should also be able to
> see this in even more detail in:
> 
> # grep radix /proc/slabinfo
> radix_tree_node   432305 482412    584   28    4 : tunables    0    0
>  0 : slabdata  17229  17229      0
> 
> again, on a system with and without your new code enabled.


Booting with init=/bin/sh and running that grep command right away at
the prompt:

With the xa_store_range() call commented out of my kernel:

radix_tree_node     9800   9968    584   56    8 : tunables    0    0    0 : slabdata    178    178      0


With xa_store_range() enabled:

radix_tree_node     9950  10136    584   56    8 : tunables    0    0    0 : slabdata    181    181      0



The head of the file says these are the field names:

# name            <active_objs> <num_objs> <objsize> <objperslab> <pagesperslab> : tunables <limit> <batchcount> <sharedfactor> : slabdata <active_slabs> <num_slabs> <sharedavail>

So I think this means that I have (9950 - 9800) * 584 = 87600 more bytes
allocated. Maybe that's a lot? But percentage-wise is seems in the
noise. E.g. We allocate one "struct sgx_epc_page" for each SGX page.
On my system I have 4GB of SGX EPC, so around 32 MB of these structures.

-Tony

Re: [PATCH v5 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[Dave Hansen]

On 9/21/21 4:48 PM, Luck, Tony wrote:
> 
> # name            <active_objs> <num_objs> <objsize> <objperslab> <pagesperslab> : tunables <limit> <batchcount> <sharedfactor> : slabdata <active_slabs> <num_slabs> <sharedavail>
> 
> So I think this means that I have (9950 - 9800) * 584 = 87600 more bytes
> allocated. Maybe that's a lot? But percentage-wise is seems in the
> noise. E.g. We allocate one "struct sgx_epc_page" for each SGX page.
> On my system I have 4GB of SGX EPC, so around 32 MB of these structures.

100k for 4GB of EPC is certainly in the noise as far as I'm concerned.

Thanks for checking this.

[PATCH v5 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Tony Luck]

A memory controller patrol scrubber can report poison in a page
that isn't currently being used.

Add "poison" field in the sgx_epc_page that can be set for an
sgx_epc_page. Check for it:
1) When sanitizing dirty pages
2) When freeing epc pages

Poison is a new field separated from flags to avoid having to make
all updates to flags atomic, or integrate poison state changes into
some other locking scheme to protect flags.

In both cases place the poisoned page on a list of poisoned epc pages
to make sure it will not be reallocated.

Add debugfs files /sys/kernel/debug/sgx/poison_page_list so that system
administrators get a list of those pages that have been dropped because
of poison.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 30 +++++++++++++++++++++++++++++-
 arch/x86/kernel/cpu/sgx/sgx.h  |  3 ++-
 2 files changed, 31 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 10892513212d..7a53ff876059 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -1,6 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0
 /*  Copyright(c) 2016-20 Intel Corporation. */
 
+#include <linux/debugfs.h>
 #include <linux/file.h>
 #include <linux/freezer.h>
 #include <linux/highmem.h>
@@ -43,6 +44,7 @@ static nodemask_t sgx_numa_mask;
 static struct sgx_numa_node *sgx_numa_nodes;
 
 static LIST_HEAD(sgx_dirty_page_list);
+static LIST_HEAD(sgx_poison_page_list);
 
 /*
  * Reset post-kexec EPC pages to the uninitialized state. The pages are removed
@@ -62,6 +64,12 @@ static void __sgx_sanitize_pages(struct list_head *dirty_page_list)
 
 		page = list_first_entry(dirty_page_list, struct sgx_epc_page, list);
 
+		if (page->poison) {
+			list_del(&page->list);
+			list_add(&page->list, &sgx_poison_page_list);
+			continue;
+		}
+
 		ret = __eremove(sgx_get_epc_virt_addr(page));
 		if (!ret) {
 			/*
@@ -626,7 +634,10 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 	spin_lock(&node->lock);
 
 	page->private = NULL;
-	list_add_tail(&page->list, &node->free_page_list);
+	if (page->poison)
+		list_add(&page->list, &sgx_poison_page_list);
+	else
+		list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
 
 	spin_unlock(&node->lock);
@@ -657,6 +668,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
 		section->pages[i].flags = 0;
+		section->pages[i].poison = 0;
 		section->pages[i].private = "dirty";
 		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
 	}
@@ -801,8 +813,21 @@ int sgx_set_attribute(unsigned long *allowed_attributes,
 }
 EXPORT_SYMBOL_GPL(sgx_set_attribute);
 
+static int poison_list_show(struct seq_file *m, void *private)
+{
+	struct sgx_epc_page *page;
+
+	list_for_each_entry(page, &sgx_poison_page_list, list)
+		seq_printf(m, "0x%lx\n", sgx_get_epc_phys_addr(page));
+
+	return 0;
+}
+
+DEFINE_SHOW_ATTRIBUTE(poison_list);
+
 static int __init sgx_init(void)
 {
+	struct dentry *dir;
 	int ret;
 	int i;
 
@@ -834,6 +859,9 @@ static int __init sgx_init(void)
 	if (sgx_vepc_init() && ret)
 		goto err_provision;
 
+	dir = debugfs_create_dir("sgx", arch_debugfs_dir);
+	debugfs_create_file("poison_page_list", 0400, dir, NULL, &poison_list_fops);
+
 	return 0;
 
 err_provision:
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index 6a55b1971956..77f3d98c9fbf 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -28,7 +28,8 @@
 
 struct sgx_epc_page {
 	unsigned int section;
-	int flags;
+	u16 flags;
+	u16 poison;
 	union {
 		void *private;
 		struct sgx_encl_page *owner;
-- 
2.31.1

[PATCH v5 4/7] x86/sgx: Add SGX infrastructure to recover from poison

[Tony Luck]

Provide a recovery function arch_memory_failure(). If the poison was
consumed synchronously then send a SIGBUS. Note that the virtual
address of the access is not included with the SIGBUS as is the case
for poison outside of SGX enclaves. This doesn't matter as addresses
of code/data inside an enclave is of little to no use to code executing
outside the (now dead) enclave.

Poison found in a free page results in the page being moved from the
free list to the poison page list.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 77 ++++++++++++++++++++++++++++++++++
 1 file changed, 77 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 7a53ff876059..8f23c8489cec 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -682,6 +682,83 @@ bool arch_is_platform_page(u64 paddr)
 }
 EXPORT_SYMBOL_GPL(arch_is_platform_page);
 
+static struct sgx_epc_page *sgx_paddr_to_page(u64 paddr)
+{
+	struct sgx_epc_section *section;
+
+	section = xa_load(&epc_page_ranges, paddr);
+	if (!section)
+		return NULL;
+
+	return &section->pages[PFN_DOWN(paddr - section->phys_addr)];
+}
+
+/*
+ * Called in process context to handle a hardware reported
+ * error in an SGX EPC page.
+ * If the MF_ACTION_REQUIRED bit is set in flags, then the
+ * context is the task that consumed the poison data. Otherwise
+ * this is called from a kernel thread unrelated to the page.
+ */
+int arch_memory_failure(unsigned long pfn, int flags)
+{
+	struct sgx_epc_page *page = sgx_paddr_to_page(pfn << PAGE_SHIFT);
+	struct sgx_epc_section *section;
+	struct sgx_numa_node *node;
+
+	/*
+	 * mm/memory-failure.c calls this routine for all errors
+	 * where there isn't a "struct page" for the address. But that
+	 * includes other address ranges besides SGX.
+	 */
+	if (!page)
+		return -ENXIO;
+
+	/*
+	 * If poison was consumed synchronously. Send a SIGBUS to
+	 * the task. Hardware has already exited the SGX enclave and
+	 * will not allow re-entry to an enclave that has a memory
+	 * error. The signal may help the task understand why the
+	 * enclave is broken.
+	 */
+	if (flags & MF_ACTION_REQUIRED)
+		force_sig(SIGBUS);
+
+	section = &sgx_epc_sections[page->section];
+	node = section->node;
+
+	spin_lock(&node->lock);
+
+	/* Already poisoned? Nothing more to do */
+	if (page->poison)
+		goto out;
+
+	page->poison = 1;
+
+	/*
+	 * If there is no owner, then the page is on a free list.
+	 * Move it to the poison page list.
+	 */
+	if (!page->private) {
+		list_del(&page->list);
+		list_add(&page->list, &sgx_poison_page_list);
+		goto out;
+	}
+
+	/*
+	 * TBD: Add additional plumbing to enable pre-emptive
+	 * action for asynchronous poison notification. Until
+	 * then just hope that the poison:
+	 * a) is not accessed - sgx_free_epc_page() will deal with it
+	 *    when the user gives it back
+	 * b) results in a recoverable machine check rather than
+	 *    a fatal one
+	 */
+out:
+	spin_unlock(&node->lock);
+	return 0;
+}
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
-- 
2.31.1

[PATCH v5 5/7] x86/sgx: Hook arch_memory_failure() into mainline code

[Tony Luck]

Add a call inside memory_failure() to check if the address is an SGX
EPC page and handle it.

Note the SGX EPC pages do not have a "struct page" entry, so the hook
goes in at the same point as the device mapping hook.

Pull the call to acquire the mutex earlier so the SGX errors are also
protected.

Make set_mce_nospec() skip SGX pages when trying to adjust
the 1:1 map.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/include/asm/processor.h  |  8 ++++++++
 arch/x86/include/asm/set_memory.h |  4 ++++
 include/linux/mm.h                | 13 +++++++++++++
 mm/memory-failure.c               | 19 +++++++++++++------
 4 files changed, 38 insertions(+), 6 deletions(-)

diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index 9ad2acaaae9b..4865f2860a4f 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -853,4 +853,12 @@ enum mds_mitigations {
 	MDS_MITIGATION_VMWERV,
 };
 
+#ifdef CONFIG_X86_SGX
+int arch_memory_failure(unsigned long pfn, int flags);
+#define arch_memory_failure arch_memory_failure
+
+bool arch_is_platform_page(u64 paddr);
+#define arch_is_platform_page arch_is_platform_page
+#endif
+
 #endif /* _ASM_X86_PROCESSOR_H */
diff --git a/arch/x86/include/asm/set_memory.h b/arch/x86/include/asm/set_memory.h
index 43fa081a1adb..ce8dd215f5b3 100644
--- a/arch/x86/include/asm/set_memory.h
+++ b/arch/x86/include/asm/set_memory.h
@@ -2,6 +2,7 @@
 #ifndef _ASM_X86_SET_MEMORY_H
 #define _ASM_X86_SET_MEMORY_H
 
+#include <linux/mm.h>
 #include <asm/page.h>
 #include <asm-generic/set_memory.h>
 
@@ -98,6 +99,9 @@ static inline int set_mce_nospec(unsigned long pfn, bool unmap)
 	unsigned long decoy_addr;
 	int rc;
 
+	/* SGX pages are not in the 1:1 map */
+	if (arch_is_platform_page(pfn << PAGE_SHIFT))
+		return 0;
 	/*
 	 * We would like to just call:
 	 *      set_memory_XX((unsigned long)pfn_to_kaddr(pfn), 1);
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 73a52aba448f..3cc63682fe47 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -3284,5 +3284,18 @@ static inline int seal_check_future_write(int seals, struct vm_area_struct *vma)
 	return 0;
 }
 
+#ifndef arch_memory_failure
+static inline int arch_memory_failure(unsigned long pfn, int flags)
+{
+	return -ENXIO;
+}
+#endif
+#ifndef arch_is_platform_page
+static inline bool arch_is_platform_page(u64 paddr)
+{
+	return false;
+}
+#endif
+
 #endif /* __KERNEL__ */
 #endif /* _LINUX_MM_H */
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 54879c339024..5693bac9509c 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -1632,21 +1632,28 @@ int memory_failure(unsigned long pfn, int flags)
 	if (!sysctl_memory_failure_recovery)
 		panic("Memory failure on page %lx", pfn);
 
+	mutex_lock(&mf_mutex);
+
 	p = pfn_to_online_page(pfn);
 	if (!p) {
+		res = arch_memory_failure(pfn, flags);
+		if (res == 0)
+			goto unlock_mutex;
+
 		if (pfn_valid(pfn)) {
 			pgmap = get_dev_pagemap(pfn, NULL);
-			if (pgmap)
-				return memory_failure_dev_pagemap(pfn, flags,
-								  pgmap);
+			if (pgmap) {
+				res = memory_failure_dev_pagemap(pfn, flags,
+								 pgmap);
+				goto unlock_mutex;
+			}
 		}
 		pr_err("Memory failure: %#lx: memory outside kernel control\n",
 			pfn);
-		return -ENXIO;
+		res = -ENXIO;
+		goto unlock_mutex;
 	}
 
-	mutex_lock(&mf_mutex);
-
 try_again:
 	if (PageHuge(p)) {
 		res = memory_failure_hugetlb(pfn, flags);
-- 
2.31.1

[PATCH v5 6/7] x86/sgx: Add hook to error injection address validation

[Tony Luck]

SGX reserved memory does not appear in the standard address maps.

Add hook to call into the SGX code to check if an address is located
in SGX memory.

There are other challenges in injecting errors into SGX. Update the
documentation with a sequence of operations to inject.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 .../firmware-guide/acpi/apei/einj.rst         | 19 +++++++++++++++++++
 drivers/acpi/apei/einj.c                      |  3 ++-
 2 files changed, 21 insertions(+), 1 deletion(-)

diff --git a/Documentation/firmware-guide/acpi/apei/einj.rst b/Documentation/firmware-guide/acpi/apei/einj.rst
index c042176e1707..55e2331a6438 100644
--- a/Documentation/firmware-guide/acpi/apei/einj.rst
+++ b/Documentation/firmware-guide/acpi/apei/einj.rst
@@ -181,5 +181,24 @@ You should see something like this in dmesg::
   [22715.834759] EDAC sbridge MC3: PROCESSOR 0:306e7 TIME 1422553404 SOCKET 0 APIC 0
   [22716.616173] EDAC MC3: 1 CE memory read error on CPU_SrcID#0_Channel#0_DIMM#0 (channel:0 slot:0 page:0x12345 offset:0x0 grain:32 syndrome:0x0 -  area:DRAM err_code:0001:0090 socket:0 channel_mask:1 rank:0)
 
+Special notes for injection into SGX enclaves:
+
+There may be a separate BIOS setup option to enable SGX injection.
+
+The injection process consists of setting some special memory controller
+trigger that will inject the error on the next write to the target
+address. But the h/w prevents any software outside of an SGX enclave
+from accessing enclave pages (even BIOS SMM mode).
+
+The following sequence can be used:
+  1) Determine physical address of enclave page
+  2) Use "notrigger=1" mode to inject (this will setup
+     the injection address, but will not actually inject)
+  3) Enter the enclave
+  4) Store data to the virtual address matching physical address from step 1
+  5) Execute CLFLUSH for that virtual address
+  6) Spin delay for 250ms
+  7) Read from the virtual address. This will trigger the error
+
 For more information about EINJ, please refer to ACPI specification
 version 4.0, section 17.5 and ACPI 5.0, section 18.6.
diff --git a/drivers/acpi/apei/einj.c b/drivers/acpi/apei/einj.c
index 2882450c443e..67c335baad52 100644
--- a/drivers/acpi/apei/einj.c
+++ b/drivers/acpi/apei/einj.c
@@ -544,7 +544,8 @@ static int einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
 	    ((region_intersects(base_addr, size, IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE)
 				!= REGION_INTERSECTS) &&
 	     (region_intersects(base_addr, size, IORESOURCE_MEM, IORES_DESC_PERSISTENT_MEMORY)
-				!= REGION_INTERSECTS)))
+				!= REGION_INTERSECTS) &&
+	     !arch_is_platform_page(base_addr)))
 		return -EINVAL;
 
 inject:
-- 
2.31.1

[PATCH v5 7/7] x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

[Tony Luck]

SGX EPC pages do not have a "struct page" associated with them so the
pfn_valid() sanity check fails and results in a warning message to
the console.

Add an additonal check to skip the warning if the address of the error
is in an SGX EPC page.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 drivers/acpi/apei/ghes.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 0c8330ed1ffd..0c5c9acc6254 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -449,7 +449,7 @@ static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 		return false;
 
 	pfn = PHYS_PFN(physical_addr);
-	if (!pfn_valid(pfn)) {
+	if (!pfn_valid(pfn) && !arch_is_platform_page(physical_addr)) {
 		pr_warn_ratelimited(FW_WARN GHES_PFX
 		"Invalid address in generic error data: %#llx\n",
 		physical_addr);
-- 
2.31.1

[PATCH v6 0/7] Basic recovery for machine checks inside SGX

[Tony Luck]

Now version 6 (what I actually meant to post as v5).

Note that I've dropped linux-kernel@vger.kernel.org and x86@kernel.org
from the distribution. Time to get some internal agreement on these
changes before bothering the x86 maintainers with yet another version.

So I'm looking for Acked-by: or Reviewed-by: on any bits of this
series that are worthy, and comments on the problems I need to fix
in the not-worthy parts.

Changes since v4 (I'm going to ignore the bogus v5 I posted):

Jarkko Sakkinen:
	+ Add linux-sgx@vger.kernel.org to Cc: list
	+ Remove explicit struct sgx_va_page *va_page type
	from argument and use in sgx_alloc_va_page(). Just
	use "void *" as this code doesn't do anything with the
	internals of struct sgx_va_page.
	+ Drop the union of all possible types for the "owner"
	field in struct sgx_epc_page (sorry Dave Hansen, this
	went in last time from your comment, but it doesn't
	seem to add much value). Back to "void *owner;"
	+ rename the xarray that tracks which addresses are
	EPC pages from "epc_page_ranges" to "sgx_epc_address_space".

Dave Hansen:
	+ Use more generic names for the globally visible
	functions that are needed in generic code:
		sgx_memory_failure -> arch_memory_failure
		sgx_is_epc_page -> arch_is_platform_page
	+ Commit comment on space used by xarray to track EPC pages.

Tony Luck:
	+ Found that ghes code spits warnings for memory addresses
	that it thinks are bad. Add a check for SGX pages.

Tony Luck (7):
  x86/sgx: Provide indication of life-cycle of EPC pages
  x86/sgx: Add infrastructure to identify SGX EPC pages
  x86/sgx: Initial poison handling for dirty and free pages
  x86/sgx: Add SGX infrastructure to recover from poison
  x86/sgx: Hook arch_memory_failure() into mainline code
  x86/sgx: Add hook to error injection address validation
  x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

 .../firmware-guide/acpi/apei/einj.rst         |  19 +++
 arch/x86/include/asm/processor.h              |   8 +
 arch/x86/include/asm/set_memory.h             |   4 +
 arch/x86/kernel/cpu/sgx/encl.c                |   5 +-
 arch/x86/kernel/cpu/sgx/encl.h                |   2 +-
 arch/x86/kernel/cpu/sgx/ioctl.c               |   2 +-
 arch/x86/kernel/cpu/sgx/main.c                | 137 ++++++++++++++++--
 arch/x86/kernel/cpu/sgx/sgx.h                 |   7 +-
 drivers/acpi/apei/einj.c                      |   3 +-
 drivers/acpi/apei/ghes.c                      |   2 +-
 include/linux/mm.h                            |  14 ++
 mm/memory-failure.c                           |  19 ++-
 12 files changed, 196 insertions(+), 26 deletions(-)


base-commit: e4e737bb5c170df6135a127739a9e6148ee3da82
-- 
2.31.1

[PATCH v6 1/7] x86/sgx: Provide indication of life-cycle of EPC pages

[Tony Luck]

SGX EPC pages go through the following life cycle:

        DIRTY ---> FREE ---> IN-USE --\
                    ^                 |
                    \-----------------/

Recovery action for poison for a DIRTY or FREE page is simple. Just
make sure never to allocate the page. IN-USE pages need some extra
handling.

It would be good to use the sgx_epc_page->owner field as an indicator
of where an EPC page is currently in that cycle (owner != NULL means
the EPC page is IN-USE). But there is one caller, sgx_alloc_va_page(),
that calls with NULL.

Since there are multiple uses of the "owner" field with different types
change the type of sgx_epc_page.owner to "void *.

Start epc_pages out with a non-NULL owner while they are in DIRTY state.

Fix up the one holdout to provide a non-NULL owner.

Refactor the allocation sequence so that changes to/from NULL
value happen together with adding/removing the epc_page from
a free list while the node->lock is held.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/encl.c  |  5 +++--
 arch/x86/kernel/cpu/sgx/encl.h  |  2 +-
 arch/x86/kernel/cpu/sgx/ioctl.c |  2 +-
 arch/x86/kernel/cpu/sgx/main.c  | 21 +++++++++++----------
 arch/x86/kernel/cpu/sgx/sgx.h   |  4 ++--
 5 files changed, 18 insertions(+), 16 deletions(-)

diff --git a/arch/x86/kernel/cpu/sgx/encl.c b/arch/x86/kernel/cpu/sgx/encl.c
index 001808e3901c..62cf20d5fbf6 100644
--- a/arch/x86/kernel/cpu/sgx/encl.c
+++ b/arch/x86/kernel/cpu/sgx/encl.c
@@ -667,6 +667,7 @@ int sgx_encl_test_and_clear_young(struct mm_struct *mm,
 
 /**
  * sgx_alloc_va_page() - Allocate a Version Array (VA) page
+ * @owner:	struct sgx_va_page connected to this VA page
  *
  * Allocate a free EPC page and convert it to a Version Array (VA) page.
  *
@@ -674,12 +675,12 @@ int sgx_encl_test_and_clear_young(struct mm_struct *mm,
  *   a VA page,
  *   -errno otherwise
  */
-struct sgx_epc_page *sgx_alloc_va_page(void)
+struct sgx_epc_page *sgx_alloc_va_page(void *owner)
 {
 	struct sgx_epc_page *epc_page;
 	int ret;
 
-	epc_page = sgx_alloc_epc_page(NULL, true);
+	epc_page = sgx_alloc_epc_page(owner, true);
 	if (IS_ERR(epc_page))
 		return ERR_CAST(epc_page);
 
diff --git a/arch/x86/kernel/cpu/sgx/encl.h b/arch/x86/kernel/cpu/sgx/encl.h
index fec43ca65065..2a972bc9b2d1 100644
--- a/arch/x86/kernel/cpu/sgx/encl.h
+++ b/arch/x86/kernel/cpu/sgx/encl.h
@@ -111,7 +111,7 @@ void sgx_encl_put_backing(struct sgx_backing *backing, bool do_write);
 int sgx_encl_test_and_clear_young(struct mm_struct *mm,
 				  struct sgx_encl_page *page);
 
-struct sgx_epc_page *sgx_alloc_va_page(void);
+struct sgx_epc_page *sgx_alloc_va_page(void *owner);
 unsigned int sgx_alloc_va_slot(struct sgx_va_page *va_page);
 void sgx_free_va_slot(struct sgx_va_page *va_page, unsigned int offset);
 bool sgx_va_page_full(struct sgx_va_page *va_page);
diff --git a/arch/x86/kernel/cpu/sgx/ioctl.c b/arch/x86/kernel/cpu/sgx/ioctl.c
index 83df20e3e633..655ce0bb069d 100644
--- a/arch/x86/kernel/cpu/sgx/ioctl.c
+++ b/arch/x86/kernel/cpu/sgx/ioctl.c
@@ -30,7 +30,7 @@ static struct sgx_va_page *sgx_encl_grow(struct sgx_encl *encl)
 		if (!va_page)
 			return ERR_PTR(-ENOMEM);
 
-		va_page->epc_page = sgx_alloc_va_page();
+		va_page->epc_page = sgx_alloc_va_page(va_page);
 		if (IS_ERR(va_page->epc_page)) {
 			err = ERR_CAST(va_page->epc_page);
 			kfree(va_page);
diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 63d3de02bbcc..69743709ec90 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -457,7 +457,7 @@ static bool __init sgx_page_reclaimer_init(void)
 	return true;
 }
 
-static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
+static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(void *owner, int nid)
 {
 	struct sgx_numa_node *node = &sgx_numa_nodes[nid];
 	struct sgx_epc_page *page = NULL;
@@ -471,6 +471,7 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
 
 	page = list_first_entry(&node->free_page_list, struct sgx_epc_page, list);
 	list_del_init(&page->list);
+	page->owner = owner;
 	sgx_nr_free_pages--;
 
 	spin_unlock(&node->lock);
@@ -480,6 +481,7 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
 
 /**
  * __sgx_alloc_epc_page() - Allocate an EPC page
+ * @owner:	the owner of the EPC page
  *
  * Iterate through NUMA nodes and reserve ia free EPC page to the caller. Start
  * from the NUMA node, where the caller is executing.
@@ -488,14 +490,14 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
  * - an EPC page:	A borrowed EPC pages were available.
  * - NULL:		Out of EPC pages.
  */
-struct sgx_epc_page *__sgx_alloc_epc_page(void)
+struct sgx_epc_page *__sgx_alloc_epc_page(void *owner)
 {
 	struct sgx_epc_page *page;
 	int nid_of_current = numa_node_id();
 	int nid = nid_of_current;
 
 	if (node_isset(nid_of_current, sgx_numa_mask)) {
-		page = __sgx_alloc_epc_page_from_node(nid_of_current);
+		page = __sgx_alloc_epc_page_from_node(owner, nid_of_current);
 		if (page)
 			return page;
 	}
@@ -506,7 +508,7 @@ struct sgx_epc_page *__sgx_alloc_epc_page(void)
 		if (nid == nid_of_current)
 			break;
 
-		page = __sgx_alloc_epc_page_from_node(nid);
+		page = __sgx_alloc_epc_page_from_node(owner, nid);
 		if (page)
 			return page;
 	}
@@ -559,7 +561,7 @@ int sgx_unmark_page_reclaimable(struct sgx_epc_page *page)
 
 /**
  * sgx_alloc_epc_page() - Allocate an EPC page
- * @owner:	the owner of the EPC page
+ * @owner:	per-caller page owner
  * @reclaim:	reclaim pages if necessary
  *
  * Iterate through EPC sections and borrow a free EPC page to the caller. When a
@@ -579,11 +581,9 @@ struct sgx_epc_page *sgx_alloc_epc_page(void *owner, bool reclaim)
 	struct sgx_epc_page *page;
 
 	for ( ; ; ) {
-		page = __sgx_alloc_epc_page();
-		if (!IS_ERR(page)) {
-			page->owner = owner;
+		page = __sgx_alloc_epc_page(owner);
+		if (!IS_ERR(page))
 			break;
-		}
 
 		if (list_empty(&sgx_active_page_list))
 			return ERR_PTR(-ENOMEM);
@@ -624,6 +624,7 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 
 	spin_lock(&node->lock);
 
+	page->owner = NULL;
 	list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
 
@@ -652,7 +653,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
 		section->pages[i].flags = 0;
-		section->pages[i].owner = NULL;
+		section->pages[i].owner = (void *)-1;
 		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
 	}
 
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index 4628acec0009..cc624778645f 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -29,7 +29,7 @@
 struct sgx_epc_page {
 	unsigned int section;
 	unsigned int flags;
-	struct sgx_encl_page *owner;
+	void *owner;
 	struct list_head list;
 };
 
@@ -77,7 +77,7 @@ static inline void *sgx_get_epc_virt_addr(struct sgx_epc_page *page)
 	return section->virt_addr + index * PAGE_SIZE;
 }
 
-struct sgx_epc_page *__sgx_alloc_epc_page(void);
+struct sgx_epc_page *__sgx_alloc_epc_page(void *owner);
 void sgx_free_epc_page(struct sgx_epc_page *page);
 
 void sgx_mark_page_reclaimable(struct sgx_epc_page *page);
-- 
2.31.1

Re: [PATCH v6 1/7] x86/sgx: Provide indication of life-cycle of EPC pages

[Jarkko Sakkinen]

On Wed, 2021-09-22 at 11:21 -0700, Tony Luck wrote:
> SGX EPC pages go through the following life cycle:
> 
>         DIRTY ---> FREE ---> IN-USE --\
>                     ^                 |
>                     \-----------------/
> 
> Recovery action for poison for a DIRTY or FREE page is simple. Just
> make sure never to allocate the page. IN-USE pages need some extra
> handling.
> 
> It would be good to use the sgx_epc_page->owner field as an indicator
> of where an EPC page is currently in that cycle (owner != NULL means
> the EPC page is IN-USE). But there is one caller, sgx_alloc_va_page(),
> that calls with NULL.
> 
> Since there are multiple uses of the "owner" field with different types
> change the type of sgx_epc_page.owner to "void *.
> 
> Start epc_pages out with a non-NULL owner while they are in DIRTY state.
> 
> Fix up the one holdout to provide a non-NULL owner.
> 
> Refactor the allocation sequence so that changes to/from NULL
> value happen together with adding/removing the epc_page from
> a free list while the node->lock is held.
> 
> Signed-off-by: Tony Luck <tony.luck@intel.com>
> ---
>  arch/x86/kernel/cpu/sgx/encl.c  |  5 +++--
>  arch/x86/kernel/cpu/sgx/encl.h  |  2 +-
>  arch/x86/kernel/cpu/sgx/ioctl.c |  2 +-
>  arch/x86/kernel/cpu/sgx/main.c  | 21 +++++++++++----------
>  arch/x86/kernel/cpu/sgx/sgx.h   |  4 ++--
>  5 files changed, 18 insertions(+), 16 deletions(-)
> 
> diff --git a/arch/x86/kernel/cpu/sgx/encl.c b/arch/x86/kernel/cpu/sgx/encl.c
> index 001808e3901c..62cf20d5fbf6 100644
> --- a/arch/x86/kernel/cpu/sgx/encl.c
> +++ b/arch/x86/kernel/cpu/sgx/encl.c
> @@ -667,6 +667,7 @@ int sgx_encl_test_and_clear_young(struct mm_struct *mm,
>  
>  /**
>   * sgx_alloc_va_page() - Allocate a Version Array (VA) page
> + * @owner:	struct sgx_va_page connected to this VA page
>   *
>   * Allocate a free EPC page and convert it to a Version Array (VA) page.
>   *
> @@ -674,12 +675,12 @@ int sgx_encl_test_and_clear_young(struct mm_struct *mm,
>   *   a VA page,
>   *   -errno otherwise
>   */
> -struct sgx_epc_page *sgx_alloc_va_page(void)
> +struct sgx_epc_page *sgx_alloc_va_page(void *owner)
>  {
>  	struct sgx_epc_page *epc_page;
>  	int ret;
>  
> -	epc_page = sgx_alloc_epc_page(NULL, true);
> +	epc_page = sgx_alloc_epc_page(owner, true);
>  	if (IS_ERR(epc_page))
>  		return ERR_CAST(epc_page);
>  
> diff --git a/arch/x86/kernel/cpu/sgx/encl.h b/arch/x86/kernel/cpu/sgx/encl.h
> index fec43ca65065..2a972bc9b2d1 100644
> --- a/arch/x86/kernel/cpu/sgx/encl.h
> +++ b/arch/x86/kernel/cpu/sgx/encl.h
> @@ -111,7 +111,7 @@ void sgx_encl_put_backing(struct sgx_backing *backing, bool do_write);
>  int sgx_encl_test_and_clear_young(struct mm_struct *mm,
>  				  struct sgx_encl_page *page);
>  
> -struct sgx_epc_page *sgx_alloc_va_page(void);
> +struct sgx_epc_page *sgx_alloc_va_page(void *owner);
>  unsigned int sgx_alloc_va_slot(struct sgx_va_page *va_page);
>  void sgx_free_va_slot(struct sgx_va_page *va_page, unsigned int offset);
>  bool sgx_va_page_full(struct sgx_va_page *va_page);
> diff --git a/arch/x86/kernel/cpu/sgx/ioctl.c b/arch/x86/kernel/cpu/sgx/ioctl.c
> index 83df20e3e633..655ce0bb069d 100644
> --- a/arch/x86/kernel/cpu/sgx/ioctl.c
> +++ b/arch/x86/kernel/cpu/sgx/ioctl.c
> @@ -30,7 +30,7 @@ static struct sgx_va_page *sgx_encl_grow(struct sgx_encl *encl)
>  		if (!va_page)
>  			return ERR_PTR(-ENOMEM);
>  
> -		va_page->epc_page = sgx_alloc_va_page();
> +		va_page->epc_page = sgx_alloc_va_page(va_page);
>  		if (IS_ERR(va_page->epc_page)) {
>  			err = ERR_CAST(va_page->epc_page);
>  			kfree(va_page);
> diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
> index 63d3de02bbcc..69743709ec90 100644
> --- a/arch/x86/kernel/cpu/sgx/main.c
> +++ b/arch/x86/kernel/cpu/sgx/main.c
> @@ -457,7 +457,7 @@ static bool __init sgx_page_reclaimer_init(void)
>  	return true;
>  }
>  
> -static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
> +static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(void *owner, int nid)
>  {
>  	struct sgx_numa_node *node = &sgx_numa_nodes[nid];
>  	struct sgx_epc_page *page = NULL;
> @@ -471,6 +471,7 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
>  
>  	page = list_first_entry(&node->free_page_list, struct sgx_epc_page, list);
>  	list_del_init(&page->list);
> +	page->owner = owner;
>  	sgx_nr_free_pages--;
>  
>  	spin_unlock(&node->lock);
> @@ -480,6 +481,7 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
>  
>  /**
>   * __sgx_alloc_epc_page() - Allocate an EPC page
> + * @owner:	the owner of the EPC page
>   *
>   * Iterate through NUMA nodes and reserve ia free EPC page to the caller. Start
>   * from the NUMA node, where the caller is executing.
> @@ -488,14 +490,14 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
>   * - an EPC page:	A borrowed EPC pages were available.
>   * - NULL:		Out of EPC pages.
>   */
> -struct sgx_epc_page *__sgx_alloc_epc_page(void)
> +struct sgx_epc_page *__sgx_alloc_epc_page(void *owner)
>  {
>  	struct sgx_epc_page *page;
>  	int nid_of_current = numa_node_id();
>  	int nid = nid_of_current;
>  
>  	if (node_isset(nid_of_current, sgx_numa_mask)) {
> -		page = __sgx_alloc_epc_page_from_node(nid_of_current);
> +		page = __sgx_alloc_epc_page_from_node(owner, nid_of_current);
>  		if (page)
>  			return page;
>  	}
> @@ -506,7 +508,7 @@ struct sgx_epc_page *__sgx_alloc_epc_page(void)
>  		if (nid == nid_of_current)
>  			break;
>  
> -		page = __sgx_alloc_epc_page_from_node(nid);
> +		page = __sgx_alloc_epc_page_from_node(owner, nid);
>  		if (page)
>  			return page;
>  	}
> @@ -559,7 +561,7 @@ int sgx_unmark_page_reclaimable(struct sgx_epc_page *page)
>  
>  /**
>   * sgx_alloc_epc_page() - Allocate an EPC page
> - * @owner:	the owner of the EPC page
> + * @owner:	per-caller page owner
>   * @reclaim:	reclaim pages if necessary
>   *
>   * Iterate through EPC sections and borrow a free EPC page to the caller. When a
> @@ -579,11 +581,9 @@ struct sgx_epc_page *sgx_alloc_epc_page(void *owner, bool reclaim)
>  	struct sgx_epc_page *page;
>  
>  	for ( ; ; ) {
> -		page = __sgx_alloc_epc_page();
> -		if (!IS_ERR(page)) {
> -			page->owner = owner;
> +		page = __sgx_alloc_epc_page(owner);
> +		if (!IS_ERR(page))
>  			break;
> -		}
>  
>  		if (list_empty(&sgx_active_page_list))
>  			return ERR_PTR(-ENOMEM);
> @@ -624,6 +624,7 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
>  
>  	spin_lock(&node->lock);
>  
> +	page->owner = NULL;
>  	list_add_tail(&page->list, &node->free_page_list);
>  	sgx_nr_free_pages++;
>  
> @@ -652,7 +653,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
>  	for (i = 0; i < nr_pages; i++) {
>  		section->pages[i].section = index;
>  		section->pages[i].flags = 0;
> -		section->pages[i].owner = NULL;
> +		section->pages[i].owner = (void *)-1;

Probably should have a named constant.

Anyway, I wonder why we want to do tricks with 'owner', when the
struct has a flags field?

Right now its use is so nice and straight forward, and most
importantly intuitive.

So what I would do instead of this, would be to add something
like

/* Pages, which are being tracked by the page reclaimer. */
#define SGX_EPC_PAGE_RECLAIMER_TRACKED	BIT(0)

/* Pages, which are allocated for use. */
#define SGX_EPC_PAGE_ALLOCATED		BIT(1)

This would be set by sgx_alloc_epc_page() and reset by
sgx_free_epc_page().

In the subsequent patch you could then instead of

       /*
        * If there is no owner, then the page is on a free list.
        * Move it to the poison page list.
        */
       if (!page->owner) {
               list_del(&page->list);
               list_add(&page->list, &sgx_poison_page_list);
               goto out;
       }

you would

       /*
        * If there is no owner, then the page is on a free list.
        * Move it to the poison page list.
        */
       if (!page->flags) {
               list_del(&page->list);
               list_add(&page->list, &sgx_poison_page_list);
               goto out;
       }

You don't actually need to compare to that flag because the
invariant would be that it is set, as long as the page is
not explicitly freed.

I think this is a better solution than in the patch set
because it does not introduce any unorthodox use of anything.

/Jarkko

Re: [PATCH v6 1/7] x86/sgx: Provide indication of life-cycle of EPC pages

[Jarkko Sakkinen]

On Thu, 2021-09-23 at 23:21 +0300, Jarkko Sakkinen wrote:
> On Wed, 2021-09-22 at 11:21 -0700, Tony Luck wrote:
> > SGX EPC pages go through the following life cycle:
> > 
> >         DIRTY ---> FREE ---> IN-USE --\
> >                     ^                 |
> >                     \-----------------/
> > 
> > Recovery action for poison for a DIRTY or FREE page is simple. Just
> > make sure never to allocate the page. IN-USE pages need some extra
> > handling.
> > 
> > It would be good to use the sgx_epc_page->owner field as an indicator
> > of where an EPC page is currently in that cycle (owner != NULL means
> > the EPC page is IN-USE). But there is one caller, sgx_alloc_va_page(),
> > that calls with NULL.
> > 
> > Since there are multiple uses of the "owner" field with different types
> > change the type of sgx_epc_page.owner to "void *.
> > 
> > Start epc_pages out with a non-NULL owner while they are in DIRTY state.
> > 
> > Fix up the one holdout to provide a non-NULL owner.
> > 
> > Refactor the allocation sequence so that changes to/from NULL
> > value happen together with adding/removing the epc_page from
> > a free list while the node->lock is held.
> > 
> > Signed-off-by: Tony Luck <tony.luck@intel.com>
> > ---
> >  arch/x86/kernel/cpu/sgx/encl.c  |  5 +++--
> >  arch/x86/kernel/cpu/sgx/encl.h  |  2 +-
> >  arch/x86/kernel/cpu/sgx/ioctl.c |  2 +-
> >  arch/x86/kernel/cpu/sgx/main.c  | 21 +++++++++++----------
> >  arch/x86/kernel/cpu/sgx/sgx.h   |  4 ++--
> >  5 files changed, 18 insertions(+), 16 deletions(-)
> > 
> > diff --git a/arch/x86/kernel/cpu/sgx/encl.c b/arch/x86/kernel/cpu/sgx/encl.c
> > index 001808e3901c..62cf20d5fbf6 100644
> > --- a/arch/x86/kernel/cpu/sgx/encl.c
> > +++ b/arch/x86/kernel/cpu/sgx/encl.c
> > @@ -667,6 +667,7 @@ int sgx_encl_test_and_clear_young(struct mm_struct *mm,
> >  
> >  /**
> >   * sgx_alloc_va_page() - Allocate a Version Array (VA) page
> > + * @owner:	struct sgx_va_page connected to this VA page
> >   *
> >   * Allocate a free EPC page and convert it to a Version Array (VA) page.
> >   *
> > @@ -674,12 +675,12 @@ int sgx_encl_test_and_clear_young(struct mm_struct *mm,
> >   *   a VA page,
> >   *   -errno otherwise
> >   */
> > -struct sgx_epc_page *sgx_alloc_va_page(void)
> > +struct sgx_epc_page *sgx_alloc_va_page(void *owner)
> >  {
> >  	struct sgx_epc_page *epc_page;
> >  	int ret;
> >  
> > -	epc_page = sgx_alloc_epc_page(NULL, true);
> > +	epc_page = sgx_alloc_epc_page(owner, true);
> >  	if (IS_ERR(epc_page))
> >  		return ERR_CAST(epc_page);
> >  
> > diff --git a/arch/x86/kernel/cpu/sgx/encl.h b/arch/x86/kernel/cpu/sgx/encl.h
> > index fec43ca65065..2a972bc9b2d1 100644
> > --- a/arch/x86/kernel/cpu/sgx/encl.h
> > +++ b/arch/x86/kernel/cpu/sgx/encl.h
> > @@ -111,7 +111,7 @@ void sgx_encl_put_backing(struct sgx_backing *backing, bool do_write);
> >  int sgx_encl_test_and_clear_young(struct mm_struct *mm,
> >  				  struct sgx_encl_page *page);
> >  
> > -struct sgx_epc_page *sgx_alloc_va_page(void);
> > +struct sgx_epc_page *sgx_alloc_va_page(void *owner);
> >  unsigned int sgx_alloc_va_slot(struct sgx_va_page *va_page);
> >  void sgx_free_va_slot(struct sgx_va_page *va_page, unsigned int offset);
> >  bool sgx_va_page_full(struct sgx_va_page *va_page);
> > diff --git a/arch/x86/kernel/cpu/sgx/ioctl.c b/arch/x86/kernel/cpu/sgx/ioctl.c
> > index 83df20e3e633..655ce0bb069d 100644
> > --- a/arch/x86/kernel/cpu/sgx/ioctl.c
> > +++ b/arch/x86/kernel/cpu/sgx/ioctl.c
> > @@ -30,7 +30,7 @@ static struct sgx_va_page *sgx_encl_grow(struct sgx_encl *encl)
> >  		if (!va_page)
> >  			return ERR_PTR(-ENOMEM);
> >  
> > -		va_page->epc_page = sgx_alloc_va_page();
> > +		va_page->epc_page = sgx_alloc_va_page(va_page);
> >  		if (IS_ERR(va_page->epc_page)) {
> >  			err = ERR_CAST(va_page->epc_page);
> >  			kfree(va_page);
> > diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
> > index 63d3de02bbcc..69743709ec90 100644
> > --- a/arch/x86/kernel/cpu/sgx/main.c
> > +++ b/arch/x86/kernel/cpu/sgx/main.c
> > @@ -457,7 +457,7 @@ static bool __init sgx_page_reclaimer_init(void)
> >  	return true;
> >  }
> >  
> > -static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
> > +static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(void *owner, int nid)
> >  {
> >  	struct sgx_numa_node *node = &sgx_numa_nodes[nid];
> >  	struct sgx_epc_page *page = NULL;
> > @@ -471,6 +471,7 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
> >  
> >  	page = list_first_entry(&node->free_page_list, struct sgx_epc_page, list);
> >  	list_del_init(&page->list);
> > +	page->owner = owner;
> >  	sgx_nr_free_pages--;
> >  
> >  	spin_unlock(&node->lock);
> > @@ -480,6 +481,7 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
> >  
> >  /**
> >   * __sgx_alloc_epc_page() - Allocate an EPC page
> > + * @owner:	the owner of the EPC page
> >   *
> >   * Iterate through NUMA nodes and reserve ia free EPC page to the caller. Start
> >   * from the NUMA node, where the caller is executing.
> > @@ -488,14 +490,14 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
> >   * - an EPC page:	A borrowed EPC pages were available.
> >   * - NULL:		Out of EPC pages.
> >   */
> > -struct sgx_epc_page *__sgx_alloc_epc_page(void)
> > +struct sgx_epc_page *__sgx_alloc_epc_page(void *owner)
> >  {
> >  	struct sgx_epc_page *page;
> >  	int nid_of_current = numa_node_id();
> >  	int nid = nid_of_current;
> >  
> >  	if (node_isset(nid_of_current, sgx_numa_mask)) {
> > -		page = __sgx_alloc_epc_page_from_node(nid_of_current);
> > +		page = __sgx_alloc_epc_page_from_node(owner, nid_of_current);
> >  		if (page)
> >  			return page;
> >  	}
> > @@ -506,7 +508,7 @@ struct sgx_epc_page *__sgx_alloc_epc_page(void)
> >  		if (nid == nid_of_current)
> >  			break;
> >  
> > -		page = __sgx_alloc_epc_page_from_node(nid);
> > +		page = __sgx_alloc_epc_page_from_node(owner, nid);
> >  		if (page)
> >  			return page;
> >  	}
> > @@ -559,7 +561,7 @@ int sgx_unmark_page_reclaimable(struct sgx_epc_page *page)
> >  
> >  /**
> >   * sgx_alloc_epc_page() - Allocate an EPC page
> > - * @owner:	the owner of the EPC page
> > + * @owner:	per-caller page owner
> >   * @reclaim:	reclaim pages if necessary
> >   *
> >   * Iterate through EPC sections and borrow a free EPC page to the caller. When a
> > @@ -579,11 +581,9 @@ struct sgx_epc_page *sgx_alloc_epc_page(void *owner, bool reclaim)
> >  	struct sgx_epc_page *page;
> >  
> >  	for ( ; ; ) {
> > -		page = __sgx_alloc_epc_page();
> > -		if (!IS_ERR(page)) {
> > -			page->owner = owner;
> > +		page = __sgx_alloc_epc_page(owner);
> > +		if (!IS_ERR(page))
> >  			break;
> > -		}
> >  
> >  		if (list_empty(&sgx_active_page_list))
> >  			return ERR_PTR(-ENOMEM);
> > @@ -624,6 +624,7 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
> >  
> >  	spin_lock(&node->lock);
> >  
> > +	page->owner = NULL;
> >  	list_add_tail(&page->list, &node->free_page_list);
> >  	sgx_nr_free_pages++;
> >  
> > @@ -652,7 +653,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
> >  	for (i = 0; i < nr_pages; i++) {
> >  		section->pages[i].section = index;
> >  		section->pages[i].flags = 0;
> > -		section->pages[i].owner = NULL;
> > +		section->pages[i].owner = (void *)-1;
> 
> Probably should have a named constant.
> 
> Anyway, I wonder why we want to do tricks with 'owner', when the
> struct has a flags field?
> 
> Right now its use is so nice and straight forward, and most
> importantly intuitive.
> 
> So what I would do instead of this, would be to add something
> like
> 
> /* Pages, which are being tracked by the page reclaimer. */
> #define SGX_EPC_PAGE_RECLAIMER_TRACKED	BIT(0)
> 
> /* Pages, which are allocated for use. */
> #define SGX_EPC_PAGE_ALLOCATED		BIT(1)
> 
> This would be set by sgx_alloc_epc_page() and reset by
> sgx_free_epc_page().
> 
> In the subsequent patch you could then instead of
> 
>        /*
>         * If there is no owner, then the page is on a free list.
>         * Move it to the poison page list.
>         */
>        if (!page->owner) {
>                list_del(&page->list);
>                list_add(&page->list, &sgx_poison_page_list);
>                goto out;
>        }
> 
> you would
> 
>        /*
>         * If there is no owner, then the page is on a free list.
>         * Move it to the poison page list.
>         */
>        if (!page->flags) {
>                list_del(&page->list);
>                list_add(&page->list, &sgx_poison_page_list);
>                goto out;
>        }
> 
> You don't actually need to compare to that flag because the
> invariant would be that it is set, as long as the page is
> not explicitly freed.
> 
> I think this is a better solution than in the patch set
> because it does not introduce any unorthodox use of anything.

And does not contain any special cases, e.g. when you debug
something you can always assume that a valid owner pointer is
always a legit sgx_encl_page instance.

/Jarkko

Re: [PATCH v6 1/7] x86/sgx: Provide indication of life-cycle of EPC pages

[Luck, Tony]

On Thu, Sep 23, 2021 at 11:24:35PM +0300, Jarkko Sakkinen wrote:
> On Thu, 2021-09-23 at 23:21 +0300, Jarkko Sakkinen wrote:
> > On Wed, 2021-09-22 at 11:21 -0700, Tony Luck wrote:
> > > -		section->pages[i].owner = NULL;
> > > +		section->pages[i].owner = (void *)-1;
> > 
> > Probably should have a named constant.
> > 
> > Anyway, I wonder why we want to do tricks with 'owner', when the
> > struct has a flags field?
> > 
> > Right now its use is so nice and straight forward, and most
> > importantly intuitive.
> > 
> > So what I would do instead of this, would be to add something
> > like
> > 
> > /* Pages, which are being tracked by the page reclaimer. */
> > #define SGX_EPC_PAGE_RECLAIMER_TRACKED	BIT(0)
> > 
> > /* Pages, which are allocated for use. */
> > #define SGX_EPC_PAGE_ALLOCATED		BIT(1)
> > 
> > This would be set by sgx_alloc_epc_page() and reset by
> > sgx_free_epc_page().
> > 
> > In the subsequent patch you could then instead of
> > 
> >        /*
> >         * If there is no owner, then the page is on a free list.
> >         * Move it to the poison page list.
> >         */
> >        if (!page->owner) {
> >                list_del(&page->list);
> >                list_add(&page->list, &sgx_poison_page_list);
> >                goto out;
> >        }
> > 
> > you would
> > 
> >        /*
> >         * If there is no owner, then the page is on a free list.
> >         * Move it to the poison page list.
> >         */
> >        if (!page->flags) {
> >                list_del(&page->list);
> >                list_add(&page->list, &sgx_poison_page_list);
> >                goto out;
> >        }
> > 
> > You don't actually need to compare to that flag because the
> > invariant would be that it is set, as long as the page is
> > not explicitly freed.
> > 
> > I think this is a better solution than in the patch set
> > because it does not introduce any unorthodox use of anything.
> 
> And does not contain any special cases, e.g. when you debug
> something you can always assume that a valid owner pointer is
> always a legit sgx_encl_page instance.
> 
Jarkko,

That's nice. It avoids having to create a fictitious owner for
the dirty pages, and for the sgx_alloc_va_page() case. Which
in turn means that the owner field in struct sgx_epc_page can
remain as "struct sgx_encl_page *owner;" (neatly avoiding DaveH's
request that it be an anonymous union of all the possible types,
because it is back to just being one type).

Thanks!  Will include in next version.

-Tony

RE: [PATCH v6 1/7] x86/sgx: Provide indication of life-cycle of EPC pages

[Luck, Tony]

> That's nice. It avoids having to create a fictitious owner for
> the dirty pages, and for the sgx_alloc_va_page() case. Which
> in turn means that the owner field in struct sgx_epc_page can
> remain as "struct sgx_encl_page *owner;" (neatly avoiding DaveH's
> request that it be an anonymous union of all the possible types,
> because it is back to just being one type).
>
> Thanks!  Will include in next version.

Also avoids a bunch of refactoring to make sure to set the owner field
while holding zone->lock.

I roughly coded it up and the old part 0001 was:

 arch/x86/kernel/cpu/sgx/encl.c  |    5 +++--
 arch/x86/kernel/cpu/sgx/encl.h  |    2 +-
 arch/x86/kernel/cpu/sgx/ioctl.c |    2 +-
 arch/x86/kernel/cpu/sgx/main.c  |   21 +++++++++++----------
 arch/x86/kernel/cpu/sgx/sgx.h   |    4 ++--
 5 files changed, 18 insertions(+), 16 deletions(-)

which is by no means huge, but the new part 0001 is

 arch/x86/kernel/cpu/sgx/main.c |    4 +++-
 arch/x86/kernel/cpu/sgx/sgx.h  |    3 +++
 2 files changed, 6 insertions(+), 1 deletion(-)

-Tony

Re: [PATCH v6 1/7] x86/sgx: Provide indication of life-cycle of EPC pages

[Jarkko Sakkinen]

On Thu, 2021-09-23 at 22:11 +0000, Luck, Tony wrote:
> > That's nice. It avoids having to create a fictitious owner for
> > the dirty pages, and for the sgx_alloc_va_page() case. Which
> > in turn means that the owner field in struct sgx_epc_page can
> > remain as "struct sgx_encl_page *owner;" (neatly avoiding DaveH's
> > request that it be an anonymous union of all the possible types,
> > because it is back to just being one type).
> > 
> > Thanks!  Will include in next version.
> 
> Also avoids a bunch of refactoring to make sure to set the owner field
> while holding zone->lock.
> 
> I roughly coded it up and the old part 0001 was:
> 
>  arch/x86/kernel/cpu/sgx/encl.c  |    5 +++--
>  arch/x86/kernel/cpu/sgx/encl.h  |    2 +-
>  arch/x86/kernel/cpu/sgx/ioctl.c |    2 +-
>  arch/x86/kernel/cpu/sgx/main.c  |   21 +++++++++++----------
>  arch/x86/kernel/cpu/sgx/sgx.h   |    4 ++--
>  5 files changed, 18 insertions(+), 16 deletions(-)
> 
> which is by no means huge, but the new part 0001 is
> 
>  arch/x86/kernel/cpu/sgx/main.c |    4 +++-
>  arch/x86/kernel/cpu/sgx/sgx.h  |    3 +++
>  2 files changed, 6 insertions(+), 1 deletion(-)
> 
> -Tony

This is good to hear. I guess it is then a no brainer to move into
this direction then.

/Jarkko

[PATCH v6 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[Tony Luck]

X86 machine check architecture reports a physical address when there
is a memory error. Handling that error requires a method to determine
whether the physical address reported is in any of the areas reserved
for EPC pages by BIOS.

SGX EPC pages do not have Linux "struct page" associated with them.

Keep track of the mapping from ranges of EPC pages to the sections
that contain them using an xarray.

Create a function arch_is_platform_page() that simply reports whether an address
is an EPC page for use elsewhere in the kernel. The ACPI error injection
code needs this function and is typically built as a module, so export it.

Note that arch_is_platform_page() will be slower than other similar "what type
is this page" functions that can simply check bits in the "struct page".
If there is some future performance critical user of this function it
may need to be implemented in a more efficient way.

Note also that the current implementation of xarray allocates a few
hundred kilobytes for this usage on a system with 4GB of SGX EPC memory
configured. This isn't ideal, but worth it for the code simplicity.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 69743709ec90..72a173b3affa 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -20,6 +20,7 @@ struct sgx_epc_section sgx_epc_sections[SGX_MAX_EPC_SECTIONS];
 static int sgx_nr_epc_sections;
 static struct task_struct *ksgxd_tsk;
 static DECLARE_WAIT_QUEUE_HEAD(ksgxd_waitq);
+static DEFINE_XARRAY(sgx_epc_address_space);
 
 /*
  * These variables are part of the state of the reclaimer, and must be accessed
@@ -649,6 +650,8 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	}
 
 	section->phys_addr = phys_addr;
+	xa_store_range(&sgx_epc_address_space, section->phys_addr,
+		       phys_addr + size - 1, section, GFP_KERNEL);
 
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
@@ -660,6 +663,12 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	return true;
 }
 
+bool arch_is_platform_page(u64 paddr)
+{
+	return !!xa_load(&sgx_epc_address_space, paddr);
+}
+EXPORT_SYMBOL_GPL(arch_is_platform_page);
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
-- 
2.31.1

[PATCH v6 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Tony Luck]

A memory controller patrol scrubber can report poison in a page
that isn't currently being used.

Add "poison" field in the sgx_epc_page that can be set for an
sgx_epc_page. Check for it:
1) When sanitizing dirty pages
2) When freeing epc pages

Poison is a new field separated from flags to avoid having to make
all updates to flags atomic, or integrate poison state changes into
some other locking scheme to protect flags.

In both cases place the poisoned page on a list of poisoned epc pages
to make sure it will not be reallocated.

Add debugfs files /sys/kernel/debug/sgx/poison_page_list so that system
administrators get a list of those pages that have been dropped because
of poison.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 30 +++++++++++++++++++++++++++++-
 arch/x86/kernel/cpu/sgx/sgx.h  |  3 ++-
 2 files changed, 31 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 72a173b3affa..91be079788ee 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -1,6 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0
 /*  Copyright(c) 2016-20 Intel Corporation. */
 
+#include <linux/debugfs.h>
 #include <linux/file.h>
 #include <linux/freezer.h>
 #include <linux/highmem.h>
@@ -43,6 +44,7 @@ static nodemask_t sgx_numa_mask;
 static struct sgx_numa_node *sgx_numa_nodes;
 
 static LIST_HEAD(sgx_dirty_page_list);
+static LIST_HEAD(sgx_poison_page_list);
 
 /*
  * Reset post-kexec EPC pages to the uninitialized state. The pages are removed
@@ -62,6 +64,12 @@ static void __sgx_sanitize_pages(struct list_head *dirty_page_list)
 
 		page = list_first_entry(dirty_page_list, struct sgx_epc_page, list);
 
+		if (page->poison) {
+			list_del(&page->list);
+			list_add(&page->list, &sgx_poison_page_list);
+			continue;
+		}
+
 		ret = __eremove(sgx_get_epc_virt_addr(page));
 		if (!ret) {
 			/*
@@ -626,7 +634,10 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 	spin_lock(&node->lock);
 
 	page->owner = NULL;
-	list_add_tail(&page->list, &node->free_page_list);
+	if (page->poison)
+		list_add(&page->list, &sgx_poison_page_list);
+	else
+		list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
 
 	spin_unlock(&node->lock);
@@ -656,6 +667,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
 		section->pages[i].flags = 0;
+		section->pages[i].poison = 0;
 		section->pages[i].owner = (void *)-1;
 		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
 	}
@@ -800,8 +812,21 @@ int sgx_set_attribute(unsigned long *allowed_attributes,
 }
 EXPORT_SYMBOL_GPL(sgx_set_attribute);
 
+static int poison_list_show(struct seq_file *m, void *private)
+{
+	struct sgx_epc_page *page;
+
+	list_for_each_entry(page, &sgx_poison_page_list, list)
+		seq_printf(m, "0x%lx\n", sgx_get_epc_phys_addr(page));
+
+	return 0;
+}
+
+DEFINE_SHOW_ATTRIBUTE(poison_list);
+
 static int __init sgx_init(void)
 {
+	struct dentry *dir;
 	int ret;
 	int i;
 
@@ -833,6 +858,9 @@ static int __init sgx_init(void)
 	if (sgx_vepc_init() && ret)
 		goto err_provision;
 
+	dir = debugfs_create_dir("sgx", arch_debugfs_dir);
+	debugfs_create_file("poison_page_list", 0400, dir, NULL, &poison_list_fops);
+
 	return 0;
 
 err_provision:
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index cc624778645f..9ba87bc3da61 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -28,7 +28,8 @@
 
 struct sgx_epc_page {
 	unsigned int section;
-	unsigned int flags;
+	u16 flags;
+	u16 poison;
 	void *owner;
 	struct list_head list;
 };
-- 
2.31.1

[PATCH v6 4/7] x86/sgx: Add SGX infrastructure to recover from poison

[Tony Luck]

Provide a recovery function sgx_memory_failure(). If the poison was
consumed synchronously then send a SIGBUS. Note that the virtual
address of the access is not included with the SIGBUS as is the case
for poison outside of SGX enclaves. This doesn't matter as addresses
of code/data inside an enclave is of little to no use to code executing
outside the (now dead) enclave.

Poison found in a free page results in the page being moved from the
free list to the poison page list.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 77 ++++++++++++++++++++++++++++++++++
 1 file changed, 77 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 91be079788ee..63d6b6d019d0 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -681,6 +681,83 @@ bool arch_is_platform_page(u64 paddr)
 }
 EXPORT_SYMBOL_GPL(arch_is_platform_page);
 
+static struct sgx_epc_page *sgx_paddr_to_page(u64 paddr)
+{
+	struct sgx_epc_section *section;
+
+	section = xa_load(&sgx_epc_address_space, paddr);
+	if (!section)
+		return NULL;
+
+	return &section->pages[PFN_DOWN(paddr - section->phys_addr)];
+}
+
+/*
+ * Called in process context to handle a hardware reported
+ * error in an SGX EPC page.
+ * If the MF_ACTION_REQUIRED bit is set in flags, then the
+ * context is the task that consumed the poison data. Otherwise
+ * this is called from a kernel thread unrelated to the page.
+ */
+int arch_memory_failure(unsigned long pfn, int flags)
+{
+	struct sgx_epc_page *page = sgx_paddr_to_page(pfn << PAGE_SHIFT);
+	struct sgx_epc_section *section;
+	struct sgx_numa_node *node;
+
+	/*
+	 * mm/memory-failure.c calls this routine for all errors
+	 * where there isn't a "struct page" for the address. But that
+	 * includes other address ranges besides SGX.
+	 */
+	if (!page)
+		return -ENXIO;
+
+	/*
+	 * If poison was consumed synchronously. Send a SIGBUS to
+	 * the task. Hardware has already exited the SGX enclave and
+	 * will not allow re-entry to an enclave that has a memory
+	 * error. The signal may help the task understand why the
+	 * enclave is broken.
+	 */
+	if (flags & MF_ACTION_REQUIRED)
+		force_sig(SIGBUS);
+
+	section = &sgx_epc_sections[page->section];
+	node = section->node;
+
+	spin_lock(&node->lock);
+
+	/* Already poisoned? Nothing more to do */
+	if (page->poison)
+		goto out;
+
+	page->poison = 1;
+
+	/*
+	 * If there is no owner, then the page is on a free list.
+	 * Move it to the poison page list.
+	 */
+	if (!page->owner) {
+		list_del(&page->list);
+		list_add(&page->list, &sgx_poison_page_list);
+		goto out;
+	}
+
+	/*
+	 * TBD: Add additional plumbing to enable pre-emptive
+	 * action for asynchronous poison notification. Until
+	 * then just hope that the poison:
+	 * a) is not accessed - sgx_free_epc_page() will deal with it
+	 *    when the user gives it back
+	 * b) results in a recoverable machine check rather than
+	 *    a fatal one
+	 */
+out:
+	spin_unlock(&node->lock);
+	return 0;
+}
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
-- 
2.31.1

[PATCH v6 5/7] x86/sgx: Hook arch_memory_failure() into mainline code

[Tony Luck]

Add a call inside memory_failure() to call the arch specific code
to check if the address is an SGX EPC page and handle it.

Note the SGX EPC pages do not have a "struct page" entry, so the hook
goes in at the same point as the device mapping hook.

Pull the call to acquire the mutex earlier so the SGX errors are also
protected.

Make set_mce_nospec() skip SGX pages when trying to adjust
the 1:1 map.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/include/asm/processor.h  |  8 ++++++++
 arch/x86/include/asm/set_memory.h |  4 ++++
 include/linux/mm.h                | 14 ++++++++++++++
 mm/memory-failure.c               | 19 +++++++++++++------
 4 files changed, 39 insertions(+), 6 deletions(-)

diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index 9ad2acaaae9b..4865f2860a4f 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -853,4 +853,12 @@ enum mds_mitigations {
 	MDS_MITIGATION_VMWERV,
 };
 
+#ifdef CONFIG_X86_SGX
+int arch_memory_failure(unsigned long pfn, int flags);
+#define arch_memory_failure arch_memory_failure
+
+bool arch_is_platform_page(u64 paddr);
+#define arch_is_platform_page arch_is_platform_page
+#endif
+
 #endif /* _ASM_X86_PROCESSOR_H */
diff --git a/arch/x86/include/asm/set_memory.h b/arch/x86/include/asm/set_memory.h
index 43fa081a1adb..ce8dd215f5b3 100644
--- a/arch/x86/include/asm/set_memory.h
+++ b/arch/x86/include/asm/set_memory.h
@@ -2,6 +2,7 @@
 #ifndef _ASM_X86_SET_MEMORY_H
 #define _ASM_X86_SET_MEMORY_H
 
+#include <linux/mm.h>
 #include <asm/page.h>
 #include <asm-generic/set_memory.h>
 
@@ -98,6 +99,9 @@ static inline int set_mce_nospec(unsigned long pfn, bool unmap)
 	unsigned long decoy_addr;
 	int rc;
 
+	/* SGX pages are not in the 1:1 map */
+	if (arch_is_platform_page(pfn << PAGE_SHIFT))
+		return 0;
 	/*
 	 * We would like to just call:
 	 *      set_memory_XX((unsigned long)pfn_to_kaddr(pfn), 1);
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 73a52aba448f..62b199ed5ec6 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -3284,5 +3284,19 @@ static inline int seal_check_future_write(int seals, struct vm_area_struct *vma)
 	return 0;
 }
 
+#ifndef arch_memory_failure
+static inline int arch_memory_failure(unsigned long pfn, int flags)
+{
+	return -ENXIO;
+}
+#endif
+
+#ifndef arch_is_platform_page
+static inline bool arch_is_platform_page(u64 paddr)
+{
+	return false;
+}
+#endif
+
 #endif /* __KERNEL__ */
 #endif /* _LINUX_MM_H */
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 54879c339024..5693bac9509c 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -1632,21 +1632,28 @@ int memory_failure(unsigned long pfn, int flags)
 	if (!sysctl_memory_failure_recovery)
 		panic("Memory failure on page %lx", pfn);
 
+	mutex_lock(&mf_mutex);
+
 	p = pfn_to_online_page(pfn);
 	if (!p) {
+		res = arch_memory_failure(pfn, flags);
+		if (res == 0)
+			goto unlock_mutex;
+
 		if (pfn_valid(pfn)) {
 			pgmap = get_dev_pagemap(pfn, NULL);
-			if (pgmap)
-				return memory_failure_dev_pagemap(pfn, flags,
-								  pgmap);
+			if (pgmap) {
+				res = memory_failure_dev_pagemap(pfn, flags,
+								 pgmap);
+				goto unlock_mutex;
+			}
 		}
 		pr_err("Memory failure: %#lx: memory outside kernel control\n",
 			pfn);
-		return -ENXIO;
+		res = -ENXIO;
+		goto unlock_mutex;
 	}
 
-	mutex_lock(&mf_mutex);
-
 try_again:
 	if (PageHuge(p)) {
 		res = memory_failure_hugetlb(pfn, flags);
-- 
2.31.1

[PATCH v6 6/7] x86/sgx: Add hook to error injection address validation

[Tony Luck]

SGX reserved memory does not appear in the standard address maps.

Add hook to call into the SGX code to check if an address is located
in SGX memory.

There are other challenges in injecting errors into SGX. Update the
documentation with a sequence of operations to inject.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 .../firmware-guide/acpi/apei/einj.rst         | 19 +++++++++++++++++++
 drivers/acpi/apei/einj.c                      |  3 ++-
 2 files changed, 21 insertions(+), 1 deletion(-)

diff --git a/Documentation/firmware-guide/acpi/apei/einj.rst b/Documentation/firmware-guide/acpi/apei/einj.rst
index c042176e1707..55e2331a6438 100644
--- a/Documentation/firmware-guide/acpi/apei/einj.rst
+++ b/Documentation/firmware-guide/acpi/apei/einj.rst
@@ -181,5 +181,24 @@ You should see something like this in dmesg::
   [22715.834759] EDAC sbridge MC3: PROCESSOR 0:306e7 TIME 1422553404 SOCKET 0 APIC 0
   [22716.616173] EDAC MC3: 1 CE memory read error on CPU_SrcID#0_Channel#0_DIMM#0 (channel:0 slot:0 page:0x12345 offset:0x0 grain:32 syndrome:0x0 -  area:DRAM err_code:0001:0090 socket:0 channel_mask:1 rank:0)
 
+Special notes for injection into SGX enclaves:
+
+There may be a separate BIOS setup option to enable SGX injection.
+
+The injection process consists of setting some special memory controller
+trigger that will inject the error on the next write to the target
+address. But the h/w prevents any software outside of an SGX enclave
+from accessing enclave pages (even BIOS SMM mode).
+
+The following sequence can be used:
+  1) Determine physical address of enclave page
+  2) Use "notrigger=1" mode to inject (this will setup
+     the injection address, but will not actually inject)
+  3) Enter the enclave
+  4) Store data to the virtual address matching physical address from step 1
+  5) Execute CLFLUSH for that virtual address
+  6) Spin delay for 250ms
+  7) Read from the virtual address. This will trigger the error
+
 For more information about EINJ, please refer to ACPI specification
 version 4.0, section 17.5 and ACPI 5.0, section 18.6.
diff --git a/drivers/acpi/apei/einj.c b/drivers/acpi/apei/einj.c
index 2882450c443e..67c335baad52 100644
--- a/drivers/acpi/apei/einj.c
+++ b/drivers/acpi/apei/einj.c
@@ -544,7 +544,8 @@ static int einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
 	    ((region_intersects(base_addr, size, IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE)
 				!= REGION_INTERSECTS) &&
 	     (region_intersects(base_addr, size, IORESOURCE_MEM, IORES_DESC_PERSISTENT_MEMORY)
-				!= REGION_INTERSECTS)))
+				!= REGION_INTERSECTS) &&
+	     !arch_is_platform_page(base_addr)))
 		return -EINVAL;
 
 inject:
-- 
2.31.1

[PATCH v6 7/7] x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

[Tony Luck]

SGX EPC pages do not have a "struct page" associated with them so the
pfn_valid() sanity check fails and results in a warning message to
the console.

Add an additonal check to skip the warning if the address of the error
is in an SGX EPC page.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 drivers/acpi/apei/ghes.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 0c8330ed1ffd..0c5c9acc6254 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -449,7 +449,7 @@ static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 		return false;
 
 	pfn = PHYS_PFN(physical_addr);
-	if (!pfn_valid(pfn)) {
+	if (!pfn_valid(pfn) && !arch_is_platform_page(physical_addr)) {
 		pr_warn_ratelimited(FW_WARN GHES_PFX
 		"Invalid address in generic error data: %#llx\n",
 		physical_addr);
-- 
2.31.1

[PATCH v7 0/7] Basic recovery for machine checks inside SGX

[Tony Luck]

Now version 7

Note that linux-kernel@vger.kernel.org and x86@kernel.org are still
dropped from the distribution. Time to get some internal agreement on these
changes before bothering the x86 maintainers with yet another version.

So I'm looking for Acked-by: or Reviewed-by: on any bits of this
series that are worthy, and comments on the problems I need to fix
in the not-worthy parts.

Changes since v6:

Jarkko Sakkinen:
	Don't use "owner" == NULL vs. != NULL as an indicator
	of whether an SGX EPC page is free vs. in-use. Just add
	a new flags bit. Note this drops most of the changes I
	had in part 0001. Remainder of the patches are largely
	unchanged except where they check for the new flags bit
	instead of owner != NULL.

Tony Luck (7):
  x86/sgx: Add new sgx_epc_page flag bit to mark in-use pages
  x86/sgx: Add infrastructure to identify SGX EPC pages
  x86/sgx: Initial poison handling for dirty and free pages
  x86/sgx: Add SGX infrastructure to recover from poison
  x86/sgx: Hook arch_memory_failure() into mainline code
  x86/sgx: Add hook to error injection address validation
  x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

 .../firmware-guide/acpi/apei/einj.rst         |  19 +++
 arch/x86/include/asm/processor.h              |   8 ++
 arch/x86/include/asm/set_memory.h             |   4 +
 arch/x86/kernel/cpu/sgx/main.c                | 121 +++++++++++++++++-
 arch/x86/kernel/cpu/sgx/sgx.h                 |   6 +-
 drivers/acpi/apei/einj.c                      |   3 +-
 drivers/acpi/apei/ghes.c                      |   2 +-
 include/linux/mm.h                            |  14 ++
 mm/memory-failure.c                           |  19 ++-
 9 files changed, 185 insertions(+), 11 deletions(-)


base-commit: 5816b3e6577eaa676ceb00a848f0fd65fe2adc29
-- 
2.31.1

[PATCH v7 1/7] x86/sgx: Add new sgx_epc_page flag bit to mark in-use pages

[Tony Luck]

SGX EPC pages go through the following life cycle:

        DIRTY ---> FREE ---> IN-USE --\
                    ^                 |
                    \-----------------/

Recovery action for poison for a DIRTY or FREE page is simple. Just
make sure never to allocate the page. IN-USE pages need some extra
handling.

Add a new flag bit SGX_EPC_PAGE_IN_USE that is set when a page
is allocated and cleared when the page is freed.

Notes:

1) These transitions are made while holding the node->lock so that
   future code that checks the flags while holding the node->lock
   can be sure that if the SGX_EPC_PAGE_IN_USE bit is set, then the
   page is on the free list.

2) Initially while the pages are on the dirty list the
   SGX_EPC_PAGE_IN_USE bit is set.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 4 +++-
 arch/x86/kernel/cpu/sgx/sgx.h  | 3 +++
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 63d3de02bbcc..d18988a46c13 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -472,6 +472,7 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
 	page = list_first_entry(&node->free_page_list, struct sgx_epc_page, list);
 	list_del_init(&page->list);
 	sgx_nr_free_pages--;
+	page->flags = SGX_EPC_PAGE_IN_USE;
 
 	spin_unlock(&node->lock);
 
@@ -626,6 +627,7 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 
 	list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
+	page->flags = 0;
 
 	spin_unlock(&node->lock);
 }
@@ -651,7 +653,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
-		section->pages[i].flags = 0;
+		section->pages[i].flags = SGX_EPC_PAGE_IN_USE;
 		section->pages[i].owner = NULL;
 		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
 	}
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index 4628acec0009..f9202d3d6278 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -26,6 +26,9 @@
 /* Pages, which are being tracked by the page reclaimer. */
 #define SGX_EPC_PAGE_RECLAIMER_TRACKED	BIT(0)
 
+/* Allocated pages */
+#define SGX_EPC_PAGE_IN_USE		BIT(1)
+
 struct sgx_epc_page {
 	unsigned int section;
 	unsigned int flags;
-- 
2.31.1

Re: [PATCH v7 1/7] x86/sgx: Add new sgx_epc_page flag bit to mark in-use pages

[Jarkko Sakkinen]

On Mon, 2021-09-27 at 14:34 -0700, Tony Luck wrote:
> SGX EPC pages go through the following life cycle:
> 
>         DIRTY ---> FREE ---> IN-USE --\
>                     ^                 |
>                     \-----------------/
> 
> Recovery action for poison for a DIRTY or FREE page is simple. Just
> make sure never to allocate the page. IN-USE pages need some extra
> handling.
> 
> Add a new flag bit SGX_EPC_PAGE_IN_USE that is set when a page
> is allocated and cleared when the page is freed.
> 
> Notes:
> 
> 1) These transitions are made while holding the node->lock so that
>    future code that checks the flags while holding the node->lock
>    can be sure that if the SGX_EPC_PAGE_IN_USE bit is set, then the
>    page is on the free list.
> 
> 2) Initially while the pages are on the dirty list the
>    SGX_EPC_PAGE_IN_USE bit is set.
> 
> Signed-off-by: Tony Luck <tony.luck@intel.com>

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>

/Jarkko

[PATCH v7 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[Tony Luck]

X86 machine check architecture reports a physical address when there
is a memory error. Handling that error requires a method to determine
whether the physical address reported is in any of the areas reserved
for EPC pages by BIOS.

SGX EPC pages do not have Linux "struct page" associated with them.

Keep track of the mapping from ranges of EPC pages to the sections
that contain them using an xarray.

Create a function arch_is_platform_page() that simply reports whether an address
is an EPC page for use elsewhere in the kernel. The ACPI error injection
code needs this function and is typically built as a module, so export it.

Note that arch_is_platform_page() will be slower than other similar "what type
is this page" functions that can simply check bits in the "struct page".
If there is some future performance critical user of this function it
may need to be implemented in a more efficient way.

Note also that the current implementation of xarray allocates a few
hundred kilobytes for this usage on a system with 4GB of SGX EPC memory
configured. This isn't ideal, but worth it for the code simplicity.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index d18988a46c13..09fa42690ff2 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -20,6 +20,7 @@ struct sgx_epc_section sgx_epc_sections[SGX_MAX_EPC_SECTIONS];
 static int sgx_nr_epc_sections;
 static struct task_struct *ksgxd_tsk;
 static DECLARE_WAIT_QUEUE_HEAD(ksgxd_waitq);
+static DEFINE_XARRAY(sgx_epc_address_space);
 
 /*
  * These variables are part of the state of the reclaimer, and must be accessed
@@ -650,6 +651,8 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	}
 
 	section->phys_addr = phys_addr;
+	xa_store_range(&sgx_epc_address_space, section->phys_addr,
+		       phys_addr + size - 1, section, GFP_KERNEL);
 
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
@@ -661,6 +664,12 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	return true;
 }
 
+bool arch_is_platform_page(u64 paddr)
+{
+	return !!xa_load(&sgx_epc_address_space, paddr);
+}
+EXPORT_SYMBOL_GPL(arch_is_platform_page);
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
-- 
2.31.1

Re: [PATCH v7 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[Jarkko Sakkinen]

On Mon, 2021-09-27 at 14:34 -0700, Tony Luck wrote:
> X86 machine check architecture reports a physical address when there
> is a memory error. Handling that error requires a method to determine
> whether the physical address reported is in any of the areas reserved
> for EPC pages by BIOS.
> 
> SGX EPC pages do not have Linux "struct page" associated with them.
> 
> Keep track of the mapping from ranges of EPC pages to the sections
> that contain them using an xarray.
> 
> Create a function arch_is_platform_page() that simply reports whether an address
> is an EPC page for use elsewhere in the kernel. The ACPI error injection
> code needs this function and is typically built as a module, so export it.
> 
> Note that arch_is_platform_page() will be slower than other similar "what type
> is this page" functions that can simply check bits in the "struct page".
> If there is some future performance critical user of this function it
> may need to be implemented in a more efficient way.
> 
> Note also that the current implementation of xarray allocates a few
> hundred kilobytes for this usage on a system with 4GB of SGX EPC memory
> configured. This isn't ideal, but worth it for the code simplicity.
> 
> Signed-off-by: Tony Luck <tony.luck@intel.com>
> ---
>  arch/x86/kernel/cpu/sgx/main.c | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
> index d18988a46c13..09fa42690ff2 100644
> --- a/arch/x86/kernel/cpu/sgx/main.c
> +++ b/arch/x86/kernel/cpu/sgx/main.c
> @@ -20,6 +20,7 @@ struct sgx_epc_section sgx_epc_sections[SGX_MAX_EPC_SECTIONS];
>  static int sgx_nr_epc_sections;
>  static struct task_struct *ksgxd_tsk;
>  static DECLARE_WAIT_QUEUE_HEAD(ksgxd_waitq);
> +static DEFINE_XARRAY(sgx_epc_address_space);
>  
>  /*
>   * These variables are part of the state of the reclaimer, and must be accessed
> @@ -650,6 +651,8 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
>  	}
>  
>  	section->phys_addr = phys_addr;
> +	xa_store_range(&sgx_epc_address_space, section->phys_addr,
> +		       phys_addr + size - 1, section, GFP_KERNEL);
>  
>  	for (i = 0; i < nr_pages; i++) {
>  		section->pages[i].section = index;
> @@ -661,6 +664,12 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
>  	return true;
>  }
>  
> +bool arch_is_platform_page(u64 paddr)
> +{
> +	return !!xa_load(&sgx_epc_address_space, paddr);
> +}
> +EXPORT_SYMBOL_GPL(arch_is_platform_page);
> +
>  /**
>   * A section metric is concatenated in a way that @low bits 12-31 define the
>   * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>

/Jarkko

[PATCH v7 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Tony Luck]

A memory controller patrol scrubber can report poison in a page
that isn't currently being used.

Add "poison" field in the sgx_epc_page that can be set for an
sgx_epc_page. Check for it:
1) When sanitizing dirty pages
2) When freeing epc pages

Poison is a new field separated from flags to avoid having to make
all updates to flags atomic, or integrate poison state changes into
some other locking scheme to protect flags.

In both cases place the poisoned page on a list of poisoned epc pages
to make sure it will not be reallocated.

Add debugfs files /sys/kernel/debug/sgx/poison_page_list so that system
administrators get a list of those pages that have been dropped because
of poison.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 31 ++++++++++++++++++++++++++++++-
 arch/x86/kernel/cpu/sgx/sgx.h  |  3 ++-
 2 files changed, 32 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 09fa42690ff2..b558c9a80af4 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -1,6 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0
 /*  Copyright(c) 2016-20 Intel Corporation. */
 
+#include <linux/debugfs.h>
 #include <linux/file.h>
 #include <linux/freezer.h>
 #include <linux/highmem.h>
@@ -43,6 +44,7 @@ static nodemask_t sgx_numa_mask;
 static struct sgx_numa_node *sgx_numa_nodes;
 
 static LIST_HEAD(sgx_dirty_page_list);
+static LIST_HEAD(sgx_poison_page_list);
 
 /*
  * Reset post-kexec EPC pages to the uninitialized state. The pages are removed
@@ -62,6 +64,12 @@ static void __sgx_sanitize_pages(struct list_head *dirty_page_list)
 
 		page = list_first_entry(dirty_page_list, struct sgx_epc_page, list);
 
+		if (page->poison) {
+			list_del(&page->list);
+			list_add(&page->list, &sgx_poison_page_list);
+			continue;
+		}
+
 		ret = __eremove(sgx_get_epc_virt_addr(page));
 		if (!ret) {
 			/*
@@ -626,7 +634,11 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 
 	spin_lock(&node->lock);
 
-	list_add_tail(&page->list, &node->free_page_list);
+	page->owner = NULL;
+	if (page->poison)
+		list_add(&page->list, &sgx_poison_page_list);
+	else
+		list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
 	page->flags = 0;
 
@@ -658,6 +670,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 		section->pages[i].section = index;
 		section->pages[i].flags = SGX_EPC_PAGE_IN_USE;
 		section->pages[i].owner = NULL;
+		section->pages[i].poison = 0;
 		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
 	}
 
@@ -801,8 +814,21 @@ int sgx_set_attribute(unsigned long *allowed_attributes,
 }
 EXPORT_SYMBOL_GPL(sgx_set_attribute);
 
+static int poison_list_show(struct seq_file *m, void *private)
+{
+	struct sgx_epc_page *page;
+
+	list_for_each_entry(page, &sgx_poison_page_list, list)
+		seq_printf(m, "0x%lx\n", sgx_get_epc_phys_addr(page));
+
+	return 0;
+}
+
+DEFINE_SHOW_ATTRIBUTE(poison_list);
+
 static int __init sgx_init(void)
 {
+	struct dentry *dir;
 	int ret;
 	int i;
 
@@ -834,6 +860,9 @@ static int __init sgx_init(void)
 	if (sgx_vepc_init() && ret)
 		goto err_provision;
 
+	dir = debugfs_create_dir("sgx", arch_debugfs_dir);
+	debugfs_create_file("poison_page_list", 0400, dir, NULL, &poison_list_fops);
+
 	return 0;
 
 err_provision:
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index f9202d3d6278..a990a4c9a00f 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -31,7 +31,8 @@
 
 struct sgx_epc_page {
 	unsigned int section;
-	unsigned int flags;
+	u16 flags;
+	u16 poison;
 	struct sgx_encl_page *owner;
 	struct list_head list;
 };
-- 
2.31.1

Re: [PATCH v7 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Jarkko Sakkinen]

On Mon, 2021-09-27 at 14:34 -0700, Tony Luck wrote:
> A memory controller patrol scrubber can report poison in a page
> that isn't currently being used.
> 
> Add "poison" field in the sgx_epc_page that can be set for an
> sgx_epc_page. Check for it:
> 1) When sanitizing dirty pages
> 2) When freeing epc pages
> 
> Poison is a new field separated from flags to avoid having to make
> all updates to flags atomic, or integrate poison state changes into
> some other locking scheme to protect flags.
> 
> In both cases place the poisoned page on a list of poisoned epc pages
> to make sure it will not be reallocated.
> 
> Add debugfs files /sys/kernel/debug/sgx/poison_page_list so that system
> administrators get a list of those pages that have been dropped because
> of poison.

So, what would a sysadmin do with that detailed information?

I would decrease the granularity a bit rather add something like
this for each node:

	/sys/devices/system/node/node[0-9]*/sgx/poisoned_size

which would give the total amount of poisoned memory in bytes
for that node.

See the series that I've recently posted:

https://lore.kernel.org/linux-sgx/20210914030422.377601-1-jarkko@kernel.org/T/#t

/Jarkko

RE: [PATCH v7 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Luck, Tony]

>> Add debugfs files /sys/kernel/debug/sgx/poison_page_list so that system
>> administrators get a list of those pages that have been dropped because
>> of poison.
>
> So, what would a sysadmin do with that detailed information?

It's going to be a rare case that there are any poisoned pages on that list
(a large enough cluster will have some systems that have uncorrected
recoverable errors in SGX EPC memory).

Even when there are some poisoned pages, there will only be a few. Systems
that have thousands of pages with uncorrected memory errors will surely crash
because one of those errors is going to either trigger an error marked as fatal,
or the error won’t be recoverable by Linux because it is in kernel memory.

A sysadmin might add a script to run during system shutdown (or periodically
during run-time) to save the poison page list. Then at startup run:

for addr in `cat saved_sgx_poison_page_list`
do
	echo $addr > /sys/devices/system/memory/hard_offline_page
done

to make poison persistent across reboots.

-Tony

Re: [PATCH v7 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Jarkko Sakkinen]

On Tue, 2021-09-28 at 15:41 +0000, Luck, Tony wrote:
> > > Add debugfs files /sys/kernel/debug/sgx/poison_page_list so that system
> > > administrators get a list of those pages that have been dropped because
> > > of poison.
> > 
> > So, what would a sysadmin do with that detailed information?
> 
> It's going to be a rare case that there are any poisoned pages on that list
> (a large enough cluster will have some systems that have uncorrected
> recoverable errors in SGX EPC memory).
> 
> Even when there are some poisoned pages, there will only be a few. Systems
> that have thousands of pages with uncorrected memory errors will surely crash
> because one of those errors is going to either trigger an error marked as fatal,
> or the error won’t be recoverable by Linux because it is in kernel memory.
> 
> A sysadmin might add a script to run during system shutdown (or periodically
> during run-time) to save the poison page list. Then at startup run:
> 
> for addr in `cat saved_sgx_poison_page_list`
> do
> 	echo $addr > /sys/devices/system/memory/hard_offline_page
> done
> 
> to make poison persistent across reboots.
> 
> -Tony

Couldn't it be a blob with 8 bytes for each address?

/Jarkko

Re: [PATCH v7 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Luck, Tony]

On Tue, Sep 28, 2021 at 11:11:30PM +0300, Jarkko Sakkinen wrote:
> On Tue, 2021-09-28 at 15:41 +0000, Luck, Tony wrote:
> > > > Add debugfs files /sys/kernel/debug/sgx/poison_page_list so that system
> > > > administrators get a list of those pages that have been dropped because
> > > > of poison.
> > > 
> > > So, what would a sysadmin do with that detailed information?
> > 
> > It's going to be a rare case that there are any poisoned pages on that list
> > (a large enough cluster will have some systems that have uncorrected
> > recoverable errors in SGX EPC memory).
> > 
> > Even when there are some poisoned pages, there will only be a few. Systems
> > that have thousands of pages with uncorrected memory errors will surely crash
> > because one of those errors is going to either trigger an error marked as fatal,
> > or the error won’t be recoverable by Linux because it is in kernel memory.
> > 
> > A sysadmin might add a script to run during system shutdown (or periodically
> > during run-time) to save the poison page list. Then at startup run:
> > 
> > for addr in `cat saved_sgx_poison_page_list`
> > do
> > 	echo $addr > /sys/devices/system/memory/hard_offline_page
> > done
> > 
> > to make poison persistent across reboots.
> > 
> > -Tony
> 
> Couldn't it be a blob with 8 bytes for each address?

It could be a blob. But that would require some perl/python
instead of simple shell to do the above persistence trick.

Or I could just drop the debugfs interface from this patch,
waiting until some use case for the data is fleshed out so
that it can be done in the most sensible way for that use case.

Untested updated patch below.

-Tony

From 551fbc5822e8faf93ff53f0a2b2448b0b98f1dde Mon Sep 17 00:00:00 2001
From: Tony Luck <tony.luck@intel.com>
Date: Mon, 27 Sep 2021 13:26:06 -0700
Subject: [PATCH] x86/sgx: Initial poison handling for dirty and free pages

A memory controller patrol scrubber can report poison in a page
that isn't currently being used.

Add "poison" field in the sgx_epc_page that can be set for an
sgx_epc_page. Check for it:
1) When sanitizing dirty pages
2) When freeing epc pages

Poison is a new field separated from flags to avoid having to make
all updates to flags atomic, or integrate poison state changes into
some other locking scheme to protect flags.

In both cases place the poisoned page on a list of poisoned epc pages
to make sure it will not be reallocated.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 14 +++++++++++++-
 arch/x86/kernel/cpu/sgx/sgx.h  |  3 ++-
 2 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 09fa42690ff2..653bace26100 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -43,6 +43,7 @@ static nodemask_t sgx_numa_mask;
 static struct sgx_numa_node *sgx_numa_nodes;
 
 static LIST_HEAD(sgx_dirty_page_list);
+static LIST_HEAD(sgx_poison_page_list);
 
 /*
  * Reset post-kexec EPC pages to the uninitialized state. The pages are removed
@@ -62,6 +63,12 @@ static void __sgx_sanitize_pages(struct list_head *dirty_page_list)
 
 		page = list_first_entry(dirty_page_list, struct sgx_epc_page, list);
 
+		if (page->poison) {
+			list_del(&page->list);
+			list_add(&page->list, &sgx_poison_page_list);
+			continue;
+		}
+
 		ret = __eremove(sgx_get_epc_virt_addr(page));
 		if (!ret) {
 			/*
@@ -626,7 +633,11 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 
 	spin_lock(&node->lock);
 
-	list_add_tail(&page->list, &node->free_page_list);
+	page->owner = NULL;
+	if (page->poison)
+		list_add(&page->list, &sgx_poison_page_list);
+	else
+		list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
 	page->flags = 0;
 
@@ -658,6 +669,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 		section->pages[i].section = index;
 		section->pages[i].flags = SGX_EPC_PAGE_IN_USE;
 		section->pages[i].owner = NULL;
+		section->pages[i].poison = 0;
 		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
 	}
 
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index f9202d3d6278..a990a4c9a00f 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -31,7 +31,8 @@
 
 struct sgx_epc_page {
 	unsigned int section;
-	unsigned int flags;
+	u16 flags;
+	u16 poison;
 	struct sgx_encl_page *owner;
 	struct list_head list;
 };
-- 
2.31.1

Re: [PATCH v7 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Jarkko Sakkinen]

On Tue, 2021-09-28 at 13:53 -0700, Luck, Tony wrote:
> On Tue, Sep 28, 2021 at 11:11:30PM +0300, Jarkko Sakkinen wrote:
> > On Tue, 2021-09-28 at 15:41 +0000, Luck, Tony wrote:
> > > > > Add debugfs files /sys/kernel/debug/sgx/poison_page_list so that system
> > > > > administrators get a list of those pages that have been dropped because
> > > > > of poison.
> > > > 
> > > > So, what would a sysadmin do with that detailed information?
> > > 
> > > It's going to be a rare case that there are any poisoned pages on that list
> > > (a large enough cluster will have some systems that have uncorrected
> > > recoverable errors in SGX EPC memory).
> > > 
> > > Even when there are some poisoned pages, there will only be a few. Systems
> > > that have thousands of pages with uncorrected memory errors will surely crash
> > > because one of those errors is going to either trigger an error marked as fatal,
> > > or the error won’t be recoverable by Linux because it is in kernel memory.
> > > 
> > > A sysadmin might add a script to run during system shutdown (or periodically
> > > during run-time) to save the poison page list. Then at startup run:
> > > 
> > > for addr in `cat saved_sgx_poison_page_list`
> > > do
> > > 	echo $addr > /sys/devices/system/memory/hard_offline_page
> > > done
> > > 
> > > to make poison persistent across reboots.
> > > 
> > > -Tony
> > 
> > Couldn't it be a blob with 8 bytes for each address?
> 
> It could be a blob. But that would require some perl/python
> instead of simple shell to do the above persistence trick.

The way I've understood it, a list of values breaks sysfs conventions.
There can be only single value per attribute. Even, if the blob is
interpreted as a list of integers, it is still a value, as far as sysfs
is concerned.

I'd also consider programs written with C, or perhaps Rust, when we
(ever) add any new sysfs for SGX. In my opinion, it makes sense to make
any uapi things we add accesible to as many tools as we can.

Such a trivially constructed blob is not enormously hard to parse in any
language, but at least I don't enjoy parsing list of strings in C code,
whereas loading a blob is effortless.

This kind of shows why the current sysfs conventions make sense in the
first place: they enforce to design attributes in the manner that they
are as reachable as possible. That's why I would follow the conventions
in a strict manner.

Finally, I would make a proper sysfs attribute out of this (and a separate
patch), which would be available per node.

/Jarkko

Re: [PATCH v7 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Luck, Tony]

On Thu, Sep 30, 2021 at 05:40:18PM +0300, Jarkko Sakkinen wrote:
> On Tue, 2021-09-28 at 13:53 -0700, Luck, Tony wrote:
> > > Couldn't it be a blob with 8 bytes for each address?
> > 
> > It could be a blob. But that would require some perl/python
> > instead of simple shell to do the above persistence trick.
> 
> The way I've understood it, a list of values breaks sysfs conventions.
> There can be only single value per attribute. Even, if the blob is
> interpreted as a list of integers, it is still a value, as far as sysfs
> is concerned.
> 
> I'd also consider programs written with C, or perhaps Rust, when we
> (ever) add any new sysfs for SGX. In my opinion, it makes sense to make
> any uapi things we add accesible to as many tools as we can.
> 
> Such a trivially constructed blob is not enormously hard to parse in any
> language, but at least I don't enjoy parsing list of strings in C code,
> whereas loading a blob is effortless.
> 
> This kind of shows why the current sysfs conventions make sense in the
> first place: they enforce to design attributes in the manner that they
> are as reachable as possible. That's why I would follow the conventions
> in a strict manner.
> 
> Finally, I would make a proper sysfs attribute out of this (and a separate
> patch), which would be available per node.

Those are all good points. I'm going to drop any interface from this
series (because that's above and beyond the goal of "basic machine check
support"). We can spend some time to come up with the right interface
and add that in a future series.

-Tony

[PATCH v7 4/7] x86/sgx: Add SGX infrastructure to recover from poison

[Tony Luck]

Provide a recovery function sgx_memory_failure(). If the poison was
consumed synchronously then send a SIGBUS. Note that the virtual
address of the access is not included with the SIGBUS as is the case
for poison outside of SGX enclaves. This doesn't matter as addresses
of code/data inside an enclave is of little to no use to code executing
outside the (now dead) enclave.

Poison found in a free page results in the page being moved from the
free list to the poison page list.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 77 ++++++++++++++++++++++++++++++++++
 1 file changed, 77 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index b558c9a80af4..9931fabb29eb 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -683,6 +683,83 @@ bool arch_is_platform_page(u64 paddr)
 }
 EXPORT_SYMBOL_GPL(arch_is_platform_page);
 
+static struct sgx_epc_page *sgx_paddr_to_page(u64 paddr)
+{
+	struct sgx_epc_section *section;
+
+	section = xa_load(&sgx_epc_address_space, paddr);
+	if (!section)
+		return NULL;
+
+	return &section->pages[PFN_DOWN(paddr - section->phys_addr)];
+}
+
+/*
+ * Called in process context to handle a hardware reported
+ * error in an SGX EPC page.
+ * If the MF_ACTION_REQUIRED bit is set in flags, then the
+ * context is the task that consumed the poison data. Otherwise
+ * this is called from a kernel thread unrelated to the page.
+ */
+int arch_memory_failure(unsigned long pfn, int flags)
+{
+	struct sgx_epc_page *page = sgx_paddr_to_page(pfn << PAGE_SHIFT);
+	struct sgx_epc_section *section;
+	struct sgx_numa_node *node;
+
+	/*
+	 * mm/memory-failure.c calls this routine for all errors
+	 * where there isn't a "struct page" for the address. But that
+	 * includes other address ranges besides SGX.
+	 */
+	if (!page)
+		return -ENXIO;
+
+	/*
+	 * If poison was consumed synchronously. Send a SIGBUS to
+	 * the task. Hardware has already exited the SGX enclave and
+	 * will not allow re-entry to an enclave that has a memory
+	 * error. The signal may help the task understand why the
+	 * enclave is broken.
+	 */
+	if (flags & MF_ACTION_REQUIRED)
+		force_sig(SIGBUS);
+
+	section = &sgx_epc_sections[page->section];
+	node = section->node;
+
+	spin_lock(&node->lock);
+
+	/* Already poisoned? Nothing more to do */
+	if (page->poison)
+		goto out;
+
+	page->poison = 1;
+
+	/*
+	 * If flags is zero, then the page is on a free list.
+	 * Move it to the poison page list.
+	 */
+	if (!page->flags) {
+		list_del(&page->list);
+		list_add(&page->list, &sgx_poison_page_list);
+		goto out;
+	}
+
+	/*
+	 * TBD: Add additional plumbing to enable pre-emptive
+	 * action for asynchronous poison notification. Until
+	 * then just hope that the poison:
+	 * a) is not accessed - sgx_free_epc_page() will deal with it
+	 *    when the user gives it back
+	 * b) results in a recoverable machine check rather than
+	 *    a fatal one
+	 */
+out:
+	spin_unlock(&node->lock);
+	return 0;
+}
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
-- 
2.31.1

[PATCH v7 5/7] x86/sgx: Hook arch_memory_failure() into mainline code

[Tony Luck]

Add a call inside memory_failure() to call the arch specific code
to check if the address is an SGX EPC page and handle it.

Note the SGX EPC pages do not have a "struct page" entry, so the hook
goes in at the same point as the device mapping hook.

Pull the call to acquire the mutex earlier so the SGX errors are also
protected.

Make set_mce_nospec() skip SGX pages when trying to adjust
the 1:1 map.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/include/asm/processor.h  |  8 ++++++++
 arch/x86/include/asm/set_memory.h |  4 ++++
 include/linux/mm.h                | 14 ++++++++++++++
 mm/memory-failure.c               | 19 +++++++++++++------
 4 files changed, 39 insertions(+), 6 deletions(-)

diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index 9ad2acaaae9b..4865f2860a4f 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -853,4 +853,12 @@ enum mds_mitigations {
 	MDS_MITIGATION_VMWERV,
 };
 
+#ifdef CONFIG_X86_SGX
+int arch_memory_failure(unsigned long pfn, int flags);
+#define arch_memory_failure arch_memory_failure
+
+bool arch_is_platform_page(u64 paddr);
+#define arch_is_platform_page arch_is_platform_page
+#endif
+
 #endif /* _ASM_X86_PROCESSOR_H */
diff --git a/arch/x86/include/asm/set_memory.h b/arch/x86/include/asm/set_memory.h
index 43fa081a1adb..ce8dd215f5b3 100644
--- a/arch/x86/include/asm/set_memory.h
+++ b/arch/x86/include/asm/set_memory.h
@@ -2,6 +2,7 @@
 #ifndef _ASM_X86_SET_MEMORY_H
 #define _ASM_X86_SET_MEMORY_H
 
+#include <linux/mm.h>
 #include <asm/page.h>
 #include <asm-generic/set_memory.h>
 
@@ -98,6 +99,9 @@ static inline int set_mce_nospec(unsigned long pfn, bool unmap)
 	unsigned long decoy_addr;
 	int rc;
 
+	/* SGX pages are not in the 1:1 map */
+	if (arch_is_platform_page(pfn << PAGE_SHIFT))
+		return 0;
 	/*
 	 * We would like to just call:
 	 *      set_memory_XX((unsigned long)pfn_to_kaddr(pfn), 1);
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 73a52aba448f..62b199ed5ec6 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -3284,5 +3284,19 @@ static inline int seal_check_future_write(int seals, struct vm_area_struct *vma)
 	return 0;
 }
 
+#ifndef arch_memory_failure
+static inline int arch_memory_failure(unsigned long pfn, int flags)
+{
+	return -ENXIO;
+}
+#endif
+
+#ifndef arch_is_platform_page
+static inline bool arch_is_platform_page(u64 paddr)
+{
+	return false;
+}
+#endif
+
 #endif /* __KERNEL__ */
 #endif /* _LINUX_MM_H */
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 3e6449f2102a..b1cbf9845c19 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -1632,21 +1632,28 @@ int memory_failure(unsigned long pfn, int flags)
 	if (!sysctl_memory_failure_recovery)
 		panic("Memory failure on page %lx", pfn);
 
+	mutex_lock(&mf_mutex);
+
 	p = pfn_to_online_page(pfn);
 	if (!p) {
+		res = arch_memory_failure(pfn, flags);
+		if (res == 0)
+			goto unlock_mutex;
+
 		if (pfn_valid(pfn)) {
 			pgmap = get_dev_pagemap(pfn, NULL);
-			if (pgmap)
-				return memory_failure_dev_pagemap(pfn, flags,
-								  pgmap);
+			if (pgmap) {
+				res = memory_failure_dev_pagemap(pfn, flags,
+								 pgmap);
+				goto unlock_mutex;
+			}
 		}
 		pr_err("Memory failure: %#lx: memory outside kernel control\n",
 			pfn);
-		return -ENXIO;
+		res = -ENXIO;
+		goto unlock_mutex;
 	}
 
-	mutex_lock(&mf_mutex);
-
 try_again:
 	if (PageHuge(p)) {
 		res = memory_failure_hugetlb(pfn, flags);
-- 
2.31.1

[PATCH v7 6/7] x86/sgx: Add hook to error injection address validation

[Tony Luck]

SGX reserved memory does not appear in the standard address maps.

Add hook to call into the SGX code to check if an address is located
in SGX memory.

There are other challenges in injecting errors into SGX. Update the
documentation with a sequence of operations to inject.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 .../firmware-guide/acpi/apei/einj.rst         | 19 +++++++++++++++++++
 drivers/acpi/apei/einj.c                      |  3 ++-
 2 files changed, 21 insertions(+), 1 deletion(-)

diff --git a/Documentation/firmware-guide/acpi/apei/einj.rst b/Documentation/firmware-guide/acpi/apei/einj.rst
index c042176e1707..55e2331a6438 100644
--- a/Documentation/firmware-guide/acpi/apei/einj.rst
+++ b/Documentation/firmware-guide/acpi/apei/einj.rst
@@ -181,5 +181,24 @@ You should see something like this in dmesg::
   [22715.834759] EDAC sbridge MC3: PROCESSOR 0:306e7 TIME 1422553404 SOCKET 0 APIC 0
   [22716.616173] EDAC MC3: 1 CE memory read error on CPU_SrcID#0_Channel#0_DIMM#0 (channel:0 slot:0 page:0x12345 offset:0x0 grain:32 syndrome:0x0 -  area:DRAM err_code:0001:0090 socket:0 channel_mask:1 rank:0)
 
+Special notes for injection into SGX enclaves:
+
+There may be a separate BIOS setup option to enable SGX injection.
+
+The injection process consists of setting some special memory controller
+trigger that will inject the error on the next write to the target
+address. But the h/w prevents any software outside of an SGX enclave
+from accessing enclave pages (even BIOS SMM mode).
+
+The following sequence can be used:
+  1) Determine physical address of enclave page
+  2) Use "notrigger=1" mode to inject (this will setup
+     the injection address, but will not actually inject)
+  3) Enter the enclave
+  4) Store data to the virtual address matching physical address from step 1
+  5) Execute CLFLUSH for that virtual address
+  6) Spin delay for 250ms
+  7) Read from the virtual address. This will trigger the error
+
 For more information about EINJ, please refer to ACPI specification
 version 4.0, section 17.5 and ACPI 5.0, section 18.6.
diff --git a/drivers/acpi/apei/einj.c b/drivers/acpi/apei/einj.c
index 2882450c443e..67c335baad52 100644
--- a/drivers/acpi/apei/einj.c
+++ b/drivers/acpi/apei/einj.c
@@ -544,7 +544,8 @@ static int einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
 	    ((region_intersects(base_addr, size, IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE)
 				!= REGION_INTERSECTS) &&
 	     (region_intersects(base_addr, size, IORESOURCE_MEM, IORES_DESC_PERSISTENT_MEMORY)
-				!= REGION_INTERSECTS)))
+				!= REGION_INTERSECTS) &&
+	     !arch_is_platform_page(base_addr)))
 		return -EINVAL;
 
 inject:
-- 
2.31.1

[PATCH v7 7/7] x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

[Tony Luck]

SGX EPC pages do not have a "struct page" associated with them so the
pfn_valid() sanity check fails and results in a warning message to
the console.

Add an additional check to skip the warning if the address of the error
is in an SGX EPC page.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 drivers/acpi/apei/ghes.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 0c8330ed1ffd..0c5c9acc6254 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -449,7 +449,7 @@ static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 		return false;
 
 	pfn = PHYS_PFN(physical_addr);
-	if (!pfn_valid(pfn)) {
+	if (!pfn_valid(pfn) && !arch_is_platform_page(physical_addr)) {
 		pr_warn_ratelimited(FW_WARN GHES_PFX
 		"Invalid address in generic error data: %#llx\n",
 		physical_addr);
-- 
2.31.1

[PATCH v8 0/7] Basic recovery for machine checks inside SGX

[Tony Luck]

Now version 8

Note that linux-kernel@vger.kernel.org and x86@kernel.org are still
dropped from the distribution. Time to get some internal agreement on these
changes before bothering the x86 maintainers with yet another version.

So I'm still looking for Acked-by: or Reviewed-by: on any bits of this
series that are worthy, and comments on the problems I need to fix
in the not-worthy parts.

Changes since v7

Parts 1 & 2: Added "Reviewed-by" tag from Jarkko (Thanks!!!)

Part 3: Jarkko had many good questions about the debugfs interface
	that was included to display addresses of pages on the SGX
	poison list. I don't have good answers to them all. While
	this was a useful hook while I was testing these patches
	(check that all the thousands of SGX pages that into which
	I had injected errors showed up on the list) it isn't needed
	for basic recovery. So I dropped the debugfs bits from the
	patch. We can revisit later when there is a clear use case
	for what should be provided.

Parts 4-7: Unchanged.

Tony Luck (7):
  x86/sgx: Add new sgx_epc_page flag bit to mark in-use pages
  x86/sgx: Add infrastructure to identify SGX EPC pages
  x86/sgx: Initial poison handling for dirty and free pages
  x86/sgx: Add SGX infrastructure to recover from poison
  x86/sgx: Hook arch_memory_failure() into mainline code
  x86/sgx: Add hook to error injection address validation
  x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

 .../firmware-guide/acpi/apei/einj.rst         |  19 ++++
 arch/x86/include/asm/processor.h              |   8 ++
 arch/x86/include/asm/set_memory.h             |   4 +
 arch/x86/kernel/cpu/sgx/main.c                | 104 +++++++++++++++++-
 arch/x86/kernel/cpu/sgx/sgx.h                 |   6 +-
 drivers/acpi/apei/einj.c                      |   3 +-
 drivers/acpi/apei/ghes.c                      |   2 +-
 include/linux/mm.h                            |  14 +++
 mm/memory-failure.c                           |  19 +++-
 9 files changed, 168 insertions(+), 11 deletions(-)


base-commit: 5816b3e6577eaa676ceb00a848f0fd65fe2adc29
-- 
2.31.1

[PATCH v8 1/7] x86/sgx: Add new sgx_epc_page flag bit to mark in-use pages

[Tony Luck]

SGX EPC pages go through the following life cycle:

        DIRTY ---> FREE ---> IN-USE --\
                    ^                 |
                    \-----------------/

Recovery action for poison for a DIRTY or FREE page is simple. Just
make sure never to allocate the page. IN-USE pages need some extra
handling.

Add a new flag bit SGX_EPC_PAGE_IN_USE that is set when a page
is allocated and cleared when the page is freed.

Notes:

1) These transitions are made while holding the node->lock so that
   future code that checks the flags while holding the node->lock
   can be sure that if the SGX_EPC_PAGE_IN_USE bit is set, then the
   page is on the free list.

2) Initially while the pages are on the dirty list the
   SGX_EPC_PAGE_IN_USE bit is set.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 4 +++-
 arch/x86/kernel/cpu/sgx/sgx.h  | 3 +++
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 63d3de02bbcc..d18988a46c13 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -472,6 +472,7 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
 	page = list_first_entry(&node->free_page_list, struct sgx_epc_page, list);
 	list_del_init(&page->list);
 	sgx_nr_free_pages--;
+	page->flags = SGX_EPC_PAGE_IN_USE;
 
 	spin_unlock(&node->lock);
 
@@ -626,6 +627,7 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 
 	list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
+	page->flags = 0;
 
 	spin_unlock(&node->lock);
 }
@@ -651,7 +653,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
-		section->pages[i].flags = 0;
+		section->pages[i].flags = SGX_EPC_PAGE_IN_USE;
 		section->pages[i].owner = NULL;
 		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
 	}
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index 4628acec0009..f9202d3d6278 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -26,6 +26,9 @@
 /* Pages, which are being tracked by the page reclaimer. */
 #define SGX_EPC_PAGE_RECLAIMER_TRACKED	BIT(0)
 
+/* Allocated pages */
+#define SGX_EPC_PAGE_IN_USE		BIT(1)
+
 struct sgx_epc_page {
 	unsigned int section;
 	unsigned int flags;
-- 
2.31.1

[PATCH v8 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[Tony Luck]

X86 machine check architecture reports a physical address when there
is a memory error. Handling that error requires a method to determine
whether the physical address reported is in any of the areas reserved
for EPC pages by BIOS.

SGX EPC pages do not have Linux "struct page" associated with them.

Keep track of the mapping from ranges of EPC pages to the sections
that contain them using an xarray.

Create a function arch_is_platform_page() that simply reports whether an
address is an EPC page for use elsewhere in the kernel. The ACPI error
injection code needs this function and is typically built as a module,
so export it.

Note that arch_is_platform_page() will be slower than other similar
"what type is this page" functions that can simply check bits in the
"struct page".  If there is some future performance critical user of
this function it may need to be implemented in a more efficient way.

Note also that the current implementation of xarray allocates a few
hundred kilobytes for this usage on a system with 4GB of SGX EPC memory
configured. This isn't ideal, but worth it for the code simplicity.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index d18988a46c13..09fa42690ff2 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -20,6 +20,7 @@ struct sgx_epc_section sgx_epc_sections[SGX_MAX_EPC_SECTIONS];
 static int sgx_nr_epc_sections;
 static struct task_struct *ksgxd_tsk;
 static DECLARE_WAIT_QUEUE_HEAD(ksgxd_waitq);
+static DEFINE_XARRAY(sgx_epc_address_space);
 
 /*
  * These variables are part of the state of the reclaimer, and must be accessed
@@ -650,6 +651,8 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	}
 
 	section->phys_addr = phys_addr;
+	xa_store_range(&sgx_epc_address_space, section->phys_addr,
+		       phys_addr + size - 1, section, GFP_KERNEL);
 
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
@@ -661,6 +664,12 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	return true;
 }
 
+bool arch_is_platform_page(u64 paddr)
+{
+	return !!xa_load(&sgx_epc_address_space, paddr);
+}
+EXPORT_SYMBOL_GPL(arch_is_platform_page);
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
-- 
2.31.1

[PATCH v8 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Tony Luck]

A memory controller patrol scrubber can report poison in a page
that isn't currently being used.

Add "poison" field in the sgx_epc_page that can be set for an
sgx_epc_page. Check for it:
1) When sanitizing dirty pages
2) When freeing epc pages

Poison is a new field separated from flags to avoid having to make
all updates to flags atomic, or integrate poison state changes into
some other locking scheme to protect flags.

In both cases place the poisoned page on a list of poisoned epc pages
to make sure it will not be reallocated.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 14 +++++++++++++-
 arch/x86/kernel/cpu/sgx/sgx.h  |  3 ++-
 2 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 09fa42690ff2..653bace26100 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -43,6 +43,7 @@ static nodemask_t sgx_numa_mask;
 static struct sgx_numa_node *sgx_numa_nodes;
 
 static LIST_HEAD(sgx_dirty_page_list);
+static LIST_HEAD(sgx_poison_page_list);
 
 /*
  * Reset post-kexec EPC pages to the uninitialized state. The pages are removed
@@ -62,6 +63,12 @@ static void __sgx_sanitize_pages(struct list_head *dirty_page_list)
 
 		page = list_first_entry(dirty_page_list, struct sgx_epc_page, list);
 
+		if (page->poison) {
+			list_del(&page->list);
+			list_add(&page->list, &sgx_poison_page_list);
+			continue;
+		}
+
 		ret = __eremove(sgx_get_epc_virt_addr(page));
 		if (!ret) {
 			/*
@@ -626,7 +633,11 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 
 	spin_lock(&node->lock);
 
-	list_add_tail(&page->list, &node->free_page_list);
+	page->owner = NULL;
+	if (page->poison)
+		list_add(&page->list, &sgx_poison_page_list);
+	else
+		list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
 	page->flags = 0;
 
@@ -658,6 +669,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 		section->pages[i].section = index;
 		section->pages[i].flags = SGX_EPC_PAGE_IN_USE;
 		section->pages[i].owner = NULL;
+		section->pages[i].poison = 0;
 		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
 	}
 
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index f9202d3d6278..a990a4c9a00f 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -31,7 +31,8 @@
 
 struct sgx_epc_page {
 	unsigned int section;
-	unsigned int flags;
+	u16 flags;
+	u16 poison;
 	struct sgx_encl_page *owner;
 	struct list_head list;
 };
-- 
2.31.1

Re: [PATCH v8 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Jarkko Sakkinen]

On Fri, 2021-10-01 at 09:47 -0700, Tony Luck wrote:
> A memory controller patrol scrubber can report poison in a page
> that isn't currently being used.
> 
> Add "poison" field in the sgx_epc_page that can be set for an
> sgx_epc_page. Check for it:
> 1) When sanitizing dirty pages
> 2) When freeing epc pages
> 
> Poison is a new field separated from flags to avoid having to make
> all updates to flags atomic, or integrate poison state changes into
> some other locking scheme to protect flags.
> 
> In both cases place the poisoned page on a list of poisoned epc pages
> to make sure it will not be reallocated.
> 
> Signed-off-by: Tony Luck <tony.luck@intel.com>
> ---
>  arch/x86/kernel/cpu/sgx/main.c | 14 +++++++++++++-
>  arch/x86/kernel/cpu/sgx/sgx.h  |  3 ++-
>  2 files changed, 15 insertions(+), 2 deletions(-)

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>

/Jarkko

[PATCH v8 4/7] x86/sgx: Add SGX infrastructure to recover from poison

[Tony Luck]

Provide a recovery function sgx_memory_failure(). If the poison was
consumed synchronously then send a SIGBUS. Note that the virtual
address of the access is not included with the SIGBUS as is the case
for poison outside of SGX enclaves. This doesn't matter as addresses
of code/data inside an enclave is of little to no use to code executing
outside the (now dead) enclave.

Poison found in a free page results in the page being moved from the
free list to the poison page list.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 77 ++++++++++++++++++++++++++++++++++
 1 file changed, 77 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 653bace26100..398c9749e4d1 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -682,6 +682,83 @@ bool arch_is_platform_page(u64 paddr)
 }
 EXPORT_SYMBOL_GPL(arch_is_platform_page);
 
+static struct sgx_epc_page *sgx_paddr_to_page(u64 paddr)
+{
+	struct sgx_epc_section *section;
+
+	section = xa_load(&sgx_epc_address_space, paddr);
+	if (!section)
+		return NULL;
+
+	return &section->pages[PFN_DOWN(paddr - section->phys_addr)];
+}
+
+/*
+ * Called in process context to handle a hardware reported
+ * error in an SGX EPC page.
+ * If the MF_ACTION_REQUIRED bit is set in flags, then the
+ * context is the task that consumed the poison data. Otherwise
+ * this is called from a kernel thread unrelated to the page.
+ */
+int arch_memory_failure(unsigned long pfn, int flags)
+{
+	struct sgx_epc_page *page = sgx_paddr_to_page(pfn << PAGE_SHIFT);
+	struct sgx_epc_section *section;
+	struct sgx_numa_node *node;
+
+	/*
+	 * mm/memory-failure.c calls this routine for all errors
+	 * where there isn't a "struct page" for the address. But that
+	 * includes other address ranges besides SGX.
+	 */
+	if (!page)
+		return -ENXIO;
+
+	/*
+	 * If poison was consumed synchronously. Send a SIGBUS to
+	 * the task. Hardware has already exited the SGX enclave and
+	 * will not allow re-entry to an enclave that has a memory
+	 * error. The signal may help the task understand why the
+	 * enclave is broken.
+	 */
+	if (flags & MF_ACTION_REQUIRED)
+		force_sig(SIGBUS);
+
+	section = &sgx_epc_sections[page->section];
+	node = section->node;
+
+	spin_lock(&node->lock);
+
+	/* Already poisoned? Nothing more to do */
+	if (page->poison)
+		goto out;
+
+	page->poison = 1;
+
+	/*
+	 * If flags is zero, then the page is on a free list.
+	 * Move it to the poison page list.
+	 */
+	if (!page->flags) {
+		list_del(&page->list);
+		list_add(&page->list, &sgx_poison_page_list);
+		goto out;
+	}
+
+	/*
+	 * TBD: Add additional plumbing to enable pre-emptive
+	 * action for asynchronous poison notification. Until
+	 * then just hope that the poison:
+	 * a) is not accessed - sgx_free_epc_page() will deal with it
+	 *    when the user gives it back
+	 * b) results in a recoverable machine check rather than
+	 *    a fatal one
+	 */
+out:
+	spin_unlock(&node->lock);
+	return 0;
+}
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
-- 
2.31.1

Re: [PATCH v8 4/7] x86/sgx: Add SGX infrastructure to recover from poison

[Jarkko Sakkinen]

On Fri, 2021-10-01 at 09:47 -0700, Tony Luck wrote:
> Provide a recovery function sgx_memory_failure(). If the poison was
> consumed synchronously then send a SIGBUS. Note that the virtual
> address of the access is not included with the SIGBUS as is the case
> for poison outside of SGX enclaves. This doesn't matter as addresses
> of code/data inside an enclave is of little to no use to code executing
> outside the (now dead) enclave.
> 
> Poison found in a free page results in the page being moved from the
> free list to the poison page list.
> 
> Signed-off-by: Tony Luck <tony.luck@intel.com>


Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>

/Jarkko

[PATCH v8 5/7] x86/sgx: Hook arch_memory_failure() into mainline code

[Tony Luck]

Add a call inside memory_failure() to call the arch specific code
to check if the address is an SGX EPC page and handle it.

Note the SGX EPC pages do not have a "struct page" entry, so the hook
goes in at the same point as the device mapping hook.

Pull the call to acquire the mutex earlier so the SGX errors are also
protected.

Make set_mce_nospec() skip SGX pages when trying to adjust
the 1:1 map.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/include/asm/processor.h  |  8 ++++++++
 arch/x86/include/asm/set_memory.h |  4 ++++
 include/linux/mm.h                | 14 ++++++++++++++
 mm/memory-failure.c               | 19 +++++++++++++------
 4 files changed, 39 insertions(+), 6 deletions(-)

diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index 9ad2acaaae9b..4865f2860a4f 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -853,4 +853,12 @@ enum mds_mitigations {
 	MDS_MITIGATION_VMWERV,
 };
 
+#ifdef CONFIG_X86_SGX
+int arch_memory_failure(unsigned long pfn, int flags);
+#define arch_memory_failure arch_memory_failure
+
+bool arch_is_platform_page(u64 paddr);
+#define arch_is_platform_page arch_is_platform_page
+#endif
+
 #endif /* _ASM_X86_PROCESSOR_H */
diff --git a/arch/x86/include/asm/set_memory.h b/arch/x86/include/asm/set_memory.h
index 43fa081a1adb..ce8dd215f5b3 100644
--- a/arch/x86/include/asm/set_memory.h
+++ b/arch/x86/include/asm/set_memory.h
@@ -2,6 +2,7 @@
 #ifndef _ASM_X86_SET_MEMORY_H
 #define _ASM_X86_SET_MEMORY_H
 
+#include <linux/mm.h>
 #include <asm/page.h>
 #include <asm-generic/set_memory.h>
 
@@ -98,6 +99,9 @@ static inline int set_mce_nospec(unsigned long pfn, bool unmap)
 	unsigned long decoy_addr;
 	int rc;
 
+	/* SGX pages are not in the 1:1 map */
+	if (arch_is_platform_page(pfn << PAGE_SHIFT))
+		return 0;
 	/*
 	 * We would like to just call:
 	 *      set_memory_XX((unsigned long)pfn_to_kaddr(pfn), 1);
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 73a52aba448f..62b199ed5ec6 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -3284,5 +3284,19 @@ static inline int seal_check_future_write(int seals, struct vm_area_struct *vma)
 	return 0;
 }
 
+#ifndef arch_memory_failure
+static inline int arch_memory_failure(unsigned long pfn, int flags)
+{
+	return -ENXIO;
+}
+#endif
+
+#ifndef arch_is_platform_page
+static inline bool arch_is_platform_page(u64 paddr)
+{
+	return false;
+}
+#endif
+
 #endif /* __KERNEL__ */
 #endif /* _LINUX_MM_H */
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 3e6449f2102a..b1cbf9845c19 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -1632,21 +1632,28 @@ int memory_failure(unsigned long pfn, int flags)
 	if (!sysctl_memory_failure_recovery)
 		panic("Memory failure on page %lx", pfn);
 
+	mutex_lock(&mf_mutex);
+
 	p = pfn_to_online_page(pfn);
 	if (!p) {
+		res = arch_memory_failure(pfn, flags);
+		if (res == 0)
+			goto unlock_mutex;
+
 		if (pfn_valid(pfn)) {
 			pgmap = get_dev_pagemap(pfn, NULL);
-			if (pgmap)
-				return memory_failure_dev_pagemap(pfn, flags,
-								  pgmap);
+			if (pgmap) {
+				res = memory_failure_dev_pagemap(pfn, flags,
+								 pgmap);
+				goto unlock_mutex;
+			}
 		}
 		pr_err("Memory failure: %#lx: memory outside kernel control\n",
 			pfn);
-		return -ENXIO;
+		res = -ENXIO;
+		goto unlock_mutex;
 	}
 
-	mutex_lock(&mf_mutex);
-
 try_again:
 	if (PageHuge(p)) {
 		res = memory_failure_hugetlb(pfn, flags);
-- 
2.31.1

[PATCH v8 6/7] x86/sgx: Add hook to error injection address validation

[Tony Luck]

SGX reserved memory does not appear in the standard address maps.

Add hook to call into the SGX code to check if an address is located
in SGX memory.

There are other challenges in injecting errors into SGX. Update the
documentation with a sequence of operations to inject.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 .../firmware-guide/acpi/apei/einj.rst         | 19 +++++++++++++++++++
 drivers/acpi/apei/einj.c                      |  3 ++-
 2 files changed, 21 insertions(+), 1 deletion(-)

diff --git a/Documentation/firmware-guide/acpi/apei/einj.rst b/Documentation/firmware-guide/acpi/apei/einj.rst
index c042176e1707..55e2331a6438 100644
--- a/Documentation/firmware-guide/acpi/apei/einj.rst
+++ b/Documentation/firmware-guide/acpi/apei/einj.rst
@@ -181,5 +181,24 @@ You should see something like this in dmesg::
   [22715.834759] EDAC sbridge MC3: PROCESSOR 0:306e7 TIME 1422553404 SOCKET 0 APIC 0
   [22716.616173] EDAC MC3: 1 CE memory read error on CPU_SrcID#0_Channel#0_DIMM#0 (channel:0 slot:0 page:0x12345 offset:0x0 grain:32 syndrome:0x0 -  area:DRAM err_code:0001:0090 socket:0 channel_mask:1 rank:0)
 
+Special notes for injection into SGX enclaves:
+
+There may be a separate BIOS setup option to enable SGX injection.
+
+The injection process consists of setting some special memory controller
+trigger that will inject the error on the next write to the target
+address. But the h/w prevents any software outside of an SGX enclave
+from accessing enclave pages (even BIOS SMM mode).
+
+The following sequence can be used:
+  1) Determine physical address of enclave page
+  2) Use "notrigger=1" mode to inject (this will setup
+     the injection address, but will not actually inject)
+  3) Enter the enclave
+  4) Store data to the virtual address matching physical address from step 1
+  5) Execute CLFLUSH for that virtual address
+  6) Spin delay for 250ms
+  7) Read from the virtual address. This will trigger the error
+
 For more information about EINJ, please refer to ACPI specification
 version 4.0, section 17.5 and ACPI 5.0, section 18.6.
diff --git a/drivers/acpi/apei/einj.c b/drivers/acpi/apei/einj.c
index 2882450c443e..67c335baad52 100644
--- a/drivers/acpi/apei/einj.c
+++ b/drivers/acpi/apei/einj.c
@@ -544,7 +544,8 @@ static int einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
 	    ((region_intersects(base_addr, size, IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE)
 				!= REGION_INTERSECTS) &&
 	     (region_intersects(base_addr, size, IORESOURCE_MEM, IORES_DESC_PERSISTENT_MEMORY)
-				!= REGION_INTERSECTS)))
+				!= REGION_INTERSECTS) &&
+	     !arch_is_platform_page(base_addr)))
 		return -EINVAL;
 
 inject:
-- 
2.31.1

[PATCH v8 7/7] x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

[Tony Luck]

SGX EPC pages do not have a "struct page" associated with them so the
pfn_valid() sanity check fails and results in a warning message to
the console.

Add an additional check to skip the warning if the address of the error
is in an SGX EPC page.

Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 drivers/acpi/apei/ghes.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 0c8330ed1ffd..0c5c9acc6254 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -449,7 +449,7 @@ static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 		return false;
 
 	pfn = PHYS_PFN(physical_addr);
-	if (!pfn_valid(pfn)) {
+	if (!pfn_valid(pfn) && !arch_is_platform_page(physical_addr)) {
 		pr_warn_ratelimited(FW_WARN GHES_PFX
 		"Invalid address in generic error data: %#llx\n",
 		physical_addr);
-- 
2.31.1

Re: [PATCH v8 0/7] Basic recovery for machine checks inside SGX

[Reinette Chatre]

Hi Tony,

On 10/1/2021 9:47 AM, Tony Luck wrote:
> Now version 8
> 
> Note that linux-kernel@vger.kernel.org and x86@kernel.org are still
> dropped from the distribution. Time to get some internal agreement on these
> changes before bothering the x86 maintainers with yet another version.
> 
> So I'm still looking for Acked-by: or Reviewed-by: on any bits of this
> series that are worthy, and comments on the problems I need to fix
> in the not-worthy parts.

Tested-by: Reinette Chatre <reinette.chatre@intel.com>

Details of testing:
I was curious how the signaling worked after reading some snippets that 
a vDSO does not generate a signal. To help my understanding I created 
the test below in the SGX selftests that implements the test sequence 
you document in patch 6/7 and with it I can see how the SIGBUS is delivered:

[BEGIN TEST OUTPUT]
#  RUN           enclave.mce ...
MCE test from pid 2746 on addr 0x7fc879644000
Set up error injection on virt 0x7fc879644000 and press any key to 
continue test ...

# mce: Test terminated unexpectedly by signal 7
[END TEST OUTPUT]

Below is the test I ran. It only implements steps 3 to 7 from the test 
sequence you document in patch 6/7. It does still require manual 
intervention to determine the physical address and trigger the error 
injection on the physical address. It also currently treats the SIGBUS 
as a test failure, which I did to help clearly see the signal, but it 
could be changed to TEST_F_SIGNAL to have a SIGBUS mean success. The 
test is thus not appropriate for the SGX selftests in its current form 
but is provided as informational to describe the testing I did. It 
applies on top of the recent SGX selftest changes from: 
https://lore.kernel.org/lkml/cover.1631731214.git.reinette.chatre@intel.com/

---8<---
  tools/testing/selftests/sgx/defines.h   |  7 +++++
  tools/testing/selftests/sgx/main.c      | 35 +++++++++++++++++++++
  tools/testing/selftests/sgx/test_encl.c | 41 +++++++++++++++++++++++++
  3 files changed, 83 insertions(+)

diff --git a/tools/testing/selftests/sgx/defines.h 
b/tools/testing/selftests/sgx/defines.h
index 02d775789ea7..2b471ba68e91 100644
--- a/tools/testing/selftests/sgx/defines.h
+++ b/tools/testing/selftests/sgx/defines.h
@@ -24,6 +24,7 @@ enum encl_op_type {
  	ENCL_OP_PUT_TO_ADDRESS,
  	ENCL_OP_GET_FROM_ADDRESS,
  	ENCL_OP_NOP,
+	ENCL_OP_MCE,
  	ENCL_OP_MAX,
  };

@@ -53,4 +54,10 @@ struct encl_op_get_from_addr {
  	uint64_t addr;
  };

+struct encl_op_mce {
+	struct encl_op_header header;
+	uint64_t addr;
+	uint64_t value;
+	uint64_t delay_cycles;
+};
  #endif /* DEFINES_H */
diff --git a/tools/testing/selftests/sgx/main.c 
b/tools/testing/selftests/sgx/main.c
index 79669c245f94..2979306a687a 100644
--- a/tools/testing/selftests/sgx/main.c
+++ b/tools/testing/selftests/sgx/main.c
@@ -594,4 +594,39 @@ TEST_F(enclave, pte_permissions)
  	EXPECT_EQ(self->run.exception_addr, 0);
  }

+TEST_F_TIMEOUT(enclave, mce, 600)
+{
+	struct encl_op_mce mce_op;
+	unsigned long data_start;
+
+	ASSERT_TRUE(setup_test_encl(ENCL_HEAP_SIZE_DEFAULT, &self->encl, 
_metadata));
+
+	memset(&self->run, 0, sizeof(self->run));
+	self->run.tcs = self->encl.encl_base;
+
+	data_start = self->encl.encl_base +
+		     encl_get_data_offset(&self->encl) +
+		     PAGE_SIZE;
+
+	printf("MCE test from pid %d on addr 0x%lx\n", getpid(), data_start);
+	/*
+	 * Sanity check to ensure it is possible to write to page that will
+	 * have its permissions manipulated.
+	 */
+
+	printf("Set up error injection on virt 0x%lx and press any key to 
continue test ...\n", data_start);
+	getchar();
+	mce_op.value = MAGIC;
+	mce_op.addr = data_start;
+	mce_op.delay_cycles = 600000000;
+	mce_op.header.type = ENCL_OP_MCE;
+
+	EXPECT_EQ(ENCL_CALL(&mce_op, &self->run, true), 0);
+
+	EXPECT_EEXIT(&self->run);
+	EXPECT_EQ(self->run.exception_vector, 0);
+	EXPECT_EQ(self->run.exception_error_code, 0);
+	EXPECT_EQ(self->run.exception_addr, 0);
+}
+
  TEST_HARNESS_MAIN
diff --git a/tools/testing/selftests/sgx/test_encl.c 
b/tools/testing/selftests/sgx/test_encl.c
index 4fca01cfd898..223a80529ba6 100644
--- a/tools/testing/selftests/sgx/test_encl.c
+++ b/tools/testing/selftests/sgx/test_encl.c
@@ -11,6 +11,11 @@
   */
  static uint8_t encl_buffer[8192] = { 1 };

+static inline void clflush(volatile void *__p)
+{
+	asm volatile("clflush %0" : "+m" (*(volatile char __force *)__p));
+}
+
  static void *memcpy(void *dest, const void *src, size_t n)
  {
  	size_t i;
@@ -35,6 +40,30 @@ static void do_encl_op_get_from_buf(void *op)
  	memcpy(&op2->value, &encl_buffer[0], 8);
  }

+static __always_inline unsigned long long read_tsc(void)
+{
+	unsigned long low, high;
+	asm volatile("rdtsc" : "=a" (low), "=d" (high) :: "ecx");
+	return (low | high << 32);
+}
+
+static __always_inline void rep_nop(void)
+{
+	asm volatile("rep;nop": : :"memory");
+}
+
+void delay_mce(unsigned cycles)
+{
+	unsigned long long start, now;
+	start = read_tsc();
+	for (;;) {
+		now = read_tsc();
+		if (now - start >= cycles)
+			break;
+		rep_nop();
+	}
+}
+
  static void do_encl_op_put_to_addr(void *_op)
  {
  	struct encl_op_put_to_addr *op = _op;
@@ -49,6 +78,17 @@ static void do_encl_op_get_from_addr(void *_op)
  	memcpy(&op->value, (void *)op->addr, 8);
  }

+static void do_encl_op_mce(void *_op)
+{
+	struct encl_op_mce *op = _op;
+
+	memcpy((void *)op->addr, &op->value, 8);
+	clflush((void *)op->addr);
+	delay_mce(op->delay_cycles);
+	memcpy(&op->value, (void *)op->addr, 8);
+}
+
+
  static void do_encl_op_nop(void *_op)
  {

@@ -62,6 +102,7 @@ void encl_body(void *rdi,  void *rsi)
  		do_encl_op_put_to_addr,
  		do_encl_op_get_from_addr,
  		do_encl_op_nop,
+		do_encl_op_mce,
  	};

  	struct encl_op_header *op = (struct encl_op_header *)rdi;

[PATCH v9 0/7] Basic recovery for machine checks inside SGX

[Tony Luck]

Posting latest version to a slightly wider audience.

The big picture is that SGX uses some memory pages that are walled off
from access by the OS. This means they:
1) Don't have "struct page" describing them
2) Don't appear in the kernel 1:1 map

But they are still backed by normal DDR memory, so errors can occur.

Parts 1-4 of this series handle the internal SGX bits to keep track of
these pages in an error context. They've had a fair amount of review
on the linux-sgx list (but if any of the 37 subscribers to that list
not named Jarkko or Reinette want to chime in with extra comments and
{Acked,Reviewed,Tested}-by that would be great).

Linux-mm reviewers can (if they like) skip to part 5 where two changes are
made: 1) Hook into memory_failure() in the same spot as device mapping 2)
Skip trying to change 1:1 map (since SGX pages aren't there).

The hooks have generic looking names rather than specifically saying
"sgx" at the suggestion of Dave Hansen. I'm not wedded to the names,
so better suggestions welcome.  I could also change to using some
"ARCH_HAS_PLATFORM_PAGES" config bits if that's the current fashion.

Rafael (and other ACPI list readers) can skip to parts 6 & 7 where there
are hooks into error injection and reporting to simply say "these odd
looking physical addresses are actually ok to use). I added some extra
notes to the einj.rst documentation on how to inject into SGX memory.

Tony Luck (7):
  x86/sgx: Add new sgx_epc_page flag bit to mark in-use pages
  x86/sgx: Add infrastructure to identify SGX EPC pages
  x86/sgx: Initial poison handling for dirty and free pages
  x86/sgx: Add SGX infrastructure to recover from poison
  x86/sgx: Hook arch_memory_failure() into mainline code
  x86/sgx: Add hook to error injection address validation
  x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

 .../firmware-guide/acpi/apei/einj.rst         |  19 ++++
 arch/x86/include/asm/processor.h              |   8 ++
 arch/x86/include/asm/set_memory.h             |   4 +
 arch/x86/kernel/cpu/sgx/main.c                | 104 +++++++++++++++++-
 arch/x86/kernel/cpu/sgx/sgx.h                 |   6 +-
 drivers/acpi/apei/einj.c                      |   3 +-
 drivers/acpi/apei/ghes.c                      |   2 +-
 include/linux/mm.h                            |  14 +++
 mm/memory-failure.c                           |  19 +++-
 9 files changed, 168 insertions(+), 11 deletions(-)


base-commit: 64570fbc14f8d7cb3fe3995f20e26bc25ce4b2cc
-- 
2.31.1

[PATCH v9 1/7] x86/sgx: Add new sgx_epc_page flag bit to mark in-use pages

[Tony Luck]

SGX EPC pages go through the following life cycle:

        DIRTY ---> FREE ---> IN-USE --\
                    ^                 |
                    \-----------------/

Recovery action for poison for a DIRTY or FREE page is simple. Just
make sure never to allocate the page. IN-USE pages need some extra
handling.

Add a new flag bit SGX_EPC_PAGE_IN_USE that is set when a page
is allocated and cleared when the page is freed.

Notes:

1) These transitions are made while holding the node->lock so that
   future code that checks the flags while holding the node->lock
   can be sure that if the SGX_EPC_PAGE_IN_USE bit is set, then the
   page is on the free list.

2) Initially while the pages are on the dirty list the
   SGX_EPC_PAGE_IN_USE bit is set.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 4 +++-
 arch/x86/kernel/cpu/sgx/sgx.h  | 3 +++
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 63d3de02bbcc..d18988a46c13 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -472,6 +472,7 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
 	page = list_first_entry(&node->free_page_list, struct sgx_epc_page, list);
 	list_del_init(&page->list);
 	sgx_nr_free_pages--;
+	page->flags = SGX_EPC_PAGE_IN_USE;
 
 	spin_unlock(&node->lock);
 
@@ -626,6 +627,7 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 
 	list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
+	page->flags = 0;
 
 	spin_unlock(&node->lock);
 }
@@ -651,7 +653,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
-		section->pages[i].flags = 0;
+		section->pages[i].flags = SGX_EPC_PAGE_IN_USE;
 		section->pages[i].owner = NULL;
 		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
 	}
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index 4628acec0009..f9202d3d6278 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -26,6 +26,9 @@
 /* Pages, which are being tracked by the page reclaimer. */
 #define SGX_EPC_PAGE_RECLAIMER_TRACKED	BIT(0)
 
+/* Allocated pages */
+#define SGX_EPC_PAGE_IN_USE		BIT(1)
+
 struct sgx_epc_page {
 	unsigned int section;
 	unsigned int flags;
-- 
2.31.1

Re: [PATCH v9 1/7] x86/sgx: Add new sgx_epc_page flag bit to mark in-use pages

[Sean Christopherson]

On Mon, Oct 11, 2021, Tony Luck wrote:
> SGX EPC pages go through the following life cycle:
> 
>         DIRTY ---> FREE ---> IN-USE --\
>                     ^                 |
>                     \-----------------/
> 
> Recovery action for poison for a DIRTY or FREE page is simple. Just
> make sure never to allocate the page. IN-USE pages need some extra
> handling.
> 
> Add a new flag bit SGX_EPC_PAGE_IN_USE that is set when a page
> is allocated and cleared when the page is freed.
> 
> Notes:
> 
> 1) These transitions are made while holding the node->lock so that
>    future code that checks the flags while holding the node->lock
>    can be sure that if the SGX_EPC_PAGE_IN_USE bit is set, then the
>    page is on the free list.
> 
> 2) Initially while the pages are on the dirty list the
>    SGX_EPC_PAGE_IN_USE bit is set.

This needs to state _why_ pages are marked as IN_USE from the get-go.  Ignoring
the "Notes", the whole changelog clearly states the the DIRTY state does _not_
require special handling, but then "Add SGX infrastructure to recover from poison"
goes and relies on it being set.

Alternatively, why not invert it and have SGX_EPC_PAGE_FREE?  That would have
clear semantics, the poison recovery code wouldn't have to assume that !flags
means "free", and the whole changelog becomes:

  Add a flag to explicitly track whether or not an EPC page is on a free list,
  memory failure recovery code needs to be able to detect if a poisoned page is
  free so that recovery can know if it's safe to "steal" the page.

[PATCH v9 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[Tony Luck]

X86 machine check architecture reports a physical address when there
is a memory error. Handling that error requires a method to determine
whether the physical address reported is in any of the areas reserved
for EPC pages by BIOS.

SGX EPC pages do not have Linux "struct page" associated with them.

Keep track of the mapping from ranges of EPC pages to the sections
that contain them using an xarray.

Create a function arch_is_platform_page() that simply reports whether an
address is an EPC page for use elsewhere in the kernel. The ACPI error
injection code needs this function and is typically built as a module,
so export it.

Note that arch_is_platform_page() will be slower than other similar
"what type is this page" functions that can simply check bits in the
"struct page".  If there is some future performance critical user of
this function it may need to be implemented in a more efficient way.

Note also that the current implementation of xarray allocates a few
hundred kilobytes for this usage on a system with 4GB of SGX EPC memory
configured. This isn't ideal, but worth it for the code simplicity.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index d18988a46c13..09fa42690ff2 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -20,6 +20,7 @@ struct sgx_epc_section sgx_epc_sections[SGX_MAX_EPC_SECTIONS];
 static int sgx_nr_epc_sections;
 static struct task_struct *ksgxd_tsk;
 static DECLARE_WAIT_QUEUE_HEAD(ksgxd_waitq);
+static DEFINE_XARRAY(sgx_epc_address_space);
 
 /*
  * These variables are part of the state of the reclaimer, and must be accessed
@@ -650,6 +651,8 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	}
 
 	section->phys_addr = phys_addr;
+	xa_store_range(&sgx_epc_address_space, section->phys_addr,
+		       phys_addr + size - 1, section, GFP_KERNEL);
 
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
@@ -661,6 +664,12 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	return true;
 }
 
+bool arch_is_platform_page(u64 paddr)
+{
+	return !!xa_load(&sgx_epc_address_space, paddr);
+}
+EXPORT_SYMBOL_GPL(arch_is_platform_page);
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
-- 
2.31.1

Re: [PATCH v9 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[kernel test robot]

[-- Attachment #1: Type: text/plain, Size: 2730 bytes --]

Hi Tony,

I love your patch! Yet something to improve:

[auto build test ERROR on rafael-pm/linux-next]
[also build test ERROR on hnaz-mm/master tip/x86/sgx v5.15-rc6 next-20211021]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Tony-Luck/x86-sgx-Add-new-sgx_epc_page-flag-bit-to-mark-in-use-pages/20211012-035926
base:   https://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git linux-next
config: x86_64-randconfig-a011-20211011 (attached as .config)
compiler: gcc-9 (Debian 9.3.0-22) 9.3.0
reproduce (this is a W=1 build):
        # https://github.com/0day-ci/linux/commit/9c7bd2907252bfbf4948be9855e3535319e1e9e4
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Tony-Luck/x86-sgx-Add-new-sgx_epc_page-flag-bit-to-mark-in-use-pages/20211012-035926
        git checkout 9c7bd2907252bfbf4948be9855e3535319e1e9e4
        # save the attached .config to linux build tree
        mkdir build_dir
        make W=1 O=build_dir ARCH=x86_64 SHELL=/bin/bash

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

   ld: arch/x86/kernel/cpu/sgx/main.o: in function `sgx_setup_epc_section':
>> arch/x86/kernel/cpu/sgx/main.c:654: undefined reference to `xa_store_range'


vim +654 arch/x86/kernel/cpu/sgx/main.c

   635	
   636	static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
   637						 unsigned long index,
   638						 struct sgx_epc_section *section)
   639	{
   640		unsigned long nr_pages = size >> PAGE_SHIFT;
   641		unsigned long i;
   642	
   643		section->virt_addr = memremap(phys_addr, size, MEMREMAP_WB);
   644		if (!section->virt_addr)
   645			return false;
   646	
   647		section->pages = vmalloc(nr_pages * sizeof(struct sgx_epc_page));
   648		if (!section->pages) {
   649			memunmap(section->virt_addr);
   650			return false;
   651		}
   652	
   653		section->phys_addr = phys_addr;
 > 654		xa_store_range(&sgx_epc_address_space, section->phys_addr,
   655			       phys_addr + size - 1, section, GFP_KERNEL);
   656	
   657		for (i = 0; i < nr_pages; i++) {
   658			section->pages[i].section = index;
   659			section->pages[i].flags = SGX_EPC_PAGE_IN_USE;
   660			section->pages[i].owner = NULL;
   661			list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
   662		}
   663	
   664		return true;
   665	}
   666	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 33705 bytes --]

[PATCH v9 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Tony Luck]

A memory controller patrol scrubber can report poison in a page
that isn't currently being used.

Add "poison" field in the sgx_epc_page that can be set for an
sgx_epc_page. Check for it:
1) When sanitizing dirty pages
2) When freeing epc pages

Poison is a new field separated from flags to avoid having to make
all updates to flags atomic, or integrate poison state changes into
some other locking scheme to protect flags.

In both cases place the poisoned page on a list of poisoned epc pages
to make sure it will not be reallocated.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 14 +++++++++++++-
 arch/x86/kernel/cpu/sgx/sgx.h  |  3 ++-
 2 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 09fa42690ff2..653bace26100 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -43,6 +43,7 @@ static nodemask_t sgx_numa_mask;
 static struct sgx_numa_node *sgx_numa_nodes;
 
 static LIST_HEAD(sgx_dirty_page_list);
+static LIST_HEAD(sgx_poison_page_list);
 
 /*
  * Reset post-kexec EPC pages to the uninitialized state. The pages are removed
@@ -62,6 +63,12 @@ static void __sgx_sanitize_pages(struct list_head *dirty_page_list)
 
 		page = list_first_entry(dirty_page_list, struct sgx_epc_page, list);
 
+		if (page->poison) {
+			list_del(&page->list);
+			list_add(&page->list, &sgx_poison_page_list);
+			continue;
+		}
+
 		ret = __eremove(sgx_get_epc_virt_addr(page));
 		if (!ret) {
 			/*
@@ -626,7 +633,11 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 
 	spin_lock(&node->lock);
 
-	list_add_tail(&page->list, &node->free_page_list);
+	page->owner = NULL;
+	if (page->poison)
+		list_add(&page->list, &sgx_poison_page_list);
+	else
+		list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
 	page->flags = 0;
 
@@ -658,6 +669,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 		section->pages[i].section = index;
 		section->pages[i].flags = SGX_EPC_PAGE_IN_USE;
 		section->pages[i].owner = NULL;
+		section->pages[i].poison = 0;
 		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
 	}
 
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index f9202d3d6278..a990a4c9a00f 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -31,7 +31,8 @@
 
 struct sgx_epc_page {
 	unsigned int section;
-	unsigned int flags;
+	u16 flags;
+	u16 poison;
 	struct sgx_encl_page *owner;
 	struct list_head list;
 };
-- 
2.31.1

Re: [PATCH v9 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Sean Christopherson]

On Mon, Oct 11, 2021, Tony Luck wrote:
> A memory controller patrol scrubber can report poison in a page
> that isn't currently being used.
> 
> Add "poison" field in the sgx_epc_page that can be set for an
> sgx_epc_page. Check for it:
> 1) When sanitizing dirty pages
> 2) When freeing epc pages
> 
> Poison is a new field separated from flags to avoid having to make
> all updates to flags atomic, or integrate poison state changes into
> some other locking scheme to protect flags.

Explain why atomic would be needed.  I lived in this code for a few years and
still had to look at the source to remember that the reclaimer can set flags
without taking node->lock.

> In both cases place the poisoned page on a list of poisoned epc pages
> to make sure it will not be reallocated.
> 
> Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
> Tested-by: Reinette Chatre <reinette.chatre@intel.com>
> Signed-off-by: Tony Luck <tony.luck@intel.com>
> ---
>  arch/x86/kernel/cpu/sgx/main.c | 14 +++++++++++++-
>  arch/x86/kernel/cpu/sgx/sgx.h  |  3 ++-
>  2 files changed, 15 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
> index 09fa42690ff2..653bace26100 100644
> --- a/arch/x86/kernel/cpu/sgx/main.c
> +++ b/arch/x86/kernel/cpu/sgx/main.c
> @@ -43,6 +43,7 @@ static nodemask_t sgx_numa_mask;
>  static struct sgx_numa_node *sgx_numa_nodes;
>  
>  static LIST_HEAD(sgx_dirty_page_list);
> +static LIST_HEAD(sgx_poison_page_list);
>  
>  /*
>   * Reset post-kexec EPC pages to the uninitialized state. The pages are removed
> @@ -62,6 +63,12 @@ static void __sgx_sanitize_pages(struct list_head *dirty_page_list)
>  
>  		page = list_first_entry(dirty_page_list, struct sgx_epc_page, list);
>  
> +		if (page->poison) {

Does this need READ_ONCE (and WRITE_ONCE in the writer) to prevent reloading
page->poison since the sanitizer doesn't hold node->lock, i.e. page->poison can
be set any time?  Honest question, I'm terrible with memory ordering rules...

> +			list_del(&page->list);
> +			list_add(&page->list, &sgx_poison_page_list);

list_move()

> +			continue;
> +		}
> +
>  		ret = __eremove(sgx_get_epc_virt_addr(page));
>  		if (!ret) {
>  			/*
> @@ -626,7 +633,11 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
>  
>  	spin_lock(&node->lock);
>  
> -	list_add_tail(&page->list, &node->free_page_list);
> +	page->owner = NULL;
> +	if (page->poison)
> +		list_add(&page->list, &sgx_poison_page_list);

sgx_poison_page_list is a global list, whereas node->lock is, well, per node.
On a system with multiple EPCs, this could corrupt sgx_poison_page_list if
multiple poisoned pages from different nodes are freed simultaneously.

> +	else
> +		list_add_tail(&page->list, &node->free_page_list);
>  	sgx_nr_free_pages++;
>  	page->flags = 0;
>  
> @@ -658,6 +669,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
>  		section->pages[i].section = index;
>  		section->pages[i].flags = SGX_EPC_PAGE_IN_USE;
>  		section->pages[i].owner = NULL;
> +		section->pages[i].poison = 0;
>  		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
>  	}
>  
> diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
> index f9202d3d6278..a990a4c9a00f 100644
> --- a/arch/x86/kernel/cpu/sgx/sgx.h
> +++ b/arch/x86/kernel/cpu/sgx/sgx.h
> @@ -31,7 +31,8 @@
>  
>  struct sgx_epc_page {
>  	unsigned int section;
> -	unsigned int flags;
> +	u16 flags;
> +	u16 poison;
>  	struct sgx_encl_page *owner;
>  	struct list_head list;
>  };
> 
> -- 
> 2.31.1
> 

Re: [PATCH v9 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Luck, Tony]

On Fri, Oct 15, 2021 at 11:07:48PM +0000, Sean Christopherson wrote:
> On Mon, Oct 11, 2021, Tony Luck wrote:
> > A memory controller patrol scrubber can report poison in a page
> > that isn't currently being used.
> > 
> > Add "poison" field in the sgx_epc_page that can be set for an
> > sgx_epc_page. Check for it:
> > 1) When sanitizing dirty pages
> > 2) When freeing epc pages
> > 
> > Poison is a new field separated from flags to avoid having to make
> > all updates to flags atomic, or integrate poison state changes into
> > some other locking scheme to protect flags.
> 
> Explain why atomic would be needed.  I lived in this code for a few years and
> still had to look at the source to remember that the reclaimer can set flags
> without taking node->lock.

Will add explanation.

> 
> > In both cases place the poisoned page on a list of poisoned epc pages
> > to make sure it will not be reallocated.
> > 
> > Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
> > Tested-by: Reinette Chatre <reinette.chatre@intel.com>
> > Signed-off-by: Tony Luck <tony.luck@intel.com>
> > ---
> >  arch/x86/kernel/cpu/sgx/main.c | 14 +++++++++++++-
> >  arch/x86/kernel/cpu/sgx/sgx.h  |  3 ++-
> >  2 files changed, 15 insertions(+), 2 deletions(-)
> > 
> > diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
> > index 09fa42690ff2..653bace26100 100644
> > --- a/arch/x86/kernel/cpu/sgx/main.c
> > +++ b/arch/x86/kernel/cpu/sgx/main.c
> > @@ -43,6 +43,7 @@ static nodemask_t sgx_numa_mask;
> >  static struct sgx_numa_node *sgx_numa_nodes;
> >  
> >  static LIST_HEAD(sgx_dirty_page_list);
> > +static LIST_HEAD(sgx_poison_page_list);
> >  
> >  /*
> >   * Reset post-kexec EPC pages to the uninitialized state. The pages are removed
> > @@ -62,6 +63,12 @@ static void __sgx_sanitize_pages(struct list_head *dirty_page_list)
> >  
> >  		page = list_first_entry(dirty_page_list, struct sgx_epc_page, list);
> >  
> > +		if (page->poison) {
> 
> Does this need READ_ONCE (and WRITE_ONCE in the writer) to prevent reloading
> page->poison since the sanitizer doesn't hold node->lock, i.e. page->poison can
> be set any time?  Honest question, I'm terrible with memory ordering rules...
> 

I think it's safe. I set page->poison in arch_memory_failure() while
holding node->lock in kthread context.  So not "at any time".

This particular read is done without holding the lock ... and is thus
racy. But there are a zillion other races early in boot before the EPC
pages get sanitized and moved to the free list. E.g. if an error is
reported before they are added to the sgx_epc_address_space xarray,
then all this code will just ignore the error as "not in Linux
controlled memory".

-Tony

[PATCH v9 4/7] x86/sgx: Add SGX infrastructure to recover from poison

[Tony Luck]

Provide a recovery function sgx_memory_failure(). If the poison was
consumed synchronously then send a SIGBUS. Note that the virtual
address of the access is not included with the SIGBUS as is the case
for poison outside of SGX enclaves. This doesn't matter as addresses
of code/data inside an enclave is of little to no use to code executing
outside the (now dead) enclave.

Poison found in a free page results in the page being moved from the
free list to the poison page list.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 77 ++++++++++++++++++++++++++++++++++
 1 file changed, 77 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 653bace26100..398c9749e4d1 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -682,6 +682,83 @@ bool arch_is_platform_page(u64 paddr)
 }
 EXPORT_SYMBOL_GPL(arch_is_platform_page);
 
+static struct sgx_epc_page *sgx_paddr_to_page(u64 paddr)
+{
+	struct sgx_epc_section *section;
+
+	section = xa_load(&sgx_epc_address_space, paddr);
+	if (!section)
+		return NULL;
+
+	return &section->pages[PFN_DOWN(paddr - section->phys_addr)];
+}
+
+/*
+ * Called in process context to handle a hardware reported
+ * error in an SGX EPC page.
+ * If the MF_ACTION_REQUIRED bit is set in flags, then the
+ * context is the task that consumed the poison data. Otherwise
+ * this is called from a kernel thread unrelated to the page.
+ */
+int arch_memory_failure(unsigned long pfn, int flags)
+{
+	struct sgx_epc_page *page = sgx_paddr_to_page(pfn << PAGE_SHIFT);
+	struct sgx_epc_section *section;
+	struct sgx_numa_node *node;
+
+	/*
+	 * mm/memory-failure.c calls this routine for all errors
+	 * where there isn't a "struct page" for the address. But that
+	 * includes other address ranges besides SGX.
+	 */
+	if (!page)
+		return -ENXIO;
+
+	/*
+	 * If poison was consumed synchronously. Send a SIGBUS to
+	 * the task. Hardware has already exited the SGX enclave and
+	 * will not allow re-entry to an enclave that has a memory
+	 * error. The signal may help the task understand why the
+	 * enclave is broken.
+	 */
+	if (flags & MF_ACTION_REQUIRED)
+		force_sig(SIGBUS);
+
+	section = &sgx_epc_sections[page->section];
+	node = section->node;
+
+	spin_lock(&node->lock);
+
+	/* Already poisoned? Nothing more to do */
+	if (page->poison)
+		goto out;
+
+	page->poison = 1;
+
+	/*
+	 * If flags is zero, then the page is on a free list.
+	 * Move it to the poison page list.
+	 */
+	if (!page->flags) {
+		list_del(&page->list);
+		list_add(&page->list, &sgx_poison_page_list);
+		goto out;
+	}
+
+	/*
+	 * TBD: Add additional plumbing to enable pre-emptive
+	 * action for asynchronous poison notification. Until
+	 * then just hope that the poison:
+	 * a) is not accessed - sgx_free_epc_page() will deal with it
+	 *    when the user gives it back
+	 * b) results in a recoverable machine check rather than
+	 *    a fatal one
+	 */
+out:
+	spin_unlock(&node->lock);
+	return 0;
+}
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
-- 
2.31.1

Re: [PATCH v9 4/7] x86/sgx: Add SGX infrastructure to recover from poison

[Sean Christopherson]

On Mon, Oct 11, 2021, Tony Luck wrote:
> +	section = &sgx_epc_sections[page->section];
> +	node = section->node;
> +
> +	spin_lock(&node->lock);
> +
> +	/* Already poisoned? Nothing more to do */
> +	if (page->poison)
> +		goto out;
> +
> +	page->poison = 1;
> +
> +	/*
> +	 * If flags is zero, then the page is on a free list.
> +	 * Move it to the poison page list.
> +	 */
> +	if (!page->flags) {

If the flag is inverted, this becomes

	if (page->flags & SGX_EPC_PAGE_FREE) {

> +		list_del(&page->list);
> +		list_add(&page->list, &sgx_poison_page_list);

list_move(), and needs the same protection for sgx_poison_page_list.

> +		goto out;
> +	}
> +
> +	/*
> +	 * TBD: Add additional plumbing to enable pre-emptive
> +	 * action for asynchronous poison notification. Until
> +	 * then just hope that the poison:
> +	 * a) is not accessed - sgx_free_epc_page() will deal with it
> +	 *    when the user gives it back
> +	 * b) results in a recoverable machine check rather than
> +	 *    a fatal one
> +	 */
> +out:
> +	spin_unlock(&node->lock);
> +	return 0;
> +}
> +
>  /**
>   * A section metric is concatenated in a way that @low bits 12-31 define the
>   * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
> 
> -- 
> 2.31.1
> 

Re: [PATCH v9 4/7] x86/sgx: Add SGX infrastructure to recover from poison

[Luck, Tony]

On Fri, Oct 15, 2021 at 11:10:32PM +0000, Sean Christopherson wrote:
> On Mon, Oct 11, 2021, Tony Luck wrote:
> > +	section = &sgx_epc_sections[page->section];
> > +	node = section->node;
> > +
> > +	spin_lock(&node->lock);
> > +
> > +	/* Already poisoned? Nothing more to do */
> > +	if (page->poison)
> > +		goto out;
> > +
> > +	page->poison = 1;
> > +
> > +	/*
> > +	 * If flags is zero, then the page is on a free list.
> > +	 * Move it to the poison page list.
> > +	 */
> > +	if (!page->flags) {
> 
> If the flag is inverted, this becomes
> 
> 	if (page->flags & SGX_EPC_PAGE_FREE) {

I like the inversion. I'll switch to SGX_EPC_PAGE_FREE

> 
> > +		list_del(&page->list);
> > +		list_add(&page->list, &sgx_poison_page_list);
> 
> list_move(), and needs the same protection for sgx_poison_page_list.

Didn't know list_move() existed. Will change all the lis_del+list_add
into list_move.

Also change the sgx_poison_page_list from global to per-node. Then
the adds will be safe (accessed while holding the node->lock).


Thanks for the review.

-Tony

[PATCH v9 5/7] x86/sgx: Hook arch_memory_failure() into mainline code

[Tony Luck]

Add a call inside memory_failure() to call the arch specific code
to check if the address is an SGX EPC page and handle it.

Note the SGX EPC pages do not have a "struct page" entry, so the hook
goes in at the same point as the device mapping hook.

Pull the call to acquire the mutex earlier so the SGX errors are also
protected.

Make set_mce_nospec() skip SGX pages when trying to adjust
the 1:1 map.

Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/include/asm/processor.h  |  8 ++++++++
 arch/x86/include/asm/set_memory.h |  4 ++++
 include/linux/mm.h                | 14 ++++++++++++++
 mm/memory-failure.c               | 19 +++++++++++++------
 4 files changed, 39 insertions(+), 6 deletions(-)

diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index 9ad2acaaae9b..4865f2860a4f 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -853,4 +853,12 @@ enum mds_mitigations {
 	MDS_MITIGATION_VMWERV,
 };
 
+#ifdef CONFIG_X86_SGX
+int arch_memory_failure(unsigned long pfn, int flags);
+#define arch_memory_failure arch_memory_failure
+
+bool arch_is_platform_page(u64 paddr);
+#define arch_is_platform_page arch_is_platform_page
+#endif
+
 #endif /* _ASM_X86_PROCESSOR_H */
diff --git a/arch/x86/include/asm/set_memory.h b/arch/x86/include/asm/set_memory.h
index 43fa081a1adb..ce8dd215f5b3 100644
--- a/arch/x86/include/asm/set_memory.h
+++ b/arch/x86/include/asm/set_memory.h
@@ -2,6 +2,7 @@
 #ifndef _ASM_X86_SET_MEMORY_H
 #define _ASM_X86_SET_MEMORY_H
 
+#include <linux/mm.h>
 #include <asm/page.h>
 #include <asm-generic/set_memory.h>
 
@@ -98,6 +99,9 @@ static inline int set_mce_nospec(unsigned long pfn, bool unmap)
 	unsigned long decoy_addr;
 	int rc;
 
+	/* SGX pages are not in the 1:1 map */
+	if (arch_is_platform_page(pfn << PAGE_SHIFT))
+		return 0;
 	/*
 	 * We would like to just call:
 	 *      set_memory_XX((unsigned long)pfn_to_kaddr(pfn), 1);
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 73a52aba448f..62b199ed5ec6 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -3284,5 +3284,19 @@ static inline int seal_check_future_write(int seals, struct vm_area_struct *vma)
 	return 0;
 }
 
+#ifndef arch_memory_failure
+static inline int arch_memory_failure(unsigned long pfn, int flags)
+{
+	return -ENXIO;
+}
+#endif
+
+#ifndef arch_is_platform_page
+static inline bool arch_is_platform_page(u64 paddr)
+{
+	return false;
+}
+#endif
+
 #endif /* __KERNEL__ */
 #endif /* _LINUX_MM_H */
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 3e6449f2102a..b1cbf9845c19 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -1632,21 +1632,28 @@ int memory_failure(unsigned long pfn, int flags)
 	if (!sysctl_memory_failure_recovery)
 		panic("Memory failure on page %lx", pfn);
 
+	mutex_lock(&mf_mutex);
+
 	p = pfn_to_online_page(pfn);
 	if (!p) {
+		res = arch_memory_failure(pfn, flags);
+		if (res == 0)
+			goto unlock_mutex;
+
 		if (pfn_valid(pfn)) {
 			pgmap = get_dev_pagemap(pfn, NULL);
-			if (pgmap)
-				return memory_failure_dev_pagemap(pfn, flags,
-								  pgmap);
+			if (pgmap) {
+				res = memory_failure_dev_pagemap(pfn, flags,
+								 pgmap);
+				goto unlock_mutex;
+			}
 		}
 		pr_err("Memory failure: %#lx: memory outside kernel control\n",
 			pfn);
-		return -ENXIO;
+		res = -ENXIO;
+		goto unlock_mutex;
 	}
 
-	mutex_lock(&mf_mutex);
-
 try_again:
 	if (PageHuge(p)) {
 		res = memory_failure_hugetlb(pfn, flags);
-- 
2.31.1

Re: [PATCH v9 5/7] x86/sgx: Hook arch_memory_failure() into mainline code

[Jarkko Sakkinen]

On Mon, 2021-10-11 at 11:59 -0700, Tony Luck wrote:
> Add a call inside memory_failure() to call the arch specific code
> to check if the address is an SGX EPC page and handle it.
> 
> Note the SGX EPC pages do not have a "struct page" entry, so the hook
> goes in at the same point as the device mapping hook.
> 
> Pull the call to acquire the mutex earlier so the SGX errors are also
> protected.
> 
> Make set_mce_nospec() skip SGX pages when trying to adjust
> the 1:1 map.
> 
> Tested-by: Reinette Chatre <reinette.chatre@intel.com>
> Signed-off-by: Tony Luck <tony.luck@intel.com>
> ---
>  arch/x86/include/asm/processor.h  |  8 ++++++++
>  arch/x86/include/asm/set_memory.h |  4 ++++
>  include/linux/mm.h                | 14 ++++++++++++++
>  mm/memory-failure.c               | 19 +++++++++++++------
>  4 files changed, 39 insertions(+), 6 deletions(-)
> 
> diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
> index 9ad2acaaae9b..4865f2860a4f 100644
> --- a/arch/x86/include/asm/processor.h
> +++ b/arch/x86/include/asm/processor.h
> @@ -853,4 +853,12 @@ enum mds_mitigations {
>         MDS_MITIGATION_VMWERV,
>  };
>  
> +#ifdef CONFIG_X86_SGX
> +int arch_memory_failure(unsigned long pfn, int flags);
> +#define arch_memory_failure arch_memory_failure
> +
> +bool arch_is_platform_page(u64 paddr);
> +#define arch_is_platform_page arch_is_platform_page
> +#endif
> +
>  #endif /* _ASM_X86_PROCESSOR_H */
> diff --git a/arch/x86/include/asm/set_memory.h b/arch/x86/include/asm/set_memory.h
> index 43fa081a1adb..ce8dd215f5b3 100644
> --- a/arch/x86/include/asm/set_memory.h
> +++ b/arch/x86/include/asm/set_memory.h
> @@ -2,6 +2,7 @@
>  #ifndef _ASM_X86_SET_MEMORY_H
>  #define _ASM_X86_SET_MEMORY_H
>  
> +#include <linux/mm.h>
>  #include <asm/page.h>
>  #include <asm-generic/set_memory.h>
>  
> @@ -98,6 +99,9 @@ static inline int set_mce_nospec(unsigned long pfn, bool unmap)
>         unsigned long decoy_addr;
>         int rc;
>  
> +       /* SGX pages are not in the 1:1 map */
> +       if (arch_is_platform_page(pfn << PAGE_SHIFT))
> +               return 0;
>         /*
>          * We would like to just call:
>          *      set_memory_XX((unsigned long)pfn_to_kaddr(pfn), 1);
> diff --git a/include/linux/mm.h b/include/linux/mm.h
> index 73a52aba448f..62b199ed5ec6 100644
> --- a/include/linux/mm.h
> +++ b/include/linux/mm.h
> @@ -3284,5 +3284,19 @@ static inline int seal_check_future_write(int seals, struct vm_area_struct *vma)
>         return 0;
>  }
>  
> +#ifndef arch_memory_failure
> +static inline int arch_memory_failure(unsigned long pfn, int flags)
> +{
> +       return -ENXIO;
> +}
> +#endif
> +
> +#ifndef arch_is_platform_page
> +static inline bool arch_is_platform_page(u64 paddr)
> +{
> +       return false;
> +}
> +#endif
> +
>  #endif /* __KERNEL__ */
>  #endif /* _LINUX_MM_H */
> diff --git a/mm/memory-failure.c b/mm/memory-failure.c
> index 3e6449f2102a..b1cbf9845c19 100644
> --- a/mm/memory-failure.c
> +++ b/mm/memory-failure.c
> @@ -1632,21 +1632,28 @@ int memory_failure(unsigned long pfn, int flags)
>         if (!sysctl_memory_failure_recovery)
>                 panic("Memory failure on page %lx", pfn);
>  
> +       mutex_lock(&mf_mutex);
> +
>         p = pfn_to_online_page(pfn);
>         if (!p) {
> +               res = arch_memory_failure(pfn, flags);
> +               if (res == 0)
> +                       goto unlock_mutex;
> +
>                 if (pfn_valid(pfn)) {
>                         pgmap = get_dev_pagemap(pfn, NULL);
> -                       if (pgmap)
> -                               return memory_failure_dev_pagemap(pfn, flags,
> -                                                                 pgmap);
> +                       if (pgmap) {
> +                               res = memory_failure_dev_pagemap(pfn, flags,
> +                                                                pgmap);
> +                               goto unlock_mutex;
> +                       }
>                 }
>                 pr_err("Memory failure: %#lx: memory outside kernel control\n",
>                         pfn);
> -               return -ENXIO;
> +               res = -ENXIO;
> +               goto unlock_mutex;
>         }
>  
> -       mutex_lock(&mf_mutex);
> -
>  try_again:
>         if (PageHuge(p)) {
>                 res = memory_failure_hugetlb(pfn, flags);

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>

/Jarkko

[PATCH v9 6/7] x86/sgx: Add hook to error injection address validation

[Tony Luck]

SGX reserved memory does not appear in the standard address maps.

Add hook to call into the SGX code to check if an address is located
in SGX memory.

There are other challenges in injecting errors into SGX. Update the
documentation with a sequence of operations to inject.

Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 .../firmware-guide/acpi/apei/einj.rst         | 19 +++++++++++++++++++
 drivers/acpi/apei/einj.c                      |  3 ++-
 2 files changed, 21 insertions(+), 1 deletion(-)

diff --git a/Documentation/firmware-guide/acpi/apei/einj.rst b/Documentation/firmware-guide/acpi/apei/einj.rst
index c042176e1707..55e2331a6438 100644
--- a/Documentation/firmware-guide/acpi/apei/einj.rst
+++ b/Documentation/firmware-guide/acpi/apei/einj.rst
@@ -181,5 +181,24 @@ You should see something like this in dmesg::
   [22715.834759] EDAC sbridge MC3: PROCESSOR 0:306e7 TIME 1422553404 SOCKET 0 APIC 0
   [22716.616173] EDAC MC3: 1 CE memory read error on CPU_SrcID#0_Channel#0_DIMM#0 (channel:0 slot:0 page:0x12345 offset:0x0 grain:32 syndrome:0x0 -  area:DRAM err_code:0001:0090 socket:0 channel_mask:1 rank:0)
 
+Special notes for injection into SGX enclaves:
+
+There may be a separate BIOS setup option to enable SGX injection.
+
+The injection process consists of setting some special memory controller
+trigger that will inject the error on the next write to the target
+address. But the h/w prevents any software outside of an SGX enclave
+from accessing enclave pages (even BIOS SMM mode).
+
+The following sequence can be used:
+  1) Determine physical address of enclave page
+  2) Use "notrigger=1" mode to inject (this will setup
+     the injection address, but will not actually inject)
+  3) Enter the enclave
+  4) Store data to the virtual address matching physical address from step 1
+  5) Execute CLFLUSH for that virtual address
+  6) Spin delay for 250ms
+  7) Read from the virtual address. This will trigger the error
+
 For more information about EINJ, please refer to ACPI specification
 version 4.0, section 17.5 and ACPI 5.0, section 18.6.
diff --git a/drivers/acpi/apei/einj.c b/drivers/acpi/apei/einj.c
index 2882450c443e..67c335baad52 100644
--- a/drivers/acpi/apei/einj.c
+++ b/drivers/acpi/apei/einj.c
@@ -544,7 +544,8 @@ static int einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
 	    ((region_intersects(base_addr, size, IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE)
 				!= REGION_INTERSECTS) &&
 	     (region_intersects(base_addr, size, IORESOURCE_MEM, IORES_DESC_PERSISTENT_MEMORY)
-				!= REGION_INTERSECTS)))
+				!= REGION_INTERSECTS) &&
+	     !arch_is_platform_page(base_addr)))
 		return -EINVAL;
 
 inject:
-- 
2.31.1

Re: [PATCH v9 6/7] x86/sgx: Add hook to error injection address validation

[Jarkko Sakkinen]

On Mon, 2021-10-11 at 11:59 -0700, Tony Luck wrote:
> SGX reserved memory does not appear in the standard address maps.
> 
> Add hook to call into the SGX code to check if an address is located
> in SGX memory.
> 
> There are other challenges in injecting errors into SGX. Update the
> documentation with a sequence of operations to inject.
> 
> Tested-by: Reinette Chatre <reinette.chatre@intel.com>
> Signed-off-by: Tony Luck <tony.luck@intel.com>
> ---
>  .../firmware-guide/acpi/apei/einj.rst         | 19 +++++++++++++++++++
>  drivers/acpi/apei/einj.c                      |  3 ++-
>  2 files changed, 21 insertions(+), 1 deletion(-)
> 
> diff --git a/Documentation/firmware-guide/acpi/apei/einj.rst b/Documentation/firmware-guide/acpi/apei/einj.rst
> index c042176e1707..55e2331a6438 100644
> --- a/Documentation/firmware-guide/acpi/apei/einj.rst
> +++ b/Documentation/firmware-guide/acpi/apei/einj.rst
> @@ -181,5 +181,24 @@ You should see something like this in dmesg::
>    [22715.834759] EDAC sbridge MC3: PROCESSOR 0:306e7 TIME 1422553404 SOCKET 0 APIC 0
>    [22716.616173] EDAC MC3: 1 CE memory read error on CPU_SrcID#0_Channel#0_DIMM#0 (channel:0 slot:0 page:0x12345 offset:0x0 grain:32 syndrome:0x0 -  area:DRAM err_code:0001:0090 socket:0
> channel_mask:1 rank:0)
>  
> +Special notes for injection into SGX enclaves:
> +
> +There may be a separate BIOS setup option to enable SGX injection.
> +
> +The injection process consists of setting some special memory controller
> +trigger that will inject the error on the next write to the target
> +address. But the h/w prevents any software outside of an SGX enclave
> +from accessing enclave pages (even BIOS SMM mode).
> +
> +The following sequence can be used:
> +  1) Determine physical address of enclave page
> +  2) Use "notrigger=1" mode to inject (this will setup
> +     the injection address, but will not actually inject)
> +  3) Enter the enclave
> +  4) Store data to the virtual address matching physical address from step 1
> +  5) Execute CLFLUSH for that virtual address
> +  6) Spin delay for 250ms
> +  7) Read from the virtual address. This will trigger the error
> +
>  For more information about EINJ, please refer to ACPI specification
>  version 4.0, section 17.5 and ACPI 5.0, section 18.6.
> diff --git a/drivers/acpi/apei/einj.c b/drivers/acpi/apei/einj.c
> index 2882450c443e..67c335baad52 100644
> --- a/drivers/acpi/apei/einj.c
> +++ b/drivers/acpi/apei/einj.c
> @@ -544,7 +544,8 @@ static int einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
>             ((region_intersects(base_addr, size, IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE)
>                                 != REGION_INTERSECTS) &&
>              (region_intersects(base_addr, size, IORESOURCE_MEM, IORES_DESC_PERSISTENT_MEMORY)
> -                               != REGION_INTERSECTS)))
> +                               != REGION_INTERSECTS) &&
> +            !arch_is_platform_page(base_addr)))
>                 return -EINVAL;
>  
>  inject:

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>

/Jarkko

[PATCH v9 7/7] x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

[Tony Luck]

SGX EPC pages do not have a "struct page" associated with them so the
pfn_valid() sanity check fails and results in a warning message to
the console.

Add an additional check to skip the warning if the address of the error
is in an SGX EPC page.

Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 drivers/acpi/apei/ghes.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 0c8330ed1ffd..0c5c9acc6254 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -449,7 +449,7 @@ static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 		return false;
 
 	pfn = PHYS_PFN(physical_addr);
-	if (!pfn_valid(pfn)) {
+	if (!pfn_valid(pfn) && !arch_is_platform_page(physical_addr)) {
 		pr_warn_ratelimited(FW_WARN GHES_PFX
 		"Invalid address in generic error data: %#llx\n",
 		physical_addr);
-- 
2.31.1

Re: [PATCH v9 7/7] x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

[Jarkko Sakkinen]

On Mon, 2021-10-11 at 11:59 -0700, Tony Luck wrote:
> SGX EPC pages do not have a "struct page" associated with them so the
> pfn_valid() sanity check fails and results in a warning message to
> the console.
> 
> Add an additional check to skip the warning if the address of the error
> is in an SGX EPC page.
> 
> Tested-by: Reinette Chatre <reinette.chatre@intel.com>
> Signed-off-by: Tony Luck <tony.luck@intel.com>
> ---
>  drivers/acpi/apei/ghes.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
> index 0c8330ed1ffd..0c5c9acc6254 100644
> --- a/drivers/acpi/apei/ghes.c
> +++ b/drivers/acpi/apei/ghes.c
> @@ -449,7 +449,7 @@ static bool ghes_do_memory_failure(u64 physical_addr, int flags)
>                 return false;
>  
>         pfn = PHYS_PFN(physical_addr);
> -       if (!pfn_valid(pfn)) {
> +       if (!pfn_valid(pfn) && !arch_is_platform_page(physical_addr)) {
>                 pr_warn_ratelimited(FW_WARN GHES_PFX
>                 "Invalid address in generic error data: %#llx\n",
>                 physical_addr);

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>

/Jarkko

Re: [PATCH v9 0/7] Basic recovery for machine checks inside SGX

[Jarkko Sakkinen]

On Mon, 2021-10-11 at 11:59 -0700, Tony Luck wrote:
> Posting latest version to a slightly wider audience.
> 
> The big picture is that SGX uses some memory pages that are walled off
> from access by the OS. This means they:
> 1) Don't have "struct page" describing them
> 2) Don't appear in the kernel 1:1 map
> 
> But they are still backed by normal DDR memory, so errors can occur.
> 
> Parts 1-4 of this series handle the internal SGX bits to keep track of
> these pages in an error context. They've had a fair amount of review
> on the linux-sgx list (but if any of the 37 subscribers to that list
> not named Jarkko or Reinette want to chime in with extra comments and
> {Acked,Reviewed,Tested}-by that would be great).
> 
> Linux-mm reviewers can (if they like) skip to part 5 where two changes are
> made: 1) Hook into memory_failure() in the same spot as device mapping 2)
> Skip trying to change 1:1 map (since SGX pages aren't there).
> 
> The hooks have generic looking names rather than specifically saying
> "sgx" at the suggestion of Dave Hansen. I'm not wedded to the names,
> so better suggestions welcome.  I could also change to using some
> "ARCH_HAS_PLATFORM_PAGES" config bits if that's the current fashion.
> 
> Rafael (and other ACPI list readers) can skip to parts 6 & 7 where there
> are hooks into error injection and reporting to simply say "these odd
> looking physical addresses are actually ok to use). I added some extra
> notes to the einj.rst documentation on how to inject into SGX memory.
> 
> Tony Luck (7):
>   x86/sgx: Add new sgx_epc_page flag bit to mark in-use pages
>   x86/sgx: Add infrastructure to identify SGX EPC pages
>   x86/sgx: Initial poison handling for dirty and free pages
>   x86/sgx: Add SGX infrastructure to recover from poison
>   x86/sgx: Hook arch_memory_failure() into mainline code
>   x86/sgx: Add hook to error injection address validation
>   x86/sgx: Add check for SGX pages to ghes_do_memory_failure()
> 
>  .../firmware-guide/acpi/apei/einj.rst         |  19 ++++
>  arch/x86/include/asm/processor.h              |   8 ++
>  arch/x86/include/asm/set_memory.h             |   4 +
>  arch/x86/kernel/cpu/sgx/main.c                | 104 +++++++++++++++++-
>  arch/x86/kernel/cpu/sgx/sgx.h                 |   6 +-
>  drivers/acpi/apei/einj.c                      |   3 +-
>  drivers/acpi/apei/ghes.c                      |   2 +-
>  include/linux/mm.h                            |  14 +++
>  mm/memory-failure.c                           |  19 +++-
>  9 files changed, 168 insertions(+), 11 deletions(-)
> 
> 
> base-commit: 64570fbc14f8d7cb3fe3995f20e26bc25ce4b2cc

I think you instructed me on this before but I've forgot it:
how do I simulate this and test how it works?

/Jarkko

RE: [PATCH v9 0/7] Basic recovery for machine checks inside SGX

[Luck, Tony]

> I think you instructed me on this before but I've forgot it:
> how do I simulate this and test how it works?

Jarkko,

You can test the non-execution paths (e.g. where the memory error is
reported by a patrol scrubber in the memory controller) by:

# echo 0x{some_SGX_EPC_ADDRESS} > /sys/devices/system/memory/hard_offline_page

The execution paths are more difficult. You need a system that can inject
errors into EPC memory. There are some hints in the Documenation changes
in part 0006.

Reinette posted some changes to sgx tests that she used to validate.

-Tony

[PATCH v10 0/7] Basic recovery for machine checks inside SGX

[Tony Luck]

v10 (based on v5.15-rc6)

Changes since v9:

ACPI reviewers (Rafael): No changes to parts 6 & 7.

MM reviewers (Horiguchi-san): No changes to part 5.

Jarkko:
	Added Reviewed-by tags to remaining patches.
	N.B. I kept the tags on parts 1, 3, 4 because
	changes based on Sean feedback didn't seem
	consequential. Please let me know if you disagree
	and see new problems introduced by me trying to
	follow Sean's feedback.

Sean:
	1) Reverse the polarity of the neutron flow (sorry,
	Dr Who fan will always autocomplete a sentence that
	begins "reverse the polarity" that way.) Actual change
	is for the new flag bit. Instead of marking in-use
	pages with the new bit, mark free pages instead. This
	avoids the weirdness where I marked the pages on the
	dirty list as "in-use", when clearly they are not.

	2) Race conditions adding poisoned pages to the global
	list of poisoned pages.
	Fixed this by changing from a global list to a per-node
	list. Additions are protected by the node->lock.

	3) Use list_move() instead of list_del(); list_add()
	Fixed both places I used this idiom.

	4) Race looking at page->poison when cleaning dirty pages.
	Added a comment documenting why losing this race isn't
	overly harmful.

Tony Luck (7):
  x86/sgx: Add new sgx_epc_page flag bit to mark free pages
  x86/sgx: Add infrastructure to identify SGX EPC pages
  x86/sgx: Initial poison handling for dirty and free pages
  x86/sgx: Add SGX infrastructure to recover from poison
  x86/sgx: Hook arch_memory_failure() into mainline code
  x86/sgx: Add hook to error injection address validation
  x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

 .../firmware-guide/acpi/apei/einj.rst         |  19 +++
 arch/x86/include/asm/processor.h              |   8 ++
 arch/x86/include/asm/set_memory.h             |   4 +
 arch/x86/kernel/cpu/sgx/main.c                | 113 +++++++++++++++++-
 arch/x86/kernel/cpu/sgx/sgx.h                 |   7 +-
 drivers/acpi/apei/einj.c                      |   3 +-
 drivers/acpi/apei/ghes.c                      |   2 +-
 include/linux/mm.h                            |  14 +++
 mm/memory-failure.c                           |  19 ++-
 9 files changed, 179 insertions(+), 10 deletions(-)


base-commit: 519d81956ee277b4419c723adfb154603c2565ba
-- 
2.31.1

[PATCH v10 1/7] x86/sgx: Add new sgx_epc_page flag bit to mark free pages

[Tony Luck]

SGX EPC pages go through the following life cycle:

        DIRTY ---> FREE ---> IN-USE --\
                    ^                 |
                    \-----------------/

Recovery action for poison for a DIRTY or FREE page is simple. Just
make sure never to allocate the page. IN-USE pages need some extra
handling.

Add a new flag bit SGX_EPC_PAGE_IS_FREE that is set when a page
is added to a free list and cleared when the page is allocated.

Notes:

1) These transitions are made while holding the node->lock so that
   future code that checks the flags while holding the node->lock
   can be sure that if the SGX_EPC_PAGE_IS_FREE bit is set, then the
   page is on the free list.

2) Initially while the pages are on the dirty list the
   SGX_EPC_PAGE_IS_FREE bit is cleared.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 2 ++
 arch/x86/kernel/cpu/sgx/sgx.h  | 3 +++
 2 files changed, 5 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 63d3de02bbcc..825aa91516c8 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -472,6 +472,7 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
 	page = list_first_entry(&node->free_page_list, struct sgx_epc_page, list);
 	list_del_init(&page->list);
 	sgx_nr_free_pages--;
+	page->flags = 0;
 
 	spin_unlock(&node->lock);
 
@@ -626,6 +627,7 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 
 	list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
+	page->flags = SGX_EPC_PAGE_IS_FREE;
 
 	spin_unlock(&node->lock);
 }
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index 4628acec0009..5906471156c5 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -26,6 +26,9 @@
 /* Pages, which are being tracked by the page reclaimer. */
 #define SGX_EPC_PAGE_RECLAIMER_TRACKED	BIT(0)
 
+/* Pages on free list */
+#define SGX_EPC_PAGE_IS_FREE		BIT(1)
+
 struct sgx_epc_page {
 	unsigned int section;
 	unsigned int flags;
-- 
2.31.1

[PATCH v10 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[Tony Luck]

X86 machine check architecture reports a physical address when there
is a memory error. Handling that error requires a method to determine
whether the physical address reported is in any of the areas reserved
for EPC pages by BIOS.

SGX EPC pages do not have Linux "struct page" associated with them.

Keep track of the mapping from ranges of EPC pages to the sections
that contain them using an xarray.

Create a function arch_is_platform_page() that simply reports whether an
address is an EPC page for use elsewhere in the kernel. The ACPI error
injection code needs this function and is typically built as a module,
so export it.

Note that arch_is_platform_page() will be slower than other similar
"what type is this page" functions that can simply check bits in the
"struct page".  If there is some future performance critical user of
this function it may need to be implemented in a more efficient way.

Note also that the current implementation of xarray allocates a few
hundred kilobytes for this usage on a system with 4GB of SGX EPC memory
configured. This isn't ideal, but worth it for the code simplicity.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 825aa91516c8..5c02cffdabc8 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -20,6 +20,7 @@ struct sgx_epc_section sgx_epc_sections[SGX_MAX_EPC_SECTIONS];
 static int sgx_nr_epc_sections;
 static struct task_struct *ksgxd_tsk;
 static DECLARE_WAIT_QUEUE_HEAD(ksgxd_waitq);
+static DEFINE_XARRAY(sgx_epc_address_space);
 
 /*
  * These variables are part of the state of the reclaimer, and must be accessed
@@ -650,6 +651,8 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	}
 
 	section->phys_addr = phys_addr;
+	xa_store_range(&sgx_epc_address_space, section->phys_addr,
+		       phys_addr + size - 1, section, GFP_KERNEL);
 
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
@@ -661,6 +664,12 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	return true;
 }
 
+bool arch_is_platform_page(u64 paddr)
+{
+	return !!xa_load(&sgx_epc_address_space, paddr);
+}
+EXPORT_SYMBOL_GPL(arch_is_platform_page);
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
-- 
2.31.1

[PATCH v10 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Tony Luck]

A memory controller patrol scrubber can report poison in a page
that isn't currently being used.

Add "poison" field in the sgx_epc_page that can be set for an
sgx_epc_page. Check for it:
1) When sanitizing dirty pages
2) When freeing epc pages

Poison is a new field separated from flags to avoid having to make all
updates to flags atomic, or integrate poison state changes into some
other locking scheme to protect flags (Currently just sgx_reclaimer_lock
which protects the SGX_EPC_PAGE_RECLAIMER_TRACKED bit in page->flags).

In both cases place the poisoned page on a per-node list of poisoned
epc pages to make sure it will not be reallocated.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 26 +++++++++++++++++++++++++-
 arch/x86/kernel/cpu/sgx/sgx.h  |  4 +++-
 2 files changed, 28 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 5c02cffdabc8..e5fcb8354bcc 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -62,6 +62,24 @@ static void __sgx_sanitize_pages(struct list_head *dirty_page_list)
 
 		page = list_first_entry(dirty_page_list, struct sgx_epc_page, list);
 
+		/*
+		 * Checking page->poison without holding the node->lock
+		 * is racy, but losing the race (i.e. poison is set just
+		 * after the check) just means __eremove() will be uselessly
+		 * called for a page that sgx_free_epc_page() will put onto
+		 * the node->sgx_poison_page_list later.
+		 */
+		if (page->poison) {
+			struct sgx_epc_section *section = &sgx_epc_sections[page->section];
+			struct sgx_numa_node *node = section->node;
+
+			spin_lock(&node->lock);
+			list_move(&page->list, &node->sgx_poison_page_list);
+			spin_unlock(&node->lock);
+
+			continue;
+		}
+
 		ret = __eremove(sgx_get_epc_virt_addr(page));
 		if (!ret) {
 			/*
@@ -626,7 +644,11 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 
 	spin_lock(&node->lock);
 
-	list_add_tail(&page->list, &node->free_page_list);
+	page->owner = NULL;
+	if (page->poison)
+		list_add(&page->list, &node->sgx_poison_page_list);
+	else
+		list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
 	page->flags = SGX_EPC_PAGE_IS_FREE;
 
@@ -658,6 +680,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 		section->pages[i].section = index;
 		section->pages[i].flags = 0;
 		section->pages[i].owner = NULL;
+		section->pages[i].poison = 0;
 		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
 	}
 
@@ -724,6 +747,7 @@ static bool __init sgx_page_cache_init(void)
 		if (!node_isset(nid, sgx_numa_mask)) {
 			spin_lock_init(&sgx_numa_nodes[nid].lock);
 			INIT_LIST_HEAD(&sgx_numa_nodes[nid].free_page_list);
+			INIT_LIST_HEAD(&sgx_numa_nodes[nid].sgx_poison_page_list);
 			node_set(nid, sgx_numa_mask);
 		}
 
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index 5906471156c5..9ec3136c7800 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -31,7 +31,8 @@
 
 struct sgx_epc_page {
 	unsigned int section;
-	unsigned int flags;
+	u16 flags;
+	u16 poison;
 	struct sgx_encl_page *owner;
 	struct list_head list;
 };
@@ -42,6 +43,7 @@ struct sgx_epc_page {
  */
 struct sgx_numa_node {
 	struct list_head free_page_list;
+	struct list_head sgx_poison_page_list;
 	spinlock_t lock;
 };
 
-- 
2.31.1

[PATCH v10 4/7] x86/sgx: Add SGX infrastructure to recover from poison

[Tony Luck]

Provide a recovery function sgx_memory_failure(). If the poison was
consumed synchronously then send a SIGBUS. Note that the virtual
address of the access is not included with the SIGBUS as is the case
for poison outside of SGX enclaves. This doesn't matter as addresses
of code/data inside an enclave is of little to no use to code executing
outside the (now dead) enclave.

Poison found in a free page results in the page being moved from the
free list to the per-node poison page list.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 76 ++++++++++++++++++++++++++++++++++
 1 file changed, 76 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index e5fcb8354bcc..231c494dfd40 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -693,6 +693,82 @@ bool arch_is_platform_page(u64 paddr)
 }
 EXPORT_SYMBOL_GPL(arch_is_platform_page);
 
+static struct sgx_epc_page *sgx_paddr_to_page(u64 paddr)
+{
+	struct sgx_epc_section *section;
+
+	section = xa_load(&sgx_epc_address_space, paddr);
+	if (!section)
+		return NULL;
+
+	return &section->pages[PFN_DOWN(paddr - section->phys_addr)];
+}
+
+/*
+ * Called in process context to handle a hardware reported
+ * error in an SGX EPC page.
+ * If the MF_ACTION_REQUIRED bit is set in flags, then the
+ * context is the task that consumed the poison data. Otherwise
+ * this is called from a kernel thread unrelated to the page.
+ */
+int arch_memory_failure(unsigned long pfn, int flags)
+{
+	struct sgx_epc_page *page = sgx_paddr_to_page(pfn << PAGE_SHIFT);
+	struct sgx_epc_section *section;
+	struct sgx_numa_node *node;
+
+	/*
+	 * mm/memory-failure.c calls this routine for all errors
+	 * where there isn't a "struct page" for the address. But that
+	 * includes other address ranges besides SGX.
+	 */
+	if (!page)
+		return -ENXIO;
+
+	/*
+	 * If poison was consumed synchronously. Send a SIGBUS to
+	 * the task. Hardware has already exited the SGX enclave and
+	 * will not allow re-entry to an enclave that has a memory
+	 * error. The signal may help the task understand why the
+	 * enclave is broken.
+	 */
+	if (flags & MF_ACTION_REQUIRED)
+		force_sig(SIGBUS);
+
+	section = &sgx_epc_sections[page->section];
+	node = section->node;
+
+	spin_lock(&node->lock);
+
+	/* Already poisoned? Nothing more to do */
+	if (page->poison)
+		goto out;
+
+	page->poison = 1;
+
+	/*
+	 * If the page is on a free list, move it to the per-node
+	 * poison page list.
+	 */
+	if (page->flags & SGX_EPC_PAGE_IS_FREE) {
+		list_move(&page->list, &node->sgx_poison_page_list);
+		goto out;
+	}
+
+	/*
+	 * TBD: Add additional plumbing to enable pre-emptive
+	 * action for asynchronous poison notification. Until
+	 * then just hope that the poison:
+	 * a) is not accessed - sgx_free_epc_page() will deal with it
+	 *    when the user gives it back
+	 * b) results in a recoverable machine check rather than
+	 *    a fatal one
+	 */
+out:
+	spin_unlock(&node->lock);
+	return 0;
+}
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
-- 
2.31.1

[PATCH v10 5/7] x86/sgx: Hook arch_memory_failure() into mainline code

[Tony Luck]

Add a call inside memory_failure() to call the arch specific code
to check if the address is an SGX EPC page and handle it.

Note the SGX EPC pages do not have a "struct page" entry, so the hook
goes in at the same point as the device mapping hook.

Pull the call to acquire the mutex earlier so the SGX errors are also
protected.

Make set_mce_nospec() skip SGX pages when trying to adjust
the 1:1 map.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/include/asm/processor.h  |  8 ++++++++
 arch/x86/include/asm/set_memory.h |  4 ++++
 include/linux/mm.h                | 14 ++++++++++++++
 mm/memory-failure.c               | 19 +++++++++++++------
 4 files changed, 39 insertions(+), 6 deletions(-)

diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index 9ad2acaaae9b..4865f2860a4f 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -853,4 +853,12 @@ enum mds_mitigations {
 	MDS_MITIGATION_VMWERV,
 };
 
+#ifdef CONFIG_X86_SGX
+int arch_memory_failure(unsigned long pfn, int flags);
+#define arch_memory_failure arch_memory_failure
+
+bool arch_is_platform_page(u64 paddr);
+#define arch_is_platform_page arch_is_platform_page
+#endif
+
 #endif /* _ASM_X86_PROCESSOR_H */
diff --git a/arch/x86/include/asm/set_memory.h b/arch/x86/include/asm/set_memory.h
index 43fa081a1adb..ce8dd215f5b3 100644
--- a/arch/x86/include/asm/set_memory.h
+++ b/arch/x86/include/asm/set_memory.h
@@ -2,6 +2,7 @@
 #ifndef _ASM_X86_SET_MEMORY_H
 #define _ASM_X86_SET_MEMORY_H
 
+#include <linux/mm.h>
 #include <asm/page.h>
 #include <asm-generic/set_memory.h>
 
@@ -98,6 +99,9 @@ static inline int set_mce_nospec(unsigned long pfn, bool unmap)
 	unsigned long decoy_addr;
 	int rc;
 
+	/* SGX pages are not in the 1:1 map */
+	if (arch_is_platform_page(pfn << PAGE_SHIFT))
+		return 0;
 	/*
 	 * We would like to just call:
 	 *      set_memory_XX((unsigned long)pfn_to_kaddr(pfn), 1);
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 73a52aba448f..62b199ed5ec6 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -3284,5 +3284,19 @@ static inline int seal_check_future_write(int seals, struct vm_area_struct *vma)
 	return 0;
 }
 
+#ifndef arch_memory_failure
+static inline int arch_memory_failure(unsigned long pfn, int flags)
+{
+	return -ENXIO;
+}
+#endif
+
+#ifndef arch_is_platform_page
+static inline bool arch_is_platform_page(u64 paddr)
+{
+	return false;
+}
+#endif
+
 #endif /* __KERNEL__ */
 #endif /* _LINUX_MM_H */
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 3e6449f2102a..b1cbf9845c19 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -1632,21 +1632,28 @@ int memory_failure(unsigned long pfn, int flags)
 	if (!sysctl_memory_failure_recovery)
 		panic("Memory failure on page %lx", pfn);
 
+	mutex_lock(&mf_mutex);
+
 	p = pfn_to_online_page(pfn);
 	if (!p) {
+		res = arch_memory_failure(pfn, flags);
+		if (res == 0)
+			goto unlock_mutex;
+
 		if (pfn_valid(pfn)) {
 			pgmap = get_dev_pagemap(pfn, NULL);
-			if (pgmap)
-				return memory_failure_dev_pagemap(pfn, flags,
-								  pgmap);
+			if (pgmap) {
+				res = memory_failure_dev_pagemap(pfn, flags,
+								 pgmap);
+				goto unlock_mutex;
+			}
 		}
 		pr_err("Memory failure: %#lx: memory outside kernel control\n",
 			pfn);
-		return -ENXIO;
+		res = -ENXIO;
+		goto unlock_mutex;
 	}
 
-	mutex_lock(&mf_mutex);
-
 try_again:
 	if (PageHuge(p)) {
 		res = memory_failure_hugetlb(pfn, flags);
-- 
2.31.1

Re: [PATCH v10 5/7] x86/sgx: Hook arch_memory_failure() into mainline code

[Naoya Horiguchi]

On Mon, Oct 18, 2021 at 01:25:40PM -0700, Tony Luck wrote:
> Add a call inside memory_failure() to call the arch specific code
> to check if the address is an SGX EPC page and handle it.
> 
> Note the SGX EPC pages do not have a "struct page" entry, so the hook
> goes in at the same point as the device mapping hook.
> 
> Pull the call to acquire the mutex earlier so the SGX errors are also
> protected.
> 
> Make set_mce_nospec() skip SGX pages when trying to adjust
> the 1:1 map.
> 
> Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
> Tested-by: Reinette Chatre <reinette.chatre@intel.com>
> Signed-off-by: Tony Luck <tony.luck@intel.com>
> ---
...
> diff --git a/include/linux/mm.h b/include/linux/mm.h
> index 73a52aba448f..62b199ed5ec6 100644
> --- a/include/linux/mm.h
> +++ b/include/linux/mm.h
> @@ -3284,5 +3284,19 @@ static inline int seal_check_future_write(int seals, struct vm_area_struct *vma)
>  	return 0;
>  }
>  
> +#ifndef arch_memory_failure
> +static inline int arch_memory_failure(unsigned long pfn, int flags)
> +{
> +	return -ENXIO;
> +}
> +#endif
> +
> +#ifndef arch_is_platform_page
> +static inline bool arch_is_platform_page(u64 paddr)
> +{
> +	return false;
> +}
> +#endif
> +

How about putting these definitions near the other related functions
in the same file (like below)?

  ...
  extern void shake_page(struct page *p);
  extern atomic_long_t num_poisoned_pages __read_mostly;
  extern int soft_offline_page(unsigned long pfn, int flags);
  
  // here?
  
  /*
   * Error handlers for various types of pages.
   */
  enum mf_result {

Otherwise, the patch looks good to me.

Reviewed-by: Naoya Horiguchi <naoya.horiguchi@nec.com>

Thanks,
Naoya Horiguchi

RE: [PATCH v10 5/7] x86/sgx: Hook arch_memory_failure() into mainline code

[Luck, Tony]

> How about putting these definitions near the other related functions
> in the same file (like below)?
>
>  ...
>  extern void shake_page(struct page *p);
>  extern atomic_long_t num_poisoned_pages __read_mostly;
>  extern int soft_offline_page(unsigned long pfn, int flags);
>  
>  // here?

Makes sense to group together with these other RAS bits.
I'll move the definitions here.
  

> Otherwise, the patch looks good to me.
>
> Reviewed-by: Naoya Horiguchi <naoya.horiguchi@nec.com>

Thanks for the review!

-Tony

[PATCH v10 6/7] x86/sgx: Add hook to error injection address validation

[Tony Luck]

SGX reserved memory does not appear in the standard address maps.

Add hook to call into the SGX code to check if an address is located
in SGX memory.

There are other challenges in injecting errors into SGX. Update the
documentation with a sequence of operations to inject.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 .../firmware-guide/acpi/apei/einj.rst         | 19 +++++++++++++++++++
 drivers/acpi/apei/einj.c                      |  3 ++-
 2 files changed, 21 insertions(+), 1 deletion(-)

diff --git a/Documentation/firmware-guide/acpi/apei/einj.rst b/Documentation/firmware-guide/acpi/apei/einj.rst
index c042176e1707..55e2331a6438 100644
--- a/Documentation/firmware-guide/acpi/apei/einj.rst
+++ b/Documentation/firmware-guide/acpi/apei/einj.rst
@@ -181,5 +181,24 @@ You should see something like this in dmesg::
   [22715.834759] EDAC sbridge MC3: PROCESSOR 0:306e7 TIME 1422553404 SOCKET 0 APIC 0
   [22716.616173] EDAC MC3: 1 CE memory read error on CPU_SrcID#0_Channel#0_DIMM#0 (channel:0 slot:0 page:0x12345 offset:0x0 grain:32 syndrome:0x0 -  area:DRAM err_code:0001:0090 socket:0 channel_mask:1 rank:0)
 
+Special notes for injection into SGX enclaves:
+
+There may be a separate BIOS setup option to enable SGX injection.
+
+The injection process consists of setting some special memory controller
+trigger that will inject the error on the next write to the target
+address. But the h/w prevents any software outside of an SGX enclave
+from accessing enclave pages (even BIOS SMM mode).
+
+The following sequence can be used:
+  1) Determine physical address of enclave page
+  2) Use "notrigger=1" mode to inject (this will setup
+     the injection address, but will not actually inject)
+  3) Enter the enclave
+  4) Store data to the virtual address matching physical address from step 1
+  5) Execute CLFLUSH for that virtual address
+  6) Spin delay for 250ms
+  7) Read from the virtual address. This will trigger the error
+
 For more information about EINJ, please refer to ACPI specification
 version 4.0, section 17.5 and ACPI 5.0, section 18.6.
diff --git a/drivers/acpi/apei/einj.c b/drivers/acpi/apei/einj.c
index 2882450c443e..67c335baad52 100644
--- a/drivers/acpi/apei/einj.c
+++ b/drivers/acpi/apei/einj.c
@@ -544,7 +544,8 @@ static int einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
 	    ((region_intersects(base_addr, size, IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE)
 				!= REGION_INTERSECTS) &&
 	     (region_intersects(base_addr, size, IORESOURCE_MEM, IORES_DESC_PERSISTENT_MEMORY)
-				!= REGION_INTERSECTS)))
+				!= REGION_INTERSECTS) &&
+	     !arch_is_platform_page(base_addr)))
 		return -EINVAL;
 
 inject:
-- 
2.31.1

[PATCH v10 7/7] x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

[Tony Luck]

SGX EPC pages do not have a "struct page" associated with them so the
pfn_valid() sanity check fails and results in a warning message to
the console.

Add an additional check to skip the warning if the address of the error
is in an SGX EPC page.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 drivers/acpi/apei/ghes.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 0c8330ed1ffd..0c5c9acc6254 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -449,7 +449,7 @@ static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 		return false;
 
 	pfn = PHYS_PFN(physical_addr);
-	if (!pfn_valid(pfn)) {
+	if (!pfn_valid(pfn) && !arch_is_platform_page(physical_addr)) {
 		pr_warn_ratelimited(FW_WARN GHES_PFX
 		"Invalid address in generic error data: %#llx\n",
 		physical_addr);
-- 
2.31.1

[PATCH v11 0/7] Basic recovery for machine checks inside SGX

[Tony Luck]

Boris,

I took this series out of lkml/x86 for a few revisions, I think
the last one posted to lkml was v5. So much has changed since then
that it might be easier to just look at this as if it were v1 and
ignore the earlier history.

First four patches add infrastructure within the SGX code to
track enclave pages (because these pages don't have a "struct
page" as they aren't directly accessible by Linux). All have
"Reviewed-by" tags from Jarkko (SGX maintainer).

Patch 5 hooks into memory_failure() to invoke recovery if
the physical address is in enclave space. This has a
"Reviewed-by" tag from Naoya Horiguchi the maintainer for
mm/memory-failure.c

Patch 6 is a hook into the error injection code and addition
to the error injection documentation explaining extra steps
needed to inject into SGX enclave memory.

Patch 7 is a hook into GHES error reporting path to recognize
that SGX enclave addresses are valid and need processing.

-Tony

Tony Luck (7):
  x86/sgx: Add new sgx_epc_page flag bit to mark free pages
  x86/sgx: Add infrastructure to identify SGX EPC pages
  x86/sgx: Initial poison handling for dirty and free pages
  x86/sgx: Add SGX infrastructure to recover from poison
  x86/sgx: Hook arch_memory_failure() into mainline code
  x86/sgx: Add hook to error injection address validation
  x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

 .../firmware-guide/acpi/apei/einj.rst         |  19 +++
 arch/x86/Kconfig                              |   1 +
 arch/x86/include/asm/processor.h              |   8 ++
 arch/x86/include/asm/set_memory.h             |   4 +
 arch/x86/kernel/cpu/sgx/main.c                | 113 +++++++++++++++++-
 arch/x86/kernel/cpu/sgx/sgx.h                 |   7 +-
 drivers/acpi/apei/einj.c                      |   3 +-
 drivers/acpi/apei/ghes.c                      |   2 +-
 include/linux/mm.h                            |  13 ++
 mm/memory-failure.c                           |  19 ++-
 10 files changed, 179 insertions(+), 10 deletions(-)


base-commit: 3906fe9bb7f1a2c8667ae54e967dc8690824f4ea
-- 
2.31.1

[PATCH v11 1/7] x86/sgx: Add new sgx_epc_page flag bit to mark free pages

[Tony Luck]

SGX EPC pages go through the following life cycle:

        DIRTY ---> FREE ---> IN-USE --\
                    ^                 |
                    \-----------------/

Recovery action for poison for a DIRTY or FREE page is simple. Just
make sure never to allocate the page. IN-USE pages need some extra
handling.

Add a new flag bit SGX_EPC_PAGE_IS_FREE that is set when a page
is added to a free list and cleared when the page is allocated.

Notes:

1) These transitions are made while holding the node->lock so that
   future code that checks the flags while holding the node->lock
   can be sure that if the SGX_EPC_PAGE_IS_FREE bit is set, then the
   page is on the free list.

2) Initially while the pages are on the dirty list the
   SGX_EPC_PAGE_IS_FREE bit is cleared.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 2 ++
 arch/x86/kernel/cpu/sgx/sgx.h  | 3 +++
 2 files changed, 5 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 63d3de02bbcc..825aa91516c8 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -472,6 +472,7 @@ static struct sgx_epc_page *__sgx_alloc_epc_page_from_node(int nid)
 	page = list_first_entry(&node->free_page_list, struct sgx_epc_page, list);
 	list_del_init(&page->list);
 	sgx_nr_free_pages--;
+	page->flags = 0;
 
 	spin_unlock(&node->lock);
 
@@ -626,6 +627,7 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 
 	list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
+	page->flags = SGX_EPC_PAGE_IS_FREE;
 
 	spin_unlock(&node->lock);
 }
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index 4628acec0009..5906471156c5 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -26,6 +26,9 @@
 /* Pages, which are being tracked by the page reclaimer. */
 #define SGX_EPC_PAGE_RECLAIMER_TRACKED	BIT(0)
 
+/* Pages on free list */
+#define SGX_EPC_PAGE_IS_FREE		BIT(1)
+
 struct sgx_epc_page {
 	unsigned int section;
 	unsigned int flags;
-- 
2.31.1

[PATCH v11 2/7] x86/sgx: Add infrastructure to identify SGX EPC pages

[Tony Luck]

X86 machine check architecture reports a physical address when there
is a memory error. Handling that error requires a method to determine
whether the physical address reported is in any of the areas reserved
for EPC pages by BIOS.

SGX EPC pages do not have Linux "struct page" associated with them.

Keep track of the mapping from ranges of EPC pages to the sections
that contain them using an xarray. N.B. adds CONFIG_XARRAY_MULTI to
the SGX dependecies. So "select" that in arch/x86/Kconfig for X86/SGX.

Create a function arch_is_platform_page() that simply reports whether an
address is an EPC page for use elsewhere in the kernel. The ACPI error
injection code needs this function and is typically built as a module,
so export it.

Note that arch_is_platform_page() will be slower than other similar
"what type is this page" functions that can simply check bits in the
"struct page".  If there is some future performance critical user of
this function it may need to be implemented in a more efficient way.

Note also that the current implementation of xarray allocates a few
hundred kilobytes for this usage on a system with 4GB of SGX EPC memory
configured. This isn't ideal, but worth it for the code simplicity.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/Kconfig               | 1 +
 arch/x86/kernel/cpu/sgx/main.c | 9 +++++++++
 2 files changed, 10 insertions(+)

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index d9830e7e1060..b3b5b5a31f89 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -1902,6 +1902,7 @@ config X86_SGX
 	select SRCU
 	select MMU_NOTIFIER
 	select NUMA_KEEP_MEMINFO if NUMA
+	select XARRAY_MULTI
 	help
 	  Intel(R) Software Guard eXtensions (SGX) is a set of CPU instructions
 	  that can be used by applications to set aside private regions of code
diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 825aa91516c8..5c02cffdabc8 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -20,6 +20,7 @@ struct sgx_epc_section sgx_epc_sections[SGX_MAX_EPC_SECTIONS];
 static int sgx_nr_epc_sections;
 static struct task_struct *ksgxd_tsk;
 static DECLARE_WAIT_QUEUE_HEAD(ksgxd_waitq);
+static DEFINE_XARRAY(sgx_epc_address_space);
 
 /*
  * These variables are part of the state of the reclaimer, and must be accessed
@@ -650,6 +651,8 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	}
 
 	section->phys_addr = phys_addr;
+	xa_store_range(&sgx_epc_address_space, section->phys_addr,
+		       phys_addr + size - 1, section, GFP_KERNEL);
 
 	for (i = 0; i < nr_pages; i++) {
 		section->pages[i].section = index;
@@ -661,6 +664,12 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 	return true;
 }
 
+bool arch_is_platform_page(u64 paddr)
+{
+	return !!xa_load(&sgx_epc_address_space, paddr);
+}
+EXPORT_SYMBOL_GPL(arch_is_platform_page);
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
-- 
2.31.1

[PATCH v11 3/7] x86/sgx: Initial poison handling for dirty and free pages

[Tony Luck]

A memory controller patrol scrubber can report poison in a page
that isn't currently being used.

Add "poison" field in the sgx_epc_page that can be set for an
sgx_epc_page. Check for it:
1) When sanitizing dirty pages
2) When freeing epc pages

Poison is a new field separated from flags to avoid having to make all
updates to flags atomic, or integrate poison state changes into some
other locking scheme to protect flags (Currently just sgx_reclaimer_lock
which protects the SGX_EPC_PAGE_RECLAIMER_TRACKED bit in page->flags).

In both cases place the poisoned page on a per-node list of poisoned
epc pages to make sure it will not be reallocated.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 26 +++++++++++++++++++++++++-
 arch/x86/kernel/cpu/sgx/sgx.h  |  4 +++-
 2 files changed, 28 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index 5c02cffdabc8..e5fcb8354bcc 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -62,6 +62,24 @@ static void __sgx_sanitize_pages(struct list_head *dirty_page_list)
 
 		page = list_first_entry(dirty_page_list, struct sgx_epc_page, list);
 
+		/*
+		 * Checking page->poison without holding the node->lock
+		 * is racy, but losing the race (i.e. poison is set just
+		 * after the check) just means __eremove() will be uselessly
+		 * called for a page that sgx_free_epc_page() will put onto
+		 * the node->sgx_poison_page_list later.
+		 */
+		if (page->poison) {
+			struct sgx_epc_section *section = &sgx_epc_sections[page->section];
+			struct sgx_numa_node *node = section->node;
+
+			spin_lock(&node->lock);
+			list_move(&page->list, &node->sgx_poison_page_list);
+			spin_unlock(&node->lock);
+
+			continue;
+		}
+
 		ret = __eremove(sgx_get_epc_virt_addr(page));
 		if (!ret) {
 			/*
@@ -626,7 +644,11 @@ void sgx_free_epc_page(struct sgx_epc_page *page)
 
 	spin_lock(&node->lock);
 
-	list_add_tail(&page->list, &node->free_page_list);
+	page->owner = NULL;
+	if (page->poison)
+		list_add(&page->list, &node->sgx_poison_page_list);
+	else
+		list_add_tail(&page->list, &node->free_page_list);
 	sgx_nr_free_pages++;
 	page->flags = SGX_EPC_PAGE_IS_FREE;
 
@@ -658,6 +680,7 @@ static bool __init sgx_setup_epc_section(u64 phys_addr, u64 size,
 		section->pages[i].section = index;
 		section->pages[i].flags = 0;
 		section->pages[i].owner = NULL;
+		section->pages[i].poison = 0;
 		list_add_tail(&section->pages[i].list, &sgx_dirty_page_list);
 	}
 
@@ -724,6 +747,7 @@ static bool __init sgx_page_cache_init(void)
 		if (!node_isset(nid, sgx_numa_mask)) {
 			spin_lock_init(&sgx_numa_nodes[nid].lock);
 			INIT_LIST_HEAD(&sgx_numa_nodes[nid].free_page_list);
+			INIT_LIST_HEAD(&sgx_numa_nodes[nid].sgx_poison_page_list);
 			node_set(nid, sgx_numa_mask);
 		}
 
diff --git a/arch/x86/kernel/cpu/sgx/sgx.h b/arch/x86/kernel/cpu/sgx/sgx.h
index 5906471156c5..9ec3136c7800 100644
--- a/arch/x86/kernel/cpu/sgx/sgx.h
+++ b/arch/x86/kernel/cpu/sgx/sgx.h
@@ -31,7 +31,8 @@
 
 struct sgx_epc_page {
 	unsigned int section;
-	unsigned int flags;
+	u16 flags;
+	u16 poison;
 	struct sgx_encl_page *owner;
 	struct list_head list;
 };
@@ -42,6 +43,7 @@ struct sgx_epc_page {
  */
 struct sgx_numa_node {
 	struct list_head free_page_list;
+	struct list_head sgx_poison_page_list;
 	spinlock_t lock;
 };
 
-- 
2.31.1

[PATCH v11 4/7] x86/sgx: Add SGX infrastructure to recover from poison

[Tony Luck]

Provide a recovery function sgx_memory_failure(). If the poison was
consumed synchronously then send a SIGBUS. Note that the virtual
address of the access is not included with the SIGBUS as is the case
for poison outside of SGX enclaves. This doesn't matter as addresses
of code/data inside an enclave is of little to no use to code executing
outside the (now dead) enclave.

Poison found in a free page results in the page being moved from the
free list to the per-node poison page list.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/kernel/cpu/sgx/main.c | 76 ++++++++++++++++++++++++++++++++++
 1 file changed, 76 insertions(+)

diff --git a/arch/x86/kernel/cpu/sgx/main.c b/arch/x86/kernel/cpu/sgx/main.c
index e5fcb8354bcc..231c494dfd40 100644
--- a/arch/x86/kernel/cpu/sgx/main.c
+++ b/arch/x86/kernel/cpu/sgx/main.c
@@ -693,6 +693,82 @@ bool arch_is_platform_page(u64 paddr)
 }
 EXPORT_SYMBOL_GPL(arch_is_platform_page);
 
+static struct sgx_epc_page *sgx_paddr_to_page(u64 paddr)
+{
+	struct sgx_epc_section *section;
+
+	section = xa_load(&sgx_epc_address_space, paddr);
+	if (!section)
+		return NULL;
+
+	return &section->pages[PFN_DOWN(paddr - section->phys_addr)];
+}
+
+/*
+ * Called in process context to handle a hardware reported
+ * error in an SGX EPC page.
+ * If the MF_ACTION_REQUIRED bit is set in flags, then the
+ * context is the task that consumed the poison data. Otherwise
+ * this is called from a kernel thread unrelated to the page.
+ */
+int arch_memory_failure(unsigned long pfn, int flags)
+{
+	struct sgx_epc_page *page = sgx_paddr_to_page(pfn << PAGE_SHIFT);
+	struct sgx_epc_section *section;
+	struct sgx_numa_node *node;
+
+	/*
+	 * mm/memory-failure.c calls this routine for all errors
+	 * where there isn't a "struct page" for the address. But that
+	 * includes other address ranges besides SGX.
+	 */
+	if (!page)
+		return -ENXIO;
+
+	/*
+	 * If poison was consumed synchronously. Send a SIGBUS to
+	 * the task. Hardware has already exited the SGX enclave and
+	 * will not allow re-entry to an enclave that has a memory
+	 * error. The signal may help the task understand why the
+	 * enclave is broken.
+	 */
+	if (flags & MF_ACTION_REQUIRED)
+		force_sig(SIGBUS);
+
+	section = &sgx_epc_sections[page->section];
+	node = section->node;
+
+	spin_lock(&node->lock);
+
+	/* Already poisoned? Nothing more to do */
+	if (page->poison)
+		goto out;
+
+	page->poison = 1;
+
+	/*
+	 * If the page is on a free list, move it to the per-node
+	 * poison page list.
+	 */
+	if (page->flags & SGX_EPC_PAGE_IS_FREE) {
+		list_move(&page->list, &node->sgx_poison_page_list);
+		goto out;
+	}
+
+	/*
+	 * TBD: Add additional plumbing to enable pre-emptive
+	 * action for asynchronous poison notification. Until
+	 * then just hope that the poison:
+	 * a) is not accessed - sgx_free_epc_page() will deal with it
+	 *    when the user gives it back
+	 * b) results in a recoverable machine check rather than
+	 *    a fatal one
+	 */
+out:
+	spin_unlock(&node->lock);
+	return 0;
+}
+
 /**
  * A section metric is concatenated in a way that @low bits 12-31 define the
  * bits 12-31 of the metric and @high bits 0-19 define the bits 32-51 of the
-- 
2.31.1

[PATCH v11 5/7] x86/sgx: Hook arch_memory_failure() into mainline code

[Tony Luck]

Add a call inside memory_failure() to call the arch specific code
to check if the address is an SGX EPC page and handle it.

Note the SGX EPC pages do not have a "struct page" entry, so the hook
goes in at the same point as the device mapping hook.

Pull the call to acquire the mutex earlier so the SGX errors are also
protected.

Make set_mce_nospec() skip SGX pages when trying to adjust
the 1:1 map.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Reviewed-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 arch/x86/include/asm/processor.h  |  8 ++++++++
 arch/x86/include/asm/set_memory.h |  4 ++++
 include/linux/mm.h                | 13 +++++++++++++
 mm/memory-failure.c               | 19 +++++++++++++------
 4 files changed, 38 insertions(+), 6 deletions(-)

diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index 9ad2acaaae9b..4865f2860a4f 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -853,4 +853,12 @@ enum mds_mitigations {
 	MDS_MITIGATION_VMWERV,
 };
 
+#ifdef CONFIG_X86_SGX
+int arch_memory_failure(unsigned long pfn, int flags);
+#define arch_memory_failure arch_memory_failure
+
+bool arch_is_platform_page(u64 paddr);
+#define arch_is_platform_page arch_is_platform_page
+#endif
+
 #endif /* _ASM_X86_PROCESSOR_H */
diff --git a/arch/x86/include/asm/set_memory.h b/arch/x86/include/asm/set_memory.h
index 43fa081a1adb..ce8dd215f5b3 100644
--- a/arch/x86/include/asm/set_memory.h
+++ b/arch/x86/include/asm/set_memory.h
@@ -2,6 +2,7 @@
 #ifndef _ASM_X86_SET_MEMORY_H
 #define _ASM_X86_SET_MEMORY_H
 
+#include <linux/mm.h>
 #include <asm/page.h>
 #include <asm-generic/set_memory.h>
 
@@ -98,6 +99,9 @@ static inline int set_mce_nospec(unsigned long pfn, bool unmap)
 	unsigned long decoy_addr;
 	int rc;
 
+	/* SGX pages are not in the 1:1 map */
+	if (arch_is_platform_page(pfn << PAGE_SHIFT))
+		return 0;
 	/*
 	 * We would like to just call:
 	 *      set_memory_XX((unsigned long)pfn_to_kaddr(pfn), 1);
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 73a52aba448f..0aa48b238db2 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -3124,6 +3124,19 @@ extern void shake_page(struct page *p);
 extern atomic_long_t num_poisoned_pages __read_mostly;
 extern int soft_offline_page(unsigned long pfn, int flags);
 
+#ifndef arch_memory_failure
+static inline int arch_memory_failure(unsigned long pfn, int flags)
+{
+	return -ENXIO;
+}
+#endif
+
+#ifndef arch_is_platform_page
+static inline bool arch_is_platform_page(u64 paddr)
+{
+	return false;
+}
+#endif
 
 /*
  * Error handlers for various types of pages.
diff --git a/mm/memory-failure.c b/mm/memory-failure.c
index 3e6449f2102a..b1cbf9845c19 100644
--- a/mm/memory-failure.c
+++ b/mm/memory-failure.c
@@ -1632,21 +1632,28 @@ int memory_failure(unsigned long pfn, int flags)
 	if (!sysctl_memory_failure_recovery)
 		panic("Memory failure on page %lx", pfn);
 
+	mutex_lock(&mf_mutex);
+
 	p = pfn_to_online_page(pfn);
 	if (!p) {
+		res = arch_memory_failure(pfn, flags);
+		if (res == 0)
+			goto unlock_mutex;
+
 		if (pfn_valid(pfn)) {
 			pgmap = get_dev_pagemap(pfn, NULL);
-			if (pgmap)
-				return memory_failure_dev_pagemap(pfn, flags,
-								  pgmap);
+			if (pgmap) {
+				res = memory_failure_dev_pagemap(pfn, flags,
+								 pgmap);
+				goto unlock_mutex;
+			}
 		}
 		pr_err("Memory failure: %#lx: memory outside kernel control\n",
 			pfn);
-		return -ENXIO;
+		res = -ENXIO;
+		goto unlock_mutex;
 	}
 
-	mutex_lock(&mf_mutex);
-
 try_again:
 	if (PageHuge(p)) {
 		res = memory_failure_hugetlb(pfn, flags);
-- 
2.31.1

[PATCH v11 6/7] x86/sgx: Add hook to error injection address validation

[Tony Luck]

SGX reserved memory does not appear in the standard address maps.

Add hook to call into the SGX code to check if an address is located
in SGX memory.

There are other challenges in injecting errors into SGX. Update the
documentation with a sequence of operations to inject.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 .../firmware-guide/acpi/apei/einj.rst         | 19 +++++++++++++++++++
 drivers/acpi/apei/einj.c                      |  3 ++-
 2 files changed, 21 insertions(+), 1 deletion(-)

diff --git a/Documentation/firmware-guide/acpi/apei/einj.rst b/Documentation/firmware-guide/acpi/apei/einj.rst
index c042176e1707..55e2331a6438 100644
--- a/Documentation/firmware-guide/acpi/apei/einj.rst
+++ b/Documentation/firmware-guide/acpi/apei/einj.rst
@@ -181,5 +181,24 @@ You should see something like this in dmesg::
   [22715.834759] EDAC sbridge MC3: PROCESSOR 0:306e7 TIME 1422553404 SOCKET 0 APIC 0
   [22716.616173] EDAC MC3: 1 CE memory read error on CPU_SrcID#0_Channel#0_DIMM#0 (channel:0 slot:0 page:0x12345 offset:0x0 grain:32 syndrome:0x0 -  area:DRAM err_code:0001:0090 socket:0 channel_mask:1 rank:0)
 
+Special notes for injection into SGX enclaves:
+
+There may be a separate BIOS setup option to enable SGX injection.
+
+The injection process consists of setting some special memory controller
+trigger that will inject the error on the next write to the target
+address. But the h/w prevents any software outside of an SGX enclave
+from accessing enclave pages (even BIOS SMM mode).
+
+The following sequence can be used:
+  1) Determine physical address of enclave page
+  2) Use "notrigger=1" mode to inject (this will setup
+     the injection address, but will not actually inject)
+  3) Enter the enclave
+  4) Store data to the virtual address matching physical address from step 1
+  5) Execute CLFLUSH for that virtual address
+  6) Spin delay for 250ms
+  7) Read from the virtual address. This will trigger the error
+
 For more information about EINJ, please refer to ACPI specification
 version 4.0, section 17.5 and ACPI 5.0, section 18.6.
diff --git a/drivers/acpi/apei/einj.c b/drivers/acpi/apei/einj.c
index 2882450c443e..67c335baad52 100644
--- a/drivers/acpi/apei/einj.c
+++ b/drivers/acpi/apei/einj.c
@@ -544,7 +544,8 @@ static int einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
 	    ((region_intersects(base_addr, size, IORESOURCE_SYSTEM_RAM, IORES_DESC_NONE)
 				!= REGION_INTERSECTS) &&
 	     (region_intersects(base_addr, size, IORESOURCE_MEM, IORES_DESC_PERSISTENT_MEMORY)
-				!= REGION_INTERSECTS)))
+				!= REGION_INTERSECTS) &&
+	     !arch_is_platform_page(base_addr)))
 		return -EINVAL;
 
 inject:
-- 
2.31.1

[PATCH v11 7/7] x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

[Tony Luck]

SGX EPC pages do not have a "struct page" associated with them so the
pfn_valid() sanity check fails and results in a warning message to
the console.

Add an additional check to skip the warning if the address of the error
is in an SGX EPC page.

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Tested-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
 drivers/acpi/apei/ghes.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
index 0c8330ed1ffd..0c5c9acc6254 100644
--- a/drivers/acpi/apei/ghes.c
+++ b/drivers/acpi/apei/ghes.c
@@ -449,7 +449,7 @@ static bool ghes_do_memory_failure(u64 physical_addr, int flags)
 		return false;
 
 	pfn = PHYS_PFN(physical_addr);
-	if (!pfn_valid(pfn)) {
+	if (!pfn_valid(pfn) && !arch_is_platform_page(physical_addr)) {
 		pr_warn_ratelimited(FW_WARN GHES_PFX
 		"Invalid address in generic error data: %#llx\n",
 		physical_addr);
-- 
2.31.1

Re: [PATCH v11 7/7] x86/sgx: Add check for SGX pages to ghes_do_memory_failure()

[Rafael J. Wysocki]

On Wed, Oct 27, 2021 at 12:01 AM Tony Luck <tony.luck@intel.com> wrote:
>
> SGX EPC pages do not have a "struct page" associated with them so the
> pfn_valid() sanity check fails and results in a warning message to
> the console.
>
> Add an additional check to skip the warning if the address of the error
> is in an SGX EPC page.
>
> Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
> Tested-by: Reinette Chatre <reinette.chatre@intel.com>
> Signed-off-by: Tony Luck <tony.luck@intel.com>

Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

> ---
>  drivers/acpi/apei/ghes.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c
> index 0c8330ed1ffd..0c5c9acc6254 100644
> --- a/drivers/acpi/apei/ghes.c
> +++ b/drivers/acpi/apei/ghes.c
> @@ -449,7 +449,7 @@ static bool ghes_do_memory_failure(u64 physical_addr, int flags)
>                 return false;
>
>         pfn = PHYS_PFN(physical_addr);
> -       if (!pfn_valid(pfn)) {
> +       if (!pfn_valid(pfn) && !arch_is_platform_page(physical_addr)) {
>                 pr_warn_ratelimited(FW_WARN GHES_PFX
>                 "Invalid address in generic error data: %#llx\n",
>                 physical_addr);
> --
> 2.31.1
>