* [PATCH for_v25 1/4] x86/sgx: Put SECS backing iff retrieving the backing succeeds
2019-12-21 0:31 [PATCH for_v25 0/4] x86/sgx: Reclaim bug fix and cleanup Sean Christopherson
@ 2019-12-21 0:31 ` Sean Christopherson
2019-12-21 0:31 ` [PATCH for_v25 2/4] x86/sgx: Use goto to handle sgx_encl_get_backing() failure in SECS flows Sean Christopherson
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Sean Christopherson @ 2019-12-21 0:31 UTC (permalink / raw)
To: Jarkko Sakkinen; +Cc: linux-sgx
Put the SECS backing if and only if retrieving the backing succeeded,
otherwise sgx_encl_put_backing() will consume uninitialized pointers.
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
---
arch/x86/kernel/cpu/sgx/reclaim.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/x86/kernel/cpu/sgx/reclaim.c b/arch/x86/kernel/cpu/sgx/reclaim.c
index 21fbfe2ed9c4..67bda5528625 100644
--- a/arch/x86/kernel/cpu/sgx/reclaim.c
+++ b/arch/x86/kernel/cpu/sgx/reclaim.c
@@ -361,9 +361,9 @@ static void sgx_reclaimer_write(struct sgx_epc_page *epc_page,
sgx_free_page(encl->secs.epc_page);
encl->secs.epc_page = NULL;
+
+ sgx_encl_put_backing(&secs_backing, true);
}
-
- sgx_encl_put_backing(&secs_backing, true);
}
}
--
2.24.1
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PATCH for_v25 2/4] x86/sgx: Use goto to handle sgx_encl_get_backing() failure in SECS flows
2019-12-21 0:31 [PATCH for_v25 0/4] x86/sgx: Reclaim bug fix and cleanup Sean Christopherson
2019-12-21 0:31 ` [PATCH for_v25 1/4] x86/sgx: Put SECS backing iff retrieving the backing succeeds Sean Christopherson
@ 2019-12-21 0:31 ` Sean Christopherson
2019-12-21 0:31 ` [PATCH for_v25 3/4] x86/sgx: Drop unused @encl parameter from __sgx_encl_ewb() Sean Christopherson
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Sean Christopherson @ 2019-12-21 0:31 UTC (permalink / raw)
To: Jarkko Sakkinen; +Cc: linux-sgx
Use a goto for the SECS backing error path in sgx_reclaimer_write() so
that the happy path is a straight shot, e.g. to make it obvious that the
backing is not being leaked. Reducing the indentation also eliminates a
line wrap.
No functional change intended.
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
---
I could go either way on this patch, included it in case you have a
preference.
arch/x86/kernel/cpu/sgx/reclaim.c | 15 ++++++++-------
1 file changed, 8 insertions(+), 7 deletions(-)
diff --git a/arch/x86/kernel/cpu/sgx/reclaim.c b/arch/x86/kernel/cpu/sgx/reclaim.c
index 67bda5528625..c5c0ba92ab63 100644
--- a/arch/x86/kernel/cpu/sgx/reclaim.c
+++ b/arch/x86/kernel/cpu/sgx/reclaim.c
@@ -355,18 +355,19 @@ static void sgx_reclaimer_write(struct sgx_epc_page *epc_page,
} else if (atomic_read(&encl->flags) & SGX_ENCL_INITIALIZED) {
ret = sgx_encl_get_backing(encl, PFN_DOWN(encl->size),
&secs_backing);
- if (!ret) {
- sgx_encl_ewb(encl->secs.epc_page,
- &secs_backing);
+ if (ret)
+ goto out;
- sgx_free_page(encl->secs.epc_page);
- encl->secs.epc_page = NULL;
+ sgx_encl_ewb(encl->secs.epc_page, &secs_backing);
- sgx_encl_put_backing(&secs_backing, true);
- }
+ sgx_free_page(encl->secs.epc_page);
+ encl->secs.epc_page = NULL;
+
+ sgx_encl_put_backing(&secs_backing, true);
}
}
+out:
mutex_unlock(&encl->lock);
}
--
2.24.1
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PATCH for_v25 3/4] x86/sgx: Drop unused @encl parameter from __sgx_encl_ewb()
2019-12-21 0:31 [PATCH for_v25 0/4] x86/sgx: Reclaim bug fix and cleanup Sean Christopherson
2019-12-21 0:31 ` [PATCH for_v25 1/4] x86/sgx: Put SECS backing iff retrieving the backing succeeds Sean Christopherson
2019-12-21 0:31 ` [PATCH for_v25 2/4] x86/sgx: Use goto to handle sgx_encl_get_backing() failure in SECS flows Sean Christopherson
@ 2019-12-21 0:31 ` Sean Christopherson
2019-12-21 0:31 ` [PATCH for_v25 4/4] x86/sgx: Pre-calculate VA slot virtual address in sgx_encl_ewb() Sean Christopherson
2020-01-02 17:06 ` [PATCH for_v25 0/4] x86/sgx: Reclaim bug fix and cleanup Jarkko Sakkinen
4 siblings, 0 replies; 6+ messages in thread
From: Sean Christopherson @ 2019-12-21 0:31 UTC (permalink / raw)
To: Jarkko Sakkinen; +Cc: linux-sgx
Remove the unused @encl parameter from __sgx_encl_ewb() to help shorten
the line lengths of its call sites.
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
---
arch/x86/kernel/cpu/sgx/reclaim.c | 11 +++++------
1 file changed, 5 insertions(+), 6 deletions(-)
diff --git a/arch/x86/kernel/cpu/sgx/reclaim.c b/arch/x86/kernel/cpu/sgx/reclaim.c
index c5c0ba92ab63..030b1c24da07 100644
--- a/arch/x86/kernel/cpu/sgx/reclaim.c
+++ b/arch/x86/kernel/cpu/sgx/reclaim.c
@@ -223,7 +223,7 @@ static void sgx_reclaimer_block(struct sgx_epc_page *epc_page)
mutex_unlock(&encl->lock);
}
-static int __sgx_encl_ewb(struct sgx_encl *encl, struct sgx_epc_page *epc_page,
+static int __sgx_encl_ewb(struct sgx_epc_page *epc_page,
struct sgx_va_page *va_page, unsigned int va_offset,
struct sgx_backing *backing)
{
@@ -292,7 +292,7 @@ static void sgx_encl_ewb(struct sgx_epc_page *epc_page,
if (sgx_va_page_full(va_page))
list_move_tail(&va_page->list, &encl->va_pages);
- ret = __sgx_encl_ewb(encl, epc_page, va_page, va_offset, backing);
+ ret = __sgx_encl_ewb(epc_page, va_page, va_offset, backing);
if (ret == SGX_NOT_TRACKED) {
ret = __etrack(sgx_epc_addr(encl->secs.epc_page));
if (ret) {
@@ -300,8 +300,7 @@ static void sgx_encl_ewb(struct sgx_epc_page *epc_page,
ENCLS_WARN(ret, "ETRACK");
}
- ret = __sgx_encl_ewb(encl, epc_page, va_page, va_offset,
- backing);
+ ret = __sgx_encl_ewb(epc_page, va_page, va_offset, backing);
if (ret == SGX_NOT_TRACKED) {
/*
* Slow path, send IPIs to kick cpus out of the
@@ -312,8 +311,8 @@ static void sgx_encl_ewb(struct sgx_epc_page *epc_page,
*/
on_each_cpu_mask(sgx_encl_ewb_cpumask(encl),
sgx_ipi_cb, NULL, 1);
- ret = __sgx_encl_ewb(encl, epc_page, va_page,
- va_offset, backing);
+ ret = __sgx_encl_ewb(epc_page, va_page, va_offset,
+ backing);
}
}
--
2.24.1
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PATCH for_v25 4/4] x86/sgx: Pre-calculate VA slot virtual address in sgx_encl_ewb()
2019-12-21 0:31 [PATCH for_v25 0/4] x86/sgx: Reclaim bug fix and cleanup Sean Christopherson
` (2 preceding siblings ...)
2019-12-21 0:31 ` [PATCH for_v25 3/4] x86/sgx: Drop unused @encl parameter from __sgx_encl_ewb() Sean Christopherson
@ 2019-12-21 0:31 ` Sean Christopherson
2020-01-02 17:06 ` [PATCH for_v25 0/4] x86/sgx: Reclaim bug fix and cleanup Jarkko Sakkinen
4 siblings, 0 replies; 6+ messages in thread
From: Sean Christopherson @ 2019-12-21 0:31 UTC (permalink / raw)
To: Jarkko Sakkinen; +Cc: linux-sgx
Now that sgx_epc_addr() is purely a calculation, calculate the VA slot
in sgx_encl_ewb() and pass it to __sgx_encl_ewb() to reduce line lengths
and avoid re-calculating the address on every EWB attempt.
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
---
arch/x86/kernel/cpu/sgx/reclaim.c | 15 +++++++--------
1 file changed, 7 insertions(+), 8 deletions(-)
diff --git a/arch/x86/kernel/cpu/sgx/reclaim.c b/arch/x86/kernel/cpu/sgx/reclaim.c
index 030b1c24da07..a33f1c45477a 100644
--- a/arch/x86/kernel/cpu/sgx/reclaim.c
+++ b/arch/x86/kernel/cpu/sgx/reclaim.c
@@ -223,8 +223,7 @@ static void sgx_reclaimer_block(struct sgx_epc_page *epc_page)
mutex_unlock(&encl->lock);
}
-static int __sgx_encl_ewb(struct sgx_epc_page *epc_page,
- struct sgx_va_page *va_page, unsigned int va_offset,
+static int __sgx_encl_ewb(struct sgx_epc_page *epc_page, void *va_slot,
struct sgx_backing *backing)
{
struct sgx_pageinfo pginfo;
@@ -237,8 +236,7 @@ static int __sgx_encl_ewb(struct sgx_epc_page *epc_page,
pginfo.metadata = (unsigned long)kmap_atomic(backing->pcmd) +
backing->pcmd_offset;
- ret = __ewb(&pginfo, sgx_epc_addr(epc_page),
- sgx_epc_addr(va_page->epc_page) + va_offset);
+ ret = __ewb(&pginfo, sgx_epc_addr(epc_page), va_slot);
kunmap_atomic((void *)(unsigned long)(pginfo.metadata -
backing->pcmd_offset));
@@ -282,6 +280,7 @@ static void sgx_encl_ewb(struct sgx_epc_page *epc_page,
struct sgx_encl *encl = encl_page->encl;
struct sgx_va_page *va_page;
unsigned int va_offset;
+ void *va_slot;
int ret;
encl_page->desc &= ~SGX_ENCL_PAGE_RECLAIMED;
@@ -289,10 +288,11 @@ static void sgx_encl_ewb(struct sgx_epc_page *epc_page,
va_page = list_first_entry(&encl->va_pages, struct sgx_va_page,
list);
va_offset = sgx_alloc_va_slot(va_page);
+ va_slot = sgx_epc_addr(va_page->epc_page) + va_offset;
if (sgx_va_page_full(va_page))
list_move_tail(&va_page->list, &encl->va_pages);
- ret = __sgx_encl_ewb(epc_page, va_page, va_offset, backing);
+ ret = __sgx_encl_ewb(epc_page, va_slot, backing);
if (ret == SGX_NOT_TRACKED) {
ret = __etrack(sgx_epc_addr(encl->secs.epc_page));
if (ret) {
@@ -300,7 +300,7 @@ static void sgx_encl_ewb(struct sgx_epc_page *epc_page,
ENCLS_WARN(ret, "ETRACK");
}
- ret = __sgx_encl_ewb(epc_page, va_page, va_offset, backing);
+ ret = __sgx_encl_ewb(epc_page, va_slot, backing);
if (ret == SGX_NOT_TRACKED) {
/*
* Slow path, send IPIs to kick cpus out of the
@@ -311,8 +311,7 @@ static void sgx_encl_ewb(struct sgx_epc_page *epc_page,
*/
on_each_cpu_mask(sgx_encl_ewb_cpumask(encl),
sgx_ipi_cb, NULL, 1);
- ret = __sgx_encl_ewb(epc_page, va_page, va_offset,
- backing);
+ ret = __sgx_encl_ewb(epc_page, va_slot, backing);
}
}
--
2.24.1
^ permalink raw reply related [flat|nested] 6+ messages in thread* Re: [PATCH for_v25 0/4] x86/sgx: Reclaim bug fix and cleanup
2019-12-21 0:31 [PATCH for_v25 0/4] x86/sgx: Reclaim bug fix and cleanup Sean Christopherson
` (3 preceding siblings ...)
2019-12-21 0:31 ` [PATCH for_v25 4/4] x86/sgx: Pre-calculate VA slot virtual address in sgx_encl_ewb() Sean Christopherson
@ 2020-01-02 17:06 ` Jarkko Sakkinen
4 siblings, 0 replies; 6+ messages in thread
From: Jarkko Sakkinen @ 2020-01-02 17:06 UTC (permalink / raw)
To: Sean Christopherson; +Cc: linux-sgx
On Fri, Dec 20, 2019 at 04:31:52PM -0800, Sean Christopherson wrote:
> One bug fix in the SECS reclaim path and loosely related cleanup.
>
> Sean Christopherson (4):
> x86/sgx: Put SECS backing iff retrieving the backing succeeds
> x86/sgx: Use goto to handle sgx_encl_get_backing() failure in SECS
> flows
> x86/sgx: Drop unused @encl parameter from __sgx_encl_ewb()
> x86/sgx: Pre-calculate VA slot virtual address in sgx_encl_ewb()
>
> arch/x86/kernel/cpu/sgx/reclaim.c | 29 ++++++++++++++---------------
> 1 file changed, 14 insertions(+), 15 deletions(-)
>
> --
> 2.24.1
>
Thank you. I merged these and added a changelog entry describing the bug
that was fixed.
/Jarkko
^ permalink raw reply [flat|nested] 6+ messages in thread