There appears to be no formal statement of what pv_irq_ops.save_fl() is
supposed to return precisely. Native returns the full flags, while lguest and
Xen only return the Interrupt Flag, and both have comments by the
implementations stating that only the Interrupt Flag is looked at. This may
have been true when initially implemented, but no longer is.
To make matters worse, the Xen PVOP leaves the upper bits undefined, making
the BUG_ON() undefined behaviour. Experimentally, this now trips for 32bit PV
guests on Broadwell hardware. The BUG_ON() is consistent for an individual
build, but not consistent for all builds. It has also been a sitting timebomb
since SMAP support was introduced.
Use native_save_fl() instead, which will obtain an accurate view of the AC
flag.
Signed-off-by: Andrew Cooper <[email protected]>
Reviewed-by: David Vrabel <[email protected]>
Tested-by: Rusty Russell <[email protected]>
CC: Thomas Gleixner <[email protected]>
CC: Ingo Molnar <[email protected]>
CC: H. Peter Anvin <[email protected]>
CC: [email protected]
CC: [email protected]
CC: Konrad Rzeszutek Wilk <[email protected]>
CC: Boris Ostrovsky <[email protected]>
CC: xen-devel <[email protected]>
CC: [email protected]
CC: [email protected]
---
arch/x86/kernel/cpu/common.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index a62cf04..4f2fded 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -291,10 +291,9 @@ __setup("nosmap", setup_disable_smap);
static __always_inline void setup_smap(struct cpuinfo_x86 *c)
{
- unsigned long eflags;
+ unsigned long eflags = native_save_fl();
/* This should have been cleared long ago */
- raw_local_save_flags(eflags);
BUG_ON(eflags & X86_EFLAGS_AC);
if (cpu_has(c, X86_FEATURE_SMAP)) {
--
1.7.10.4
On 06/03/2015 02:31 AM, Andrew Cooper wrote:
> There appears to be no formal statement of what pv_irq_ops.save_fl() is
> supposed to return precisely. Native returns the full flags, while lguest and
> Xen only return the Interrupt Flag, and both have comments by the
> implementations stating that only the Interrupt Flag is looked at. This may
> have been true when initially implemented, but no longer is.
>
> To make matters worse, the Xen PVOP leaves the upper bits undefined, making
> the BUG_ON() undefined behaviour. Experimentally, this now trips for 32bit PV
> guests on Broadwell hardware. The BUG_ON() is consistent for an individual
> build, but not consistent for all builds. It has also been a sitting timebomb
> since SMAP support was introduced.
>
> Use native_save_fl() instead, which will obtain an accurate view of the AC
> flag.
Could we fix the Xen pvops wrapper instead to not do things like this?
-hpa
On 04/06/15 07:38, H. Peter Anvin wrote:
> On 06/03/2015 02:31 AM, Andrew Cooper wrote:
>> There appears to be no formal statement of what pv_irq_ops.save_fl() is
>> supposed to return precisely. Native returns the full flags, while lguest and
>> Xen only return the Interrupt Flag, and both have comments by the
>> implementations stating that only the Interrupt Flag is looked at. This may
>> have been true when initially implemented, but no longer is.
>>
>> To make matters worse, the Xen PVOP leaves the upper bits undefined, making
>> the BUG_ON() undefined behaviour. Experimentally, this now trips for 32bit PV
>> guests on Broadwell hardware. The BUG_ON() is consistent for an individual
>> build, but not consistent for all builds. It has also been a sitting timebomb
>> since SMAP support was introduced.
>>
>> Use native_save_fl() instead, which will obtain an accurate view of the AC
>> flag.
> Could we fix the Xen pvops wrapper instead to not do things like this?
>
> -hpa
>
>
We could, and I have a patch for that, but the check would still then be
broken in lguest, and it makes a hotpath rather longer.
Either pv_irq_ops.save_fl() gets defined to handle all flags, and Xen &
lguest need correcting in this regard, or save_fl() gets defined to
handle the interrupt flag only, and this becomes the single problematic
caller in the codebase.
The problem with expanding save_fl() to handle all flags is that
restore_fl() should follow suit, and there are a number of system flags
are inapplicable in such a situation.
~Andrew
Andrew Cooper <[email protected]> writes:
> On 04/06/15 07:38, H. Peter Anvin wrote:
>> On 06/03/2015 02:31 AM, Andrew Cooper wrote:
>>> There appears to be no formal statement of what pv_irq_ops.save_fl() is
>>> supposed to return precisely. Native returns the full flags, while lguest and
>>> Xen only return the Interrupt Flag, and both have comments by the
>>> implementations stating that only the Interrupt Flag is looked at. This may
>>> have been true when initially implemented, but no longer is.
>>>
>>> To make matters worse, the Xen PVOP leaves the upper bits undefined, making
>>> the BUG_ON() undefined behaviour. Experimentally, this now trips for 32bit PV
>>> guests on Broadwell hardware. The BUG_ON() is consistent for an individual
>>> build, but not consistent for all builds. It has also been a sitting timebomb
>>> since SMAP support was introduced.
>>>
>>> Use native_save_fl() instead, which will obtain an accurate view of the AC
>>> flag.
>> Could we fix the Xen pvops wrapper instead to not do things like this?
>>
>> -hpa
>>
>>
>
> We could, and I have a patch for that, but the check would still then be
> broken in lguest, and it makes a hotpath rather longer.
>
> Either pv_irq_ops.save_fl() gets defined to handle all flags, and Xen &
> lguest need correcting in this regard, or save_fl() gets defined to
> handle the interrupt flag only, and this becomes the single problematic
> caller in the codebase.
>
> The problem with expanding save_fl() to handle all flags is that
> restore_fl() should follow suit, and there are a number of system flags
> are inapplicable in such a situation.
Yeah, hard cases make bad law.
I'm not too unhappy with this fix; ideally we'd rename save_fl and
restore_fl to save_eflags_if and restore_eflags_if too.
Cheers,
Rusty.
On 06/04/2015 12:55 PM, Rusty Russell wrote:
>
> Yeah, hard cases make bad law.
>
> I'm not too unhappy with this fix; ideally we'd rename save_fl and
> restore_fl to save_eflags_if and restore_eflags_if too.
>
I would be fine with this... but please document what the bloody
semantics of pvops is actually supposed to be.
-hpa
"H. Peter Anvin" <[email protected]> writes:
> On 06/04/2015 12:55 PM, Rusty Russell wrote:
>>
>> Yeah, hard cases make bad law.
>>
>> I'm not too unhappy with this fix; ideally we'd rename save_fl and
>> restore_fl to save_eflags_if and restore_eflags_if too.
>>
>
> I would be fine with this... but please document what the bloody
> semantics of pvops is actually supposed to be.
*cough*.
Lightly compile tested...
Subject: x86: rename save_fl/restore_fl paravirt ops to highlight eflags.
From: Rusty Russell <[email protected]>
As the comment in arch/x86/include/asm/paravirt_types.h says:
* Get/set interrupt state. save_fl and restore_fl are only
* expected to use X86_EFLAGS_IF; all other bits
* returned from save_fl are undefined, and may be ignored by
* restore_fl.
Signed-off-by: Rusty Russell <[email protected]>
diff --git a/arch/x86/include/asm/paravirt.h b/arch/x86/include/asm/paravirt.h
index 8957810ad7d1..5ad7b0e62a77 100644
--- a/arch/x86/include/asm/paravirt.h
+++ b/arch/x86/include/asm/paravirt.h
@@ -801,12 +801,12 @@ static __always_inline void __ticket_unlock_kick(struct arch_spinlock *lock,
static inline notrace unsigned long arch_local_save_flags(void)
{
- return PVOP_CALLEE0(unsigned long, pv_irq_ops.save_fl);
+ return PVOP_CALLEE0(unsigned long, pv_irq_ops.save_eflags_if);
}
static inline notrace void arch_local_irq_restore(unsigned long f)
{
- PVOP_VCALLEE1(pv_irq_ops.restore_fl, f);
+ PVOP_VCALLEE1(pv_irq_ops.restore_eflags_if, f);
}
static inline notrace void arch_local_irq_disable(void)
diff --git a/arch/x86/include/asm/paravirt_types.h b/arch/x86/include/asm/paravirt_types.h
index f7b0b5c112f2..05425c8544f1 100644
--- a/arch/x86/include/asm/paravirt_types.h
+++ b/arch/x86/include/asm/paravirt_types.h
@@ -204,8 +204,8 @@ struct pv_irq_ops {
* NOTE: These functions callers expect the callee to preserve
* more registers than the standard C calling convention.
*/
- struct paravirt_callee_save save_fl;
- struct paravirt_callee_save restore_fl;
+ struct paravirt_callee_save save_eflags_if;
+ struct paravirt_callee_save restore_eflags_if;
struct paravirt_callee_save irq_disable;
struct paravirt_callee_save irq_enable;
diff --git a/arch/x86/kernel/paravirt.c b/arch/x86/kernel/paravirt.c
index c614dd492f5f..d42e33b66ee6 100644
--- a/arch/x86/kernel/paravirt.c
+++ b/arch/x86/kernel/paravirt.c
@@ -321,8 +321,8 @@ struct pv_time_ops pv_time_ops = {
};
__visible struct pv_irq_ops pv_irq_ops = {
- .save_fl = __PV_IS_CALLEE_SAVE(native_save_fl),
- .restore_fl = __PV_IS_CALLEE_SAVE(native_restore_fl),
+ .save_eflags_if = __PV_IS_CALLEE_SAVE(native_save_fl),
+ .restore_eflags_if = __PV_IS_CALLEE_SAVE(native_restore_fl),
.irq_disable = __PV_IS_CALLEE_SAVE(native_irq_disable),
.irq_enable = __PV_IS_CALLEE_SAVE(native_irq_enable),
.safe_halt = native_safe_halt,
diff --git a/arch/x86/kernel/paravirt_patch_32.c b/arch/x86/kernel/paravirt_patch_32.c
index d9f32e6d6ab6..cf20c1b37f43 100644
--- a/arch/x86/kernel/paravirt_patch_32.c
+++ b/arch/x86/kernel/paravirt_patch_32.c
@@ -2,8 +2,8 @@
DEF_NATIVE(pv_irq_ops, irq_disable, "cli");
DEF_NATIVE(pv_irq_ops, irq_enable, "sti");
-DEF_NATIVE(pv_irq_ops, restore_fl, "push %eax; popf");
-DEF_NATIVE(pv_irq_ops, save_fl, "pushf; pop %eax");
+DEF_NATIVE(pv_irq_ops, restore_eflags_if, "push %eax; popf");
+DEF_NATIVE(pv_irq_ops, save_eflags_if, "pushf; pop %eax");
DEF_NATIVE(pv_cpu_ops, iret, "iret");
DEF_NATIVE(pv_cpu_ops, irq_enable_sysexit, "sti; sysexit");
DEF_NATIVE(pv_mmu_ops, read_cr2, "mov %cr2, %eax");
@@ -38,8 +38,8 @@ unsigned native_patch(u8 type, u16 clobbers, void *ibuf,
switch (type) {
PATCH_SITE(pv_irq_ops, irq_disable);
PATCH_SITE(pv_irq_ops, irq_enable);
- PATCH_SITE(pv_irq_ops, restore_fl);
- PATCH_SITE(pv_irq_ops, save_fl);
+ PATCH_SITE(pv_irq_ops, restore_eflags_if);
+ PATCH_SITE(pv_irq_ops, save_eflags_if);
PATCH_SITE(pv_cpu_ops, iret);
PATCH_SITE(pv_cpu_ops, irq_enable_sysexit);
PATCH_SITE(pv_mmu_ops, read_cr2);
diff --git a/arch/x86/kernel/paravirt_patch_64.c b/arch/x86/kernel/paravirt_patch_64.c
index a1da6737ba5b..24eaaa5f9aa6 100644
--- a/arch/x86/kernel/paravirt_patch_64.c
+++ b/arch/x86/kernel/paravirt_patch_64.c
@@ -4,8 +4,8 @@
DEF_NATIVE(pv_irq_ops, irq_disable, "cli");
DEF_NATIVE(pv_irq_ops, irq_enable, "sti");
-DEF_NATIVE(pv_irq_ops, restore_fl, "pushq %rdi; popfq");
-DEF_NATIVE(pv_irq_ops, save_fl, "pushfq; popq %rax");
+DEF_NATIVE(pv_irq_ops, restore_eflags_if, "pushq %rdi; popfq");
+DEF_NATIVE(pv_irq_ops, save_eflags_if, "pushfq; popq %rax");
DEF_NATIVE(pv_mmu_ops, read_cr2, "movq %cr2, %rax");
DEF_NATIVE(pv_mmu_ops, read_cr3, "movq %cr3, %rax");
DEF_NATIVE(pv_mmu_ops, write_cr3, "movq %rdi, %cr3");
@@ -45,8 +45,8 @@ unsigned native_patch(u8 type, u16 clobbers, void *ibuf,
end = end_##ops##_##x; \
goto patch_site
switch(type) {
- PATCH_SITE(pv_irq_ops, restore_fl);
- PATCH_SITE(pv_irq_ops, save_fl);
+ PATCH_SITE(pv_irq_ops, restore_eflags_if);
+ PATCH_SITE(pv_irq_ops, save_eflags_if);
PATCH_SITE(pv_irq_ops, irq_enable);
PATCH_SITE(pv_irq_ops, irq_disable);
PATCH_SITE(pv_cpu_ops, irq_enable_sysexit);
diff --git a/arch/x86/kernel/vsmp_64.c b/arch/x86/kernel/vsmp_64.c
index ee22c1d93ae5..c7f09f3fb31d 100644
--- a/arch/x86/kernel/vsmp_64.c
+++ b/arch/x86/kernel/vsmp_64.c
@@ -78,8 +78,8 @@ static unsigned __init_or_module vsmp_patch(u8 type, u16 clobbers, void *ibuf,
switch (type) {
case PARAVIRT_PATCH(pv_irq_ops.irq_enable):
case PARAVIRT_PATCH(pv_irq_ops.irq_disable):
- case PARAVIRT_PATCH(pv_irq_ops.save_fl):
- case PARAVIRT_PATCH(pv_irq_ops.restore_fl):
+ case PARAVIRT_PATCH(pv_irq_ops.save_eflags_if):
+ case PARAVIRT_PATCH(pv_irq_ops.restore_eflags_if):
return paravirt_patch_default(type, clobbers, ibuf, addr, len);
default:
return native_patch(type, clobbers, ibuf, addr, len);
@@ -119,8 +119,8 @@ static void __init set_vsmp_pv_ops(void)
/* Setup irq ops and turn on vSMP IRQ fastpath handling */
pv_irq_ops.irq_disable = PV_CALLEE_SAVE(vsmp_irq_disable);
pv_irq_ops.irq_enable = PV_CALLEE_SAVE(vsmp_irq_enable);
- pv_irq_ops.save_fl = PV_CALLEE_SAVE(vsmp_save_fl);
- pv_irq_ops.restore_fl = PV_CALLEE_SAVE(vsmp_restore_fl);
+ pv_irq_ops.save_eflags_if = PV_CALLEE_SAVE(vsmp_save_fl);
+ pv_irq_ops.restore_eflags_if = PV_CALLEE_SAVE(vsmp_restore_fl);
pv_init_ops.patch = vsmp_patch;
ctl &= ~(1 << 4);
}
diff --git a/arch/x86/lguest/boot.c b/arch/x86/lguest/boot.c
index 8f9a133cc099..5f35bf3ff0b0 100644
--- a/arch/x86/lguest/boot.c
+++ b/arch/x86/lguest/boot.c
@@ -1426,8 +1426,8 @@ __init void lguest_init(void)
*/
/* Interrupt-related operations */
- pv_irq_ops.save_fl = PV_CALLEE_SAVE(lguest_save_fl);
- pv_irq_ops.restore_fl = __PV_IS_CALLEE_SAVE(lg_restore_fl);
+ pv_irq_ops.save_eflags_if = PV_CALLEE_SAVE(lguest_save_fl);
+ pv_irq_ops.restore_eflags_if = __PV_IS_CALLEE_SAVE(lg_restore_fl);
pv_irq_ops.irq_disable = PV_CALLEE_SAVE(lguest_irq_disable);
pv_irq_ops.irq_enable = __PV_IS_CALLEE_SAVE(lg_irq_enable);
pv_irq_ops.safe_halt = lguest_safe_halt;
diff --git a/arch/x86/lguest/head_32.S b/arch/x86/lguest/head_32.S
index d5ae63f5ec5d..03b94d38fbd7 100644
--- a/arch/x86/lguest/head_32.S
+++ b/arch/x86/lguest/head_32.S
@@ -64,7 +64,7 @@ LGUEST_PATCH(pushf, movl lguest_data+LGUEST_DATA_irq_enabled, %eax)
/*G:033
* But using those wrappers is inefficient (we'll see why that doesn't matter
- * for save_fl and irq_disable later). If we write our routines carefully in
+ * for lguest_save_fl and irq_disable later). If we write our routines carefully in
* assembler, we can avoid clobbering any registers and avoid jumping through
* the wrapper functions.
*
diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index 46957ead3060..d9511df0d63c 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -1074,8 +1074,8 @@ void xen_setup_vcpu_info_placement(void)
* percpu area for all cpus, so make use of it. Note that for
* PVH we want to use native IRQ mechanism. */
if (have_vcpu_info_placement && !xen_pvh_domain()) {
- pv_irq_ops.save_fl = __PV_IS_CALLEE_SAVE(xen_save_fl_direct);
- pv_irq_ops.restore_fl = __PV_IS_CALLEE_SAVE(xen_restore_fl_direct);
+ pv_irq_ops.save_eflags_if = __PV_IS_CALLEE_SAVE(xen_save_fl_direct);
+ pv_irq_ops.restore_eflags_if = __PV_IS_CALLEE_SAVE(xen_restore_fl_direct);
pv_irq_ops.irq_disable = __PV_IS_CALLEE_SAVE(xen_irq_disable_direct);
pv_irq_ops.irq_enable = __PV_IS_CALLEE_SAVE(xen_irq_enable_direct);
pv_mmu_ops.read_cr2 = xen_read_cr2_direct;
@@ -1102,8 +1102,8 @@ static unsigned xen_patch(u8 type, u16 clobbers, void *insnbuf,
switch (type) {
SITE(pv_irq_ops, irq_enable);
SITE(pv_irq_ops, irq_disable);
- SITE(pv_irq_ops, save_fl);
- SITE(pv_irq_ops, restore_fl);
+ SITE(pv_irq_ops, save_eflags_if);
+ SITE(pv_irq_ops, restore_eflags_if);
#undef SITE
patch_site:
diff --git a/arch/x86/xen/irq.c b/arch/x86/xen/irq.c
index a1207cb6472a..a7f58c48c2e1 100644
--- a/arch/x86/xen/irq.c
+++ b/arch/x86/xen/irq.c
@@ -115,8 +115,8 @@ static void xen_halt(void)
}
static const struct pv_irq_ops xen_irq_ops __initconst = {
- .save_fl = PV_CALLEE_SAVE(xen_save_fl),
- .restore_fl = PV_CALLEE_SAVE(xen_restore_fl),
+ .save_eflags_if = PV_CALLEE_SAVE(xen_save_fl),
+ .restore_eflags_if = PV_CALLEE_SAVE(xen_restore_fl),
.irq_disable = PV_CALLEE_SAVE(xen_irq_disable),
.irq_enable = PV_CALLEE_SAVE(xen_irq_enable),
* Rusty Russell <[email protected]> wrote:
> As the comment in arch/x86/include/asm/paravirt_types.h says:
>
> * Get/set interrupt state. save_fl and restore_fl are only
> * expected to use X86_EFLAGS_IF; all other bits
> * returned from save_fl are undefined, and may be ignored by
> * restore_fl.
There should be no 'may' about this: under CONFIG_PARAVIRT_DEBUG=y the reminder of
the flags should be cleared (or all bits should be set: that will instantly crash
if restored verbatim), so that we trip up generic code that relies on non-IF
restoration?
Thanks,
Ingo
On 05/06/15 03:58, Rusty Russell wrote:
>
> Subject: x86: rename save_fl/restore_fl paravirt ops to highlight eflags.
> From: Rusty Russell <[email protected]>
>
> As the comment in arch/x86/include/asm/paravirt_types.h says:
>
> * Get/set interrupt state. save_fl and restore_fl are only
> * expected to use X86_EFLAGS_IF; all other bits
> * returned from save_fl are undefined, and may be ignored by
> * restore_fl.
>
> Signed-off-by: Rusty Russell <[email protected]>
[...]
> --- a/arch/x86/xen/enlighten.c
> +++ b/arch/x86/xen/enlighten.c
> @@ -1074,8 +1074,8 @@ void xen_setup_vcpu_info_placement(void)
> * percpu area for all cpus, so make use of it. Note that for
> * PVH we want to use native IRQ mechanism. */
> if (have_vcpu_info_placement && !xen_pvh_domain()) {
> - pv_irq_ops.save_fl = __PV_IS_CALLEE_SAVE(xen_save_fl_direct);
> - pv_irq_ops.restore_fl = __PV_IS_CALLEE_SAVE(xen_restore_fl_direct);
> + pv_irq_ops.save_eflags_if = __PV_IS_CALLEE_SAVE(xen_save_fl_direct);
> + pv_irq_ops.restore_eflags_if = __PV_IS_CALLEE_SAVE(xen_restore_fl_direct);
> pv_irq_ops.irq_disable = __PV_IS_CALLEE_SAVE(xen_irq_disable_direct);
> pv_irq_ops.irq_enable = __PV_IS_CALLEE_SAVE(xen_irq_enable_direct);
> pv_mmu_ops.read_cr2 = xen_read_cr2_direct;
> @@ -1102,8 +1102,8 @@ static unsigned xen_patch(u8 type, u16 clobbers, void *insnbuf,
> switch (type) {
> SITE(pv_irq_ops, irq_enable);
> SITE(pv_irq_ops, irq_disable);
> - SITE(pv_irq_ops, save_fl);
> - SITE(pv_irq_ops, restore_fl);
> + SITE(pv_irq_ops, save_eflags_if);
> + SITE(pv_irq_ops, restore_eflags_if);
> #undef SITE
>
> patch_site:
> diff --git a/arch/x86/xen/irq.c b/arch/x86/xen/irq.c
Acked-by: David Vrabel <[email protected]>
Thanks.
David
Ping?
None of the discussion on this thread altered the contents of this
patch, and the bug is still present.
~Andrew
On 03/06/15 10:31, Andrew Cooper wrote:
> There appears to be no formal statement of what pv_irq_ops.save_fl() is
> supposed to return precisely. Native returns the full flags, while lguest and
> Xen only return the Interrupt Flag, and both have comments by the
> implementations stating that only the Interrupt Flag is looked at. This may
> have been true when initially implemented, but no longer is.
>
> To make matters worse, the Xen PVOP leaves the upper bits undefined, making
> the BUG_ON() undefined behaviour. Experimentally, this now trips for 32bit PV
> guests on Broadwell hardware. The BUG_ON() is consistent for an individual
> build, but not consistent for all builds. It has also been a sitting timebomb
> since SMAP support was introduced.
>
> Use native_save_fl() instead, which will obtain an accurate view of the AC
> flag.
>
> Signed-off-by: Andrew Cooper <[email protected]>
> Reviewed-by: David Vrabel <[email protected]>
> Tested-by: Rusty Russell <[email protected]>
> CC: Thomas Gleixner <[email protected]>
> CC: Ingo Molnar <[email protected]>
> CC: H. Peter Anvin <[email protected]>
> CC: [email protected]
> CC: [email protected]
> CC: Konrad Rzeszutek Wilk <[email protected]>
> CC: Boris Ostrovsky <[email protected]>
> CC: xen-devel <[email protected]>
> CC: [email protected]
> CC: [email protected]
> ---
> arch/x86/kernel/cpu/common.c | 3 +--
> 1 file changed, 1 insertion(+), 2 deletions(-)
>
> diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
> index a62cf04..4f2fded 100644
> --- a/arch/x86/kernel/cpu/common.c
> +++ b/arch/x86/kernel/cpu/common.c
> @@ -291,10 +291,9 @@ __setup("nosmap", setup_disable_smap);
>
> static __always_inline void setup_smap(struct cpuinfo_x86 *c)
> {
> - unsigned long eflags;
> + unsigned long eflags = native_save_fl();
>
> /* This should have been cleared long ago */
> - raw_local_save_flags(eflags);
> BUG_ON(eflags & X86_EFLAGS_AC);
>
> if (cpu_has(c, X86_FEATURE_SMAP)) {
Commit-ID: 581b7f158fe0383b492acd1ce3fb4e99d4e57808
Gitweb: http://git.kernel.org/tip/581b7f158fe0383b492acd1ce3fb4e99d4e57808
Author: Andrew Cooper <[email protected]>
AuthorDate: Wed, 3 Jun 2015 10:31:14 +0100
Committer: Thomas Gleixner <[email protected]>
CommitDate: Thu, 19 Nov 2015 11:07:49 +0100
x86/cpu: Fix SMAP check in PVOPS environments
There appears to be no formal statement of what pv_irq_ops.save_fl() is
supposed to return precisely. Native returns the full flags, while lguest and
Xen only return the Interrupt Flag, and both have comments by the
implementations stating that only the Interrupt Flag is looked at. This may
have been true when initially implemented, but no longer is.
To make matters worse, the Xen PVOP leaves the upper bits undefined, making
the BUG_ON() undefined behaviour. Experimentally, this now trips for 32bit PV
guests on Broadwell hardware. The BUG_ON() is consistent for an individual
build, but not consistent for all builds. It has also been a sitting timebomb
since SMAP support was introduced.
Use native_save_fl() instead, which will obtain an accurate view of the AC
flag.
Signed-off-by: Andrew Cooper <[email protected]>
Reviewed-by: David Vrabel <[email protected]>
Tested-by: Rusty Russell <[email protected]>
Cc: Rusty Russell <[email protected]>
Cc: Konrad Rzeszutek Wilk <[email protected]>
Cc: Boris Ostrovsky <[email protected]>
Cc: <[email protected]>
Cc: Xen-devel <[email protected]>
CC: [email protected]
Link: http://lkml.kernel.org/r/[email protected]
Signed-off-by: Thomas Gleixner <[email protected]>
---
arch/x86/kernel/cpu/common.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index 4ddd780..c2b7522 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -273,10 +273,9 @@ __setup("nosmap", setup_disable_smap);
static __always_inline void setup_smap(struct cpuinfo_x86 *c)
{
- unsigned long eflags;
+ unsigned long eflags = native_save_fl();
/* This should have been cleared long ago */
- raw_local_save_flags(eflags);
BUG_ON(eflags & X86_EFLAGS_AC);
if (cpu_has(c, X86_FEATURE_SMAP)) {