Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1538633 > unrolled thread

[PATCH 0/2] Fix paravirt fail

Started byPeter Zijlstra <peterz@infradead.org>
First post2016-12-08 16:50 +0100
Last post2016-12-12 08:00 +0100
Articles 7 — 3 participants

Back to article view | Back to linux.kernel


Contents

  [PATCH 0/2] Fix paravirt fail Peter Zijlstra <peterz@infradead.org> - 2016-12-08 16:50 +0100
    [PATCH 1/2] x86,paravirt: Fix native_patch() Peter Zijlstra <peterz@infradead.org> - 2016-12-08 16:50 +0100
      [tip:locking/core] x86/paravirt: Fix native_patch() tip-bot for Peter Zijlstra <tipbot@zytor.com> - 2016-12-12 08:00 +0100
    [PATCH 2/2] x86, paravirt: Fix bool return type for PVOP_CALL Peter Zijlstra <peterz@infradead.org> - 2016-12-08 16:50 +0100
      Re: [PATCH 2/2] x86, paravirt: Fix bool return type for PVOP_CALL Pan Xinhui <xinhui@linux.vnet.ibm.com> - 2016-12-08 17:50 +0100
        Re: [PATCH 2/2] x86, paravirt: Fix bool return type for PVOP_CALL Peter Zijlstra <peterz@infradead.org> - 2016-12-08 18:00 +0100
      [tip:locking/core] x86/paravirt: Fix bool return type for  PVOP_CALL() tip-bot for Peter Zijlstra <tipbot@zytor.com> - 2016-12-12 08:00 +0100

#1538633 — [PATCH 0/2] Fix paravirt fail

FromPeter Zijlstra <peterz@infradead.org>
Date2016-12-08 16:50 +0100
Subject[PATCH 0/2] Fix paravirt fail
Message-ID<sM97k-7Se-15@gated-at.bofh.it>
Two patches that cure fallout from commit:

  3cded4179481 ("x86/paravirt: Optimize native pv_lock_ops.vcpu_is_preempted()")

[toc] | [next] | [standalone]


#1538637 — [PATCH 1/2] x86,paravirt: Fix native_patch()

FromPeter Zijlstra <peterz@infradead.org>
Date2016-12-08 16:50 +0100
Subject[PATCH 1/2] x86,paravirt: Fix native_patch()
Message-ID<sM97k-7Se-51@gated-at.bofh.it>
In reply to#1538633
While chasing a regression I noticed we potentially patch the wrong
code in native_patch().

If we do not select the native code sequence, we must use the default
patcher, not fall-through the switch case.

Fixes: 3cded4179481 ("x86/paravirt: Optimize native pv_lock_ops.vcpu_is_preempted()")
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
---
 arch/x86/kernel/paravirt_patch_32.c |    4 ++++
 arch/x86/kernel/paravirt_patch_64.c |    4 ++++
 2 files changed, 8 insertions(+)

--- a/arch/x86/kernel/paravirt_patch_32.c
+++ b/arch/x86/kernel/paravirt_patch_32.c
@@ -56,15 +56,19 @@ unsigned native_patch(u8 type, u16 clobb
 				end   = end_pv_lock_ops_queued_spin_unlock;
 				goto patch_site;
 			}
+			goto patch_default;
+
 		case PARAVIRT_PATCH(pv_lock_ops.vcpu_is_preempted):
 			if (pv_is_native_vcpu_is_preempted()) {
 				start = start_pv_lock_ops_vcpu_is_preempted;
 				end   = end_pv_lock_ops_vcpu_is_preempted;
 				goto patch_site;
 			}
+			goto patch_default;
 #endif
 
 	default:
+patch_default:
 		ret = paravirt_patch_default(type, clobbers, ibuf, addr, len);
 		break;
 
--- a/arch/x86/kernel/paravirt_patch_64.c
+++ b/arch/x86/kernel/paravirt_patch_64.c
@@ -68,15 +68,19 @@ unsigned native_patch(u8 type, u16 clobb
 				end   = end_pv_lock_ops_queued_spin_unlock;
 				goto patch_site;
 			}
+			goto patch_default;
+
 		case PARAVIRT_PATCH(pv_lock_ops.vcpu_is_preempted):
 			if (pv_is_native_vcpu_is_preempted()) {
 				start = start_pv_lock_ops_vcpu_is_preempted;
 				end   = end_pv_lock_ops_vcpu_is_preempted;
 				goto patch_site;
 			}
+			goto patch_default;
 #endif
 
 	default:
+patch_default:
 		ret = paravirt_patch_default(type, clobbers, ibuf, addr, len);
 		break;
 

[toc] | [prev] | [next] | [standalone]


#1540144 — [tip:locking/core] x86/paravirt: Fix native_patch()

Fromtip-bot for Peter Zijlstra <tipbot@zytor.com>
Date2016-12-12 08:00 +0100
Subject[tip:locking/core] x86/paravirt: Fix native_patch()
Message-ID<sNsKB-11W-3@gated-at.bofh.it>
In reply to#1538637
Commit-ID:  45dbea5f55c05980cbb4c30047c71a820cd3f282
Gitweb:     http://git.kernel.org/tip/45dbea5f55c05980cbb4c30047c71a820cd3f282
Author:     Peter Zijlstra <peterz@infradead.org>
AuthorDate: Thu, 8 Dec 2016 16:42:14 +0100
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Sun, 11 Dec 2016 13:09:19 +0100

x86/paravirt: Fix native_patch()

While chasing a regression I noticed we potentially patch the wrong
code in native_patch().

If we do not select the native code sequence, we must use the default
patcher, not fall-through the switch case.

Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Alok Kataria <akataria@vmware.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Chris Wright <chrisw@sous-sol.org>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Pan Xinhui <xinhui.pan@linux.vnet.ibm.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Anvin <hpa@zytor.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kernel test robot <xiaolong.ye@intel.com>
Fixes: 3cded4179481 ("x86/paravirt: Optimize native pv_lock_ops.vcpu_is_preempted()")
Link: http://lkml.kernel.org/r/20161208154349.270616999@infradead.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/kernel/paravirt_patch_32.c | 4 ++++
 arch/x86/kernel/paravirt_patch_64.c | 4 ++++
 2 files changed, 8 insertions(+)

diff --git a/arch/x86/kernel/paravirt_patch_32.c b/arch/x86/kernel/paravirt_patch_32.c
index ff03dbd..33cdec2 100644
--- a/arch/x86/kernel/paravirt_patch_32.c
+++ b/arch/x86/kernel/paravirt_patch_32.c
@@ -58,15 +58,19 @@ unsigned native_patch(u8 type, u16 clobbers, void *ibuf,
 				end   = end_pv_lock_ops_queued_spin_unlock;
 				goto patch_site;
 			}
+			goto patch_default;
+
 		case PARAVIRT_PATCH(pv_lock_ops.vcpu_is_preempted):
 			if (pv_is_native_vcpu_is_preempted()) {
 				start = start_pv_lock_ops_vcpu_is_preempted;
 				end   = end_pv_lock_ops_vcpu_is_preempted;
 				goto patch_site;
 			}
+			goto patch_default;
 #endif
 
 	default:
+patch_default:
 		ret = paravirt_patch_default(type, clobbers, ibuf, addr, len);
 		break;
 
diff --git a/arch/x86/kernel/paravirt_patch_64.c b/arch/x86/kernel/paravirt_patch_64.c
index e61dd97..b0fceff 100644
--- a/arch/x86/kernel/paravirt_patch_64.c
+++ b/arch/x86/kernel/paravirt_patch_64.c
@@ -70,15 +70,19 @@ unsigned native_patch(u8 type, u16 clobbers, void *ibuf,
 				end   = end_pv_lock_ops_queued_spin_unlock;
 				goto patch_site;
 			}
+			goto patch_default;
+
 		case PARAVIRT_PATCH(pv_lock_ops.vcpu_is_preempted):
 			if (pv_is_native_vcpu_is_preempted()) {
 				start = start_pv_lock_ops_vcpu_is_preempted;
 				end   = end_pv_lock_ops_vcpu_is_preempted;
 				goto patch_site;
 			}
+			goto patch_default;
 #endif
 
 	default:
+patch_default:
 		ret = paravirt_patch_default(type, clobbers, ibuf, addr, len);
 		break;
 

[toc] | [prev] | [next] | [standalone]


#1538650 — [PATCH 2/2] x86, paravirt: Fix bool return type for PVOP_CALL

FromPeter Zijlstra <peterz@infradead.org>
Date2016-12-08 16:50 +0100
Subject[PATCH 2/2] x86, paravirt: Fix bool return type for PVOP_CALL
Message-ID<sM97l-7Se-75@gated-at.bofh.it>
In reply to#1538633
Commit 3cded4179481 ("x86/paravirt: Optimize native
pv_lock_ops.vcpu_is_preempted()") introduced a paravirt op with bool
return type [*]

It turns out that the PVOP_CALL*() macros miscompile when rettype is
bool. Code that looked like:

   83 ef 01                sub    $0x1,%edi
   ff 15 32 a0 d8 00       callq  *0xd8a032(%rip)        # ffffffff81e28120 <pv_lock_ops+0x20>
   84 c0                   test   %al,%al

ended up looking like so after PVOP_CALL1() was applied:

   83 ef 01                sub    $0x1,%edi
   48 63 ff                movslq %edi,%rdi
   ff 14 25 20 81 e2 81    callq  *0xffffffff81e28120
   48 85 c0                test   %rax,%rax

Note how it tests the whole of %rax, even though a typical bool return
function only sets %al, like:

  0f 95 c0                setne  %al
  c3                      retq

This is because ____PVOP_CALL() does:

		__ret = (rettype)__eax;

and while regular integer type casts truncate the result, a cast to
bool tests for any !0 value. Fix this by explicitly truncating to
sizeof(rettype) before casting.


[*] The actual bug should've been exposed in commit 446f3dc8cc0a
("locking/core, x86/paravirt: Implement vcpu_is_preempted(cpu) for KVM
and Xen guests") but that didn't properly implement the paravirt call.

Cc: Peter Anvin <hpa@zytor.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Chris Wright <chrisw@sous-sol.org>
Cc: Alok Kataria <akataria@vmware.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: virtualization@lists.linux-foundation.org
Cc: Pan Xinhui <xinhui.pan@linux.vnet.ibm.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Fixes: 3cded4179481 ("x86/paravirt: Optimize native pv_lock_ops.vcpu_is_preempted()")
Reported-by: kernel test robot <xiaolong.ye@intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
---
 arch/x86/include/asm/paravirt_types.h |   14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

--- a/arch/x86/include/asm/paravirt_types.h
+++ b/arch/x86/include/asm/paravirt_types.h
@@ -508,6 +508,18 @@ int paravirt_disable_iospace(void);
 #define PVOP_TEST_NULL(op)	((void)op)
 #endif
 
+#define PVOP_RETMASK(rettype)						\
+	({	unsigned long __mask = ~0UL;				\
+		switch (sizeof(rettype)) {				\
+		case 1: __mask =       0xffUL; break;			\
+		case 2: __mask =     0xffffUL; break;			\
+		case 4: __mask = 0xffffffffUL; break;			\
+		default: break;						\
+		}							\
+		__mask;							\
+	})
+
+
 #define ____PVOP_CALL(rettype, op, clbr, call_clbr, extra_clbr,		\
 		      pre, post, ...)					\
 	({								\
@@ -535,7 +547,7 @@ int paravirt_disable_iospace(void);
 				       paravirt_clobber(clbr),		\
 				       ##__VA_ARGS__			\
 				     : "memory", "cc" extra_clbr);	\
-			__ret = (rettype)__eax;				\
+			__ret = (rettype)(__eax & PVOP_RETMASK(rettype));	\
 		}							\
 		__ret;							\
 	})

[toc] | [prev] | [next] | [standalone]


#1538724 — Re: [PATCH 2/2] x86, paravirt: Fix bool return type for PVOP_CALL

FromPan Xinhui <xinhui@linux.vnet.ibm.com>
Date2016-12-08 17:50 +0100
SubjectRe: [PATCH 2/2] x86, paravirt: Fix bool return type for PVOP_CALL
Message-ID<sMa3n-8rP-21@gated-at.bofh.it>
In reply to#1538650
hi, Peter
	I think I know the point.

then could we just let __eax rettype(here is bool), not unsigned long?
I does not do tests for my thoughts.

@@ -461,7 +461,9 @@ int paravirt_disable_iospace(void);
  #define PVOP_VCALL_ARGS                                                        \
         unsigned long __eax = __eax, __edx = __edx, __ecx = __ecx;      \
         register void *__sp asm("esp")
-#define PVOP_CALL_ARGS                 PVOP_VCALL_ARGS
+#define PVOP_CALL_ARGS 					\
+       rettype __eax = __eax, __edx = __edx, __ecx = __ecx;    \
+       register void *__sp asm("esp")

[toc] | [prev] | [next] | [standalone]


#1538731 — Re: [PATCH 2/2] x86, paravirt: Fix bool return type for PVOP_CALL

FromPeter Zijlstra <peterz@infradead.org>
Date2016-12-08 18:00 +0100
SubjectRe: [PATCH 2/2] x86, paravirt: Fix bool return type for PVOP_CALL
Message-ID<sMad4-8vg-15@gated-at.bofh.it>
In reply to#1538724
On Fri, Dec 09, 2016 at 12:40:35AM +0800, Pan Xinhui wrote:
> 
> hi, Peter
> 	I think I know the point.
> 
> then could we just let __eax rettype(here is bool), not unsigned long?
> I does not do tests for my thoughts.
> 
> @@ -461,7 +461,9 @@ int paravirt_disable_iospace(void);
>  #define PVOP_VCALL_ARGS                                                        \
>         unsigned long __eax = __eax, __edx = __edx, __ecx = __ecx;      \
>         register void *__sp asm("esp")
> -#define PVOP_CALL_ARGS                 PVOP_VCALL_ARGS
> +#define PVOP_CALL_ARGS 					\
> +       rettype __eax = __eax, __edx = __edx, __ecx = __ecx;    \
> +       register void *__sp asm("esp")

Doesn't work on i386 where eax is also an argument register.

[toc] | [prev] | [next] | [standalone]


#1540145 — [tip:locking/core] x86/paravirt: Fix bool return type for PVOP_CALL()

Fromtip-bot for Peter Zijlstra <tipbot@zytor.com>
Date2016-12-12 08:00 +0100
Subject[tip:locking/core] x86/paravirt: Fix bool return type for PVOP_CALL()
Message-ID<sNsKB-11W-7@gated-at.bofh.it>
In reply to#1538650
Commit-ID:  11f254dbb3a2e3f0d8552d0dd37f4faa432b6b16
Gitweb:     http://git.kernel.org/tip/11f254dbb3a2e3f0d8552d0dd37f4faa432b6b16
Author:     Peter Zijlstra <peterz@infradead.org>
AuthorDate: Thu, 8 Dec 2016 16:42:15 +0100
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Sun, 11 Dec 2016 13:09:20 +0100

x86/paravirt: Fix bool return type for PVOP_CALL()

Commit:

  3cded4179481 ("x86/paravirt: Optimize native pv_lock_ops.vcpu_is_preempted()")

introduced a paravirt op with bool return type [*]

It turns out that the PVOP_CALL*() macros miscompile when rettype is
bool. Code that looked like:

   83 ef 01                sub    $0x1,%edi
   ff 15 32 a0 d8 00       callq  *0xd8a032(%rip)        # ffffffff81e28120 <pv_lock_ops+0x20>
   84 c0                   test   %al,%al

ended up looking like so after PVOP_CALL1() was applied:

   83 ef 01                sub    $0x1,%edi
   48 63 ff                movslq %edi,%rdi
   ff 14 25 20 81 e2 81    callq  *0xffffffff81e28120
   48 85 c0                test   %rax,%rax

Note how it tests the whole of %rax, even though a typical bool return
function only sets %al, like:

  0f 95 c0                setne  %al
  c3                      retq

This is because ____PVOP_CALL() does:

		__ret = (rettype)__eax;

and while regular integer type casts truncate the result, a cast to
bool tests for any !0 value. Fix this by explicitly truncating to
sizeof(rettype) before casting.

[*] The actual bug should've been exposed in commit:
      446f3dc8cc0a ("locking/core, x86/paravirt: Implement vcpu_is_preempted(cpu) for KVM and Xen guests")
    but that didn't properly implement the paravirt call.

Reported-by: kernel test robot <xiaolong.ye@intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Alok Kataria <akataria@vmware.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Chris Wright <chrisw@sous-sol.org>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Pan Xinhui <xinhui.pan@linux.vnet.ibm.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Anvin <hpa@zytor.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Thomas Gleixner <tglx@linutronix.de>
Fixes: 3cded4179481 ("x86/paravirt: Optimize native pv_lock_ops.vcpu_is_preempted()")
Link: http://lkml.kernel.org/r/20161208154349.346057680@infradead.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/include/asm/paravirt_types.h | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/paravirt_types.h b/arch/x86/include/asm/paravirt_types.h
index 2614bd7..3f2bc0f 100644
--- a/arch/x86/include/asm/paravirt_types.h
+++ b/arch/x86/include/asm/paravirt_types.h
@@ -510,6 +510,18 @@ int paravirt_disable_iospace(void);
 #define PVOP_TEST_NULL(op)	((void)op)
 #endif
 
+#define PVOP_RETMASK(rettype)						\
+	({	unsigned long __mask = ~0UL;				\
+		switch (sizeof(rettype)) {				\
+		case 1: __mask =       0xffUL; break;			\
+		case 2: __mask =     0xffffUL; break;			\
+		case 4: __mask = 0xffffffffUL; break;			\
+		default: break;						\
+		}							\
+		__mask;							\
+	})
+
+
 #define ____PVOP_CALL(rettype, op, clbr, call_clbr, extra_clbr,		\
 		      pre, post, ...)					\
 	({								\
@@ -537,7 +549,7 @@ int paravirt_disable_iospace(void);
 				       paravirt_clobber(clbr),		\
 				       ##__VA_ARGS__			\
 				     : "memory", "cc" extra_clbr);	\
-			__ret = (rettype)__eax;				\
+			__ret = (rettype)(__eax & PVOP_RETMASK(rettype));	\
 		}							\
 		__ret;							\
 	})

[toc] | [prev] | [standalone]


Back to top | Article view | linux.kernel


csiph-web