Re: [PATCH] x86/vdso: Fix vDSO build if a retpoline is emitted

2018-08-16 Thread Matthew Rickard 




On 17/08/2018 5:41 AM, Andy Lutomirski wrote:

Currently, if the vDSO ends up containing an indirect branch or
call, GCC will emit the "external thunk" style of retpoline, and it
will fail to link.

Fix it by building the vDSO with inline retpoline thunks.

I haven't seen any reports of this triggering on an unpatched
kernel.

Cc: Matt Rickard 
Cc: Jason Vas Dias 
Cc: David Woodhouse 
Cc: Peter Zijlstra 
Cc: Andi Kleen 
Fixes: commit 76b043848fd2 ("x86/retpoline: Add initial retpoline support")
Signed-off-by: Andy Lutomirski 
---

This is a bit of a marginal candidate for the merge window.  There are
a couple of reports of people hitting this while hacking on the vDSO,
but I haven't seen any report of anyone hitting this on a stock kernel.

Feel free to tell me to buzz of and send it later :)

  Makefile | 4 
  arch/x86/entry/vdso/Makefile | 6 --
  2 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/Makefile b/Makefile
index a0650bf79606..7bab2e90e4e1 100644
--- a/Makefile
+++ b/Makefile
@@ -507,9 +507,13 @@ KBUILD_AFLAGS += $(call cc-option, -no-integrated-as)
  endif
  
  RETPOLINE_CFLAGS_GCC := -mindirect-branch=thunk-extern -mindirect-branch-register

+RETPOLINE_VDSO_CFLAGS_GCC := -mindirect-branch=thunk-inline 
-mindirect-branch-register
  RETPOLINE_CFLAGS_CLANG := -mretpoline-external-thunk
+RETPOLINE_VDSO_CFLAGS_CLANG := -mretpoline
  RETPOLINE_CFLAGS := $(call cc-option,$(RETPOLINE_CFLAGS_GCC),$(call 
cc-option,$(RETPOLINE_CFLAGS_CLANG)))
+RETPOLINE_VDSO_CFLAGS := $(call cc-option,$(RETPOLINE_VDSO_CFLAGS_GCC),$(call 
cc-option,$(RETPOLINE_VDSO_CFLAGS_CLANG)))
  export RETPOLINE_CFLAGS
+export RETPOLINE_VDSO_CFLAGS
  
  KBUILD_CFLAGS	+= $(call cc-option,-fno-PIE)

  KBUILD_AFLAGS += $(call cc-option,-fno-PIE)
diff --git a/arch/x86/entry/vdso/Makefile b/arch/x86/entry/vdso/Makefile
index 9f695f517747..fa3f439f0a92 100644
--- a/arch/x86/entry/vdso/Makefile
+++ b/arch/x86/entry/vdso/Makefile
@@ -68,9 +68,9 @@ $(obj)/vdso-image-%.c: $(obj)/vdso%.so.dbg $(obj)/vdso%.so 
$(obj)/vdso2c FORCE
  CFL := $(PROFILING) -mcmodel=small -fPIC -O2 -fasynchronous-unwind-tables 
-m64 \
 $(filter -g%,$(KBUILD_CFLAGS)) $(call cc-option, -fno-stack-protector) 
\
 -fno-omit-frame-pointer -foptimize-sibling-calls \
-   -DDISABLE_BRANCH_PROFILING -DBUILD_VDSO
+   -DDISABLE_BRANCH_PROFILING -DBUILD_VDSO $(RETPOLINE_VDSO_CFLAGS)
  
-$(vobjs): KBUILD_CFLAGS := $(filter-out $(GCC_PLUGINS_CFLAGS),$(KBUILD_CFLAGS)) $(CFL)

+$(vobjs): KBUILD_CFLAGS := $(filter-out $(GCC_PLUGINS_CFLAGS) 
$(RETPOLINE_CFLAGS),$(KBUILD_CFLAGS)) $(CFL)
  
  #

  # vDSO code runs in userspace and -pg doesn't help with profiling anyway.
@@ -132,11 +132,13 @@ KBUILD_CFLAGS_32 := $(filter-out 
-mcmodel=kernel,$(KBUILD_CFLAGS_32))
  KBUILD_CFLAGS_32 := $(filter-out -fno-pic,$(KBUILD_CFLAGS_32))
  KBUILD_CFLAGS_32 := $(filter-out -mfentry,$(KBUILD_CFLAGS_32))
  KBUILD_CFLAGS_32 := $(filter-out $(GCC_PLUGINS_CFLAGS),$(KBUILD_CFLAGS_32))
+KBUILD_CFLAGS_32 := $(filter-out $(RETPOLINE_CFLAGS),$(KBUILD_CFLAGS_32))
  KBUILD_CFLAGS_32 += -m32 -msoft-float -mregparm=0 -fpic
  KBUILD_CFLAGS_32 += $(call cc-option, -fno-stack-protector)
  KBUILD_CFLAGS_32 += $(call cc-option, -foptimize-sibling-calls)
  KBUILD_CFLAGS_32 += -fno-omit-frame-pointer
  KBUILD_CFLAGS_32 += -DDISABLE_BRANCH_PROFILING
+KBUILD_CFLAGS_32 += $(RETPOLINE_VDSO_CFLAGS)
  $(obj)/vdso32.so.dbg: KBUILD_CFLAGS = $(KBUILD_CFLAGS_32)
  
  $(obj)/vdso32.so.dbg: FORCE \




That solves my issue, thanks.
Acked-by: Matt Rickard

Re: [PATCH] x86/vdso: Fix vDSO build if a retpoline is emitted

2018-08-16 Thread Matthew Rickard 




On 17/08/2018 5:41 AM, Andy Lutomirski wrote:

Currently, if the vDSO ends up containing an indirect branch or
call, GCC will emit the "external thunk" style of retpoline, and it
will fail to link.

Fix it by building the vDSO with inline retpoline thunks.

I haven't seen any reports of this triggering on an unpatched
kernel.

Cc: Matt Rickard 
Cc: Jason Vas Dias 
Cc: David Woodhouse 
Cc: Peter Zijlstra 
Cc: Andi Kleen 
Fixes: commit 76b043848fd2 ("x86/retpoline: Add initial retpoline support")
Signed-off-by: Andy Lutomirski 
---

This is a bit of a marginal candidate for the merge window.  There are
a couple of reports of people hitting this while hacking on the vDSO,
but I haven't seen any report of anyone hitting this on a stock kernel.

Feel free to tell me to buzz of and send it later :)

  Makefile | 4 
  arch/x86/entry/vdso/Makefile | 6 --
  2 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/Makefile b/Makefile
index a0650bf79606..7bab2e90e4e1 100644
--- a/Makefile
+++ b/Makefile
@@ -507,9 +507,13 @@ KBUILD_AFLAGS += $(call cc-option, -no-integrated-as)
  endif
  
  RETPOLINE_CFLAGS_GCC := -mindirect-branch=thunk-extern -mindirect-branch-register

+RETPOLINE_VDSO_CFLAGS_GCC := -mindirect-branch=thunk-inline 
-mindirect-branch-register
  RETPOLINE_CFLAGS_CLANG := -mretpoline-external-thunk
+RETPOLINE_VDSO_CFLAGS_CLANG := -mretpoline
  RETPOLINE_CFLAGS := $(call cc-option,$(RETPOLINE_CFLAGS_GCC),$(call 
cc-option,$(RETPOLINE_CFLAGS_CLANG)))
+RETPOLINE_VDSO_CFLAGS := $(call cc-option,$(RETPOLINE_VDSO_CFLAGS_GCC),$(call 
cc-option,$(RETPOLINE_VDSO_CFLAGS_CLANG)))
  export RETPOLINE_CFLAGS
+export RETPOLINE_VDSO_CFLAGS
  
  KBUILD_CFLAGS	+= $(call cc-option,-fno-PIE)

  KBUILD_AFLAGS += $(call cc-option,-fno-PIE)
diff --git a/arch/x86/entry/vdso/Makefile b/arch/x86/entry/vdso/Makefile
index 9f695f517747..fa3f439f0a92 100644
--- a/arch/x86/entry/vdso/Makefile
+++ b/arch/x86/entry/vdso/Makefile
@@ -68,9 +68,9 @@ $(obj)/vdso-image-%.c: $(obj)/vdso%.so.dbg $(obj)/vdso%.so 
$(obj)/vdso2c FORCE
  CFL := $(PROFILING) -mcmodel=small -fPIC -O2 -fasynchronous-unwind-tables 
-m64 \
 $(filter -g%,$(KBUILD_CFLAGS)) $(call cc-option, -fno-stack-protector) 
\
 -fno-omit-frame-pointer -foptimize-sibling-calls \
-   -DDISABLE_BRANCH_PROFILING -DBUILD_VDSO
+   -DDISABLE_BRANCH_PROFILING -DBUILD_VDSO $(RETPOLINE_VDSO_CFLAGS)
  
-$(vobjs): KBUILD_CFLAGS := $(filter-out $(GCC_PLUGINS_CFLAGS),$(KBUILD_CFLAGS)) $(CFL)

+$(vobjs): KBUILD_CFLAGS := $(filter-out $(GCC_PLUGINS_CFLAGS) 
$(RETPOLINE_CFLAGS),$(KBUILD_CFLAGS)) $(CFL)
  
  #

  # vDSO code runs in userspace and -pg doesn't help with profiling anyway.
@@ -132,11 +132,13 @@ KBUILD_CFLAGS_32 := $(filter-out 
-mcmodel=kernel,$(KBUILD_CFLAGS_32))
  KBUILD_CFLAGS_32 := $(filter-out -fno-pic,$(KBUILD_CFLAGS_32))
  KBUILD_CFLAGS_32 := $(filter-out -mfentry,$(KBUILD_CFLAGS_32))
  KBUILD_CFLAGS_32 := $(filter-out $(GCC_PLUGINS_CFLAGS),$(KBUILD_CFLAGS_32))
+KBUILD_CFLAGS_32 := $(filter-out $(RETPOLINE_CFLAGS),$(KBUILD_CFLAGS_32))
  KBUILD_CFLAGS_32 += -m32 -msoft-float -mregparm=0 -fpic
  KBUILD_CFLAGS_32 += $(call cc-option, -fno-stack-protector)
  KBUILD_CFLAGS_32 += $(call cc-option, -foptimize-sibling-calls)
  KBUILD_CFLAGS_32 += -fno-omit-frame-pointer
  KBUILD_CFLAGS_32 += -DDISABLE_BRANCH_PROFILING
+KBUILD_CFLAGS_32 += $(RETPOLINE_VDSO_CFLAGS)
  $(obj)/vdso32.so.dbg: KBUILD_CFLAGS = $(KBUILD_CFLAGS_32)
  
  $(obj)/vdso32.so.dbg: FORCE \




That solves my issue, thanks.
Acked-by: Matt Rickard

Re: [PATCH] x86/vdso: Fix vDSO build if a retpoline is emitted

2018-08-16 Thread David Woodhouse 


On Thu, 2018-08-16 at 12:41 -0700, Andy Lutomirski wrote:
> Currently, if the vDSO ends up containing an indirect branch or
> call, GCC will emit the "external thunk" style of retpoline, and it
> will fail to link.
> 
> Fix it by building the vDSO with inline retpoline thunks.
> 
> I haven't seen any reports of this triggering on an unpatched
> kernel.
> 
> Cc: Matt Rickard 
> Cc: Jason Vas Dias 
> Cc: David Woodhouse 
> Cc: Peter Zijlstra 
> Cc: Andi Kleen 
> Fixes: commit 76b043848fd2 ("x86/retpoline: Add initial retpoline support")
> Signed-off-by: Andy Lutomirski 

Acked-by: David Woodhouse 

Thanks.

smime.p7s
Description: S/MIME cryptographic signature

Re: [PATCH] x86/vdso: Fix vDSO build if a retpoline is emitted

2018-08-16 Thread David Woodhouse 


On Thu, 2018-08-16 at 12:41 -0700, Andy Lutomirski wrote:
> Currently, if the vDSO ends up containing an indirect branch or
> call, GCC will emit the "external thunk" style of retpoline, and it
> will fail to link.
> 
> Fix it by building the vDSO with inline retpoline thunks.
> 
> I haven't seen any reports of this triggering on an unpatched
> kernel.
> 
> Cc: Matt Rickard 
> Cc: Jason Vas Dias 
> Cc: David Woodhouse 
> Cc: Peter Zijlstra 
> Cc: Andi Kleen 
> Fixes: commit 76b043848fd2 ("x86/retpoline: Add initial retpoline support")
> Signed-off-by: Andy Lutomirski 

Acked-by: David Woodhouse 

Thanks.

smime.p7s
Description: S/MIME cryptographic signature