linux-kernel - Re: [PATCH 03/11] objtool: Convert ANNOTATE_RETPOLINE

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20240913204711.2041299-1-jthoughton@google.com>
Date: Fri, 13 Sep 2024 20:47:11 +0000
From: James Houghton <jthoughton@...gle.com>
To: peterz@...radead.org
Cc: seanjc@...gle.com, jpoimboe@...hat.com, kvm@...r.kernel.org, 
	linux-kernel@...r.kernel.org, pbonzini@...hat.com, tglx@...utronix.de, 
	x86@...nel.org
Subject: Re: [PATCH 03/11] objtool: Convert ANNOTATE_RETPOLINE_SAFE to ANNOTATE

Hi Peter,

On Wed, 6 Dec 2023, Sean Christopherson wrote:
> On Mon, Dec 04, 2023, Peter Zijlstra wrote:
> > 
> > --- a/arch/x86/include/asm/nospec-branch.h
> > +++ b/arch/x86/include/asm/nospec-branch.h
> > @@ -193,12 +193,7 @@
> >   * objtool the subsequent indirect jump/call is vouched safe for retpoline
> >   * builds.
> >   */
> > -.macro ANNOTATE_RETPOLINE_SAFE
> > -.Lhere_\@:
> > -	.pushsection .discard.retpoline_safe
> > -	.long .Lhere_\@
> > -	.popsection
> > -.endm
> > +#define ANNOTATE_RETPOLINE_SAFE	ANNOTATE type=ANNOTYPE_RETPOLINE_SAFE
> >  
> >  /*
> >   * (ab)use RETPOLINE_SAFE on RET to annotate away 'bare' RET instructions
> > @@ -317,11 +312,7 @@
> >  
> >  #else /* __ASSEMBLY__ */
> >  
> > -#define ANNOTATE_RETPOLINE_SAFE					\
> > -	"999:\n\t"						\
> > -	".pushsection .discard.retpoline_safe\n\t"		\
> > -	".long 999b\n\t"					\
> > -	".popsection\n\t"
> > +#define ANNOTATE_RETPOLINE_SAFE ASM_ANNOTATE(ANNOTYPE_RETPOLINE_SAFE)
> 
> This fails for some of my builds that end up with CONFIG_OBJTOOl=n.  Adding a
> stub for ASM_ANNOTATE() gets me past that:
> 
> @@ -156,6 +171,7 @@
>  #define STACK_FRAME_NON_STANDARD(func)
>  #define STACK_FRAME_NON_STANDARD_FP(func)
>  #define ANNOTATE_NOENDBR
> +#define ASM_ANNOTATE(x)
>  #define ASM_REACHABLE
>  #else
>  #define ANNOTATE_INTRA_FUNCTION_CALL
> 
> but then I run into other issues:
> 
> arch/x86/kernel/relocate_kernel_32.S: Assembler messages:
> arch/x86/kernel/relocate_kernel_32.S:96: Error: Parameter named `type' does not exist for macro `annotate'
> arch/x86/kernel/relocate_kernel_32.S:166: Error: Parameter named `type' does not exist for macro `annotate'
> arch/x86/kernel/relocate_kernel_32.S:174: Error: Parameter named `type' does not exist for macro `annotate'
> arch/x86/kernel/relocate_kernel_32.S:200: Error: Parameter named `type' does not exist for macro `annotate'
> arch/x86/kernel/relocate_kernel_32.S:220: Error: Parameter named `type' does not exist for macro `annotate'
> arch/x86/kernel/relocate_kernel_32.S:285: Error: Parameter named `type' does not exist for macro `annotate'

Sean pointed me at this series recently. It seems like these compile errors
(and some others) go away with the following diff:

diff --git a/arch/x86/include/asm/nospec-branch.h b/arch/x86/include/asm/nospec-branch.h
index 0bebdcad7ba1..036ab199859a 100644
--- a/arch/x86/include/asm/nospec-branch.h
+++ b/arch/x86/include/asm/nospec-branch.h
@@ -489,7 +489,7 @@ static inline void call_depth_return_thunk(void) {}
 	"       .align 16\n"					\
 	"903:	lea    4(%%esp), %%esp;\n"			\
 	"       pushl  %[thunk_target];\n"			\
-	"       ret;\n"						\
+	"       ret;\n",					\
 	X86_FEATURE_RETPOLINE,					\
 	"lfence;\n"						\
 	ANNOTATE_RETPOLINE_SAFE					\
diff --git a/include/linux/objtool.h b/include/linux/objtool.h
index f6f80bfefe3b..e811b2ff3a9c 100644
--- a/include/linux/objtool.h
+++ b/include/linux/objtool.h
@@ -159,6 +159,7 @@
 #define STACK_FRAME_NON_STANDARD_FP(func)
 #define ANNOTATE_NOENDBR
 #define ASM_REACHABLE
+#define ASM_ANNOTATE(x)
 #else
 #define ANNOTATE_INTRA_FUNCTION_CALL
 .macro UNWIND_HINT type:req sp_reg=0 sp_offset=0 signal=0
@@ -169,7 +170,7 @@
 .endm
 .macro REACHABLE
 .endm
-.macro ANNOTATE
+.macro ANNOTATE type:req
 .endm
 #endif

This series applies on top of the latest kvm-x86/next with only a few trivial
conflicts, so I hope you are able to send a new version.

I could send one for you, but I have no idea how to properly test it.