[PATCH] x86 UEFI: Fix broken build for UEFI

[PATCH] x86 UEFI: Fix broken build for UEFI

[Zhenzhong Duan]

UEFI loads EFI applications to dynamic runtime addresses, so it requires
all applications to be compiled as PIC (position independent code).

The new introduced single-step #DB tests series bring some compile time
absolute address, fixed it with RIP relative address.

Fixes: 9734b4236294 ("x86/debug: Add framework for single-step #DB tests")
Fixes: 6bfb9572ec04 ("x86/debug: Test IN instead of RDMSR for single-step #DB emulation test")
Fixes: bc0dd8bdc627 ("x86/debug: Add single-step #DB + STI/MOVSS blocking tests")
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
 x86/debug.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/x86/debug.c b/x86/debug.c
index 20ff8ebacc16..13d1f9629e1d 100644
--- a/x86/debug.c
+++ b/x86/debug.c
@@ -145,7 +145,7 @@ static unsigned long singlestep_basic(void)
 		"and $~(1<<8),%%rax\n\t"
 		"1:push %%rax\n\t"
 		"popf\n\t"
-		"lea 1b, %0\n\t"
+		"lea 1b(%%rip), %0\n\t"
 		: "=r" (start) : : "rax"
 	);
 	return start;
@@ -186,7 +186,7 @@ static unsigned long singlestep_emulated_instructions(void)
 		"movl $0x3fd, %%edx\n\t"
 		"inb %%dx, %%al\n\t"
 		"popf\n\t"
-		"lea 1b,%0\n\t"
+		"lea 1b(%%rip),%0\n\t"
 		: "=r" (start) : : "rax", "ebx", "ecx", "edx"
 	);
 	return start;
@@ -223,7 +223,7 @@ static unsigned long singlestep_with_sti_blocking(void)
 		"1:and $~(1<<8),%%rax\n\t"
 		"push %%rax\n\t"
 		"popf\n\t"
-		"lea 1b,%0\n\t"
+		"lea 1b(%%rip),%0\n\t"
 		: "=r" (start_rip) : : "rax"
 	);
 	return start_rip;
@@ -259,7 +259,7 @@ static unsigned long singlestep_with_movss_blocking(void)
 		"and $~(1<<8),%%rax\n\t"
 		"1: push %%rax\n\t"
 		"popf\n\t"
-		"lea 1b,%0\n\t"
+		"lea 1b(%%rip),%0\n\t"
 		: "=r" (start_rip) : : "rax"
 	);
 	return start_rip;
@@ -302,7 +302,7 @@ static unsigned long singlestep_with_movss_blocking_and_icebp(void)
 		"1:and $~(1<<8),%%rax\n\t"
 		"push %%rax\n\t"
 		"popf\n\t"
-		"lea 1b,%0\n\t"
+		"lea 1b(%%rip),%0\n\t"
 		: "=r" (start) : : "rax"
 	);
 	return start;
@@ -346,7 +346,7 @@ static unsigned long singlestep_with_movss_blocking_and_dr7_gd(void)
 		"and $~(1<<8),%%rax\n\t"
 		"push %%rax\n\t"
 		"popf\n\t"
-		"lea 1b,%0\n\t"
+		"lea 1b(%%rip),%0\n\t"
 		: "=r" (start_rip) : : "rax"
 	);
 	return start_rip;
-- 
2.25.1

Re: [PATCH] x86 UEFI: Fix broken build for UEFI

[Sean Christopherson]

On Thu, Feb 10, 2022, Zhenzhong Duan wrote:
> UEFI loads EFI applications to dynamic runtime addresses, so it requires
> all applications to be compiled as PIC (position independent code).
> 
> The new introduced single-step #DB tests series bring some compile time
> absolute address, fixed it with RIP relative address.
> 
> Fixes: 9734b4236294 ("x86/debug: Add framework for single-step #DB tests")
> Fixes: 6bfb9572ec04 ("x86/debug: Test IN instead of RDMSR for single-step #DB emulation test")
> Fixes: bc0dd8bdc627 ("x86/debug: Add single-step #DB + STI/MOVSS blocking tests")
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---

Hrm.  It would be nice if we could enabled -fPIC by default for tests that support
it, having to compile twice is going to be annoying...

Reviewed-and-tested-by: Sean Christopherson <seanjc@google.com>

RE: [PATCH] x86 UEFI: Fix broken build for UEFI

[Duan, Zhenzhong]

>-----Original Message-----
>From: Sean Christopherson <seanjc@google.com>
>Sent: Friday, February 11, 2022 12:31 AM
>To: Duan, Zhenzhong <zhenzhong.duan@intel.com>
>Cc: kvm@vger.kernel.org; pbonzini@redhat.com
>Subject: Re: [PATCH] x86 UEFI: Fix broken build for UEFI
>
>On Thu, Feb 10, 2022, Zhenzhong Duan wrote:
>> UEFI loads EFI applications to dynamic runtime addresses, so it
>> requires all applications to be compiled as PIC (position independent code).
>>
>> The new introduced single-step #DB tests series bring some compile
>> time absolute address, fixed it with RIP relative address.
>>
>> Fixes: 9734b4236294 ("x86/debug: Add framework for single-step #DB
>> tests")
>> Fixes: 6bfb9572ec04 ("x86/debug: Test IN instead of RDMSR for
>> single-step #DB emulation test")
>> Fixes: bc0dd8bdc627 ("x86/debug: Add single-step #DB + STI/MOVSS
>> blocking tests")
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>> ---
>
>Hrm.  It would be nice if we could enabled -fPIC by default for tests that
>support it, having to compile twice is going to be annoying...

Agree it's annoying, but enable -fPIC for non-UEFI build may not help find
this build error, because the error happen in link stage to generate xxx.so,
while non-UEFI build will not generate xxx.so.

Thanks
Zhenzhong

Re: [PATCH] x86 UEFI: Fix broken build for UEFI

[Alexandru Elisei]

Hi,

On Thu, Feb 10, 2022 at 05:20:44PM +0800, Zhenzhong Duan wrote:
> UEFI loads EFI applications to dynamic runtime addresses, so it requires
> all applications to be compiled as PIC (position independent code).
> 
> The new introduced single-step #DB tests series bring some compile time
> absolute address, fixed it with RIP relative address.

With this patch the error:

ld: x86/debug.o: relocation R_X86_64_32S against `.text' can not be used when making a shared object; recompile with -fPIC

disappears and I can now build kvm-unit-tests for x86_64 when configured
with --target-efi:

Tested-by: Alexandru Elisei <alexandru.elisei@arm.com>

Thanks,
Alex

> 
> Fixes: 9734b4236294 ("x86/debug: Add framework for single-step #DB tests")
> Fixes: 6bfb9572ec04 ("x86/debug: Test IN instead of RDMSR for single-step #DB emulation test")
> Fixes: bc0dd8bdc627 ("x86/debug: Add single-step #DB + STI/MOVSS blocking tests")
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---
>  x86/debug.c | 12 ++++++------
>  1 file changed, 6 insertions(+), 6 deletions(-)
> 
> diff --git a/x86/debug.c b/x86/debug.c
> index 20ff8ebacc16..13d1f9629e1d 100644
> --- a/x86/debug.c
> +++ b/x86/debug.c
> @@ -145,7 +145,7 @@ static unsigned long singlestep_basic(void)
>  		"and $~(1<<8),%%rax\n\t"
>  		"1:push %%rax\n\t"
>  		"popf\n\t"
> -		"lea 1b, %0\n\t"
> +		"lea 1b(%%rip), %0\n\t"
>  		: "=r" (start) : : "rax"
>  	);
>  	return start;
> @@ -186,7 +186,7 @@ static unsigned long singlestep_emulated_instructions(void)
>  		"movl $0x3fd, %%edx\n\t"
>  		"inb %%dx, %%al\n\t"
>  		"popf\n\t"
> -		"lea 1b,%0\n\t"
> +		"lea 1b(%%rip),%0\n\t"
>  		: "=r" (start) : : "rax", "ebx", "ecx", "edx"
>  	);
>  	return start;
> @@ -223,7 +223,7 @@ static unsigned long singlestep_with_sti_blocking(void)
>  		"1:and $~(1<<8),%%rax\n\t"
>  		"push %%rax\n\t"
>  		"popf\n\t"
> -		"lea 1b,%0\n\t"
> +		"lea 1b(%%rip),%0\n\t"
>  		: "=r" (start_rip) : : "rax"
>  	);
>  	return start_rip;
> @@ -259,7 +259,7 @@ static unsigned long singlestep_with_movss_blocking(void)
>  		"and $~(1<<8),%%rax\n\t"
>  		"1: push %%rax\n\t"
>  		"popf\n\t"
> -		"lea 1b,%0\n\t"
> +		"lea 1b(%%rip),%0\n\t"
>  		: "=r" (start_rip) : : "rax"
>  	);
>  	return start_rip;
> @@ -302,7 +302,7 @@ static unsigned long singlestep_with_movss_blocking_and_icebp(void)
>  		"1:and $~(1<<8),%%rax\n\t"
>  		"push %%rax\n\t"
>  		"popf\n\t"
> -		"lea 1b,%0\n\t"
> +		"lea 1b(%%rip),%0\n\t"
>  		: "=r" (start) : : "rax"
>  	);
>  	return start;
> @@ -346,7 +346,7 @@ static unsigned long singlestep_with_movss_blocking_and_dr7_gd(void)
>  		"and $~(1<<8),%%rax\n\t"
>  		"push %%rax\n\t"
>  		"popf\n\t"
> -		"lea 1b,%0\n\t"
> +		"lea 1b(%%rip),%0\n\t"
>  		: "=r" (start_rip) : : "rax"
>  	);
>  	return start_rip;
> -- 
> 2.25.1
> 

Re: [PATCH] x86 UEFI: Fix broken build for UEFI

[Paolo Bonzini]

On 2/14/22 12:14, Alexandru Elisei wrote:
> Hi,
> 
> On Thu, Feb 10, 2022 at 05:20:44PM +0800, Zhenzhong Duan wrote:
>> UEFI loads EFI applications to dynamic runtime addresses, so it requires
>> all applications to be compiled as PIC (position independent code).
>>
>> The new introduced single-step #DB tests series bring some compile time
>> absolute address, fixed it with RIP relative address.
> 
> With this patch the error:
> 
> ld: x86/debug.o: relocation R_X86_64_32S against `.text' can not be used when making a shared object; recompile with -fPIC
> 
> disappears and I can now build kvm-unit-tests for x86_64 when configured
> with --target-efi:
> 
> Tested-by: Alexandru Elisei <alexandru.elisei@arm.com>

Applied, thanks.

Paolo