[v2,1/4] x86/signal: Introduce helpers to get the maximum signal frame size

  Signal frames do not have a fixed format and can vary in size when a number
of things change: support XSAVE features, 32 vs. 64-bit apps. Add the code
to support a runtime method for userspace to dynamically discover how large
a signal stack needs to be.

Introduce a new variable, max_frame_size, and helper functions for the
calculation to be used in a new user interface. Set max_frame_size to a
system-wide worst-case value, instead of storing multiple app-specific
values.

Locate the body of the helper function -- fpu__get_fpstate_sigframe_size()
in fpu/signal.c for its relevance.

Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
Reviewed-by: Len Brown <len.brown@intel.com>
Acked-by: H.J. Lu <hjl.tools@gmail.com>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
---
Change from v1:
* Took stack alignment into account for sigframe size (Dave Martin)
---
 arch/x86/include/asm/fpu/signal.h |  2 +
 arch/x86/include/asm/sigframe.h   | 23 ++++++++++++
 arch/x86/kernel/cpu/common.c      |  3 ++
 arch/x86/kernel/fpu/signal.c      | 20 ++++++++++
 arch/x86/kernel/signal.c          | 61 ++++++++++++++++++++++++++++++-
 5 files changed, 107 insertions(+), 2 deletions(-)
  

On Thu, Nov 19, 2020 at 11:02:34AM -0800, Chang S. Bae wrote:
> Signal frames do not have a fixed format and can vary in size when a number
> of things change: support XSAVE features, 32 vs. 64-bit apps. Add the code
> to support a runtime method for userspace to dynamically discover how large
> a signal stack needs to be.
> 
> Introduce a new variable, max_frame_size, and helper functions for the
> calculation to be used in a new user interface. Set max_frame_size to a
> system-wide worst-case value, instead of storing multiple app-specific
> values.
> 
> Locate the body of the helper function -- fpu__get_fpstate_sigframe_size()
> in fpu/signal.c for its relevance.

This sentence is strange and not needed.

> diff --git a/arch/x86/include/asm/sigframe.h b/arch/x86/include/asm/sigframe.h
> index 84eab2724875..ac77f3f90bc9 100644
> --- a/arch/x86/include/asm/sigframe.h
> +++ b/arch/x86/include/asm/sigframe.h
> @@ -52,6 +52,15 @@ struct rt_sigframe_ia32 {
>  	char retcode[8];
>  	/* fp state follows here */
>  };
> +
> +#define SIZEOF_sigframe_ia32	sizeof(struct sigframe_ia32)
> +#define SIZEOF_rt_sigframe_ia32	sizeof(struct rt_sigframe_ia32)
> +
> +#else
> +
> +#define SIZEOF_sigframe_ia32	0
> +#define SIZEOF_rt_sigframe_ia32	0
> +
>  #endif /* defined(CONFIG_X86_32) || defined(CONFIG_IA32_EMULATION) */
>  
>  #ifdef CONFIG_X86_64
> @@ -81,8 +90,22 @@ struct rt_sigframe_x32 {
>  	/* fp state follows here */
>  };
>  
> +#define SIZEOF_rt_sigframe_x32	sizeof(struct rt_sigframe_x32)
> +
>  #endif /* CONFIG_X86_X32_ABI */
>  
> +#define SIZEOF_rt_sigframe	sizeof(struct rt_sigframe)
> +
> +#else
> +
> +#define SIZEOF_rt_sigframe	0
> +
>  #endif /* CONFIG_X86_64 */
>  
> +#ifndef SIZEOF_rt_sigframe_x32
> +#define SIZEOF_rt_sigframe_x32	0
> +#endif

Those are defined here to be used in only one place -
init_sigframe_size() - where there already is ifdeffery. Just use the
normal sizeof() operator there instead of adding more gunk here.

> diff --git a/arch/x86/kernel/fpu/signal.c b/arch/x86/kernel/fpu/signal.c
> index a4ec65317a7f..9f009525f551 100644
> --- a/arch/x86/kernel/fpu/signal.c
> +++ b/arch/x86/kernel/fpu/signal.c
> @@ -507,6 +507,26 @@ fpu__alloc_mathframe(unsigned long sp, int ia32_frame,
>  
>  	return sp;
>  }
> +
> +unsigned long fpu__get_fpstate_sigframe_size(void)
> +{
> +	unsigned long xstate_size = xstate_sigframe_size();
> +	unsigned long fsave_header_size = 0;
> +
> +	/*
> +	 * This space is needed on (most) 32-bit kernels, or when a 32-bit
> +	 * app is running on a 64-bit kernel. To keep things simple, just
> +	 * assume the worst case and always include space for 'freg_state',
> +	 * even for 64-bit apps on 64-bit kernels. This wastes a bit of
> +	 * space, but keeps the code simple.
> +	 */
> +	if ((IS_ENABLED(CONFIG_IA32_EMULATION) ||
> +	     IS_ENABLED(CONFIG_X86_32)) && use_fxsr())
> +		fsave_header_size = sizeof(struct fregs_state);
> +
> +	return fsave_header_size + xstate_size;
> +}

I guess this can be simplified to:

unsigned long fpu__get_fpstate_sigframe_size(void)
{
        unsigned long ret = xstate_sigframe_size();

        /*
         * This space is needed on (most) 32-bit kernels, or when a 32-bit
         * app is running on a 64-bit kernel. To keep things simple, just
         * assume the worst case and always include space for 'freg_state',
         * even for 64-bit apps on 64-bit kernels. This wastes a bit of
         * space, but keeps the code simple.
         */
        if ((IS_ENABLED(CONFIG_IA32_EMULATION) ||
             IS_ENABLED(CONFIG_X86_32)) && use_fxsr())
                ret += sizeof(struct fregs_state);

        return ret;
}

Also, this function simply gives you the xstate size, there's no need
for "sigframe" in the name.
  

> On Nov 25, 2020, at 20:17, Borislav Petkov <bp@alien8.de> wrote:
> 
> On Thu, Nov 19, 2020 at 11:02:34AM -0800, Chang S. Bae wrote:
>> diff --git a/arch/x86/include/asm/sigframe.h b/arch/x86/include/asm/sigframe.h
>> index 84eab2724875..ac77f3f90bc9 100644
>> --- a/arch/x86/include/asm/sigframe.h
>> +++ b/arch/x86/include/asm/sigframe.h
>> @@ -52,6 +52,15 @@ struct rt_sigframe_ia32 {
>> 	char retcode[8];
>> 	/* fp state follows here */
>> };
>> +
>> +#define SIZEOF_sigframe_ia32	sizeof(struct sigframe_ia32)
>> +#define SIZEOF_rt_sigframe_ia32	sizeof(struct rt_sigframe_ia32)
>> +
>> +#else
>> +
>> +#define SIZEOF_sigframe_ia32	0
>> +#define SIZEOF_rt_sigframe_ia32	0
>> +
>> #endif /* defined(CONFIG_X86_32) || defined(CONFIG_IA32_EMULATION) */
>> 
>> #ifdef CONFIG_X86_64
>> @@ -81,8 +90,22 @@ struct rt_sigframe_x32 {
>> 	/* fp state follows here */
>> };
>> 
>> +#define SIZEOF_rt_sigframe_x32	sizeof(struct rt_sigframe_x32)
>> +
>> #endif /* CONFIG_X86_X32_ABI */
>> 
>> +#define SIZEOF_rt_sigframe	sizeof(struct rt_sigframe)
>> +
>> +#else
>> +
>> +#define SIZEOF_rt_sigframe	0
>> +
>> #endif /* CONFIG_X86_64 */
>> 
>> +#ifndef SIZEOF_rt_sigframe_x32
>> +#define SIZEOF_rt_sigframe_x32	0
>> +#endif
> 
> Those are defined here to be used in only one place -
> init_sigframe_size() - where there already is ifdeffery. Just use the
> normal sizeof() operator there instead of adding more gunk here.

[ Just want to clarify your comment. ]

Admittedly, this is an (ugly) workaround to avoid compile errors.

E.g. when code is written like this in the function:

	if (IS_ENABLED(CONFIG_X86_X32_ABI))
		size = max(size, sizeof(struct rt_sigframe_x32));

and compile with CONFIG_X86_X32_ABI=n, got such a message:

	"invalid application of 'sizeof' to incomplete type 'struct
	sigframe_ia32’"

While the coding-style doc [1] seems to mention this:

   "However, this approach still allows the C compiler to see the
    code inside the block, and check it for correctness (syntax, 
    types, symbol references, etc). Thus, you still have to use an 
    #ifdef if the code inside the block references symbols that 
    will not exist if the condition is not met.”

In general, putting #ifdef in a C file is advised to avoid. I wonder
whether it is okay to include #ifdef in the C file in this case.

Thanks,
Chang

[1] https://www.kernel.org/doc/html/v5.9/process/coding-style.html
  

On Mon, Nov 30, 2020 at 08:40:32PM +0000, Bae, Chang Seok wrote:
> In general, putting #ifdef in a C file is advised to avoid.

Well, in this case it is arguably better to have the ifdeffery there,
where it is being used.

> I wonder whether it is okay to include #ifdef in the C file in this
> case.

Yes, it's not like this will be the only place where you have #ifdef in
a .c file:

[ ~/kernel/linux> git grep "#ifdef" *.c | wc -l
29401