AT_BASE_PLATFORM (v2)

AT_BASE_PLATFORM (v2)

[Nathan Lynch]

Background:
Some IBM POWER-based systems have the ability to run in a
"compatibility" mode which mostly appears to the OS as a different
processor from the actual hardware.  This feature of the platform is
useful for live partition migration and for backwards compatibility
with old kernels on new hardware.  For example, a Power6 system may
appear to be a Power5+, which makes the AT_PLATFORM value "power5+".

Problem:
Booting a system in a compatibility mode means that ld.so may load
libraries that are inappropriately tuned for the real
microarchitecture, and apps that use JIT techniques do not have the
right information for generating tuned code.  While the AT_PLATFORM
auxiliary vector entry correctly indicates the ISA supported, it does
not accurately reflect the underlying microarchitecture in this case,
and there is no good way for userspace to get this information.

Proposed solution:
Add an AT_BASE_PLATFORM auxiliary vector entry which indicates the
microarchitecture.  This entry uses the same string format as
AT_PLATFORM, and is readily usable by ld.so and other applications.

Other solutions that have been suggested but found wanting:

- Use a bit in AT_HWCAP to indicate compat mode -- this is not
expressive enough.  It's not possible to derive the microarchitecture
from the combination of AT_PLATFORM's value and a single bit.

- Use dsocaps -- this seems to be a ld.so-specific interface and not
easily usable by other programs.  ld.so/glibc is not the only program
that can use knowledge of the microarchitecture.

The following two patches:
- add the base support to binfmt_elf.c for AT_BASE_PLATFORM
- implement AT_BASE_PLATFORM for powerpc

Changes since v1:
- increment AT_VECTOR_SIZE_BASE
- define AT_BASE_PLATFORM in generic code instead of powerpc

[PATCH] elf loader support for auxvec base platform string

[Nathan Lynch]

Some IBM POWER-based platforms have the ability to run in a
mode which mostly appears to the OS as a different processor from the
actual hardware.  For example, a Power6 system may appear to be a
Power5+, which makes the AT_PLATFORM value "power5+".  This means that
programs are restricted to the ISA supported by Power5+;
Power6-specific instructions are treated as illegal.

However, some applications (virtual machines, optimized libraries) can
benefit from knowledge of the underlying CPU model.  A new aux vector
entry, AT_BASE_PLATFORM, will denote the actual hardware.  For
example, on a Power6 system in Power5+ compatibility mode, AT_PLATFORM
will be "power5+" and AT_BASE_PLATFORM will be "power6".  The idea is
that AT_PLATFORM indicates the instruction set supported, while
AT_BASE_PLATFORM indicates the underlying microarchitecture.

If the architecture has defined ELF_BASE_PLATFORM, copy that value to
the user stack in the same manner as ELF_PLATFORM.

Signed-off-by: Nathan Lynch <ntl@pobox.com>
---
 fs/binfmt_elf.c        |   23 +++++++++++++++++++++++
 include/linux/auxvec.h |    5 ++++-
 2 files changed, 27 insertions(+), 1 deletions(-)

diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
index d48ff5f..834c2c4 100644
--- a/fs/binfmt_elf.c
+++ b/fs/binfmt_elf.c
@@ -131,6 +131,10 @@ static int padzero(unsigned long elf_bss)
 #define STACK_ALLOC(sp, len) ({ sp -= len ; sp; })
 #endif
 
+#ifndef ELF_BASE_PLATFORM
+#define ELF_BASE_PLATFORM NULL
+#endif
+
 static int
 create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
 		unsigned long load_addr, unsigned long interp_load_addr)
@@ -142,7 +146,9 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
 	elf_addr_t __user *envp;
 	elf_addr_t __user *sp;
 	elf_addr_t __user *u_platform;
+	elf_addr_t __user *u_base_platform;
 	const char *k_platform = ELF_PLATFORM;
+	const char *k_base_platform = ELF_BASE_PLATFORM;
 	int items;
 	elf_addr_t *elf_info;
 	int ei_index = 0;
@@ -172,6 +178,19 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
 			return -EFAULT;
 	}
 
+	/*
+	 * If this architecture has a "base" platform capability
+	 * string, copy it to userspace.
+	 */
+	u_base_platform = NULL;
+	if (k_base_platform) {
+		size_t len = strlen(k_base_platform) + 1;
+
+		u_base_platform = (elf_addr_t __user *)STACK_ALLOC(p, len);
+		if (__copy_to_user(u_base_platform, k_base_platform, len))
+			return -EFAULT;
+	}
+
 	/* Create the ELF interpreter info */
 	elf_info = (elf_addr_t *)current->mm->saved_auxv;
 	/* update AT_VECTOR_SIZE_BASE if the number of NEW_AUX_ENT() changes */
@@ -208,6 +227,10 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
 		NEW_AUX_ENT(AT_PLATFORM,
 			    (elf_addr_t)(unsigned long)u_platform);
 	}
+	if (k_base_platform) {
+		NEW_AUX_ENT(AT_BASE_PLATFORM,
+			    (elf_addr_t)(unsigned long)u_base_platform);
+	}
 	if (bprm->interp_flags & BINPRM_FLAGS_EXECFD) {
 		NEW_AUX_ENT(AT_EXECFD, bprm->interp_data);
 	}
diff --git a/include/linux/auxvec.h b/include/linux/auxvec.h
index ad89545..1adc61d 100644
--- a/include/linux/auxvec.h
+++ b/include/linux/auxvec.h
@@ -26,8 +26,11 @@
 
 #define AT_SECURE 23   /* secure mode boolean */
 
+#define AT_BASE_PLATFORM 38	/* string identifying real platform, may
+				 * differ from AT_PLATFORM. */
+
 #ifdef __KERNEL__
-#define AT_VECTOR_SIZE_BASE (14 + 2) /* NEW_AUX_ENT entries in auxiliary table */
+#define AT_VECTOR_SIZE_BASE (14 + 3) /* NEW_AUX_ENT entries in auxiliary table */
 #endif
 
 #endif /* _LINUX_AUXVEC_H */
-- 
1.5.6.2

Re: [PATCH] elf loader support for auxvec base platform string

[Benjamin Herrenschmidt]

Hi Linus, Andrew !

Should I seek somebody's ack before merging a patch like the one below ?

I'm a bit reluctant to merge via the powerpc.git tree some changes to
generic files without at least an ack from somebody else :-)

There have been some debate on whether this AT_BASE_PLATFORM is the
right approach, though I haven't seen them reach any useful conclusion
and our toolchain people internally are screaming for it...

Cheers,
Ben.

On Tue, 2008-07-15 at 18:58 -0500, Nathan Lynch wrote:
> Some IBM POWER-based platforms have the ability to run in a
> mode which mostly appears to the OS as a different processor from the
> actual hardware.  For example, a Power6 system may appear to be a
> Power5+, which makes the AT_PLATFORM value "power5+".  This means that
> programs are restricted to the ISA supported by Power5+;
> Power6-specific instructions are treated as illegal.
> 
> However, some applications (virtual machines, optimized libraries) can
> benefit from knowledge of the underlying CPU model.  A new aux vector
> entry, AT_BASE_PLATFORM, will denote the actual hardware.  For
> example, on a Power6 system in Power5+ compatibility mode, AT_PLATFORM
> will be "power5+" and AT_BASE_PLATFORM will be "power6".  The idea is
> that AT_PLATFORM indicates the instruction set supported, while
> AT_BASE_PLATFORM indicates the underlying microarchitecture.
> 
> If the architecture has defined ELF_BASE_PLATFORM, copy that value to
> the user stack in the same manner as ELF_PLATFORM.
> 
> Signed-off-by: Nathan Lynch <ntl@pobox.com>
> ---
>  fs/binfmt_elf.c        |   23 +++++++++++++++++++++++
>  include/linux/auxvec.h |    5 ++++-
>  2 files changed, 27 insertions(+), 1 deletions(-)
> 
> diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
> index d48ff5f..834c2c4 100644
> --- a/fs/binfmt_elf.c
> +++ b/fs/binfmt_elf.c
> @@ -131,6 +131,10 @@ static int padzero(unsigned long elf_bss)
>  #define STACK_ALLOC(sp, len) ({ sp -= len ; sp; })
>  #endif
>  
> +#ifndef ELF_BASE_PLATFORM
> +#define ELF_BASE_PLATFORM NULL
> +#endif
> +
>  static int
>  create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
>  		unsigned long load_addr, unsigned long interp_load_addr)
> @@ -142,7 +146,9 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
>  	elf_addr_t __user *envp;
>  	elf_addr_t __user *sp;
>  	elf_addr_t __user *u_platform;
> +	elf_addr_t __user *u_base_platform;
>  	const char *k_platform = ELF_PLATFORM;
> +	const char *k_base_platform = ELF_BASE_PLATFORM;
>  	int items;
>  	elf_addr_t *elf_info;
>  	int ei_index = 0;
> @@ -172,6 +178,19 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
>  			return -EFAULT;
>  	}
>  
> +	/*
> +	 * If this architecture has a "base" platform capability
> +	 * string, copy it to userspace.
> +	 */
> +	u_base_platform = NULL;
> +	if (k_base_platform) {
> +		size_t len = strlen(k_base_platform) + 1;
> +
> +		u_base_platform = (elf_addr_t __user *)STACK_ALLOC(p, len);
> +		if (__copy_to_user(u_base_platform, k_base_platform, len))
> +			return -EFAULT;
> +	}
> +
>  	/* Create the ELF interpreter info */
>  	elf_info = (elf_addr_t *)current->mm->saved_auxv;
>  	/* update AT_VECTOR_SIZE_BASE if the number of NEW_AUX_ENT() changes */
> @@ -208,6 +227,10 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
>  		NEW_AUX_ENT(AT_PLATFORM,
>  			    (elf_addr_t)(unsigned long)u_platform);
>  	}
> +	if (k_base_platform) {
> +		NEW_AUX_ENT(AT_BASE_PLATFORM,
> +			    (elf_addr_t)(unsigned long)u_base_platform);
> +	}
>  	if (bprm->interp_flags & BINPRM_FLAGS_EXECFD) {
>  		NEW_AUX_ENT(AT_EXECFD, bprm->interp_data);
>  	}
> diff --git a/include/linux/auxvec.h b/include/linux/auxvec.h
> index ad89545..1adc61d 100644
> --- a/include/linux/auxvec.h
> +++ b/include/linux/auxvec.h
> @@ -26,8 +26,11 @@
>  
>  #define AT_SECURE 23   /* secure mode boolean */
>  
> +#define AT_BASE_PLATFORM 38	/* string identifying real platform, may
> +				 * differ from AT_PLATFORM. */
> +
>  #ifdef __KERNEL__
> -#define AT_VECTOR_SIZE_BASE (14 + 2) /* NEW_AUX_ENT entries in auxiliary table */
> +#define AT_VECTOR_SIZE_BASE (14 + 3) /* NEW_AUX_ENT entries in auxiliary table */
>  #endif
>  
>  #endif /* _LINUX_AUXVEC_H */

Re: [PATCH] elf loader support for auxvec base platform string

[Andrew Morton]

On Thu, 17 Jul 2008 16:35:39 +1000 Benjamin Herrenschmidt <benh@kernel.crashing.org> wrote:

> Hi Linus, Andrew !
> 
> Should I seek somebody's ack before merging a patch like the one below ?

I think it's good to do so.

> I'm a bit reluctant to merge via the powerpc.git tree some changes to
> generic files without at least an ack from somebody else :-)

It tends to happen.  People often don't notice unless it a) crashes or
b) spits warnings or c) screws up my tree or d) all the above plus
more.

> There have been some debate on whether this AT_BASE_PLATFORM is the
> right approach, though I haven't seen them reach any useful conclusion
> and our toolchain people internally are screaming for it...
> 
> Cheers,
> Ben.
> 
> On Tue, 2008-07-15 at 18:58 -0500, Nathan Lynch wrote:
> > Some IBM POWER-based platforms have the ability to run in a
> > mode which mostly appears to the OS as a different processor from the
> > actual hardware.  For example, a Power6 system may appear to be a
> > Power5+, which makes the AT_PLATFORM value "power5+".  This means that
> > programs are restricted to the ISA supported by Power5+;
> > Power6-specific instructions are treated as illegal.
> > 
> > However, some applications (virtual machines, optimized libraries) can
> > benefit from knowledge of the underlying CPU model.  A new aux vector
> > entry, AT_BASE_PLATFORM, will denote the actual hardware.  For
> > example, on a Power6 system in Power5+ compatibility mode, AT_PLATFORM
> > will be "power5+" and AT_BASE_PLATFORM will be "power6".  The idea is
> > that AT_PLATFORM indicates the instruction set supported, while
> > AT_BASE_PLATFORM indicates the underlying microarchitecture.
> > 
> > If the architecture has defined ELF_BASE_PLATFORM, copy that value to
> > the user stack in the same manner as ELF_PLATFORM.
> > 
> > Signed-off-by: Nathan Lynch <ntl@pobox.com>
> > ---
> >  fs/binfmt_elf.c        |   23 +++++++++++++++++++++++
> >  include/linux/auxvec.h |    5 ++++-
> >  2 files changed, 27 insertions(+), 1 deletions(-)
> > 
> > diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
> > index d48ff5f..834c2c4 100644
> > --- a/fs/binfmt_elf.c
> > +++ b/fs/binfmt_elf.c
> > @@ -131,6 +131,10 @@ static int padzero(unsigned long elf_bss)
> >  #define STACK_ALLOC(sp, len) ({ sp -= len ; sp; })
> >  #endif
> >  
> > +#ifndef ELF_BASE_PLATFORM
> > +#define ELF_BASE_PLATFORM NULL
> > +#endif

Please add a comment which explains what this is.

Please also add a comment telling the world in which header file the
architecture *must* define this macro and then ensure that that header is
included into this file by reliable means.  asm/elf.h looks OK.

> >  static int
> >  create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
> >  		unsigned long load_addr, unsigned long interp_load_addr)
> > @@ -142,7 +146,9 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
> >  	elf_addr_t __user *envp;
> >  	elf_addr_t __user *sp;
> >  	elf_addr_t __user *u_platform;
> > +	elf_addr_t __user *u_base_platform;
> >  	const char *k_platform = ELF_PLATFORM;
> > +	const char *k_base_platform = ELF_BASE_PLATFORM;
> >  	int items;
> >  	elf_addr_t *elf_info;
> >  	int ei_index = 0;
> > @@ -172,6 +178,19 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
> >  			return -EFAULT;
> >  	}
> >  
> > +	/*
> > +	 * If this architecture has a "base" platform capability
> > +	 * string, copy it to userspace.
> > +	 */
> > +	u_base_platform = NULL;
> > +	if (k_base_platform) {
> > +		size_t len = strlen(k_base_platform) + 1;
> > +
> > +		u_base_platform = (elf_addr_t __user *)STACK_ALLOC(p, len);
> > +		if (__copy_to_user(u_base_platform, k_base_platform, len))
> > +			return -EFAULT;
> > +	}

>From my reading, this change will result in no additional code
generation on non-powerpc architectures.  This is good.  If poss, could
you please verify that theory and perhaps drop a note in the changelog
about that?


Apart from that - acked-by-me

Re: [PATCH] elf loader support for auxvec base platform string

[Nathan Lynch]

Andrew Morton wrote:
> > On Tue, 2008-07-15 at 18:58 -0500, Nathan Lynch wrote:
> > > diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
> > > index d48ff5f..834c2c4 100644
> > > --- a/fs/binfmt_elf.c
> > > +++ b/fs/binfmt_elf.c
> > > @@ -131,6 +131,10 @@ static int padzero(unsigned long elf_bss)
> > >  #define STACK_ALLOC(sp, len) ({ sp -= len ; sp; })
> > >  #endif
> > >  
> > > +#ifndef ELF_BASE_PLATFORM
> > > +#define ELF_BASE_PLATFORM NULL
> > > +#endif
> 
> Please add a comment which explains what this is.
> 
> Please also add a comment telling the world in which header file the
> architecture *must* define this macro and then ensure that that header is
> included into this file by reliable means.  asm/elf.h looks OK.

Okay.


> > > @@ -172,6 +178,19 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
> > >  			return -EFAULT;
> > >  	}
> > >  
> > > +	/*
> > > +	 * If this architecture has a "base" platform capability
> > > +	 * string, copy it to userspace.
> > > +	 */
> > > +	u_base_platform = NULL;
> > > +	if (k_base_platform) {
> > > +		size_t len = strlen(k_base_platform) + 1;
> > > +
> > > +		u_base_platform = (elf_addr_t __user *)STACK_ALLOC(p, len);
> > > +		if (__copy_to_user(u_base_platform, k_base_platform, len))
> > > +			return -EFAULT;
> > > +	}
> 
> >From my reading, this change will result in no additional code
> generation on non-powerpc architectures.  This is good.  If poss, could
> you please verify that theory and perhaps drop a note in the changelog
> about that?

That was the intent, yes.  However:

$ scripts/bloat-o-meter vmlinux-x86-{before,after}
add/remove: 0/0 grow/shrink: 2/0 up/down: 18/0 (18)
function                                     old     new   delta
init_mm                                      784     800     +16
load_elf_binary                            11946   11948      +2

(x86_64_defconfig, gcc 4.2.3)

The init_mm/mm_struct bloat is expected (although I'd like to avoid
that), but evidently it has some small effect on load_elf_binary.

[PATCH v3] elf loader support for auxvec base platform string

[Nathan Lynch]

Some IBM POWER-based platforms have the ability to run in a
mode which mostly appears to the OS as a different processor from the
actual hardware.  For example, a Power6 system may appear to be a
Power5+, which makes the AT_PLATFORM value "power5+".  This means that
programs are restricted to the ISA supported by Power5+;
Power6-specific instructions are treated as illegal.

However, some applications (virtual machines, optimized libraries) can
benefit from knowledge of the underlying CPU model.  A new aux vector
entry, AT_BASE_PLATFORM, will denote the actual hardware.  For
example, on a Power6 system in Power5+ compatibility mode, AT_PLATFORM
will be "power5+" and AT_BASE_PLATFORM will be "power6".  The idea is
that AT_PLATFORM indicates the instruction set supported, while
AT_BASE_PLATFORM indicates the underlying microarchitecture.

If the architecture has defined ELF_BASE_PLATFORM, copy that value to
the user stack in the same manner as ELF_PLATFORM.

Signed-off-by: Nathan Lynch <ntl@pobox.com>

---

Added comment explaining ELF_BASE_PLATFORM.

 fs/binfmt_elf.c        |   28 ++++++++++++++++++++++++++++
 include/linux/auxvec.h |    5 ++++-
 2 files changed, 32 insertions(+), 1 deletions(-)

diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
index d48ff5f..d8a7cc0 100644
--- a/fs/binfmt_elf.c
+++ b/fs/binfmt_elf.c
@@ -131,6 +131,15 @@ static int padzero(unsigned long elf_bss)
 #define STACK_ALLOC(sp, len) ({ sp -= len ; sp; })
 #endif
 
+#ifndef ELF_BASE_PLATFORM
+/*
+ * AT_BASE_PLATFORM indicates the "real" hardware/microarchitecture.
+ * If the arch defines ELF_BASE_PLATFORM (in asm/elf.h), the value
+ * will be copied to the user stack in the same manner as AT_PLATFORM.
+ */
+#define ELF_BASE_PLATFORM NULL
+#endif
+
 static int
 create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
 		unsigned long load_addr, unsigned long interp_load_addr)
@@ -142,7 +151,9 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
 	elf_addr_t __user *envp;
 	elf_addr_t __user *sp;
 	elf_addr_t __user *u_platform;
+	elf_addr_t __user *u_base_platform;
 	const char *k_platform = ELF_PLATFORM;
+	const char *k_base_platform = ELF_BASE_PLATFORM;
 	int items;
 	elf_addr_t *elf_info;
 	int ei_index = 0;
@@ -172,6 +183,19 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
 			return -EFAULT;
 	}
 
+	/*
+	 * If this architecture has a "base" platform capability
+	 * string, copy it to userspace.
+	 */
+	u_base_platform = NULL;
+	if (k_base_platform) {
+		size_t len = strlen(k_base_platform) + 1;
+
+		u_base_platform = (elf_addr_t __user *)STACK_ALLOC(p, len);
+		if (__copy_to_user(u_base_platform, k_base_platform, len))
+			return -EFAULT;
+	}
+
 	/* Create the ELF interpreter info */
 	elf_info = (elf_addr_t *)current->mm->saved_auxv;
 	/* update AT_VECTOR_SIZE_BASE if the number of NEW_AUX_ENT() changes */
@@ -208,6 +232,10 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
 		NEW_AUX_ENT(AT_PLATFORM,
 			    (elf_addr_t)(unsigned long)u_platform);
 	}
+	if (k_base_platform) {
+		NEW_AUX_ENT(AT_BASE_PLATFORM,
+			    (elf_addr_t)(unsigned long)u_base_platform);
+	}
 	if (bprm->interp_flags & BINPRM_FLAGS_EXECFD) {
 		NEW_AUX_ENT(AT_EXECFD, bprm->interp_data);
 	}
diff --git a/include/linux/auxvec.h b/include/linux/auxvec.h
index ad89545..1adc61d 100644
--- a/include/linux/auxvec.h
+++ b/include/linux/auxvec.h
@@ -26,8 +26,11 @@
 
 #define AT_SECURE 23   /* secure mode boolean */
 
+#define AT_BASE_PLATFORM 38	/* string identifying real platform, may
+				 * differ from AT_PLATFORM. */
+
 #ifdef __KERNEL__
-#define AT_VECTOR_SIZE_BASE (14 + 2) /* NEW_AUX_ENT entries in auxiliary table */
+#define AT_VECTOR_SIZE_BASE (14 + 3) /* NEW_AUX_ENT entries in auxiliary table */
 #endif
 
 #endif /* _LINUX_AUXVEC_H */
-- 
1.5.6.2

Re: [PATCH v3] elf loader support for auxvec base platform string

[Andrew Morton]

On Thu, 17 Jul 2008 17:19:32 -0500
Nathan Lynch <ntl@pobox.com> wrote:

> Some IBM POWER-based platforms have the ability to run in a
> mode which mostly appears to the OS as a different processor from the
> actual hardware.  For example, a Power6 system may appear to be a
> Power5+, which makes the AT_PLATFORM value "power5+".  This means that
> programs are restricted to the ISA supported by Power5+;
> Power6-specific instructions are treated as illegal.
> 
> However, some applications (virtual machines, optimized libraries) can
> benefit from knowledge of the underlying CPU model.  A new aux vector
> entry, AT_BASE_PLATFORM, will denote the actual hardware.  For
> example, on a Power6 system in Power5+ compatibility mode, AT_PLATFORM
> will be "power5+" and AT_BASE_PLATFORM will be "power6".  The idea is
> that AT_PLATFORM indicates the instruction set supported, while
> AT_BASE_PLATFORM indicates the underlying microarchitecture.
> 
> If the architecture has defined ELF_BASE_PLATFORM, copy that value to
> the user stack in the same manner as ELF_PLATFORM.
> 
> Signed-off-by: Nathan Lynch <ntl@pobox.com>
> 
> ---
> 
> Added comment explaining ELF_BASE_PLATFORM.
> 
>  fs/binfmt_elf.c        |   28 ++++++++++++++++++++++++++++
>  include/linux/auxvec.h |    5 ++++-
>  2 files changed, 32 insertions(+), 1 deletions(-)
> 
> diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
> index d48ff5f..d8a7cc0 100644
> --- a/fs/binfmt_elf.c
> +++ b/fs/binfmt_elf.c
> @@ -131,6 +131,15 @@ static int padzero(unsigned long elf_bss)
>  #define STACK_ALLOC(sp, len) ({ sp -= len ; sp; })
>  #endif
>  
> +#ifndef ELF_BASE_PLATFORM
> +/*
> + * AT_BASE_PLATFORM indicates the "real" hardware/microarchitecture.
> + * If the arch defines ELF_BASE_PLATFORM (in asm/elf.h), the value
> + * will be copied to the user stack in the same manner as AT_PLATFORM.
> + */
> +#define ELF_BASE_PLATFORM NULL
> +#endif
> +
>  static int
>  create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
>  		unsigned long load_addr, unsigned long interp_load_addr)
> @@ -142,7 +151,9 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
>  	elf_addr_t __user *envp;
>  	elf_addr_t __user *sp;
>  	elf_addr_t __user *u_platform;
> +	elf_addr_t __user *u_base_platform;
>  	const char *k_platform = ELF_PLATFORM;
> +	const char *k_base_platform = ELF_BASE_PLATFORM;
>  	int items;
>  	elf_addr_t *elf_info;
>  	int ei_index = 0;
> @@ -172,6 +183,19 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
>  			return -EFAULT;
>  	}
>  
> +	/*
> +	 * If this architecture has a "base" platform capability
> +	 * string, copy it to userspace.
> +	 */
> +	u_base_platform = NULL;
> +	if (k_base_platform) {
> +		size_t len = strlen(k_base_platform) + 1;
> +
> +		u_base_platform = (elf_addr_t __user *)STACK_ALLOC(p, len);
> +		if (__copy_to_user(u_base_platform, k_base_platform, len))
> +			return -EFAULT;
> +	}
> +
>  	/* Create the ELF interpreter info */
>  	elf_info = (elf_addr_t *)current->mm->saved_auxv;
>  	/* update AT_VECTOR_SIZE_BASE if the number of NEW_AUX_ENT() changes */
> @@ -208,6 +232,10 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
>  		NEW_AUX_ENT(AT_PLATFORM,
>  			    (elf_addr_t)(unsigned long)u_platform);
>  	}
> +	if (k_base_platform) {
> +		NEW_AUX_ENT(AT_BASE_PLATFORM,
> +			    (elf_addr_t)(unsigned long)u_base_platform);
> +	}
>  	if (bprm->interp_flags & BINPRM_FLAGS_EXECFD) {
>  		NEW_AUX_ENT(AT_EXECFD, bprm->interp_data);
>  	}
> diff --git a/include/linux/auxvec.h b/include/linux/auxvec.h
> index ad89545..1adc61d 100644
> --- a/include/linux/auxvec.h
> +++ b/include/linux/auxvec.h
> @@ -26,8 +26,11 @@
>  
>  #define AT_SECURE 23   /* secure mode boolean */
>  
> +#define AT_BASE_PLATFORM 38	/* string identifying real platform, may
> +				 * differ from AT_PLATFORM. */
> +
>  #ifdef __KERNEL__
> -#define AT_VECTOR_SIZE_BASE (14 + 2) /* NEW_AUX_ENT entries in auxiliary table */
> +#define AT_VECTOR_SIZE_BASE (14 + 3) /* NEW_AUX_ENT entries in auxiliary table */
>  #endif
>  
>  #endif /* _LINUX_AUXVEC_H */

OK.

But it conflicts directly with the already-queued
execve-filename-document-and-export-via-auxiliary-vector.patch (which
various potential reviewers blithely deleted - don't come complaining
to me!):


From: John Reiser <jreiser@BitWagon.com>

The Linux kernel puts the filename argument of execve() into the
new address space.  Many developers are surprised to learn this.
Those who know and could use it, object "But it's not documented."
Those who want to use it dislike the expression
  (char *)(1+ strlen(env[-1+ n_env]) + env[-1+ n_env])
because it requires locating the last original environment variable,
and assumes that the filename follows the characters.

This patch documents the insertion of the filename, and makes it easier to
find by adding a new tag AT_EXECFN in the ElfXX_auxv_t; see <elf.h>.

In many cases readlink("/proc/self/exe",) gives the same answer.  But if all
the original pages get unmapped, then the kernel erases the symlink for
/proc/self/exe.  This can happen when a program decompressor does a good job
of cleaning up after uncompressing directly to memory, so that the address
space of the target program looks the same as if compression had never
happened.  One example is http://upx.sourceforge.net .

One notable use of the underlying concept (what path containED the executable)
is glibc expanding $ORIGIN in DT_RUNPATH.  In practice for the near term, it
may be a good idea for user-mode code to use both /proc/self/exe and AT_EXECFN
as fall-back methods for each other.  /proc/self/exe can fail due to
unmapping, AT_EXECFN can fail because it won't be present on non-new systems. 
The auxvec or {AT_EXECFN}.d_val also can get overwritten, although in nearly
all cases this would be the result of a bug.

The runtime cost is one NEW_AUX_ENT using two words of stack space.  The
underlying value is maintained already as bprm->exec; setup_arg_pages() in
fs/exec.c slides it for stack_shift, etc.

Signed-off-by: John Reiser <jreiser@BitWagon.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Jakub Jelinek <jakub@redhat.com>
Cc: Ulrich Drepper <drepper@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 fs/binfmt_elf.c        |    1 +
 include/linux/auxvec.h |    4 +++-
 2 files changed, 4 insertions(+), 1 deletion(-)

diff -puN fs/binfmt_elf.c~execve-filename-document-and-export-via-auxiliary-vector fs/binfmt_elf.c
--- a/fs/binfmt_elf.c~execve-filename-document-and-export-via-auxiliary-vector
+++ a/fs/binfmt_elf.c
@@ -204,6 +204,7 @@ create_elf_tables(struct linux_binprm *b
 	NEW_AUX_ENT(AT_GID, tsk->gid);
 	NEW_AUX_ENT(AT_EGID, tsk->egid);
  	NEW_AUX_ENT(AT_SECURE, security_bprm_secureexec(bprm));
+	NEW_AUX_ENT(AT_EXECFN, bprm->exec);
 	if (k_platform) {
 		NEW_AUX_ENT(AT_PLATFORM,
 			    (elf_addr_t)(unsigned long)u_platform);
diff -puN include/linux/auxvec.h~execve-filename-document-and-export-via-auxiliary-vector include/linux/auxvec.h
--- a/include/linux/auxvec.h~execve-filename-document-and-export-via-auxiliary-vector
+++ a/include/linux/auxvec.h
@@ -26,8 +26,10 @@
 
 #define AT_SECURE 23   /* secure mode boolean */
 
+#define AT_EXECFN  31	/* filename of program */
 #ifdef __KERNEL__
-#define AT_VECTOR_SIZE_BASE (14 + 2) /* NEW_AUX_ENT entries in auxiliary table */
+#define AT_VECTOR_SIZE_BASE 17 /* NEW_AUX_ENT entries in auxiliary table */
+  /* number of "#define AT_.*" above, minus {AT_NULL, AT_IGNORE, AT_NOTELF} */
 #endif
 
 #endif /* _LINUX_AUXVEC_H */
_

Re: [PATCH v3] elf loader support for auxvec base platform string

[John Reiser]

Andrew Morton wrote:
> On Thu, 17 Jul 2008 17:19:32 -0500
> Nathan Lynch <ntl@pobox.com> wrote:
> 
> 
>> [snip]
>> A new aux vector entry, AT_BASE_PLATFORM, will denote the actual hardware.
[snip]

> OK.
> 
> But it conflicts directly with the already-queued
> execve-filename-document-and-export-via-auxiliary-vector.patch

[snip]

It seems to me that most of the patch conflicts are mechanical
and could be merged mechanically.

However I believe that the documentation change to this comment is important:
-----
>  #ifdef __KERNEL__
> -#define AT_VECTOR_SIZE_BASE (14 + 2) /* NEW_AUX_ENT entries in auxiliary table */
> +#define AT_VECTOR_SIZE_BASE 17 /* NEW_AUX_ENT entries in auxiliary table */
> +  /* number of "#define AT_.*" above, minus {AT_NULL, AT_IGNORE, AT_NOTELF} */
>  #endif
-----
I scratched my head for a while to figure out that AT_NOTELF also was
a subtraction as far as AT_VECTOR_SIZE_BASE was concerned.


-- 
John Reiser, jreiser@BitWagon.com

Re: [PATCH v3] elf loader support for auxvec base platform string

[Nathan Lynch]

John Reiser wrote:
> Andrew Morton wrote:
> > On Thu, 17 Jul 2008 17:19:32 -0500
> > Nathan Lynch <ntl@pobox.com> wrote:
> > 
> > 
> >> [snip]
> >> A new aux vector entry, AT_BASE_PLATFORM, will denote the actual hardware.
> [snip]
> 
> > OK.
> > 
> > But it conflicts directly with the already-queued
> > execve-filename-document-and-export-via-auxiliary-vector.patch

Okay, I can rebase on -mm.


> It seems to me that most of the patch conflicts are mechanical
> and could be merged mechanically.
> 
> However I believe that the documentation change to this comment is important:
> -----
> >  #ifdef __KERNEL__
> > -#define AT_VECTOR_SIZE_BASE (14 + 2) /* NEW_AUX_ENT entries in auxiliary table */
> > +#define AT_VECTOR_SIZE_BASE 17 /* NEW_AUX_ENT entries in auxiliary table */
> > +  /* number of "#define AT_.*" above, minus {AT_NULL, AT_IGNORE, AT_NOTELF} */
> >  #endif
> -----
> I scratched my head for a while to figure out that AT_NOTELF also was
> a subtraction as far as AT_VECTOR_SIZE_BASE was concerned.

John, from your patch:

+#define AT_EXECFN  31	/* filename of program */

How did you arrive at 31 for the value of AT_EXECFN?  I haven't been
able to find out how AT_* values are "allocated", or what the reason
is for the gap between AT_SECURE and AT_SYSINFO.

Re: [PATCH v3] elf loader support for auxvec base platform string

[John Reiser]

Nathan Lynch wrote:

> +#define AT_EXECFN  31	/* filename of program */
> 
> How did you arrive at 31 for the value of AT_EXECFN?  I haven't been
> able to find out how AT_* values are "allocated", or what the reason
> is for the gap between AT_SECURE and AT_SYSINFO.

The numbers are chosen by experience, taste, and fiat.
Hopefully new choices do not conflict with existing ones,
but there is no formal "issuing authority."
In history the auxiliary vector has been not well standardized
and many times has been hidden from view of applications, although some
SysV-based systems have made it visible as a fourth argument to main().
Linux has /proc/pid/auxv, although the implementation suffers
from being exposed to overwriting by the user.  From long experience
at virtualization in user mode, I favor better access, more use,
and better understanding of what the auxiliary vector provides.

AT_SYSINFO at 32 was chosen to avoid conflicts with [0,31]
partly on the theory that the first 32 might be considered
to be reserved for use across all UNIX-like systems,
while AT_SYSINFO definitely was Linux-specific.
I chose AT_EXECFN at 31 because I considered the concept
to be applicable to any system having execve(), even if AT_EXECFN
is not universally implemented.  I had not seen any new tags
below 32 in a long time.  The concept of AT_EXECFN allows
a nice interface for a virtualizer. The somewhat-related
AT_EXECFD already exists below 32.

Elsewhere, I've staked out use of a new AT_WINE_PRELOAD_INFO
at 30.  Avoid that one, please. :-)

-- 
John Reiser, jreiser@BitWagon.com

Re: [PATCH v3] elf loader support for auxvec base platform string

[Andrew Morton]

On Fri, 18 Jul 2008 13:31:29 -0700
John Reiser <jreiser@BitWagon.com> wrote:

> Elsewhere, I've staked out use of a new AT_WINE_PRELOAD_INFO
> at 30.  Avoid that one, please. :-)

The reliable way in which to reserve these numbers is to patch the
header file.

Re: [PATCH v3] elf loader support for auxvec base platform string

[Benjamin Herrenschmidt]

On Fri, 2008-07-18 at 13:28 -0500, Nathan Lynch wrote:
> John Reiser wrote:
> > Andrew Morton wrote:
> > > On Thu, 17 Jul 2008 17:19:32 -0500
> > > Nathan Lynch <ntl@pobox.com> wrote:
> > > 
> > > 
> > >> [snip]
> > >> A new aux vector entry, AT_BASE_PLATFORM, will denote the actual hardware.
> > [snip]
> > 
> > > OK.
> > > 
> > > But it conflicts directly with the already-queued
> > > execve-filename-document-and-export-via-auxiliary-vector.patch
> 
> Okay, I can rebase on -mm.

Andrew, we have one other patch (the powerpc bits) on top of that one.
Do you want to carry both in -mm on top of John's patch ? We would like
that in .27 though, I don't know what your merge plans are for John's
patch.

Cheers,
Ben.

Re: [PATCH v3] elf loader support for auxvec base platform string

[Andrew Morton]

On Mon, 21 Jul 2008 13:24:15 +1000 Benjamin Herrenschmidt <benh@kernel.crashing.org> wrote:

> On Fri, 2008-07-18 at 13:28 -0500, Nathan Lynch wrote:
> > John Reiser wrote:
> > > Andrew Morton wrote:
> > > > On Thu, 17 Jul 2008 17:19:32 -0500
> > > > Nathan Lynch <ntl@pobox.com> wrote:
> > > > 
> > > > 
> > > >> [snip]
> > > >> A new aux vector entry, AT_BASE_PLATFORM, will denote the actual hardware.
> > > [snip]
> > > 
> > > > OK.
> > > > 
> > > > But it conflicts directly with the already-queued
> > > > execve-filename-document-and-export-via-auxiliary-vector.patch
> > 
> > Okay, I can rebase on -mm.
> 
> Andrew, we have one other patch (the powerpc bits) on top of that one.
> Do you want to carry both in -mm on top of John's patch ? We would like
> that in .27 though, I don't know what your merge plans are for John's
> patch.
> 

How about I send John's patch Linuswards right now?

Re: [PATCH v3] elf loader support for auxvec base platform string

[Benjamin Herrenschmidt]

> > Andrew, we have one other patch (the powerpc bits) on top of that one.
> > Do you want to carry both in -mm on top of John's patch ? We would like
> > that in .27 though, I don't know what your merge plans are for John's
> > patch.
> > 
> 
> How about I send John's patch Linuswards right now?

No objection.

Cheers,
Ben.

[PATCH v4] elf loader support for auxvec base platform string

[Nathan Lynch]

Some IBM POWER-based platforms have the ability to run in a
mode which mostly appears to the OS as a different processor from the
actual hardware.  For example, a Power6 system may appear to be a
Power5+, which makes the AT_PLATFORM value "power5+".  This means that
programs are restricted to the ISA supported by Power5+;
Power6-specific instructions are treated as illegal.

However, some applications (virtual machines, optimized libraries) can
benefit from knowledge of the underlying CPU model.  A new aux vector
entry, AT_BASE_PLATFORM, will denote the actual hardware.  For
example, on a Power6 system in Power5+ compatibility mode, AT_PLATFORM
will be "power5+" and AT_BASE_PLATFORM will be "power6".  The idea is
that AT_PLATFORM indicates the instruction set supported, while
AT_BASE_PLATFORM indicates the underlying microarchitecture.

If the architecture has defined ELF_BASE_PLATFORM, copy that value to
the user stack in the same manner as ELF_PLATFORM.

Signed-off-by: Nathan Lynch <ntl@pobox.com>

---

Andrew Morton wrote:
> OK.
> 
> But it conflicts directly with the already-queued
> execve-filename-document-and-export-via-auxiliary-vector.patch (which
> various potential reviewers blithely deleted - don't come complaining
> to me!):
> 

Rebased to -mm to resolve conflicts with
execve-filename-document-and-export-via-auxiliary-vector.patch, and
changed AT_BASE_PLATFORM to lowest unclaimed value (24).


 fs/binfmt_elf.c        |   28 ++++++++++++++++++++++++++++
 include/linux/auxvec.h |    6 +++++-
 2 files changed, 33 insertions(+), 1 deletions(-)

diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
index bad7d87..180f92b 100644
--- a/fs/binfmt_elf.c
+++ b/fs/binfmt_elf.c
@@ -131,6 +131,15 @@ static int padzero(unsigned long elf_bss)
 #define STACK_ALLOC(sp, len) ({ sp -= len ; sp; })
 #endif
 
+#ifndef ELF_BASE_PLATFORM
+/*
+ * AT_BASE_PLATFORM indicates the "real" hardware/microarchitecture.
+ * If the arch defines ELF_BASE_PLATFORM (in asm/elf.h), the value
+ * will be copied to the user stack in the same manner as AT_PLATFORM.
+ */
+#define ELF_BASE_PLATFORM NULL
+#endif
+
 static int
 create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
 		unsigned long load_addr, unsigned long interp_load_addr)
@@ -142,7 +151,9 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
 	elf_addr_t __user *envp;
 	elf_addr_t __user *sp;
 	elf_addr_t __user *u_platform;
+	elf_addr_t __user *u_base_platform;
 	const char *k_platform = ELF_PLATFORM;
+	const char *k_base_platform = ELF_BASE_PLATFORM;
 	int items;
 	elf_addr_t *elf_info;
 	int ei_index = 0;
@@ -172,6 +183,19 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
 			return -EFAULT;
 	}
 
+	/*
+	 * If this architecture has a "base" platform capability
+	 * string, copy it to userspace.
+	 */
+	u_base_platform = NULL;
+	if (k_base_platform) {
+		size_t len = strlen(k_base_platform) + 1;
+
+		u_base_platform = (elf_addr_t __user *)STACK_ALLOC(p, len);
+		if (__copy_to_user(u_base_platform, k_base_platform, len))
+			return -EFAULT;
+	}
+
 	/* Create the ELF interpreter info */
 	elf_info = (elf_addr_t *)current->mm->saved_auxv;
 	/* update AT_VECTOR_SIZE_BASE if the number of NEW_AUX_ENT() changes */
@@ -209,6 +233,10 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
 		NEW_AUX_ENT(AT_PLATFORM,
 			    (elf_addr_t)(unsigned long)u_platform);
 	}
+	if (k_base_platform) {
+		NEW_AUX_ENT(AT_BASE_PLATFORM,
+			    (elf_addr_t)(unsigned long)u_base_platform);
+	}
 	if (bprm->interp_flags & BINPRM_FLAGS_EXECFD) {
 		NEW_AUX_ENT(AT_EXECFD, bprm->interp_data);
 	}
diff --git a/include/linux/auxvec.h b/include/linux/auxvec.h
index 0da17d1..d7afa9d 100644
--- a/include/linux/auxvec.h
+++ b/include/linux/auxvec.h
@@ -26,9 +26,13 @@
 
 #define AT_SECURE 23   /* secure mode boolean */
 
+#define AT_BASE_PLATFORM 24	/* string identifying real platform, may
+				 * differ from AT_PLATFORM. */
+
 #define AT_EXECFN  31	/* filename of program */
+
 #ifdef __KERNEL__
-#define AT_VECTOR_SIZE_BASE 17 /* NEW_AUX_ENT entries in auxiliary table */
+#define AT_VECTOR_SIZE_BASE 18 /* NEW_AUX_ENT entries in auxiliary table */
   /* number of "#define AT_.*" above, minus {AT_NULL, AT_IGNORE, AT_NOTELF} */
 #endif
 
-- 
1.5.6.2

Re: [PATCH v4] elf loader support for auxvec base platform string

[Benjamin Herrenschmidt]

On Mon, 2008-07-21 at 13:48 -0500, Nathan Lynch wrote:
> Some IBM POWER-based platforms have the ability to run in a
> mode which mostly appears to the OS as a different processor from the
> actual hardware.  For example, a Power6 system may appear to be a
> Power5+, which makes the AT_PLATFORM value "power5+".  This means that
> programs are restricted to the ISA supported by Power5+;
> Power6-specific instructions are treated as illegal.
> 
> However, some applications (virtual machines, optimized libraries) can
> benefit from knowledge of the underlying CPU model.  A new aux vector
> entry, AT_BASE_PLATFORM, will denote the actual hardware.  For
> example, on a Power6 system in Power5+ compatibility mode, AT_PLATFORM
> will be "power5+" and AT_BASE_PLATFORM will be "power6".  The idea is
> that AT_PLATFORM indicates the instruction set supported, while
> AT_BASE_PLATFORM indicates the underlying microarchitecture.
> 
> If the architecture has defined ELF_BASE_PLATFORM, copy that value to
> the user stack in the same manner as ELF_PLATFORM.
> 
> Signed-off-by: Nathan Lynch <ntl@pobox.com>

Andrew, I'll merge that after John patch shows up if you give me your
Ack :-)

Cheers,
Ben.

> ---
> 
> Andrew Morton wrote:
> > OK.
> > 
> > But it conflicts directly with the already-queued
> > execve-filename-document-and-export-via-auxiliary-vector.patch (which
> > various potential reviewers blithely deleted - don't come complaining
> > to me!):
> > 
> 
> Rebased to -mm to resolve conflicts with
> execve-filename-document-and-export-via-auxiliary-vector.patch, and
> changed AT_BASE_PLATFORM to lowest unclaimed value (24).
> 
> 
>  fs/binfmt_elf.c        |   28 ++++++++++++++++++++++++++++
>  include/linux/auxvec.h |    6 +++++-
>  2 files changed, 33 insertions(+), 1 deletions(-)
> 
> diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
> index bad7d87..180f92b 100644
> --- a/fs/binfmt_elf.c
> +++ b/fs/binfmt_elf.c
> @@ -131,6 +131,15 @@ static int padzero(unsigned long elf_bss)
>  #define STACK_ALLOC(sp, len) ({ sp -= len ; sp; })
>  #endif
>  
> +#ifndef ELF_BASE_PLATFORM
> +/*
> + * AT_BASE_PLATFORM indicates the "real" hardware/microarchitecture.
> + * If the arch defines ELF_BASE_PLATFORM (in asm/elf.h), the value
> + * will be copied to the user stack in the same manner as AT_PLATFORM.
> + */
> +#define ELF_BASE_PLATFORM NULL
> +#endif
> +
>  static int
>  create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
>  		unsigned long load_addr, unsigned long interp_load_addr)
> @@ -142,7 +151,9 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
>  	elf_addr_t __user *envp;
>  	elf_addr_t __user *sp;
>  	elf_addr_t __user *u_platform;
> +	elf_addr_t __user *u_base_platform;
>  	const char *k_platform = ELF_PLATFORM;
> +	const char *k_base_platform = ELF_BASE_PLATFORM;
>  	int items;
>  	elf_addr_t *elf_info;
>  	int ei_index = 0;
> @@ -172,6 +183,19 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
>  			return -EFAULT;
>  	}
>  
> +	/*
> +	 * If this architecture has a "base" platform capability
> +	 * string, copy it to userspace.
> +	 */
> +	u_base_platform = NULL;
> +	if (k_base_platform) {
> +		size_t len = strlen(k_base_platform) + 1;
> +
> +		u_base_platform = (elf_addr_t __user *)STACK_ALLOC(p, len);
> +		if (__copy_to_user(u_base_platform, k_base_platform, len))
> +			return -EFAULT;
> +	}
> +
>  	/* Create the ELF interpreter info */
>  	elf_info = (elf_addr_t *)current->mm->saved_auxv;
>  	/* update AT_VECTOR_SIZE_BASE if the number of NEW_AUX_ENT() changes */
> @@ -209,6 +233,10 @@ create_elf_tables(struct linux_binprm *bprm, struct elfhdr *exec,
>  		NEW_AUX_ENT(AT_PLATFORM,
>  			    (elf_addr_t)(unsigned long)u_platform);
>  	}
> +	if (k_base_platform) {
> +		NEW_AUX_ENT(AT_BASE_PLATFORM,
> +			    (elf_addr_t)(unsigned long)u_base_platform);
> +	}
>  	if (bprm->interp_flags & BINPRM_FLAGS_EXECFD) {
>  		NEW_AUX_ENT(AT_EXECFD, bprm->interp_data);
>  	}
> diff --git a/include/linux/auxvec.h b/include/linux/auxvec.h
> index 0da17d1..d7afa9d 100644
> --- a/include/linux/auxvec.h
> +++ b/include/linux/auxvec.h
> @@ -26,9 +26,13 @@
>  
>  #define AT_SECURE 23   /* secure mode boolean */
>  
> +#define AT_BASE_PLATFORM 24	/* string identifying real platform, may
> +				 * differ from AT_PLATFORM. */
> +
>  #define AT_EXECFN  31	/* filename of program */
> +
>  #ifdef __KERNEL__
> -#define AT_VECTOR_SIZE_BASE 17 /* NEW_AUX_ENT entries in auxiliary table */
> +#define AT_VECTOR_SIZE_BASE 18 /* NEW_AUX_ENT entries in auxiliary table */
>    /* number of "#define AT_.*" above, minus {AT_NULL, AT_IGNORE, AT_NOTELF} */
>  #endif
>  

Re: [PATCH] elf loader support for auxvec base platform string

[Linus Torvalds]

On Thu, 17 Jul 2008, Benjamin Herrenschmidt wrote:
> 
> Should I seek somebody's ack before merging a patch like the one below ?
> 
> I'm a bit reluctant to merge via the powerpc.git tree some changes to
> generic files without at least an ack from somebody else :-)

Gaah. Generally we don't, but you're right to ask. The ELF loading keeps 
on accumulating these things, and I'm not sure we have the right process 
for things like this. They're all individually small and insignificant, 
and they are all individually never going away and making the ELF loader 
messier and messier.

I dunno. 

		Linus

Re: [PATCH] elf loader support for auxvec base platform string

[Nathan Lynch]

Linus Torvalds wrote:
> 
> 
> On Thu, 17 Jul 2008, Benjamin Herrenschmidt wrote:
> > 
> > Should I seek somebody's ack before merging a patch like the one below ?
> > 
> > I'm a bit reluctant to merge via the powerpc.git tree some changes to
> > generic files without at least an ack from somebody else :-)
> 
> Gaah. Generally we don't, but you're right to ask. The ELF loading keeps 
> on accumulating these things, and I'm not sure we have the right process 
> for things like this. They're all individually small and insignificant, 
> and they are all individually never going away and making the ELF loader 
> messier and messier.

FWIW, I was initially reluctant to do it this way, but the ELF aux
vector seems to be the only reasonable mechanism for getting this
information to all interested users.  And the only reason the patch
touches the generic code is because we have to copy a string to
userspace (just like AT_PLATFORM).  Otherwise it could be done in
powerpc's ARCH_DLINFO.

Re: [PATCH] elf loader support for auxvec base platform string

[Benjamin Herrenschmidt]

On Thu, 2008-07-17 at 09:10 -0700, Linus Torvalds wrote:
> 
> On Thu, 17 Jul 2008, Benjamin Herrenschmidt wrote:
> > 
> > Should I seek somebody's ack before merging a patch like the one below ?
> > 
> > I'm a bit reluctant to merge via the powerpc.git tree some changes to
> > generic files without at least an ack from somebody else :-)
> 
> Gaah. Generally we don't, but you're right to ask. The ELF loading keeps 
> on accumulating these things, and I'm not sure we have the right process 
> for things like this. They're all individually small and insignificant, 
> and they are all individually never going away and making the ELF loader 
> messier and messier.

Yeah, annoying. Oh well, this one at least is now getting Andrew's
scrutinity...

Cheers,
Ben.

[PATCH] enable AT_BASE_PLATFORM aux vector for powerpc

[Nathan Lynch]

Stash the first platform string matched by identify_cpu() in
powerpc_base_platform, and supply that to the ELF loader for the value
of AT_BASE_PLATFORM.

Signed-off-by: Nathan Lynch <ntl@pobox.com>
---
 arch/powerpc/kernel/cputable.c |   11 +++++++++++
 include/asm-powerpc/cputable.h |    2 ++
 include/asm-powerpc/elf.h      |    8 ++++++++
 3 files changed, 21 insertions(+), 0 deletions(-)

diff --git a/arch/powerpc/kernel/cputable.c b/arch/powerpc/kernel/cputable.c
index f7f3c21..89d8731 100644
--- a/arch/powerpc/kernel/cputable.c
+++ b/arch/powerpc/kernel/cputable.c
@@ -23,6 +23,9 @@
 struct cpu_spec* cur_cpu_spec = NULL;
 EXPORT_SYMBOL(cur_cpu_spec);
 
+/* The platform string corresponding to the real PVR */
+const char *powerpc_base_platform;
+
 /* NOTE:
  * Unlike ppc32, ppc64 will only call this once for the boot CPU, it's
  * the responsibility of the appropriate CPU save/restore functions to
@@ -1632,6 +1635,14 @@ struct cpu_spec * __init identify_cpu(unsigned long offset, unsigned int pvr)
 			} else
 				*t = *s;
 			*PTRRELOC(&cur_cpu_spec) = &the_cpu_spec;
+
+			/*
+			 * Set the base platform string once; assumes
+			 * we're called with real pvr first.
+			 */
+			if (powerpc_base_platform == NULL)
+				powerpc_base_platform = t->platform;
+
 #if defined(CONFIG_PPC64) || defined(CONFIG_BOOKE)
 			/* ppc64 and booke expect identify_cpu to also call
 			 * setup_cpu for that processor. I will consolidate
diff --git a/include/asm-powerpc/cputable.h b/include/asm-powerpc/cputable.h
index 2a3e907..ef8a248 100644
--- a/include/asm-powerpc/cputable.h
+++ b/include/asm-powerpc/cputable.h
@@ -127,6 +127,8 @@ extern struct cpu_spec *identify_cpu(unsigned long offset, unsigned int pvr);
 extern void do_feature_fixups(unsigned long value, void *fixup_start,
 			      void *fixup_end);
 
+extern const char *powerpc_base_platform;
+
 #endif /* __ASSEMBLY__ */
 
 /* CPU kernel features */
diff --git a/include/asm-powerpc/elf.h b/include/asm-powerpc/elf.h
index 8966467..80d1f39 100644
--- a/include/asm-powerpc/elf.h
+++ b/include/asm-powerpc/elf.h
@@ -217,6 +217,14 @@ typedef elf_vrregset_t elf_fpxregset_t;
 
 #define ELF_PLATFORM	(cur_cpu_spec->platform)
 
+/* While ELF_PLATFORM indicates the ISA supported by the platform, it
+ * may not accurately reflect the underlying behavior of the hardware
+ * (as in the case of running in Power5+ compatibility mode on a
+ * Power6 machine).  ELF_BASE_PLATFORM allows ld.so to load libraries
+ * that are tuned for the real hardware.
+ */
+#define ELF_BASE_PLATFORM (powerpc_base_platform)
+
 #ifdef __powerpc64__
 # define ELF_PLAT_INIT(_r, load_addr)	do {	\
 	_r->gpr[2] = load_addr; 		\
-- 
1.5.6.2