AT_BASE_PLATFORM (v2)

Linux Kernel Mailing List, post #230,129
Author:
Date:
Subject:
 Nathan Lynch
 2008-07-15 18:58:49
 AT_BASE_PLATFORM (v2)
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
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Author:
Date:
Subject:
 Nathan Lynch
 2008-07-15 18:58:50
 [PATCH] elf loader support for auxvec base platform string
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 <[email protected]>
---
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

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Author:
Date:
Subject:
 Benjamin Herrenschmidt
 2008-07-17 16:35:39
 Re: [PATCH] elf loader support for auxvec base platform string
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 <[email protected]>
> ---
> 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 */

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/