[PATCH 1/2] MIPS: Allow the auxv's elf_platform entry to be set.

[PATCH 1/2] MIPS: Allow the auxv's elf_platform entry to be set.

[David Daney]

The userspace runtime linker uses the elf_platform to find the
libraries optimized for the current CPU archecture variant.  First we
need to allow it to be set to something other than NULL.  Follow-on
patches will set some values for specific CPUs.

GLIBC already does the right thing.  The kernel just needs to supply
good data.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
---
 arch/mips/include/asm/elf.h  |   14 +++++++-------
 arch/mips/kernel/cpu-probe.c |    1 +
 2 files changed, 8 insertions(+), 7 deletions(-)

diff --git a/arch/mips/include/asm/elf.h b/arch/mips/include/asm/elf.h
index 9266f02..1c3dbf0 100644
--- a/arch/mips/include/asm/elf.h
+++ b/arch/mips/include/asm/elf.h
@@ -334,14 +334,14 @@ extern int dump_task_fpu(struct task_struct *, elf_fpregset_t *);
 
 #define ELF_HWCAP       (0)
 
-/* This yields a string that ld.so will use to load implementation
-   specific libraries for optimization.  This is more specific in
-   intent than poking at uname or /proc/cpuinfo.
-
-   For the moment, we have only optimizations for the Intel generations,
-   but that could change... */
+/*
+ * This yields a string that ld.so will use to load implementation
+ * specific libraries for optimization.  This is more specific in
+ * intent than poking at uname or /proc/cpuinfo.
+ */
 
-#define ELF_PLATFORM  (NULL)
+#define ELF_PLATFORM  __elf_platform
+extern const char *__elf_platform;
 
 /*
  * See comments in asm-alpha/elf.h, this is the same thing
diff --git a/arch/mips/kernel/cpu-probe.c b/arch/mips/kernel/cpu-probe.c
index aa25fd0..91ba1c8 100644
--- a/arch/mips/kernel/cpu-probe.c
+++ b/arch/mips/kernel/cpu-probe.c
@@ -917,6 +917,7 @@ static inline void cpu_probe_cavium(struct cpuinfo_mips *c, unsigned int cpu)
 }
 
 const char *__cpu_name[NR_CPUS];
+const char *__elf_platform;
 
 __cpuinit void cpu_probe(void)
 {
-- 
1.6.0.6

[PATCH 2/2] MIPS: Set __elf_platform for Octeon.

[David Daney]

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
---
 arch/mips/kernel/cpu-probe.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/arch/mips/kernel/cpu-probe.c b/arch/mips/kernel/cpu-probe.c
index 91ba1c8..906a2d1 100644
--- a/arch/mips/kernel/cpu-probe.c
+++ b/arch/mips/kernel/cpu-probe.c
@@ -908,6 +908,8 @@ static inline void cpu_probe_cavium(struct cpuinfo_mips *c, unsigned int cpu)
 	case PRID_IMP_CAVIUM_CN52XX:
 		c->cputype = CPU_CAVIUM_OCTEON;
 		__cpu_name[cpu] = "Cavium Octeon";
+		if (cpu == 0)
+			__elf_platform = "octeon";
 		break;
 	default:
 		printk(KERN_INFO "Unknown Octeon chip!\n");
-- 
1.6.0.6

Re: [PATCH 1/2] MIPS: Allow the auxv's elf_platform entry to be set.

[Ralf Baechle]

On Thu, Jan 28, 2010 at 04:52:12PM -0800, David Daney wrote:

> The userspace runtime linker uses the elf_platform to find the
> libraries optimized for the current CPU archecture variant.  First we
> need to allow it to be set to something other than NULL.  Follow-on
> patches will set some values for specific CPUs.
> 
> GLIBC already does the right thing.  The kernel just needs to supply
> good data.

Thanks, queued for 2.6.34.

  Ralf

Re: [PATCH 2/2] MIPS: Set __elf_platform for Octeon.

[Ralf Baechle]

On Thu, Jan 28, 2010 at 04:52:13PM -0800, David Daney wrote:

Thanks, queued for 2.6.34.

  Ralf