Break ELF_PLATFORM and stack pointer randomization dependency

Currently arch_align_stack() is used by fs/binfmt_elf.c to randomize
stack pointer inside a page. But this happens only if ELF_PLATFORM
symbol is defined.

ELF_PLATFORM is normally set if the architecture wants ld.so to load
implementation specific libraries for optimization. And currently a
lot of architectures just yield this symbol to NULL.

This is the case for MIPS architecture where ELF_PLATFORM is NULL but
arch_align_stack() has been redefined to do stack inside page
randomization. So in this case no randomization is actually done.

This patch breaks this dependency which seems to be useless and allows
platforms such MIPS to do the randomization.

Signed-off-by: Franck Bui-Huu <[e-mail]>
Cc: Ingo Molnar <[e-mail]>
Cc: Arjan van de Ven <[e-mail]>
Cc: Ralf Baechle <[e-mail]>
Signed-off-by: Andrew Morton <[e-mail]>
Signed-off-by: Linus Torvalds <[e-mail]>

fs/binfmt_elf.c

···
 	struct vm_area_struct *vma;
 
 	/*
+	 * In some cases (e.g. Hyper-Threading), we want to avoid L1
+	 * evictions by the processes running on the same package. One
+	 * thing we can do is to shuffle the initial stack for them.
+	 */
+
+	p = arch_align_stack(p);
+
+	/*
 	 * If this architecture has a platform capability string, copy it
 	 * to userspace.  In some cases (Sparc), this info is impossible
 	 * for userspace to get any other way, in others (i386) it is
···
 	u_platform = NULL;
 	if (k_platform) {
 		size_t len = strlen(k_platform) + 1;
-
-		/*
-		 * In some cases (e.g. Hyper-Threading), we want to avoid L1
-		 * evictions by the processes running on the same package. One
-		 * thing we can do is to shuffle the initial stack for them.
-		 */
-
-		p = arch_align_stack(p);
 
 		u_platform = (elf_addr_t __user *)STACK_ALLOC(p, len);
 		if (__copy_to_user(u_platform, k_platform, len))