Re: [PATCH v8 3/6] module: use relative references for __ksymtab entries

2018-03-11 Thread Ard Biesheuvel
On 11 March 2018 at 12:38, Ard Biesheuvel  wrote:
> An ordinary arm64 defconfig build has ~64 KB worth of __ksymtab
> entries, each consisting of two 64-bit fields containing absolute
> references, to the symbol itself and to a char array containing
> its name, respectively.
>
> When we build the same configuration with KASLR enabled, we end
> up with an additional ~192 KB of relocations in the .init section,
> i.e., one 24 byte entry for each absolute reference, which all need
> to be processed at boot time.
>
> Given how the struct kernel_symbol that describes each entry is
> completely local to module.c (except for the references emitted
> by EXPORT_SYMBOL() itself), we can easily modify it to contain
> two 32-bit relative references instead. This reduces the size of
> the __ksymtab section by 50% for all 64-bit architectures, and
> gets rid of the runtime relocations entirely for architectures
> implementing KASLR, either via standard PIE linking (arm64) or
> using custom host tools (x86).
>
> Note that the binary search involving __ksymtab contents relies
> on each section being sorted by symbol name. This is implemented
> based on the input section names, not the names in the ksymtab
> entries, so this patch does not interfere with that.
>
> Given that the use of place-relative relocations requires support
> both in the toolchain and in the module loader, we cannot enable
> this feature for all architectures. So make it dependent on whether
> CONFIG_HAVE_ARCH_PREL32_RELOCATIONS is defined.
>
> Cc: Arnd Bergmann 
> Cc: Andrew Morton 
> Cc: Ingo Molnar 
> Cc: Kees Cook 
> Cc: Thomas Garnier 
> Cc: Nicolas Pitre 
> Acked-by: Jessica Yu 
> Signed-off-by: Ard Biesheuvel 
> ---
>  arch/x86/include/asm/Kbuild   |  1 +
>  arch/x86/include/asm/export.h |  5 ---
>  include/asm-generic/export.h  | 12 -
>  include/linux/compiler.h  | 19 
>  include/linux/export.h| 46 +++-
>  kernel/module.c   | 32 +++---
>  6 files changed, 91 insertions(+), 24 deletions(-)
>
...
> diff --git a/include/linux/compiler.h b/include/linux/compiler.h
> index ab4711c63601..0a9328ea9dbd 100644
> --- a/include/linux/compiler.h
> +++ b/include/linux/compiler.h
> @@ -280,6 +280,25 @@ unsigned long read_word_at_a_time(const void *addr)
>
>  #endif /* __KERNEL__ */
>
> +/*
> + * Force the compiler to emit 'sym' as a symbol, so that we can reference
> + * it from inline assembler. Necessary in case 'sym' could be inlined
> + * otherwise, or eliminated entirely due to lack of references that are
> + * visible to the compiler.
> + */
> +#define __ADDRESSABLE(sym) \
> +   static void * const __attribute__((section(".discard"), used))  \
> +   __PASTE(__addressable_##sym, __LINE__) = (void *)
> +

kernelci.org tells me that I need to drop the 'const' here, or we may
end up with .discard sections with conflicting attributes (r/o vs r/w)
in some cases (CONFIG_DEBUG_FORCE_WEAK_PER_CPU=y)


[PATCH v8 3/6] module: use relative references for __ksymtab entries

2018-03-11 Thread Ard Biesheuvel
An ordinary arm64 defconfig build has ~64 KB worth of __ksymtab
entries, each consisting of two 64-bit fields containing absolute
references, to the symbol itself and to a char array containing
its name, respectively.

When we build the same configuration with KASLR enabled, we end
up with an additional ~192 KB of relocations in the .init section,
i.e., one 24 byte entry for each absolute reference, which all need
to be processed at boot time.

Given how the struct kernel_symbol that describes each entry is
completely local to module.c (except for the references emitted
by EXPORT_SYMBOL() itself), we can easily modify it to contain
two 32-bit relative references instead. This reduces the size of
the __ksymtab section by 50% for all 64-bit architectures, and
gets rid of the runtime relocations entirely for architectures
implementing KASLR, either via standard PIE linking (arm64) or
using custom host tools (x86).

Note that the binary search involving __ksymtab contents relies
on each section being sorted by symbol name. This is implemented
based on the input section names, not the names in the ksymtab
entries, so this patch does not interfere with that.

Given that the use of place-relative relocations requires support
both in the toolchain and in the module loader, we cannot enable
this feature for all architectures. So make it dependent on whether
CONFIG_HAVE_ARCH_PREL32_RELOCATIONS is defined.

Cc: Arnd Bergmann 
Cc: Andrew Morton 
Cc: Ingo Molnar 
Cc: Kees Cook 
Cc: Thomas Garnier 
Cc: Nicolas Pitre 
Acked-by: Jessica Yu 
Signed-off-by: Ard Biesheuvel 
---
 arch/x86/include/asm/Kbuild   |  1 +
 arch/x86/include/asm/export.h |  5 ---
 include/asm-generic/export.h  | 12 -
 include/linux/compiler.h  | 19 
 include/linux/export.h| 46 +++-
 kernel/module.c   | 32 +++---
 6 files changed, 91 insertions(+), 24 deletions(-)

diff --git a/arch/x86/include/asm/Kbuild b/arch/x86/include/asm/Kbuild
index de690c2d2e33..a0ab9ab61c75 100644
--- a/arch/x86/include/asm/Kbuild
+++ b/arch/x86/include/asm/Kbuild
@@ -8,5 +8,6 @@ generated-y += xen-hypercalls.h
 
 generic-y += dma-contiguous.h
 generic-y += early_ioremap.h
+generic-y += export.h
 generic-y += mcs_spinlock.h
 generic-y += mm-arch-hooks.h
diff --git a/arch/x86/include/asm/export.h b/arch/x86/include/asm/export.h
deleted file mode 100644
index 2a51d66689c5..
--- a/arch/x86/include/asm/export.h
+++ /dev/null
@@ -1,5 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 */
-#ifdef CONFIG_64BIT
-#define KSYM_ALIGN 16
-#endif
-#include 
diff --git a/include/asm-generic/export.h b/include/asm-generic/export.h
index 719db1968d81..97ce606459ae 100644
--- a/include/asm-generic/export.h
+++ b/include/asm-generic/export.h
@@ -5,12 +5,10 @@
 #define KSYM_FUNC(x) x
 #endif
 #ifdef CONFIG_64BIT
-#define __put .quad
 #ifndef KSYM_ALIGN
 #define KSYM_ALIGN 8
 #endif
 #else
-#define __put .long
 #ifndef KSYM_ALIGN
 #define KSYM_ALIGN 4
 #endif
@@ -25,6 +23,16 @@
 #define KSYM(name) name
 #endif
 
+.macro __put, val, name
+#ifdef CONFIG_HAVE_ARCH_PREL32_RELOCATIONS
+   .long   \val - ., \name - .
+#elif defined(CONFIG_64BIT)
+   .quad   \val, \name
+#else
+   .long   \val, \name
+#endif
+.endm
+
 /*
  * note on .section use: @progbits vs %progbits nastiness doesn't matter,
  * since we immediately emit into those sections anyway.
diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index ab4711c63601..0a9328ea9dbd 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -280,6 +280,25 @@ unsigned long read_word_at_a_time(const void *addr)
 
 #endif /* __KERNEL__ */
 
+/*
+ * Force the compiler to emit 'sym' as a symbol, so that we can reference
+ * it from inline assembler. Necessary in case 'sym' could be inlined
+ * otherwise, or eliminated entirely due to lack of references that are
+ * visible to the compiler.
+ */
+#define __ADDRESSABLE(sym) \
+   static void * const __attribute__((section(".discard"), used))  \
+   __PASTE(__addressable_##sym, __LINE__) = (void *)
+
+/**
+ * offset_to_ptr - convert a relative memory offset to an absolute pointer
+ * @off:   the address of the 32-bit offset value
+ */
+static inline void *offset_to_ptr(const int *off)
+{
+   return (void *)((unsigned long)off + *off);
+}
+
 #endif /* __ASSEMBLY__ */
 
 #ifndef __optimize
diff --git a/include/linux/export.h b/include/linux/export.h
index 25005b55b079..04c78e6bfec9 100644
--- a/include/linux/export.h
+++ b/include/linux/export.h
@@ -24,12 +24,6 @@
 #define VMLINUX_SYMBOL_STR(x) __VMLINUX_SYMBOL_STR(x)
 
 #ifndef __ASSEMBLY__
-struct kernel_symbol
-{
-   unsigned long value;
-   const char *name;
-};
-
 #ifdef MODULE
 extern struct module __this_module;
 #define THIS_MODULE (&__this_module)
@@