Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1710825 > unrolled thread

[PATCH 0/5] add support for relative references in special sections

Back to article view | Back to linux.kernel

Contents

  [PATCH 0/5] add support for relative references in special sections Ard Biesheuvel <ard.biesheuvel@linaro.org> - 2017-08-14 13:00 +0200
    [PATCH 2/5] module: use relative references for __ksymtab entries Ard Biesheuvel <ard.biesheuvel@linaro.org> - 2017-08-14 13:00 +0200
    Re: [PATCH 0/5] add support for relative references in special  sections Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com> - 2017-08-18 08:00 +0200
      Re: [PATCH 0/5] add support for relative references in special sections Ard Biesheuvel <ard.biesheuvel@linaro.org> - 2017-08-18 08:20 +0200
        Re: [PATCH 0/5] add support for relative references in special  sections Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com> - 2017-08-18 08:30 +0200
          Re: [PATCH 0/5] add support for relative references in special sections Ard Biesheuvel <ard.biesheuvel@linaro.org> - 2017-08-18 08:40 +0200

#1710825 — [PATCH 0/5] add support for relative references in special sections

This adds support for emitting special sections such as initcall arrays,
PCI fixups and tracepoints as relative references rather than absolute
references. This reduces the size by 50% on 64-bit architectures, but
more importantly, it removes the need for carrying relocation metadata
for these sections in relocatables kernels (e.g., for KASLR) that need
to fix up these absolute references at boot time. On arm64, this reduces
the vmlinux footprint of such a reference by 8x (8 byte absolute reference
+ 24 byte RELA entry vs 4 byte relative reference)

Patch #2 was sent out before as a single patch. This series supersedes
the previous submission. This version makes relative ksymtab entries
dependent on the new Kconfig symbol HAVE_ARCH_PREL32_RELOCATIONS rather
than trying to infer from kbuild test robot replies for which architectures
it should be blacklisted.

Patch #1 introduces the new Kconfig symbol HAVE_ARCH_PREL32_RELOCATIONS,
and sets it for the main architectures that are expected to benefit most
from this feature, i.e., 64-bit architectures, and ones that use runtime
relocation.

Patches #3 - #5 implement relative references for initcallls, PCI fixups
and tracepoints, respectively, all of which produce sections with order
~1000 entries on an arm64 defconfig kernel with tracing enabled. This
means we save about 28 KB of vmlinux space for each of these patches.

For the arm64 kernel, all patches combined reduce the size of vmlinux
by about 300 KB.

Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Thomas Garnier <thgarnie@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "Serge E. Hallyn" <serge@hallyn.com>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Petr Mladek <pmladek@suse.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Morris <james.l.morris@oracle.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Nicolas Pitre <nico@linaro.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Jessica Yu <jeyu@kernel.org>

Ard Biesheuvel (5):
  arch: enable relative relocations for arm64, power, x86, s390 and x86
  module: use relative references for __ksymtab entries
  init: allow initcall tables to be emitted using relative references
  drivers: pci: add support for relative addressing in quirk tables
  kernel: tracepoints: add support for relative references

 arch/Kconfig                    | 10 +++++
 arch/arm64/Kconfig              |  1 +
 arch/arm64/kernel/vmlinux.lds.S |  2 +-
 arch/powerpc/Kconfig            |  1 +
 arch/s390/Kconfig               |  1 +
 arch/x86/Kconfig                |  1 +
 arch/x86/include/asm/Kbuild     |  1 +
 arch/x86/include/asm/export.h   |  4 --
 drivers/pci/quirks.c            | 13 ++++--
 include/asm-generic/export.h    | 12 ++++-
 include/linux/compiler.h        | 11 +++++
 include/linux/export.h          | 47 +++++++++++++++-----
 include/linux/init.h            | 44 +++++++++++++-----
 include/linux/pci.h             | 20 +++++++++
 include/linux/tracepoint.h      | 19 ++++++--
 init/main.c                     | 32 ++++++-------
 kernel/module.c                 | 33 +++++++++++---
 kernel/printk/printk.c          |  4 +-
 kernel/tracepoint.c             | 19 +++++---
 security/security.c             |  4 +-
 20 files changed, 212 insertions(+), 67 deletions(-)
 delete mode 100644 arch/x86/include/asm/export.h

-- 
2.11.0

[toc] | [next] | [standalone]

#1710826 — [PATCH 2/5] module: use relative references for __ksymtab entries

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 dependend on whether
CONFIG_HAVE_ARCH_PREL32_RELOCATIONS is defined.

Cc: Jessica Yu <jeyu@kernel.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Thomas Garnier <thgarnie@google.com>
Cc: Nicolas Pitre <nico@linaro.org>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
---
 arch/x86/include/asm/Kbuild   |  1 +
 arch/x86/include/asm/export.h |  4 --
 include/asm-generic/export.h  | 12 ++++-
 include/linux/compiler.h      | 11 +++++
 include/linux/export.h        | 47 +++++++++++++++-----
 kernel/module.c               | 33 +++++++++++---
 6 files changed, 85 insertions(+), 23 deletions(-)

diff --git a/arch/x86/include/asm/Kbuild b/arch/x86/include/asm/Kbuild
index 5d6a53fd7521..3e8a88dcaa1d 100644
--- a/arch/x86/include/asm/Kbuild
+++ b/arch/x86/include/asm/Kbuild
@@ -9,5 +9,6 @@ generated-y += xen-hypercalls.h
 generic-y += clkdev.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 138de56b13eb..000000000000
--- a/arch/x86/include/asm/export.h
+++ /dev/null
@@ -1,4 +0,0 @@
-#ifdef CONFIG_64BIT
-#define KSYM_ALIGN 16
-#endif
-#include <asm-generic/export.h>
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 eca8ad75e28b..5644d2e653f0 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -590,4 +590,15 @@ static __always_inline void __write_once_size(volatile void *p, void *res, int s
 	(_________p1); \
 })
 
+/*
+ * 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
+ * visibile to the compiler.
+ */
+#define __ADDRESSABLE(sym) \
+	static void * __attribute__((section(".discard.text"), used))	\
+		__PASTE(__discard_##sym,__LINE__)(void)			\
+			{ return (void *)&sym; }			\
+
 #endif /* __LINUX_COMPILER_H */
diff --git a/include/linux/export.h b/include/linux/export.h
index 1a1dfdb2a5c6..986d02e57253 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)
@@ -60,17 +54,48 @@ extern struct module __this_module;
 #define __CRC_SYMBOL(sym, sec)
 #endif
 
+#ifdef CONFIG_HAVE_ARCH_PREL32_RELOCATIONS
+/*
+ * Emit the ksymtab entry as a pair of relative references: this reduces
+ * the size by half on 64-bit architectures, and eliminates the need for
+ * absolute relocations that require runtime processing on relocatable
+ * kernels.
+ */
+#define __KSYMTAB_ENTRY(sym, sec)					\
+	__ADDRESSABLE(sym)						\
+	asm("	.section \"___ksymtab" sec "+" #sym "\", \"a\"	\n"	\
+	    "	.balign	8					\n"	\
+	    VMLINUX_SYMBOL_STR(__ksymtab_##sym) ":		\n"	\
+	    "	.long "	VMLINUX_SYMBOL_STR(sym) "- .		\n"	\
+	    "	.long "	VMLINUX_SYMBOL_STR(__kstrtab_##sym) "- .\n"	\
+	    "	.previous					\n")
+
+struct kernel_symbol
+{
+	signed int value_offset;
+	signed int name_offset;
+};
+#else
+#define __KSYMTAB_ENTRY(sym, sec)					\
+	static const struct kernel_symbol __ksymtab_##sym		\
+	__attribute__((section("___ksymtab" sec "+" #sym), used))	\
+	= { (unsigned long)&sym, __kstrtab_##sym }
+
+struct kernel_symbol
+{
+	unsigned long value;
+	const char *name;
+};
+#endif
+
 /* For every exported symbol, place a struct in the __ksymtab section */
 #define ___EXPORT_SYMBOL(sym, sec)					\
 	extern typeof(sym) sym;						\
 	__CRC_SYMBOL(sym, sec)						\
 	static const char __kstrtab_##sym[]				\
-	__attribute__((section("__ksymtab_strings"), aligned(1)))	\
+	__attribute__((section("__ksymtab_strings"), used, aligned(1)))	\
 	= VMLINUX_SYMBOL_STR(sym);					\
-	static const struct kernel_symbol __ksymtab_##sym		\
-	__used								\
-	__attribute__((section("___ksymtab" sec "+" #sym), used))	\
-	= { (unsigned long)&sym, __kstrtab_##sym }
+	__KSYMTAB_ENTRY(sym, sec)
 
 #if defined(__KSYM_DEPS__)
 
diff --git a/kernel/module.c b/kernel/module.c
index 40f983cbea81..a45423dcc32d 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -539,12 +539,31 @@ static bool check_symbol(const struct symsearch *syms,
 	return true;
 }
 
+static unsigned long kernel_symbol_value(const struct kernel_symbol *sym)
+{
+#ifdef CONFIG_HAVE_ARCH_PREL32_RELOCATIONS
+	return (unsigned long)&sym->value_offset + sym->value_offset;
+#else
+	return sym->value;
+#endif
+}
+
+static const char *kernel_symbol_name(const struct kernel_symbol *sym)
+{
+#ifdef CONFIG_HAVE_ARCH_PREL32_RELOCATIONS
+	return (const char *)((unsigned long)&sym->name_offset +
+			      sym->name_offset);
+#else
+	return sym->name;
+#endif
+}
+
 static int cmp_name(const void *va, const void *vb)
 {
 	const char *a;
 	const struct kernel_symbol *b;
 	a = va; b = vb;
-	return strcmp(a, b->name);
+	return strcmp(a, kernel_symbol_name(b));
 }
 
 static bool find_symbol_in_section(const struct symsearch *syms,
@@ -2190,7 +2209,7 @@ void *__symbol_get(const char *symbol)
 		sym = NULL;
 	preempt_enable();
 
-	return sym ? (void *)sym->value : NULL;
+	return sym ? (void *)kernel_symbol_value(sym) : NULL;
 }
 EXPORT_SYMBOL_GPL(__symbol_get);
 
@@ -2220,10 +2239,12 @@ static int verify_export_symbols(struct module *mod)
 
 	for (i = 0; i < ARRAY_SIZE(arr); i++) {
 		for (s = arr[i].sym; s < arr[i].sym + arr[i].num; s++) {
-			if (find_symbol(s->name, &owner, NULL, true, false)) {
+			if (find_symbol(kernel_symbol_name(s), &owner, NULL,
+					true, false)) {
 				pr_err("%s: exports duplicate symbol %s"
 				       " (owned by %s)\n",
-				       mod->name, s->name, module_name(owner));
+				       mod->name, kernel_symbol_name(s),
+				       module_name(owner));
 				return -ENOEXEC;
 			}
 		}
@@ -2272,7 +2293,7 @@ static int simplify_symbols(struct module *mod, const struct load_info *info)
 			ksym = resolve_symbol_wait(mod, info, name);
 			/* Ok if resolved.  */
 			if (ksym && !IS_ERR(ksym)) {
-				sym[i].st_value = ksym->value;
+				sym[i].st_value = kernel_symbol_value(ksym);
 				break;
 			}
 
@@ -2532,7 +2553,7 @@ static int is_exported(const char *name, unsigned long value,
 		ks = lookup_symbol(name, __start___ksymtab, __stop___ksymtab);
 	else
 		ks = lookup_symbol(name, mod->syms, mod->syms + mod->num_syms);
-	return ks != NULL && ks->value == value;
+	return ks != NULL && kernel_symbol_value(ks) == value;
 }
 
 /* As per nm */
-- 
2.11.0

[toc] | [prev] | [next] | [standalone]

#1714589 — Re: [PATCH 0/5] add support for relative references in special sections

On (08/14/17 11:52), Ard Biesheuvel wrote:
> This adds support for emitting special sections such as initcall arrays,
> PCI fixups and tracepoints as relative references rather than absolute
> references. This reduces the size by 50% on 64-bit architectures, but
> more importantly, it removes the need for carrying relocation metadata
> for these sections in relocatables kernels (e.g., for KASLR) that need
> to fix up these absolute references at boot time. On arm64, this reduces
> the vmlinux footprint of such a reference by 8x (8 byte absolute reference
> + 24 byte RELA entry vs 4 byte relative reference)
[..]

a side note,
checkpatch complaints quite a lot.

	-ss

[toc] | [prev] | [next] | [standalone]

#1714597

Hi Sergey,

Thanks for taking a look

On 18 August 2017 at 06:56, Sergey Senozhatsky
<sergey.senozhatsky.work@gmail.com> wrote:
> On (08/14/17 11:52), Ard Biesheuvel wrote:
>> This adds support for emitting special sections such as initcall arrays,
>> PCI fixups and tracepoints as relative references rather than absolute
>> references. This reduces the size by 50% on 64-bit architectures, but
>> more importantly, it removes the need for carrying relocation metadata
>> for these sections in relocatables kernels (e.g., for KASLR) that need
>> to fix up these absolute references at boot time. On arm64, this reduces
>> the vmlinux footprint of such a reference by 8x (8 byte absolute reference
>> + 24 byte RELA entry vs 4 byte relative reference)
> [..]
>
> a side note,
> checkpatch complaints quite a lot.
>

Yeah, fair point. Many of them are debatable or completely bogus, though:

ERROR: "foo * bar" should be "foo *bar"
#114: FILE: include/linux/compiler.h:600:
+ static void * __attribute__((section(".discard.text"), used)) \

I think it is rather common to keep whitespace between * and what
follows if it is not the identifier.

ERROR: Macros with complex values should be enclosed in parentheses
#147: FILE: include/linux/export.h:64:
+#define __KSYMTAB_ENTRY(sym, sec) \
+ __ADDRESSABLE(sym) \
+ asm(" .section \"___ksymtab" sec "+" #sym "\", \"a\" \n" \
+    " .balign 8 \n" \
+    VMLINUX_SYMBOL_STR(__ksymtab_##sym) ": \n" \
+    " .long " VMLINUX_SYMBOL_STR(sym) "- . \n" \
+    " .long " VMLINUX_SYMBOL_STR(__kstrtab_##sym) "- .\n" \
+    " .previous \n")

WARNING: do not add new typedefs
#29: FILE: include/linux/init.h:114:
+typedef signed int initcall_entry_t;

WARNING: do not add new typedefs
#36: FILE: include/linux/init.h:121:
+typedef initcall_t initcall_entry_t;

This is a typedef that accompanies the existing typedef for
initcall_t, and it is the only way to parameterise this code without a
ton of changes to duplicate all extern declarations of initcall
arrays.

ERROR: Macros with multiple statements should be enclosed in a do - while loop
#55: FILE: include/linux/init.h:181:
+#define ___define_initcall(fn, id, __sec) \
+ __ADDRESSABLE(fn) \
+ asm(".section \"" #__sec ".init\", \"a\" \n" \
+ "__initcall_" #fn #id ": \n" \
+    ".long " VMLINUX_SYMBOL_STR(fn) " - . \n" \
+    ".previous \n");

This one is bogus. This macro is only used in file scope.

WARNING: externs should be avoided in .c files
#109: FILE: init/main.c:837:
+extern initcall_entry_t __initcall0_start[];

This changes the type of the existing initcall array declarations, and
I don't think moving them to a .h file for no other reason than to
appease checkpatch is pointless.

WARNING: unnecessary whitespace before a quoted newline
#63: FILE: include/linux/pci.h:1759:
+ asm(".section " #sec ", \"a\" \n" \

WARNING: unnecessary whitespace before a quoted newline
#64: FILE: include/linux/pci.h:1760:
+    ".balign 16 \n" \

WARNING: unnecessary whitespace before a quoted newline
#65: FILE: include/linux/pci.h:1761:
+    ".short " #vendor ", " #device " \n" \

These are not newlines that end up as strings in the program, and I
think it makes little sense to make the inline asm less readable by
putting the \n right after the text.

WARNING: braces {} are not necessary for single statement blocks
#83: FILE: kernel/tracepoint.c:515:
+ for (iter = begin; iter < (signed int *)end; iter++) {
+ fct((struct tracepoint *)((unsigned long)iter + *iter), priv);
+ }

This simply mirrors the existing code in the other branch of the if ()

I will clean up the meaningful ones in v2, but please don't expect
this series to be checkpatch clean: it simply doesn't deal with inline
asm very well, and some of this code predates checkpatch by a decade,
and I'd rather not mix up rather tricky functional changes with
checkpatch cleanup duty.

-- 
Ard.

[toc] | [prev] | [next] | [standalone]

#1714608 — Re: [PATCH 0/5] add support for relative references in special sections

Hi Ard,

On (08/18/17 07:12), Ard Biesheuvel wrote:
> Hi Sergey,
> 
> Thanks for taking a look
> 
> On 18 August 2017 at 06:56, Sergey Senozhatsky
> <sergey.senozhatsky.work@gmail.com> wrote:
> > On (08/14/17 11:52), Ard Biesheuvel wrote:
> >> This adds support for emitting special sections such as initcall arrays,
> >> PCI fixups and tracepoints as relative references rather than absolute
> >> references. This reduces the size by 50% on 64-bit architectures, but
> >> more importantly, it removes the need for carrying relocation metadata
> >> for these sections in relocatables kernels (e.g., for KASLR) that need
> >> to fix up these absolute references at boot time. On arm64, this reduces
> >> the vmlinux footprint of such a reference by 8x (8 byte absolute reference
> >> + 24 byte RELA entry vs 4 byte relative reference)
> > [..]
> >
> > a side note,
> > checkpatch complaints quite a lot.
> >
[..]
> I will clean up the meaningful ones in v2, but please don't expect
> this series to be checkpatch clean: it simply doesn't deal with inline
> asm very well, and some of this code predates checkpatch by a decade,
> and I'd rather not mix up rather tricky functional changes with
> checkpatch cleanup duty.

sure. thanks.

I'm running two x86 boxes with the patch set applied, for
several days, with no issues being observed. it does save
some memory (well, several pages in my case) even on "tiny"
kernels configs.

	-ss

[toc] | [prev] | [next] | [standalone]

#1714610

On 18 August 2017 at 07:29, Sergey Senozhatsky
<sergey.senozhatsky.work@gmail.com> wrote:
> Hi Ard,
>
> On (08/18/17 07:12), Ard Biesheuvel wrote:
>> Hi Sergey,
>>
>> Thanks for taking a look
>>
>> On 18 August 2017 at 06:56, Sergey Senozhatsky
>> <sergey.senozhatsky.work@gmail.com> wrote:
>> > On (08/14/17 11:52), Ard Biesheuvel wrote:
>> >> This adds support for emitting special sections such as initcall arrays,
>> >> PCI fixups and tracepoints as relative references rather than absolute
>> >> references. This reduces the size by 50% on 64-bit architectures, but
>> >> more importantly, it removes the need for carrying relocation metadata
>> >> for these sections in relocatables kernels (e.g., for KASLR) that need
>> >> to fix up these absolute references at boot time. On arm64, this reduces
>> >> the vmlinux footprint of such a reference by 8x (8 byte absolute reference
>> >> + 24 byte RELA entry vs 4 byte relative reference)
>> > [..]
>> >
>> > a side note,
>> > checkpatch complaints quite a lot.
>> >
> [..]
>> I will clean up the meaningful ones in v2, but please don't expect
>> this series to be checkpatch clean: it simply doesn't deal with inline
>> asm very well, and some of this code predates checkpatch by a decade,
>> and I'd rather not mix up rather tricky functional changes with
>> checkpatch cleanup duty.
>
> sure. thanks.
>
> I'm running two x86 boxes with the patch set applied, for
> several days, with no issues being observed. it does save
> some memory (well, several pages in my case) even on "tiny"
> kernels configs.
>

That is good to hear. Thanks.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web