[PATCH 0/6] Trim unused exported symbols

[PATCH 0/6] Trim unused exported symbols

[Nicolas Pitre]

The kernel and some modules make many symbols available for
other modules to use via EXPORT_SYMBOL() and variants. Depending
on the set of modules being selected in your kernel configuration,
many of those exported symbols might never be used.

One direct effect of EXPORT_SYMBOL() entries is to force the referenced
code and the code it references into the kernel binary with no
possibilityes for dead code elimination, eeven though the referenced
code is never executed.

This patch series provides the option to omit those exported symbols
from the kernel and modules that are never referenced by any of the
selected modules in the current kernel configuration.  In turns this
allows for optimizing the compiled code and reducing final binaryes'
size. When using LTO the binari size reduction is even more effective.

Also, by removing some entry points, it could be argued that this has
some security advantages by reducing the attack surface of the kernel.
But this is not my main motivation for doing this so I'll let security
experts judge the merit of this claim.

Here's some numbers showing the effect of this series.

Kernel v4.5-rc2 x86 defconfig:

$ size vmlinux
   text    data     bss     dec     hex filename
12362563        1856456 1101824 15320843         e9c70b vmlinux

$ wc -l Module.symvers
8806 Module.symvers

Kernel v4.5-rc2 x86 defconfig + CONFIG_TRIM_UNUSED_EXPSYMS=y:

$ size vmlinux
   text    data     bss     dec     hex filename
12059848        1856456 1101824 15018128         e52890 vmlinux

$ wc -l Module.symvers
225 Module.symvers

Because the x86 defconfig only contains 18 modules, the number of
needed exported symbols is only 225 out of a possible 8806 for this
configuration.  The kernel text size shrank by about 2.4%.

More numbers, on ARM this time.

Kernel v4.5-rc2 arm defconfig (same as multi_v7_defconfig):

$ size vmlinux
   text    data     bss     dec     hex filename
13664222        1554068  351368 15569658         ed92fa vmlinux

$ wc -l Module.symvers
10044 Module.symvers

Kernel v4.5-rc2 arm defconfig + CONFIG_TRIM_UNUSED_EXPSYMS=y:

$ size vmlinux
   text    data     bss     dec     hex filename
13255051        1554132  351240 15160423         e75467 vmlinux

$ wc -l Module.symvers
2703 Module.symvers

This time many more modules (279 of them) are part of the build
configuration. Still, only 2703 out of 10044 exported symbols are
required. And despite a smaller number of omitted exports, the kernel
shrank by 3%.

Now let's add LTO into the mix.

Kernel v4.5-rc2 arm defconfig + LTO:

$ size vmlinux
   text    data     bss     dec     hex filename
12813766        1538324  344356 14696446         e03ffe vmlinux

$ wc -l Module.symvers
8415 Module.symvers

Kernel v4.5-rc2 arm defconfig + LTO + CONFIG_TRIM_UNUSED_EXPSYMS=y:

$ size vmlinux
   text    data     bss     dec     hex filename
12197437        1536052  338660 14072149         d6b955 vmlinux

$ wc -l Module.symvers
1742 Module.symvers

This time the kernel shrank by 5%.

Finally, let's have a look at a configuration that is potentially more
representative of an embedded target.

Kernel v4.5-rc2 arm realview_defconfig + LTO:

$ size vmlinux
   text    data     bss     dec     hex filename
4422942  209640  126880 4759462  489fa6 vmlinux

$ wc -l Module.symvers
5597 Module.symvers

Kernel v4.5-rc2 arm realview_defconfig + LTO + CONFIG_TRIM_UNUSED_EXPSYMS=y:

$ size vmlinux
   text    data     bss     dec     hex filename
3823485  205416  125800 4154701  3f654d vmlinux

$ wc -l Module.symvers
52 Module.symvers

Here we reduced the kernel text by about 13.6%. Disabling module
support altogether would reduce it by 13.9% i.e. only 0.3% difference.
This means the overhead of using modules on embedded targets is greatly
reduced.

I know that Viro posted a series allowing for assembly code to declare
exported symbols but this doesn't conflict (we don't touch the same
files) and making this work with his series is very trivial.

diffstat:


 Makefile                  | 14 +++++++
 include/linux/export.h    | 22 ++++++++++-
 init/Kconfig              | 16 ++++++++
 scripts/Kbuild.include    |  3 +-
 scripts/Makefile.build    | 16 ++++++--
 scripts/adjust_expsyms.sh | 97 +++++++++++++++++++++++++++++++++++++++++++++++
 scripts/basic/fixdep.c    | 80 +++++++++++++++++++++++++++++---------
 7 files changed, 224 insertions(+), 24 deletions(-)

[PATCH 1/6] kbuild: record needed exported symbols for modules

[Nicolas Pitre]

Kernel modules are partially linked object files with some undefined
symbols that are expected to be matched with EXPORT_SYMBOL() entries
from elsewhere.

Each .tmp_versions/*.mod file currently contains two line of text
separated by a newline character. The first line has the actual module
file name while the second line has a list of object files constituting
that module. Those files are parsed by modpost (scripts/mod/sumversion.c),
scripts/Makefile.modpost, scripts/Makefile.modsign, etc.  Only the
modpost utility cares about the second line while the others retrieve
only the first line.

Therefore we can add a third line to record the list of undefined symbols
aka required EXPORT_SYMBOL() entries for each module into that file
without breaking anything. Like for the second line, symbols are separated
by a blank and the list is terminated with a newline character.

To avoid needless build overhead, the undefined symbols extraction is
performed only when CONFIG_TRIM_UNUSED_EXPSYMS is selected.

Signed-off-by: Nicolas Pitre <nico@linaro.org>
---
 scripts/Makefile.build | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/scripts/Makefile.build b/scripts/Makefile.build
index 2c47f9c305..1d8b07ea3e 100644
--- a/scripts/Makefile.build
+++ b/scripts/Makefile.build
@@ -253,6 +253,13 @@ define rule_cc_o_c
 	mv -f $(dot-target).tmp $(dot-target).cmd
 endef
 
+# List module undefined symbols
+ifdef CONFIG_TRIM_UNUSED_EXPSYMS
+cmd_undef_syms = $(NM) $@ | sed -n 's/^ \+U \(.*\)/\1/p' | xargs echo
+else
+cmd_undef_syms = echo
+endif
+
 # Built-in and composite module parts
 $(obj)/%.o: $(src)/%.c $(recordmcount_source) FORCE
 	$(call cmd,force_checksrc)
@@ -263,7 +270,8 @@ $(obj)/%.o: $(src)/%.c $(recordmcount_source) FORCE
 $(single-used-m): $(obj)/%.o: $(src)/%.c $(recordmcount_source) FORCE
 	$(call cmd,force_checksrc)
 	$(call if_changed_rule,cc_o_c)
-	@{ echo $(@:.o=.ko); echo $@; } > $(MODVERDIR)/$(@F:.o=.mod)
+	@{ echo $(@:.o=.ko); echo $@; \
+	   $(cmd_undef_syms); } > $(MODVERDIR)/$(@F:.o=.mod)
 
 quiet_cmd_cc_lst_c = MKLST   $@
       cmd_cc_lst_c = $(CC) $(c_flags) -g -c -o $*.o $< && \
@@ -393,7 +401,8 @@ $(call multi_depend, $(multi-used-y), .o, -objs -y)
 
 $(multi-used-m): FORCE
 	$(call if_changed,link_multi-m)
-	@{ echo $(@:.o=.ko); echo $(link_multi_deps); } > $(MODVERDIR)/$(@F:.o=.mod)
+	@{ echo $(@:.o=.ko); echo $(link_multi_deps); \
+	   $(cmd_undef_syms); } > $(MODVERDIR)/$(@F:.o=.mod)
 $(call multi_depend, $(multi-used-m), .o, -objs -y -m)
 
 targets += $(multi-used-y) $(multi-used-m)
-- 
2.5.0

Re: [PATCH 1/6] kbuild: record needed exported symbols for modules

[Sam Ravnborg]

>  
> +# List module undefined symbols
> +ifdef CONFIG_TRIM_UNUSED_EXPSYMS
> +cmd_undef_syms = $(NM) $@ | sed -n 's/^ \+U \(.*\)/\1/p' | xargs echo
> +else
> +cmd_undef_syms = echo
> +endif
Use ":" as a "do nothing" cammand. Then you do not emit an empty line.

	Sam

Re: [PATCH 1/6] kbuild: record needed exported symbols for modules

[Nicolas Pitre]

On Mon, 8 Feb 2016, Sam Ravnborg wrote:

> >  
> > +# List module undefined symbols
> > +ifdef CONFIG_TRIM_UNUSED_EXPSYMS
> > +cmd_undef_syms = $(NM) $@ | sed -n 's/^ \+U \(.*\)/\1/p' | xargs echo
> > +else
> > +cmd_undef_syms = echo
> > +endif
> Use ":" as a "do nothing" cammand. Then you do not emit an empty line.

I wanted to emit that empty line though. This way it is less likely that 
it'll be reused for other purposes.


Nicolas

Re: [PATCH 1/6] kbuild: record needed exported symbols for modules

[Al Viro]

On Mon, Feb 08, 2016 at 03:28:30PM -0500, Nicolas Pitre wrote:
> +ifdef CONFIG_TRIM_UNUSED_EXPSYMS
> +cmd_undef_syms = $(NM) $@ | sed -n 's/^ \+U \(.*\)/\1/p' | xargs echo

Umm...  sed -nre 's/^ +U //p', perhaps?

Re: [PATCH 1/6] kbuild: record needed exported symbols for modules

[Nicolas Pitre]

On Tue, 9 Feb 2016, Al Viro wrote:

> On Mon, Feb 08, 2016 at 03:28:30PM -0500, Nicolas Pitre wrote:
> > +ifdef CONFIG_TRIM_UNUSED_EXPSYMS
> > +cmd_undef_syms = $(NM) $@ | sed -n 's/^ \+U \(.*\)/\1/p' | xargs echo
> 
> Umm...  sed -nre 's/^ +U //p', perhaps?

Yep, I like it better.

Although I might be tempted by sed -n 's/^ \+U //p' to avoid the "non 
portable" mention that comes with -r (do we actually care?).


Nicolas

[PATCH 2/6] allow for per-symbol configurable EXPORT_SYMBOL()

[Nicolas Pitre]

Similar to include/generated/autoconf.h, include/generated/expsyms.h will
contain a list of defines for each EXPORT_SYMBOL() that we want active.
The format is:

  #define __EXPSYM_<symbol_name> 1

This list will be auto-generated with another patch.  For now we only
include the preprocessor magic to automatically create or omit the
corresponding struct kernel_symbol declaration.

Given the content of include/generated/expsyms.h may not be known in
advance, an empty file is created early on to let the build proceed.

Signed-off-by: Nicolas Pitre <nico@linaro.org>
---
 Makefile               |  1 +
 include/linux/export.h | 22 ++++++++++++++++++++--
 2 files changed, 21 insertions(+), 2 deletions(-)

diff --git a/Makefile b/Makefile
index 6c1a3c2479..3a264129f8 100644
--- a/Makefile
+++ b/Makefile
@@ -986,6 +986,7 @@ prepare2: prepare3 outputmakefile asm-generic
 prepare1: prepare2 $(version_h) include/generated/utsrelease.h \
                    include/config/auto.conf
 	$(cmd_crmodverdir)
+	$(Q)touch include/generated/expsyms.h
 
 archprepare: archheaders archscripts prepare1 scripts_basic
 
diff --git a/include/linux/export.h b/include/linux/export.h
index 96e45ea463..f8e4d9da35 100644
--- a/include/linux/export.h
+++ b/include/linux/export.h
@@ -38,7 +38,7 @@ extern struct module __this_module;
 
 #ifdef CONFIG_MODULES
 
-#ifndef __GENKSYMS__
+#if defined(__KERNEL__) && !defined(__GENKSYMS__)
 #ifdef CONFIG_MODVERSIONS
 /* Mark the CRC weak since genksyms apparently decides not to
  * generate a checksums for some symbols */
@@ -53,7 +53,7 @@ extern struct module __this_module;
 #endif
 
 /* For every exported symbol, place a struct in the __ksymtab section */
-#define __EXPORT_SYMBOL(sym, sec)				\
+#define ___EXPORT_SYMBOL(sym, sec)				\
 	extern typeof(sym) sym;					\
 	__CRC_SYMBOL(sym, sec)					\
 	static const char __kstrtab_##sym[]			\
@@ -65,6 +65,24 @@ extern struct module __this_module;
 	__attribute__((section("___ksymtab" sec "+" #sym), unused))	\
 	= { (unsigned long)&sym, __kstrtab_##sym }
 
+#ifdef CONFIG_TRIM_UNUSED_EXPSYMS
+
+#include <linux/kconfig.h>
+#include <generated/expsyms.h>
+
+#define __EXPORT_SYMBOL(sym, sec)				\
+	__cond_export_sym(sym, sec, config_enabled(__EXPSYM_##sym))
+#define __cond_export_sym(sym, sec, conf)			\
+	___cond_export_sym(sym, sec, conf)
+#define ___cond_export_sym(sym, sec, enabled)			\
+	__cond_export_sym_##enabled(sym, sec)
+#define __cond_export_sym_1(sym, sec) ___EXPORT_SYMBOL(sym, sec)
+#define __cond_export_sym_0(sym, sec) /* nothing */
+
+#else
+#define __EXPORT_SYMBOL ___EXPORT_SYMBOL
+#endif
+
 #define EXPORT_SYMBOL(sym)					\
 	__EXPORT_SYMBOL(sym, "")
 
-- 
2.5.0

Re: [PATCH 2/6] allow for per-symbol configurable EXPORT_SYMBOL()

[Sam Ravnborg]

On Mon, Feb 08, 2016 at 03:28:31PM -0500, Nicolas Pitre wrote:
> Similar to include/generated/autoconf.h, include/generated/expsyms.h will
> contain a list of defines for each EXPORT_SYMBOL() that we want active.
> The format is:
> 
>   #define __EXPSYM_<symbol_name> 1

Please spell it out.
EXP could be EXPECT...


> +	$(Q)touch include/generated/expsyms.h

Same goes here. Make it explicit what this is about.
It is obvious today that "exp" is synonym for export_symbol.
But in one year less so.

Same goes for CONFIG_ symbol etc.

	Sam
> 

[PATCH 3/6] fixdep: minor cleanup

[Nicolas Pitre]

Simple code reorganization to make coming changes more obvious.

Signed-off-by: Nicolas Pitre <nico@linaro.org>
---
 scripts/basic/fixdep.c | 17 ++++++++---------
 1 file changed, 8 insertions(+), 9 deletions(-)

diff --git a/scripts/basic/fixdep.c b/scripts/basic/fixdep.c
index 5b327c67a8..63f129021d 100644
--- a/scripts/basic/fixdep.c
+++ b/scripts/basic/fixdep.c
@@ -236,21 +236,20 @@ static void parse_config_file(const char *map, size_t len)
 			continue;
 		p += 7;
 		for (q = p; q < map + len; q++) {
-			if (!(isalnum(*q) || *q == '_'))
-				goto found;
+			if (!(isalnum(*q) || *q == '_')) {
+				if (!memcmp(q - 7, "_MODULE", 7))
+					q -= 7;
+				if (q - p > 0)
+					use_config(p, q - p);
+				break;
+			}
 		}
 		continue;
 
-	found:
-		if (!memcmp(q - 7, "_MODULE", 7))
-			q -= 7;
-		if (q - p < 0)
-			continue;
-		use_config(p, q - p);
 	}
 }
 
-/* test is s ends in sub */
+/* test if s ends in sub */
 static int strrcmp(const char *s, const char *sub)
 {
 	int slen = strlen(s);
-- 
2.5.0

[PATCH 4/6] fixdep: add fine grained build dependencies for exported symbols

[Nicolas Pitre]

Like for kconfig options, we now have the ability to compile in and out
individual EXPORT_SYMBOL() declarations based on the content of
include/generated/expsyms.h.  However we don't want the entire world
to be rebuilt whenever that file is touched.

Let's apply the same build dependency trick used with config symbols to
those EXPORT_SYMBOL() instances. The key in this case is the actual symbol
name being exported. All symbol names are collapsed to lowercase path
names whose time stamp provide fine grained dependencies.

Because of the lowercasing, there might be name collisions triggering
spurious rebuilds for similar symbols. But this shouldn't be a big issue
in practice. (This is the case for CONFIG symbols and I didn't want to
be different here, whatever the original reason for doing so.)

To avoid needless build overhead, the exported symbol name parsing is
performed only when CONFIG_TRIM_UNUSED_EXPSYMS is selected. A new cmdline
argument is added to that effect.

Signed-off-by: Nicolas Pitre <nico@linaro.org>
---
 scripts/Kbuild.include |  3 ++-
 scripts/Makefile.build |  3 ++-
 scripts/basic/fixdep.c | 67 +++++++++++++++++++++++++++++++++++++++++---------
 3 files changed, 60 insertions(+), 13 deletions(-)

diff --git a/scripts/Kbuild.include b/scripts/Kbuild.include
index 1db6d73c8d..13b4032d42 100644
--- a/scripts/Kbuild.include
+++ b/scripts/Kbuild.include
@@ -257,7 +257,8 @@ if_changed = $(if $(strip $(any-prereq) $(arg-check)),                       \
 if_changed_dep = $(if $(strip $(any-prereq) $(arg-check) ),                  \
 	@set -e;                                                             \
 	$(echo-cmd) $(cmd_$(1));                                             \
-	scripts/basic/fixdep $(depfile) $@ '$(make-cmd)' > $(dot-target).tmp;\
+	scripts/basic/fixdep $(if $(CONFIG_TRIM_UNUSED_EXPSYMS),-e)	     \
+		 $(depfile) $@ '$(make-cmd)' > $(dot-target).tmp;	     \
 	rm -f $(depfile);                                                    \
 	mv -f $(dot-target).tmp $(dot-target).cmd)
 
diff --git a/scripts/Makefile.build b/scripts/Makefile.build
index 1d8b07ea3e..f2c4d93726 100644
--- a/scripts/Makefile.build
+++ b/scripts/Makefile.build
@@ -247,7 +247,8 @@ define rule_cc_o_c
 	$(cmd_modversions)						  \
 	$(call echo-cmd,record_mcount)					  \
 	$(cmd_record_mcount)						  \
-	scripts/basic/fixdep $(depfile) $@ '$(call make-cmd,cc_o_c)' >    \
+	scripts/basic/fixdep $(if $(CONFIG_TRIM_UNUSED_EXPSYMS),-e)	  \
+	                     $(depfile) $@ '$(call make-cmd,cc_o_c)' >    \
 	                                              $(dot-target).tmp;  \
 	rm -f $(depfile);						  \
 	mv -f $(dot-target).tmp $(dot-target).cmd
diff --git a/scripts/basic/fixdep.c b/scripts/basic/fixdep.c
index 63f129021d..71099ffe67 100644
--- a/scripts/basic/fixdep.c
+++ b/scripts/basic/fixdep.c
@@ -120,13 +120,20 @@
 #define INT_NFIG ntohl(0x4e464947)
 #define INT_FIG_ ntohl(0x4649475f)
 
+#define INT_EXPO ntohl(0x4558504f)
+#define INT_XPOR ntohl(0x58504f52)
+#define INT_PORT ntohl(0x504f5254)
+#define INT_ORT_ ntohl(0x4f52545f)
+
+int parse_export_symbol;
 char *target;
 char *depfile;
 char *cmdline;
 
 static void usage(void)
 {
-	fprintf(stderr, "Usage: fixdep <depfile> <target> <cmdline>\n");
+	fprintf(stderr, "Usage: fixdep [-e] <depfile> <target> <cmdline>\n");
+	fprintf(stderr, " -e  parse EXPORT_SYMBOL_* too.\n");
 	exit(1);
 }
 
@@ -140,6 +147,7 @@ static void print_cmdline(void)
 
 struct item {
 	struct item	*next;
+	const char	*type;
 	unsigned int	len;
 	unsigned int	hash;
 	char		name[0];
@@ -161,12 +169,12 @@ static unsigned int strhash(const char *str, unsigned int sz)
 /*
  * Lookup a value in the configuration string.
  */
-static int is_defined_config(const char *name, int len, unsigned int hash)
+static int is_defined_string(const char *type, const char *name, int len, unsigned int hash)
 {
 	struct item *aux;
 
 	for (aux = hashtab[hash % HASHSZ]; aux; aux = aux->next) {
-		if (aux->hash == hash && aux->len == len &&
+		if (aux->hash == hash && aux->len == len && aux->type == type &&
 		    memcmp(aux->name, name, len) == 0)
 			return 1;
 	}
@@ -176,7 +184,7 @@ static int is_defined_config(const char *name, int len, unsigned int hash)
 /*
  * Add a new value to the configuration string.
  */
-static void define_config(const char *name, int len, unsigned int hash)
+static void define_string(const char *type, const char *name, int len, unsigned int hash)
 {
 	struct item *aux = malloc(sizeof(*aux) + len);
 
@@ -186,25 +194,26 @@ static void define_config(const char *name, int len, unsigned int hash)
 	}
 	memcpy(aux->name, name, len);
 	aux->len = len;
+	aux->type = type;
 	aux->hash = hash;
 	aux->next = hashtab[hash % HASHSZ];
 	hashtab[hash % HASHSZ] = aux;
 }
 
 /*
- * Record the use of a CONFIG_* word.
+ * Record the use of a CONFIG_* word, or the argument to EXPORT_MODULE*()
  */
-static void use_config(const char *m, int slen)
+static void use_dep(const char *type, const char *m, int slen)
 {
 	unsigned int hash = strhash(m, slen);
 	int c, i;
 
-	if (is_defined_config(m, slen, hash))
+	if (is_defined_string(type, m, slen, hash))
 	    return;
 
-	define_config(m, slen, hash);
+	define_string(type, m, slen, hash);
 
-	printf("    $(wildcard include/config/");
+	printf("    $(wildcard include/%s/", type);
 	for (i = 0; i < slen; i++) {
 		c = m[i];
 		if (c == '_')
@@ -228,7 +237,17 @@ static void parse_config_file(const char *map, size_t len)
 		if (*m == INT_ONFI) { p = (char *) m-1; goto conf; }
 		if (*m == INT_NFIG) { p = (char *) m-2; goto conf; }
 		if (*m == INT_FIG_) { p = (char *) m-3; goto conf; }
+
+		if (!parse_export_symbol)
+			continue;
+
+		if (*m == INT_EXPO) { p = (char *) m  ; goto export; }
+		if (*m == INT_XPOR) { p = (char *) m-1; goto export; }
+		if (*m == INT_PORT) { p = (char *) m-2; goto export; }
+		if (*m == INT_ORT_) { p = (char *) m-3; goto export; }
+
 		continue;
+
 	conf:
 		if (p > map + len - 7)
 			continue;
@@ -240,12 +259,34 @@ static void parse_config_file(const char *map, size_t len)
 				if (!memcmp(q - 7, "_MODULE", 7))
 					q -= 7;
 				if (q - p > 0)
-					use_config(p, q - p);
+					use_dep("config", p, q - p);
 				break;
 			}
 		}
 		continue;
 
+	export:
+		if (p[-1] == '_')
+			continue;
+		if (p > map + len - 7 || memcmp(p, "EXPORT_", 7) != 0)
+			continue;
+		p += 7;
+		if (p <= map + len - 8 && memcmp(p, "PER_CPU_", 8) == 0)
+			p += 8;
+		if (p > map + len - 6 || memcmp(p, "SYMBOL", 6) != 0)
+			continue;
+		p += 6;
+		while (p < map + len && *p != '(' && *p != '\n')
+			p++;
+		q = p + 1;
+		while (q < map + len && *q != ')' && *q != '\n' && *q != '#')
+			q++;
+		if (q >= map + len || *p != '(' || *q != ')')
+			continue;
+		while (isblank(*++p));
+		while (isblank(q[-1])) q--;
+		if (q - p > 0)
+			use_dep("config/expsym", p, q - p);
 	}
 }
 
@@ -330,6 +371,7 @@ static void parse_dep_file(void *map, size_t len)
 			if (strrcmp(s, "include/generated/autoconf.h") &&
 			    strrcmp(s, "arch/um/include/uml-config.h") &&
 			    strrcmp(s, "include/linux/kconfig.h") &&
+			    strrcmp(s, "include/generated/expsyms.h") &&
 			    strrcmp(s, ".ver")) {
 				/*
 				 * Do not list the source file as dependency,
@@ -429,7 +471,10 @@ int main(int argc, char *argv[])
 {
 	traps();
 
-	if (argc != 4)
+	if (argc == 5 && !strcmp(argv[1], "-e")) {
+		parse_export_symbol = 1;
+		argv++;
+	} else if (argc != 4)
 		usage();
 
 	depfile = argv[1];
-- 
2.5.0

[PATCH 5/6] create/adjust generated/expsyms.h

[Nicolas Pitre]

Given the list of exported symbols needed by all modules, we can create
a header file containing preprocessor defines for each of those symbols.
Also, when some symbols are added and/or removed from the list, we can
update the time on the corresponding files used as build dependencies for
those symbols. And finally, if any symbol did change state, the
corresponding source files must be rebuilt.

The insertion or removal of an EXPORT_SYMBOL() entry within a module may
create or remove the need for another exported symbol.  This is why this
operation has to be repeated until the list of needed exported symbols
becomes stable. Only then the final kernel and modules link take place.

Signed-off-by: Nicolas Pitre <nico@linaro.org>
---
 Makefile                  | 13 +++++++
 scripts/adjust_expsyms.sh | 97 +++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 110 insertions(+)
 create mode 100755 scripts/adjust_expsyms.sh

diff --git a/Makefile b/Makefile
index 3a264129f8..f904c1d3e8 100644
--- a/Makefile
+++ b/Makefile
@@ -921,6 +921,10 @@ quiet_cmd_link-vmlinux = LINK    $@
 # Include targets which we want to
 # execute if the rest of the kernel build went well.
 vmlinux: scripts/link-vmlinux.sh $(vmlinux-deps) FORCE
+ifdef CONFIG_TRIM_UNUSED_EXPSYMS
+	$(Q)$(CONFIG_SHELL) scripts/adjust_expsyms.sh \
+	    "$(MAKE) KBUILD_MODULES=1 -f $(srctree)/Makefile vmlinux_recursive"
+endif
 ifdef CONFIG_HEADERS_CHECK
 	$(Q)$(MAKE) -f $(srctree)/Makefile headers_check
 endif
@@ -935,6 +939,15 @@ ifdef CONFIG_GDB_SCRIPTS
 endif
 	+$(call if_changed,link-vmlinux)
 
+vmlinux_recursive: $(vmlinux-deps)
+	$(Q)$(CONFIG_SHELL) scripts/adjust_expsyms.sh \
+	    "$(MAKE) KBUILD_MODULES=1 -f $(srctree)/Makefile vmlinux_recursive"
+PHONY += vmlinux_recursive
+
+# standalone target for easier testing
+include/generated/expsyms.h: FORCE
+	$(Q)$(CONFIG_SHELL) scripts/adjust_expsyms.sh true
+
 # The actual objects are generated when descending,
 # make sure no implicit rule kicks in
 $(sort $(vmlinux-deps)): $(vmlinux-dirs) ;
diff --git a/scripts/adjust_expsyms.sh b/scripts/adjust_expsyms.sh
new file mode 100755
index 0000000000..f363f4b8ea
--- /dev/null
+++ b/scripts/adjust_expsyms.sh
@@ -0,0 +1,97 @@
+#!/bin/sh
+
+# Script to create/update include/generated/expsyms.h and dependency files
+#
+# Copyright:	(C) 2016  Linaro Limited
+# Created by:	Nicolas Pitre, January 2016
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License version 2 as
+# published by the Free Software Foundation.
+
+# Create/update the include/generated/expsyms.h file from the list
+# of all module's needed symbols as recorded on the third line of
+# .tmp_versions/*.mod files.
+#
+# For each symbol being added or removed, the corresponding dependency
+# file's timestamp is updated to force a rebuild of the affected source
+# file. All arguments passed to this script are assumed to be a command
+# to be exec'd that will trigger a rebuild of those files.
+
+set -e
+
+cur_expsym_file="include/generated/expsyms.h"
+new_expsym_file="include/generated/expsyms.h.tmpnew"
+
+info() { [ "$quiet" != "silent_" ] && printf "  %-7s %s\n" "$1" "$2"; }
+
+info "CHK" "$cur_expsym_file"
+
+# Use "make V=1" to debug this script.
+case "$KBUILD_VERBOSE" in
+*1*)
+	set -x
+	;;
+esac
+
+# We need access to CONFIG_ symbols
+case "${KCONFIG_CONFIG}" in
+*/*)
+	. "${KCONFIG_CONFIG}"
+	;;
+*)
+	# Force using a file from the current directory
+	. "./${KCONFIG_CONFIG}"
+esac
+
+# In case it doesn't exist yet...
+[ -e "$cur_expsym_file" ] || touch "$cur_expsym_file"
+
+# Generate a new expsym list file with symbols needed by the current
+# set of modules.
+cat > "$new_expsym_file" << EOT
+/*
+ * Automatically generated file; DO NOT EDIT.
+ */
+
+EOT
+sed -ns -e '3s/ /\n/gp' "$MODVERDIR"/*.mod | sort -u |
+while read sym; do
+	if [ -n "$CONFIG_HAVE_UNDERSCORE_SYMBOL_PREFIX" ]; then
+		sym=$(sed 's/_\(.*\)$/\1/' <<< "$sym")
+	fi
+	echo "#define __EXPSYM_${sym} 1"
+done >> "$new_expsym_file"
+
+# Special case for modversions (see modpost.c)
+if [ -n "$CONFIG_MODVERSIONS" ]; then
+	echo "#define __EXPSYM_module_layout 1" >> "$new_expsym_file"
+fi
+
+# Extract changes between old and new list and touch corresponding
+# dependency files.
+# Note: sort -m doesn't work well with underscore prefixed symbols so we
+# use 'cat ... | sort' instead.
+changed=0
+while read sympath; do
+	[ -z "$sympath" ] && continue
+	depfile="include/config/expsym/${sympath}.h"
+	mkdir -p "$(dirname "$depfile")"
+	touch "$depfile"
+	changed=$((changed + 1))
+done <<< "$(
+	cat "$cur_expsym_file" "$new_expsym_file" | sort | uniq -u |
+	sed -n 's/^#define __EXPSYM_\(.*\) 1/\1/p' | tr "A-Z_" "a-z/"  )"
+
+if [ $changed -gt 0 ]; then
+	# Replace the old list with tne new one
+	old=$(grep -c "^#define __EXPSYM_" "$cur_expsym_file" || true)
+	new=$(grep -c "^#define __EXPSYM_" "$new_expsym_file" || true)
+	info "EXPSYM" "symbols: $old old, $new new, $changed changed"
+	info "UPD" "$cur_expsym_file"
+	mv -f "$new_expsym_file" "$cur_expsym_file"
+	# Then trigger a rebuild of affected files
+	exec $@
+else
+	rm -f "$new_expsym_file"
+fi
-- 
2.5.0

Re: [PATCH 5/6] create/adjust generated/expsyms.h

[Sam Ravnborg]

On Mon, Feb 08, 2016 at 03:28:34PM -0500, Nicolas Pitre wrote:
> Given the list of exported symbols needed by all modules, we can create
> a header file containing preprocessor defines for each of those symbols.
> Also, when some symbols are added and/or removed from the list, we can
> update the time on the corresponding files used as build dependencies for
> those symbols. And finally, if any symbol did change state, the
> corresponding source files must be rebuilt.
> 
> The insertion or removal of an EXPORT_SYMBOL() entry within a module may
> create or remove the need for another exported symbol.  This is why this
> operation has to be repeated until the list of needed exported symbols
> becomes stable. Only then the final kernel and modules link take place.

Could this magic with vmlinux_recursive have been implemented in a more
obvious way in link-vmlinux.sh?
One of the purposes with link-vmlinux.sh was to make the final link
stage more readable and this patch goes in the other direction.

	Sam

Re: [PATCH 5/6] create/adjust generated/expsyms.h

[Nicolas Pitre]

On Mon, 8 Feb 2016, Sam Ravnborg wrote:

> On Mon, Feb 08, 2016 at 03:28:34PM -0500, Nicolas Pitre wrote:
> > Given the list of exported symbols needed by all modules, we can create
> > a header file containing preprocessor defines for each of those symbols.
> > Also, when some symbols are added and/or removed from the list, we can
> > update the time on the corresponding files used as build dependencies for
> > those symbols. And finally, if any symbol did change state, the
> > corresponding source files must be rebuilt.
> > 
> > The insertion or removal of an EXPORT_SYMBOL() entry within a module may
> > create or remove the need for another exported symbol.  This is why this
> > operation has to be repeated until the list of needed exported symbols
> > becomes stable. Only then the final kernel and modules link take place.
> 
> Could this magic with vmlinux_recursive have been implemented in a more
> obvious way in link-vmlinux.sh?
> One of the purposes with link-vmlinux.sh was to make the final link
> stage more readable and this patch goes in the other direction.

I played with different alternatives and this is really the best I came 
up with.  Those alternatives were much uglier.

The adjust_expsyms.sh script must be run only when all vmlinux 
prerequisites have been built i.e descending in all subdirs is done, but 
before actually linking it. And then if some changes in the list of 
exported symbols was detected then all those prerequisites have to be 
re-evaluated again.  The adjust_expsyms.sh invocation could be done from 
link-vmlinux.sh but there would still be a need for a special Make 
target to revisit all dependencies before resuming with the link.

If you have any suggestions for improving this though please feel free 
to share them.


Nicolas

[PATCH 6/6] kconfig option for TRIM_UNUSED_EXPSYMS

[Nicolas Pitre]

The config option to enable it all.

Signed-off-by: Nicolas Pitre <nico@linaro.org>
---
 init/Kconfig | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/init/Kconfig b/init/Kconfig
index 22320804fb..893c4979a9 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -1990,6 +1990,22 @@ config MODULE_COMPRESS_XZ
 
 endchoice
 
+config TRIM_UNUSED_EXPSYMS
+	bool "Trim unused exported symbols"
+	depends on MODULES && !UNUSED_SYMBOLS
+	help
+	  The kernel and some modules make many symbols available for
+	  other modules to use via EXPORT_SYMBOL() and variants. Depending
+	  on the set of modules being selected in your kernel configuration,
+	  many of those exported symbols might never be used.
+
+	  This option allows for unused exported symbols to be dropped from
+	  the build. In turn, this provides the compiler more opportunities
+	  (especially when using LTO) for optimizing the code and reducing
+	  binary size.  This might have some security advantages as well.
+
+	  If unsure say N.
+
 endif # MODULES
 
 config MODULES_TREE_LOOKUP
-- 
2.5.0

Re: [PATCH 6/6] kconfig option for TRIM_UNUSED_EXPSYMS

[Rusty Russell]

Nicolas Pitre <nicolas.pitre@linaro.org> writes:
> The config option to enable it all.
>
> Signed-off-by: Nicolas Pitre <nico@linaro.org>

Nice series!  In case you need it, here's my Ack.  Despite the overlap
with modules, it's all build trickery, so best via Sam's tree.

Acked-by: Rusty Russell <rusty@rustcorp.com.au>

Thanks,
Rusty.

Re: [PATCH 6/6] kconfig option for TRIM_UNUSED_EXPSYMS

[Nicolas Pitre]

On Tue, 9 Feb 2016, Rusty Russell wrote:

> Nicolas Pitre <nicolas.pitre@linaro.org> writes:
> > The config option to enable it all.
> >
> > Signed-off-by: Nicolas Pitre <nico@linaro.org>
> 
> Nice series!  In case you need it, here's my Ack.  Despite the overlap
> with modules, it's all build trickery, so best via Sam's tree.
> 
> Acked-by: Rusty Russell <rusty@rustcorp.com.au>

Thanks


Nicolas

Re: [PATCH 6/6] kconfig option for TRIM_UNUSED_EXPSYMS

[Sam Ravnborg]

> 
> so best via Sam's tree.
That was some years ago ;-)

	Sam

Re: [PATCH 6/6] kconfig option for TRIM_UNUSED_EXPSYMS

[Christoph Hellwig]

On Mon, Feb 08, 2016 at 03:28:35PM -0500, Nicolas Pitre wrote:
> The config option to enable it all.

Just enable it by default..

Re: [PATCH 6/6] kconfig option for TRIM_UNUSED_EXPSYMS

[Austin S. Hemmelgarn]

On 2016-02-09 08:28, Christoph Hellwig wrote:
> On Mon, Feb 08, 2016 at 03:28:35PM -0500, Nicolas Pitre wrote:
>> The config option to enable it all.
>
> Just enable it by default..
>
I doubt we care too much about third party modules, but enabling this by 
default will likely cause at least some of them to stop working.  At the 
very least, we need the config option, even if it's enabled by default.

Re: [PATCH 6/6] kconfig option for TRIM_UNUSED_EXPSYMS

[Nicolas Pitre]

On Tue, 9 Feb 2016, Christoph Hellwig wrote:

> On Mon, Feb 08, 2016 at 03:28:35PM -0500, Nicolas Pitre wrote:
> > The config option to enable it all.
> 
> Just enable it by default..

:-)

Viro took the opposite view with this patch:

http://lkml.org/lkml/2016/2/3/1068

where he's ensuring that no exports are dropped, even if obviously 
they're not used (otherwise some complaints would have come about).  So 
I'm not sure if having this turned on by default is a good thing.

But I don't mind being convinced otherwise.


Nicolas

Re: [PATCH 6/6] kconfig option for TRIM_UNUSED_EXPSYMS

[Rusty Russell]

Nicolas Pitre <nicolas.pitre@linaro.org> writes:
> On Tue, 9 Feb 2016, Christoph Hellwig wrote:
>
>> On Mon, Feb 08, 2016 at 03:28:35PM -0500, Nicolas Pitre wrote:
>> > The config option to enable it all.
>> 
>> Just enable it by default..
>
> :-)
>
> Viro took the opposite view with this patch:
>
> http://lkml.org/lkml/2016/2/3/1068
>
> where he's ensuring that no exports are dropped, even if obviously 
> they're not used (otherwise some complaints would have come about).  So 
> I'm not sure if having this turned on by default is a good thing.
>
> But I don't mind being convinced otherwise.

People do build out-of-tree modules.  Dropping support for that would
simplify our lives, but seems fairly controversial :)

Cheers,
Rusty.