kunit: add macro to allow conditionally exposing static symbols to tests
Subsystems (4)
cover kunit: add macro to allow conditionally exposing static symbols to tests
Currently in order to test a static function, tests must be included in the same translation unit as the function. However, this can cause issues with including implementation and test code in the same file. As an alternative, the first patch in this series creates a macro that will set a function to be static or not depending on whether CONFIG_KUNIT is enabled. This allows the function to be visible during testing and static otherwise.
As an example, the current status quo to test static functions is:
=== test.c ===
static void test_case(struct kunit *test)
{
KUNIT_EXPECT_EQ(test, my_func_to_test(), 2);
}
Then the tests are included in the implementation file as a workaround to the issue of testing static functions:
=== implementation.c ===
static int my_func_to_test() {...}
...
#include "test.c"
Instead, the function could be defined with this new macro:
=== implementation.c ===
VISIBLE_IF_KUNIT int my_func_to_test() {...}
The first patch also creates a macro that will export a symbol into a kunit testing namespace only if CONFIG_KUNIT is enabled. This follows the logic above and allows symbols to be conditionally exported based on the testing status.
The second patch in the series updates the policy_unpack test in AppArmor to show an example of how to use both of these macros in order to address the issue of testing static functions.
Rae Moar (2):
kunit: add macro to allow conditionally exposing static symbols to
tests
apparmor: test: make static symbols visible during kunit testing
include/kunit/visibility.h | 32 ++++++++++ security/apparmor/Kconfig | 4 +- security/apparmor/Makefile | 2 + security/apparmor/include/policy_unpack.h | 50 ++++++++++++++++ security/apparmor/policy_unpack.c | 72 +++++++---------------- security/apparmor/policy_unpack_test.c | 5 ++ 6 files changed, 112 insertions(+), 53 deletions(-) create mode 100644 include/kunit/visibility.h
base-commit: 11e76194937b506caf1b49512c42d5c2588681d7
Signature · 1 line
2.38.1.273.g43a17bfeac-goog
PATCH 1/2 kunit: add macro to allow conditionally exposing static symbols to tests
Create two macros:
VISIBLE_IF_KUNIT - A macro that sets symbols to be static if CONFIG_KUNIT is not enabled. Otherwise if CONFIG_KUNIT is enabled there is no change to the symbol definition.
EXPORT_SYMBOL_IF_KUNIT(symbol) - Exports symbol into EXPORTED_FOR_KUNIT_TESTING namespace only if CONFIG_KUNIT is enabled. Must use MODULE_IMPORT_NS(EXPORTED_FOR_KUNIT_TESTING) in test file in order to use symbols.
--- include/kunit/visibility.h | 32 ++++++++++++++++++++++++++++++++ 1 file changed, 32 insertions(+) create mode 100644 include/kunit/visibility.h
include/kunit/visibility.h new file +32 -0
@@ -0,0 +1,32 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* +* KUnit API to allow symbols to be conditionally visible during KUnit +* testing +* +* Copyright (C) 2019, Google LLC. +* Author: Brendan Higgins <[email protected]> +*/ + +#ifndef _KUNIT_VISIBILITY_H +#define _KUNIT_VISIBILITY_H + +/** +* VISIBLE_IF_KUNIT - A macro that sets symbols to be static if CONFIG_KUNIT +* is not enabled. Otherwise if CONFIG_KUNIT is enabled there is no change +* to the symbol definition. +* +* EXPORT_SYMBOL_IF_KUNIT(symbol) - Exports symbol into +* EXPORTED_FOR_KUNIT_TESTING namespace only if CONFIG_KUNIT is +* enabled. Must use MODULE_IMPORT_NS(EXPORTED_FOR_KUNIT_TESTING) +* in test file in order to use symbols. +*/ +#if IS_ENABLED (CONFIG_KUNIT ) +#define VISIBLE_IF_KUNIT +#define EXPORT_SYMBOL_IF_KUNIT (symbol) EXPORT_SYMBOL_NS(symbol, \ + EXPORTED_FOR_KUNIT_TESTING) +#else +#define VISIBLE_IF_KUNIT static +#define EXPORT_SYMBOL_IF_KUNIT (symbol) +#endif + +#endif /* _KUNIT_VISIBILITY_H */
Signature · 1 line
2.38.1.273.g43a17bfeac-goog
Quoted 6 lines · Rae Moar
Create two macros:
VISIBLE_IF_KUNIT- A macro that sets symbols to be static ifCONFIG_KUNITis not enabled. Otherwise ifCONFIG_KUNITis enabled there is no change to the symbol definition.
Quoted 6 lines
EXPORT_SYMBOL_IF_KUNIT(symbol)- Exports symbol intoEXPORTED_FOR_KUNIT_TESTINGnamespace only ifCONFIG_KUNITis enabled. Must useMODULE_IMPORT_NS(EXPORTED_FOR_KUNIT_TESTING)in test file in order to use symbols.Signed-off-by Rae Moar <[email protected]>
This looks fine
Quoted 43 lines
--- include/kunit/visibility.h | 32 ++++++++++++++++++++++++++++++++ 1 file changed, 32 insertions(+)create mode 100644
include/kunit/visibility.hinclude/kunit/visibility.h new file +32
-0@@ -0,0 +1,32 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* +* KUnit API to allow symbols to be conditionally visible during KUnit +* testing +* +* Copyright (C) 2019, Google LLC. +* Author: Brendan Higgins <[email protected]> +*/ + +#ifndef _KUNIT_VISIBILITY_H +#define _KUNIT_VISIBILITY_H + +/** +* VISIBLE_IF_KUNIT - A macro that sets symbols to be static if CONFIG_KUNIT +* is not enabled. Otherwise if CONFIG_KUNIT is enabled there is no change +* to the symbol definition. +* +* EXPORT_SYMBOL_IF_KUNIT(symbol) - Exports symbol into +* EXPORTED_FOR_KUNIT_TESTING namespace only if CONFIG_KUNIT is +* enabled. Must use MODULE_IMPORT_NS(EXPORTED_FOR_KUNIT_TESTING) +* in test file in order to use symbols. +*/ +#if IS_ENABLED (CONFIG_KUNIT ) +#define VISIBLE_IF_KUNIT +#define EXPORT_SYMBOL_IF_KUNIT (symbol) EXPORT_SYMBOL_NS(symbol, \ + EXPORTED_FOR_KUNIT_TESTING) +#else +#define VISIBLE_IF_KUNIT static +#define EXPORT_SYMBOL_IF_KUNIT (symbol) +#endif + +#endif /* _KUNIT_VISIBILITY_H */
Quoted 7 lines · Rae Moar
Create two macros:
VISIBLE_IF_KUNIT- A macro that sets symbols to be static ifCONFIG_KUNITis not enabled. Otherwise ifCONFIG_KUNITis enabled there is no change to the symbol definition.
Quoted 7 lines
EXPORT_SYMBOL_IF_KUNIT(symbol)- Exports symbol intoEXPORTED_FOR_KUNIT_TESTINGnamespace only ifCONFIG_KUNITis enabled. Must useMODULE_IMPORT_NS(EXPORTED_FOR_KUNIT_TESTING)in test file in order to use symbols.Signed-off-by Rae Moar <[email protected]>---
This looks good to me overall.
Two thoughts (though they're not _problems_ with this patch, so shouldn't necessarily hold it up).
How many KUnit headers do we want for these sorts of things? We've currently got 'test.h' (for actual tests) and 'test-bug.h' for the kunit_fail_current_test() style hooks. And there'll be 'static-stub.h' coming.
Maybe it'd make sense to merge some of these, if it turns out the same code always needs to access them? On the other hand, I actually quite like having them separate like this. (Though the split of assertions into assert.h and perhaps renaming test-bug.h might be worth doing down the line.) No immediate action item, but something perhaps worth thinking about.
Secondly, do we want to support individual subsystems and/or tests to have their own symbol namespaces. Again, not a problem for the moment, but we may need to deal with this if we need to export multiple copies of very-generically-named functions. Any thoughts (e.g., an EXPORT_SYMBOL_IF_KUNIT_NS macro which allows the user to specify the namespace?)
Regardless, this is
Cheers,
-- David
Quoted 10 lines
include/kunit/visibility.h | 32 ++++++++++++++++++++++++++++++++ 1 file changed, 32 insertions(+) create mode 100644 include/kunit/visibility.hinclude/kunit/visibility.h new file +0
-0
include/kunit/visibility.h line 7
Quoted 7 lines
include/kunit/visibility.h new file +7
-0@@ -0,0 +1,32 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* +* KUnit API to allow symbols to be conditionally visible during KUnit +* testing +* +* Copyright (C) 2019, Google LLC. +* Author: Brendan Higgins <[email protected]>
Please update the date here, and use your name/email for the Author section.
Quoted 28 lines
include/kunit/visibility.h new file +25
-1+*/ + +#ifndef _KUNIT_VISIBILITY_H +#define _KUNIT_VISIBILITY_H + +/** +* VISIBLE_IF_KUNIT - A macro that sets symbols to be static if CONFIG_KUNIT +* is not enabled. Otherwise if CONFIG_KUNIT is enabled there is no change +* to the symbol definition. +* +* EXPORT_SYMBOL_IF_KUNIT(symbol) - Exports symbol into +* EXPORTED_FOR_KUNIT_TESTING namespace only if CONFIG_KUNIT is +* enabled. Must use MODULE_IMPORT_NS(EXPORTED_FOR_KUNIT_TESTING) +* in test file in order to use symbols. +*/ +#if IS_ENABLED (CONFIG_KUNIT ) +#define VISIBLE_IF_KUNIT +#define EXPORT_SYMBOL_IF_KUNIT (symbol) EXPORT_SYMBOL_NS(symbol, \ + EXPORTED_FOR_KUNIT_TESTING) +#else +#define VISIBLE_IF_KUNIT static +#define EXPORT_SYMBOL_IF_KUNIT (symbol) +#endif + +#endif /* _KUNIT_VISIBILITY_H */ --2.38.1.273.g43a17bfeac-goog
PATCH 2/2 apparmor: test: make static symbols visible during kunit testing
In reply to Rae MoarUse macros, VISIBLE_IF_KUNIT and EXPORT_SYMBOL_IF_KUNIT, to allow static symbols to be conditionally set to be visible during KUnit testing. Remove the need to include testing file in the implementation file. Provide example of how static symbols can be dealt with in testing.
--- security/apparmor/Kconfig | 4 +- security/apparmor/Makefile | 2 + security/apparmor/include/policy_unpack.h | 50 ++++++++++++++++ security/apparmor/policy_unpack.c | 72 +++++++---------------- security/apparmor/policy_unpack_test.c | 5 ++ 5 files changed, 80 insertions(+), 53 deletions(-)
security/apparmor/Kconfig +2 -2
@@ -106,8 +106,8 @@ config SECURITY_APPARMOR_PARANOID_LOAD Disabling the check will speed up policy loads. config SECURITY_APPARMOR_KUNIT_TEST - bool "Build KUnit tests for policy_unpack.c" if !KUNIT_ALL_TESTS - depends on KUNIT=y && SECURITY_APPARMOR + tristate "Build KUnit tests for policy_unpack.c" if !KUNIT_ALL_TESTS + depends on KUNIT && SECURITY_APPARMOR default KUNIT_ALL_TESTS help This builds the AppArmor KUnit tests.
security/apparmor/Makefile +2 -0
@@ -8,6 +8,8 @@ apparmor-y := apparmorfs.o audit.o capability.o task.o ipc.o lib.o match.o \ resource.o secid.o file.o policy_ns.o label.o mount.o net.o apparmor-$(CONFIG_SECURITY_APPARMOR_HASH) += crypto.o +obj-$(CONFIG_SECURITY_APPARMOR_KUNIT_TEST) += policy_unpack_test.o + clean-files := capability_names.h rlim_names.h net_names.h # Build a lower case string table of address family names
security/apparmor/include/policy_unpack.h +50 -0
60 lines not on this page · raw message
security/apparmor/policy_unpack.c +21 -51
175 lines not on this page · raw message
security/apparmor/policy_unpack_test.c +5 -0
@@ -4,6 +4,7 @@*/ #include <kunit/test.h> +#include <kunit/visibility.h> #include "include/policy.h" #include "include/policy_unpack.h"
@@ -43,6 +44,8 @@#define TEST_ARRAY_BUF_OFFSET \ (TEST_NAMED_ARRAY_BUF_OFFSET + 3 + strlen(TEST_ARRAY_NAME) + 1) +MODULE_IMPORT_NS (EXPORTED_FOR_KUNIT_TESTING ); +struct policy_unpack_fixture {struct aa_ext *e;size_t e_size;
@@ -605,3 +608,5 @@ static struct kunit_suite apparmor_policy_unpack_test_module = { };kunit_test_suite (apparmor_policy_unpack_test_module); + +MODULE_LICENSE ("GPL" );
Signature · 1 line
2.38.1.273.g43a17bfeac-goog
John Johansen
maintainerRe: [PATCH v1 2/2] apparmor: test: make static symbols visible during kunit testing
Quoted 78 lines · Rae Moar
Use macros,
VISIBLE_IF_KUNITandEXPORT_SYMBOL_IF_KUNIT, to allow static symbols to be conditionally set to be visible during KUnit testing. Remove the need to include testing file in the implementation file. Provide example of how static symbols can be dealt with in testing.Signed-off-by Rae Moar <[email protected]>--- security/apparmor/Kconfig | 4 +- security/apparmor/Makefile | 2 + security/apparmor/include/policy_unpack.h | 50 ++++++++++++++++ security/apparmor/policy_unpack.c | 72 +++++++---------------- security/apparmor/policy_unpack_test.c | 5 ++ 5 files changed, 80 insertions(+), 53 deletions(-)security/apparmor/Kconfig +2
-2@@ -106,8 +106,8 @@ config SECURITY_APPARMOR_PARANOID_LOAD Disabling the check will speed up policy loads. config SECURITY_APPARMOR_KUNIT_TEST - bool "Build KUnit tests for policy_unpack.c" if !KUNIT_ALL_TESTS - depends on KUNIT=y && SECURITY_APPARMOR + tristate "Build KUnit tests for policy_unpack.c" if !KUNIT_ALL_TESTS + depends on KUNIT && SECURITY_APPARMOR default KUNIT_ALL_TESTS help This builds the AppArmor KUnit tests.security/apparmor/Makefile +2
-0@@ -8,6 +8,8 @@ apparmor-y := apparmorfs.o audit.o capability.o task.o ipc.o lib.o match.o \ resource.o secid.o file.o policy_ns.o label.o mount.o net.o apparmor-$(CONFIG_SECURITY_APPARMOR_HASH) += crypto.o +obj-$(CONFIG_SECURITY_APPARMOR_KUNIT_TEST) += policy_unpack_test.o + clean-files := capability_names.h rlim_names.h net_names.h # Build a lower case string table of address family namessecurity/apparmor/include/policy_unpack.h +27
-0@@ -48,6 +48,43 @@ enum { AAFS_LOADDATA_NDENTS /* count of entries */ }; +/* +* The AppArmor interface treats data as a type byte followed by the +* actual data. The interface has the notion of a named entry +* which has a name (AA_NAME typecode followed by name string) followed by +* the entries typecode and data. Named types allow for optional +* elements and extensions to be added and tested for without breaking +* backwards compatibility. +*/ + +enum aa_code { +AA_U8 , +AA_U16 , +AA_U32 , +AA_U64 , +AA_NAME ,/* same as string except it is items name */ +AA_STRING , +AA_BLOB , +AA_STRUCT , +AA_STRUCTEND , +AA_LIST , +AA_LISTEND , +AA_ARRAY , +AA_ARRAYEND , +}; + +/* +* aa_ext is the read of the buffer containing the serialized profile. The
security/apparmor/include/policy_unpack.h enum { · line 87
Quoted 10 lines
security/apparmor/include/policy_unpack.h +10
-0+* data is copied into a kernel buffer in apparmorfs and then handed off to +* the unpack routines. +*/ +struct aa_ext { +void *start; +void *end; +void *pos;/* pointer to current position in the buffer */ +u32 version; +}; +
hrmmm, I prefer these symbols to be only available to the unpack code but can live with them being more widely available.
Quoted 8 lines
security/apparmor/include/policy_unpack.h +1
-0/* * struct aa_loaddata - buffer of policy raw_data set * @@ -126,4 +163,17 @@ static inline void aa_put_loaddata(struct aa_loaddata *data)kref_put (&data->count, aa_loaddata_kref); } +#if IS_ENABLED (CONFIG_KUNIT )
security/apparmor/include/policy_unpack.h static inline void aa_put_loaddata(struct aa_loaddata *data) · line 176
Quoted 10 lines
security/apparmor/include/policy_unpack.h +10
-0+bool inbounds (struct aa_ext *e,size_t size); +size_t unpack_u16_chunk (struct aa_ext *e,char **chunk); +bool unpack_X (struct aa_ext *e,enum aa_code code); +bool unpack_nameX (struct aa_ext *e,enum aa_code code,const char *name); +bool unpack_u32 (struct aa_ext *e,u32 *data,const char *name); +bool unpack_u64 (struct aa_ext *e,u64 *data,const char *name); +size_t unpack_array (struct aa_ext *e,const char *name); +size_t unpack_blob (struct aa_ext *e,char **blob,const char *name); +int unpack_str (struct aa_ext *e,const char **string,const char *name); +int unpack_strdup (struct aa_ext *e,char **string,const char *name);
So this is a problem. If this symbols are going to be visible outside of the unpack code they need to be prefixed with aa_ to help avoid collisions with other kernel code.
Quoted 183 lines
security/apparmor/include/policy_unpack.h +2
-0+#endif +#endif /* __POLICY_INTERFACE_H */ security/apparmor/policy_unpack.c +20
-47@@ -14,6 +14,7 @@*/ #include <asm/unaligned.h> +#include <kunit/visibility.h> #include <linux/ctype.h> #include <linux/errno.h> #include <linux/zlib.h> @@ -37,43 +38,6 @@#define v7 7#define v8 8/* full network masking */ -/* -* The AppArmor interface treats data as a type byte followed by the -* actual data. The interface has the notion of a named entry -* which has a name (AA_NAME typecode followed by name string) followed by -* the entries typecode and data. Named types allow for optional -* elements and extensions to be added and tested for without breaking -* backwards compatibility. -*/ - -enum aa_code { -AA_U8 , -AA_U16 , -AA_U32 , -AA_U64 , -AA_NAME ,/* same as string except it is items name */ -AA_STRING , -AA_BLOB , -AA_STRUCT , -AA_STRUCTEND , -AA_LIST , -AA_LISTEND , -AA_ARRAY , -AA_ARRAYEND , -}; - -/* -* aa_ext is the read of the buffer containing the serialized profile. The -* data is copied into a kernel buffer in apparmorfs and then handed off to -* the unpack routines. -*/ -struct aa_ext { -void *start; -void *end; -void *pos;/* pointer to current position in the buffer */ -u32 version; -}; -/* audit callback for unpack fields */ static void audit_cb (struct audit_buffer *ab,void *va) {@@ -199,10 +163,11 @@ struct aa_loaddata *aa_loaddata_alloc(size_t size) }/* test if read will be in packed data bounds */ -static bool inbounds (struct aa_ext *e,size_t size) +VISIBLE_IF_KUNIT bool inbounds(struct aa_ext *e, size_t size) {return (size <= e->end - e->pos); } +EXPORT_SYMBOL_IF_KUNIT (inbounds);static void *kvmemdup (const void *src,size_t len) {@@ -220,7 +185,7 @@ static void *kvmemdup(const void *src, size_t len)* * Returns: the size of chunk found with the read head at the end of the chunk. */ -static size_t unpack_u16_chunk (struct aa_ext *e,char **chunk) +VISIBLE_IF_KUNIT size_t unpack_u16_chunk(struct aa_ext *e, char **chunk) {size_t size =0 ;void *pos = e->pos;@@ -239,9 +204,10 @@ static size_t unpack_u16_chunk(struct aa_ext *e, char **chunk) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_u16_chunk);/* unpack control byte */ -static bool unpack_X (struct aa_ext *e,enum aa_code code) +VISIBLE_IF_KUNIT bool unpack_X(struct aa_ext *e, enum aa_code code) {if (!inbounds (e,1 ))return false;@@ -250,6 +216,7 @@ static bool unpack_X(struct aa_ext *e, enum aa_code code) e->pos++;return true; } +EXPORT_SYMBOL_IF_KUNIT (unpack_X);/** * unpack_nameX - check is the next element is of type X with a name of @name @@ -267,7 +234,7 @@ static bool unpack_X(struct aa_ext *e, enum aa_code code)* * Returns: false if either match fails, the read head does not move */ -static bool unpack_nameX (struct aa_ext *e,enum aa_code code,const char *name) +VISIBLE_IF_KUNIT bool unpack_nameX(struct aa_ext *e, enum aa_code code, const char *name) {/* * May need to reset pos if name or type doesn't match @@ -296,6 +263,7 @@ static bool unpack_nameX(struct aa_ext *e, enum aa_code code, const char *name) e->pos = pos;return false; } +EXPORT_SYMBOL_IF_KUNIT (unpack_nameX);static bool unpack_u8 (struct aa_ext *e,u8 *data,const char *name) {@@ -315,7 +283,7 @@ static bool unpack_u8(struct aa_ext *e, u8 *data, const char *name)return false; } -static bool unpack_u32 (struct aa_ext *e,u32 *data,const char *name) +VISIBLE_IF_KUNIT bool unpack_u32(struct aa_ext *e, u32 *data, const char *name) {void *pos = e->pos;@@ -332,8 +300,9 @@ static bool unpack_u32(struct aa_ext *e, u32 *data, const char *name) e->pos = pos;return false; } +EXPORT_SYMBOL_IF_KUNIT (unpack_u32); -static bool unpack_u64 (struct aa_ext *e,u64 *data,const char *name) +VISIBLE_IF_KUNIT bool unpack_u64(struct aa_ext *e, u64 *data, const char *name) {void *pos = e->pos;@@ -350,8 +319,9 @@ static bool unpack_u64(struct aa_ext *e, u64 *data, const char *name) e->pos = pos;return false; } +EXPORT_SYMBOL_IF_KUNIT (unpack_u64); -static size_t unpack_array (struct aa_ext *e,const char *name) +VISIBLE_IF_KUNIT size_t unpack_array(struct aa_ext *e, const char *name) {void *pos = e->pos;@@ -368,8 +338,9 @@ static size_t unpack_array(struct aa_ext *e, const char *name) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_array); -static size_t unpack_blob (struct aa_ext *e,char **blob,const char *name) +VISIBLE_IF_KUNIT size_t unpack_blob(struct aa_ext *e, char **blob, const char *name) {void *pos = e->pos;@@ -390,8 +361,9 @@ static size_t unpack_blob(struct aa_ext *e, char **blob, const char *name) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_blob); -static int unpack_str (struct aa_ext *e,const char **string,const char *name) +VISIBLE_IF_KUNIT int unpack_str(struct aa_ext *e, const char **string, const char *name) {char *src_str;size_t size =0 ;@@ -413,8 +385,9 @@ static int unpack_str(struct aa_ext *e, const char **string, const char *name) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_str); -static int unpack_strdup (struct aa_ext *e,char **string,const char *name) +VISIBLE_IF_KUNIT int unpack_strdup(struct aa_ext *e, char **string, const char *name) {const char *tmp;void *pos = e->pos;
security/apparmor/policy_unpack.c static int unpack_strdup(struct aa_ext *e, char **string, const char *name) · line 409
Quoted 5 lines
security/apparmor/policy_unpack.c +1
-0@@ -432,6 +405,7 @@ static int unpack_strdup(struct aa_ext *e, char **string, const char *name)return res; } +EXPORT_SYMBOL_IF_KUNIT (unpack_strdup);
Again if the symbols are going to be exported they need the aa_ prefix
But I am not sure this is worth doing, exporting a lot of symbols just so the test code can be built as a module doesn't seem worth it to me.
Quoted 37 lines
security/apparmor/policy_unpack.c +0
-4/** @@ -1251,7 +1225,3 @@ int aa_unpack(struct aa_loaddata *udata, struct list_head *lh,return error; } - -#ifdef CONFIG_SECURITY_APPARMOR_KUNIT_TEST -#include "policy_unpack_test.c" -#endif /* CONFIG_SECURITY_APPARMOR_KUNIT_TEST */ security/apparmor/policy_unpack_test.c +5
-0@@ -4,6 +4,7 @@*/ #include <kunit/test.h> +#include <kunit/visibility.h> #include "include/policy.h" #include "include/policy_unpack.h" @@ -43,6 +44,8 @@#define TEST_ARRAY_BUF_OFFSET \ (TEST_NAMED_ARRAY_BUF_OFFSET + 3 + strlen(TEST_ARRAY_NAME) + 1) +MODULE_IMPORT_NS (EXPORTED_FOR_KUNIT_TESTING ); +struct policy_unpack_fixture {struct aa_ext *e;size_t e_size;@@ -605,3 +608,5 @@ static struct kunit_suite apparmor_policy_unpack_test_module = { };kunit_test_suite (apparmor_policy_unpack_test_module); + +MODULE_LICENSE ("GPL" );
Re: [PATCH v1 2/2] apparmor: test: make static symbols visible during kunit testing
Quoted 101 lines · John Johansen
Quoted 78 lines · Rae Moar
Use macros,
VISIBLE_IF_KUNITandEXPORT_SYMBOL_IF_KUNIT, to allow static symbols to be conditionally set to be visible during KUnit testing. Remove the need to include testing file in the implementation file. Provide example of how static symbols can be dealt with in testing.Signed-off-by Rae Moar <[email protected]>--- security/apparmor/Kconfig | 4 +- security/apparmor/Makefile | 2 + security/apparmor/include/policy_unpack.h | 50 ++++++++++++++++ security/apparmor/policy_unpack.c | 72 +++++++---------------- security/apparmor/policy_unpack_test.c | 5 ++ 5 files changed, 80 insertions(+), 53 deletions(-)security/apparmor/Kconfig +2
-2@@ -106,8 +106,8 @@ config SECURITY_APPARMOR_PARANOID_LOAD Disabling the check will speed up policy loads. config SECURITY_APPARMOR_KUNIT_TEST - bool "Build KUnit tests for policy_unpack.c" if !KUNIT_ALL_TESTS - depends on KUNIT=y && SECURITY_APPARMOR + tristate "Build KUnit tests for policy_unpack.c" if !KUNIT_ALL_TESTS + depends on KUNIT && SECURITY_APPARMOR default KUNIT_ALL_TESTS help This builds the AppArmor KUnit tests.security/apparmor/Makefile +2
-0@@ -8,6 +8,8 @@ apparmor-y := apparmorfs.o audit.o capability.o task.o ipc.o lib.o match.o \ resource.o secid.o file.o policy_ns.o label.o mount.o net.o apparmor-$(CONFIG_SECURITY_APPARMOR_HASH) += crypto.o +obj-$(CONFIG_SECURITY_APPARMOR_KUNIT_TEST) += policy_unpack_test.o + clean-files := capability_names.h rlim_names.h net_names.h # Build a lower case string table of address family namessecurity/apparmor/include/policy_unpack.h +27
-0@@ -48,6 +48,43 @@ enum { AAFS_LOADDATA_NDENTS /* count of entries */ }; +/* +* The AppArmor interface treats data as a type byte followed by the +* actual data. The interface has the notion of a named entry +* which has a name (AA_NAME typecode followed by name string) followed by +* the entries typecode and data. Named types allow for optional +* elements and extensions to be added and tested for without breaking +* backwards compatibility. +*/ + +enum aa_code { +AA_U8 , +AA_U16 , +AA_U32 , +AA_U64 , +AA_NAME ,/* same as string except it is items name */ +AA_STRING , +AA_BLOB , +AA_STRUCT , +AA_STRUCTEND , +AA_LIST , +AA_LISTEND , +AA_ARRAY , +AA_ARRAYEND , +}; + +/* +* aa_ext is the read of the buffer containing the serialized profile. The security/apparmor/include/policy_unpack.h enum { · line 87
Quoted 10 lines
security/apparmor/include/policy_unpack.h +10
-0+* data is copied into a kernel buffer in apparmorfs and then handed off to +* the unpack routines. +*/ +struct aa_ext { +void *start; +void *end; +void *pos;/* pointer to current position in the buffer */ +u32 version; +}; +hrmmm, I prefer these symbols to be only available to the unpack code but can live with them being more widely available.
Quoted 7 lines
security/apparmor/include/policy_unpack.h +0
-0/* * struct aa_loaddata - buffer of policy raw_data set * @@ -126,4 +163,17 @@ static inline void aa_put_loaddata(struct aa_loaddata *data)kref_put (&data->count, aa_loaddata_kref); }
Quoted 16 lines
security/apparmor/include/policy_unpack.h static inline void aa_put_loaddata(struct aa_loaddata *data) · line 176
Quoted 11 lines
security/apparmor/include/policy_unpack.h +11
-0+#if IS_ENABLED (CONFIG_KUNIT ) +bool inbounds (struct aa_ext *e,size_t size); +size_t unpack_u16_chunk (struct aa_ext *e,char **chunk); +bool unpack_X (struct aa_ext *e,enum aa_code code); +bool unpack_nameX (struct aa_ext *e,enum aa_code code,const char *name); +bool unpack_u32 (struct aa_ext *e,u32 *data,const char *name); +bool unpack_u64 (struct aa_ext *e,u64 *data,const char *name); +size_t unpack_array (struct aa_ext *e,const char *name); +size_t unpack_blob (struct aa_ext *e,char **blob,const char *name); +int unpack_str (struct aa_ext *e,const char **string,const char *name); +int unpack_strdup (struct aa_ext *e,char **string,const char *name);So this is a problem. If this symbols are going to be visible outside of the unpack code they need to be prefixed with
aa_to help avoid collisions with other kernel code.
Hmm... I agree we need some sort of way of restricting access to these symbols.
As-is, they're _exported_ to a different symbol namespace, so it shouldn't be a problem during linking when built as a module, nor if KUnit is disabled (due to the preprocessor step).
One option is to put these in a separate header (that only the test and policy-unpack code include), but even that doesn't solve the linking problem when built-in.
So I guess namespacing is the only option which solves all of these problems. (It'd be nice if the symbol namespacing system worked for built-ins as well as modules...)
Quoted 184 lines
Quoted 184 lines
security/apparmor/include/policy_unpack.h +2
-0+#endif +#endif /* __POLICY_INTERFACE_H */ security/apparmor/policy_unpack.c +20
-47@@ -14,6 +14,7 @@*/ #include <asm/unaligned.h> +#include <kunit/visibility.h> #include <linux/ctype.h> #include <linux/errno.h> #include <linux/zlib.h> @@ -37,43 +38,6 @@#define v7 7#define v8 8/* full network masking */ -/* -* The AppArmor interface treats data as a type byte followed by the -* actual data. The interface has the notion of a named entry -* which has a name (AA_NAME typecode followed by name string) followed by -* the entries typecode and data. Named types allow for optional -* elements and extensions to be added and tested for without breaking -* backwards compatibility. -*/ - -enum aa_code { -AA_U8 , -AA_U16 , -AA_U32 , -AA_U64 , -AA_NAME ,/* same as string except it is items name */ -AA_STRING , -AA_BLOB , -AA_STRUCT , -AA_STRUCTEND , -AA_LIST , -AA_LISTEND , -AA_ARRAY , -AA_ARRAYEND , -}; - -/* -* aa_ext is the read of the buffer containing the serialized profile. The -* data is copied into a kernel buffer in apparmorfs and then handed off to -* the unpack routines. -*/ -struct aa_ext { -void *start; -void *end; -void *pos;/* pointer to current position in the buffer */ -u32 version; -}; -/* audit callback for unpack fields */ static void audit_cb (struct audit_buffer *ab,void *va) {@@ -199,10 +163,11 @@ struct aa_loaddata *aa_loaddata_alloc(size_t size) }/* test if read will be in packed data bounds */ -static bool inbounds (struct aa_ext *e,size_t size) +VISIBLE_IF_KUNIT bool inbounds(struct aa_ext *e, size_t size) {return (size <= e->end - e->pos); } +EXPORT_SYMBOL_IF_KUNIT (inbounds);static void *kvmemdup (const void *src,size_t len) {@@ -220,7 +185,7 @@ static void *kvmemdup(const void *src, size_t len)* * Returns: the size of chunk found with the read head at the end of the chunk. */ -static size_t unpack_u16_chunk (struct aa_ext *e,char **chunk) +VISIBLE_IF_KUNIT size_t unpack_u16_chunk(struct aa_ext *e, char **chunk) {size_t size =0 ;void *pos = e->pos;@@ -239,9 +204,10 @@ static size_t unpack_u16_chunk(struct aa_ext *e, char **chunk) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_u16_chunk);/* unpack control byte */ -static bool unpack_X (struct aa_ext *e,enum aa_code code) +VISIBLE_IF_KUNIT bool unpack_X(struct aa_ext *e, enum aa_code code) {if (!inbounds (e,1 ))return false;@@ -250,6 +216,7 @@ static bool unpack_X(struct aa_ext *e, enum aa_code code) e->pos++;return true; } +EXPORT_SYMBOL_IF_KUNIT (unpack_X);/** * unpack_nameX - check is the next element is of type X with a name of @name @@ -267,7 +234,7 @@ static bool unpack_X(struct aa_ext *e, enum aa_code code)* * Returns: false if either match fails, the read head does not move */ -static bool unpack_nameX (struct aa_ext *e,enum aa_code code,const char *name) +VISIBLE_IF_KUNIT bool unpack_nameX(struct aa_ext *e, enum aa_code code, const char *name) {/* * May need to reset pos if name or type doesn't match @@ -296,6 +263,7 @@ static bool unpack_nameX(struct aa_ext *e, enum aa_code code, const char *name) e->pos = pos;return false; } +EXPORT_SYMBOL_IF_KUNIT (unpack_nameX);static bool unpack_u8 (struct aa_ext *e,u8 *data,const char *name) {@@ -315,7 +283,7 @@ static bool unpack_u8(struct aa_ext *e, u8 *data, const char *name)return false; } -static bool unpack_u32 (struct aa_ext *e,u32 *data,const char *name) +VISIBLE_IF_KUNIT bool unpack_u32(struct aa_ext *e, u32 *data, const char *name) {void *pos = e->pos;@@ -332,8 +300,9 @@ static bool unpack_u32(struct aa_ext *e, u32 *data, const char *name) e->pos = pos;return false; } +EXPORT_SYMBOL_IF_KUNIT (unpack_u32); -static bool unpack_u64 (struct aa_ext *e,u64 *data,const char *name) +VISIBLE_IF_KUNIT bool unpack_u64(struct aa_ext *e, u64 *data, const char *name) {void *pos = e->pos;@@ -350,8 +319,9 @@ static bool unpack_u64(struct aa_ext *e, u64 *data, const char *name) e->pos = pos;return false; } +EXPORT_SYMBOL_IF_KUNIT (unpack_u64); -static size_t unpack_array (struct aa_ext *e,const char *name) +VISIBLE_IF_KUNIT size_t unpack_array(struct aa_ext *e, const char *name) {void *pos = e->pos;@@ -368,8 +338,9 @@ static size_t unpack_array(struct aa_ext *e, const char *name) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_array); -static size_t unpack_blob (struct aa_ext *e,char **blob,const char *name) +VISIBLE_IF_KUNIT size_t unpack_blob(struct aa_ext *e, char **blob, const char *name) {void *pos = e->pos;@@ -390,8 +361,9 @@ static size_t unpack_blob(struct aa_ext *e, char **blob, const char *name) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_blob); -static int unpack_str (struct aa_ext *e,const char **string,const char *name) +VISIBLE_IF_KUNIT int unpack_str(struct aa_ext *e, const char **string, const char *name) {char *src_str;size_t size =0 ;@@ -413,8 +385,9 @@ static int unpack_str(struct aa_ext *e, const char **string, const char *name) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_str); -static int unpack_strdup (struct aa_ext *e,char **string,const char *name) +VISIBLE_IF_KUNIT int unpack_strdup(struct aa_ext *e, char **string, const char *name) {const char *tmp;void *pos = e->pos;@@ -432,6 +405,7 @@ static int unpack_strdup(struct aa_ext *e, char **string, const char *name)
Quoted 9 lines
security/apparmor/policy_unpack.c static int unpack_strdup(struct aa_ext *e, char **string, const char *name) · line 409
Quoted 4 lines
security/apparmor/policy_unpack.c +1
-0return res; } +EXPORT_SYMBOL_IF_KUNIT (unpack_strdup);Again if the symbols are going to be exported they need the
aa_prefixBut I am not sure this is worth doing, exporting a lot of symbols just so the test code can be built as a module doesn't seem worth it to me.
Again, agreed that we need to namespace these for the non-module case (the symbol namespacing should be okay otherwise).
One of the reasons behind doing this is that there are a few KUnit users who can only run tests which are built as modules. In particular, Android and (IIRC) Red Hat are both configuring all of their kernels with KUnit built as a module, and distributing the KUnit and KUnit test modules in a different package.
If we kept things the way there are, then it'd not be possible to unconditionally _build_ the apparmor tests, but only load and run them on demand (due to the way they're built into the apparmor module, they'd always run when it loads). This is a no-go for Android/Red Hat, so they won't ship or run the apparmor tests. (There are some other tests with the same problem, notably amdgpu, but apparmor seemed a nice first trial-user, as it were, having a small but non-trivial number of symbols to export.)
Thoughts?
Quoted 38 lines
Quoted 37 lines
security/apparmor/policy_unpack.c +0
-4/** @@ -1251,7 +1225,3 @@ int aa_unpack(struct aa_loaddata *udata, struct list_head *lh,return error; } - -#ifdef CONFIG_SECURITY_APPARMOR_KUNIT_TEST -#include "policy_unpack_test.c" -#endif /* CONFIG_SECURITY_APPARMOR_KUNIT_TEST */ security/apparmor/policy_unpack_test.c +5
-0@@ -4,6 +4,7 @@*/ #include <kunit/test.h> +#include <kunit/visibility.h> #include "include/policy.h" #include "include/policy_unpack.h" @@ -43,6 +44,8 @@#define TEST_ARRAY_BUF_OFFSET \ (TEST_NAMED_ARRAY_BUF_OFFSET + 3 + strlen(TEST_ARRAY_NAME) + 1) +MODULE_IMPORT_NS (EXPORTED_FOR_KUNIT_TESTING ); +struct policy_unpack_fixture {struct aa_ext *e;size_t e_size;@@ -605,3 +608,5 @@ static struct kunit_suite apparmor_policy_unpack_test_module = { };kunit_test_suite (apparmor_policy_unpack_test_module); + +MODULE_LICENSE ("GPL" );
John Johansen
maintainerRe: [PATCH v1 2/2] apparmor: test: make static symbols visible during kunit testing
Quoted 120 lines · David Gow
Quoted 101 lines · John Johansen
Quoted 78 lines · Rae Moar
Use macros,
VISIBLE_IF_KUNITandEXPORT_SYMBOL_IF_KUNIT, to allow static symbols to be conditionally set to be visible during KUnit testing. Remove the need to include testing file in the implementation file. Provide example of how static symbols can be dealt with in testing.Signed-off-by Rae Moar <[email protected]>--- security/apparmor/Kconfig | 4 +- security/apparmor/Makefile | 2 + security/apparmor/include/policy_unpack.h | 50 ++++++++++++++++ security/apparmor/policy_unpack.c | 72 +++++++---------------- security/apparmor/policy_unpack_test.c | 5 ++ 5 files changed, 80 insertions(+), 53 deletions(-)security/apparmor/Kconfig +2
-2@@ -106,8 +106,8 @@ config SECURITY_APPARMOR_PARANOID_LOAD Disabling the check will speed up policy loads. config SECURITY_APPARMOR_KUNIT_TEST - bool "Build KUnit tests for policy_unpack.c" if !KUNIT_ALL_TESTS - depends on KUNIT=y && SECURITY_APPARMOR + tristate "Build KUnit tests for policy_unpack.c" if !KUNIT_ALL_TESTS + depends on KUNIT && SECURITY_APPARMOR default KUNIT_ALL_TESTS help This builds the AppArmor KUnit tests.security/apparmor/Makefile +2
-0@@ -8,6 +8,8 @@ apparmor-y := apparmorfs.o audit.o capability.o task.o ipc.o lib.o match.o \ resource.o secid.o file.o policy_ns.o label.o mount.o net.o apparmor-$(CONFIG_SECURITY_APPARMOR_HASH) += crypto.o +obj-$(CONFIG_SECURITY_APPARMOR_KUNIT_TEST) += policy_unpack_test.o + clean-files := capability_names.h rlim_names.h net_names.h # Build a lower case string table of address family namessecurity/apparmor/include/policy_unpack.h +27
-0@@ -48,6 +48,43 @@ enum { AAFS_LOADDATA_NDENTS /* count of entries */ }; +/* +* The AppArmor interface treats data as a type byte followed by the +* actual data. The interface has the notion of a named entry +* which has a name (AA_NAME typecode followed by name string) followed by +* the entries typecode and data. Named types allow for optional +* elements and extensions to be added and tested for without breaking +* backwards compatibility. +*/ + +enum aa_code { +AA_U8 , +AA_U16 , +AA_U32 , +AA_U64 , +AA_NAME ,/* same as string except it is items name */ +AA_STRING , +AA_BLOB , +AA_STRUCT , +AA_STRUCTEND , +AA_LIST , +AA_LISTEND , +AA_ARRAY , +AA_ARRAYEND , +}; + +/* +* aa_ext is the read of the buffer containing the serialized profile. The security/apparmor/include/policy_unpack.h enum { · line 87
Quoted 10 lines
security/apparmor/include/policy_unpack.h +10
-0+* data is copied into a kernel buffer in apparmorfs and then handed off to +* the unpack routines. +*/ +struct aa_ext { +void *start; +void *end; +void *pos;/* pointer to current position in the buffer */ +u32 version; +}; +hrmmm, I prefer these symbols to be only available to the unpack code but can live with them being more widely available.
Quoted 7 lines
security/apparmor/include/policy_unpack.h +0
-0/* * struct aa_loaddata - buffer of policy raw_data set * @@ -126,4 +163,17 @@ static inline void aa_put_loaddata(struct aa_loaddata *data)kref_put (&data->count, aa_loaddata_kref); }Quoted 16 lines
security/apparmor/include/policy_unpack.h static inline void aa_put_loaddata(struct aa_loaddata *data) · line 176
Quoted 11 lines
security/apparmor/include/policy_unpack.h +11
-0+#if IS_ENABLED (CONFIG_KUNIT ) +bool inbounds (struct aa_ext *e,size_t size); +size_t unpack_u16_chunk (struct aa_ext *e,char **chunk); +bool unpack_X (struct aa_ext *e,enum aa_code code); +bool unpack_nameX (struct aa_ext *e,enum aa_code code,const char *name); +bool unpack_u32 (struct aa_ext *e,u32 *data,const char *name); +bool unpack_u64 (struct aa_ext *e,u64 *data,const char *name); +size_t unpack_array (struct aa_ext *e,const char *name); +size_t unpack_blob (struct aa_ext *e,char **blob,const char *name); +int unpack_str (struct aa_ext *e,const char **string,const char *name); +int unpack_strdup (struct aa_ext *e,char **string,const char *name);So this is a problem. If this symbols are going to be visible outside of the unpack code they need to be prefixed with
aa_to help avoid collisions with other kernel code.
Quoted 3 lines
Hmm... I agree we need some sort of way of restricting access to these symbols.
As-is, they're _exported_ to a different symbol namespace, so it
if by exported you mean static. This particular set of symbols is not exported in a header and each function is static. They are accessible to the kunit tests because of some what gross
#ifdef CONFIG_SECURITY_APPARMOR_KUNIT_TEST
#include "policy_unpack_test.c"
#endif /* CONFIG_SECURITY_APPARMOR_KUNIT_TEST */
at the end of the file. And of course this means these tests can't be built as a module
Quoted 3 lines
shouldn't be a problem during linking when built as a module, nor if KUnit is disabled (due to the preprocessor step).
right
Quoted 4 lines
One option is to put these in a separate header (that only the test and policy-unpack code include), but even that doesn't solve the linking problem when built-in.
a separate header might be as a pseudo documentation step but the reality is that I can't see anyone else including the existing header so its not really required.
Quoted 4 lines
So I guess namespacing is the only option which solves all of these problems. (It'd be nice if the symbol namespacing system worked for built-ins as well as modules...)
yeah
Quoted 194 lines
Quoted 184 lines
Quoted 184 lines
security/apparmor/include/policy_unpack.h +2
-0+#endif +#endif /* __POLICY_INTERFACE_H */ security/apparmor/policy_unpack.c +20
-47@@ -14,6 +14,7 @@*/ #include <asm/unaligned.h> +#include <kunit/visibility.h> #include <linux/ctype.h> #include <linux/errno.h> #include <linux/zlib.h> @@ -37,43 +38,6 @@#define v7 7#define v8 8/* full network masking */ -/* -* The AppArmor interface treats data as a type byte followed by the -* actual data. The interface has the notion of a named entry -* which has a name (AA_NAME typecode followed by name string) followed by -* the entries typecode and data. Named types allow for optional -* elements and extensions to be added and tested for without breaking -* backwards compatibility. -*/ - -enum aa_code { -AA_U8 , -AA_U16 , -AA_U32 , -AA_U64 , -AA_NAME ,/* same as string except it is items name */ -AA_STRING , -AA_BLOB , -AA_STRUCT , -AA_STRUCTEND , -AA_LIST , -AA_LISTEND , -AA_ARRAY , -AA_ARRAYEND , -}; - -/* -* aa_ext is the read of the buffer containing the serialized profile. The -* data is copied into a kernel buffer in apparmorfs and then handed off to -* the unpack routines. -*/ -struct aa_ext { -void *start; -void *end; -void *pos;/* pointer to current position in the buffer */ -u32 version; -}; -/* audit callback for unpack fields */ static void audit_cb (struct audit_buffer *ab,void *va) {@@ -199,10 +163,11 @@ struct aa_loaddata *aa_loaddata_alloc(size_t size) }/* test if read will be in packed data bounds */ -static bool inbounds (struct aa_ext *e,size_t size) +VISIBLE_IF_KUNIT bool inbounds(struct aa_ext *e, size_t size) {return (size <= e->end - e->pos); } +EXPORT_SYMBOL_IF_KUNIT (inbounds);static void *kvmemdup (const void *src,size_t len) {@@ -220,7 +185,7 @@ static void *kvmemdup(const void *src, size_t len)* * Returns: the size of chunk found with the read head at the end of the chunk. */ -static size_t unpack_u16_chunk (struct aa_ext *e,char **chunk) +VISIBLE_IF_KUNIT size_t unpack_u16_chunk(struct aa_ext *e, char **chunk) {size_t size =0 ;void *pos = e->pos;@@ -239,9 +204,10 @@ static size_t unpack_u16_chunk(struct aa_ext *e, char **chunk) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_u16_chunk);/* unpack control byte */ -static bool unpack_X (struct aa_ext *e,enum aa_code code) +VISIBLE_IF_KUNIT bool unpack_X(struct aa_ext *e, enum aa_code code) {if (!inbounds (e,1 ))return false;@@ -250,6 +216,7 @@ static bool unpack_X(struct aa_ext *e, enum aa_code code) e->pos++;return true; } +EXPORT_SYMBOL_IF_KUNIT (unpack_X);/** * unpack_nameX - check is the next element is of type X with a name of @name @@ -267,7 +234,7 @@ static bool unpack_X(struct aa_ext *e, enum aa_code code)* * Returns: false if either match fails, the read head does not move */ -static bool unpack_nameX (struct aa_ext *e,enum aa_code code,const char *name) +VISIBLE_IF_KUNIT bool unpack_nameX(struct aa_ext *e, enum aa_code code, const char *name) {/* * May need to reset pos if name or type doesn't match @@ -296,6 +263,7 @@ static bool unpack_nameX(struct aa_ext *e, enum aa_code code, const char *name) e->pos = pos;return false; } +EXPORT_SYMBOL_IF_KUNIT (unpack_nameX);static bool unpack_u8 (struct aa_ext *e,u8 *data,const char *name) {@@ -315,7 +283,7 @@ static bool unpack_u8(struct aa_ext *e, u8 *data, const char *name)return false; } -static bool unpack_u32 (struct aa_ext *e,u32 *data,const char *name) +VISIBLE_IF_KUNIT bool unpack_u32(struct aa_ext *e, u32 *data, const char *name) {void *pos = e->pos;@@ -332,8 +300,9 @@ static bool unpack_u32(struct aa_ext *e, u32 *data, const char *name) e->pos = pos;return false; } +EXPORT_SYMBOL_IF_KUNIT (unpack_u32); -static bool unpack_u64 (struct aa_ext *e,u64 *data,const char *name) +VISIBLE_IF_KUNIT bool unpack_u64(struct aa_ext *e, u64 *data, const char *name) {void *pos = e->pos;@@ -350,8 +319,9 @@ static bool unpack_u64(struct aa_ext *e, u64 *data, const char *name) e->pos = pos;return false; } +EXPORT_SYMBOL_IF_KUNIT (unpack_u64); -static size_t unpack_array (struct aa_ext *e,const char *name) +VISIBLE_IF_KUNIT size_t unpack_array(struct aa_ext *e, const char *name) {void *pos = e->pos;@@ -368,8 +338,9 @@ static size_t unpack_array(struct aa_ext *e, const char *name) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_array); -static size_t unpack_blob (struct aa_ext *e,char **blob,const char *name) +VISIBLE_IF_KUNIT size_t unpack_blob(struct aa_ext *e, char **blob, const char *name) {void *pos = e->pos;@@ -390,8 +361,9 @@ static size_t unpack_blob(struct aa_ext *e, char **blob, const char *name) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_blob); -static int unpack_str (struct aa_ext *e,const char **string,const char *name) +VISIBLE_IF_KUNIT int unpack_str(struct aa_ext *e, const char **string, const char *name) {char *src_str;size_t size =0 ;@@ -413,8 +385,9 @@ static int unpack_str(struct aa_ext *e, const char **string, const char *name) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_str); -static int unpack_strdup (struct aa_ext *e,char **string,const char *name) +VISIBLE_IF_KUNIT int unpack_strdup(struct aa_ext *e, char **string, const char *name) {const char *tmp;void *pos = e->pos;@@ -432,6 +405,7 @@ static int unpack_strdup(struct aa_ext *e, char **string, const char *name)Quoted 9 lines
security/apparmor/policy_unpack.c static int unpack_strdup(struct aa_ext *e, char **string, const char *name) · line 409
Quoted 4 lines
security/apparmor/policy_unpack.c +1
-0return res; } +EXPORT_SYMBOL_IF_KUNIT (unpack_strdup);Again if the symbols are going to be exported they need the
aa_prefixBut I am not sure this is worth doing, exporting a lot of symbols just so the test code can be built as a module doesn't seem worth it to me.
Quoted 9 lines
Again, agreed that we need to namespace these for the non-module case (the symbol namespacing should be okay otherwise).
One of the reasons behind doing this is that there are a few KUnit users who can only run tests which are built as modules. In particular, Android and (IIRC) Red Hat are both configuring all of their kernels with KUnit built as a module, and distributing the KUnit and KUnit test modules in a different package.
ah, okay
Quoted 11 lines
If we kept things the way there are, then it'd not be possible to unconditionally _build_ the apparmor tests, but only load and run them on demand (due to the way they're built into the apparmor module, they'd always run when it loads). This is a no-go for Android/Red Hat, so they won't ship or run the apparmor tests. (There are some other tests with the same problem, notably amdgpu, but apparmor seemed a nice first trial-user, as it were, having a small but non-trivial number of symbols to export.)
Thoughts?
I'm not opposed to exporting them. It has a cost and I wanted there to be a real need before doing it. Since that need has been explained, my only requirement is to make sure they are properly namespaced.
Quoted 38 lines
Quoted 38 lines
Quoted 37 lines
security/apparmor/policy_unpack.c +0
-4/** @@ -1251,7 +1225,3 @@ int aa_unpack(struct aa_loaddata *udata, struct list_head *lh,return error; } - -#ifdef CONFIG_SECURITY_APPARMOR_KUNIT_TEST -#include "policy_unpack_test.c" -#endif /* CONFIG_SECURITY_APPARMOR_KUNIT_TEST */ security/apparmor/policy_unpack_test.c +5
-0@@ -4,6 +4,7 @@*/ #include <kunit/test.h> +#include <kunit/visibility.h> #include "include/policy.h" #include "include/policy_unpack.h" @@ -43,6 +44,8 @@#define TEST_ARRAY_BUF_OFFSET \ (TEST_NAMED_ARRAY_BUF_OFFSET + 3 + strlen(TEST_ARRAY_NAME) + 1) +MODULE_IMPORT_NS (EXPORTED_FOR_KUNIT_TESTING ); +struct policy_unpack_fixture {struct aa_ext *e;size_t e_size;@@ -605,3 +608,5 @@ static struct kunit_suite apparmor_policy_unpack_test_module = { };kunit_test_suite (apparmor_policy_unpack_test_module); + +MODULE_LICENSE ("GPL" );
Re: [PATCH v1 2/2] apparmor: test: make static symbols visible during kunit testing
Quoted 370 lines · John Johansen
Quoted 120 lines · David Gow
Quoted 101 lines · John Johansen
Quoted 78 lines · Rae Moar
Use macros,
VISIBLE_IF_KUNITandEXPORT_SYMBOL_IF_KUNIT, to allow static symbols to be conditionally set to be visible during KUnit testing. Remove the need to include testing file in the implementation file. Provide example of how static symbols can be dealt with in testing.Signed-off-by Rae Moar <[email protected]>--- security/apparmor/Kconfig | 4 +- security/apparmor/Makefile | 2 + security/apparmor/include/policy_unpack.h | 50 ++++++++++++++++ security/apparmor/policy_unpack.c | 72 +++++++---------------- security/apparmor/policy_unpack_test.c | 5 ++ 5 files changed, 80 insertions(+), 53 deletions(-)security/apparmor/Kconfig +2
-2@@ -106,8 +106,8 @@ config SECURITY_APPARMOR_PARANOID_LOAD Disabling the check will speed up policy loads. config SECURITY_APPARMOR_KUNIT_TEST - bool "Build KUnit tests for policy_unpack.c" if !KUNIT_ALL_TESTS - depends on KUNIT=y && SECURITY_APPARMOR + tristate "Build KUnit tests for policy_unpack.c" if !KUNIT_ALL_TESTS + depends on KUNIT && SECURITY_APPARMOR default KUNIT_ALL_TESTS help This builds the AppArmor KUnit tests.security/apparmor/Makefile +2
-0@@ -8,6 +8,8 @@ apparmor-y := apparmorfs.o audit.o capability.o task.o ipc.o lib.o match.o \ resource.o secid.o file.o policy_ns.o label.o mount.o net.o apparmor-$(CONFIG_SECURITY_APPARMOR_HASH) += crypto.o +obj-$(CONFIG_SECURITY_APPARMOR_KUNIT_TEST) += policy_unpack_test.o + clean-files := capability_names.h rlim_names.h net_names.h # Build a lower case string table of address family namessecurity/apparmor/include/policy_unpack.h +27
-0@@ -48,6 +48,43 @@ enum { AAFS_LOADDATA_NDENTS /* count of entries */ }; +/* +* The AppArmor interface treats data as a type byte followed by the +* actual data. The interface has the notion of a named entry +* which has a name (AA_NAME typecode followed by name string) followed by +* the entries typecode and data. Named types allow for optional +* elements and extensions to be added and tested for without breaking +* backwards compatibility. +*/ + +enum aa_code { +AA_U8 , +AA_U16 , +AA_U32 , +AA_U64 , +AA_NAME ,/* same as string except it is items name */ +AA_STRING , +AA_BLOB , +AA_STRUCT , +AA_STRUCTEND , +AA_LIST , +AA_LISTEND , +AA_ARRAY , +AA_ARRAYEND , +}; + +/* +* aa_ext is the read of the buffer containing the serialized profile. The security/apparmor/include/policy_unpack.h enum { · line 87
Quoted 10 lines
security/apparmor/include/policy_unpack.h +10
-0+* data is copied into a kernel buffer in apparmorfs and then handed off to +* the unpack routines. +*/ +struct aa_ext { +void *start; +void *end; +void *pos;/* pointer to current position in the buffer */ +u32 version; +}; +hrmmm, I prefer these symbols to be only available to the unpack code but can live with them being more widely available.
Quoted 7 lines
security/apparmor/include/policy_unpack.h +0
-0/* * struct aa_loaddata - buffer of policy raw_data set * @@ -126,4 +163,17 @@ static inline void aa_put_loaddata(struct aa_loaddata *data)kref_put (&data->count, aa_loaddata_kref); }Quoted 16 lines
security/apparmor/include/policy_unpack.h static inline void aa_put_loaddata(struct aa_loaddata *data) · line 176
Quoted 11 lines
security/apparmor/include/policy_unpack.h +11
-0+#if IS_ENABLED (CONFIG_KUNIT ) +bool inbounds (struct aa_ext *e,size_t size); +size_t unpack_u16_chunk (struct aa_ext *e,char **chunk); +bool unpack_X (struct aa_ext *e,enum aa_code code); +bool unpack_nameX (struct aa_ext *e,enum aa_code code,const char *name); +bool unpack_u32 (struct aa_ext *e,u32 *data,const char *name); +bool unpack_u64 (struct aa_ext *e,u64 *data,const char *name); +size_t unpack_array (struct aa_ext *e,const char *name); +size_t unpack_blob (struct aa_ext *e,char **blob,const char *name); +int unpack_str (struct aa_ext *e,const char **string,const char *name); +int unpack_strdup (struct aa_ext *e,char **string,const char *name);So this is a problem. If this symbols are going to be visible outside of the unpack code they need to be prefixed with
aa_to help avoid collisions with other kernel code.Quoted 3 lines
Hmm... I agree we need some sort of way of restricting access to these symbols.
As-is, they're _exported_ to a different symbol namespace, so it
if by exported you mean static. This particular set of symbols is not exported in a header and each function is static. They are accessible to the kunit tests because of some what gross
#ifdef
CONFIG_SECURITY_APPARMOR_KUNIT_TEST
#include "policy_unpack_test.c"
#endif /*CONFIG_SECURITY_APPARMOR_KUNIT_TEST*/at the end of the file. And of course this means these tests can't be built as a module
Quoted 3 lines
shouldn't be a problem during linking when built as a module, nor if KUnit is disabled (due to the preprocessor step).
right
Quoted 4 lines
One option is to put these in a separate header (that only the test and policy-unpack code include), but even that doesn't solve the linking problem when built-in.
a separate header might be as a pseudo documentation step but the reality is that I can't see anyone else including the existing header so its not really required.
Quoted 4 lines
So I guess namespacing is the only option which solves all of these problems. (It'd be nice if the symbol namespacing system worked for built-ins as well as modules...)
yeah
Quoted 194 lines
Quoted 184 lines
Quoted 184 lines
security/apparmor/include/policy_unpack.h +2
-0+#endif +#endif /* __POLICY_INTERFACE_H */ security/apparmor/policy_unpack.c +20
-47@@ -14,6 +14,7 @@*/ #include <asm/unaligned.h> +#include <kunit/visibility.h> #include <linux/ctype.h> #include <linux/errno.h> #include <linux/zlib.h> @@ -37,43 +38,6 @@#define v7 7#define v8 8/* full network masking */ -/* -* The AppArmor interface treats data as a type byte followed by the -* actual data. The interface has the notion of a named entry -* which has a name (AA_NAME typecode followed by name string) followed by -* the entries typecode and data. Named types allow for optional -* elements and extensions to be added and tested for without breaking -* backwards compatibility. -*/ - -enum aa_code { -AA_U8 , -AA_U16 , -AA_U32 , -AA_U64 , -AA_NAME ,/* same as string except it is items name */ -AA_STRING , -AA_BLOB , -AA_STRUCT , -AA_STRUCTEND , -AA_LIST , -AA_LISTEND , -AA_ARRAY , -AA_ARRAYEND , -}; - -/* -* aa_ext is the read of the buffer containing the serialized profile. The -* data is copied into a kernel buffer in apparmorfs and then handed off to -* the unpack routines. -*/ -struct aa_ext { -void *start; -void *end; -void *pos;/* pointer to current position in the buffer */ -u32 version; -}; -/* audit callback for unpack fields */ static void audit_cb (struct audit_buffer *ab,void *va) {@@ -199,10 +163,11 @@ struct aa_loaddata *aa_loaddata_alloc(size_t size) }/* test if read will be in packed data bounds */ -static bool inbounds (struct aa_ext *e,size_t size) +VISIBLE_IF_KUNIT bool inbounds(struct aa_ext *e, size_t size) {return (size <= e->end - e->pos); } +EXPORT_SYMBOL_IF_KUNIT (inbounds);static void *kvmemdup (const void *src,size_t len) {@@ -220,7 +185,7 @@ static void *kvmemdup(const void *src, size_t len)* * Returns: the size of chunk found with the read head at the end of the chunk. */ -static size_t unpack_u16_chunk (struct aa_ext *e,char **chunk) +VISIBLE_IF_KUNIT size_t unpack_u16_chunk(struct aa_ext *e, char **chunk) {size_t size =0 ;void *pos = e->pos;@@ -239,9 +204,10 @@ static size_t unpack_u16_chunk(struct aa_ext *e, char **chunk) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_u16_chunk);/* unpack control byte */ -static bool unpack_X (struct aa_ext *e,enum aa_code code) +VISIBLE_IF_KUNIT bool unpack_X(struct aa_ext *e, enum aa_code code) {if (!inbounds (e,1 ))return false;@@ -250,6 +216,7 @@ static bool unpack_X(struct aa_ext *e, enum aa_code code) e->pos++;return true; } +EXPORT_SYMBOL_IF_KUNIT (unpack_X);/** * unpack_nameX - check is the next element is of type X with a name of @name @@ -267,7 +234,7 @@ static bool unpack_X(struct aa_ext *e, enum aa_code code)* * Returns: false if either match fails, the read head does not move */ -static bool unpack_nameX (struct aa_ext *e,enum aa_code code,const char *name) +VISIBLE_IF_KUNIT bool unpack_nameX(struct aa_ext *e, enum aa_code code, const char *name) {/* * May need to reset pos if name or type doesn't match @@ -296,6 +263,7 @@ static bool unpack_nameX(struct aa_ext *e, enum aa_code code, const char *name) e->pos = pos;return false; } +EXPORT_SYMBOL_IF_KUNIT (unpack_nameX);static bool unpack_u8 (struct aa_ext *e,u8 *data,const char *name) {@@ -315,7 +283,7 @@ static bool unpack_u8(struct aa_ext *e, u8 *data, const char *name)return false; } -static bool unpack_u32 (struct aa_ext *e,u32 *data,const char *name) +VISIBLE_IF_KUNIT bool unpack_u32(struct aa_ext *e, u32 *data, const char *name) {void *pos = e->pos;@@ -332,8 +300,9 @@ static bool unpack_u32(struct aa_ext *e, u32 *data, const char *name) e->pos = pos;return false; } +EXPORT_SYMBOL_IF_KUNIT (unpack_u32); -static bool unpack_u64 (struct aa_ext *e,u64 *data,const char *name) +VISIBLE_IF_KUNIT bool unpack_u64(struct aa_ext *e, u64 *data, const char *name) {void *pos = e->pos;@@ -350,8 +319,9 @@ static bool unpack_u64(struct aa_ext *e, u64 *data, const char *name) e->pos = pos;return false; } +EXPORT_SYMBOL_IF_KUNIT (unpack_u64); -static size_t unpack_array (struct aa_ext *e,const char *name) +VISIBLE_IF_KUNIT size_t unpack_array(struct aa_ext *e, const char *name) {void *pos = e->pos;@@ -368,8 +338,9 @@ static size_t unpack_array(struct aa_ext *e, const char *name) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_array); -static size_t unpack_blob (struct aa_ext *e,char **blob,const char *name) +VISIBLE_IF_KUNIT size_t unpack_blob(struct aa_ext *e, char **blob, const char *name) {void *pos = e->pos;@@ -390,8 +361,9 @@ static size_t unpack_blob(struct aa_ext *e, char **blob, const char *name) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_blob); -static int unpack_str (struct aa_ext *e,const char **string,const char *name) +VISIBLE_IF_KUNIT int unpack_str(struct aa_ext *e, const char **string, const char *name) {char *src_str;size_t size =0 ;@@ -413,8 +385,9 @@ static int unpack_str(struct aa_ext *e, const char **string, const char *name) e->pos = pos;return 0 ; } +EXPORT_SYMBOL_IF_KUNIT (unpack_str); -static int unpack_strdup (struct aa_ext *e,char **string,const char *name) +VISIBLE_IF_KUNIT int unpack_strdup(struct aa_ext *e, char **string, const char *name) {const char *tmp;void *pos = e->pos;@@ -432,6 +405,7 @@ static int unpack_strdup(struct aa_ext *e, char **string, const char *name)Quoted 9 lines
security/apparmor/policy_unpack.c static int unpack_strdup(struct aa_ext *e, char **string, const char *name) · line 409
Quoted 4 lines
security/apparmor/policy_unpack.c +1
-0return res; } +EXPORT_SYMBOL_IF_KUNIT (unpack_strdup);Again if the symbols are going to be exported they need the
aa_prefixBut I am not sure this is worth doing, exporting a lot of symbols just so the test code can be built as a module doesn't seem worth it to me.
Quoted 9 lines
Again, agreed that we need to namespace these for the non-module case (the symbol namespacing should be okay otherwise).
One of the reasons behind doing this is that there are a few KUnit users who can only run tests which are built as modules. In particular, Android and (IIRC) Red Hat are both configuring all of their kernels with KUnit built as a module, and distributing the KUnit and KUnit test modules in a different package.
ah, okay
Quoted 9 lines
If we kept things the way there are, then it'd not be possible to unconditionally _build_ the apparmor tests, but only load and run them on demand (due to the way they're built into the apparmor module, they'd always run when it loads). This is a no-go for Android/Red Hat, so they won't ship or run the apparmor tests. (There are some other tests with the same problem, notably amdgpu, but apparmor seemed a nice first trial-user, as it were, having a small but non-trivial number of symbols to export.)
Quoted 6 lines
Quoted 2 lines
Thoughts?
I'm not opposed to exporting them. It has a cost and I wanted there to be a real need before doing it. Since that need has been explained, my only requirement is to make sure they are properly namespaced.
Hello!
Thanks to David and John for the helpful comments and sorry for the delay in my response. It seems that there has been somewhat of an agreement that these changes are acceptable but the symbols that will be exported must be properly namespaced with the aa_ prefix. This makes a lot of sense to me and I will go ahead and make a v2 with those changes.
-Rae
Quoted 39 lines
Quoted 38 lines
Quoted 38 lines
Quoted 37 lines
security/apparmor/policy_unpack.c +0
-4/** @@ -1251,7 +1225,3 @@ int aa_unpack(struct aa_loaddata *udata, struct list_head *lh,return error; } - -#ifdef CONFIG_SECURITY_APPARMOR_KUNIT_TEST -#include "policy_unpack_test.c" -#endif /* CONFIG_SECURITY_APPARMOR_KUNIT_TEST */ security/apparmor/policy_unpack_test.c +5
-0@@ -4,6 +4,7 @@*/ #include <kunit/test.h> +#include <kunit/visibility.h> #include "include/policy.h" #include "include/policy_unpack.h" @@ -43,6 +44,8 @@#define TEST_ARRAY_BUF_OFFSET \ (TEST_NAMED_ARRAY_BUF_OFFSET + 3 + strlen(TEST_ARRAY_NAME) + 1) +MODULE_IMPORT_NS (EXPORTED_FOR_KUNIT_TESTING ); +struct policy_unpack_fixture {struct aa_ext *e;size_t e_size;@@ -605,3 +608,5 @@ static struct kunit_suite apparmor_policy_unpack_test_module = { };kunit_test_suite (apparmor_policy_unpack_test_module); + +MODULE_LICENSE ("GPL" );
John Johansen
maintainersorry for the delay on this
Quoted 3 lines · Rae Moar
Currently in order to test a static function, tests must be included in the same translation unit as the function. However, this can cause issues with including implementation and test code in the same file. As an alternative,
And the issue is? An ugly include of C file that only happens if kunit is built. Also that you can't make the kunit code a module but is that really an issue?
Quoted 53 lines
the first patch in this series creates a macro that will set a function to be static or not depending on whether
CONFIG_KUNITis enabled. This allows the function to be visible during testing and static otherwise.As an example, the current status quo to test static functions is:
===
test.c===static void
test_case(struct kunit *test){ KUNIT_EXPECT_EQ(test, my_func_to_test(), 2); }Then the tests are included in the implementation file as a workaround to the issue of testing static functions:
===
implementation.c===static int
my_func_to_test(){...}
...
#include "test.c"Instead, the function could be defined with this new macro:
===
implementation.c===
VISIBLE_IF_KUNITintmy_func_to_test(){...}The first patch also creates a macro that will export a symbol into a kunit testing namespace only if
CONFIG_KUNITis enabled. This follows the logic above and allows symbols to be conditionally exported based on the testing status.The second patch in the series updates the
policy_unpacktest in AppArmor to show an example of how to use both of these macros in order to address the issue of testing static functions.Rae Moar (2):
kunit: add macro to allow conditionally exposing static symbols to
tests
apparmor: test: make static symbols visible during kunit testinginclude/kunit/visibility.h | 32 ++++++++++ security/apparmor/Kconfig | 4 +- security/apparmor/Makefile | 2 + security/apparmor/include/policy_unpack.h | 50 ++++++++++++++++ security/apparmor/policy_unpack.c | 72 +++++++---------------- security/apparmor/policy_unpack_test.c | 5 ++ 6 files changed, 112 insertions(+), 53 deletions(-)create mode 100644
include/kunit/visibility.hbase-commit:
11e76194937b506caf1b49512c42d5c2588681d7