Skip to contentSkip to conversation

kernel module detection (own implementation)

PATCH Superseded Petr Vorel 3 participants 10 messages Jan 19, 2021 – Jan 20, 2021 Library code

v3 of 5 Superseded Latest is v5 →

lore
Share & tools
lore

Petr Vorel

author
#0 lore raw

v3 cover kernel module detection (own implementation)

Hi,

changes v2->v3:
* treat '-' and '_' as the same (follow kmod implementation)
* drop unused MODULES_DIR definition
* code readability (more use of SAFE_ASPRINTF())

It'd deserve a test.

Kind regards,
Petr

Petr Vorel (2):
lib: Fix kernel module detection on BusyBox
zram: Fix module detection on BusyBox

 lib/tst_kernel.c                              | 99 ++++++++++++++++---
 .../kernel/device-drivers/zram/zram_lib.sh    |  6 +-
 2 files changed, 89 insertions(+), 16 deletions(-)
Signature · 1 line
2.30.0
author
#1 lore raw

PATCH v3 1/2 lib: Fix kernel module detection on BusyBox

BusyBox modprobe implementation does not support -n switch.

It does support -D, which could be used, *but* unless is busybox binary configured with CONFIG_MODPROBE_SMALL=y (IMHO the default).

We could use modinfo and grep output for 'filename:', but we agreed on ML that having our own implementation will be the best as it also does not require modinfo as external dependency.

Implementation searches for for module presence in /lib/modules/$(uname -r)/modules.{dep,builtin}. On Android expect files in /system/lib/modules directory.

Also treat '-' and '_' in module name as the same (follow kmod implementation).

On Android still assume all drivers are available because modules.* files might not be available. We could search modules in /system/lib/modules, but to to determine built-in drivers we need modules.builtin (it's required also by Busybox mod{info,probe} implementation).

This fixes many tests on BusyBox, e.g. *all* network tests (tests using tst_net.sh) after 305a78e4c ("tst_net.sh: Require veth for netns").

---
 lib/tst_kernel.c | 99 ++++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 87 insertions(+), 12 deletions(-)
lib/tst_kernel.c +87 -12
index 57fa4b2be..279c8936c 100644
@@ -1,5 +1,6 @@
 /*
  * Copyright (c) 2017 Cyril Hrubis <[email protected]>
+ * Copyright (c) 2020-2021 Petr Vorel <[email protected]>
  *
  * This program is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -17,8 +18,11 @@
 
 #include <sys/personality.h>
 #include <sys/utsname.h>
+#include <limits.h>
+
 #include "test.h"
 #include "tst_kernel.h"
+#include "old_safe_stdio.h"
 
 static int get_kernel_bits_from_uname(struct utsname *buf)
 {
@@ -81,20 +85,91 @@ int tst_kernel_bits(void)
 	return kernel_bits;
 }
 
-int tst_check_driver(const char *name)
+int tst_search_driver(const char *driver, const char *file)
 {
-#ifndef __ANDROID__
-	const char * const argv[] = { "modprobe", "-n", name, NULL };
-	int res = tst_cmd_(NULL, argv, "/dev/null", "/dev/null",
-			       TST_CMD_PASS_RETVAL);
-
-	/* 255 - it looks like modprobe not available */
-	return (res == 255) ? 0 : res;
-#else
-	/* Android modprobe may not have '-n', or properly installed
-	 * module.*.bin files to determine built-in drivers. Assume
-	 * all drivers are available.
+	struct stat st;
+	char *path = NULL, *search = NULL;
+	char buf[PATH_MAX], module[PATH_MAX];
+	FILE *f;
+
+	struct utsname uts;
+
+	if (uname(&uts)) {
+		tst_brkm(TBROK | TERRNO, NULL, "uname() failed");
+		return -1;
+	}
+	SAFE_ASPRINTF(NULL, &path, "/lib/modules/%s/%s", uts.release, file);
+
+	if (stat(path, &st) || !(S_ISREG(st.st_mode) || S_ISLNK(st.st_mode))) {
+		tst_resm(TWARN, "expected file %s does not exist or not a file", path);
+		return -1;
+	}
+
+	if (access(path, R_OK)) {
+		tst_resm(TWARN, "file %s cannot be read", path);
+		return -1;
+	}
+
+	SAFE_ASPRINTF(NULL, &search, "/%s.ko", driver);
+
+	f = SAFE_FOPEN(NULL, path, "r");
+
+	while (fgets(buf, sizeof(buf), f)) {
+		if (sscanf(buf, "%s", module) != 1)
+			continue;
+
+		if (strstr(module, search) != NULL) {
+			SAFE_FCLOSE(NULL, f);
+			return 0;
+		}
+	}
+
+	SAFE_FCLOSE(NULL, f);
+
+	return -1;
+}
+
+int tst_check_driver_(const char *driver)
+{
+	if (!tst_search_driver(driver, "modules.dep") ||
+		!tst_search_driver(driver, "modules.builtin"))
+		return 0;
+
+	return 1;
+}
+
+int tst_check_driver(const char *driver)
+{
+#ifdef __ANDROID__
+	/*
+	 * Android may not have properly installed modules.* files. We could
+	 * search modules in /system/lib/modules, but to to determine built-in
+	 * drivers we need modules.builtin. Therefore assume all drivers are
+	 * available.
 	 */
 	return 0;
 #endif
+
+	if (!tst_check_driver_(driver))
+		return 0;
+
+	if (strrchr(driver, '-') || strrchr(driver, '_')) {
+		char *driver2 = strdup(driver);
+		char *ix = driver2;
+		char find = '-', replace = '_';
+
+		if (strrchr(driver, '_')) {
+			find = '_';
+			replace = '-';
+		}
+
+		while ((ix = strchr(ix, find)) != NULL) {
+			*ix++ = replace;
+		}
+
+		if (!tst_check_driver_(driver2))
+			return 0;
+	}
+
+	return 1;
 }
Signature · 1 line
2.30.0
#2 lore raw

PATCH v3 1/2 lib: Fix kernel module detection on BusyBox

Quoted 71 lines · Petr Vorel

BusyBox modprobe implementation does not support -n switch.

It does support -D, which could be used, *but* unless is busybox binary configured with CONFIG_MODPROBE_SMALL=y (IMHO the default).

We could use modinfo and grep output for 'filename:', but we agreed on ML that having our own implementation will be the best as it also does not require modinfo as external dependency.

Implementation searches for for module presence in /lib/modules/$(uname -r)/modules.{dep,builtin}. On Android expect files in /system/lib/modules directory.

Also treat '-' and '_' in module name as the same (follow kmod implementation).

On Android still assume all drivers are available because modules.* files might not be available. We could search modules in /system/lib/modules, but to to determine built-in drivers we need modules.builtin (it's required also by Busybox mod{info,probe} implementation).

This fixes many tests on BusyBox, e.g. *all* network tests (tests using tst_net.sh) after 305a78e4c ("tst_net.sh: Require veth for netns").

---
 lib/tst_kernel.c | 99 ++++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 87 insertions(+), 12 deletions(-)
lib/tst_kernel.c +7 -12
index 57fa4b2be..279c8936c 100644
@@ -1,5 +1,6 @@
 /*
  * Copyright (c) 2017 Cyril Hrubis <[email protected]>
+ * Copyright (c) 2020-2021 Petr Vorel <[email protected]>
  *
  * This program is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -17,8 +18,11 @@
 
 #include <sys/personality.h>
 #include <sys/utsname.h>
+#include <limits.h>
+
 #include "test.h"
 #include "tst_kernel.h"
+#include "old_safe_stdio.h"
 
 static int get_kernel_bits_from_uname(struct utsname *buf)
 {
@@ -81,20 +85,91 @@ int tst_kernel_bits(void)
 	return kernel_bits;
 }
 
-int tst_check_driver(const char *name)
+int tst_search_driver(const char *driver, const char *file)
 {
-#ifndef __ANDROID__
-	const char * const argv[] = { "modprobe", "-n", name, NULL };
-	int res = tst_cmd_(NULL, argv, "/dev/null", "/dev/null",
-			       TST_CMD_PASS_RETVAL);
-
-	/* 255 - it looks like modprobe not available */
-	return (res == 255) ? 0 : res;
-#else
-	/* Android modprobe may not have '-n', or properly installed
-	 * module.*.bin files to determine built-in drivers. Assume
-	 * all drivers are available.
+	struct stat st;
+	char *path = NULL, *search = NULL;

lib/tst_kernel.c int tst_kernel_bits(void) · line 101

Quoted 10 lines
lib/tst_kernel.c +10 -0
+	char buf[PATH_MAX], module[PATH_MAX];
+	FILE *f;
+
+	struct utsname uts;
+
+	if (uname(&uts)) {
+		tst_brkm(TBROK | TERRNO, NULL, "uname() failed");
+		return -1;
+	}
+	SAFE_ASPRINTF(NULL, &path, "/lib/modules/%s/%s", uts.release, file);

This is just the ramdisk location, the on-disk location is /vendor/lib/modules/. I also think that the ramdisk one goes away after we switch over 2nd stage init. Is there a test I can run that uses these functions now to make sure this works?

Also, unfortunately (and sadly) we may have to do something Android specific downstream as the /vendor/lib/modules and /lib/modules only started to appear as of android 11 :(.

Once you share how I can test, I'm happy to test and add my Tested-by for Android.

+cc: [email protected]

Quoted 35 lines
lib/tst_kernel.c +35 -0
+
+	if (stat(path, &st) || !(S_ISREG(st.st_mode) || S_ISLNK(st.st_mode))) {
+		tst_resm(TWARN, "expected file %s does not exist or not a file", path);
+		return -1;
+	}
+
+	if (access(path, R_OK)) {
+		tst_resm(TWARN, "file %s cannot be read", path);
+		return -1;
+	}
+
+	SAFE_ASPRINTF(NULL, &search, "/%s.ko", driver);
+
+	f = SAFE_FOPEN(NULL, path, "r");
+
+	while (fgets(buf, sizeof(buf), f)) {
+		if (sscanf(buf, "%s", module) != 1)
+			continue;
+
+		if (strstr(module, search) != NULL) {
+			SAFE_FCLOSE(NULL, f);
+			return 0;
+		}
+	}
+
+	SAFE_FCLOSE(NULL, f);
+
+	return -1;
+}
+
+int tst_check_driver_(const char *driver)
+{
+	if (!tst_search_driver(driver, "modules.dep") ||
+		!tst_search_driver(driver, "modules.builtin"))
+		return 0;

lib/tst_kernel.c int tst_kernel_bits(void) · line 146

Quoted 10 lines
lib/tst_kernel.c +10 -0
+
+	return 1;
+}
+
+int tst_check_driver(const char *driver)
+{
+#ifdef __ANDROID__
+	/*
+	 * Android may not have properly installed modules.* files. We could
+	 * search modules in /system/lib/modules, but to to determine built-in

the appropriate location would be /lib/modules OR /vendor/lib/modules.

Quoted 35 lines
lib/tst_kernel.c +25 -0
+	 * drivers we need modules.builtin. Therefore assume all drivers are
+	 * available.
 	 */
 	return 0;
 #endif
+
+	if (!tst_check_driver_(driver))
+		return 0;
+
+	if (strrchr(driver, '-') || strrchr(driver, '_')) {
+		char *driver2 = strdup(driver);
+		char *ix = driver2;
+		char find = '-', replace = '_';
+
+		if (strrchr(driver, '_')) {
+			find = '_';
+			replace = '-';
+		}
+
+		while ((ix = strchr(ix, find)) != NULL) {
+			*ix++ = replace;
+		}
+
+		if (!tst_check_driver_(driver2))
+			return 0;
+	}
+
+	return 1;
 }
Signature · 5 lines
2.30.0


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
author
#3 lore raw

PATCH v3 1/2 lib: Fix kernel module detection on BusyBox

Hi Sandeep,

thanks for your comments!

Quoted 6 lines

lib/tst_kernel.c int tst_kernel_bits(void) · line 101

Quoted 1 line
lib/tst_kernel.c +1 -0
+	SAFE_ASPRINTF(NULL, &path, "/lib/modules/%s/%s", uts.release, file);

This is just the ramdisk location, the on-disk location is /vendor/lib/modules/. I also think that the ramdisk one goes away after we switch over 2nd stage init. Is there a test I can run that uses these functions now to make sure this works?

Any C based test which defines needs_drivers (e.g. fsetxattr02, ioctl08, uevent01, ...) (or shell based tests with TST_NEEDS_DRIVERS, but you probably don't run any shell test).

Quoted 3 lines

Also, unfortunately (and sadly) we may have to do something Android specific downstream as the /vendor/lib/modules and /lib/modules only started to appear as of android 11 :(.

Feel free to send a patch upstream. If not that much complicated and you're willing to maintain it, it might get to upstream (depends on other maintainers, but we're quite open).

Quoted 43 lines

Once you share how I can test, I'm happy to test and add my Tested-by for Android.

+cc: [email protected]

Quoted 39 lines
lib/tst_kernel.c +39 -0
+
+	if (stat(path, &st) || !(S_ISREG(st.st_mode) || S_ISLNK(st.st_mode))) {
+		tst_resm(TWARN, "expected file %s does not exist or not a file", path);
+		return -1;
+	}
+
+	if (access(path, R_OK)) {
+		tst_resm(TWARN, "file %s cannot be read", path);
+		return -1;
+	}
+
+	SAFE_ASPRINTF(NULL, &search, "/%s.ko", driver);
+
+	f = SAFE_FOPEN(NULL, path, "r");
+
+	while (fgets(buf, sizeof(buf), f)) {
+		if (sscanf(buf, "%s", module) != 1)
+			continue;
+
+		if (strstr(module, search) != NULL) {
+			SAFE_FCLOSE(NULL, f);
+			return 0;
+		}
+	}
+
+	SAFE_FCLOSE(NULL, f);
+
+	return -1;
+}
+
+int tst_check_driver_(const char *driver)
+{
+	if (!tst_search_driver(driver, "modules.dep") ||
+		!tst_search_driver(driver, "modules.builtin"))
+		return 0;
+
+	return 1;
+}
+
Quoted 8 lines

lib/tst_kernel.c int tst_kernel_bits(void) · line 146

Quoted 6 lines
lib/tst_kernel.c +6 -0
+int tst_check_driver(const char *driver)
+{
+#ifdef __ANDROID__
+	/*
+	 * Android may not have properly installed modules.* files. We could
+	 * search modules in /system/lib/modules, but to to determine built-in

the appropriate location would be /lib/modules OR /vendor/lib/modules.

OK. I tested only old aosp (Oreo, Nougat and KitKat).
Out of curiosity, does have Android 11 modules.{builtin,dep}? If yes, it'd make sense to treat it as Linux (apply "modules always available" approach only if files aren't available).

Kind regards,
Petr

Quoted 5 lines
Quoted 5 lines
lib/tst_kernel.c +2 -0
+	 * drivers we need modules.builtin. Therefore assume all drivers are
+	 * available.
 	 */
 	return 0;
 #endif
author
#4 lore raw

PATCH v3 1/2 lib: Fix kernel module detection on BusyBox

In reply to Petr Vorel

Hi,

...

Quoted 1 line

+int tst_search_driver(const char *driver, const char *file)

As Joerg pointed out, this must be static.

...

Quoted 1 line

+int tst_check_driver_(const char *driver)

And this one as well.

lib/tst_kernel.c int tst_kernel_bits(void) · line 139

Quoted 7 lines
lib/tst_kernel.c +7 -0
+{
+	if (!tst_search_driver(driver, "modules.dep") ||
+		!tst_search_driver(driver, "modules.builtin"))
+		return 0;
+
+	return 1;
+}

...

Kind regards,
Petr

#5 lore raw

PATCH v3 1/2 lib: Fix kernel module detection on BusyBox

In reply to Petr Vorel

Hi!

Quoted 48 lines

#include <sys/personality.h>
#include <sys/utsname.h>

lib/tst_kernel.c +24 -12
+#include <limits.h>
+
 #include "test.h"
 #include "tst_kernel.h"
+#include "old_safe_stdio.h"
 
 static int get_kernel_bits_from_uname(struct utsname *buf)
 {
@@ -81,20 +85,91 @@ int tst_kernel_bits(void)
 	return kernel_bits;
 }
 
-int tst_check_driver(const char *name)
+int tst_search_driver(const char *driver, const char *file)
 {
-#ifndef __ANDROID__
-	const char * const argv[] = { "modprobe", "-n", name, NULL };
-	int res = tst_cmd_(NULL, argv, "/dev/null", "/dev/null",
-			       TST_CMD_PASS_RETVAL);
-
-	/* 255 - it looks like modprobe not available */
-	return (res == 255) ? 0 : res;
-#else
-	/* Android modprobe may not have '-n', or properly installed
-	 * module.*.bin files to determine built-in drivers. Assume
-	 * all drivers are available.
+	struct stat st;
+	char *path = NULL, *search = NULL;
+	char buf[PATH_MAX], module[PATH_MAX];
+	FILE *f;
+
+	struct utsname uts;
+
+	if (uname(&uts)) {
+		tst_brkm(TBROK | TERRNO, NULL, "uname() failed");
+		return -1;
+	}
+	SAFE_ASPRINTF(NULL, &path, "/lib/modules/%s/%s", uts.release, file);
+
+	if (stat(path, &st) || !(S_ISREG(st.st_mode) || S_ISLNK(st.st_mode))) {
+		tst_resm(TWARN, "expected file %s does not exist or not a file", path);
+		return -1;
+	}
+
+	if (access(path, R_OK)) {
+		tst_resm(TWARN, "file %s cannot be read", path);

lib/tst_kernel.c int tst_kernel_bits(void) · line 119

Quoted 10 lines
lib/tst_kernel.c +10 -0
+		return -1;
+	}
+
+	SAFE_ASPRINTF(NULL, &search, "/%s.ko", driver);
+
+	f = SAFE_FOPEN(NULL, path, "r");
+
+	while (fgets(buf, sizeof(buf), f)) {
+		if (sscanf(buf, "%s", module) != 1)
+			continue;

What is the point in the sscanf() here?

Why can't we just strstr(buf, search) here?

Quoted 38 lines
lib/tst_kernel.c +35 -0
+		if (strstr(module, search) != NULL) {
+			SAFE_FCLOSE(NULL, f);
+			return 0;
+		}
+	}
+
+	SAFE_FCLOSE(NULL, f);
+
+	return -1;
+}
+
+int tst_check_driver_(const char *driver)
+{
+	if (!tst_search_driver(driver, "modules.dep") ||
+		!tst_search_driver(driver, "modules.builtin"))
+		return 0;
+
+	return 1;
+}
+
+int tst_check_driver(const char *driver)
+{
+#ifdef __ANDROID__
+	/*
+	 * Android may not have properly installed modules.* files. We could
+	 * search modules in /system/lib/modules, but to to determine built-in
+	 * drivers we need modules.builtin. Therefore assume all drivers are
+	 * available.
 	 */
 	return 0;
 #endif
+
+	if (!tst_check_driver_(driver))
+		return 0;
+
+	if (strrchr(driver, '-') || strrchr(driver, '_')) {
+		char *driver2 = strdup(driver);
+		char *ix = driver2;

lib/tst_kernel.c int tst_kernel_bits(void) · line 168

Quoted 10 lines
lib/tst_kernel.c +10 -0
+		char find = '-', replace = '_';
+
+		if (strrchr(driver, '_')) {
+			find = '_';
+			replace = '-';
+		}
+
+		while ((ix = strchr(ix, find)) != NULL) {
+			*ix++ = replace;
+		}

Just:

		while ((ix = strchr(ix, find))
			*ix++ = replace;

lib/tst_kernel.c int tst_kernel_bits(void) · line 170

Quoted 1 line
lib/tst_kernel.c +1 -0
+		if (!tst_check_driver_(driver2))

free(driver2) ?

Quoted 8 lines
lib/tst_kernel.c +4 -0
+			return 0;
+	}
+
+	return 1;
 }
Signature · 1 line
2.30.0
Signature · 2 lines
Cyril Hrubis
[email protected]
author
#6 lore raw

PATCH v3 1/2 lib: Fix kernel module detection on BusyBox

Hi Cyril,

...

Quoted 7 lines

lib/tst_kernel.c int tst_kernel_bits(void) · line 119

Quoted 3 lines
lib/tst_kernel.c +3 -0
+	while (fgets(buf, sizeof(buf), f)) {
+		if (sscanf(buf, "%s", module) != 1)
+			continue;

What is the point in the sscanf() here?

Why can't we just strstr(buf, search) here?

modules.dep has format:
module:[dependency [another-dependency ...]]

e.g.:
kernel/arch/x86/kernel/cpu/mce/mce-inject.ko.xz:
kernel/arch/x86/crypto/twofish-x86_64.ko.xz: kernel/crypto/twofish_common.ko.xz kernel/arch/x86/crypto/aesni-intel.ko.xz: kernel/crypto/crypto_simd.ko.xz kernel/crypto/cryptd.ko.xz kernel/arch/x86/crypto/glue_helper.ko.xz

First reading "%s" reads only first module with ':' separator. Searching without it could find first module which is as dependency (e.g. "/twofish_common.ko.xz" instead of "/twofish-x86_64.ko.xz"). Although that shouldn't be a problem, because it's very unlikely that module dependency is missing. Do you want me to drop sscanf() or put some comment?

Also man modules.dep(5) warns about using text format as "their format is subject to change in the future". Hopefully we can depend on it. Or should we use binary format?

...

Quoted 18 lines
Quoted 6 lines
lib/tst_kernel.c +6 -0
+	if (!tst_check_driver_(driver))
+		return 0;
+
+	if (strrchr(driver, '-') || strrchr(driver, '_')) {
+		char *driver2 = strdup(driver);
+		char *ix = driver2;

lib/tst_kernel.c int tst_kernel_bits(void) · line 168

Quoted 10 lines
lib/tst_kernel.c +10 -0
+		char find = '-', replace = '_';
+
+		if (strrchr(driver, '_')) {
+			find = '_';
+			replace = '-';
+		}
+
+		while ((ix = strchr(ix, find)) != NULL) {
+			*ix++ = replace;
+		}

Just:

Quoted 6 lines
		while ((ix = strchr(ix, find))
			*ix++ = replace;

lib/tst_kernel.c int tst_kernel_bits(void) · line 170

Quoted 1 line
lib/tst_kernel.c +1 -0
+		if (!tst_check_driver_(driver2))

free(driver2) ?

Oops, you're right. + path and search path in tst_search_driver() Below are new versions.

Kind regards,
Petr

static int tst_search_driver(const char *driver, const char *file)

{
	struct stat st;
	char *path = NULL, *search = NULL;
	char buf[PATH_MAX], module[PATH_MAX];
	FILE *f;
	int ret = -1;

	struct utsname uts;

	if (uname(&uts)) {
		tst_brkm(TBROK | TERRNO, NULL, "uname() failed");
		return -1;
	}
	SAFE_ASPRINTF(NULL, &path, "/lib/modules/%s/%s", uts.release, file);

	if (stat(path, &st) || !(S_ISREG(st.st_mode) || S_ISLNK(st.st_mode))) {
		tst_resm(TWARN, "expected file %s does not exist or not a file", path);
		return -1;
	}

	if (access(path, R_OK)) {
		tst_resm(TWARN, "file %s cannot be read", path);
		return -1;
	}

	SAFE_ASPRINTF(NULL, &search, "/%s.ko", driver);

	f = SAFE_FOPEN(NULL, path, "r");

	while (fgets(buf, sizeof(buf), f)) {
		if (sscanf(buf, "%s", module) != 1)
			continue;

		if (strstr(module, search) != NULL) {
			ret = 0;
			break;
		}
	}

	SAFE_FCLOSE(NULL, f);
	free(search);
	free(path);

	return ret;
}

static int tst_check_driver_(const char *driver)

{
	if (!tst_search_driver(driver, "modules.dep") ||
		!tst_search_driver(driver, "modules.builtin"))
		return 0;

	return 1;
}

int tst_check_driver(const char *driver)
{
#ifdef __ANDROID__
/*
* Android may not have properly installed modules.* files. We could
* search modules in /system/lib/modules, but to to determine built-in
* drivers we need modules.builtin. Therefore assume all drivers are
* available.
*/
return 0;
#endif

	if (!tst_check_driver_(driver))
		return 0;

	int ret = 1;

	if (strrchr(driver, '-') || strrchr(driver, '_')) {
		char *driver2 = strdup(driver);
		char *ix = driver2;
		char find = '-', replace = '_';

		if (strrchr(driver, '_')) {
			find = '_';
			replace = '-';
		}

		while ((ix = strchr(ix, find)))
			*ix++ = replace;

		ret = tst_check_driver_(driver2);
		free(driver2);
	}

	return ret;
}
#7 lore raw

PATCH v3 1/2 lib: Fix kernel module detection on BusyBox

Hi!

Quoted 8 lines

modules.dep has format:
module:[dependency [another-dependency ...]]

e.g.:
kernel/arch/x86/kernel/cpu/mce/mce-inject.ko.xz:
kernel/arch/x86/crypto/twofish-x86_64.ko.xz: kernel/crypto/twofish_common.ko.xz kernel/arch/x86/crypto/aesni-intel.ko.xz: kernel/crypto/crypto_simd.ko.xz kernel/crypto/cryptd.ko.xz kernel/arch/x86/crypto/glue_helper.ko.xz

Quoted 5 lines

First reading "%s" reads only first module with ':' separator. Searching without it could find first module which is as dependency (e.g. "/twofish_common.ko.xz" instead of "/twofish-x86_64.ko.xz"). Although that shouldn't be a problem, because it's very unlikely that module dependency is missing. Do you want me to drop sscanf() or put some comment?

Well it would be probably cleaner to do something as:

	/* Cut dependencies after : */
	if ((sep = strchr(buf, ':')))
		*sep = 0;

No need to copy the string just because we neet to cut part of it.

Quoted 3 lines

Also man modules.dep(5) warns about using text format as "their format is subject to change in the future". Hopefully we can depend on it. Or should we use binary format?

That depends on how complicated is to parse the binary format...

Signature · 2 lines
Cyril Hrubis
[email protected]
author
#8 lore raw

PATCH v3 1/2 lib: Fix kernel module detection on BusyBox

Hi,

Quoted 14 lines
Quoted 8 lines

modules.dep has format:
module:[dependency [another-dependency ...]]

e.g.:
kernel/arch/x86/kernel/cpu/mce/mce-inject.ko.xz:
kernel/arch/x86/crypto/twofish-x86_64.ko.xz: kernel/crypto/twofish_common.ko.xz kernel/arch/x86/crypto/aesni-intel.ko.xz: kernel/crypto/crypto_simd.ko.xz kernel/crypto/cryptd.ko.xz kernel/arch/x86/crypto/glue_helper.ko.xz

Quoted 5 lines

First reading "%s" reads only first module with ':' separator. Searching without it could find first module which is as dependency (e.g. "/twofish_common.ko.xz" instead of "/twofish-x86_64.ko.xz"). Although that shouldn't be a problem, because it's very unlikely that module dependency is missing. Do you want me to drop sscanf() or put some comment?

Quoted 7 lines

Well it would be probably cleaner to do something as:

	/* Cut dependencies after : */
	if ((sep = strchr(buf, ':')))
		*sep = 0;

No need to copy the string just because we neet to cut part of it.

+1, thanks!

Quoted 5 lines
Quoted 3 lines

Also man modules.dep(5) warns about using text format as "their format is subject to change in the future". Hopefully we can depend on it. Or should we use binary format?

That depends on how complicated is to parse the binary format...

I'll have a look.

I was also thinking whether whole thing really works on toolchains which use BusyBox. BusyBox's depmod implementation does not generate modules.builtin (it generates only modules.{alias,dep,symbols}, i.e. only few text files, no binary). But the only distro I know which installs BusyBox depmod/modinfo/modprobe instead of kmod is Alpine Linux and even this distro clearly generates all modules.* files for its kernel package with kmod.

It looks to me that kernel supports generating modules.builtin.modinfo since v5.2-rc1 (898490c010b5 "moduleparam: Save information about built-in modules in separate file") and kmod used this change in v27 (in 60084cf "libkmod: Add parser for modules.builtin.modinfo" released a year ago). Thus built-in modules are impossible to detect on old kernels no matter we use kmod or our own implementation.

But built-in modules are mostly problem for embedded: we require modules like btrfs, loop, tun, uinput, ... which are loadable modules on typical linux distros.

Thus generally using modules.{dep,builtin} seems to me as a good approach for linux distros including embedded ones and we could also use binary variants.

Other possible improvements:
* last resort effort search in /lib/modules/$(uname -r) or a directory defined in environment variable (with or without version) - fallback for old android or some embedded (again - not working for built-in modules)
* environment variable to skip the detection.

Kind regards,
Petr

author
#9 lore raw

PATCH v3 2/2 zram: Fix module detection on BusyBox

In reply to Petr Vorel

BusyBox modinfo implementation does not exit with 0 when module not found. Our own API implementation used for module detection in tst_check_driver() was fixed in previous commit thus use it.

---
 testcases/kernel/device-drivers/zram/zram_lib.sh | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
testcases/kernel/device-drivers/zram/zram_lib.sh +2 -4
index 3f4d1d55f..bdbf2453a 100755
@@ -1,6 +1,6 @@
 #!/bin/sh
 # Copyright (c) 2015 Oracle and/or its affiliates. All Rights Reserved.
-# Copyright (c) 2019 Petr Vorel <[email protected]>
+# Copyright (c) 2019-2021 Petr Vorel <[email protected]>
 # Author: Alexey Kodanev <[email protected]>
 
 dev_makeswap=-1
@@ -9,6 +9,7 @@ dev_mounted=-1
 TST_NEEDS_TMPDIR=1
 TST_SETUP="zram_load"
 TST_CLEANUP="zram_cleanup"
+TST_NEEDS_DRIVERS="zram"
 . tst_test.sh
 
 zram_cleanup()
@@ -210,6 +211,3 @@ zram_mount()
 
 	tst_res TPASS "mount of zram device(s) succeeded"
 }
-
-modinfo zram > /dev/null 2>&1 ||
-	tst_brk TCONF "zram not configured in kernel"
Signature · 1 line
2.30.0