Skip to contentSkip to conversation

Clang -Wformat warning fixes

PATCH Applied Bill Wendling 11 participants 34 messages Jun 9, 2022 – Jul 11, 2022

Subsystems (14)
lore
Share & tools
lore
#0 lore raw

cover Clang -Wformat warning fixes

This patch set fixes some clang warnings when -Wformat is enabled.

Bill Wendling (12):
x86/mce: use correct format characters x86/CPU/AMD: use correct format characters x86/e820: use correct format characters
blk-cgroup: use correct format characters
fs: quota: use correct format characters
PNP: use correct format characters driver/char: use correct format characters
cdrom: use correct format characters
ALSA: seq: use correct format characters
ALSA: seq: use correct format characters
ALSA: control: use correct format characters
netfilter: conntrack: use correct format characters

 arch/x86/kernel/cpu/mce/amd.c       | 9 +++++----
 arch/x86/kernel/cpu/mce/core.c      | 2 +-
 arch/x86/kernel/e820.c              | 4 ++--
 drivers/cdrom/cdrom.c               | 2 +-
 drivers/char/mem.c                  | 2 +-
 drivers/pnp/interface.c             | 2 +-
 fs/quota/dquot.c                    | 2 +-
 mm/backing-dev.c                    | 2 +-
 net/netfilter/nf_conntrack_helper.c | 2 +-
 scripts/Makefile.extrawarn          | 4 ++--
 sound/core/control.c                | 2 +-
 sound/core/seq/seq_clientmgr.c      | 2 +-
 sound/core/sound.c                  | 2 +-
 13 files changed, 19 insertions(+), 18 deletions(-)
Signature · 1 line
2.36.1.255.ge46751e96f-goog
#1 lore raw

PATCH 1/12 x86/mce: use correct format characters

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

arch/x86/kernel/cpu/mce/core.c:295:9: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

                panic(msg);
                      ^~~

Use a string literal for the format string.

---
 arch/x86/kernel/cpu/mce/core.c | 2 +-
 scripts/Makefile.extrawarn     | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)
arch/x86/kernel/cpu/mce/core.c +1 -1
index 2c8ec5c71712..3d411b7c85ad 100644
@@ -292,7 +292,7 @@ static noinstr void mce_panic(const char *msg, struct mce *final, char *exp)
 	if (!fake_panic) {
 		if (panic_timeout == 0)
 			panic_timeout = mca_cfg.panic_timeout;
-		panic(msg);
+		panic("%s", msg);
 	} else
 		pr_emerg(HW_ERR "Fake kernel panic: %s\n", msg);
 
Signature · 1 line
2.36.1.255.ge46751e96f-goog
#4 lore raw

PATCH 2/12 x86/CPU/AMD: use correct format characters

In reply to Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

arch/x86/kernel/cpu/mce/amd.c:1119:67: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

        err = kobject_init_and_add(&b->kobj, &threshold_ktype, tb->kobj, get_name(cpu, bank, b));
                                                                         ^~~~~~~~~~~~~~~~~~~~~~
arch/x86/kernel/cpu/mce/amd.c:1151:47: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]
        err = kobject_add(&b->blocks->kobj, b->kobj, b->blocks->kobj.name);
                                                     ^~~~~~~~~~~~~~~~~~~~
arch/x86/kernel/cpu/mce/amd.c:1157:42: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]
                err = kobject_add(&pos->kobj, b->kobj, pos->kobj.name);
                                                       ^~~~~~~~~~~~~~
arch/x86/kernel/cpu/mce/amd.c:1187:43: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]
                        err = kobject_add(b->kobj, &dev->kobj, name);
                                                               ^~~~
                                                               "%s",

Use a string literal for the format string.

---
 arch/x86/kernel/cpu/mce/amd.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)
arch/x86/kernel/cpu/mce/amd.c +5 -4
index 1c87501e0fa3..d19bf0eb0abe 100644
@@ -1116,7 +1116,8 @@ static int allocate_threshold_blocks(unsigned int cpu, struct threshold_bank *tb
 	else
 		tb->blocks = b;
 
-	err = kobject_init_and_add(&b->kobj, &threshold_ktype, tb->kobj, get_name(cpu, bank, b));
+	err = kobject_init_and_add(&b->kobj, &threshold_ktype, tb->kobj, "%s",
+				   get_name(cpu, bank, b));
 	if (err)
 		goto out_free;
 recurse:
@@ -1148,13 +1149,13 @@ static int __threshold_add_blocks(struct threshold_bank *b)
 	struct threshold_block *tmp = NULL;
 	int err = 0;
 
-	err = kobject_add(&b->blocks->kobj, b->kobj, b->blocks->kobj.name);
+	err = kobject_add(&b->blocks->kobj, b->kobj, "%s", b->blocks->kobj.name);
 	if (err)
 		return err;
 
 	list_for_each_entry_safe(pos, tmp, head, miscj) {
 
-		err = kobject_add(&pos->kobj, b->kobj, pos->kobj.name);
+		err = kobject_add(&pos->kobj, b->kobj, "%s", pos->kobj.name);
 		if (err) {
 			list_for_each_entry_safe_reverse(pos, tmp, head, miscj)
 				kobject_del(&pos->kobj);
@@ -1184,7 +1185,7 @@ static int threshold_create_bank(struct threshold_bank **bp, unsigned int cpu,
 		if (nb && nb->bank4) {
 			/* yes, use it */
 			b = nb->bank4;
-			err = kobject_add(b->kobj, &dev->kobj, name);
+			err = kobject_add(b->kobj, &dev->kobj, "%s", name);
 			if (err)
 				goto out;
 
Signature · 1 line
2.36.1.255.ge46751e96f-goog
#5 lore raw

PATCH 3/12 x86/e820: use correct format characters

In reply to Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

arch/x86/kernel/e820.c:877:15: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

        early_printk(msg);
                     ^~~
arch/x86/kernel/e820.c:878:8: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]
        panic(msg);
              ^~~

Use a string literal for the format string.

---
 arch/x86/kernel/e820.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/x86/kernel/e820.c +2 -2
index f267205f2d5a..ca4634a0bdb5 100644
@@ -874,8 +874,8 @@ unsigned long __init e820__end_of_low_ram_pfn(void)
 
 static void __init early_panic(char *msg)
 {
-	early_printk(msg);
-	panic(msg);
+	early_printk("%s", msg);
+	panic("%s", msg);
 }
 
 static int userdef __initdata;
Signature · 1 line
2.36.1.255.ge46751e96f-goog
#6 lore raw

PATCH 4/12 blk-cgroup: use correct format characters

In reply to Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

mm/backing-dev.c:880:57: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

        dev = device_create(bdi_class, NULL, MKDEV(0, 0), bdi, bdi->dev_name);
                                                               ^~~~~~~~~~~~~

Use a string literal for the format string.

---
 mm/backing-dev.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/backing-dev.c +1 -1
index ff60bd7d74e0..7b7786dceff3 100644
@@ -877,7 +877,7 @@ int bdi_register_va(struct backing_dev_info *bdi, const char *fmt, va_list args)
 		return 0;
 
 	vsnprintf(bdi->dev_name, sizeof(bdi->dev_name), fmt, args);
-	dev = device_create(bdi_class, NULL, MKDEV(0, 0), bdi, bdi->dev_name);
+	dev = device_create(bdi_class, NULL, MKDEV(0, 0), bdi, "%s", bdi->dev_name);
 	if (IS_ERR(dev))
 		return PTR_ERR(dev);
 
Signature · 1 line
2.36.1.255.ge46751e96f-goog
#8 lore raw

PATCH 5/12 fs: quota: use correct format characters

In reply to Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

fs/quota/dquot.c:206:22: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

                    request_module(module_names[qm].qm_mod_name))
                                   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~

Use a string literal for the format string.

---
 fs/quota/dquot.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/quota/dquot.c +1 -1
index a74aef99bd3d..3b613de3b371 100644
@@ -203,7 +203,7 @@ static struct quota_format_type *find_quota_format(int id)
 			     module_names[qm].qm_fmt_id != id; qm++)
 			;
 		if (!module_names[qm].qm_fmt_id ||
-		    request_module(module_names[qm].qm_mod_name))
+		    request_module("%s", module_names[qm].qm_mod_name))
 			return NULL;
 
 		spin_lock(&dq_list_lock);
Signature · 1 line
2.36.1.255.ge46751e96f-goog
#9 lore raw

PATCH 6/12 PNP: use correct format characters

In reply to Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

drivers/pnp/interface.c:273:22: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

                pnp_printf(buffer, pnp_resource_type_name(res));
                                   ^~~~~~~~~~~~~~~~~~~~~~~~~~~

Use a string literal for the format string.

---
 drivers/pnp/interface.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/pnp/interface.c +1 -1
index 44efcdb87e6f..553221a0c89a 100644
@@ -270,7 +270,7 @@ static ssize_t resources_show(struct device *dmdev,
 	list_for_each_entry(pnp_res, &dev->resources, list) {
 		res = &pnp_res->res;
 
-		pnp_printf(buffer, pnp_resource_type_name(res));
+		pnp_printf(buffer, "%s", pnp_resource_type_name(res));
 
 		if (res->flags & IORESOURCE_DISABLED) {
 			pnp_printf(buffer, " disabled\n");
Signature · 1 line
2.36.1.255.ge46751e96f-goog
#10 lore raw

PATCH 7/12 driver/char: use correct format characters

In reply to Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

drivers/char/mem.c:775:16: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

                              NULL, devlist[minor].name);
                                    ^~~~~~~~~~~~~~~~~~~

Use a string literal for the format string.

---
 drivers/char/mem.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/char/mem.c +1 -1
index 84ca98ed1dad..32d821ba9e4d 100644
@@ -772,7 +772,7 @@ static int __init chr_dev_init(void)
 			continue;
 
 		device_create(mem_class, NULL, MKDEV(MEM_MAJOR, minor),
-			      NULL, devlist[minor].name);
+			      NULL, "%s", devlist[minor].name);
 	}
 
 	return tty_init();
Signature · 1 line
2.36.1.255.ge46751e96f-goog
#13 lore raw

PATCH 8/12 cdrom: use correct format characters

In reply to Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

drivers/cdrom/cdrom.c:3454:48: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

        ret = scnprintf(info + *pos, max_size - *pos, header);
                                                      ^~~~~~

Use a string literal for the format string.

---
 drivers/cdrom/cdrom.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/cdrom/cdrom.c +1 -1
index 416f723a2dbb..52b40120c76e 100644
@@ -3451,7 +3451,7 @@ static int cdrom_print_info(const char *header, int val, char *info,
 	struct cdrom_device_info *cdi;
 	int ret;
 
-	ret = scnprintf(info + *pos, max_size - *pos, header);
+	ret = scnprintf(info + *pos, max_size - *pos, "%s", header);
 	if (!ret)
 		return 1;
 
Signature · 1 line
2.36.1.255.ge46751e96f-goog
#16 lore raw

PATCH 9/12 ALSA: seq: use correct format characters

In reply to Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

sound/core/seq/seq_clientmgr.c:2414:22: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

        snd_iprintf(buffer, msg);
                            ^~~

Use a string literal for the format string.

---
 sound/core/seq/seq_clientmgr.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
sound/core/seq/seq_clientmgr.c +1 -1
index 2e9d695d336c..2340f3e14eeb 100644
@@ -2411,7 +2411,7 @@ static void snd_seq_info_dump_subscribers(struct snd_info_buffer *buffer,
 		up_read(&group->list_mutex);
 		return;
 	}
-	snd_iprintf(buffer, msg);
+	snd_iprintf(buffer, "%s", msg);
 	list_for_each(p, &group->list_head) {
 		if (is_src)
 			s = list_entry(p, struct snd_seq_subscribers, src_list);
Signature · 1 line
2.36.1.255.ge46751e96f-goog
#17 lore raw

PATCH 10/12 ALSA: seq: use correct format characters

In reply to Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

sound/core/sound.c:79:17: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

        request_module(str);
                       ^~~

Use a string literal for the format string.

---
 sound/core/sound.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
sound/core/sound.c +1 -1
index df5571d98629..7866f29621bf 100644
@@ -76,7 +76,7 @@ static void snd_request_other(int minor)
 	case SNDRV_MINOR_TIMER:		str = "snd-timer";	break;
 	default:			return;
 	}
-	request_module(str);
+	request_module("%s", str);
 }
 
 #endif	/* modular kernel */
Signature · 1 line
2.36.1.255.ge46751e96f-goog
#18 lore raw

PATCH 11/12 ALSA: control: use correct format characters

In reply to Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

sound/core/control.c:2062:24: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

        return request_module(module_name);
                              ^~~~~~~~~~~

Use a string literal for the format string.

---
 sound/core/control.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
sound/core/control.c +1 -1
index a25c0d64d104..a1778137147d 100644
@@ -2059,7 +2059,7 @@ int snd_ctl_request_layer(const char *module_name)
 	up_read(&snd_ctl_layer_rwsem);
 	if (lops)
 		return 0;
-	return request_module(module_name);
+	return request_module("%s", module_name);
 }
 EXPORT_SYMBOL_GPL(snd_ctl_request_layer);
 
Signature · 1 line
2.36.1.255.ge46751e96f-goog
#19 lore raw

PATCH 12/12 netfilter: conntrack: use correct format characters

In reply to Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

net/netfilter/nf_conntrack_helper.c:168:18: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

                request_module(mod_name);
                               ^~~~~~~~

Use a string literal for the format string.

---
 net/netfilter/nf_conntrack_helper.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/netfilter/nf_conntrack_helper.c +1 -1
index c12a87ebc3ee..1e0424d37abc 100644
@@ -165,7 +165,7 @@ nf_nat_helper_try_module_get(const char *name, u16 l3num, u8 protonum)
 	if (!nat) {
 		snprintf(mod_name, sizeof(mod_name), "%s", h->nat_mod_name);
 		rcu_read_unlock();
-		request_module(mod_name);
+		request_module("%s", mod_name);
 
 		rcu_read_lock();
 		nat = nf_conntrack_nat_helper_find(mod_name);
Signature · 1 line
2.36.1.255.ge46751e96f-goog
maintainer
#22 lore raw
In reply to Bill Wendling
Quoted 2 lines · On Thu, 9 Jun 2022 22:16:19 +0000 Bill Wendling <[email protected]> wrote:

This patch set fixes some clang warnings when -Wformat is enabled.

tldr:

- printk(msg);
+ printk("%s", msg);

the only reason to make this change is where msg' could contain a %'. Generally, it came from userspace. Otherwise these changes are a useless consumer of runtime resources.

I think it would be better to quieten clang in some fashion.

#23 lore raw
Quoted 11 lines · Andrew Morton
Quoted 2 lines · On Thu, 9 Jun 2022 22:16:19 +0000 Bill Wendling <[email protected]> wrote:

This patch set fixes some clang warnings when -Wformat is enabled.

tldr:

(quoted diff) +1 -1
-       printk(msg);
+       printk("%s", msg);
Quoted 2 lines

the only reason to make this change is where msg' could contain a %'. Generally, it came from userspace.

It helps kernel developers not accidentally to insert an unescaped '%' in their messages, potentially exposing their code to an attack vector.

Quoted 2 lines

Otherwise these changes are a
useless consumer of runtime resources.

Calling a "printf" style function is already insanely expensive. :-) I understand that it's not okay blithely to increase runtime resources simply because it's already slow, but in this case it's worthwhile.

Quoted 1 line

I think it would be better to quieten clang in some fashion.

The "printk" and similar functions all have the "__printf" attribute. I don't know of a modification to that attribute which can turn off this type of check.

-bw

#24 lore raw
Quoted 10 lines · Bill Wendling
Quoted 9 lines · Andrew Morton
Quoted 1 line · On Thu, 9 Jun 2022 22:16:19 +0000 Bill Wendling <[email protected]> wrote:

This patch set fixes some clang warnings when -Wformat is enabled.

tldr:

(quoted diff) +1 -1
-       printk(msg);
+       printk("%s", msg);
Quoted 8 lines
Quoted 2 lines

Otherwise these changes are a
useless consumer of runtime resources.

Calling a "printf" style function is already insanely expensive. [...]
The "printk" and similar functions all have the "__printf" attribute. I don't know of a modification to that attribute which can turn off this type of check.

Perhaps you can split vprintk_store in the middle (after the call to vsnprintf), and offer the second half as a function of its own (e.g. "puts"). Then the tldr could be

- printk(msg);
+ puts(msg);

#2 lore raw

Re: [PATCH 01/12] x86/mce: use correct format characters

In reply to Bill Wendling
Quoted 8 lines · Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

arch/x86/kernel/cpu/mce/core.c:295:9: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

                panic(msg);
                      ^~~
Quoted 7 lines

Use a string literal for the format string.

---
 arch/x86/kernel/cpu/mce/core.c | 2 +-
 scripts/Makefile.extrawarn     | 4 ++--

Where is the scripts/ change?

Quoted 15 lines
 2 files changed, 3 insertions(+), 3 deletions(-)
arch/x86/kernel/cpu/mce/core.c +1 -1
index 2c8ec5c71712..3d411b7c85ad 100644
@@ -292,7 +292,7 @@ static noinstr void mce_panic(const char *msg, struct mce *final, char *exp)
 	if (!fake_panic) {
 		if (panic_timeout == 0)
 			panic_timeout = mca_cfg.panic_timeout;
-		panic(msg);
+		panic("%s", msg);
 	} else
 		pr_emerg(HW_ERR "Fake kernel panic: %s\n", msg);
 
Signature · 1 line
~Randy
#25 lore raw
In reply to Jan Engelhardt
Quoted 20 lines · Jan Engelhardt
Quoted 10 lines · Bill Wendling
Quoted 9 lines · Andrew Morton
Quoted 1 line · On Thu, 9 Jun 2022 22:16:19 +0000 Bill Wendling <[email protected]> wrote:

This patch set fixes some clang warnings when -Wformat is enabled.

tldr:

(quoted diff) +1 -1
-       printk(msg);
+       printk("%s", msg);
Quoted 8 lines
Quoted 2 lines

Otherwise these changes are a
useless consumer of runtime resources.

Calling a "printf" style function is already insanely expensive. [...]
The "printk" and similar functions all have the "__printf" attribute. I don't know of a modification to that attribute which can turn off this type of check.

Quoted 6 lines

Perhaps you can split vprintk_store in the middle (after the call to vsnprintf), and offer the second half as a function of its own (e.g. "puts"). Then the tldr could be

- printk(msg);
+ puts(msg);

That might be a nice compromise. Andrew, what do you think?

-bw

#3 lore raw

Re: [PATCH 01/12] x86/mce: use correct format characters

In reply to Randy Dunlap
Quoted 11 lines · Randy Dunlap
Quoted 10 lines · Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

arch/x86/kernel/cpu/mce/core.c:295:9: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

                panic(msg);
                      ^~~

Use a string literal for the format string.

Quoted 8 lines
Quoted 5 lines
---
 arch/x86/kernel/cpu/mce/core.c | 2 +-
 scripts/Makefile.extrawarn     | 4 ++--

Where is the scripts/ change?

I'm sorry about this. The change in that file was a mistake that I reverted, but I forgot to change this part.

-bw

Quoted 18 lines
Quoted 7 lines
 2 files changed, 3 insertions(+), 3 deletions(-)
arch/x86/kernel/cpu/mce/core.c +0 -0
index 2c8ec5c71712..3d411b7c85ad 100644

arch/x86/kernel/cpu/mce/core.c static noinstr void mce_panic(const char *msg, struct mce *final, char *exp) · line 298

Quoted 8 lines
arch/x86/kernel/cpu/mce/core.c +1 -1
@@ -292,7 +292,7 @@ static noinstr void mce_panic(const char *msg, struct mce *final, char *exp)
      if (!fake_panic) {
              if (panic_timeout == 0)
                      panic_timeout = mca_cfg.panic_timeout;
-             panic(msg);
+             panic("%s", msg);
      } else
              pr_emerg(HW_ERR "Fake kernel panic: %s\n", msg);
 

--
~Randy

#33 lore raw
In reply to Andrew Morton
Quoted 4 lines · Andrew Morton
Quoted 1 line · On Thu, 9 Jun 2022 22:16:19 +0000 Bill Wendling <[email protected]> wrote:

This patch set fixes some clang warnings when -Wformat is enabled.

It looks like this series fixes -Wformat-security, which while being a member of the -Wformat group, is intentionally disabled in the kernel and somewhat orthogonal to enabling -Wformat with Clang.

-Wformat is a group flag (like -Wall) that enables multiple other flags implicitly. Reading through clang/include/clang/Basic/DiagnosticGroups.td in clang's sources, it looks like:

1. -Wformat is a group flag.
2. -Wformat-security is a member of the -Wformat group; enabling -Wformat will enable -Wformat-security.
3. -Wformat itself is a member of -Wmost (never heard of -Wmost, but w/e). So -Wmost will enable -Wformat will enable -Wformat-security.
4. -Wmost is itself a member of -Wall. -Wall enables -Wmost enables -Wformat enables -Wformat security.

Looking now at Kbuild:
1. Makefile:523 adds -Wall to KBUILD_CFLAGS.
2. The same assignment expression but on line 526 immediately disables -Wformat-security via -Wno-format-security.
3. scripts/Makefile.extrawarn disables -Wformat via -Wno-format only for clang (via guard of CONFIG_CC_IS_CLANG).

We _want_ -Wformat enabled for clang so that developers aren't sending patches that trigger -Wformat with GCC (if they didn't happen to test their code with both). It's disabled for clang until we can build the kernel cleanly with it enabled, which we'd like to do.

I don't think that we need to enable -Wformat-security to build with -Wformat for clang.

I suspect based on Randy's comment on patch 1/12 that perhaps -Wformat was _added_ to KBUILD_CFLAGS in scripts/Makefile.extrawarn rather than -Wno-format being _removed_. The former would re-enable -Wformat-security due to the grouping logic described above. The latter is probably closer to our ultimate goal of enabling -Wformat coverage for clang (or rather not disabling the coverage via -Wno-format; a double negative).

I'm pretty sure the kernel doesn't support %n in format strings...see the comment above vsnprintf in lib/vsprintf.c. Are there other attacks other than %n that -Wformat-security guards against? Maybe there's some context on the commit that added -Wno-format-security to the kernel? Regardless, I don't think enabling -Wformat-security is a blocker for enabling -Wformat (or...disabling -Wno-format...two sides of the same coin) for clang.

Quoted 12 lines
Quoted 1 line

tldr:

(quoted diff) +1 -1
-       printk(msg);
+       printk("%s", msg);

the only reason to make this change is where msg' could contain a %'. Generally, it came from userspace. Otherwise these changes are a useless consumer of runtime resources.

I think it would be better to quieten clang in some fashion.

Signature · 2 lines
Thanks,
~Nick Desaulniers
maintainer
#26 lore raw
In reply to Bill Wendling
Quoted 21 lines · On Thu, 9 Jun 2022 16:16:16 -0700 Bill Wendling <[email protected]> wrote:
Quoted 20 lines · Jan Engelhardt
Quoted 18 lines · Bill Wendling
Quoted 11 lines · Andrew Morton
Quoted 1 line · On Thu, 9 Jun 2022 22:16:19 +0000 Bill Wendling <[email protected]> wrote:

This patch set fixes some clang warnings when -Wformat is enabled.

tldr:

(quoted diff) +1 -1
-       printk(msg);
+       printk("%s", msg);

Otherwise these changes are a
useless consumer of runtime resources.

Calling a "printf" style function is already insanely expensive. [...]
The "printk" and similar functions all have the "__printf" attribute. I don't know of a modification to that attribute which can turn off this type of check.

Quoted 9 lines
Quoted 6 lines

Perhaps you can split vprintk_store in the middle (after the call to vsnprintf), and offer the second half as a function of its own (e.g. "puts"). Then the tldr could be

- printk(msg);
+ puts(msg);

That might be a nice compromise. Andrew, what do you think?

Sure. It's surprising that we don't already have a puts().

#11 lore raw

Re: [PATCH 07/12] driver/char: use correct format characters

In reply to Bill Wendling
Quoted 1 line · Bill Wendling

From: Bill Wendling <[email protected]>

Why isn't that matching your From: line in the email?

Quoted 2 lines

When compiling with -Wformat, clang emits the following warnings:

Is that ever a default build option for the kernel?

Quoted 18 lines

drivers/char/mem.c:775:16: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

                              NULL, devlist[minor].name);
                                    ^~~~~~~~~~~~~~~~~~~

Use a string literal for the format string.

---
 drivers/char/mem.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/char/mem.c +0 -0
index 84ca98ed1dad..32d821ba9e4d 100644

drivers/char/mem.c static int __init chr_dev_init(void) · line 775

Quoted 5 lines
drivers/char/mem.c +1 -1
@@ -772,7 +772,7 @@ static int __init chr_dev_init(void)
 			continue;
 
 		device_create(mem_class, NULL, MKDEV(MEM_MAJOR, minor),
-			      NULL, devlist[minor].name);
+			      NULL, "%s", devlist[minor].name);

Please explain how this static string can ever be user controlled.

thanks,

greg k-h

#27 lore raw
In reply to Bill Wendling
Quoted 21 lines · Bill Wendling
Quoted 20 lines · Jan Engelhardt
Quoted 18 lines · Bill Wendling
Quoted 11 lines · Andrew Morton
Quoted 1 line · On Thu, 9 Jun 2022 22:16:19 +0000 Bill Wendling <[email protected]> wrote:

This patch set fixes some clang warnings when -Wformat is enabled.

tldr:

(quoted diff) +1 -1
-       printk(msg);
+       printk("%s", msg);

Otherwise these changes are a
useless consumer of runtime resources.

Calling a "printf" style function is already insanely expensive. [...]
The "printk" and similar functions all have the "__printf" attribute. I don't know of a modification to that attribute which can turn off this type of check.

Quoted 8 lines
Quoted 6 lines

Perhaps you can split vprintk_store in the middle (after the call to vsnprintf), and offer the second half as a function of its own (e.g. "puts"). Then the tldr could be

- printk(msg);
+ puts(msg);

That might be a nice compromise. Andrew, what do you think?

You would need to do that for all of the dev_printk() variants, so I doubt that would ever be all that useful as almost no one should be using a "raw" printk() these days.

thanks,

greg k-h

#7 lore raw

Re: [PATCH 04/12] blk-cgroup: use correct format characters

In reply to Bill Wendling

mm/backing-dev.c int bdi_register_va(struct backing_dev_info *bdi, const char *fmt, va_list args) · line 880

Quoted 3 lines · Bill Wendling
mm/backing-dev.c +1 -1
 	vsnprintf(bdi->dev_name, sizeof(bdi->dev_name), fmt, args);
-	dev = device_create(bdi_class, NULL, MKDEV(0, 0), bdi, bdi->dev_name);
+	dev = device_create(bdi_class, NULL, MKDEV(0, 0), bdi, "%s", bdi->dev_name);

Please avoid the overly long line. But given that bdi names aren't user input this warning seems to shoot from the hip a bit.

#29 lore raw
In reply to Bill Wendling

From: Bill Wendling

Quoted 21 lines

Sent: 09 June 2022 23:49

Quoted 11 lines · Andrew Morton
Quoted 2 lines · On Thu, 9 Jun 2022 22:16:19 +0000 Bill Wendling <[email protected]> wrote:

This patch set fixes some clang warnings when -Wformat is enabled.

tldr:

(quoted diff) +1 -1
-       printk(msg);
+       printk("%s", msg);
Quoted 2 lines

the only reason to make this change is where msg' could contain a %'. Generally, it came from userspace.

It helps kernel developers not accidentally to insert an unescaped '%' in their messages, potentially exposing their code to an attack vector.

Quoted 6 lines
Quoted 2 lines

Otherwise these changes are a
useless consumer of runtime resources.

Calling a "printf" style function is already insanely expensive. :-) I understand that it's not okay blithely to increase runtime resources simply because it's already slow, but in this case it's worthwhile.

Yep, IMHO definitely should be fixed.
It is even possible that using "%s" is faster because the printf code doesn't have to scan the string for format effectors.

Quoted 5 lines
Quoted 1 line

I think it would be better to quieten clang in some fashion.

The "printk" and similar functions all have the "__printf" attribute. I don't know of a modification to that attribute which can turn off this type of check.

And you wouldn't want to for these cases.

The only problems arise when the format is calculated (or passed in from a caller).
But that is likely to be dangerous - reading formats from files (eg for language translation) isn't a good idea at all.

David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)

#30 lore raw
Quoted 8 lines · David Laight
Quoted 4 lines

Calling a "printf" style function is already insanely expensive. :-) I understand that it's not okay blithely to increase runtime resources simply because it's already slow, but in this case it's worthwhile.

Yep, IMHO definitely should be fixed.
It is even possible that using "%s" is faster because the printf code doesn't have to scan the string for format effectors.

I see no special handling; the vsnprintf function just loops over fmt as usual and I see no special casing of fmt by e.g. strcmp(fmt, "%s") == 0 to take a shortcut.

#31 lore raw

From: Jan Engelhardt

Quoted 9 lines

Sent: 10 June 2022 09:32

Quoted 5 lines · David Laight
Quoted 4 lines

Calling a "printf" style function is already insanely expensive. :-) I understand that it's not okay blithely to increase runtime resources simply because it's already slow, but in this case it's worthwhile.

Quoted 7 lines
Quoted 3 lines

Yep, IMHO definitely should be fixed.
It is even possible that using "%s" is faster because the printf code doesn't have to scan the string for format effectors.

I see no special handling; the vsnprintf function just loops over fmt as usual and I see no special casing of fmt by e.g. strcmp(fmt, "%s") == 0 to take a shortcut.

Consider the difference between:

	printf("fubar");
and
	printf("%s", "fubar");

In the former all of "fubar" is checked for '%'. In the latter only the length of "fubar" has to be counted.

FWIW the patch description should probably by:
use "%s" when formatting a single string.
(or something to that effect).

David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)

#32 lore raw
Quoted 8 lines · David Laight
Quoted 7 lines
Quoted 3 lines

Yep, IMHO definitely should be fixed.
It is even possible that using "%s" is faster because the printf code doesn't have to scan the string for format effectors.

I see no special handling; the vsnprintf function just loops over fmt as usual and I see no special casing of fmt by e.g. strcmp(fmt, "%s") == 0 to take a shortcut.

Quoted 6 lines

Consider the difference between:

	printf("fubar");
and
	printf("%s", "fubar");

In the former all of "fubar" is checked for '%'. In the latter only the length of "fubar" has to be counted.

To check the length of "fubar", printf first needs to know that there even is an argument to be pulled from the stack, which it does by evaluating the format string.

So, in fairness, it's more like:

Quoted 1 line
Quoted 1 line

In the latter, all of "%s" is checked for '%'.

#28 lore raw
In reply to Greg Kroah-Hartman
Quoted 10 lines · Greg Kroah-Hartman
Quoted 9 lines · Bill Wendling
Quoted 8 lines · Jan Engelhardt
Quoted 7 lines · Bill Wendling
Quoted 6 lines · Andrew Morton
Quoted 1 line · On Thu, 9 Jun 2022 22:16:19 +0000 Bill Wendling <[email protected]> wrote:

This patch set fixes some clang warnings when -Wformat is enabled.

tldr:

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 5 lines
(quoted diff) +1 -1
-       printk(msg);
+       printk("%s", msg);

Otherwise these changes are a
useless consumer of runtime resources.

Calling a "printf" style function is already insanely expensive.

I expect the printk code itself dominates, not the % scan cost.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Perhaps you can split vprintk_store in the middle (after the call to vsnprintf), and offer the second half as a function of its own (e.g. "puts"). Then the tldr could be

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

- printk(msg);
+ puts(msg);

That might be a nice compromise. Andrew, what do you think?

You would need to do that for all of the dev_printk() variants, so I doubt that would ever be all that useful as almost no one should be using a "raw" printk() these days.

True. The kernel has ~20K variants like that.

$ git grep -P '\b(?:(?:\w+_){1,3}(?:alert|emerg|crit|err|warn|notice|info|cont|debug|dbg)|printk)\s*\(".*"\s*\)\s*;' | wc -l
21160

That doesn't include the ~3K uses like

#define foo "bar"
	printk(foo);
$ git grep -P '\b(?:(?:\w+_){1,3}(?:alert|emerg|crit|err|warn|info|notice|debug|dbg|cont)|printk)\s*\((?:\s*\w+){1,3}\s*\)\s*;'|wc -l
2922

There are apparently only a few hundred uses of variants like:

	printk("%s", foo)
$ git grep -P '\b(?:(?:\w+_){1,3}(?:alert|emerg|crit|err|warn|info|notice|debug|dbg|cont)|printk)\s*\(\s*"%s(?:\\n)?"\s*,\s*(?:".*"|\w+)\s*\)\s*;' | wc -l
305

unless I screwed up my greps (which of course is quite possible)

maintainer
#14 lore raw

Re: [PATCH 08/12] cdrom: use correct format characters

In reply to Bill Wendling
Quoted 32 lines · Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

drivers/cdrom/cdrom.c:3454:48: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

        ret = scnprintf(info + *pos, max_size - *pos, header);
                                                      ^~~~~~

Use a string literal for the format string.

---
 drivers/cdrom/cdrom.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/cdrom/cdrom.c +1 -1
index 416f723a2dbb..52b40120c76e 100644
@@ -3451,7 +3451,7 @@ static int cdrom_print_info(const char *header, int val, char *info,
 	struct cdrom_device_info *cdi;
 	int ret;
 
-	ret = scnprintf(info + *pos, max_size - *pos, header);
+	ret = scnprintf(info + *pos, max_size - *pos, "%s", header);
 	if (!ret)
 		return 1;
 
Signature · 1 line
2.36.1.255.ge46751e96f-goog

Hi Bill,

Thank you for the patch, much appreciated.

Looking at this though, all callers of cdrom_print_info() provide 'header' as a string literal defined within the driver, when making the call. Therefore, I'm not convinced this change is necessary for cdrom.c - that said, in this particular use case I don't think it would hurt either.

I've followed the other responses on parts of this series, so I understand that a different solution is potentially in the works. Thought I'd respond anyway though out of courtesy.

All the best,
Phil (Uniform CDROM Maintainer)

#12 lore raw

Re: [PATCH 07/12] driver/char: use correct format characters

In reply to Greg Kroah-Hartman
Quoted 6 lines · Greg Kroah-Hartman
Quoted 1 line · Bill Wendling

From: Bill Wendling <[email protected]>

Why isn't that matching your From: line in the email?

There must be something wrong with my .gitconfig file. I"ll check into it.

Quoted 5 lines
Quoted 2 lines

When compiling with -Wformat, clang emits the following warnings:

Is that ever a default build option for the kernel?

We want to enable -Wformat for clang. I believe that these specific warnings have been disabled, but I'm confused as to why, because they're valid warnings. When I compiled with the warning enabled, there were only a few (12) places that needed changes, so thought that patches would be a nice cleanup, even though the warning itself is disabled.

Quoted 19 lines
Quoted 19 lines

drivers/char/mem.c:775:16: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

                              NULL, devlist[minor].name);
                                    ^~~~~~~~~~~~~~~~~~~

Use a string literal for the format string.

---
 drivers/char/mem.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/char/mem.c +0 -0
index 84ca98ed1dad..32d821ba9e4d 100644
@@ -772,7 +772,7 @@ static int __init chr_dev_init(void)
                      continue;
 
Quoted 6 lines

drivers/char/mem.c static int __init chr_dev_init(void) · line 775

Quoted 3 lines
drivers/char/mem.c +1 -1
              device_create(mem_class, NULL, MKDEV(MEM_MAJOR, minor),
-                           NULL, devlist[minor].name);
+                           NULL, "%s", devlist[minor].name);

Please explain how this static string can ever be user controlled.

All someone would need to do is accidentally insert an errant '%' in one of the strings for this function call to perform unexpected actions---at the very least reading memory that's not allocated and may contain garbage, thereby decreasing performance and possibly overrunning some buffer. Perhaps in this specific scenario it's unlikely, but "device_create()" is used in a lot more places than here. This patch is a general code cleanup.

-bw

#15 lore raw

Re: [PATCH 08/12] cdrom: use correct format characters

In reply to Phillip Potter
Quoted 45 lines · Phillip Potter
Quoted 30 lines · Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

drivers/cdrom/cdrom.c:3454:48: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

        ret = scnprintf(info + *pos, max_size - *pos, header);
                                                      ^~~~~~

Use a string literal for the format string.

---
 drivers/cdrom/cdrom.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/cdrom/cdrom.c +1 -2
index 416f723a2dbb..52b40120c76e 100644
@@ -3451,7 +3451,7 @@ static int cdrom_print_info(const char *header, int val, char *info,
      struct cdrom_device_info *cdi;
      int ret;
 
-     ret = scnprintf(info + *pos, max_size - *pos, header);
+     ret = scnprintf(info + *pos, max_size - *pos, "%s", header);
      if (!ret)
              return 1;
 
--
Quoted 2 lines

2.36.1.255.ge46751e96f-goog

Hi Bill,

Thank you for the patch, much appreciated.

Looking at this though, all callers of cdrom_print_info() provide 'header' as a string literal defined within the driver, when making the call. Therefore, I'm not convinced this change is necessary for cdrom.c - that said, in this particular use case I don't think it would hurt either.

Quoted 4 lines

I've followed the other responses on parts of this series, so I understand that a different solution is potentially in the works. Thought I'd respond anyway though out of courtesy.

Thanks, Phillip.

I pointed out in a separate response that this specific warning is disabled by default, but when I ran into while hacking stuff there weren't a lot of places where the warning popped up (at least for x86 builds) and thought it would be a nice cleanup. I understand if you don't think this patch is necessary for your code. There are some places where visual inspection of the code is "good enough" to ensure that nothing untoward will happen (Greg pointed out a similar thing in an mm/ file).

Cheers!
-bw

#20 lore raw

Re: [PATCH 12/12] netfilter: conntrack: use correct format characters

In reply to Bill Wendling
Quoted 9 lines · Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

net/netfilter/nf_conntrack_helper.c:168:18: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

                request_module(mod_name);
                               ^~~~~~~~

Use a string literal for the format string.

Should I handle this through the netfilter tree? Or you prefer a different path?

Thanks

maintainer Applied
#21 lore raw

Re: [PATCH 12/12] netfilter: conntrack: use correct format characters

In reply to Bill Wendling
Quoted 9 lines · Bill Wendling

From: Bill Wendling <[email protected]>

When compiling with -Wformat, clang emits the following warnings:

net/netfilter/nf_conntrack_helper.c:168:18: error: format string is not a string literal (potentially insecure) [-Werror,-Wformat-security]

                request_module(mod_name);
                               ^~~~~~~~

Use a string literal for the format string.

Applied this patch to nf-next, thanks