Skip to contentSkip to conversation

kernfs: Convert from strlcpy() to strscpy()

PATCH In review Kees Cook 2 participants 7 messages Dec 12, 2023 – Dec 22, 2023

Subsystems (4)
lore
Share & tools
lore

Kees Cook

author
#0 lore raw

v3 cover kernfs: Convert from strlcpy() to strscpy()

Hi,

One of the last users of strlcpy() is kernfs, which has some complex calling hierarchies that needed to be carefully examined. This series refactors the strlcpy() calls into strscpy() calls, and bubbles up all changes in return value checking for callers. Future work in kernfs and sysfs will see the replacement of open-coded string handling with the seq_buf API, but we need to do one thing at a time.

Thanks!

-kees

v3: don't need to account for scnprintf() returning negative (christophe.jaillet)
v2: https://lore.kernel.org/all/[email protected]/
v1: https://lore.kernel.org/linux-hardening/[email protected]/

Kees Cook (3):
kernfs: Convert kernfs_walk_ns() from strlcpy() to strscpy()
kernfs: Convert kernfs_name_locked() from strlcpy() to strscpy()
kernfs: Convert kernfs_path_from_node_locked() from strlcpy() to
strscpy()

 fs/kernfs/dir.c           | 50 +++++++++++++++++++--------------------
 kernel/cgroup/cgroup-v1.c |  2 +-
 kernel/cgroup/cgroup.c    |  4 ++--
 kernel/cgroup/cpuset.c    |  2 +-
 4 files changed, 29 insertions(+), 29 deletions(-)
Signature · 1 line
2.34.1
author
#1 lore raw

PATCH v3 1/3 kernfs: Convert kernfs_walk_ns() from strlcpy() to strscpy()

strlcpy() reads the entire source buffer first. This read may exceed the destination size limit. This is both inefficient and can lead to linear read overflows if a source string is not NUL-terminated[1]. Additionally, it returns the size of the source string, not the resulting size of the destination string. In an effort to remove strlcpy() completely[2], replace strlcpy() here with strscpy().

---
 fs/kernfs/dir.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
fs/kernfs/dir.c +3 -3
index 8b2bd65d70e7..37353901ede1 100644
@@ -850,16 +850,16 @@ static struct kernfs_node *kernfs_walk_ns(struct kernfs_node *parent,
 					  const unsigned char *path,
 					  const void *ns)
 {
-	size_t len;
+	ssize_t len;
 	char *p, *name;
 
 	lockdep_assert_held_read(&kernfs_root(parent)->kernfs_rwsem);
 
 	spin_lock_irq(&kernfs_pr_cont_lock);
 
-	len = strlcpy(kernfs_pr_cont_buf, path, sizeof(kernfs_pr_cont_buf));
+	len = strscpy(kernfs_pr_cont_buf, path, sizeof(kernfs_pr_cont_buf));
 
-	if (len >= sizeof(kernfs_pr_cont_buf)) {
+	if (len < 0) {
 		spin_unlock_irq(&kernfs_pr_cont_lock);
 		return NULL;
 	}
Signature · 1 line
2.34.1
maintainer Acked-by on v3 1/3
#2 lore raw

Re: [PATCH v3 1/3] kernfs: Convert kernfs_walk_ns() from strlcpy() to strscpy()

Quoted 7 lines · Kees Cook

strlcpy() reads the entire source buffer first. This read may exceed the destination size limit. This is both inefficient and can lead to linear read overflows if a source string is not NUL-terminated[1]. Additionally, it returns the size of the source string, not the resulting size of the destination string. In an effort to remove strlcpy() completely[2], replace strlcpy() here with strscpy().

Quoted 7 lines

Thanks.

Signature · 1 line
tejun
author
#3 lore raw

PATCH v3 2/3 kernfs: Convert kernfs_name_locked() from strlcpy() to strscpy()

In reply to Kees Cook

strlcpy() reads the entire source buffer first. This read may exceed the destination size limit. This is both inefficient and can lead to linear read overflows if a source string is not NUL-terminated[1]. Additionally, it returns the size of the source string, not the resulting size of the destination string. In an effort to remove strlcpy() completely[2], replace strlcpy() here with strscpy().

Nothing actually checks the return value coming from kernfs_name_locked(), so this has no impact on error paths. The caller hierarchy is:

kernfs_name_locked()
kernfs_name()
pr_cont_kernfs_name()
return value ignored
cgroup_name()
current_css_set_cg_links_read()
return value ignored
print_page_owner_memcg()
return value ignored

---
 fs/kernfs/dir.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
fs/kernfs/dir.c +5 -5
index 37353901ede1..8c0e5442597e 100644
@@ -54,9 +54,9 @@ static bool kernfs_lockdep(struct kernfs_node *kn)
 static int kernfs_name_locked(struct kernfs_node *kn, char *buf, size_t buflen)
 {
 	if (!kn)
-		return strlcpy(buf, "(null)", buflen);
+		return strscpy(buf, "(null)", buflen);
 
-	return strlcpy(buf, kn->parent ? kn->name : "/", buflen);
+	return strscpy(buf, kn->parent ? kn->name : "/", buflen);
 }
 
 /* kernfs_node_depth - compute depth from @from to @to */
@@ -182,12 +182,12 @@ static int kernfs_path_from_node_locked(struct kernfs_node *kn_to,
  * @buflen: size of @buf
  *
  * Copies the name of @kn into @buf of @buflen bytes.  The behavior is
- * similar to strlcpy().
+ * similar to strscpy().
  *
  * Fills buffer with "(null)" if @kn is %NULL.
  *
- * Return: the length of @kn's name and if @buf isn't long enough,
- * it's filled up to @buflen-1 and nul terminated.
+ * Return: the resulting length of @buf. If @buf isn't long enough,
+ * it's filled up to @buflen-1 and nul terminated, and returns -E2BIG.
  *
  * This function can be called from any context.
  */
Signature · 1 line
2.34.1
maintainer Acked-by on v3 2/3
#4 lore raw

Re: [PATCH v3 2/3] kernfs: Convert kernfs_name_locked() from strlcpy() to strscpy()

Quoted 20 lines · Kees Cook

strlcpy() reads the entire source buffer first. This read may exceed the destination size limit. This is both inefficient and can lead to linear read overflows if a source string is not NUL-terminated[1]. Additionally, it returns the size of the source string, not the resulting size of the destination string. In an effort to remove strlcpy() completely[2], replace strlcpy() here with strscpy().

Nothing actually checks the return value coming from kernfs_name_locked(), so this has no impact on error paths. The caller hierarchy is:

kernfs_name_locked()
kernfs_name()
pr_cont_kernfs_name()
return value ignored
cgroup_name()
current_css_set_cg_links_read()
return value ignored
print_page_owner_memcg()
return value ignored

Quoted 7 lines

Thanks.

Signature · 1 line
tejun
author
#5 lore raw

PATCH v3 3/3 kernfs: Convert kernfs_path_from_node_locked() from strlcpy() to strscpy()

In reply to Kees Cook

One of the last remaining users of strlcpy() in the kernel is kernfs_path_from_node_locked(), which passes back the problematic "length we _would_ have copied" return value to indicate truncation. Convert the chain of all callers to use the negative return value (some of which already doing this explicitly). All callers were already also checking for negative return values, so the risk to missed checks looks very low.

In this analysis, it was found that cgroup1_release_agent() actually didn't handle the "too large" condition, so this is technically also a bug fix. :)

Here's the chain of callers, and resolution identifying each one as now handling the correct return value:

kernfs_path_from_node_locked()
kernfs_path_from_node()
pr_cont_kernfs_path()
returns void
kernfs_path()
sysfs_warn_dup()
return value ignored
cgroup_path()
blkg_path()
bfq_bic_update_cgroup()
return value ignored
TRACE_IOCG_PATH()
return value ignored
TRACE_CGROUP_PATH()
return value ignored
perf_event_cgroup()
return value ignored
task_group_path()
return value ignored
damon_sysfs_memcg_path_eq()
return value ignored
get_mm_memcg_path()
return value ignored
lru_gen_seq_show()
return value ignored
cgroup_path_from_kernfs_id()
return value ignored
cgroup_show_path()
already converted "too large" error to negative value
cgroup_path_ns_locked()
cgroup_path_ns()
bpf_iter_cgroup_show_fdinfo()
return value ignored
cgroup1_release_agent()
wasn't checking "too large" error
proc_cgroup_show()
already converted "too large" to negative value

---
 fs/kernfs/dir.c           | 34 +++++++++++++++++-----------------
 kernel/cgroup/cgroup-v1.c |  2 +-
 kernel/cgroup/cgroup.c    |  4 ++--
 kernel/cgroup/cpuset.c    |  2 +-
 4 files changed, 21 insertions(+), 21 deletions(-)
fs/kernfs/dir.c +17 -17
index 8c0e5442597e..8ec73f6cf6ec 100644
@@ -127,7 +127,7 @@ static struct kernfs_node *kernfs_common_ancestor(struct kernfs_node *a,
  *
  * [3] when @kn_to is %NULL result will be "(null)"
  *
- * Return: the length of the full path.  If the full length is equal to or
+ * Return: the length of the constructed path.  If the path would have been
  * greater than @buflen, @buf contains the truncated path with the trailing
  * '\0'.  On error, -errno is returned.
  */
@@ -138,16 +138,17 @@ static int kernfs_path_from_node_locked(struct kernfs_node *kn_to,
 	struct kernfs_node *kn, *common;
 	const char parent_str[] = "/..";
 	size_t depth_from, depth_to, len = 0;
+	ssize_t copied;
 	int i, j;
 
 	if (!kn_to)
-		return strlcpy(buf, "(null)", buflen);
+		return strscpy(buf, "(null)", buflen);
 
 	if (!kn_from)
 		kn_from = kernfs_root(kn_to)->kn;
 
 	if (kn_from == kn_to)
-		return strlcpy(buf, "/", buflen);
+		return strscpy(buf, "/", buflen);
 
 	common = kernfs_common_ancestor(kn_from, kn_to);
 	if (WARN_ON(!common))
@@ -158,18 +159,19 @@ static int kernfs_path_from_node_locked(struct kernfs_node *kn_to,
 
 	buf[0] = '\0';
 
-	for (i = 0; i < depth_from; i++)
-		len += strlcpy(buf + len, parent_str,
-			       len < buflen ? buflen - len : 0);
+	for (i = 0; i < depth_from; i++) {
+		copied = strscpy(buf + len, parent_str, buflen - len);
+		if (copied < 0)
+			return copied;
+		len += copied;
+	}
 
 	/* Calculate how many bytes we need for the rest */
 	for (i = depth_to - 1; i >= 0; i--) {
 		for (kn = kn_to, j = 0; j < i; j++)
 			kn = kn->parent;
-		len += strlcpy(buf + len, "/",
-			       len < buflen ? buflen - len : 0);
-		len += strlcpy(buf + len, kn->name,
-			       len < buflen ? buflen - len : 0);
+
+		len += scnprintf(buf + len, buflen - len, "/%s", kn->name);
 	}
 
 	return len;
@@ -214,7 +216,7 @@ int kernfs_name(struct kernfs_node *kn, char *buf, size_t buflen)
  * path (which includes '..'s) as needed to reach from @from to @to is
  * returned.
  *
- * Return: the length of the full path.  If the full length is equal to or
+ * Return: the length of the constructed path.  If the path would have been
  * greater than @buflen, @buf contains the truncated path with the trailing
  * '\0'.  On error, -errno is returned.
  */
@@ -265,12 +267,10 @@ void pr_cont_kernfs_path(struct kernfs_node *kn)
 	sz = kernfs_path_from_node(kn, NULL, kernfs_pr_cont_buf,
 				   sizeof(kernfs_pr_cont_buf));
 	if (sz < 0) {
-		pr_cont("(error)");
-		goto out;
-	}
-
-	if (sz >= sizeof(kernfs_pr_cont_buf)) {
-		pr_cont("(name too long)");
+		if (sz == -E2BIG)
+			pr_cont("(name too long)");
+		else
+			pr_cont("(error)");
 		goto out;
 	}
 
kernel/cgroup/cgroup-v1.c +1 -1
index 76db6c67e39a..9cb00ebe9ac6 100644
@@ -802,7 +802,7 @@ void cgroup1_release_agent(struct work_struct *work)
 		goto out_free;
 
 	ret = cgroup_path_ns(cgrp, pathbuf, PATH_MAX, &init_cgroup_ns);
-	if (ret < 0 || ret >= PATH_MAX)
+	if (ret < 0)
 		goto out_free;
 
 	argv[0] = agentbuf;
kernel/cgroup/cgroup.c +2 -2
index 4b9ff41ca603..8d2674c6aaef 100644
@@ -1893,7 +1893,7 @@ int cgroup_show_path(struct seq_file *sf, struct kernfs_node *kf_node,
 	len = kernfs_path_from_node(kf_node, ns_cgroup->kn, buf, PATH_MAX);
 	spin_unlock_irq(&css_set_lock);
 
-	if (len >= PATH_MAX)
+	if (len == -E2BIG)
 		len = -ERANGE;
 	else if (len > 0) {
 		seq_escape(sf, buf, " \t\n\\");
@@ -6301,7 +6301,7 @@ int proc_cgroup_show(struct seq_file *m, struct pid_namespace *ns,
 		if (cgroup_on_dfl(cgrp) || !(tsk->flags & PF_EXITING)) {
 			retval = cgroup_path_ns_locked(cgrp, buf, PATH_MAX,
 						current->nsproxy->cgroup_ns);
-			if (retval >= PATH_MAX)
+			if (retval == -E2BIG)
 				retval = -ENAMETOOLONG;
 			if (retval < 0)
 				goto out_unlock;
kernel/cgroup/cpuset.c +1 -1
index 615daaf87f1f..fb29158ae825 100644
@@ -4941,7 +4941,7 @@ int proc_cpuset_show(struct seq_file *m, struct pid_namespace *ns,
 	retval = cgroup_path_ns(css->cgroup, buf, PATH_MAX,
 				current->nsproxy->cgroup_ns);
 	css_put(css);
-	if (retval >= PATH_MAX)
+	if (retval == -E2BIG)
 		retval = -ENAMETOOLONG;
 	if (retval < 0)
 		goto out_free;
Signature · 1 line
2.34.1
maintainer
#6 lore raw

Re: [PATCH v3 3/3] kernfs: Convert kernfs_path_from_node_locked() from strlcpy() to strscpy()

Hello,

On Tue, Dec 12, 2023 at 01:17:40PM -0800, Kees Cook wrote:
...

Quoted 9 lines
fs/kernfs/dir.c +1 -1
@@ -127,7 +127,7 @@ static struct kernfs_node *kernfs_common_ancestor(struct kernfs_node *a,
  *
  * [3] when @kn_to is %NULL result will be "(null)"
  *
- * Return: the length of the full path.  If the full length is equal to or
+ * Return: the length of the constructed path.  If the path would have been
  * greater than @buflen, @buf contains the truncated path with the trailing
  * '\0'.  On error, -errno is returned.
  */

...

Quoted 1 line

/* Calculate how many bytes we need for the rest */

We probably should drop this comment.

fs/kernfs/dir.c static int kernfs_path_from_node_locked(struct kernfs_node *kn_to, · line 174

Quoted 9 lines
fs/kernfs/dir.c +2 -4
 	for (i = depth_to - 1; i >= 0; i--) {
 		for (kn = kn_to, j = 0; j < i; j++)
 			kn = kn->parent;
-		len += strlcpy(buf + len, "/",
-			       len < buflen ? buflen - len : 0);
-		len += strlcpy(buf + len, kn->name,
-			       len < buflen ? buflen - len : 0);
+
+		len += scnprintf(buf + len, buflen - len, "/%s", kn->name);

scnprintf doesn't return -E2BIG on overflow, right? It just returns the truncated length, so the overflow behavior would be different depending on where this function overflows, right? Not a huge problem but it may be better to keep calling strscpy to keep things consistent?

kernel/cgroup/cgroup.c int cgroup_show_path(struct seq_file *sf, struct kernfs_node *kf_node, · line 1897

Quoted 9 lines
kernel/cgroup/cgroup.c +1 -1
@@ -1893,7 +1893,7 @@ int cgroup_show_path(struct seq_file *sf, struct kernfs_node *kf_node,
 	len = kernfs_path_from_node(kf_node, ns_cgroup->kn, buf, PATH_MAX);
 	spin_unlock_irq(&css_set_lock);
 
-	if (len >= PATH_MAX)
+	if (len == -E2BIG)
 		len = -ERANGE;

I'd just pass up -E2BIG.

kernel/cgroup/cgroup.c int proc_cgroup_show(struct seq_file *m, struct pid_namespace *ns, · line 6305

Quoted 9 lines
kernel/cgroup/cgroup.c +1 -1
 	else if (len > 0) {
 		seq_escape(sf, buf, " \t\n\\");
@@ -6301,7 +6301,7 @@ int proc_cgroup_show(struct seq_file *m, struct pid_namespace *ns,
 		if (cgroup_on_dfl(cgrp) || !(tsk->flags & PF_EXITING)) {
 			retval = cgroup_path_ns_locked(cgrp, buf, PATH_MAX,
 						current->nsproxy->cgroup_ns);
-			if (retval >= PATH_MAX)
+			if (retval == -E2BIG)
 				retval = -ENAMETOOLONG;

Ditto.

kernel/cgroup/cpuset.c int proc_cpuset_show(struct seq_file *m, struct pid_namespace *ns, · line 4944

Quoted 10 lines
kernel/cgroup/cpuset.c +1 -1
index 615daaf87f1f..fb29158ae825 100644
@@ -4941,7 +4941,7 @@ int proc_cpuset_show(struct seq_file *m, struct pid_namespace *ns,
 	retval = cgroup_path_ns(css->cgroup, buf, PATH_MAX,
 				current->nsproxy->cgroup_ns);
 	css_put(css);
-	if (retval >= PATH_MAX)
+	if (retval == -E2BIG)

Ditto.

Thanks.

Signature · 1 line
tejun