Skip to contentSkip to conversation

bpf: Add missing access_ok call to copy_user_syms

PATCH Applied Jiri Olsa 5 participants 5 messages Jun 16 – Jun 22

Subsystems (5)
lore
Share & tools
lore

Jiri Olsa

author reviewer
#0 lore raw

PATCH 1/1 bpf: Add missing access_ok call to copy_user_syms

As reported by sashiko we use __get_user without prior access_ok call on the user space pointer. Adding the missing call for the whole pointer array.

Plus removing the err check in the error path, because it's not needed and also we can return -ENOMEM directly from the first kvmalloc_array fail path.

Cc: [email protected]
[1] https://lore.kernel.org/bpf/[email protected]/

Fixes 0236fec57a15 ("bpf: Resolve symbols with ftrace_lookup_symbols for kprobe multi link")
---
 kernel/trace/bpf_trace.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
kernel/trace/bpf_trace.c +6 -5
index 82f8feea6931..75495a5c3507 100644
@@ -2376,9 +2376,12 @@ static int copy_user_syms(struct user_syms *us, unsigned long __user *usyms, u32
 	int err = -ENOMEM;
 	unsigned int i;
 
+	if (!access_ok(usyms, cnt * sizeof(*usyms)))
+		return -EFAULT;
+
 	syms = kvmalloc_array(cnt, sizeof(*syms), GFP_KERNEL);
 	if (!syms)
-		goto error;
+		return -ENOMEM;
 
 	buf = kvmalloc_array(cnt, KSYM_NAME_LEN, GFP_KERNEL);
 	if (!buf)
@@ -2403,10 +2406,8 @@ static int copy_user_syms(struct user_syms *us, unsigned long __user *usyms, u32
 	return 0;
 
 error:
-	if (err) {
-		kvfree(syms);
-		kvfree(buf);
-	}
+	kvfree(syms);
+	kvfree(buf);
 	return err;
 }
 
Signature · 1 line
2.54.0
#1 lore raw
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] Missing access_ok() checks for user pointers uoffsets, uref_ctr_offsets, and ucookies before using __get_user() in bpf_uprobe_multi_link_attach. --

commit cd20e04dceb618df71198c96a9e801e033dcb0a0
Author: Jiri Olsa <[email protected]>
bpf: Add missing access_ok call to copy_user_syms

The patch adds a missing access_ok() check for the user pointer array usyms in copy_user_syms() before accessing it via __get_user(). It also cleans up the error path by returning -ENOMEM directly upon memory allocation failure instead of using a goto.

Quoted 20 lines
kernel/trace/bpf_trace.c +4 -1
index a02bd258677ee..c274109d020f3 100644
@@ -2331,9 +2331,12 @@ static int copy_user_syms(struct user_syms *us, unsigned long __user *usyms, u32
 	int err = -ENOMEM;
 	unsigned int i;
 
+	if (!access_ok(usyms, cnt * sizeof(*usyms)))
+		return -EFAULT;
+
 	syms = kvmalloc_array(cnt, sizeof(*syms), GFP_KERNEL);
 	if (!syms)
-		goto error;
+		return -ENOMEM;
 
 	buf = kvmalloc_array(cnt, KSYM_NAME_LEN, GFP_KERNEL);
 	if (!buf)
@@ -2358,10 +2361,8 @@ static int copy_user_syms(struct user_syms *us, unsigned long __user *usyms, u32
 	return 0;

kernel/trace/bpf_trace.c static int copy_user_syms(struct user_syms *us, unsigned long __user *usyms, u32 · line 2367

Quoted 10 lines
kernel/trace/bpf_trace.c +2 -4
 
 error:
-	if (err) {
-		kvfree(syms);
-		kvfree(buf);
-	}
+	kvfree(syms);
+	kvfree(buf);
 	return err;
 }

[Severity: High]
This is a pre-existing issue, but while reviewing this patch I noticed a similar pattern in bpf_uprobe_multi_link_attach() within the same file.

kernel/trace/bpf_trace.c:bpf_uprobe_multi_link_attach() {
    ...
	for (i = 0; i < cnt; i++) {
		if (__get_user(uprobes[i].offset, uoffsets + i)) {
			err = -EFAULT;
			goto error_free;
		}
		if (uprobes[i].offset < 0) {
			err = -EINVAL;
			goto error_free;
		}
		if (uref_ctr_offsets && __get_user(uprobes[i].ref_ctr_offset, uref_ctr_offsets + i)) {
			err = -EFAULT;
			goto error_free;
		}
		if (ucookies && __get_user(uprobes[i].cookie, ucookies + i)) {
			err = -EFAULT;
			goto error_free;
		}
    ...
}

Are the user-provided array pointers uoffsets, uref_ctr_offsets, and ucookies missing overarching access_ok() validations before being accessed in this loop?

Since __get_user() explicitly bypasses the access_ok() validation performed by get_user(), reading these pointers without a prior check could potentially allow an attacker to pass arbitrary kernel memory addresses and trigger unauthorized reads or crashes.

Does this function need the same fix as copy_user_syms()?

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
#2 lore raw
Quoted 5 lines · On Tue, Jun 16, 2026 at 08:42:03AM +0000, [email protected] wrote:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] Missing access_ok() checks for user pointers uoffsets, uref_ctr_offsets, and ucookies before using __get_user() in bpf_uprobe_multi_link_attach. --

it's fixed in separate post in here:
https://lore.kernel.org/bpf/[email protected]/

jirka

Quoted 81 lines

commit cd20e04dceb618df71198c96a9e801e033dcb0a0
Author: Jiri Olsa <[email protected]>
bpf: Add missing access_ok call to copy_user_syms

The patch adds a missing access_ok() check for the user pointer array usyms in copy_user_syms() before accessing it via __get_user(). It also cleans up the error path by returning -ENOMEM directly upon memory allocation failure instead of using a goto.

Quoted 20 lines
kernel/trace/bpf_trace.c +4 -1
index a02bd258677ee..c274109d020f3 100644
@@ -2331,9 +2331,12 @@ static int copy_user_syms(struct user_syms *us, unsigned long __user *usyms, u32
 	int err = -ENOMEM;
 	unsigned int i;
 
+	if (!access_ok(usyms, cnt * sizeof(*usyms)))
+		return -EFAULT;
+
 	syms = kvmalloc_array(cnt, sizeof(*syms), GFP_KERNEL);
 	if (!syms)
-		goto error;
+		return -ENOMEM;
 
 	buf = kvmalloc_array(cnt, KSYM_NAME_LEN, GFP_KERNEL);
 	if (!buf)
@@ -2358,10 +2361,8 @@ static int copy_user_syms(struct user_syms *us, unsigned long __user *usyms, u32
 	return 0;

kernel/trace/bpf_trace.c static int copy_user_syms(struct user_syms *us, unsigned long __user *usyms, u32 · line 2367

Quoted 10 lines
kernel/trace/bpf_trace.c +2 -4
 
 error:
-	if (err) {
-		kvfree(syms);
-		kvfree(buf);
-	}
+	kvfree(syms);
+	kvfree(buf);
 	return err;
 }

[Severity: High]
This is a pre-existing issue, but while reviewing this patch I noticed a similar pattern in bpf_uprobe_multi_link_attach() within the same file.

kernel/trace/bpf_trace.c:bpf_uprobe_multi_link_attach() {
    ...
	for (i = 0; i < cnt; i++) {
		if (__get_user(uprobes[i].offset, uoffsets + i)) {
			err = -EFAULT;
			goto error_free;
		}
		if (uprobes[i].offset < 0) {
			err = -EINVAL;
			goto error_free;
		}
		if (uref_ctr_offsets && __get_user(uprobes[i].ref_ctr_offset, uref_ctr_offsets + i)) {
			err = -EFAULT;
			goto error_free;
		}
		if (ucookies && __get_user(uprobes[i].cookie, ucookies + i)) {
			err = -EFAULT;
			goto error_free;
		}
    ...
}

Are the user-provided array pointers uoffsets, uref_ctr_offsets, and ucookies missing overarching access_ok() validations before being accessed in this loop?

Since __get_user() explicitly bypasses the access_ok() validation performed by get_user(), reading these pointers without a prior check could potentially allow an attacker to pass arbitrary kernel memory addresses and trigger unauthorized reads or crashes.

Does this function need the same fix as copy_user_syms()?

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
reviewer Reviewed-by on v1 1/1
#3 lore raw
In reply to Jiri Olsa
Quoted 6 lines · Jiri Olsa

As reported by sashiko we use __get_user without prior access_ok call on the user space pointer. Adding the missing call for the whole pointer array.

Plus removing the err check in the error path, because it's not needed and also we can return -ENOMEM directly from the first kvmalloc_array fail path.

Quoted 6 lines

Cc: [email protected]
[1] https://lore.kernel.org/bpf/[email protected]/

Fixes 0236fec57a15 ("bpf: Resolve symbols with ftrace_lookup_symbols for kprobe multi link")
Quoted 35 lines
---
 kernel/trace/bpf_trace.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
kernel/trace/bpf_trace.c +6 -5
index 82f8feea6931..75495a5c3507 100644
@@ -2376,9 +2376,12 @@ static int copy_user_syms(struct user_syms *us, unsigned long __user *usyms, u32
 	int err = -ENOMEM;
 	unsigned int i;
 
+	if (!access_ok(usyms, cnt * sizeof(*usyms)))
+		return -EFAULT;
+
 	syms = kvmalloc_array(cnt, sizeof(*syms), GFP_KERNEL);
 	if (!syms)
-		goto error;
+		return -ENOMEM;
 
 	buf = kvmalloc_array(cnt, KSYM_NAME_LEN, GFP_KERNEL);
 	if (!buf)
@@ -2403,10 +2406,8 @@ static int copy_user_syms(struct user_syms *us, unsigned long __user *usyms, u32
 	return 0;
 
 error:
-	if (err) {
-		kvfree(syms);
-		kvfree(buf);
-	}
+	kvfree(syms);
+	kvfree(buf);
 	return err;
 }
 
#4 lore raw
In reply to Jiri Olsa
This patch was applied to bpf/bpf.git (master) by Alexei Starovoitov <[email protected]>:

Hello:

This patch was applied to bpf/bpf.git (master)
by Alexei Starovoitov <[email protected]>:

Quoted 6 lines · On Tue, 16 Jun 2026 10:30:56 +0200 you wrote:

As reported by sashiko we use __get_user without prior access_ok call on the user space pointer. Adding the missing call for the whole pointer array.

Plus removing the err check in the error path, because it's not needed and also we can return -ENOMEM directly from the first kvmalloc_array fail path.

Quoted 8 lines

Cc: [email protected]
[1] https://lore.kernel.org/bpf/[email protected]/

Fixes 0236fec57a15 ("bpf: Resolve symbols with ftrace_lookup_symbols for kprobe multi link")

[...]

Here is the summary with links:
- [bpf] bpf: Add missing access_ok call to copy_user_syms

    https://git.kernel.org/bpf/bpf/c/d5dc200c3a3f

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html