Skip to contentSkip to conversation

module: fix [e_shstrndx].sh_size=0 OOB access

PATCH Superseded Alexey Dobriyan 2 participants 5 messages May 3, 2022 – May 4, 2022

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 module: fix [e_shstrndx].sh_size=0 OOB access

It is trivial to craft a module to trigger OOB access in this line:

	if (info->secstrings[strhdr->sh_size - 1] != '\0') {
BUG: unable to handle page fault for address: ffffc90000aa0fff
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 100000067 P4D 100000067 PUD 100066067 PMD 10436f067 PTE 0
Oops: 0000 [#1] PREEMPT SMP PTI
CPU: 7 PID: 1215 Comm: insmod Not tainted 5.18.0-rc5-00007-g9bf578647087-dirty #10
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-4.fc34 04/01/2014
RIP: 0010:load_module+0x19b/0x2391

---

 kernel/module.c |    4 ++++
 1 file changed, 4 insertions(+)
kernel/module.c +4 -0
@@ -3033,6 +3033,10 @@ static int elf_validity_check(struct load_info *info)
 	 * strings in the section safe.
 	 */
 	info->secstrings = (void *)info->hdr + strhdr->sh_offset;
+	if (strhdr->sh_size == 0) {
+		pr_err("empty section name table\n");
+		goto no_exec;
+	}
 	if (info->secstrings[strhdr->sh_size - 1] != '\0') {
 		pr_err("ELF Spec violation: section name table isn't null terminated\n");
 		goto no_exec;
#1 lore raw
Quoted 13 lines · Alexey Dobriyan

It is trivial to craft a module to trigger OOB access in this line:

	if (info->secstrings[strhdr->sh_size - 1] != '\0') {
BUG: unable to handle page fault for address: ffffc90000aa0fff
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 100000067 P4D 100000067 PUD 100066067 PMD 10436f067 PTE 0
Oops: 0000 [#1] PREEMPT SMP PTI
CPU: 7 PID: 1215 Comm: insmod Not tainted 5.18.0-rc5-00007-g9bf578647087-dirty #10
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-4.fc34 04/01/2014
RIP: 0010:load_module+0x19b/0x2391
Quoted 1 line

Thanks! Can you resend and Cc stable? This seems like an issue present for a long time, can you identify the commit which introduced the issue?

Luis

#2 lore raw
Quoted 14 lines · Luis Chamberlain
Quoted 13 lines · Alexey Dobriyan

It is trivial to craft a module to trigger OOB access in this line:

	if (info->secstrings[strhdr->sh_size - 1] != '\0') {
BUG: unable to handle page fault for address: ffffc90000aa0fff
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 100000067 P4D 100000067 PUD 100066067 PMD 10436f067 PTE 0
Oops: 0000 [#1] PREEMPT SMP PTI
CPU: 7 PID: 1215 Comm: insmod Not tainted 5.18.0-rc5-00007-g9bf578647087-dirty #10
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-4.fc34 04/01/2014
RIP: 0010:load_module+0x19b/0x2391
Quoted 3 lines
Quoted 1 line

Thanks! Can you resend and Cc stable?

Stable will pick it up simply because patch does apply I assume?

Quoted 2 lines

This seems like an issue present for
a long time, can you identify the commit which introduced the issue?

I thought it was ancient code but no.

#3 lore raw

PATCH v2 1/1 module: fix [e_shstrndx].sh_size=0 OOB access

In reply to Luis Chamberlain

It is trivial to craft a module to trigger OOB access in this line:

	if (info->secstrings[strhdr->sh_size - 1] != '\0') {
BUG: unable to handle page fault for address: ffffc90000aa0fff
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 100000067 P4D 100000067 PUD 100066067 PMD 10436f067 PTE 0
Oops: 0000 [#1] PREEMPT SMP PTI
CPU: 7 PID: 1215 Comm: insmod Not tainted 5.18.0-rc5-00007-g9bf578647087-dirty #10
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-4.fc34 04/01/2014
RIP: 0010:load_module+0x19b/0x2391
Fixes ec2a29593c83 ("module: harden ELF info handling")

---

 kernel/module.c |    4 ++++
 1 file changed, 4 insertions(+)
kernel/module.c +4 -0
@@ -3033,6 +3033,10 @@ static int elf_validity_check(struct load_info *info)
 	 * strings in the section safe.
 	 */
 	info->secstrings = (void *)info->hdr + strhdr->sh_offset;
+	if (strhdr->sh_size == 0) {
+		pr_err("empty section name table\n");
+		goto no_exec;
+	}
 	if (info->secstrings[strhdr->sh_size - 1] != '\0') {
 		pr_err("ELF Spec violation: section name table isn't null terminated\n");
 		goto no_exec;

---

 kernel/module.c |    4 ++++
 1 file changed, 4 insertions(+)
kernel/module.c +4 -0
@@ -3033,6 +3033,10 @@ static int elf_validity_check(struct load_info *info)
 	 * strings in the section safe.
 	 */
 	info->secstrings = (void *)info->hdr + strhdr->sh_offset;
+	if (strhdr->sh_size == 0) {
+		pr_err("empty section name table\n");
+		goto no_exec;
+	}
 	if (info->secstrings[strhdr->sh_size - 1] != '\0') {
 		pr_err("ELF Spec violation: section name table isn't null terminated\n");
 		goto no_exec;
#4 lore raw
Quoted 13 lines · Alexey Dobriyan

It is trivial to craft a module to trigger OOB access in this line:

	if (info->secstrings[strhdr->sh_size - 1] != '\0') {
BUG: unable to handle page fault for address: ffffc90000aa0fff
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 100000067 P4D 100000067 PUD 100066067 PMD 10436f067 PTE 0
Oops: 0000 [#1] PREEMPT SMP PTI
CPU: 7 PID: 1215 Comm: insmod Not tainted 5.18.0-rc5-00007-g9bf578647087-dirty #10
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-4.fc34 04/01/2014
RIP: 0010:load_module+0x19b/0x2391
Quoted 2 lines
Fixes ec2a29593c83 ("module: harden ELF info handling")

Thanks! I rebased your patch onto modules-next [0] and applied onto modules-testing and pushed out there. If that doesn't break I'll then push to modules-next as well.

Luis