Skip to contentSkip to conversation

s390/qeth: validate user buffer length in SNMP and ARP query ioctls

PATCH Applied Hidayath Khan 4 participants 4 messages Jul 30 – Aug 4

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 s390/qeth: validate user buffer length in SNMP and ARP query ioctls

qeth_snmp_command() and qeth_l3_arp_query() allocate a buffer sized by a user-supplied length (udata_len) without checking a lower bound, then set udata_offset to a fixed non-zero value and pass both to a reply callback. The callback bounds-checks the copy with

if ((udata_len - udata_offset) < len)

Both fields are u32, so a udata_len smaller than udata_offset makes the subtraction wrap and the check pass, and the following memcpy() writes past the allocation. A udata_len of 0 also yields ZERO_SIZE_PTR from kzalloc(), which the existing NULL check does not catch.

Reject buffers smaller than udata_offset before allocating, so the callback subtraction can no longer underflow.

Fixes 4a71df50047f ("qeth: new qeth device driver")
---
 drivers/s390/net/qeth_core_main.c | 3 +++
 drivers/s390/net/qeth_l3_main.c   | 5 +++++
 2 files changed, 8 insertions(+)
drivers/s390/net/qeth_core_main.c +3 -0
index f18eed9df3c7..c3257b213360 100644
@@ -4710,6 +4710,9 @@ static int qeth_snmp_command(struct qeth_card *card, char __user *udata)
 	if (req_len > QETH_BUFSIZE)
 		return -EINVAL;
 
+	if (qinfo.udata_len < sizeof(struct qeth_snmp_ureq_hdr))
+		return -EINVAL;
+
 	iob = qeth_get_adapter_cmd(card, IPA_SETADP_SET_SNMP_CONTROL, req_len);
 	if (!iob)
 		return -ENOMEM;
drivers/s390/net/qeth_l3_main.c +5 -0
index 1542bfc9f561..f1ac9950dcb4 100644
@@ -1415,6 +1415,11 @@ static int qeth_l3_arp_query(struct qeth_card *card, char __user *udata)
 		rc = -EFAULT;
 		goto out;
 	}
+
+	if (qinfo.udata_len < QETH_QARP_ENTRIES_OFFSET) {
+		rc = -EINVAL;
+		goto out;
+	}
 	qinfo.udata = kzalloc(qinfo.udata_len, GFP_KERNEL);
 	if (!qinfo.udata) {
 		rc = -ENOMEM;

base-commit: 58c1c294d685baf94801839334df20501711eb8e

Signature · 1 line
2.52.0
Reviewed-by on v1 1/1
#1 lore raw
Quoted 15 lines · Hidayath Khan

qeth_snmp_command() and qeth_l3_arp_query() allocate a buffer sized by a user-supplied length (udata_len) without checking a lower bound, then set udata_offset to a fixed non-zero value and pass both to a reply callback. The callback bounds-checks the copy with

if ((udata_len - udata_offset) < len)

Both fields are u32, so a udata_len smaller than udata_offset makes the subtraction wrap and the check pass, and the following memcpy() writes past the allocation. A udata_len of 0 also yields ZERO_SIZE_PTR from kzalloc(), which the existing NULL check does not catch.

Reject buffers smaller than udata_offset before allocating, so the callback subtraction can no longer underflow.

Quoted 8 lines
Fixes 4a71df50047f ("qeth: new qeth device driver")
---
 drivers/s390/net/qeth_core_main.c | 3 +++
 drivers/s390/net/qeth_l3_main.c   | 5 +++++
 2 files changed, 8 insertions(+)
#2 lore raw
In reply to Hidayath Khan
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] Unvalidated length fields from hardware responses lead to kernel heap out-of-bounds reads and memory disclosure. --

--- Patch [1]: [PATCH net] s390/qeth: validate user buffer length in SNMP and ARP query ioctls --- commit f9f8c7e047077e9dbc90a35127afc75e1690d27f
Author: Hidayath Khan <[email protected]>

s390/qeth: validate user buffer length in SNMP and ARP query ioctls

This commit validates the user-supplied length (udata_len) against a lower bound (udata_offset) before allocating the buffer in SNMP and ARP query ioctls. This prevents underflow during callback bounds checking and fixes a ZERO_SIZE_PTR allocation when udata_len is 0.

drivers/s390/net/qeth_core_main.c static int qeth_snmp_command(struct qeth_card *card, char __user *udata) · line 4715

Quoted 11 lines
drivers/s390/net/qeth_core_main.c +3 -0
index f18eed9df3c7a..c3257b213360c 100644
@@ -4710,6 +4710,9 @@ static int qeth_snmp_command(struct qeth_card *card, char __user *udata)
 	if (req_len > QETH_BUFSIZE)
 		return -EINVAL;
 
+	if (qinfo.udata_len < sizeof(struct qeth_snmp_ureq_hdr))
+		return -EINVAL;
+

[Severity: High]
This is a pre-existing issue, but does this expose the kernel to a heap out-of-bounds read if the hardware returns an invalid length?

In qeth_snmp_command_cb(), the response length is calculated from the firmware's header:

	data_len = adp_cmd->hdr.cmdlength - sizeof(adp_cmd->hdr);
    ...
	if ((qinfo->udata_len - qinfo->udata_offset) < data_len) {
		return -ENOSPC;
	}
    ...
	memcpy(qinfo->udata + qinfo->udata_offset, snmp_data, data_len);

The check ensures data_len fits within the userspace buffer (qinfo->udata_len), but doesn't verify if data_len exceeds the actual hardware response buffer length. If a compromised or malicious firmware returns a large cmdlength (and userspace provided a large udata_len), could this memcpy() read past the end of the hardware response buffer and leak adjacent kernel memory back to userspace?

Quoted 8 lines
drivers/s390/net/qeth_core_main.c +0 -0
 	iob = qeth_get_adapter_cmd(card, IPA_SETADP_SET_SNMP_CONTROL, req_len);
 	if (!iob)
 		return -ENOMEM;
drivers/s390/net/qeth_l3_main.c +0 -0
index 1542bfc9f561b..f1ac9950dcb4e 100644

drivers/s390/net/qeth_l3_main.c static int qeth_l3_arp_query(struct qeth_card *card, char __user *udata) · line 1422

Quoted 8 lines
drivers/s390/net/qeth_l3_main.c +5 -0
@@ -1415,6 +1415,11 @@ static int qeth_l3_arp_query(struct qeth_card *card, char __user *udata)
 		rc = -EFAULT;
 		goto out;
 	}
+
+	if (qinfo.udata_len < QETH_QARP_ENTRIES_OFFSET) {
+		rc = -EINVAL;
+		goto out;
+	}

[Severity: High]
This isn't a bug introduced by this patch, but is there a similar out-of-bounds read issue when processing the ARP entries?

In qeth_l3_arp_query_cb(), the code trusts the firmware-provided qdata->no_entries without checking if the entries actually fit within the received hardware response buffer:

	for (e = 0; e < qdata->no_entries; ++e) {
        ...
		esize = get_arp_entry_size(card, qdata, etype, do_strip_entries);
        ...
		if ((qinfo->udata_len - qinfo->udata_offset) < esize) {
			return -ENOSPC;
		}

		memcpy(qinfo->udata + qinfo->udata_offset,
			&qdata->data + entrybytes_done + stripped_bytes,
			esize);
		entrybytes_done += esize + stripped_bytes;

Like the SNMP issue above, the bounds check only verifies that there is enough space in the destination user buffer (qinfo->udata_len), but does not check if we are reading past the bounds of the hardware response buffer. Could a malicious firmware response with a large no_entries cause a kernel heap out-of-bounds read?

Quoted 3 lines
drivers/s390/net/qeth_l3_main.c +0 -0
 	qinfo.udata = kzalloc(qinfo.udata_len, GFP_KERNEL);
 	if (!qinfo.udata) {
 		rc = -ENOMEM;
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
#3 lore raw
In reply to Hidayath Khan
This patch was applied to netdev/net.git (main) by Jakub Kicinski <[email protected]>:

Hello:

This patch was applied to netdev/net.git (main)
by Jakub Kicinski <[email protected]>:

Quoted 8 lines · On Thu, 30 Jul 2026 16:22:16 +0200 you wrote:

qeth_snmp_command() and qeth_l3_arp_query() allocate a buffer sized by a user-supplied length (udata_len) without checking a lower bound, then set udata_offset to a fixed non-zero value and pass both to a reply callback. The callback bounds-checks the copy with

if ((udata_len - udata_offset) < len)

[...]

Here is the summary with links:
- [net] s390/qeth: validate user buffer length in SNMP and ARP query ioctls

    https://git.kernel.org/netdev/net/c/d141f087b1af

You are awesome, thank you!

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