6.6.145-rc1 review

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Zijun Hu <[email protected]>

[ Upstream commit 6b8cbcf08de0db62254d1981f83db0f94681ccd9 ]

hu->serdev is NULL for hci_uart attached via non-serdev paths, but qca_dmp_hdr() unconditionally dereferences hu->serdev->dev.driver->name, causing a NULL pointer dereference.

Fix by guarding the dereference with a NULL check and falling back to "hci_ldisc_qca" for the non-serdev case.

---
 drivers/bluetooth/hci_qca.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index ac737ee143614c..2986b18655a150 100644
@@ -1008,7 +1008,7 @@ static void qca_dmp_hdr(struct hci_dev *hdev, struct sk_buff *skb)
 	skb_put_data(skb, buf, strlen(buf));
 
 	snprintf(buf, sizeof(buf), "Driver: %s\n",
-		hu->serdev->dev.driver->name);
+		 hu->serdev ? hu->serdev->dev.driver->name : "hci_ldisc_qca");
 	skb_put_data(skb, buf, strlen(buf));
 }
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Samuel Moelius <[email protected]>

[ Upstream commit c38fbcdc407925c7088f7e5f11c1fff73d2d35a2 ]

Read Local Codec Capabilities returns a sequence of capability elements. Each element starts with a one-byte length followed by that many payload bytes.

hci_read_codec_capabilities() checks that the skb contains the length byte, but then validates only caps->len against the remaining skb length. A malformed controller response with one remaining byte and caps->len set to one passes that check even though the element needs two bytes. The parser then records a two-byte capability and copies one byte beyond the advertised response payload into the codec list.

Validate the full element size, including the length byte, before adding it to the accumulated capability length. This preserves all well-formed capability elements and drops only truncated controller responses.

---
 net/bluetooth/hci_codec.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 3cc135bb1d30ca..5bc5003c387c66 100644
@@ -100,7 +100,7 @@ static void hci_read_codec_capabilities(struct hci_dev *hdev, __u8 transport,
 				caps = (void *)skb->data;
 				if (skb->len < sizeof(*caps))
 					goto error;
-				if (skb->len < caps->len)
+				if (skb->len < sizeof(caps->len) + caps->len)
 					goto error;
 				len += sizeof(caps->len) + caps->len;
 				skb_pull(skb,  sizeof(caps->len) + caps->len);

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Samuel Moelius <[email protected]>

[ Upstream commit 88c2404a3c59c3126453919388dbd5ed98ed01bd ]

The VHCI force_devcoredump debugfs hook accepts a small test record from userspace. It validates the requested terminal state only after registering, initializing and appending a Bluetooth devcoredump.

As a result, an invalid state returns -EINVAL but still leaves queued devcoredump work behind. With a non-zero timeout field, the rejected write can still emit a devcoredump after the timeout expires.

Reject unsupported states before allocating the skb or changing the HCI devcoredump state machine.

---
 drivers/bluetooth/hci_vhci.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

index 0935045051699f..c0588fe110a2d4 100644
@@ -337,7 +337,17 @@ static ssize_t force_devcd_write(struct file *file, const char __user *user_buf,
 	if (copy_from_user(&dump_data, user_buf, count))
 		return -EFAULT;
 
+	switch (dump_data.state) {
+	case HCI_DEVCOREDUMP_DONE:
+	case HCI_DEVCOREDUMP_ABORT:
+	case HCI_DEVCOREDUMP_TIMEOUT:
+		break;
+	default:
+		return -EINVAL;
+	}
+
 	data_size = count - offsetof(struct devcoredump_test_data, data);
+
 	skb = alloc_skb(data_size, GFP_ATOMIC);
 	if (!skb)
 		return -ENOMEM;

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Maxwell Doose <[email protected]>

[ Upstream commit 89009392c80da5da00876c8334ff20028e6e3eb6 ]

The current code uses debug prints conditionally compiled with #ifdef DEBUG. However, that code, when compiled, causes compiler errors due to incompatible formatters and undefined variables, notably:

fs/efs/file.c: In function `efs_get_block':
fs/efs/file.c:26:35: error: block' undeclared (first use in this function); did you mean iblock'?
26 | __func__, block, inode->i_blocks, inode->i_size);
| ^~~~~

and:

fs/efs/file.c: In function `efs_bmap':
./include/linux/kern_levels.h:5:25: error: format %ld' expects argument of type long int', but argument 4 has type blkcnt_t' {aka long long unsigned int'} [-Werror=format=]

   5 | #define KERN_SOH        "\001"          /* ASCII Start Of Header */
     |                         ^~~~~~

which also extends to the other formatters. As this part of the code has been dead for just about 14 years now, it has not been modernized to stay compatible with the most recent gcc compilers. Fix these issues by removing the debug prints.

---
 fs/efs/file.c | 21 +++------------------
 1 file changed, 3 insertions(+), 18 deletions(-)

index 9e641da6fab276..9153dfe79bbcc9 100644
@@ -18,16 +18,9 @@ int efs_get_block(struct inode *inode, sector_t iblock,
 
 	if (create)
 		return error;
-	if (iblock >= inode->i_blocks) {
-#ifdef DEBUG
-		/*
-		 * i have no idea why this happens as often as it does
-		 */
-		pr_warn("%s(): block %d >= %ld (filesize %ld)\n",
-			__func__, block, inode->i_blocks, inode->i_size);
-#endif
+	if (iblock >= inode->i_blocks)
 		return 0;
-	}
+
 	phys = efs_map_block(inode, iblock);
 	if (phys)
 		map_bh(bh_result, inode->i_sb, phys);

@@ -42,16 +35,8 @@ int efs_bmap(struct inode *inode, efs_block_t block) {
 	}
 
 	/* are we about to read past the end of a file ? */
-	if (!(block < inode->i_blocks)) {
-#ifdef DEBUG
-		/*
-		 * i have no idea why this happens as often as it does
-		 */
-		pr_warn("%s(): block %d >= %ld (filesize %ld)\n",
-			__func__, block, inode->i_blocks, inode->i_size);
-#endif
+	if (!(block < inode->i_blocks))
 		return 0;
-	}
 
 	return efs_map_block(inode, block);
 }

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Patrisious Haddad <[email protected]>

[ Upstream commit 666031fed8f0fdfc29b20d125a628c1b0a04cdaf ]

Raw RSS QP restrack tracking wasn't working to begin with as it was only tracking the first raw RSS QP which was added, since at creation the raw RSS QP number is reserved so the QP number for this qp type was always zero.
The following raw RSS QP additions were always failing silently.

Since the fix isn't trivial and there were no users that required or complained about this issue we are dropping this for now instead of fixing.

---
 drivers/infiniband/hw/mlx5/qp.c | 1 +
 1 file changed, 1 insertion(+)

index 0a9ae84600b204..df4acd9de4918c 100644
@@ -3090,6 +3090,7 @@ static int create_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd,
 	int err;
 
 	if (params->is_rss_raw) {
+		rdma_restrack_no_track(&qp->ibqp.res);
 		err = create_rss_raw_qp_tir(dev, pd, qp, params);
 		goto out;
 	}

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Maher Sanalla <[email protected]>

[ Upstream commit d881d60223aac8fdc12b227d89c76e131e92a9cd ]

set_rq_size() computes the RQ WQE size as "1 << rq_wqe_shift" based on the user-provided rq_wqe_shift, which is only checked to be greater than 32, so shifts of 32 are still accepted. A shift of 31 also overflows a signed integer, leading to undefined behavior.

Use check_shl_overflow() to compute the RQ WQE size and reject any invalid values.

---
 drivers/infiniband/hw/mlx5/qp.c | 11 ++++-------
 1 file changed, 4 insertions(+), 7 deletions(-)

index df4acd9de4918c..6bbf12732051d8 100644
@@ -451,16 +451,13 @@ static int set_rq_size(struct mlx5_ib_dev *dev, struct ib_qp_cap *cap,
 
 		if (ucmd) {
 			qp->rq.wqe_cnt = ucmd->rq_wqe_count;
-			if (ucmd->rq_wqe_shift > BITS_PER_BYTE * sizeof(ucmd->rq_wqe_shift))
-				return -EINVAL;
 			qp->rq.wqe_shift = ucmd->rq_wqe_shift;
-			if ((1 << qp->rq.wqe_shift) /
-				    sizeof(struct mlx5_wqe_data_seg) <
-			    wq_sig)
+			if (check_shl_overflow(1, qp->rq.wqe_shift, &wqe_size))
+				return -EINVAL;
+			if (wqe_size / sizeof(struct mlx5_wqe_data_seg) < wq_sig)
 				return -EINVAL;
 			qp->rq.max_gs =
-				(1 << qp->rq.wqe_shift) /
-					sizeof(struct mlx5_wqe_data_seg) -
+				wqe_size / sizeof(struct mlx5_wqe_data_seg) -
 				wq_sig;
 			qp->rq.max_post = qp->rq.wqe_cnt;
 		} else {

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Leon Romanovsky <[email protected]>

[ Upstream commit 449ae7927152e46acbe5f19f97eafdae6d3a96b1 ]

Free the UAR index returned by the hardware.

---
 drivers/infiniband/hw/mlx5/main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index d523bc9052ede3..10422dee6db818 100644
@@ -2282,7 +2282,7 @@ static int uar_mmap(struct mlx5_ib_dev *dev, enum mlx5_ib_mmap_cmd cmd,
 	if (!dyn_uar)
 		return err;
 
-	mlx5_cmd_uar_dealloc(dev->mdev, idx, context->devx_uid);
+	mlx5_cmd_uar_dealloc(dev->mdev, uar_index, context->devx_uid);
 
 free_bfreg:
 	mlx5_ib_free_bfreg(dev, bfregi, bfreg_dyn_idx);

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: HyeongJun An <[email protected]>

[ Upstream commit 0b08baeccdcf52fad328ad645f5b4fbee04eea34 ]

hdac_hdmi_set_pin_port_mux() uses the written enum value to index the texts array before calling snd_soc_dapm_put_enum_double(), which validates that the value is within the enum item range.

An out-of-range value can therefore make the driver read past the texts array before the helper rejects the write. Move the lookup after the helper has accepted the value.

---
 sound/soc/codecs/hdac_hdmi.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

index 0ddfb0cb376fde..7707f0b5562d4d 100644
@@ -906,12 +906,14 @@ static int hdac_hdmi_set_pin_port_mux(struct snd_kcontrol *kcontrol,
 	struct hdac_device *hdev = dev_to_hdac_dev(dapm->dev);
 	struct hdac_hdmi_priv *hdmi = hdev_to_hdmi_priv(hdev);
 	struct hdac_hdmi_pcm *pcm;
-	const char *cvt_name =  e->texts[ucontrol->value.enumerated.item[0]];
+	const char *cvt_name;
 
 	ret = snd_soc_dapm_put_enum_double(kcontrol, ucontrol);
 	if (ret < 0)
 		return ret;
 
+	cvt_name = e->texts[ucontrol->value.enumerated.item[0]];
+
 	if (port == NULL)
 		return -EINVAL;
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: HyeongJun An <[email protected]>

[ Upstream commit 3cd17e4e2871114d5579fa7bc8da66faf7fc1930 ]

fsl_audmix_put_mix_clk_src() and fsl_audmix_put_out_src() convert the user-provided enum item with snd_soc_enum_item_to_val() before checking whether the item is within the enum's item count.

The generic snd_soc_put_enum_double() helper performs that validation, but these callbacks use the converted value first: the clock-source path tests it with BIT(), and the output-source path indexes the prms transition table with it.

Reject out-of-range enum items before converting them.

---
 sound/soc/fsl/fsl_audmix.c | 6 ++++++
 1 file changed, 6 insertions(+)

index d8e44470f501e7..8bc02ceda13c2b 100644
@@ -117,6 +117,9 @@ static int fsl_audmix_put_mix_clk_src(struct snd_kcontrol *kcontrol,
 	unsigned int *item = ucontrol->value.enumerated.item;
 	unsigned int reg_val, val, mix_clk;
 
+	if (item[0] >= e->items)
+		return -EINVAL;
+
 	/* Get current state */
 	reg_val = snd_soc_component_read(comp, FSL_AUDMIX_CTR);
 	mix_clk = ((reg_val & FSL_AUDMIX_CTR_MIXCLK_MASK)

@@ -157,6 +160,9 @@ static int fsl_audmix_put_out_src(struct snd_kcontrol *kcontrol,
 	unsigned int reg_val, val, mask = 0, ctr = 0;
 	int ret;
 
+	if (item[0] >= e->items)
+		return -EINVAL;
+
 	/* Get current state */
 	reg_val = snd_soc_component_read(comp, FSL_AUDMIX_CTR);
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: HyeongJun An <[email protected]>

[ Upstream commit 1d8aabb413b5638670dfd1162169edc0ba276a2e ]

tegra_ahub_put_value_enum() reads e->values[item[0]] before checking whether item[0] is within the enum item range. The existing check therefore happens too late to prevent an out-of-range read of the values array.

Move the check before the array access.

---
 sound/soc/tegra/tegra210_ahub.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

index 140cb27f73287e..f7918bcd918d47 100644
@@ -60,13 +60,15 @@ static int tegra_ahub_put_value_enum(struct snd_kcontrol *kctl,
 	struct soc_enum *e = (struct soc_enum *)kctl->private_value;
 	struct snd_soc_dapm_update update[TEGRA_XBAR_UPDATE_MAX_REG] = { };
 	unsigned int *item = uctl->value.enumerated.item;
-	unsigned int value = e->values[item[0]];
+	unsigned int value;
 	unsigned int i, bit_pos, reg_idx = 0, reg_val = 0;
 	int change = 0;
 
 	if (item[0] >= e->items)
 		return -EINVAL;
 
+	value = e->values[item[0]];
+
 	if (value) {
 		/* Get the register index and value to set */
 		reg_idx = (value - 1) / (8 * cmpnt->val_bytes);

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Jamal Hadi Salim <[email protected]>

[ Upstream commit f294fc71c4a0fa4964f6428a1b4e7929c1d83125 ]

The flow classifier falls back to addr_fold() for fields that are missing from packet headers. In map mode, userspace controls mask, xor, rshift, addend and divisor, and can observe the resulting classid through class statistics. This allows a tc classifier in a user/network namespace to recover the 32-bit folded value of skb->sk, skb_dst() or skb_nfct().

Align with standard kernel practices for pointer hashing and replace the XOR folding with a keyed siphash (which is cryptographically secure)

---
 net/sched/cls_flow.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

index d92ffdaf546c39..9ba13953d3a5e6 100644
@@ -21,6 +21,7 @@
 #include <net/inet_sock.h>
 
 #include <net/pkt_cls.h>
+#include <linux/siphash.h>
 #include <net/ip.h>
 #include <net/route.h>
 #include <net/flow_dissector.h>

@@ -57,11 +58,15 @@ struct flow_filter {
 	struct rcu_work		rwork;
 };
 
+static siphash_aligned_key_t flow_keys_secret __read_mostly;
+
 static inline u32 addr_fold(void *addr)
 {
-	unsigned long a = (unsigned long)addr;
-
-	return (a & 0xFFFFFFFF) ^ (BITS_PER_LONG > 32 ? a >> 32 : 0);
+#ifdef CONFIG_64BIT
+	return (u32)siphash_1u64((u64)addr, &flow_keys_secret);
+#else
+	return (u32)siphash_1u32((u32)addr, &flow_keys_secret);
+#endif
 }
 
 static u32 flow_get_src(const struct sk_buff *skb, const struct flow_keys *flow)

@@ -596,6 +601,7 @@ static int flow_init(struct tcf_proto *tp)
 		return -ENOBUFS;
 	INIT_LIST_HEAD(&head->filters);
 	rcu_assign_pointer(tp->root, head);
+	net_get_random_once(&flow_keys_secret, sizeof(flow_keys_secret));
 	return 0;
 }
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Kuniyuki Iwashima <[email protected]>

[ Upstream commit 2821e85c058f81c9948a2fb1a634f7b47457d51c ]

rocker_router_fib_event() calls fib_rule_get() during RCU dump.

If the fib_rule is dying, refcount_inc() will complain about it.

Let's call refcount_inc_not_zero() in fib_rules_dump().

---
 include/net/fib_rules.h | 5 +++++
 net/core/fib_rules.c    | 6 +++++-
 2 files changed, 10 insertions(+), 1 deletion(-)

index 82da359bca035d..c47f6f6d3cada6 100644
@@ -107,6 +107,11 @@ static inline void fib_rule_get(struct fib_rule *rule)
 	refcount_inc(&rule->refcnt);
 }
 
+static inline bool fib_rule_get_safe(struct fib_rule *rule)
+{
+	return refcount_inc_not_zero(&rule->refcnt);
+}
+
 static inline void fib_rule_put(struct fib_rule *rule)
 {
 	if (refcount_dec_and_test(&rule->refcnt))

index 75282222e0b456..e31873c87c0a36 100644
@@ -328,7 +328,7 @@ int fib_rules_lookup(struct fib_rules_ops *ops, struct flowi *fl,
 
 		if (err != -EAGAIN) {
 			if ((arg->flags & FIB_LOOKUP_NOREF) ||
-			    likely(refcount_inc_not_zero(&rule->refcnt))) {
+			    likely(fib_rule_get_safe(rule))) {
 				arg->rule = rule;
 				goto out;
 			}

@@ -386,8 +386,12 @@ int fib_rules_dump(struct net *net, struct notifier_block *nb, int family,
 	if (!ops)
 		return -EAFNOSUPPORT;
 	list_for_each_entry_rcu(rule, &ops->rules_list, list) {
+		if (!fib_rule_get_safe(rule))
+			continue;
+
 		err = call_fib_rule_notifier(nb, FIB_EVENT_RULE_ADD,
 					     rule, family, extack);
+		fib_rule_put(rule);
 		if (err)
 			break;
 	}

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Xiang Mei <[email protected]>

[ Upstream commit f3e02edd8322b31b8e6517faa6ba053bf29d1e26 ]

ccm_tx_work_expired() re-arms itself via queue_delayed_work() using the configured exp_interval converted by interval_to_us(). When exp_interval is BR_CFM_CCM_INTERVAL_NONE or out of range, interval_to_us() returns 0, causing the worker to fire immediately in a tight loop that allocates skbs until OOM.

Fix this by validating exp_interval at configuration time:

- Constrain IFLA_BRIDGE_CFM_CC_CONFIG_EXP_INTERVAL to the valid range [BR_CFM_CCM_INTERVAL_3_3_MS, BR_CFM_CCM_INTERVAL_10_MIN] in the netlink policy so userspace cannot set an invalid value.

- Reject starting CCM TX in br_cfm_cc_ccm_tx() when exp_interval has not yet been configured (defaults to 0 from kzalloc).

---
 net/bridge/br_cfm.c         | 6 ++++++
 net/bridge/br_cfm_netlink.c | 4 +++-
 2 files changed, 9 insertions(+), 1 deletion(-)

index a3c755d0a09de9..a89aec83141648 100644
@@ -805,6 +805,12 @@ int br_cfm_cc_ccm_tx(struct net_bridge *br, const u32 instance,
 		goto save;
 	}
 
+	if (!interval_to_us(mep->cc_config.exp_interval)) {
+		NL_SET_ERR_MSG_MOD(extack,
+				   "Invalid CCM interval");
+		return -EINVAL;
+	}
+
 	/* Start delayed work to transmit CCM frames. It is done with zero delay
 	 * to send first frame immediately
 	 */

index 2faab44652e7c0..91b9922dc3f25e 100644
@@ -34,7 +34,9 @@ br_cfm_cc_config_policy[IFLA_BRIDGE_CFM_CC_CONFIG_MAX + 1] = {
 	[IFLA_BRIDGE_CFM_CC_CONFIG_UNSPEC]	 = { .type = NLA_REJECT },
 	[IFLA_BRIDGE_CFM_CC_CONFIG_INSTANCE]	 = { .type = NLA_U32 },
 	[IFLA_BRIDGE_CFM_CC_CONFIG_ENABLE]	 = { .type = NLA_U32 },
-	[IFLA_BRIDGE_CFM_CC_CONFIG_EXP_INTERVAL] = { .type = NLA_U32 },
+	[IFLA_BRIDGE_CFM_CC_CONFIG_EXP_INTERVAL] =
+		NLA_POLICY_RANGE(NLA_U32, BR_CFM_CCM_INTERVAL_3_3_MS,
+				 BR_CFM_CCM_INTERVAL_10_MIN),
 	[IFLA_BRIDGE_CFM_CC_CONFIG_EXP_MAID]	 = {
 	.type = NLA_BINARY, .len = CFM_MAID_LENGTH },
 };

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Xin Long <[email protected]>

[ Upstream commit e9361d0ca55c4af12aac09e2572852fa91046229 ]

sctp_verify_asconf() and sctp_verify_param() only validate ADD_IP, DEL_IP, and SET_PRIMARY parameters against a fixed minimum size of sizeof(struct sctp_addip_param) + sizeof(struct sctp_paramhdr). This ensures the outer parameter is large enough to contain an embedded address parameter header, but does not verify that the embedded address parameter's declared length fits within the bounds of the outer parameter.

Later, sctp_process_param() and sctp_process_asconf_param() extract the embedded address parameter and pass it to af->from_addr_param(), which uses the address parameter length to parse the variable-length address payload. A malformed peer can therefore advertise an embedded address parameter length that exceeds the remaining bytes in the enclosing parameter.

Validate that addr_param->p.length does not exceed the space available after the sctp_addip_param header before processing the embedded address parameter. Reject malformed parameters when the embedded address length extends beyond the enclosing parameter bounds.

This prevents out-of-bounds reads when parsing malformed parameters carried in INIT or ASCONF processing paths.

---
 net/sctp/sm_make_chunk.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

index eb2ed7db3fe16b..611bddad36d4ac 100644
@@ -2662,6 +2662,9 @@ static int sctp_process_param(struct sctp_association *asoc,
 			goto fall_through;
 
 		addr_param = param.v + sizeof(struct sctp_addip_param);
+		if (ntohs(addr_param->p.length) >
+		    ntohs(param.p->length) - sizeof(struct sctp_addip_param))
+			break;
 
 		af = sctp_get_af_specific(param_type2af(addr_param->p.type));
 		if (!af)

@@ -3060,13 +3063,16 @@ static __be16 sctp_process_asconf_param(struct sctp_association *asoc,
 	union sctp_addr	addr;
 	struct sctp_af *af;
 
-	addr_param = (void *)asconf_param + sizeof(*asconf_param);
-
 	if (asconf_param->param_hdr.type != SCTP_PARAM_ADD_IP &&
 	    asconf_param->param_hdr.type != SCTP_PARAM_DEL_IP &&
 	    asconf_param->param_hdr.type != SCTP_PARAM_SET_PRIMARY)
 		return SCTP_ERROR_UNKNOWN_PARAM;
 
+	addr_param = (void *)asconf_param + sizeof(*asconf_param);
+	if (ntohs(addr_param->p.length) >
+	    ntohs(asconf_param->param_hdr.length) - sizeof(*asconf_param))
+		return SCTP_ERROR_PROTO_VIOLATION;
+
 	switch (addr_param->p.type) {
 	case SCTP_PARAM_IPV6_ADDRESS:
 		if (!asoc->peer.ipv6_address)

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Victor Nogueira <[email protected]>

[ Upstream commit 90b662ea25f5e83bb3b8ccec5b93ced810b92fb8 ]

update_vf() is called from two places for the same class during a single dequeue when the class's child qdisc (e.g. codel/fq_codel) drops its last packets while dequeuing:

1. The child calls qdisc_tree_reduce_backlog(), which, now that the child is empty, invokes hfsc_qlen_notify() -> update_vf(cl, 0, 0) and turns the class passive (cl_nactive is decremented up the hierarchy).

2. hfsc_dequeue() then calls update_vf(cl, qdisc_pkt_len(skb), cur_time) to charge the dequeued bytes.

On the second call the class is already passive, but its child qdisc is still empty, so update_vf() arms go_passive again:

      if (cl->qdisc->q.qlen == 0 && cl->cl_flags & HFSC_FSC)
              go_passive = 1;

The leaf is then skipped by the cl_nactive == 0 check inside the loop, which does not clear go_passive, so the stale go_passive propagates to the parent and decrements its cl_nactive a second time. A parent that still has other active children is driven to cl_nactive == 0 and removed from the vttree, even though those siblings are still backlogged. They are never dequeued again and the qdisc stalls.

Fix this by only arming go_passive when the class is actually active, so an already-passive class no longer triggers a second passive transition. The byte accounting (cl->cl_total += len) still runs for every ancestor, so dequeued bytes continue to be counted exactly once.

---
 net/sched/sch_hfsc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 11ae998cda3a85..71f96a8980a7cd 100644
@@ -753,7 +753,7 @@ update_vf(struct hfsc_class *cl, unsigned int len, u64 cur_time)
 	u64 f; /* , myf_bound, delta; */
 	int go_passive = 0;
 
-	if (cl->qdisc->q.qlen == 0 && cl->cl_flags & HFSC_FSC)
+	if (cl->qdisc->q.qlen == 0 && cl->cl_flags & HFSC_FSC && cl->cl_nactive)
 		go_passive = 1;
 
 	for (; cl->cl_parent != NULL; cl = cl->cl_parent) {

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Michael Bommarito <[email protected]>

[ Upstream commit 86b0c540e2ea397cde021eecd24145f7c16a3d4e ]

TIPCv2 registers mutating generic-netlink operations without admin permission flags. Generic netlink only checks CAP_NET_ADMIN when an operation sets GENL_ADMIN_PERM or GENL_UNS_ADMIN_PERM, so a local unprivileged process can currently change TIPC state through commands such as TIPC_NL_NET_SET, TIPC_NL_KEY_SET, TIPC_NL_KEY_FLUSH, and bearer enable/disable.

The legacy TIPC netlink API already checks netlink_net_capable(..., CAP_NET_ADMIN) for administrative commands. Give the TIPCv2 mutators the equivalent generic-netlink gate. Use GENL_UNS_ADMIN_PERM, which maps to the same namespace-aware CAP_NET_ADMIN check that netlink_net_capable() performs, so the behaviour matches the legacy path and keeps working for CAP_NET_ADMIN holders in a non-initial user namespace (containers).

A QEMU/KASAN repro run as uid/gid 65534 with zero effective capabilities previously succeeded in changing the network id and node identity, setting and flushing key material, and enabling/disabling a UDP bearer. With this patch applied the same operations fail with -EPERM.

---
 net/tipc/netlink.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

index 1a9a5bdaccf4fc..8336a9664703fe 100644
@@ -152,11 +152,13 @@ static const struct genl_ops tipc_genl_v2_ops[] = {
 	{
 		.cmd	= TIPC_NL_BEARER_DISABLE,
 		.validate = GENL_DONT_VALIDATE_STRICT | GENL_DONT_VALIDATE_DUMP,
+		.flags	= GENL_UNS_ADMIN_PERM,
 		.doit	= tipc_nl_bearer_disable,
 	},
 	{
 		.cmd	= TIPC_NL_BEARER_ENABLE,
 		.validate = GENL_DONT_VALIDATE_STRICT | GENL_DONT_VALIDATE_DUMP,
+		.flags	= GENL_UNS_ADMIN_PERM,
 		.doit	= tipc_nl_bearer_enable,
 	},
 	{

@@ -168,11 +170,13 @@ static const struct genl_ops tipc_genl_v2_ops[] = {
 	{
 		.cmd	= TIPC_NL_BEARER_ADD,
 		.validate = GENL_DONT_VALIDATE_STRICT | GENL_DONT_VALIDATE_DUMP,
+		.flags	= GENL_UNS_ADMIN_PERM,
 		.doit	= tipc_nl_bearer_add,
 	},
 	{
 		.cmd	= TIPC_NL_BEARER_SET,
 		.validate = GENL_DONT_VALIDATE_STRICT | GENL_DONT_VALIDATE_DUMP,
+		.flags	= GENL_UNS_ADMIN_PERM,
 		.doit	= tipc_nl_bearer_set,
 	},
 	{

@@ -197,11 +201,13 @@ static const struct genl_ops tipc_genl_v2_ops[] = {
 	{
 		.cmd	= TIPC_NL_LINK_SET,
 		.validate = GENL_DONT_VALIDATE_STRICT | GENL_DONT_VALIDATE_DUMP,
+		.flags	= GENL_UNS_ADMIN_PERM,
 		.doit	= tipc_nl_node_set_link,
 	},
 	{
 		.cmd	= TIPC_NL_LINK_RESET_STATS,
 		.validate = GENL_DONT_VALIDATE_STRICT | GENL_DONT_VALIDATE_DUMP,
+		.flags	= GENL_UNS_ADMIN_PERM,
 		.doit   = tipc_nl_node_reset_link_stats,
 	},
 	{

@@ -213,6 +219,7 @@ static const struct genl_ops tipc_genl_v2_ops[] = {
 	{
 		.cmd	= TIPC_NL_MEDIA_SET,
 		.validate = GENL_DONT_VALIDATE_STRICT | GENL_DONT_VALIDATE_DUMP,
+		.flags	= GENL_UNS_ADMIN_PERM,
 		.doit	= tipc_nl_media_set,
 	},
 	{

@@ -228,6 +235,7 @@ static const struct genl_ops tipc_genl_v2_ops[] = {
 	{
 		.cmd	= TIPC_NL_NET_SET,
 		.validate = GENL_DONT_VALIDATE_STRICT | GENL_DONT_VALIDATE_DUMP,
+		.flags	= GENL_UNS_ADMIN_PERM,
 		.doit	= tipc_nl_net_set,
 	},
 	{

@@ -238,6 +246,7 @@ static const struct genl_ops tipc_genl_v2_ops[] = {
 	{
 		.cmd	= TIPC_NL_MON_SET,
 		.validate = GENL_DONT_VALIDATE_STRICT | GENL_DONT_VALIDATE_DUMP,
+		.flags	= GENL_UNS_ADMIN_PERM,
 		.doit	= tipc_nl_node_set_monitor,
 	},
 	{

@@ -255,6 +264,7 @@ static const struct genl_ops tipc_genl_v2_ops[] = {
 	{
 		.cmd	= TIPC_NL_PEER_REMOVE,
 		.validate = GENL_DONT_VALIDATE_STRICT | GENL_DONT_VALIDATE_DUMP,
+		.flags	= GENL_UNS_ADMIN_PERM,
 		.doit	= tipc_nl_peer_rm,
 	},
 #ifdef CONFIG_TIPC_MEDIA_UDP

@@ -269,11 +279,13 @@ static const struct genl_ops tipc_genl_v2_ops[] = {
 	{
 		.cmd	= TIPC_NL_KEY_SET,
 		.validate = GENL_DONT_VALIDATE_STRICT | GENL_DONT_VALIDATE_DUMP,
+		.flags	= GENL_UNS_ADMIN_PERM,
 		.doit	= tipc_nl_node_set_key,
 	},
 	{
 		.cmd	= TIPC_NL_KEY_FLUSH,
 		.validate = GENL_DONT_VALIDATE_STRICT | GENL_DONT_VALIDATE_DUMP,
+		.flags	= GENL_UNS_ADMIN_PERM,
 		.doit	= tipc_nl_node_flush_key,
 	},
 #endif

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Michael Bommarito <[email protected]>

[ Upstream commit ab3e10b44ba5411779aac7afd2477917dd77750f ]

tipc_sk_conn_proto_rcv() subtracts the peer-supplied connection ack count from the unsigned 16-bit send counter snt_unacked without checking that it does not exceed the number of messages actually outstanding:

	tsk->snt_unacked -= msg_conn_ack(hdr);

msg_conn_ack() is read straight from a received CONN_MANAGER/CONN_ACK message. If the ack count is larger than snt_unacked, the subtraction wraps to a near-maximum value, leaving tsk_conn_cong() permanently true and starving the connection of further transmits.

Validate the ACK count at the start of the CONN_ACK block and drop the message if it acknowledges more messages than are outstanding. A peer (or, for a local connection, the connected peer socket) can otherwise wedge a TIPC connection's send side by sending an oversized connection ack.

---
 net/tipc/socket.c | 3 +++
 1 file changed, 3 insertions(+)

index e5931e56ef094e..31c6b61f0b0fd9 100644
@@ -1367,6 +1367,9 @@ static void tipc_sk_conn_proto_rcv(struct tipc_sock *tsk, struct sk_buff *skb,
 			__skb_queue_tail(xmitq, skb);
 		return;
 	} else if (mtyp == CONN_ACK) {
+		if (tsk->snt_unacked < msg_conn_ack(hdr))
+			goto exit;
+
 		was_cong = tsk_conn_cong(tsk);
 		tipc_sk_push_backlog(tsk, msg_nagle_ack(hdr));
 		tsk->snt_unacked -= msg_conn_ack(hdr);

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Michael Bommarito <[email protected]>

[ Upstream commit 2afb648f7b99216c687db1f89739c995e1144153 ]

tipc_update_nametbl() inserts a binding advertised by a peer node using the lower and upper service-range bounds taken directly from the wire, without checking that lower <= upper. The local bind path validates the ordering (tipc_uaddr_valid()), but the name-distribution path does not.

A binding with lower > upper is inserted at the far end of the service-range rbtree (keyed on lower) where no lookup or withdrawal can ever match it (service_range_foreach_match() requires sr->lower <= end). The publication, its service_range node and the augmented rbtree entry are then leaked for the lifetime of the namespace, and there is no per-peer cap equivalent to TIPC_MAX_PUBL on locally created bindings.

Reject inverted ranges in the network path as well. A peer node can otherwise leak unbounded binding-table memory by sending PUBLICATION items with lower > upper.

---
 net/tipc/name_distr.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

index 190b49c5cbc3ec..ba4f4906e13b70 100644
@@ -280,12 +280,21 @@ static bool tipc_update_nametbl(struct net *net, struct distr_item *i,
 				u32 node, u32 dtype)
 {
 	struct publication *p = NULL;
+	u32 lower = ntohl(i->lower);
+	u32 upper = ntohl(i->upper);
 	struct tipc_socket_addr sk;
-	struct tipc_uaddr ua;
 	u32 key = ntohl(i->key);
+	struct tipc_uaddr ua;
+
+	/* A peer-advertised binding with lower > upper can never be matched
+	 * or withdrawn and would leak the publication; the local bind path
+	 * rejects such ranges, so reject ranges learned from the network too.
+	 */
+	if (lower > upper)
+		return false;
 
 	tipc_uaddr(&ua, TIPC_SERVICE_RANGE, TIPC_CLUSTER_SCOPE,
-		   ntohl(i->type), ntohl(i->lower), ntohl(i->upper));
+		   ntohl(i->type), lower, upper);
 	sk.ref = ntohl(i->port);
 	sk.node = node;
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Dave Jiang <[email protected]>

[ Upstream commit dfe28c8592538152e9611341dae6f7be1735b3f1 ]

alloc_mock_res() calls kzalloc() without checking the return value. Add scope based resource management to deal with the allocated memory cleanly.

---
 tools/testing/cxl/test/cxl.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

index ed1361c8b96c48..0bb93bff74be0f 100644
@@ -396,12 +396,16 @@ static void depopulate_all_mock_resources(void)
 
 static struct cxl_mock_res *alloc_mock_res(resource_size_t size, int align)
 {
-	struct cxl_mock_res *res = kzalloc(sizeof(*res), GFP_KERNEL);
 	struct genpool_data_align data = {
 		.align = align,
 	};
 	unsigned long phys;
 
+	struct cxl_mock_res *res __free(kfree) = kzalloc(sizeof(*res),
+							 GFP_KERNEL);
+	if (!res)
+		return NULL;
+
 	INIT_LIST_HEAD(&res->list);
 	phys = gen_pool_alloc_algo(cxl_mock_pool, size,
 				   gen_pool_first_fit_align, &data);

@@ -416,7 +420,7 @@ static struct cxl_mock_res *alloc_mock_res(resource_size_t size, int align)
 	list_add(&res->list, &mock_res);
 	mutex_unlock(&mock_res_lock);
 
-	return res;
+	return no_free_ptr(res);
 }
 
 static int populate_cedt(void)

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Felix Gu <[email protected]>

[ Upstream commit 7891c64c0520519782470ba29bac8a5761e295d8 ]

The sg_cleanup path used list[i] instead of list[j] when unmapping DMA buffers, leaking successfully mapped entries and repeatedly unmapping the failed one.

---
 drivers/crypto/marvell/octeontx/otx_cptvf_reqmgr.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

index c80baf1ad90b2f..89030e2711ce41 100644
@@ -157,8 +157,8 @@ static inline int setup_sgio_components(struct pci_dev *pdev,
 sg_cleanup:
 	for (j = 0; j < i; j++) {
 		if (list[j].dma_addr) {
-			dma_unmap_single(&pdev->dev, list[i].dma_addr,
-					 list[i].size, DMA_BIDIRECTIONAL);
+			dma_unmap_single(&pdev->dev, list[j].dma_addr,
+					 list[j].size, DMA_BIDIRECTIONAL);
 		}
 
 		list[j].dma_addr = 0;

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Felix Gu <[email protected]>

[ Upstream commit 9dbf173bd32d5f81b005008b682bfb50aa093455 ]

The sg_cleanup error path used list[i] instead of list[j] when unmapping DMA buffers, leaking successfully mapped entries and repeatedly unmapping the failed one.

---
 drivers/crypto/cavium/cpt/cptvf_reqmanager.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

index 153004bdfb5cde..753bdb2ae33f8d 100644
@@ -108,8 +108,8 @@ static int setup_sgio_components(struct cpt_vf *cptvf, struct buf_ptr *list,
 sg_cleanup:
 	for (j = 0; j < i; j++) {
 		if (list[j].dma_addr) {
-			dma_unmap_single(&pdev->dev, list[i].dma_addr,
-					 list[i].size, DMA_BIDIRECTIONAL);
+			dma_unmap_single(&pdev->dev, list[j].dma_addr,
+					 list[j].size, DMA_BIDIRECTIONAL);
 		}
 
 		list[j].dma_addr = 0;

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Herbert Xu <[email protected]>

[ Upstream commit 606ba888b98e0d26a2c4e5c8dc0542e3ad8f0f3a ]

When the rng module is removed the default RNG will be leaked. Call crypto_del_default_rng to free it if possible.

---
 crypto/rng.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

index e462031fef68c6..c49115a5b2668d 100644
@@ -259,5 +259,16 @@ void crypto_unregister_rngs(struct rng_alg *algs, int count)
 }
 EXPORT_SYMBOL_GPL(crypto_unregister_rngs);
 
+static void __exit rng_exit(void)
+{
+	int err;
+
+	err = crypto_del_default_rng();
+	if (err)
+		pr_err("Failed delete default RNG: %d\n", err);
+}
+
+module_exit(rng_exit);
+
 MODULE_LICENSE("GPL");
 MODULE_DESCRIPTION("Random Number Generator");

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Lars Pöschel <[email protected]>

[ Upstream commit 47f3b5365536e8c38f264824ab15fdb74454e066 ]

The method the driver uses to determine the size of the FIFO has a problem. What it currently does is this:
It stops the SPI hardware and writes to the TX FIFO register until TX FIFO FULL asserts in the status register. But the hardware does not only have the FIFO, it also has a shift register which can hold a byte. This can be seen, when writing a byte to the FIFO (while the SPI hardware is stopped,) the TX FIFO EMPTY is still empty. So, if we have a FIFO size of 16 for example, the current method returns a 17.
This is a problem, at least when using the driver in irq mode. The same size determined for the TX FIFO is also assumed for the RX FIFO. When a SPI transaction wants to write the amount of the FIFO size or more bytes, the following happens, for example with 16 bytes FIFO size:
The driver stops the SPI hardware and writes 17 bytes to the TX FIFO and starts the SPI hardware and goes sleep.
The hardware then shifts out 17 bytes (FIFO + shift register) and simultaneously reads bytes into the RX FIFO, but it only has 16 places, so it looses one byte. Then TX FIFO empty asserts, wakes the driver again, which has a fast path and reads 16 bytes from the RX FIFO, but before reading the last 17th byte (which is lost) it does this:

	sr = xspi->read_fn(xspi->regs + XSPI_SR_OFFSET);
	if (!(sr & XSPI_SR_RX_EMPTY_MASK)) {
		xilinx_spi_rx(xspi);
		rx_words--;
	}

It reads the status register and checks if the RX FIFO is not empty. But it is empty in our case. So this check spins in a while loop forever locking the driver.

This patch fixes the logic to determine the FIFO size.

---
 drivers/spi/spi-xilinx.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

index b5405d9974ba6b..36362bee7fff81 100644
@@ -370,11 +370,18 @@ static int xilinx_spi_find_buffer_size(struct xilinx_spi *xspi)
 		xspi->regs + XIPIF_V123B_RESETR_OFFSET);
 
 	/* Fill the Tx FIFO with as many words as possible */
-	do {
+	while (1) {
 		xspi->write_fn(0, xspi->regs + XSPI_TXD_OFFSET);
 		sr = xspi->read_fn(xspi->regs + XSPI_SR_OFFSET);
+		if (sr & XSPI_SR_TX_FULL_MASK)
+			break;
+
 		n_words++;
-	} while (!(sr & XSPI_SR_TX_FULL_MASK));
+	}
+
+	/* Handle the NO FIFO case separately */
+	if (!n_words)
+		return 1;
 
 	return n_words;
 }

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Guangshuo Li <[email protected]>

[ Upstream commit 69b4141b428bcf2cf7a863950c0d6e5c5ae89ac1 ]

The PLL lock failure path in adau1372_set_power() unwinds by putting the regmap back in cache-only mode, asserting the optional power-down GPIO and disabling mclk.

adau1372_enable_pll() enables CLK_CTRL.PLL_EN before polling the PLL lock bit. If the lock fails on a board without a power-down GPIO, the error path disables mclk and returns an error, but leaves PLL_EN set in the hardware register. The normal power-off path already handles the no-GPIO case by explicitly clearing PLL_EN.

Mirror that cleanup in the PLL lock failure path and clear PLL_EN while the regmap is still live, before switching it back to cache-only mode.

---
 sound/soc/codecs/adau1372.c | 5 +++++
 1 file changed, 5 insertions(+)

index 25f123110b5be4..7dc0b5c79e3998 100644
@@ -812,6 +812,11 @@ static int adau1372_set_power(struct adau1372 *adau1372, bool enable)
 		if (adau1372->use_pll) {
 			ret = adau1372_enable_pll(adau1372);
 			if (ret) {
+				if (!adau1372->pd_gpio)
+					regmap_update_bits(adau1372->regmap,
+							   ADAU1372_REG_CLK_CTRL,
+							   ADAU1372_CLK_CTRL_PLL_EN,
+							   0);
 				regcache_cache_only(adau1372->regmap, true);
 				if (adau1372->pd_gpio)
 					gpiod_set_value(adau1372->pd_gpio, 1);

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Lucas Tsai <[email protected]>

[ Upstream commit ba61aed9a34671222d1149acfc2f0179a9ce7e80 ]

If dts property power-supplies has multiple values, then accessing to psy->supplied_from[i-1] in __power_supply_populate_supplied_from will overrun supplied_from array.

---
 drivers/power/supply/power_supply_core.c | 11 +++--------
 1 file changed, 3 insertions(+), 8 deletions(-)

index 1893d37dd575de..4f350db05efa09 100644
@@ -262,18 +262,13 @@ static int power_supply_check_supplies(struct power_supply *psy)
 	if (cnt == 1)
 		return 0;
 
-	/* All supplies found, allocate char ** array for filling */
-	psy->supplied_from = devm_kzalloc(&psy->dev, sizeof(*psy->supplied_from),
+	/* All supplies found, allocate char * array for filling */
+	psy->supplied_from = devm_kcalloc(&psy->dev,
+					  cnt - 1, sizeof(*psy->supplied_from),
 					  GFP_KERNEL);
 	if (!psy->supplied_from)
 		return -ENOMEM;
 
-	*psy->supplied_from = devm_kcalloc(&psy->dev,
-					   cnt - 1, sizeof(**psy->supplied_from),
-					   GFP_KERNEL);
-	if (!*psy->supplied_from)
-		return -ENOMEM;
-
 	return power_supply_populate_supplied_from(psy);
 }
 #else

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Chuck Lever <[email protected]>

[ Upstream commit 81246a65303d9635266b1334490142caaf86a11f ]

ACCEPT and DONE are the two downcalls of the handshake genl family, both intended for use by the trusted handshake agent (tlshd). ACCEPT already requires GENL_ADMIN_PERM; DONE has no privilege check at all.

The fd-lookup in handshake_nl_done_doit() only confirms that some pending handshake request exists for the supplied sockfd; it does not authenticate the sender. An unprivileged process that guesses or observes a valid sockfd can therefore submit a DONE with HANDSHAKE_A_DONE_STATUS == 0, leaving the kernel consumer to proceed as if the handshake succeeded. A non-zero status on a forged DONE tears down a legitimate in-flight handshake before tlshd can report its real result.

---
 Documentation/netlink/specs/handshake.yaml | 1 +
 net/handshake/genl.c                       | 2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)

index 6d89e30f5fd52b..0888c39ec2bcd0 100644
@@ -113,6 +113,7 @@ operations:
       name: done
       doc: Handler reports handshake completion
       attribute-set: done
+      flags: [admin-perm]
       do:
         request:
           attributes:

index 233be5cbfec91a..8fdec426e580a0 100644
@@ -36,7 +36,7 @@ static const struct genl_split_ops handshake_nl_ops[] = {
 		.doit		= handshake_nl_done_doit,
 		.policy		= handshake_done_nl_policy,
 		.maxattr	= HANDSHAKE_A_DONE_REMOTE_AUTH,
-		.flags		= GENL_CMD_CAP_DO,
+		.flags		= GENL_ADMIN_PERM | GENL_CMD_CAP_DO,
 	},
 };
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Victor Nogueira <[email protected]>

[ Upstream commit 097f6fc7b1ae362dd7a9444b2572162fda73b284 ]

Whenever fq_codel drops packets during peek, it calls qdisc_tree_reduce_backlog. An issue arises because it calls qdisc_tree_reduce_backlog before it reincrements the qlen. If qlen drops to zero, but peek returns an skb, the parent's qlen_notify callback will be executed even though fq_codel still has 1 packet on the queue and, thus, will mistakenly deactivate the parent's class causing issues like a recent report [1] and a wild memory access in qfq:

[   29.371146][  T360] Oops: general protection fault, probably for non-canonical address 0xfbd59c0000000024: 0000 [#1] SMP KASAN NOPTI
[   29.371666][  T360] KASAN: maybe wild-memory-access in range [0xdead000000000120-0xdead000000000127]
[   29.371987][  T360] CPU: 6 UID: 0 PID: 360 Comm: tc Not tainted 7.1.0-rc5-00285-gc530e5b2dbc6-dirty #82 PREEMPT(full)
[   29.372384][  T360] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[   29.372620][  T360] RIP: 0010:qfq_deactivate_agg (include/linux/list.h:1029 (discriminator 2) include/linux/list.h:1043 (discriminator 2) net/sched/sch_qfq.c:1369 (discriminator 2) net/sched/sch_qfq.c:1395 (discriminator 2)) sch_qfq
[   29.373544][  T360] RSP: 0018:ffff888102417370 EFLAGS: 00010216
[   29.373800][  T360] RAX: 0000000000000000 RBX: ffff88811224d568 RCX: dffffc0000000000
[   29.374079][  T360] RDX: 1ffff11021fe1543 RSI: ffff88810ff0aa00 RDI: dffffc0000000000
[   29.374368][  T360] RBP: ffff88811224c280 R08: dead000000000122 R09: 1bd5a00000000024
[   29.374649][  T360] R10: fffffbfff7940329 R11: fffffbfff7940329 R12: 0000000000000000
[   29.374926][  T360] R13: dead000000000100 R14: ffff88811224d580 R15: ffff88811224d578
[   29.375207][  T360] FS:  00007f5b794e5780(0000) GS:ffff88815d1e9000(0000) knlGS:0000000000000000
[   29.375545][  T360] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   29.375823][  T360] CR2: 000055ffb091f000 CR3: 000000010a305000 CR4: 0000000000750ef0
[   29.376103][  T360] PKRU: 55555554
[   29.376258][  T360] Call Trace:
[   29.376401][  T360]  <TASK>
...
[   29.376885][  T360] qfq_reset_qdisc (net/sched/sch_qfq.c:357 net/sched/sch_qfq.c:1487) sch_qfq
[   29.377074][  T360]  qdisc_reset (net/sched/sch_generic.c:1057)
[   29.377414][  T360]  __qdisc_destroy (net/sched/sch_generic.c:1096)
[   29.377600][  T360]  qdisc_graft (net/sched/sch_api.c:1062 net/sched/sch_api.c:1053 net/sched/sch_api.c:1159)
[   29.378593][  T360]  tc_get_qdisc (net/sched/sch_api.c:1528 net/sched/sch_api.c:1556)

Fix this by only calling qdisc_tree_reduce_backlog in peek after the qlen is restored.

[1] http://lore.kernel.org/netdev/CAN2cbVe79oj0O9==m4+4x3v+O+qzRagA=2=wkrp9i9=CqYvyZA@mail.gmail.com/

---
 net/sched/sch_fq_codel.c | 41 ++++++++++++++++++++++++++++++++++++++--
 1 file changed, 39 insertions(+), 2 deletions(-)

index 056895df178541..e059d6c145741b 100644
@@ -278,7 +278,7 @@ static void drop_func(struct sk_buff *skb, void *ctx)
 	qdisc_qstats_drop(sch);
 }
 
-static struct sk_buff *fq_codel_dequeue(struct Qdisc *sch)
+static struct sk_buff *__fq_codel_dequeue(struct Qdisc *sch)
 {
 	struct fq_codel_sched_data *q = qdisc_priv(sch);
 	struct sk_buff *skb;

@@ -315,12 +315,49 @@ static struct sk_buff *fq_codel_dequeue(struct Qdisc *sch)
 	qdisc_bstats_update(sch, skb);
 	flow->deficit -= qdisc_pkt_len(skb);
 
+	return skb;
+}
+
+static void fq_codel_dequeue_drop(struct Qdisc *sch)
+{
+	struct fq_codel_sched_data *q = qdisc_priv(sch);
+
 	if (q->cstats.drop_count) {
 		qdisc_tree_reduce_backlog(sch, q->cstats.drop_count,
 					  q->cstats.drop_len);
 		q->cstats.drop_count = 0;
 		q->cstats.drop_len = 0;
 	}
+}
+
+static struct sk_buff *fq_codel_dequeue(struct Qdisc *sch)
+{
+	struct sk_buff *skb;
+
+	skb =  __fq_codel_dequeue(sch);
+
+	fq_codel_dequeue_drop(sch);
+
+	return skb;
+}
+
+static struct sk_buff *fq_codel_peek(struct Qdisc *sch)
+{
+	struct sk_buff *skb = skb_peek(&sch->gso_skb);
+
+	if (!skb) {
+		skb = __fq_codel_dequeue(sch);
+
+		if (skb) {
+			__skb_queue_head(&sch->gso_skb, skb);
+			/* it's still part of the queue */
+			qdisc_qstats_backlog_inc(sch, skb);
+			sch->q.qlen++;
+		}
+
+		fq_codel_dequeue_drop(sch);
+	}
+
 	return skb;
 }
 

@@ -707,7 +744,7 @@ static struct Qdisc_ops fq_codel_qdisc_ops __read_mostly = {
 	.priv_size	=	sizeof(struct fq_codel_sched_data),
 	.enqueue	=	fq_codel_enqueue,
 	.dequeue	=	fq_codel_dequeue,
-	.peek		=	qdisc_peek_dequeued,
+	.peek		=	fq_codel_peek,
 	.init		=	fq_codel_init,
 	.reset		=	fq_codel_reset,
 	.destroy	=	fq_codel_destroy,

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Victor Nogueira <[email protected]>

[ Upstream commit 52f1da34c9f4d5bdc1e8b44242da5c7ba8db85f3 ]

Whenever codel drops packets during peek, it calls qdisc_tree_reduce_backlog. An issue arises because it calls qdisc_tree_reduce_backlog before it reincrements the qlen. If qlen drops to zero, but peek returns an skb, the parent's qlen_notify callback will be executed even though codel still has 1 packet on the queue and, thus, will mistakenly deactivate the parent's class causing issues like a wild memory access when qfq has codel as a child:

[   36.339843][  T370] Oops: general protection fault, probably for non-canonical address 0xfbd59c0000000024: 0000 [#1] SMP KASAN NOPTI
[   36.340408][  T370] KASAN: maybe wild-memory-access in range [0xdead000000000120-0xdead000000000127]
[   36.340737][  T370] CPU: 2 UID: 0 PID: 370 Comm: tc Not tainted 7.1.0-rc5-00287-g66e13b626592 #87 PREEMPT(full)
[   36.341113][  T370] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[   36.341357][  T370] RIP: 0010:qfq_deactivate_agg (include/linux/list.h:1029 (discriminator 2) include/linux/list.h:1043 (discriminator 2) net/sched/sch_qfq.c:1369 (discriminator 2) net/sched/sch_qfq.c:1395 (discriminator 2)) sch_qfq
[   36.342221][  T370] RSP: 0018:ffff8881100ef370 EFLAGS: 00010216
[   36.342422][  T370] RAX: 0000000000000000 RBX: ffff8881058a9568 RCX: dffffc0000000000
[   36.342664][  T370] RDX: 1ffff11021064dc3 RSI: ffff888108326e00 RDI: dffffc0000000000
[   36.342905][  T370] RBP: ffff8881058a8280 R08: dead000000000122 R09: 1bd5a00000000024
[   36.343140][  T370] R10: fffffbfff2940329 R11: fffffbfff2940329 R12: 0000000000000000
[   36.343383][  T370] R13: dead000000000100 R14: ffff8881058a9580 R15: ffff8881058a9578
[   36.343631][  T370] FS:  00007fc04b0ca780(0000) GS:ffff888184fef000(0000) knlGS:0000000000000000
[   36.343911][  T370] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   36.344116][  T370] CR2: 0000557c02c02000 CR3: 000000010e0ba000 CR4: 0000000000750ef0
[   36.344359][  T370] PKRU: 55555554
[   36.344481][  T370] Call Trace:
...
[   36.345054][  T370] qfq_reset_qdisc (net/sched/sch_qfq.c:357 net/sched/sch_qfq.c:1487) sch_qfq
[   36.345222][  T370]  qdisc_reset (net/sched/sch_generic.c:1057)
[   36.345503][  T370]  __qdisc_destroy (net/sched/sch_generic.c:1096)
[   36.345677][  T370]  qdisc_graft (net/sched/sch_api.c:1062 net/sched/sch_api.c:1053 net/sched/sch_api.c:1159)
[   36.346335][  T370]  tc_get_qdisc (net/sched/sch_api.c:1528 net/sched/sch_api.c:1556)

Fix this by only calling qdisc_tree_reduce_backlog in peek after the qlen is restored.

---
 net/sched/sch_codel.c | 48 ++++++++++++++++++++++++++++++++++++++-----
 1 file changed, 43 insertions(+), 5 deletions(-)

index 63c02040b426ae..0dfdc8b6569955 100644
@@ -86,7 +86,7 @@ static void drop_func(struct sk_buff *skb, void *ctx)
 	qdisc_qstats_drop(sch);
 }
 
-static struct sk_buff *codel_qdisc_dequeue(struct Qdisc *sch)
+static struct sk_buff *__codel_qdisc_dequeue(struct Qdisc *sch)
 {
 	struct codel_sched_data *q = qdisc_priv(sch);
 	struct sk_buff *skb;

@@ -95,13 +95,51 @@ static struct sk_buff *codel_qdisc_dequeue(struct Qdisc *sch)
 			    &q->stats, qdisc_pkt_len, codel_get_enqueue_time,
 			    drop_func, dequeue_func);
 
+	if (skb)
+		qdisc_bstats_update(sch, skb);
+	return skb;
+}
+
+static void codel_dequeue_drop(struct Qdisc *sch)
+{
+	struct codel_sched_data *q = qdisc_priv(sch);
+
 	if (q->stats.drop_count) {
-		qdisc_tree_reduce_backlog(sch, q->stats.drop_count, q->stats.drop_len);
+		qdisc_tree_reduce_backlog(sch, q->stats.drop_count,
+					  q->stats.drop_len);
 		q->stats.drop_count = 0;
 		q->stats.drop_len = 0;
 	}
-	if (skb)
-		qdisc_bstats_update(sch, skb);
+}
+
+static struct sk_buff *codel_qdisc_dequeue(struct Qdisc *sch)
+{
+	struct sk_buff *skb;
+
+	skb = __codel_qdisc_dequeue(sch);
+
+	codel_dequeue_drop(sch);
+
+	return skb;
+}
+
+static struct sk_buff *codel_peek(struct Qdisc *sch)
+{
+	struct sk_buff *skb = skb_peek(&sch->gso_skb);
+
+	if (!skb) {
+		skb = __codel_qdisc_dequeue(sch);
+
+		if (skb) {
+			__skb_queue_head(&sch->gso_skb, skb);
+			/* it's still part of the queue */
+			qdisc_qstats_backlog_inc(sch, skb);
+			sch->q.qlen++;
+		}
+
+		codel_dequeue_drop(sch);
+	}
+
 	return skb;
 }
 

@@ -276,7 +314,7 @@ static struct Qdisc_ops codel_qdisc_ops __read_mostly = {
 
 	.enqueue	=	codel_qdisc_enqueue,
 	.dequeue	=	codel_qdisc_dequeue,
-	.peek		=	qdisc_peek_dequeued,
+	.peek		=	codel_peek,
 	.init		=	codel_init,
 	.reset		=	codel_reset,
 	.change 	=	codel_change,

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Aditya Garg <[email protected]>

[ Upstream commit 5985474e1cb4034680fac2145497a94b0860be50 ]

mana_gd_create_mana_wq_cq() leaves queue->id as 0 (from kzalloc_obj()) until mana_create_wq_obj() assigns the firmware-returned id. If creation fails before that, cleanup calls mana_gd_destroy_cq() with id 0, NULLing gc->cq_table[0] and silently breaking whichever real CQ owns that slot.

Initialize queue->id to INVALID_QUEUE_ID right after allocation, matching mana_gd_create_eq(). The existing (id >= max_num_cqs) guard then short-circuits cleanly.

---
 drivers/net/ethernet/microsoft/mana/gdma_main.c | 2 ++
 1 file changed, 2 insertions(+)

index d3c9a3020fbf6e..8a91b056931fa9 100644
@@ -840,6 +840,8 @@ int mana_gd_create_mana_wq_cq(struct gdma_dev *gd,
 	if (!queue)
 		return -ENOMEM;
 
+	queue->id = INVALID_QUEUE_ID;
+
 	gmi = &queue->mem_info;
 	err = mana_gd_alloc_memory(gc, spec->queue_size, gmi);
 	if (err)

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Aditya Garg <[email protected]>

[ Upstream commit f8fd56977eeea3d6939b1a9cd8bd36f1779b3ad0 ]

mana_create_txq() has several error paths (after mana_alloc_queues() or mana_create_wq_obj() failure) where tx_qp[i].tx_object stays as the INVALID_MANA_HANDLE sentinel set at allocation. mana_destroy_txq() then unconditionally calls mana_destroy_wq_obj() with (u64)-1, which firmware rejects and logs an error.

Mirror the RX-side pattern in mana_destroy_rxq() and skip the destroy when the handle is still INVALID_MANA_HANDLE.

---
 drivers/net/ethernet/microsoft/mana/mana_en.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

index 3cf4ad1d91f67e..0249ba7e3b92cd 100644
@@ -1880,7 +1880,8 @@ static void mana_destroy_txq(struct mana_port_context *apc)
 			netif_napi_del(napi);
 			apc->tx_qp[i].txq.napi_initialized = false;
 		}
-		mana_destroy_wq_obj(apc, GDMA_SQ, apc->tx_qp[i].tx_object);
+		if (apc->tx_qp[i].tx_object != INVALID_MANA_HANDLE)
+			mana_destroy_wq_obj(apc, GDMA_SQ, apc->tx_qp[i].tx_object);
 
 		mana_deinit_cq(apc, &apc->tx_qp[i].tx_cq);
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sun Jian <[email protected]>

[ Upstream commit 6001896f00984d317fb75160ba05c4a885fbe2a0 ]

Generic XDP devmap multi redirect uses skb_clone() for intermediate destinations and sends the last destination with the original skb. This can leave multiple destinations sharing the same packet data.

This becomes visible after generic devmap egress-program support was
added: a devmap egress program may mutate packet data, and another destination sharing the same data can observe that mutation.

Native XDP broadcast redirect does not have this issue because xdpf_clone() copies the frame data for each destination. Generic XDP should provide the same per-destination isolation before running a devmap egress program.

Fix this by making cloned skbs private before running the generic devmap egress program. Use skb_copy() instead of skb_unshare() so allocation failure does not consume the skb and the existing caller error paths keep their ownership semantics.

---
 kernel/bpf/devmap.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)

index 3f710d5af55d4b..8ce3b152dcd5b9 100644
@@ -687,6 +687,18 @@ int dev_map_generic_redirect(struct bpf_dtab_netdev *dst, struct sk_buff *skb,
 	if (unlikely(err))
 		return err;
 
+	if (dst->xdp_prog && skb_cloned(skb)) {
+		struct sk_buff *nskb;
+
+		nskb = skb_copy(skb, GFP_ATOMIC);
+		if (!nskb)
+			return -ENOMEM;
+
+		nskb->mac_len = skb->mac_len;
+		consume_skb(skb);
+		skb = nskb;
+	}
+
 	/* Redirect has already succeeded semantically at this point, so we just
 	 * return 0 even if packet is dropped. Helper below takes care of
 	 * freeing skb.

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Dragos Tatulea <[email protected]>

[ Upstream commit d7b0413b35715d7b32cb12d4d424613eff85ed2b ]

The max_macs devlink param is checked against the FW max value only at param register time (driver load) and inside the validate callback (devlink param set). The stored DRIVERINIT value persists across FW resets and devlink reloads without any further checks against the max.

If the FW link type changes from Ethernet to IB and a FW reset happens, the MAX cap for log_max_current_uc_list will become zero, but the previously stored max_macs value remains and is unconditionally programmed into the HCA caps in handle_hca_cap(). FW will then return a syndrome during SET_HCA_CAP:

mlx5_cmd_out_err:839:(pid 3831): SET_HCA_CAP(0x109) op_mod(0x0) failed, status bad parameter(0x3), syndrome (0x537801), err(-22) set_hca_cap:907:(pid 3831): handle_hca_cap failed

This results in a failure to register the RDMA device.

This patch skips programming log_max_current_uc_list when the MAX capability is 0 (in case of IB).

---
 drivers/net/ethernet/mellanox/mlx5/core/main.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

index 80f3493fc544c4..9b86632270936b 100644
@@ -570,7 +570,6 @@ static int handle_hca_cap(struct mlx5_core_dev *dev, void *set_ctx)
 {
 	struct mlx5_profile *prof = &dev->profile;
 	void *set_hca_cap;
-	int max_uc_list;
 	int err;
 
 	err = mlx5_core_get_caps(dev, MLX5_CAP_GENERAL);

@@ -653,10 +652,13 @@ static int handle_hca_cap(struct mlx5_core_dev *dev, void *set_ctx)
 		MLX5_SET(cmd_hca_cap, set_hca_cap, roce,
 			 mlx5_is_roce_on(dev));
 
-	max_uc_list = max_uc_list_get_devlink_param(dev);
-	if (max_uc_list > 0)
-		MLX5_SET(cmd_hca_cap, set_hca_cap, log_max_current_uc_list,
-			 ilog2(max_uc_list));
+	if (MLX5_CAP_GEN_MAX(dev, log_max_current_uc_list)) {
+		int max_uc_list = max_uc_list_get_devlink_param(dev);
+
+		if (max_uc_list > 0)
+			MLX5_SET(cmd_hca_cap, set_hca_cap,
+				 log_max_current_uc_list, ilog2(max_uc_list));
+	}
 
 	return set_caps(dev, set_ctx, MLX5_SET_HCA_CAP_OP_MOD_GENERAL_DEVICE);
 }

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Ruoyu Wang <[email protected]>

[ Upstream commit 05f789fa90d95d5771230e78453cedff2486039d ]

t7xx_port_ctrl_tx() clones each skb fragment before passing it to the port transmit path. The clone is used immediately to set cloned->len, so an skb_clone() failure results in a NULL pointer dereference.

Check the clone before using it. If previous fragments were already queued, preserve the driver's existing partial-write behavior by returning the number of bytes submitted so far.

---
 drivers/net/wwan/t7xx/t7xx_port_wwan.c | 2 ++
 1 file changed, 2 insertions(+)

index 17389c8f6600a3..597a06e86f6ec0 100644
@@ -74,6 +74,8 @@ static int t7xx_port_ctrl_tx(struct wwan_port *port, struct sk_buff *skb)
 
 	while (cur) {
 		cloned = skb_clone(cur, GFP_KERNEL);
+		if (!cloned)
+			return cnt ? cnt : -ENOMEM;
 		cloned->len = skb_headlen(cur);
 		ret = t7xx_port_send_skb(port_private, cloned, 0, 0);
 		if (ret) {

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Ovidiu Panait <[email protected]>

[ Upstream commit fd615abd53110f0f815984e99e7cc51ca6b7d979 ]

Under heavy network traffic, we observed sporadic TX queue timeouts on the Raspberry Pi 4. The timeouts can be reproduced by stress testing the TX path with multiple concurrent iperf UDP streams:

iperf3 -c <ip> -u -b0 -P16 -t60
NETDEV WATCHDOG: CPU: 0: transmit queue 0 timed out 2044 ms NETDEV WATCHDOG: CPU: 3: transmit queue 0 timed out 2004 ms

Investigation showed that the timeouts are caused by the priority-based arbiter. Under heavy load the highest priority queue starves the lower priority ones, causing timeouts. The TX strict priority arbiter is not suitable for the default use case where all the traffic gets spread across all the TX queues.

Therefore, to fix this, switch the TX DMA arbiter to Weighted Round-Robin, which services all queues, so they do not stall. The weights were chosen to follow the existing priority scheme: q0 gets the smallest weight, while q1-4 get the bulk of the TX bandwidth.

---
 .../net/ethernet/broadcom/genet/bcmgenet.c    | 23 +++++++------------
 1 file changed, 8 insertions(+), 15 deletions(-)

index d8e4456d111e04..482f56aed980de 100644
@@ -41,9 +41,8 @@
 
 #include "bcmgenet.h"
 
-/* Default highest priority queue for multi queue support */
-#define GENET_Q1_PRIORITY	0
-#define GENET_Q0_PRIORITY	1
+#define GENET_Q0_WEIGHT		1
+#define GENET_Q1_WEIGHT		4
 
 #define GENET_Q0_RX_BD_CNT	\
 	(TOTAL_DESC - priv->hw_params->rx_queues * priv->hw_params->rx_bds_per_q)

@@ -2086,13 +2085,6 @@ static netdev_tx_t bcmgenet_xmit(struct sk_buff *skb, struct net_device *dev)
 	int i;
 
 	index = skb_get_queue_mapping(skb);
-	/* Mapping strategy:
-	 * queue_mapping = 0, unclassified, packet xmited through ring 0
-	 * queue_mapping = 1, goes to ring 1. (highest priority queue)
-	 * queue_mapping = 2, goes to ring 2.
-	 * queue_mapping = 3, goes to ring 3.
-	 * queue_mapping = 4, goes to ring 4.
-	 */
 	ring = &priv->tx_rings[index];
 	txq = netdev_get_tx_queue(dev, index);
 

@@ -2792,8 +2784,9 @@ static void bcmgenet_fini_tx_napi(struct bcmgenet_priv *priv)
 
 /* Initialize Tx queues
  *
- * Queues 1-4 are priority-based, each one has 32 descriptors,
- * with queue 1 being the highest priority queue.
+ * Queues 1-4 are the priority queues, each one has 32 descriptors.
+ * The weighted round-robin arbiter gives them a larger share of TX
+ * bandwidth than the default queue 0.
  *
  * Queue 0 is the default Tx queue with
  * GENET_Q0_TX_BD_CNT = 256 - 4 * 32 = 128 descriptors.

@@ -2821,8 +2814,8 @@ static void bcmgenet_init_tx_queues(struct net_device *dev)
 	dma_ctrl = 0;
 	ring_cfg = 0;
 
-	/* Enable strict priority arbiter mode */
-	bcmgenet_tdma_writel(priv, DMA_ARBITER_SP, DMA_ARB_CTRL);
+	/* Enable Weighted Round-Robin arbiter mode */
+	bcmgenet_tdma_writel(priv, DMA_ARBITER_WRR, DMA_ARB_CTRL);
 
 	/* Initialize Tx priority queues */
 	for (i = 0; i <= priv->hw_params->tx_queues; i++) {

@@ -2832,7 +2825,7 @@ static void bcmgenet_init_tx_queues(struct net_device *dev)
 		ring_cfg |= (1 << i);
 		dma_ctrl |= (1 << (i + DMA_RING_BUF_EN_SHIFT));
 		dma_priority[DMA_PRIO_REG_INDEX(i)] |=
-			(i ? GENET_Q1_PRIORITY : GENET_Q0_PRIORITY)
+			(i ? GENET_Q1_WEIGHT : GENET_Q0_WEIGHT)
 			<< DMA_PRIO_REG_SHIFT(i);
 	}
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Runyu Xiao <[email protected]>

[ Upstream commit 47186409c092cd7dd70350999186c700233e854d ]

kcm_attach() replaces a live lower TCP socket's sk_data_ready and sk_write_space callbacks with KCM handlers, and kcm_unattach() restores them later. Those callback-pointer updates are still plain stores even though the same fields can be read and invoked concurrently on other CPUs.

If another CPU observes an older callback snapshot after the live field has already been restored, callback execution can run with a mismatched target and sk_user_data state, leading to stale or misdirected wakeups.

Use WRITE_ONCE() for the callback replacement and restore operations so these shared callback fields follow the same visibility contract already established by the earlier 4022 fixes.

---
 net/kcm/kcmsock.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

index 829eb67240a998..a47afaaea2a13c 100644
@@ -1281,8 +1281,8 @@ static int kcm_attach(struct socket *sock, struct socket *csock,
 	psock->save_write_space = csk->sk_write_space;
 	psock->save_state_change = csk->sk_state_change;
 	csk->sk_user_data = psock;
-	csk->sk_data_ready = psock_data_ready;
-	csk->sk_write_space = psock_write_space;
+	WRITE_ONCE(csk->sk_data_ready, psock_data_ready);
+	WRITE_ONCE(csk->sk_write_space, psock_write_space);
 	csk->sk_state_change = psock_state_change;
 
 	write_unlock_bh(&csk->sk_callback_lock);

@@ -1358,8 +1358,8 @@ static void kcm_unattach(struct kcm_psock *psock)
 	 */
 	write_lock_bh(&csk->sk_callback_lock);
 	csk->sk_user_data = NULL;
-	csk->sk_data_ready = psock->save_data_ready;
-	csk->sk_write_space = psock->save_write_space;
+	WRITE_ONCE(csk->sk_data_ready, psock->save_data_ready);
+	WRITE_ONCE(csk->sk_write_space, psock->save_write_space);
 	csk->sk_state_change = psock->save_state_change;
 	strp_stop(&psock->strp);
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Florian Westphal <[email protected]>

[ Upstream commit 64d7d5abe2160bba369b4a8f06bdf5630573bab0 ]

rcu_derefence_raw() should not have been used here, it concealed this bug. Its used because struct rb_node lacks __rcu annotated pointers, so plain rcu_derefence causes sparse warnings.

The major tradeoff is that rcu_derefence_raw() doesn't warn when the caller isn't in a rcu read section.

Extend the rcu read lock scope accordingly and cause sparse warnings, those warnings are the lesser evil.

---
 net/netfilter/nf_conncount.c | 6 +++---
 net/openvswitch/conntrack.c  | 2 +-
 2 files changed, 4 insertions(+), 4 deletions(-)

index ae9ad439449fa5..b9265a8cdb2bc4 100644
@@ -499,7 +499,7 @@ count_tree(struct net *net,
 	hash = jhash2(key, data->keylen, conncount_rnd) % CONNCOUNT_SLOTS;
 	root = &data->root[hash];
 
-	parent = rcu_dereference_raw(root->rb_node);
+	parent = rcu_dereference(root->rb_node);
 	while (parent) {
 		int diff;
 

@@ -507,9 +507,9 @@ count_tree(struct net *net,
 
 		diff = key_diff(key, rbconn->key, data->keylen);
 		if (diff < 0) {
-			parent = rcu_dereference_raw(parent->rb_left);
+			parent = rcu_dereference(parent->rb_left);
 		} else if (diff > 0) {
-			parent = rcu_dereference_raw(parent->rb_right);
+			parent = rcu_dereference(parent->rb_right);
 		} else {
 			int ret;
 

index 32485b36773755..82ffb1ac86b0cb 100644
@@ -1774,10 +1774,10 @@ static int ovs_ct_limit_get_zone_limit(struct net *net,
 		} else {
 			rcu_read_lock();
 			limit = ct_limit_get(info, zone);
-			rcu_read_unlock();
 
 			err = __ovs_ct_limit_get_zone_limit(
 				net, info->data, zone, limit, reply);
+			rcu_read_unlock();
 			if (err)
 				return err;
 		}

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Takashi Iwai <[email protected]>

[ Upstream commit b113a891252c3fa4fab11ec8c2894a22ecaf278c ]

snd_power_ref_and_wait() takes the power refcount and doesn't leave it no matter whether it returns an error or not. However, the majority of callers don't expect but just returns without unreferencing in the caller side upon errors.

For addressing the potential refcount unbalance, rather correct the behavior of snd_power_ref_wait() to unreference upon returning an error.

Note that the problem above is likely negligible; the function returns an error only when the sound card is being shutdown, hence it doesn't matter about the power refcount any longer at such a state.

---
 sound/core/init.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

index 81186af3ac625c..0dc1d25be18404 100644
@@ -1154,7 +1154,7 @@ EXPORT_SYMBOL(snd_card_file_remove);
  * typically around calling control ops.
  *
  * The caller needs to pull down the refcount via snd_power_unref() later
- * no matter whether the error is returned from this function or not.
+ * when this function returns 0.
  *
  * Return: Zero if successful, or a negative error code.
  */

@@ -1167,7 +1167,11 @@ int snd_power_ref_and_wait(struct snd_card *card)
 		       card->shutdown ||
 		       snd_power_get_state(card) == SNDRV_CTL_POWER_D0,
 		       snd_power_unref(card), snd_power_ref(card));
-	return card->shutdown ? -ENODEV : 0;
+	if (card->shutdown) {
+		snd_power_unref(card);
+		return  -ENODEV;
+	}
+	return 0;
 }
 EXPORT_SYMBOL_GPL(snd_power_ref_and_wait);
 

@@ -1184,7 +1188,8 @@ int snd_power_wait(struct snd_card *card)
 	int ret;
 
 	ret = snd_power_ref_and_wait(card);
-	snd_power_unref(card);
+	if (!ret)
+		snd_power_unref(card);
 	return ret;
 }
 EXPORT_SYMBOL(snd_power_wait);

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Jian Zhang <[email protected]>

[ Upstream commit 6d9a4aaaa8b2612b5ef9d581e2f286a458b71ee1 ]

Cifs files may be put into fileinfo_put_wq during umounting cifs. After umount done, cifsFileInfo_put_final is called, which cause following BUG:

BUG: kernel NULL pointer dereference, address: 0000000000000000
...
[  134.222152]  list_lru_add+0x64/0x1a0
[  134.222399]  ? cifs_put_tcon+0x171/0x340 [cifs]
[  134.222772]  d_lru_add+0x44/0x60
[  134.222997]  dput+0x1fc/0x210
[  134.223213]  cifsFileInfo_put_final+0x11a/0x140 [cifs]
[  134.223576]  process_one_work+0x17c/0x320
[  134.223843]  worker_thread+0x188/0x280
[  134.224084]  ? __pfx_worker_thread+0x10/0x10
[  134.224366]  kthread+0xcc/0x100
[  134.224576]  ? __pfx_kthread+0x10/0x10
[  134.224827]  ret_from_fork+0x30/0x50
[  134.225063]  ? __pfx_kthread+0x10/0x10
[  134.225328]  ret_from_fork_asm+0x1b/0x30

This can be reproduce by following:
unshare -n bash -c "
mkdir -p ${CIFS_MNT}
ip netns attach root 1
ip link add eth0 type veth peer veth0 netns root
ip link set eth0 up
ip -n root link set veth0 up
ip addr add 192.168.0.2/24 dev eth0
ip -n root addr add 192.168.0.1/24 dev veth0
ip route add default via 192.168.0.1 dev eth0
ip netns exec root sysctl net.ipv4.ip_forward=1
ip netns exec root iptables -t nat -A POSTROUTING -s 192.168.0.2 -o
${DEV} -j MASQUERADE
mount -t cifs ${CIFS_PATH} ${CIFS_MNT} -o vers=3.0,sec=ntlmssp,credentials=${CIFS_CRED},rsize=65536,wsize=65536,cache=none,echo_interval=1 touch ${CIFS_MNT}/a.txt
ip netns exec root iptables -t nat -D POSTROUTING -s 192.168.0.2 -o
${DEV} -j MASQUERADE
"
umount ${CIFS_MNT}

---
 fs/smb/client/connect.c | 3 +++
 1 file changed, 3 insertions(+)

index 78feea9c262ef0..fe5d21f174c47e 100644
@@ -3968,6 +3968,9 @@ cifs_umount(struct cifs_sb_info *cifs_sb)
 	}
 	spin_unlock(&cifs_sb->tlink_tree_lock);
 
+	flush_workqueue(serverclose_wq);
+	flush_workqueue(fileinfo_put_wq);
+
 	kfree(cifs_sb->prepath);
 	call_rcu(&cifs_sb->rcu, delayed_free);
 }

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Huiwen He <[email protected]>

[ Upstream commit 7acbaa16b99edaf8ef432229d4b7a6f3b666767d ]

Currently, repeated lsattr calls on a regular CIFS file without the compressed attribute may show random flags:

	$ touch test.bin
	$ lsattr test.bin
	s-S-ia-A-EjI---------m test.bin
	$ lsattr test.bin
	------d-cEjI---------m test.bin

The lsattr reproducer depends on the previous contents of its userspace buffer, so it may not reproduce on every setup. A deterministic reproducer is to initialize the ioctl argument before FS_IOC_GETFLAGS on a file without the compressed attribute:

        int flags = 0x7fffffff;
        ioctl(fd, FS_IOC_GETFLAGS, &flags);

On an affected kernel, flags remains 0x7fffffff. With the fix, it is set to 0.

This happens because when the cached inode does not have the compressed bit set, the CIFS fallback path in FS_IOC_GETFLAGS returns success without calling put_user() to write the zero flags value into the user buffer. As a result, the caller observes stale contents from its own buffer.

Fix this by always writing the visible flags value back to the user buffer before returning success, even when the value is zero.

---
 fs/smb/client/ioctl.c | 12 +++++-------
 1 file changed, 5 insertions(+), 7 deletions(-)

index 8ea532cdc4302b..8a668b827570a4 100644
@@ -376,13 +376,11 @@ long cifs_ioctl(struct file *filep, unsigned int command, unsigned long arg)
 			}
 #endif /* CONFIG_CIFS_ALLOW_INSECURE_LEGACY */
 #endif /* CONFIG_CIFS_POSIX */
-			rc = 0;
-			if (CIFS_I(inode)->cifsAttrs & ATTR_COMPRESSED) {
-				/* add in the compressed bit */
-				ExtAttrBits = FS_COMPR_FL;
-				rc = put_user(ExtAttrBits & FS_FL_USER_VISIBLE,
-					      (int __user *)arg);
-			}
+			if (CIFS_I(inode)->cifsAttrs & FILE_ATTRIBUTE_COMPRESSED)
+				ExtAttrBits |= FS_COMPR_FL;
+
+			rc = put_user(ExtAttrBits & FS_FL_USER_VISIBLE,
+				      (int __user *)arg);
 			break;
 		case FS_IOC_SETFLAGS:
 			if (pSMBFile == NULL)

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Leo Yan <[email protected]>

[ Upstream commit 55ffbe8a15b1254f44d56952fb425a10e3f15c31 ]

ASAN reports stack-buffer-overflow due to the uninitialized op_name.

Initialize it to fix the issue.

---
 tools/testing/selftests/bpf/prog_tests/verifier_log.c | 1 +
 1 file changed, 1 insertion(+)

index 8337c6bc5b95b6..abc473c3741f53 100644
@@ -317,6 +317,7 @@ static void verif_btf_log_subtest(bool bad_btf)
 	res = load_btf(&opts, true);
 	ASSERT_EQ(res, -ENOSPC, "half_log_fd");
 	ASSERT_EQ(strlen(logs.buf), 24, "log_fixed_25");
+	strscpy(op_name, "log_fixed", sizeof(op_name));
 	ASSERT_STRNEQ(logs.buf, logs.reference, 24, op_name);
 
 	/* validate rolling verifier log logic: try all variations of log buf

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Leon Hwang <[email protected]>

[ Upstream commit ca0f587c029afa66227f7b932450b1c417403394 ]

When TCP over IPv4 via INET6 API, bpf_get/setsockopt with ipv4 will fail, because sk->sk_family is AF_INET6. With ipv6 will success, not take effect, because inet_csk(sk)->icsk_af_ops is ipv6_mapped and use ip_queue_xmit, inet_sk(sk)->tos.

To relax this restriction, allow getting/setting tos for those possible ipv4-mapped ipv6 sockets.

---
 net/core/filter.c | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)

index 4a9c1367ea4d5e..5f99b2144c257f 100644
@@ -5455,11 +5455,24 @@ static int sol_tcp_sockopt(struct sock *sk, int optname,
 				 KERNEL_SOCKPTR(optval), *optlen);
 }
 
+static bool sk_allows_sol_ip_sockopt(struct sock *sk)
+{
+	switch (sk->sk_family) {
+	case AF_INET:
+		return true;
+	case AF_INET6:
+		/* Allow getting/setting sockopt for possible ipv4-mapped ipv6 socket. */
+		return sk->sk_type != SOCK_RAW && !ipv6_only_sock(sk);
+	default:
+		return false;
+	}
+}
+
 static int sol_ip_sockopt(struct sock *sk, int optname,
 			  char *optval, int *optlen,
 			  bool getopt)
 {
-	if (sk->sk_family != AF_INET)
+	if (!sk_allows_sol_ip_sockopt(sk))
 		return -EINVAL;
 
 	switch (optname) {

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Weiming Shi <[email protected]>

[ Upstream commit 0c0a8ed85349dae298712d79cb276acfeb794d82 ]

When the scatterlist ring is full or nearly full, bpf_msg_push_data() enters a copy fallback path and computes copy + len for the page allocation size. Since len comes from BPF with arg3_type = ARG_ANYTHING and both are u32, a crafted len can wrap the sum to a small value, causing an undersized allocation followed by an out-of-bounds memcpy.

 BUG: unable to handle page fault for address: ffffed104089a402
 Oops: Oops: 0000 [#1] SMP KASAN NOPTI
 Call Trace:
  __asan_memcpy (mm/kasan/shadow.c:105)
  bpf_msg_push_data (net/core/filter.c:2852 net/core/filter.c:2788)
  bpf_prog_9ed8b5711920a7d7+0x2e/0x36
  sk_psock_msg_verdict (net/core/skmsg.c:934)
  tcp_bpf_sendmsg (net/ipv4/tcp_bpf.c:421 net/ipv4/tcp_bpf.c:584)
  __sys_sendto (net/socket.c:2206)
  do_syscall_64 (arch/x86/entry/syscall_64.c:94)
  entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:130)

Add an overflow check before the allocation.

---
 net/core/filter.c | 3 +++
 1 file changed, 3 insertions(+)

index 5f99b2144c257f..be51e8281109ac 100644
@@ -2822,6 +2822,9 @@ BPF_CALL_4(bpf_msg_push_data, struct sk_msg *, msg, u32, start,
 	if (!space || (space == 1 && start != offset))
 		copy = msg->sg.data[i].length;
 
+	if (unlikely(copy + len < copy))
+		return -EINVAL;
+
 	page = alloc_pages(__GFP_NOWARN | GFP_ATOMIC | __GFP_COMP,
 			   get_order(copy + len));
 	if (unlikely(!page))

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sechang Lim <[email protected]>

[ Upstream commit a48802fb2cd2d1e23651989f8ff4d15e9d5dad54 ]

start and len are u32, so

	u64 last = start + len;

evaluates start + len in 32-bit and wraps before storing it in last. The bounds check

	if (start >= offset + l || last > msg->sg.size)
		return -EINVAL;

can then be passed with an out-of-range start/len, after which the pop loop runs off the end of the scatterlist and sk_msg_shift_left() calls put_page() on the empty msg->sg.end slot:

  Oops: general protection fault, probably for non-canonical address
  0xdffffc0000000001: 0000 [#1] SMP KASAN PTI
  KASAN: null-ptr-deref in range [0x0000000000000008-0x000000000000000f]
  RIP: 0010:sk_msg_shift_left net/core/filter.c:2957 [inline]
  RIP: 0010:____bpf_msg_pop_data net/core/filter.c:3103 [inline]
  RIP: 0010:bpf_msg_pop_data+0x753/0x1a10 net/core/filter.c:2984
  Call Trace:
   <TASK>
   bpf_prog_4cc92c278f4d5d56+0x1b1/0x1e8
   bpf_prog_run_pin_on_cpu+0x107/0x320 include/linux/filter.h:746
   sk_psock_msg_verdict+0x357/0x7f0 net/core/skmsg.c:934
   tcp_bpf_send_verdict net/ipv4/tcp_bpf.c:420 [inline]
   tcp_bpf_sendmsg+0x766/0x1ae0 net/ipv4/tcp_bpf.c:583
   __sock_sendmsg+0x153/0x1c0 net/socket.c:802
   __sys_sendto+0x326/0x430 net/socket.c:2265
   __x64_sys_sendto+0xe3/0x100 net/socket.c:2268
   do_syscall_64+0x14c/0x480
   entry_SYSCALL_64_after_hwframe+0x77/0x7f
   </TASK>

Widen the addition with a (u64) cast so the bound is evaluated in 64-bit and a len near U32_MAX no longer wraps below msg->sg.size.

While here, change pop from int to u32. It counts bytes against the unsigned scatterlist lengths and can never be negative, so the signed type only invites sign-confusion in the pop loop.

---
 net/core/filter.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

index be51e8281109ac..3e01ce9f01dfd0 100644
@@ -2988,8 +2988,8 @@ BPF_CALL_4(bpf_msg_pop_data, struct sk_msg *, msg, u32, start,
 	   u32, len, u64, flags)
 {
 	u32 i = 0, l = 0, space, offset = 0;
-	u64 last = start + len;
-	int pop;
+	u64 last = (u64)start + len;
+	u32 pop;
 
 	if (unlikely(flags))
 		return -EINVAL;

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Yadan Fan <[email protected]>

[ Upstream commit 1b001b16bc88f3f7817e228acfd91ee01bdcfcce ]

maar_res_walk() uses wi->num_cfg as the index into the fixed-size wi->cfg array, but checks whether the array is full only after it has filled the selected entry. If walk_system_ram_range() reports more than 16 memory ranges, the overflow call writes one struct maar_config past the end of the array before WARN_ON() prevents num_cfg from advancing.

Move the full-array check before taking the array slot and return non-zero when the scratch array is full, so walk_system_ram_range() terminates the walk instead of invoking the callback for further ranges.

---
 arch/mips/mm/init.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

index 6e368a4658b544..de3713b42cbc2d 100644
@@ -281,9 +281,15 @@ static int maar_res_walk(unsigned long start_pfn, unsigned long nr_pages,
 			 void *data)
 {
 	struct maar_walk_info *wi = data;
-	struct maar_config *cfg = &wi->cfg[wi->num_cfg];
+	struct maar_config *cfg;
 	unsigned int maar_align;
 
+	/* Ensure we don't overflow the cfg array */
+	if (WARN_ON(wi->num_cfg >= ARRAY_SIZE(wi->cfg)))
+		return -1;
+
+	cfg = &wi->cfg[wi->num_cfg];
+
 	/* MAAR registers hold physical addresses right shifted by 4 bits */
 	maar_align = BIT(MIPS_MAAR_ADDR_SHIFT + 4);
 

@@ -292,9 +298,7 @@ static int maar_res_walk(unsigned long start_pfn, unsigned long nr_pages,
 	cfg->upper = ALIGN_DOWN(PFN_PHYS(start_pfn + nr_pages), maar_align) - 1;
 	cfg->attrs = MIPS_MAAR_S;
 
-	/* Ensure we don't overflow the cfg array */
-	if (!WARN_ON(wi->num_cfg >= ARRAY_SIZE(wi->cfg)))
-		wi->num_cfg++;
+	wi->num_cfg++;
 
 	return 0;
 }

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Aboorva Devarajan <[email protected]>

[ Upstream commit 81e3a86030462824a67d697739cf3f387f4ba350 ]

fsl_emb_pmu_del() unconditionally calls put_cpu_var(cpu_hw_events) at the 'out:' label, but only calls the matching get_cpu_var() after the 'i < 0' early-return check. When event->hw.idx is negative the function jumps to 'out:' without having taken get_cpu_var(), and the trailing put_cpu_var() then issues an unmatched preempt_enable(), underflowing preempt_count.

On a CONFIG_PREEMPT=y kernel preempt_count would underflow and eventually present as a 'scheduling while atomic' BUG.

Move put_cpu_var() to pair with get_cpu_var() so the percpu access is correctly bracketed and the 'out:' label only handles perf_pmu_enable.

---
 arch/powerpc/perf/core-fsl-emb.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

index 1a53ab08447cbe..31037df4178ca2 100644
@@ -366,9 +366,10 @@ static void fsl_emb_pmu_del(struct perf_event *event, int flags)
 
 	cpuhw->n_events--;
 
+	put_cpu_var(cpu_hw_events);
+
  out:
 	perf_pmu_enable(event->pmu);
-	put_cpu_var(cpu_hw_events);
 }
 
 static void fsl_emb_pmu_start(struct perf_event *event, int ef_flags)

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Aboorva Devarajan <[email protected]>

[ Upstream commit 0ecd26e93e698c8327521910fc6296f5b84a4b92 ]

pnv_kexec_wait_secondaries_down() calls get_cpu() to obtain the current CPU id but never calls the matching put_cpu(), leaking one preempt_disable() nesting level on every invocation.

In practice the imbalance does not trigger a visible splat because the kexec teardown path is a one-way trip: IRQs are already disabled, no schedule() occurs after the leak, and default_machine_kexec() overwrites preempt_count with HARDIRQ_OFFSET before jumping into kexec_sequence() which never returns. However the bookkeeping is still wrong.

The function only needs the current CPU id, and this path runs with interrupts disabled and the CPU pinned, so the preempt_disable() side-effect of get_cpu() is unnecessary. Replace it with raw_smp_processor_id().

---
 arch/powerpc/platforms/powernv/setup.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

index 4dbb47ddbdcc4b..06ed5e2aa26584 100644
@@ -396,7 +396,8 @@ static void pnv_kexec_wait_secondaries_down(void)
 {
 	int my_cpu, i, notified = -1;
 
-	my_cpu = get_cpu();
+	/* Called with interrupts disabled, so the CPU is pinned. */
+	my_cpu = raw_smp_processor_id();
 
 	for_each_online_cpu(i) {
 		uint8_t status;

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Aboorva Devarajan <[email protected]>

[ Upstream commit 5c86f1c1f972761a04bf22f4c0618d1aa714185b ]

kexec_prepare_cpus_wait() calls get_cpu() internally to obtain the current CPU id. kexec_prepare_cpus() calls kexec_prepare_cpus_wait() twice -- once for KEXEC_STATE_IRQS_OFF and once for KEXEC_STATE_REAL_MODE -- but only issues a single put_cpu() at the end, leaving preempt_count elevated by one extra nesting level.

In practice the imbalance does not trigger a 'scheduling while atomic' splat because the kexec path is a one-way trip: IRQs are already disabled, no schedule() occurs after the leak, and default_machine_kexec() overwrites preempt_count with HARDIRQ_OFFSET before jumping into kexec_sequence() which never returns. However the bookkeeping is still wrong.

kexec_prepare_cpus() calls local_irq_disable()/hard_irq_disable() before invoking kexec_prepare_cpus_wait(), so the CPU is already pinned and the get_cpu()/put_cpu() preempt_disable() bracketing is unnecessary. Only the current CPU id is needed, so replace get_cpu() with raw_smp_processor_id() and drop the now-unneeded put_cpu().

---
 arch/powerpc/kexec/core_64.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

index ec7e909d7dcb7e..f8ae6c035aa103 100644
@@ -167,7 +167,7 @@ static void kexec_prepare_cpus_wait(int wait_state)
 	int my_cpu, i, notified=-1;
 
 	hw_breakpoint_disable();
-	my_cpu = get_cpu();
+	my_cpu = raw_smp_processor_id();
 	/* Make sure each CPU has at least made it to the state we need.
 	 *
 	 * FIXME: There is a (slim) chance of a problem if not all of the CPUs

@@ -265,8 +265,6 @@ static void kexec_prepare_cpus(void)
 	/* after we tell the others to go down */
 	if (ppc_md.kexec_cpu_down)
 		ppc_md.kexec_cpu_down(0, 0);
-
-	put_cpu();
 }
 
 #else /* ! SMP */

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Gui-Dong Han <[email protected]>

[ Upstream commit c1201b37f666f6466ab1fd3a381c2b7a4b7e9fee ]

The negative-key race fix added release/acquire ordering for key use.

Publish payload before state; read state before payload.

keyring_search_iterator() still uses READ_ONCE() before match callbacks. An asymmetric match callback calls asymmetric_key_ids(), which reads key->payload.data[asym_key_ids].

Use key_read_state() there to complete that ordering.

---
 security/keys/keyring.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index df2580072cfe16..e105349794f23e 100644
@@ -576,7 +576,7 @@ static int keyring_search_iterator(const void *object, void *iterator_data)
 	struct keyring_search_context *ctx = iterator_data;
 	const struct key *key = keyring_ptr_to_key(object);
 	unsigned long kflags = READ_ONCE(key->flags);
-	short state = READ_ONCE(key->state);
+	short state = key_read_state(key);
 
 	kenter("{%d}", key->serial);
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Eric Dumazet <[email protected]>

[ Upstream commit f4c3d89fc986b0da196ddfc6cfe0ea5d5d08bec6 ]

Syzbot reported a slab-use-after-free in ipvlan_hard_header() when called from tipc_l2_send_msg().

The root cause is that tipc_disable_l2_media() calls synchronize_net() while b->media_ptr is still valid. This allows concurrent RCU readers to obtain the device pointer after synchronize_net() has finished. The pointer is cleared later in bearer_disable(), but without any subsequent synchronization, allowing the device to be freed while still in use by readers.

Fix this by clearing b->media_ptr in tipc_disable_l2_media() before calling synchronize_net().

This is safe to do now because the call order in bearer_disable() was reversed in 0d051bf93c06 ("tipc: make bearer packet filtering generic") to call tipc_node_delete_links() (which needs the pointer) before disable_media().

https: //lore.kernel.org/netdev/[email protected]/T/#u

---
 net/tipc/bearer.c | 1 +
 1 file changed, 1 insertion(+)

index fec638e494c9dd..ae57ac1174f812 100644
@@ -482,6 +482,7 @@ void tipc_disable_l2_media(struct tipc_bearer *b)
 	dev = (struct net_device *)rtnl_dereference(b->media_ptr);
 	dev_remove_pack(&b->pt);
 	RCU_INIT_POINTER(dev->tipc_ptr, NULL);
+	RCU_INIT_POINTER(b->media_ptr, NULL);
 	synchronize_net();
 	dev_put(dev);
 }

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Eric Dumazet <[email protected]>

[ Upstream commit 2bf43d0e2e6a27d52a7d624e2d6b9116972e8a22 ]

When MTU is large, ip6_default_advmss() can return IPV6_MAXPLEN (65535). This is interpreted by TCP as mss_clamp, allowing the MSS to reach 65535.

However, 0xFFFF is also used as a magic value GSO_BY_FRAGS in the kernel. If a TCP packet with gso_size=0xFFFF is passed to skb_segment(), it will be mistakenly treated as GSO_BY_FRAGS, leading to a NULL pointer dereference because local TCP packets do not use frag_list.

Fix this by returning min(IPV6_MAXPLEN, GSO_BY_FRAGS - 1) (65534) from ip6_default_advmss() when MTU is large.

Also update the stale comment in ip6_default_advmss() which suggested that IPV6_MAXPLEN is returned to mean "any MSS".

---
 net/ipv6/route.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

index 6d80e17c04c0d8..c3257e8d575485 100644
@@ -3251,11 +3251,11 @@ static unsigned int ip6_default_advmss(const struct dst_entry *dst)
 	/*
 	 * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
 	 * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
-	 * IPV6_MAXPLEN is also valid and means: "any MSS,
-	 * rely only on pmtu discovery"
+	 * Limit the default MSS to GSO_BY_FRAGS - 1 to avoid
+	 * collision with the GSO_BY_FRAGS magic value (0xFFFF).
 	 */
 	if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
-		mtu = IPV6_MAXPLEN;
+		mtu = min_t(unsigned int, IPV6_MAXPLEN, GSO_BY_FRAGS - 1);
 	return mtu;
 }
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Brett Creeley <[email protected]>

[ Upstream commit 7678e69079c10b2fb10977f28f44ddb22971ea5b ]

The current check will fail if SR-IOV is not initialized for the physical function; this is because is_physfn is 0 if sriov_init() isn't run or fails. Change the check that prevents getting the link down count to use is_virtfn instead so that VFs don't get this functionality, which was the original intent.

---
 drivers/net/ethernet/pensando/ionic/ionic_ethtool.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

index b07dd56b0c76d2..38c2695bc529ef 100644
@@ -114,8 +114,15 @@ static void ionic_get_link_ext_stats(struct net_device *netdev,
 {
 	struct ionic_lif *lif = netdev_priv(netdev);
 
-	if (lif->ionic->pdev->is_physfn)
-		stats->link_down_events = lif->link_down_count;
+	if (lif->ionic->pdev->is_virtfn)
+		return;
+
+	if (!lif->ionic->idev.port_info) {
+		netdev_err_once(netdev, "port_info not initialized\n");
+		return;
+	}
+
+	stats->link_down_events = lif->link_down_count;
 }
 
 static int ionic_get_link_ksettings(struct net_device *netdev,

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Guangshuo Li <[email protected]>

[ Upstream commit 65b655f65c3ca1ab5d598d3832bb0ff531725858 ]

same_client_has_lease() returns an opinfo pointer from ci->m_op_list after dropping ci->m_lock without taking a reference.

smb_grant_oplock() then dereferences that pointer in copy_lease() and when checking breaking_cnt. A concurrent close can remove the old lease from ci->m_op_list and drop the last reference before the caller uses the returned pointer, leading to a use-after-free.

Take a reference when same_client_has_lease() selects an existing lease, drop any previous match while scanning, and release the returned reference in smb_grant_oplock() after copying the lease state.

---
 fs/smb/server/oplock.c | 6 ++++++
 1 file changed, 6 insertions(+)

index bba17ea84023e2..1d2e06e34a9a35 100644
@@ -531,7 +531,12 @@ static struct oplock_info *same_client_has_lease(struct ksmbd_inode *ci,
 
 		ret = compare_guid_key(opinfo, client_guid, lctx->lease_key);
 		if (ret) {
+			if (!atomic_inc_not_zero(&opinfo->refcount))
+				continue;
+			if (m_opinfo)
+				opinfo_put(m_opinfo);
 			m_opinfo = opinfo;
+
 			/* skip upgrading lease about breaking lease */
 			if (atomic_read(&opinfo->breaking_cnt))
 				continue;

@@ -1249,6 +1254,7 @@ int smb_grant_oplock(struct ksmbd_work *work, int req_op_level, u64 pid,
 			if (atomic_read(&m_opinfo->breaking_cnt))
 				opinfo->o_lease->flags =
 					SMB2_LEASE_FLAG_BREAK_IN_PROGRESS_LE;
+			opinfo_put(m_opinfo);
 			goto out;
 		}
 	}

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Charles Keepax <[email protected]>

[ Upstream commit b6ef1a74b3ec254f87a6a3c554fe8f8083ebd37c ]

Currently the code checks if a firmware was received, however it does not verify that the firmware size is larger than the firmware header. As the firmware pointer is dereferenced as a pointer to the header structure this could lead to an out of bounds memory access. Add the missing check.

---
 drivers/mfd/cs42l43.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 1cea3f8f467d43..c1f7b2cdb45b67 100644
@@ -666,7 +666,7 @@ static void cs42l43_mcu_load_firmware(const struct firmware *firmware, void *con
 	unsigned int loadaddr, val;
 	int ret;
 
-	if (!firmware) {
+	if (!firmware || firmware->size < sizeof(*hdr)) {
 		dev_err(cs42l43->dev, "Failed to load firmware\n");
 		cs42l43->firmware_error = -ENODEV;
 		goto err;

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Aleksandr Nogikh <[email protected]>

[ Upstream commit ff6f26c58421614b02694ac9d219ac61d924bc68 ]

A circular locking dependency involves INODE_ALLOC_SYSTEM_INODE, EXTENT_ALLOC_SYSTEM_INODE, and ORPHAN_DIR_SYSTEM_INODE.

1. ocfs2_mknod() acquires INODE_ALLOC then EXTENT_ALLOC.

2. ocfs2_dio_end_io_write() acquires EXTENT_ALLOC for unwritten extents, then ORPHAN_DIR via ocfs2_del_inode_from_orphan() while still holding EXTENT_ALLOC.

3. ocfs2_wipe_inode() acquires ORPHAN_DIR then INODE_ALLOC via ocfs2_remove_inode.

Break the cycle in ocfs2_dio_end_io_write() by freeing the allocation contexts (releasing EXTENT_ALLOC) before acquiring ORPHAN_DIR.

WARNING: possible circular locking dependency detected
------------------------------------------------------
is trying to acquire lock:
ffff8881e78b33a0 (&ocfs2_sysfile_lock_key[INODE_ALLOC_SYSTEM_INODE]){+.+.}-{4:4}, at:
ocfs2_evict_inode+0x1539/0x43b0 fs/ocfs2/inode.c:1299

but task is already holding lock:
ffff8881e78b4fa0 (&ocfs2_sysfile_lock_key[ORPHAN_DIR_SYSTEM_INODE]){+.+.}-{4:4}, at:
ocfs2_evict_inode+0xe97/0x43b0 fs/ocfs2/inode.c:1299

the existing dependency chain (in reverse order) is:

-> #2 (&ocfs2_sysfile_lock_key[ORPHAN_DIR_SYSTEM_INODE]){+.+.}-{4:4}:

       inode_lock include/linux/fs.h:1029 [inline]
       ocfs2_del_inode_from_orphan+0x12e/0x7a0 fs/ocfs2/namei.c:2728
       ocfs2_dio_end_io+0xf9c/0x1370 fs/ocfs2/aops.c:2418
       dio_complete+0x25b/0x790 fs/direct-io.c:281

-> #1 (&ocfs2_sysfile_lock_key[EXTENT_ALLOC_SYSTEM_INODE]){+.+.}-{4:4}:

       inode_lock include/linux/fs.h:1029 [inline]
       ocfs2_reserve_suballoc_bits+0x16d/0x4840 fs/ocfs2/suballoc.c:882
       ocfs2_reserve_new_metadata_blocks+0x415/0x9a0
       fs/ocfs2/suballoc.c:1078
       ocfs2_mknod+0x10f3/0x2260 fs/ocfs2/namei.c:351

-> #0 (&ocfs2_sysfile_lock_key[INODE_ALLOC_SYSTEM_INODE]){+.+.}-{4:4}:

       __lock_acquire+0x15a5/0x2cf0 kernel/locking/lockdep.c:5237
       lock_acquire+0x106/0x350 kernel/locking/lockdep.c:5868
       down_write+0x96/0x200 kernel/locking/rwsem.c:1625
       inode_lock include/linux/fs.h:1029 [inline]
       ocfs2_remove_inode fs/ocfs2/inode.c:733 [inline]
       ocfs2_wipe_inode fs/ocfs2/inode.c:896 [inline]
       ocfs2_delete_inode fs/ocfs2/inode.c:1157 [inline]
       ocfs2_evict_inode+0x1539/0x43b0 fs/ocfs2/inode.c:1299

Chain exists of:
  &ocfs2_sysfile_lock_key[INODE_ALLOC_SYSTEM_INODE] -->
  &ocfs2_sysfile_lock_key[EXTENT_ALLOC_SYSTEM_INODE] -->
  &ocfs2_sysfile_lock_key[ORPHAN_DIR_SYSTEM_INODE]

 Possible unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  lock(&ocfs2_sysfile_lock_key[ORPHAN_DIR_SYSTEM_INODE]);
                               lock(&ocfs2_sysfile_lock_key[EXTENT_ALLOC_SYSTEM_INODE]);
                               lock(&ocfs2_sysfile_lock_key[ORPHAN_DIR_SYSTEM_INODE]);

lock(&ocfs2_sysfile_lock_key[INODE_ALLOC_SYSTEM_INODE]);

*** DEADLOCK ***

---
 fs/ocfs2/aops.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

index 5365fc7675de46..680d7fa7e1482e 100644
@@ -2400,6 +2400,15 @@ static int ocfs2_dio_end_io_write(struct inode *inode,
 unlock:
 	up_write(&oi->ip_alloc_sem);
 
+	if (data_ac) {
+		ocfs2_free_alloc_context(data_ac);
+		data_ac = NULL;
+	}
+	if (meta_ac) {
+		ocfs2_free_alloc_context(meta_ac);
+		meta_ac = NULL;
+	}
+
 	/* everything looks good, let's start the cleanup */
 	if (!ret && dwc->dw_orphaned) {
 		BUG_ON(dwc->dw_writer_pid != task_pid_nr(current));

@@ -2411,10 +2420,6 @@ static int ocfs2_dio_end_io_write(struct inode *inode,
 	ocfs2_inode_unlock(inode, 1);
 	brelse(di_bh);
 out:
-	if (data_ac)
-		ocfs2_free_alloc_context(data_ac);
-	if (meta_ac)
-		ocfs2_free_alloc_context(meta_ac);
 	ocfs2_run_deallocs(osb, &dealloc);
 	ocfs2_dio_free_write_ctx(inode, dwc);
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Yizhou Zhao <[email protected]>

[ Upstream commit 7d54894a1ee265a72d70f7cae1da6cc774cccc71 ]

The rdma->state field is modified without holding req_lock in both recv_done() and p9_cm_event_handler(), while rdma_request() accesses the same field under the req_lock spinlock. This inconsistent locking creates a race condition:

- recv_done() running in softirq completion context sets rdma->state = P9_RDMA_FLUSHING without acquiring req_lock

- p9_cm_event_handler() modifies rdma->state at multiple points (ADDR_RESOLVED, ROUTE_RESOLVED, ESTABLISHED, CLOSED) without req_lock

- rdma_request() uses spin_lock_irqsave(&rdma->req_lock, flags) to protect the read-modify-write of rdma->state

The race can cause lost state transitions: recv_done() or the CM event handler could set state to FLUSHING/CLOSED while rdma_request() is concurrently checking or modifying state under the lock, leading to the FLUSHING transition being silently overwritten by CLOSING. This corrupts the connection state machine and can cause use-after-free on RDMA request objects during teardown.

Fix by adding req_lock protection to all rdma->state modifications in recv_done() and p9_cm_event_handler(), matching the pattern already used in rdma_request(). Use spin_lock_irqsave/spin_unlock_irqrestore in the CM event handler since it can race with recv_done() which runs in softirq context.

Tested with a kernel module that races two threads (simulating rdma_request and recv_done/CM handler) on rdma->state with proper
locking: 5.5M+ FLUSHING writes over 27M iterations with 0 lost transitions.

Message-ID: <[email protected]>

---
 net/9p/trans_rdma.c | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

index b84748baf9cbe3..d9f53f1da2c21a 100644
@@ -240,25 +240,36 @@ p9_cm_event_handler(struct rdma_cm_id *id, struct rdma_cm_event *event)
 {
 	struct p9_client *c = id->context;
 	struct p9_trans_rdma *rdma = c->trans;
+	unsigned long flags;
+
 	switch (event->event) {
 	case RDMA_CM_EVENT_ADDR_RESOLVED:
+		spin_lock_irqsave(&rdma->req_lock, flags);
 		BUG_ON(rdma->state != P9_RDMA_INIT);
 		rdma->state = P9_RDMA_ADDR_RESOLVED;
+		spin_unlock_irqrestore(&rdma->req_lock, flags);
 		break;
 
 	case RDMA_CM_EVENT_ROUTE_RESOLVED:
+		spin_lock_irqsave(&rdma->req_lock, flags);
 		BUG_ON(rdma->state != P9_RDMA_ADDR_RESOLVED);
 		rdma->state = P9_RDMA_ROUTE_RESOLVED;
+		spin_unlock_irqrestore(&rdma->req_lock, flags);
 		break;
 
 	case RDMA_CM_EVENT_ESTABLISHED:
+		spin_lock_irqsave(&rdma->req_lock, flags);
 		BUG_ON(rdma->state != P9_RDMA_ROUTE_RESOLVED);
 		rdma->state = P9_RDMA_CONNECTED;
+		spin_unlock_irqrestore(&rdma->req_lock, flags);
 		break;
 
 	case RDMA_CM_EVENT_DISCONNECTED:
-		if (rdma)
+		if (rdma) {
+			spin_lock_irqsave(&rdma->req_lock, flags);
 			rdma->state = P9_RDMA_CLOSED;
+			spin_unlock_irqrestore(&rdma->req_lock, flags);
+		}
 		c->status = Disconnected;
 		break;
 

@@ -296,6 +307,7 @@ recv_done(struct ib_cq *cq, struct ib_wc *wc)
 	struct p9_req_t *req;
 	int err = 0;
 	int16_t tag;
+	unsigned long flags;
 
 	req = NULL;
 	ib_dma_unmap_single(rdma->cm_id->device, c->busa, client->msize,

@@ -332,7 +344,10 @@ recv_done(struct ib_cq *cq, struct ib_wc *wc)
  err_out:
 	p9_debug(P9_DEBUG_ERROR, "req %p err %d status %d\n",
 			req, err, wc->status);
-	rdma->state = P9_RDMA_FLUSHING;
+	spin_lock_irqsave(&rdma->req_lock, flags);
+	if (rdma->state < P9_RDMA_FLUSHING)
+		rdma->state = P9_RDMA_FLUSHING;
+	spin_unlock_irqrestore(&rdma->req_lock, flags);
 	client->status = Disconnected;
 	goto out;
 }

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Alexandru Hossu <[email protected]>

[ Upstream commit 26813881181deb3a32fbb59eadb2599cbe8423f6 ]

In nvec_rx_completed(), when an incomplete RX transfer is detected, nvec_msg_free() is called to return the message back to the pool by clearing its 'used' atomic flag. Immediately after this, the code accesses nvec->rx->data[0] to check the message type.

Since nvec_msg_free() marks the pool slot as available via atomic_set(), any concurrent or subsequent call to nvec_msg_alloc() could claim that same slot and overwrite its data[] array. Reading nvec->rx->data[0] after freeing the message is therefore a use-after-free.

Fix this by saving the message type byte before calling nvec_msg_free(), then using the saved value for the battery quirk check.

---
 drivers/staging/nvec/nvec.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

index 2823cacde13093..0ed7266b033eb5 100644
@@ -489,6 +489,8 @@ static void nvec_tx_completed(struct nvec_chip *nvec)
 static void nvec_rx_completed(struct nvec_chip *nvec)
 {
 	if (nvec->rx->pos != nvec_msg_size(nvec->rx)) {
+		unsigned char msg_type = nvec->rx->data[0];
+
 		dev_err(nvec->dev, "RX incomplete: Expected %u bytes, got %u\n",
 			(uint)nvec_msg_size(nvec->rx),
 			(uint)nvec->rx->pos);

@@ -497,7 +499,7 @@ static void nvec_rx_completed(struct nvec_chip *nvec)
 		nvec->state = 0;
 
 		/* Battery quirk - Often incomplete, and likes to crash */
-		if (nvec->rx->data[0] == NVEC_BAT)
+		if (msg_type == NVEC_BAT)
 			complete(&nvec->ec_transfer);
 
 		return;

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Yingchao Deng <[email protected]>

[ Upstream commit 551bb2fd5e4ed63d33aa11f07102cce5179b7595 ]

In cti_plat_process_filter_sigs(), after allocating a temporary cti_trig_grp struct via kzalloc_obj(), the code never assigns tg->nr_sigs = nr_filter_sigs. Since kzalloc zero-initialises the struct, tg->nr_sigs remains 0. cti_plat_read_trig_group() guards with:

    if (!tgrp->nr_sigs)
        return 0;

so it returns immediately without reading any signal indices from DT.

Fix by assigning tg->nr_sigs before calling cti_plat_read_trig_group().

---
 drivers/hwtracing/coresight/coresight-cti-platform.c | 1 +
 1 file changed, 1 insertion(+)

index ccef04f27f12fa..438f9c7ddccdc3 100644
@@ -329,6 +329,7 @@ static int cti_plat_process_filter_sigs(struct cti_drvdata *drvdata,
 	if (!tg)
 		return -ENOMEM;
 
+	tg->nr_sigs = nr_filter_sigs;
 	err = cti_plat_read_trig_group(tg, fwnode, CTI_DT_FILTER_OUT_SIGS);
 	if (!err)
 		drvdata->config.trig_out_filter |= tg->used_mask;

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Leo Yan <[email protected]>

[ Upstream commit 0ec0a8785d21f63db520bd9d2a67c55e855d36a8 ]

It is a typo to use trcvmidcctlr0 to save and restore TRCVMIDCCTLR1. Use trcvmidcctlr1 instead.

---
 drivers/hwtracing/coresight/coresight-etm4x-core.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

index c5928f63475a06..331df260d07451 100644
@@ -1837,7 +1837,7 @@ static int __etm4_cpu_save(struct etmv4_drvdata *drvdata)
 
 	state->trcvmidcctlr0 = etm4x_read32(csa, TRCVMIDCCTLR0);
 	if (drvdata->numvmidc > 4)
-		state->trcvmidcctlr0 = etm4x_read32(csa, TRCVMIDCCTLR1);
+		state->trcvmidcctlr1 = etm4x_read32(csa, TRCVMIDCCTLR1);
 
 	state->trcclaimset = etm4x_read32(csa, TRCCLAIMCLR);
 

@@ -1963,7 +1963,7 @@ static void __etm4_cpu_restore(struct etmv4_drvdata *drvdata)
 
 	etm4x_relaxed_write32(csa, state->trcvmidcctlr0, TRCVMIDCCTLR0);
 	if (drvdata->numvmidc > 4)
-		etm4x_relaxed_write32(csa, state->trcvmidcctlr0, TRCVMIDCCTLR1);
+		etm4x_relaxed_write32(csa, state->trcvmidcctlr1, TRCVMIDCCTLR1);
 
 	etm4x_relaxed_write32(csa, state->trcclaimset, TRCCLAIMSET);
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Carlos Bilbao <[email protected]>

[ Upstream commit c855c9921da72e535c24737c748f603a52d03f7e ]

Reconfiguring ASPM when a device transitions to low-power state can enable L1.1/L1.2 substates on the PCIe link at a time when the device is sleeping and may be unable to exit them. ASPM should be reconfigured on D0 entry (resume), not on the way down.

pci_set_low_power_state() calls pcie_aspm_pm_state_change() after writing D3hot to PCI_PM_CTRL. pcie_aspm_pm_state_change() resets link->aspm_capable to link->aspm_support and then calls pcie_config_aspm_path(), which can enable ASPM L1.1/L1.2 substates on the PCIe link. If the device cannot recover the link from L1.2 while in D3hot, subsequent config space reads return 0xFFFF ("device inaccessible") and pci_power_up() fails with messages like:

vfio-pci 0000:5d:00.0: Unable to change power state from D3hot to D0, device inaccessible

This was observed on NVIDIA H100 SXM5 GPUs bound to vfio-pci when Linux runtime PM suspends them to D3hot: the GPU becomes permanently inaccessible and disappears from the PCIe bus.

The call to pcie_aspm_pm_state_change() in pci_set_low_power_state() was restored by f93e71aea6c6 ("Revert "PCI/ASPM: Remove pcie_aspm_pm_state_change()""), which reverted 08d0cc5f3426 ("PCI/ASPM:
Remove pcie_aspm_pm_state_change()"). The revert was necessary because the removal broke suspend/resume on certain platforms that required ASPM to be reconfigured on D0 entry. However, the revert restored the call in both pci_set_full_power_state() (D0 entry) and pci_set_low_power_state() (low-power entry).

Only the D0-entry call is needed to fix the suspend/resume regression. The low-power-entry call is harmful: reconfiguring ASPM immediately after putting a device into D3hot can enable link substates that the device or platform cannot exit while the device is sleeping.

Remove the pcie_aspm_pm_state_change() call from pci_set_low_power_state(). ASPM will still be reconfigured correctly when the device returns to D0 via pci_set_full_power_state().

---
 drivers/pci/pci.c | 3 ---
 1 file changed, 3 deletions(-)

index e7648a1abfdbf3..c0cfc3f44f4bf9 100644
@@ -1518,9 +1518,6 @@ static int pci_set_low_power_state(struct pci_dev *dev, pci_power_t state, bool
 				     pci_power_name(dev->current_state),
 				     pci_power_name(state));
 
-	if (dev->bus->self)
-		pcie_aspm_pm_state_change(dev->bus->self, locked);
-
 	return 0;
 }
 

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Guixin Liu <[email protected]>

[ Upstream commit 113e86bc58a918f85d250723436a4d541a873358 ]

Convert the magic numbers associated with PCI ROM into named definitions. Some of these definitions will be used in the second fix patch.

Stable-dep-of: 538796b807fc ("PCI: Check ROM header and data structure addr before accessing")

---
 drivers/pci/rom.c | 40 ++++++++++++++++++++++++++++------------
 1 file changed, 28 insertions(+), 12 deletions(-)

index e18d3a4383ba6b..d4a141bd148b4b 100644
@@ -5,13 +5,28 @@
  * (C) Copyright 2004 Jon Smirl <[email protected]>
  * (C) Copyright 2004 Silicon Graphics, Inc. Jesse Barnes <[email protected]>
  */
+
+#include <linux/bits.h>
 #include <linux/kernel.h>
 #include <linux/export.h>
 #include <linux/pci.h>
+#include <linux/sizes.h>
 #include <linux/slab.h>
 
 #include "pci.h"
 
+#define PCI_ROM_HEADER_SIZE			0x1A
+#define PCI_ROM_POINTER_TO_DATA_STRUCT		0x18
+#define PCI_ROM_LAST_IMAGE_INDICATOR		0x15
+#define PCI_ROM_LAST_IMAGE_INDICATOR_BIT	BIT(7)
+#define PCI_ROM_IMAGE_LEN			0x10
+#define PCI_ROM_IMAGE_SECTOR_SIZE		SZ_512
+#define PCI_ROM_IMAGE_SIGNATURE			0xAA55
+
+/* Data structure signature is "PCIR" in ASCII representation */
+#define PCI_ROM_DATA_STRUCT_SIGNATURE		0x52494350
+#define PCI_ROM_DATA_STRUCT_LEN			0x0A
+
 /**
  * pci_enable_rom - enable ROM decoding for a PCI device
  * @pdev: PCI device to enable

@@ -91,26 +106,27 @@ static size_t pci_get_rom_size(struct pci_dev *pdev, void __iomem *rom,
 	do {
 		void __iomem *pds;
 		/* Standard PCI ROMs start out with these bytes 55 AA */
-		if (readw(image) != 0xAA55) {
-			pci_info(pdev, "Invalid PCI ROM header signature: expecting 0xaa55, got %#06x\n",
-				 readw(image));
+		if (readw(image) != PCI_ROM_IMAGE_SIGNATURE) {
+			pci_info(pdev, "Invalid PCI ROM header signature: expecting %#06x, got %#06x\n",
+				 PCI_ROM_IMAGE_SIGNATURE, readw(image));
 			break;
 		}
-		/* get the PCI data structure and check its "PCIR" signature */
-		pds = image + readw(image + 24);
-		if (readl(pds) != 0x52494350) {
-			pci_info(pdev, "Invalid PCI ROM data signature: expecting 0x52494350, got %#010x\n",
-				 readl(pds));
+		/* Get the PCI data structure and check its "PCIR" signature */
+		pds = image + readw(image + PCI_ROM_POINTER_TO_DATA_STRUCT);
+		if (readl(pds) != PCI_ROM_DATA_STRUCT_SIGNATURE) {
+			pci_info(pdev, "Invalid PCI ROM data signature: expecting %#010x, got %#010x\n",
+				 PCI_ROM_DATA_STRUCT_SIGNATURE, readl(pds));
 			break;
 		}
-		last_image = readb(pds + 21) & 0x80;
-		length = readw(pds + 16);
-		image += length * 512;
+		last_image = readb(pds + PCI_ROM_LAST_IMAGE_INDICATOR) &
+				   PCI_ROM_LAST_IMAGE_INDICATOR_BIT;
+		length = readw(pds + PCI_ROM_IMAGE_LEN);
+		image += length * PCI_ROM_IMAGE_SECTOR_SIZE;
 		/* Avoid iterating through memory outside the resource window */
 		if (image >= rom + size)
 			break;
 		if (!last_image) {
-			if (readw(image) != 0xAA55) {
+			if (readw(image) != PCI_ROM_IMAGE_SIGNATURE) {
 				pci_info(pdev, "No more image in the PCI ROM\n");
 				break;
 			}

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Guixin Liu <[email protected]>

[ Upstream commit 538796b807fcfb81b2ce40cc97a614fd8588feb5 ]

We meet a crash when running stress-ng on x86_64 machine:

  BUG: unable to handle page fault for address: ffa0000007f40000
  RIP: 0010:pci_get_rom_size+0x52/0x220
  Call Trace:
  <TASK>
    pci_map_rom+0x80/0x130
    pci_read_rom+0x4b/0xe0
    kernfs_file_read_iter+0x96/0x180
    vfs_read+0x1b1/0x300

Our analysis reveals that the ROM space's start address is 0xffa0000007f30000, and size is 0x10000. Because of broken ROM space, before calling readl(pds), the pds's value is 0xffa0000007f3ffff, which is already pointed to the ROM space end, invoking readl() would read 4 bytes therefore cause an out-of-bounds access and trigger a crash. Fix this by adding image header and data structure checking.

We also found another crash on arm64 machine:

  Unable to handle kernel paging request at virtual address ffff8000dd1393ff
  Mem abort info:
  ESR = 0x0000000096000021
  EC = 0x25: DABT (current EL), IL = 32 bits
  SET = 0, FnV = 0
  EA = 0, S1PTW = 0
  FSC = 0x21: alignment fault

The call trace is the same with x86_64, but the crash reason is that the data structure addr is not aligned with 4, and arm64 machine report "alignment fault". Fix this by adding alignment checking.

---
 drivers/pci/rom.c | 123 +++++++++++++++++++++++++++++++++++++++-------
 1 file changed, 105 insertions(+), 18 deletions(-)

index d4a141bd148b4b..f2105c6ceef524 100644
@@ -6,9 +6,12 @@
  * (C) Copyright 2004 Silicon Graphics, Inc. Jesse Barnes <[email protected]>
  */
 
+#include <linux/align.h>
 #include <linux/bits.h>
 #include <linux/kernel.h>
 #include <linux/export.h>
+#include <linux/io.h>
+#include <linux/overflow.h>
 #include <linux/pci.h>
 #include <linux/sizes.h>
 #include <linux/slab.h>

@@ -27,6 +30,15 @@
 #define PCI_ROM_DATA_STRUCT_SIGNATURE		0x52494350
 #define PCI_ROM_DATA_STRUCT_LEN			0x0A
 
+/*
+ * Per PCI Firmware r3.3, sec 5.1.3, a conformant PCI Data Structure is at
+ * least 24 bytes (0x18), large enough to cover every fixed field this
+ * driver reads (up to the Indicator byte at offset 0x15).  Reject smaller
+ * device-claimed lengths so the follow-up readers in pci_get_rom_size()
+ * cannot escape the mapped ROM window.
+ */
+#define PCI_ROM_DATA_STRUCT_MIN_LEN		0x18
+
 /**
  * pci_enable_rom - enable ROM decoding for a PCI device
  * @pdev: PCI device to enable

@@ -84,6 +96,91 @@ void pci_disable_rom(struct pci_dev *pdev)
 }
 EXPORT_SYMBOL_GPL(pci_disable_rom);
 
+static bool pci_rom_header_valid(struct pci_dev *pdev, void __iomem *image,
+				 void __iomem *rom, size_t size,
+				 bool expect_valid)
+{
+	unsigned long rom_end = (unsigned long)rom + size - 1;
+	unsigned long header_end;
+	u16 signature;
+
+	/*
+	 * Per PCI Firmware r3.3, sec 5.1, each image must start on a
+	 * 512-byte boundary and must contain the PCI Expansion ROM header.
+	 * Because @rom is page-aligned (returned by ioremap()), checking
+	 * 512-byte alignment of @image is equivalent to enforcing the
+	 * spec's sector-aligned layout within the ROM.  This also
+	 * satisfies the natural-alignment requirement of readw() on archs
+	 * such as arm64 that disallow unaligned IOMEM access.
+	 */
+	if (!IS_ALIGNED((unsigned long)image, PCI_ROM_IMAGE_SECTOR_SIZE))
+		return false;
+
+	if (check_add_overflow((unsigned long)image, PCI_ROM_HEADER_SIZE - 1,
+				&header_end))
+		return false;
+
+	if (image < rom || header_end > rom_end)
+		return false;
+
+	/* Standard PCI ROMs start out with these bytes 55 AA */
+	signature = readw(image);
+	if (signature != PCI_ROM_IMAGE_SIGNATURE) {
+		if (expect_valid) {
+			pci_info(pdev, "Invalid PCI ROM header signature: expecting %#06x, got %#06x\n",
+				 PCI_ROM_IMAGE_SIGNATURE, signature);
+		} else {
+			pci_info(pdev, "No more images in PCI ROM\n");
+		}
+		return false;
+	}
+
+	return true;
+}
+
+static bool pci_rom_data_struct_valid(struct pci_dev *pdev, void __iomem *pds,
+				      void __iomem *rom, size_t size)
+{
+	unsigned long rom_end = (unsigned long)rom + size - 1;
+	unsigned long end;
+	u32 signature;
+	u16 data_len;
+
+	/*
+	 * Some CPU architectures require IOMEM access addresses to be
+	 * aligned, for example arm64, so since we're about to call
+	 * readl(), check here for 4-byte alignment.
+	 */
+	if (!IS_ALIGNED((unsigned long)pds, 4))
+		return false;
+
+	if (check_add_overflow((unsigned long)pds, PCI_ROM_DATA_STRUCT_LEN + 1,
+				&end))
+		return false;
+
+	if (pds < rom || end > rom_end)
+		return false;
+
+	signature = readl(pds);
+	if (signature != PCI_ROM_DATA_STRUCT_SIGNATURE) {
+		pci_info(pdev, "Invalid PCI ROM data signature: expecting %#010x, got %#010x\n",
+			 PCI_ROM_DATA_STRUCT_SIGNATURE, signature);
+		return false;
+	}
+
+	data_len = readw(pds + PCI_ROM_DATA_STRUCT_LEN);
+	if (data_len < PCI_ROM_DATA_STRUCT_MIN_LEN || data_len == U16_MAX)
+		return false;
+
+	if (check_add_overflow((unsigned long)pds, data_len - 1, &end))
+		return false;
+
+	if (end > rom_end)
+		return false;
+
+	return true;
+}
+
 /**
  * pci_get_rom_size - obtain the actual size of the ROM image
  * @pdev: target PCI device

@@ -99,38 +196,28 @@ static size_t pci_get_rom_size(struct pci_dev *pdev, void __iomem *rom,
 			       size_t size)
 {
 	void __iomem *image;
-	int last_image;
 	unsigned int length;
+	bool last_image;
 
 	image = rom;
 	do {
 		void __iomem *pds;
-		/* Standard PCI ROMs start out with these bytes 55 AA */
-		if (readw(image) != PCI_ROM_IMAGE_SIGNATURE) {
-			pci_info(pdev, "Invalid PCI ROM header signature: expecting %#06x, got %#06x\n",
-				 PCI_ROM_IMAGE_SIGNATURE, readw(image));
+		if (!pci_rom_header_valid(pdev, image, rom, size, true))
 			break;
-		}
+
 		/* Get the PCI data structure and check its "PCIR" signature */
 		pds = image + readw(image + PCI_ROM_POINTER_TO_DATA_STRUCT);
-		if (readl(pds) != PCI_ROM_DATA_STRUCT_SIGNATURE) {
-			pci_info(pdev, "Invalid PCI ROM data signature: expecting %#010x, got %#010x\n",
-				 PCI_ROM_DATA_STRUCT_SIGNATURE, readl(pds));
+		if (!pci_rom_data_struct_valid(pdev, pds, rom, size))
 			break;
-		}
+
 		last_image = readb(pds + PCI_ROM_LAST_IMAGE_INDICATOR) &
 				   PCI_ROM_LAST_IMAGE_INDICATOR_BIT;
 		length = readw(pds + PCI_ROM_IMAGE_LEN);
 		image += length * PCI_ROM_IMAGE_SECTOR_SIZE;
-		/* Avoid iterating through memory outside the resource window */
-		if (image >= rom + size)
+
+		if (!last_image &&
+		    !pci_rom_header_valid(pdev, image, rom, size, false))
 			break;
-		if (!last_image) {
-			if (readw(image) != PCI_ROM_IMAGE_SIGNATURE) {
-				pci_info(pdev, "No more image in the PCI ROM\n");
-				break;
-			}
-		}
 	} while (length && !last_image);
 
 	/* never return a size larger than the PCI resource window */

2.53.0

6.6-stable review patch. If anyone has any objections, please let me know.

------------------

From: Rafael J. Wysocki <[email protected]>

[ Upstream commit cc966553e6ff0849978b5754531b768b0ff54985 ]

Prevent leaking a wakeup source object after removing the driver by adding appropriate cleanup code to its remove callback function.