Skip to contentSkip to conversation

usbnet: ipheth: race between ipheth_close and error handling

PATCH New Sasha Levin 1 participant 9 messages Aug 23, 2024

Subsystems (5)
lore
Share & tools
lore

Sasha Levin

author maintainer
#0 lore raw

PATCH 1/9 usbnet: ipheth: race between ipheth_close and error handling

From: Oliver Neukum <[email protected]>

[ Upstream commit e5876b088ba03a62124266fa20d00e65533c7269 ]

ipheth_sndbulk_callback() can submit carrier_work as a part of its error handling. That means that the driver must make sure that the work is cancelled after it has made sure that no more URB can terminate with an error condition.

Hence the order of actions in ipheth_close() needs to be inverted.

---
 drivers/net/usb/ipheth.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/usb/ipheth.c +1 -1
index 06d9f19ca142a..0774d753dd316 100644
@@ -353,8 +353,8 @@ static int ipheth_close(struct net_device *net)
 {
 	struct ipheth_device *dev = netdev_priv(net);
 
-	cancel_delayed_work_sync(&dev->carrier_work);
 	netif_stop_queue(net);
+	cancel_delayed_work_sync(&dev->carrier_work);
 	return 0;
 }
 
Signature · 1 line
2.43.0
author maintainer
#1 lore raw

PATCH 2/9 usbnet: ipheth: fix carrier detection in modes 1 and 4

From: Foster Snowhill <[email protected]>

[ Upstream commit 67927a1b255d883881be9467508e0af9a5e0be9d ]

Apart from the standard "configurations", "interfaces" and "alternate interface settings" in USB, iOS devices also have a notion of "modes". In different modes, the device exposes a different set of available configurations.

Depending on the iOS version, and depending on the current mode, the length and contents of the carrier state control message differs:

* 1 byte (seen on iOS 4.2.1, 8.4):
* 03: carrier off (mode 0)
* 04: carrier on (mode 0)
* 3 bytes (seen on iOS 10.3.4, 15.7.6):
* 03 03 03: carrier off (mode 0)
* 04 04 03: carrier on (mode 0)
* 4 bytes (seen on iOS 16.5, 17.6):
* 03 03 03 00: carrier off (mode 0)
* 04 03 03 00: carrier off (mode 1)
* 06 03 03 00: carrier off (mode 4)
* 04 04 03 04: carrier on (mode 0 and 1)
* 06 04 03 04: carrier on (mode 4)

Before this change, the driver always used the first byte of the response to determine carrier state.

From this larger sample, the first byte seems to indicate the number of available USB configurations in the current mode (with the exception of the default mode 0), and in some cases (namely mode 1 and 4) does not correlate with the carrier state.

Previous logic erroneously counted 04 03 03 00 as "carrier on" and 06 04 03 04 as "carrier off" on iOS versions that support mode 1 and mode 4 respectively.

Only modes 0, 1 and 4 expose the USB Ethernet interfaces necessary for the ipheth driver.

Check the second byte of the control message where possible, and fall back to checking the first byte on older iOS versions.

---
 drivers/net/usb/ipheth.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/net/usb/ipheth.c +3 -2
index 0774d753dd316..208336c871be5 100644
@@ -253,13 +253,14 @@ static int ipheth_carrier_set(struct ipheth_device *dev)
 			0x02, /* index */
 			dev->ctrl_buf, IPHETH_CTRL_BUF_SIZE,
 			IPHETH_CTRL_TIMEOUT);
-	if (retval < 0) {
+	if (retval <= 0) {
 		dev_err(&dev->intf->dev, "%s: usb_control_msg: %d\n",
 			__func__, retval);
 		return retval;
 	}
 
-	if (dev->ctrl_buf[0] == IPHETH_CARRIER_ON) {
+	if ((retval == 1 && dev->ctrl_buf[0] == IPHETH_CARRIER_ON) ||
+	    (retval >= 2 && dev->ctrl_buf[1] == IPHETH_CARRIER_ON)) {
 		netif_carrier_on(dev->net);
 		if (dev->tx_urb->status != -EINPROGRESS)
 			netif_wake_queue(dev->net);
Signature · 1 line
2.43.0
author maintainer
#2 lore raw

PATCH 3/9 net: ethernet: use ip_hdrlen() instead of bit shift

In reply to Sasha Levin

From: Moon Yeounsu <[email protected]>

[ Upstream commit 9a039eeb71a42c8b13408a1976e300f3898e1be0 ]

ip_hdr(skb)->ihl << 2 is the same as ip_hdrlen(skb) Therefore, we should use a well-defined function not a bit shift to find the header length.

It also compresses two lines to a single line.

---
 drivers/net/ethernet/jme.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)
drivers/net/ethernet/jme.c +4 -6
index e9efe074edc11..80858206c5147 100644
@@ -947,15 +947,13 @@ jme_udpsum(struct sk_buff *skb)
 	if (skb->protocol != htons(ETH_P_IP))
 		return csum;
 	skb_set_network_header(skb, ETH_HLEN);
-	if ((ip_hdr(skb)->protocol != IPPROTO_UDP) ||
-	    (skb->len < (ETH_HLEN +
-			(ip_hdr(skb)->ihl << 2) +
-			sizeof(struct udphdr)))) {
+
+	if (ip_hdr(skb)->protocol != IPPROTO_UDP ||
+	    skb->len < (ETH_HLEN + ip_hdrlen(skb) + sizeof(struct udphdr))) {
 		skb_reset_network_header(skb);
 		return csum;
 	}
-	skb_set_transport_header(skb,
-			ETH_HLEN + (ip_hdr(skb)->ihl << 2));
+	skb_set_transport_header(skb, ETH_HLEN + ip_hdrlen(skb));
 	csum = udp_hdr(skb)->check;
 	skb_reset_transport_header(skb);
 	skb_reset_network_header(skb);
Signature · 1 line
2.43.0
author maintainer
#3 lore raw

PATCH 4/9 net: phy: vitesse: repair vsc73xx autonegotiation

In reply to Sasha Levin

From: Pawel Dembicki <[email protected]>

[ Upstream commit de7a670f8defe4ed2115552ad23dea0f432f7be4 ]

When the vsc73xx mdio bus work properly, the generic autonegotiation configuration works well.

---
 drivers/net/phy/vitesse.c | 14 --------------
 1 file changed, 14 deletions(-)
drivers/net/phy/vitesse.c +0 -14
index bb680352708a6..3f594c8784e20 100644
@@ -232,16 +232,6 @@ static int vsc739x_config_init(struct phy_device *phydev)
 	return 0;
 }
 
-static int vsc73xx_config_aneg(struct phy_device *phydev)
-{
-	/* The VSC73xx switches does not like to be instructed to
-	 * do autonegotiation in any way, it prefers that you just go
-	 * with the power-on/reset defaults. Writing some registers will
-	 * just make autonegotiation permanently fail.
-	 */
-	return 0;
-}
-
 /* This adds a skew for both TX and RX clocks, so the skew should only be
  * applied to "rgmii-id" interfaces. It may not work as expected
  * on "rgmii-txid", "rgmii-rxid" or "rgmii" interfaces. */
@@ -424,7 +414,6 @@ static struct phy_driver vsc82xx_driver[] = {
 	.phy_id_mask    = 0x000ffff0,
 	/* PHY_GBIT_FEATURES */
 	.config_init    = vsc738x_config_init,
-	.config_aneg    = vsc73xx_config_aneg,
 	.read_page      = vsc73xx_read_page,
 	.write_page     = vsc73xx_write_page,
 }, {
@@ -433,7 +422,6 @@ static struct phy_driver vsc82xx_driver[] = {
 	.phy_id_mask    = 0x000ffff0,
 	/* PHY_GBIT_FEATURES */
 	.config_init    = vsc738x_config_init,
-	.config_aneg    = vsc73xx_config_aneg,
 	.read_page      = vsc73xx_read_page,
 	.write_page     = vsc73xx_write_page,
 }, {
@@ -442,7 +430,6 @@ static struct phy_driver vsc82xx_driver[] = {
 	.phy_id_mask    = 0x000ffff0,
 	/* PHY_GBIT_FEATURES */
 	.config_init    = vsc739x_config_init,
-	.config_aneg    = vsc73xx_config_aneg,
 	.read_page      = vsc73xx_read_page,
 	.write_page     = vsc73xx_write_page,
 }, {
@@ -451,7 +438,6 @@ static struct phy_driver vsc82xx_driver[] = {
 	.phy_id_mask    = 0x000ffff0,
 	/* PHY_GBIT_FEATURES */
 	.config_init    = vsc739x_config_init,
-	.config_aneg    = vsc73xx_config_aneg,
 	.read_page      = vsc73xx_read_page,
 	.write_page     = vsc73xx_write_page,
 }, {
Signature · 1 line
2.43.0
author maintainer
#4 lore raw

PATCH 5/9 powerpc/mm: Fix boot warning with hugepages and CONFIG_DEBUG_VIRTUAL

In reply to Sasha Levin

From: Christophe Leroy <[email protected]>

[ Upstream commit e7e846dc6c73fbc94ae8b4ec20d05627646416f2 ]

Booting with CONFIG_DEBUG_VIRTUAL leads to following warning when passing hugepage reservation on command line:

Kernel command line: hugepagesz=1g hugepages=1 hugepagesz=64m hugepages=1 hugepagesz=256m hugepages=1 noreboot
HugeTLB: allocating 1 of page size 1.00 GiB failed. Only allocated 0 hugepages.

  ------------[ cut here ]------------
  WARNING: CPU: 0 PID: 0 at arch/powerpc/include/asm/io.h:948 __alloc_bootmem_huge_page+0xd4/0x284
  Modules linked in:
  CPU: 0 PID: 0 Comm: swapper Not tainted 6.10.0-rc6-00396-g6b0e82791bd0-dirty #936
  Hardware name: MPC8544DS e500v2 0x80210030 MPC8544 DS
  NIP:  c1020240 LR: c10201d0 CTR: 00000000
  REGS: c13fdd30 TRAP: 0700   Not tainted  (6.10.0-rc6-00396-g6b0e82791bd0-dirty)
  MSR:  00021000 <CE,ME>  CR: 44084288  XER: 20000000

GPR00: c10201d0 c13fde20 c130b560 e8000000 e8001000 00000000 00000000 c1420000
GPR08: 00000000 00028001 00000000 00000004 44084282 01066ac0 c0eb7c9c efffe149
GPR16: c0fc4228 0000005f ffffffff c0eb7d0c c0eb7cc0 c0eb7ce0 ffffffff 00000000
GPR24: c1441cec efffe153 e8001000 c14240c0 00000000 c1441d64 00000000 e8000000 NIP [c1020240] __alloc_bootmem_huge_page+0xd4/0x284
LR [c10201d0] __alloc_bootmem_huge_page+0x64/0x284

  Call Trace:
  [c13fde20] [c10201d0] __alloc_bootmem_huge_page+0x64/0x284 (unreliable)
  [c13fde50] [c10207b8] hugetlb_hstate_alloc_pages+0x8c/0x3e8
  [c13fdeb0] [c1021384] hugepages_setup+0x240/0x2cc
  [c13fdef0] [c1000574] unknown_bootoption+0xfc/0x280
  [c13fdf30] [c0078904] parse_args+0x200/0x4c4
  [c13fdfa0] [c1000d9c] start_kernel+0x238/0x7d0
  [c13fdff0] [c0000434] set_ivor+0x12c/0x168
  Code: 554aa33e 7c042840 3ce0c142 80a7427c 5109a016 50caa016 7c9a2378 7fdcf378 4180000c 7c052040 41810160 7c095040 <0fe00000> 38c00000 40800108 3c60c0eb
  ---[ end trace 0000000000000000 ]---

This is due to virt_addr_valid() using high_memory before it is set.

high_memory is set in mem_init() using max_low_pfn, but max_low_pfn is available long before, it is set in mem_topology_setup(). So just like commit daa9ada2093e ("powerpc/mm: Fix boot crash with FLATMEM") moved the setting of max_mapnr immediately after the call to mem_topology_setup(), the same can be done for high_memory.

---
 arch/powerpc/kernel/setup-common.c | 1 +
 arch/powerpc/mm/mem.c              | 2 --
 2 files changed, 1 insertion(+), 2 deletions(-)
arch/powerpc/kernel/setup-common.c +1 -0
index 1e417c3eedfef..b99b35ba2412d 100644
@@ -908,6 +908,7 @@ void __init setup_arch(char **cmdline_p)
 	mem_topology_setup();
 	/* Set max_mapnr before paging_init() */
 	set_max_mapnr(max_pfn);
+	high_memory = (void *)__va(max_low_pfn * PAGE_SIZE);
 
 	/*
 	 * Release secondary cpus out of their spinloops at 0x60 now that
arch/powerpc/mm/mem.c +0 -2
index 08e3422eb7926..cd6a68a64d87c 100644
@@ -292,8 +292,6 @@ void __init mem_init(void)
 		swiotlb_init(0);
 #endif
 
-	high_memory = (void *) __va(max_low_pfn * PAGE_SIZE);
-
 	kasan_late_init();
 
 	memblock_free_all();
Signature · 1 line
2.43.0
author maintainer
#5 lore raw

PATCH 6/9 Squashfs: sanity check symbolic link size

In reply to Sasha Levin

From: Phillip Lougher <[email protected]>

[ Upstream commit 810ee43d9cd245d138a2733d87a24858a23f577d ]

Syzkiller reports a "KMSAN: uninit-value in pick_link" bug.

This is caused by an uninitialised page, which is ultimately caused by a corrupted symbolic link size read from disk.

The reason why the corrupted symlink size causes an uninitialised page is due to the following sequence of events:

1. squashfs_read_inode() is called to read the symbolic link from disk. This assigns the corrupted value 3875536935 to inode->i_size.

2. Later squashfs_symlink_read_folio() is called, which assigns this corrupted value to the length variable, which being a signed int, overflows producing a negative number.

3. The following loop that fills in the page contents checks that the copied bytes is less than length, which being negative means the loop is skipped, producing an uninitialised page.

This patch adds a sanity check which checks that the symbolic link size is not larger than expected.

--

V2: fix spelling mistake.

---
 fs/squashfs/inode.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
fs/squashfs/inode.c +6 -1
index 24463145b3513..f31649080a881 100644
@@ -276,8 +276,13 @@ int squashfs_read_inode(struct inode *inode, long long ino)
 		if (err < 0)
 			goto failed_read;
 
-		set_nlink(inode, le32_to_cpu(sqsh_ino->nlink));
 		inode->i_size = le32_to_cpu(sqsh_ino->symlink_size);
+		if (inode->i_size > PAGE_SIZE) {
+			ERROR("Corrupted symlink\n");
+			return -EINVAL;
+		}
+
+		set_nlink(inode, le32_to_cpu(sqsh_ino->nlink));
 		inode->i_op = &squashfs_symlink_inode_ops;
 		inode_nohighmem(inode);
 		inode->i_data.a_ops = &squashfs_symlink_aops;
Signature · 1 line
2.43.0
author maintainer
#6 lore raw

PATCH 7/9 of/irq: Prevent device address out-of-bounds read in interrupt map walk

In reply to Sasha Levin

From: Stefan Wiehler <[email protected]>

[ Upstream commit b739dffa5d570b411d4bdf4bb9b8dfd6b7d72305 ]

When of_irq_parse_raw() is invoked with a device address smaller than the interrupt parent node (from #address-cells property), KASAN detects the following out-of-bounds read when populating the initial match table (dyndbg="func of_irq_parse_* +p"):

OF: of_irq_parse_one: dev=/soc@0/picasso/watchdog, index=0
OF: parent=/soc@0/pci@878000000000/gpio0@17,0, intsize=2
OF: intspec=4
OF: of_irq_parse_raw: ipar=/soc@0/pci@878000000000/gpio0@17,0, size=2
OF: -> addrsize=3
==================================================================

  BUG: KASAN: slab-out-of-bounds in of_irq_parse_raw+0x2b8/0x8d0
  Read of size 4 at addr ffffff81beca5608 by task bash/764

  CPU: 1 PID: 764 Comm: bash Tainted: G           O       6.1.67-484c613561-nokia_sm_arm64 #1
  Hardware name: Unknown Unknown Product/Unknown Product, BIOS 2023.01-12.24.03-dirty 01/01/2023
  Call trace:
   dump_backtrace+0xdc/0x130
   show_stack+0x1c/0x30
   dump_stack_lvl+0x6c/0x84
   print_report+0x150/0x448
   kasan_report+0x98/0x140
   __asan_load4+0x78/0xa0
   of_irq_parse_raw+0x2b8/0x8d0
   of_irq_parse_one+0x24c/0x270
   parse_interrupts+0xc0/0x120
   of_fwnode_add_links+0x100/0x2d0
   fw_devlink_parse_fwtree+0x64/0xc0
   device_add+0xb38/0xc30
   of_device_add+0x64/0x90
   of_platform_device_create_pdata+0xd0/0x170
   of_platform_bus_create+0x244/0x600
   of_platform_notify+0x1b0/0x254
   blocking_notifier_call_chain+0x9c/0xd0
   __of_changeset_entry_notify+0x1b8/0x230
   __of_changeset_apply_notify+0x54/0xe4
   of_overlay_fdt_apply+0xc04/0xd94
   ...

  The buggy address belongs to the object at ffffff81beca5600
   which belongs to the cache kmalloc-128 of size 128
  The buggy address is located 8 bytes inside of
   128-byte region [ffffff81beca5600, ffffff81beca5680)

  The buggy address belongs to the physical page:
  page:00000000230d3d03 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x1beca4
  head:00000000230d3d03 order:1 compound_mapcount:0 compound_pincount:0
  flags: 0x8000000000010200(slab|head|zone=2)
  raw: 8000000000010200 0000000000000000 dead000000000122 ffffff810000c300
  raw: 0000000000000000 0000000000200020 00000001ffffffff 0000000000000000
  page dumped because: kasan: bad access detected

  Memory state around the buggy address:
   ffffff81beca5500: 04 fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
   ffffff81beca5580: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
  >ffffff81beca5600: 00 fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
                        ^
   ffffff81beca5680: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
   ffffff81beca5700: 00 00 00 00 00 00 fc fc fc fc fc fc fc fc fc fc
  ==================================================================
  OF:  -> got it !

Prevent the out-of-bounds read by copying the device address into a buffer of sufficient size.

---
 drivers/of/irq.c | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)
drivers/of/irq.c +11 -4
index 352e14b007e78..ad0cb49e233ac 100644
@@ -288,7 +288,8 @@ int of_irq_parse_one(struct device_node *device, int index, struct of_phandle_ar
 	struct device_node *p;
 	const __be32 *addr;
 	u32 intsize;
-	int i, res;
+	int i, res, addr_len;
+	__be32 addr_buf[3] = { 0 };
 
 	pr_debug("of_irq_parse_one: dev=%pOF, index=%d\n", device, index);
 
@@ -297,13 +298,19 @@ int of_irq_parse_one(struct device_node *device, int index, struct of_phandle_ar
 		return of_irq_parse_oldworld(device, index, out_irq);
 
 	/* Get the reg property (if any) */
-	addr = of_get_property(device, "reg", NULL);
+	addr = of_get_property(device, "reg", &addr_len);
+
+	/* Prevent out-of-bounds read in case of longer interrupt parent address size */
+	if (addr_len > (3 * sizeof(__be32)))
+		addr_len = 3 * sizeof(__be32);
+	if (addr)
+		memcpy(addr_buf, addr, addr_len);
 
 	/* Try the new-style interrupts-extended first */
 	res = of_parse_phandle_with_args(device, "interrupts-extended",
 					"#interrupt-cells", index, out_irq);
 	if (!res)
-		return of_irq_parse_raw(addr, out_irq);
+		return of_irq_parse_raw(addr_buf, out_irq);
 
 	/* Look for the interrupt parent. */
 	p = of_irq_find_parent(device);
@@ -333,7 +340,7 @@ int of_irq_parse_one(struct device_node *device, int index, struct of_phandle_ar
 
 
 	/* Check if there are any interrupt-map translations to process */
-	res = of_irq_parse_raw(addr, out_irq);
+	res = of_irq_parse_raw(addr_buf, out_irq);
  out:
 	of_node_put(p);
 	return res;
Signature · 1 line
2.43.0
author maintainer
#7 lore raw

PATCH 8/9 lib/generic-radix-tree.c: Fix rare race in __genradix_ptr_alloc()

In reply to Sasha Levin

From: Kent Overstreet <[email protected]>

[ Upstream commit b2f11c6f3e1fc60742673b8675c95b78447f3dae ]

If we need to increase the tree depth, allocate a new node, and then race with another thread that increased the tree depth before us, we'll still have a preallocated node that might be used later.

If we then use that node for a new non-root node, it'll still have a pointer to the old root instead of being zeroed - fix this by zeroing it in the cmpxchg failure path.

---
 lib/generic-radix-tree.c | 2 ++
 1 file changed, 2 insertions(+)
lib/generic-radix-tree.c +2 -0
index f25eb111c0516..34d3ac52de894 100644
@@ -131,6 +131,8 @@ void *__genradix_ptr_alloc(struct __genradix *radix, size_t offset,
 		if ((v = cmpxchg_release(&radix->root, r, new_root)) == r) {
 			v = new_root;
 			new_node = NULL;
+		} else {
+			new_node->children[0] = NULL;
 		}
 	}
 
Signature · 1 line
2.43.0
author maintainer
#8 lore raw

PATCH 9/9 btrfs: update target inode's ctime on unlink

In reply to Sasha Levin

From: Jeff Layton <[email protected]>

[ Upstream commit 3bc2ac2f8f0b78a13140fc72022771efe0c9b778 ]

Unlink changes the link count on the target inode. POSIX mandates that the ctime must also change when this occurs.

According to https://pubs.opengroup.org/onlinepubs/9699919799/functions/unlink.html:

"Upon successful completion, unlink() shall mark for update the last data
modification and last file status change timestamps of the parent directory. Also, if the file's link count is not 0, the last file status change timestamp of the file shall be marked for update."

[ add link to the opengroup docs ]
---
 fs/btrfs/inode.c | 1 +
 1 file changed, 1 insertion(+)
fs/btrfs/inode.c +1 -0
index 1f99d7dced17a..4a6c85df7b3ce 100644
@@ -3704,6 +3704,7 @@ static int __btrfs_unlink_inode(struct btrfs_trans_handle *trans,
 
 	btrfs_i_size_write(dir, dir->vfs_inode.i_size - name_len * 2);
 	inode_inc_iversion(&inode->vfs_inode);
+	inode_set_ctime_current(&inode->vfs_inode);
 	inode_inc_iversion(&dir->vfs_inode);
 	inode->vfs_inode.i_ctime = dir->vfs_inode.i_mtime =
 		dir->vfs_inode.i_ctime = current_time(&inode->vfs_inode);
Signature · 1 line
2.43.0