Skip to contentSkip to conversation

clk: zynqmp: Fix stack-out-of-bounds in strncpy`

PATCH In review Sasha Levin 2 participants 30 messages Oct 13, 2022

Subsystems (12)
lore
Share & tools
lore

Sasha Levin

author maintainer
#0 lore raw

PATCH 1/27 clk: zynqmp: Fix stack-out-of-bounds in strncpy`

From: Ian Nam <[email protected]>

[ Upstream commit dd80fb2dbf1cd8751efbe4e53e54056f56a9b115 ]

"BUG: KASAN: stack-out-of-bounds in strncpy+0x30/0x68"

Linux-ATF interface is using 16 bytes of SMC payload. In case clock name is longer than 15 bytes, string terminated NULL character will not be received by Linux. Add explicit NULL character at last byte to fix issues when clock name is longer.

This fixes below bug reported by KASAN:

==================================================================

 BUG: KASAN: stack-out-of-bounds in strncpy+0x30/0x68
 Read of size 1 at addr ffff0008c89a7410 by task swapper/0/1

 CPU: 1 PID: 1 Comm: swapper/0 Not tainted 5.4.0-00396-g81ef9e7-dirty #3
 Hardware name: Xilinx Versal vck190 Eval board revA (QSPI) (DT)
 Call trace:
  dump_backtrace+0x0/0x1e8
  show_stack+0x14/0x20
  dump_stack+0xd4/0x108
  print_address_description.isra.0+0xbc/0x37c
  __kasan_report+0x144/0x198
  kasan_report+0xc/0x18
  __asan_load1+0x5c/0x68
  strncpy+0x30/0x68
  zynqmp_clock_probe+0x238/0x7b8
  platform_drv_probe+0x6c/0xc8
  really_probe+0x14c/0x418
  driver_probe_device+0x74/0x130
  __device_attach_driver+0xc4/0xe8
  bus_for_each_drv+0xec/0x150
  __device_attach+0x160/0x1d8
  device_initial_probe+0x10/0x18
  bus_probe_device+0xe0/0xf0
  device_add+0x528/0x950
  of_device_add+0x5c/0x80
  of_platform_device_create_pdata+0x120/0x168
  of_platform_bus_create+0x244/0x4e0
  of_platform_populate+0x50/0xe8
  zynqmp_firmware_probe+0x370/0x3a8
  platform_drv_probe+0x6c/0xc8
  really_probe+0x14c/0x418
  driver_probe_device+0x74/0x130
  device_driver_attach+0x94/0xa0
  __driver_attach+0x70/0x108
  bus_for_each_dev+0xe4/0x158
  driver_attach+0x30/0x40
  bus_add_driver+0x21c/0x2b8
  driver_register+0xbc/0x1d0
  __platform_driver_register+0x7c/0x88
  zynqmp_firmware_driver_init+0x1c/0x24
  do_one_initcall+0xa4/0x234
  kernel_init_freeable+0x1b0/0x24c
  kernel_init+0x10/0x110
  ret_from_fork+0x10/0x18

 The buggy address belongs to the page:
 page:ffff0008f9be1c88 refcount:0 mapcount:0 mapping:0000000000000000 index:0x0
 raw: 0008d00000000000 ffff0008f9be1c90 ffff0008f9be1c90 0000000000000000
 raw: 0000000000000000 0000000000000000 00000000ffffffff
 page dumped because: kasan: bad access detected

addr ffff0008c89a7410 is located in stack of task swapper/0/1 at offset 112 in frame:
zynqmp_clock_probe+0x0/0x7b8

this frame has 3 objects:
[32, 44) 'response'
[64, 80) 'ret_payload'
[96, 112) 'name'

 Memory state around the buggy address:
  ffff0008c89a7300: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  ffff0008c89a7380: 00 00 00 00 f1 f1 f1 f1 00 04 f2 f2 00 00 f2 f2
 >ffff0008c89a7400: 00 00 f3 f3 00 00 00 00 00 00 00 00 00 00 00 00
                          ^
  ffff0008c89a7480: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  ffff0008c89a7500: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 ==================================================================
---
 drivers/clk/zynqmp/clkc.c | 7 +++++++
 1 file changed, 7 insertions(+)
drivers/clk/zynqmp/clkc.c +7 -0
index 6f057ab9df03..4ef201ac8252 100644
@@ -679,6 +679,13 @@ static void zynqmp_get_clock_info(void)
 				  FIELD_PREP(CLK_ATTR_NODE_INDEX, i);
 
 		zynqmp_pm_clock_get_name(clock[i].clk_id, &name);
+
+		/*
+		 * Terminate with NULL character in case name provided by firmware
+		 * is longer and truncated due to size limit.
+		 */
+		name.name[sizeof(name.name) - 1] = '\0';
+
 		if (!strcmp(name.name, RESERVED_CLK_NAME))
 			continue;
 		strncpy(clock[i].clk_name, name.name, MAX_NAME_LEN);
Signature · 7 lines
2.35.1


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
author maintainer
#1 lore raw

PATCH 2/27 media: cx88: Fix a null-ptr-deref bug in buffer_prepare()

From: Zheyu Ma <[email protected]>

[ Upstream commit 2b064d91440b33fba5b452f2d1b31f13ae911d71 ]

When the driver calls cx88_risc_buffer() to prepare the buffer, the function call may fail, resulting in a empty buffer and null-ptr-deref later in buffer_queue().

The following log can reveal it:

[   41.822762] general protection fault, probably for non-canonical address 0xdffffc0000000000: 0000 [#1] PREEMPT SMP KASAN PTI
[   41.824488] KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007]
[   41.828027] RIP: 0010:buffer_queue+0xc2/0x500
[   41.836311] Call Trace:
[   41.836945]  __enqueue_in_driver+0x141/0x360
[   41.837262]  vb2_start_streaming+0x62/0x4a0
[   41.838216]  vb2_core_streamon+0x1da/0x2c0
[   41.838516]  __vb2_init_fileio+0x981/0xbc0
[   41.839141]  __vb2_perform_fileio+0xbf9/0x1120
[   41.840072]  vb2_fop_read+0x20e/0x400
[   41.840346]  v4l2_read+0x215/0x290
[   41.840603]  vfs_read+0x162/0x4c0

Fix this by checking the return value of cx88_risc_buffer()

[hverkuil: fix coding style issues]

---
 drivers/media/pci/cx88/cx88-vbi.c   |  9 +++---
 drivers/media/pci/cx88/cx88-video.c | 43 +++++++++++++++--------------
 2 files changed, 26 insertions(+), 26 deletions(-)
drivers/media/pci/cx88/cx88-vbi.c +4 -5
index 58489ea0c1da..7cf2271866d0 100644
@@ -144,11 +144,10 @@ static int buffer_prepare(struct vb2_buffer *vb)
 		return -EINVAL;
 	vb2_set_plane_payload(vb, 0, size);
 
-	cx88_risc_buffer(dev->pci, &buf->risc, sgt->sgl,
-			 0, VBI_LINE_LENGTH * lines,
-			 VBI_LINE_LENGTH, 0,
-			 lines);
-	return 0;
+	return cx88_risc_buffer(dev->pci, &buf->risc, sgt->sgl,
+				0, VBI_LINE_LENGTH * lines,
+				VBI_LINE_LENGTH, 0,
+				lines);
 }
 
 static void buffer_finish(struct vb2_buffer *vb)
drivers/media/pci/cx88/cx88-video.c +22 -21

70 lines not on this page · raw message

Signature · 1 line
2.35.1
author maintainer
#2 lore raw

PATCH 3/27 clk: zynqmp: pll: rectify rate rounding in zynqmp_pll_round_rate

In reply to Sasha Levin

From: Quanyang Wang <[email protected]>

[ Upstream commit 30eaf02149ecc3c5815e45d27187bf09e925071d ]

The function zynqmp_pll_round_rate is used to find a most appropriate PLL frequency which the hardware can generate according to the desired frequency. For example, if the desired frequency is 297MHz, considering the limited range from PS_PLL_VCO_MIN (1.5GHz) to PS_PLL_VCO_MAX (3.0GHz) of PLL, zynqmp_pll_round_rate should return 1.872GHz (297MHz * 5).

There are two problems with the current code of zynqmp_pll_round_rate:

1) When the rate is below PS_PLL_VCO_MIN, it can't find a correct rate when the parameter "rate" is an integer multiple of *prate, in other words, if "f" is zero, zynqmp_pll_round_rate won't return a valid frequency which is from PS_PLL_VCO_MIN to PS_PLL_VCO_MAX. For example, *prate is 33MHz and the rate is 660MHz, zynqmp_pll_round_rate will not boost up rate and just return 660MHz, and this will cause clk_calc_new_rates failure since zynqmp_pll_round_rate returns an invalid rate out of its boundaries.

2) Even if the rate is higher than PS_PLL_VCO_MIN, there is still a risk that zynqmp_pll_round_rate returns an invalid rate because the function DIV_ROUND_CLOSEST makes some loss in the fractional part. If the parent clock *prate is 33333333Hz and we want to set the PLL rate to 1.5GHz, this function will return 1499999985Hz by using the formula below:
value = *prate * DIV_ROUND_CLOSEST(rate, *prate)).
This value is also invalid since it's slightly smaller than PS_PLL_VCO_MIN. because DIV_ROUND_CLOSEST makes some loss in the fractional part.

---
 drivers/clk/zynqmp/pll.c | 31 +++++++++++++++----------------
 1 file changed, 15 insertions(+), 16 deletions(-)
drivers/clk/zynqmp/pll.c +15 -16
index 18fee827602a..3a2a694e5bf3 100644
@@ -98,26 +98,25 @@ static long zynqmp_pll_round_rate(struct clk_hw *hw, unsigned long rate,
 				  unsigned long *prate)
 {
 	u32 fbdiv;
-	long rate_div, f;
+	u32 mult, div;
 
-	/* Enable the fractional mode if needed */
-	rate_div = (rate * FRAC_DIV) / *prate;
-	f = rate_div % FRAC_DIV;
-	if (f) {
-		if (rate > PS_PLL_VCO_MAX) {
-			fbdiv = rate / PS_PLL_VCO_MAX;
-			rate = rate / (fbdiv + 1);
-		}
-		if (rate < PS_PLL_VCO_MIN) {
-			fbdiv = DIV_ROUND_UP(PS_PLL_VCO_MIN, rate);
-			rate = rate * fbdiv;
-		}
-		return rate;
+	/* Let rate fall inside the range PS_PLL_VCO_MIN ~ PS_PLL_VCO_MAX */
+	if (rate > PS_PLL_VCO_MAX) {
+		div = DIV_ROUND_UP(rate, PS_PLL_VCO_MAX);
+		rate = rate / div;
+	}
+	if (rate < PS_PLL_VCO_MIN) {
+		mult = DIV_ROUND_UP(PS_PLL_VCO_MIN, rate);
+		rate = rate * mult;
 	}
 
 	fbdiv = DIV_ROUND_CLOSEST(rate, *prate);
-	fbdiv = clamp_t(u32, fbdiv, PLL_FBDIV_MIN, PLL_FBDIV_MAX);
-	return *prate * fbdiv;
+	if (fbdiv < PLL_FBDIV_MIN || fbdiv > PLL_FBDIV_MAX) {
+		fbdiv = clamp_t(u32, fbdiv, PLL_FBDIV_MIN, PLL_FBDIV_MAX);
+		rate = *prate * fbdiv;
+	}
+
+	return rate;
 }
 
 /**
Signature · 1 line
2.35.1
author maintainer
#3 lore raw

PATCH 4/27 scsi: 3w-9xxx: Avoid disabling device if failing to enable it

In reply to Sasha Levin

From: Letu Ren <[email protected]>

[ Upstream commit 7eff437b5ee1309b34667844361c6bbb5c97df05 ]

The original code will "goto out_disable_device" and call pci_disable_device() if pci_enable_device() fails. The kernel will generate a warning message like "3w-9xxx 0000:00:05.0: disabling already-disabled device".

We shouldn't disable a device that failed to be enabled. A simple return is fine.

---
 drivers/scsi/3w-9xxx.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/scsi/3w-9xxx.c +1 -1
index 3337b1e80412..f6f92033132a 100644
@@ -2014,7 +2014,7 @@ static int twa_probe(struct pci_dev *pdev, const struct pci_device_id *dev_id)
 	retval = pci_enable_device(pdev);
 	if (retval) {
 		TW_PRINTK(host, TW_DRIVER, 0x34, "Failed to enable pci device");
-		goto out_disable_device;
+		return -ENODEV;
 	}
 
 	pci_set_master(pdev);
Signature · 1 line
2.35.1
author maintainer
#4 lore raw

PATCH 5/27 nbd: Fix hung when signal interrupts nbd_start_device_ioctl()

In reply to Sasha Levin

From: Shigeru Yoshida <[email protected]>

[ Upstream commit 1de7c3cf48fc41cd95adb12bd1ea9033a917798a ]

syzbot reported hung task [1]. The following program is a simplified version of the reproducer:

int main(void)

{
	int sv[2], fd;

	if (socketpair(AF_UNIX, SOCK_STREAM, 0, sv) < 0)
		return 1;
	if ((fd = open("/dev/nbd0", 0)) < 0)
		return 1;
	if (ioctl(fd, NBD_SET_SIZE_BLOCKS, 0x81) < 0)
		return 1;
	if (ioctl(fd, NBD_SET_SOCK, sv[0]) < 0)
		return 1;
	if (ioctl(fd, NBD_DO_IT) < 0)
		return 1;
	return 0;
}

When signal interrupt nbd_start_device_ioctl() waiting the condition atomic_read(&config->recv_threads) == 0, the task can hung because it waits the completion of the inflight IOs.

This patch fixes the issue by clearing queue, not just shutdown, when signal interrupt nbd_start_device_ioctl().

---
 drivers/block/nbd.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/block/nbd.c +4 -2
index 09323b0510f0..610dc6a36a9d 100644
@@ -1327,10 +1327,12 @@ static int nbd_start_device_ioctl(struct nbd_device *nbd, struct block_device *b
 	mutex_unlock(&nbd->config_lock);
 	ret = wait_event_interruptible(config->recv_wq,
 					 atomic_read(&config->recv_threads) == 0);
-	if (ret)
+	if (ret) {
 		sock_shutdown(nbd);
-	flush_workqueue(nbd->recv_workq);
+		nbd_clear_que(nbd);
+	}
 
+	flush_workqueue(nbd->recv_workq);
 	mutex_lock(&nbd->config_lock);
 	nbd_bdev_reset(bdev);
 	/* user requested, ignore socket errors */
Signature · 1 line
2.35.1
author maintainer
#5 lore raw

PATCH 6/27 staging: rtl8712: Fix return type for implementation of ndo_start_xmit

In reply to Sasha Levin

From: GUO Zihua <[email protected]>

[ Upstream commit 307d343620e1fc7a6a2b7a1cdadb705532c9b6a5 ]

CFI (Control Flow Integrity) is a safety feature allowing the system to detect and react should a potential control flow hijacking occurs. In particular, the Forward-Edge CFI protects indirect function calls by ensuring the prototype of function that is actually called matches the definition of the function hook.

Since Linux now supports CFI, it will be a good idea to fix mismatched return type for implementation of hooks. Otherwise this would get cought out by CFI and cause a panic.

Use enums from netdev_tx_t as return value instead, then change return type to netdev_tx_t.

---
 drivers/staging/rtl8712/xmit_linux.c | 6 +++---
 drivers/staging/rtl8712/xmit_osdep.h | 2 +-
 2 files changed, 4 insertions(+), 4 deletions(-)
drivers/staging/rtl8712/xmit_linux.c +3 -3
index 1f67d86c606f..8ec8edbb7fd8 100644
@@ -147,7 +147,7 @@ void r8712_xmit_complete(struct _adapter *padapter, struct xmit_frame *pxframe)
 	pxframe->pkt = NULL;
 }
 
-int r8712_xmit_entry(_pkt *pkt, struct  net_device *netdev)
+netdev_tx_t r8712_xmit_entry(_pkt *pkt, struct  net_device *netdev)
 {
 	struct xmit_frame *xmitframe = NULL;
 	struct _adapter *adapter = netdev_priv(netdev);
@@ -172,11 +172,11 @@ int r8712_xmit_entry(_pkt *pkt, struct  net_device *netdev)
 	}
 	xmitpriv->tx_pkts++;
 	xmitpriv->tx_bytes += xmitframe->attrib.last_txcmdsz;
-	return 0;
+	return NETDEV_TX_OK;
 _xmit_entry_drop:
 	if (xmitframe)
 		r8712_free_xmitframe(xmitpriv, xmitframe);
 	xmitpriv->tx_drop++;
 	dev_kfree_skb_any(pkt);
-	return 0;
+	return NETDEV_TX_OK;
 }
drivers/staging/rtl8712/xmit_osdep.h +1 -1
index 21f6b31e0f50..3f2f5edd2d91 100644
@@ -34,7 +34,7 @@ struct sta_xmit_priv;
 struct xmit_frame;
 struct xmit_buf;
 
-int r8712_xmit_entry(_pkt *pkt, struct  net_device *pnetdev);
+netdev_tx_t r8712_xmit_entry(_pkt *pkt, struct  net_device *pnetdev);
 void r8712_SetFilter(struct work_struct *work);
 int r8712_xmit_resource_alloc(struct _adapter *padapter,
 			   struct xmit_buf *pxmitbuf);
Signature · 1 line
2.35.1
#6 lore raw

Re: [PATCH AUTOSEL 5.4 06/27] staging: rtl8712: Fix return type for implementation of ndo_start_xmit

Quoted 10 lines · Sasha Levin

From: GUO Zihua <[email protected]>

[ Upstream commit 307d343620e1fc7a6a2b7a1cdadb705532c9b6a5 ]

CFI (Control Flow Integrity) is a safety feature allowing the system to detect and react should a potential control flow hijacking occurs. In particular, the Forward-Edge CFI protects indirect function calls by ensuring the prototype of function that is actually called matches the definition of the function hook.

Quoted 3 lines

Since Linux now supports CFI, it will be a good idea to fix mismatched return type for implementation of hooks. Otherwise this would get cought out by CFI and cause a panic.

And another that should be dropped from all stable branches, thanks.

greg k-h

author maintainer
#7 lore raw

Re: [PATCH AUTOSEL 5.4 06/27] staging: rtl8712: Fix return type for implementation of ndo_start_xmit

Quoted 11 lines · Greg Kroah-Hartman
Quoted 10 lines · Sasha Levin

From: GUO Zihua <[email protected]>

[ Upstream commit 307d343620e1fc7a6a2b7a1cdadb705532c9b6a5 ]

CFI (Control Flow Integrity) is a safety feature allowing the system to detect and react should a potential control flow hijacking occurs. In particular, the Forward-Edge CFI protects indirect function calls by ensuring the prototype of function that is actually called matches the definition of the function hook.

Quoted 5 lines
Quoted 3 lines

Since Linux now supports CFI, it will be a good idea to fix mismatched return type for implementation of hooks. Otherwise this would get cought out by CFI and cause a panic.

And another that should be dropped from all stable branches, thanks.

Ack, I'll drop this and the rest of the staging patches you've pointed out.

Signature · 2 lines
Thanks,
Sasha
author maintainer
#8 lore raw

PATCH 7/27 staging: rtl8192e: Fix return type for implementation of ndo_start_xmit

In reply to Sasha Levin

From: GUO Zihua <[email protected]>

[ Upstream commit 513d9a61156d79dd0979c4ad400c8587f52cbb9d ]

CFI (Control Flow Integrity) is a safety feature allowing the system to detect and react should a potential control flow hijacking occurs. In particular, the Forward-Edge CFI protects indirect function calls by ensuring the prototype of function that is actually called matches the definition of the function hook.

Since Linux now supports CFI, it will be a good idea to fix mismatched return type for implementation of hooks. Otherwise this would get cought out by CFI and cause a panic.

Use enums from netdev_tx_t as return value instead, then change return type to netdev_tx_t. Note that rtllib_xmit_inter() would return 1 only on allocation failure and the queue is stopped if that happens, meeting the documented requirement if NETDEV_TX_BUSY should be returned by ndo_start_xmit.

---
 drivers/staging/rtl8192e/rtllib.h    | 2 +-
 drivers/staging/rtl8192e/rtllib_tx.c | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)
drivers/staging/rtl8192e/rtllib.h +1 -1
index 49bf3ad31f91..0fd000d0cc2f 100644
@@ -1938,7 +1938,7 @@ int rtllib_encrypt_fragment(
 	struct sk_buff *frag,
 	int hdr_len);
 
-int rtllib_xmit(struct sk_buff *skb,  struct net_device *dev);
+netdev_tx_t rtllib_xmit(struct sk_buff *skb,  struct net_device *dev);
 void rtllib_txb_free(struct rtllib_txb *txb);
 
 /* rtllib_rx.c */
drivers/staging/rtl8192e/rtllib_tx.c +2 -2
index 8cddb2e12dc4..34abc4655aaf 100644
@@ -964,9 +964,9 @@ static int rtllib_xmit_inter(struct sk_buff *skb, struct net_device *dev)
 
 }
 
-int rtllib_xmit(struct sk_buff *skb, struct net_device *dev)
+netdev_tx_t rtllib_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	memset(skb->cb, 0, sizeof(skb->cb));
-	return rtllib_xmit_inter(skb, dev);
+	return rtllib_xmit_inter(skb, dev) ? NETDEV_TX_BUSY : NETDEV_TX_OK;
 }
 EXPORT_SYMBOL(rtllib_xmit);
Signature · 1 line
2.35.1
#9 lore raw

Re: [PATCH AUTOSEL 5.4 07/27] staging: rtl8192e: Fix return type for implementation of ndo_start_xmit

Quoted 10 lines · Sasha Levin

From: GUO Zihua <[email protected]>

[ Upstream commit 513d9a61156d79dd0979c4ad400c8587f52cbb9d ]

CFI (Control Flow Integrity) is a safety feature allowing the system to detect and react should a potential control flow hijacking occurs. In particular, the Forward-Edge CFI protects indirect function calls by ensuring the prototype of function that is actually called matches the definition of the function hook.

Quoted 3 lines

Since Linux now supports CFI, it will be a good idea to fix mismatched return type for implementation of hooks. Otherwise this would get cought out by CFI and cause a panic.

Not needed in any stable branch, please drop from all.

thanks,

greg k-h

author maintainer
#10 lore raw

PATCH 8/27 power: supply: adp5061: fix out-of-bounds read in adp5061_get_chg_type()

In reply to Sasha Levin

From: Wei Yongjun <[email protected]>

[ Upstream commit 9d47e01b9d807808224347935562f7043a358054 ]

ADP5061_CHG_STATUS_1_CHG_STATUS is masked with 0x07, which means a length of 8, but adp5061_chg_type array size is 4, may end up reading 4 elements beyond the end of the adp5061_chg_type[] array.

---
 drivers/power/supply/adp5061.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/power/supply/adp5061.c +3 -3
index 003557043ab3..daee1161c305 100644
@@ -427,11 +427,11 @@ static int adp5061_get_chg_type(struct adp5061_state *st,
 	if (ret < 0)
 		return ret;
 
-	chg_type = adp5061_chg_type[ADP5061_CHG_STATUS_1_CHG_STATUS(status1)];
-	if (chg_type > ADP5061_CHG_FAST_CV)
+	chg_type = ADP5061_CHG_STATUS_1_CHG_STATUS(status1);
+	if (chg_type >= ARRAY_SIZE(adp5061_chg_type))
 		val->intval = POWER_SUPPLY_STATUS_UNKNOWN;
 	else
-		val->intval = chg_type;
+		val->intval = adp5061_chg_type[chg_type];
 
 	return ret;
 }
Signature · 1 line
2.35.1
author maintainer
#11 lore raw

PATCH 9/27 staging: vt6655: fix potential memory leak

In reply to Sasha Levin

From: Nam Cao <[email protected]>

[ Upstream commit c8ff91535880d41b49699b3829fb6151942de29e ]

In function device_init_td0_ring, memory is allocated for member td_info of priv->apTD0Rings[i], with i increasing from 0. In case of allocation failure, the memory is freed in reversed order, with i decreasing to 0. However, the case i=0 is left out and thus memory is leaked.

Modify the memory freeing loop to include the case i=0.

---
 drivers/staging/vt6655/device_main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/staging/vt6655/device_main.c +1 -1
index 082302944c37..02bf33ace1eb 100644
@@ -670,7 +670,7 @@ static int device_init_td0_ring(struct vnt_private *priv)
 	return 0;
 
 err_free_desc:
-	while (--i) {
+	while (i--) {
 		desc = &priv->apTD0Rings[i];
 		kfree(desc->td_info);
 	}
Signature · 1 line
2.35.1
author maintainer
#12 lore raw

PATCH 10/27 ata: libahci_platform: Sanity check the DT child nodes number

In reply to Sasha Levin

From: Serge Semin <[email protected]>

[ Upstream commit 3c132ea6508b34956e5ed88d04936983ec230601 ]

Having greater than AHCI_MAX_PORTS (32) ports detected isn't that critical from the further AHCI-platform initialization point of view since exceeding the ports upper limit will cause allocating more resources than will be used afterwards. But detecting too many child DT-nodes doesn't seem right since it's very unlikely to have it on an ordinary platform. In accordance with the AHCI specification there can't be more than 32 ports implemented at least due to having the CAP.NP field of 5 bits wide and the PI register of dword size. Thus if such situation is found the DTB must have been corrupted and the data read from it shouldn't be reliable. Let's consider that as an erroneous situation and halt further resources allocation.

Note it's logically more correct to have the nports set only after the initialization value is checked for being sane. So while at it let's make sure nports is assigned with a correct value.

---
 drivers/ata/libahci_platform.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)
drivers/ata/libahci_platform.c +12 -2
index 8a963d2a951d..c0ac25b80a1f 100644
@@ -451,14 +451,24 @@ struct ahci_host_priv *ahci_platform_get_resources(struct platform_device *pdev,
 		}
 	}
 
-	hpriv->nports = child_nodes = of_get_child_count(dev->of_node);
+	/*
+	 * Too many sub-nodes most likely means having something wrong with
+	 * the firmware.
+	 */
+	child_nodes = of_get_child_count(dev->of_node);
+	if (child_nodes > AHCI_MAX_PORTS) {
+		rc = -EINVAL;
+		goto err_out;
+	}
 
 	/*
 	 * If no sub-node was found, we still need to set nports to
 	 * one in order to be able to use the
 	 * ahci_platform_[en|dis]able_[phys|regulators] functions.
 	 */
-	if (!child_nodes)
+	if (child_nodes)
+		hpriv->nports = child_nodes;
+	else
 		hpriv->nports = 1;
 
 	hpriv->phys = devm_kcalloc(dev, hpriv->nports, sizeof(*hpriv->phys), GFP_KERNEL);
Signature · 1 line
2.35.1
author maintainer
#13 lore raw

PATCH 11/27 bcache: fix set_at_max_writeback_rate() for multiple attached devices

In reply to Sasha Levin

From: Coly Li <[email protected]>

[ Upstream commit d2d05b88035d2d51a5bb6c5afec88a0880c73df4 ]

Inside set_at_max_writeback_rate() the calculation in following if() check is wrong,

	if (atomic_inc_return(&c->idle_counter) <
	    atomic_read(&c->attached_dev_nr) * 6)

Because each attached backing device has its own writeback thread running and increasing c->idle_counter, the counter increates much faster than expected. The correct calculation should be,

	(counter / dev_nr) < dev_nr * 6

which equals to,
counter < dev_nr * dev_nr * 6

This patch fixes the above mistake with correct calculation, and helper routine idle_counter_exceeded() is added to make code be more clear.

---
 drivers/md/bcache/writeback.c | 73 +++++++++++++++++++++++++----------
 1 file changed, 52 insertions(+), 21 deletions(-)
drivers/md/bcache/writeback.c +52 -21
index 0b02210ab435..5767ff6c13e3 100644
@@ -119,27 +119,61 @@ static void __update_writeback_rate(struct cached_dev *dc)
 	dc->writeback_rate_target = target;
 }
 
+static bool idle_counter_exceeded(struct cache_set *c)
+{
+	int counter, dev_nr;
+
+	/*
+	 * If c->idle_counter is overflow (idel for really long time),
+	 * reset as 0 and not set maximum rate this time for code
+	 * simplicity.
+	 */
+	counter = atomic_inc_return(&c->idle_counter);
+	if (counter <= 0) {
+		atomic_set(&c->idle_counter, 0);
+		return false;
+	}
+
+	dev_nr = atomic_read(&c->attached_dev_nr);
+	if (dev_nr == 0)
+		return false;
+
+	/*
+	 * c->idle_counter is increased by writeback thread of all
+	 * attached backing devices, in order to represent a rough
+	 * time period, counter should be divided by dev_nr.
+	 * Otherwise the idle time cannot be larger with more backing
+	 * device attached.
+	 * The following calculation equals to checking
+	 *	(counter / dev_nr) < (dev_nr * 6)
+	 */
+	if (counter < (dev_nr * dev_nr * 6))
+		return false;
+
+	return true;
+}
+
+/*
+ * Idle_counter is increased every time when update_writeback_rate() is
+ * called. If all backing devices attached to the same cache set have
+ * identical dc->writeback_rate_update_seconds values, it is about 6
+ * rounds of update_writeback_rate() on each backing device before
+ * c->at_max_writeback_rate is set to 1, and then max wrteback rate set
+ * to each dc->writeback_rate.rate.
+ * In order to avoid extra locking cost for counting exact dirty cached
+ * devices number, c->attached_dev_nr is used to calculate the idle
+ * throushold. It might be bigger if not all cached device are in write-
+ * back mode, but it still works well with limited extra rounds of
+ * update_writeback_rate().
+ */
 static bool set_at_max_writeback_rate(struct cache_set *c,
 				       struct cached_dev *dc)
 {
 	/* Don't set max writeback rate if gc is running */
 	if (!c->gc_mark_valid)
 		return false;
-	/*
-	 * Idle_counter is increased everytime when update_writeback_rate() is
-	 * called. If all backing devices attached to the same cache set have
-	 * identical dc->writeback_rate_update_seconds values, it is about 6
-	 * rounds of update_writeback_rate() on each backing device before
-	 * c->at_max_writeback_rate is set to 1, and then max wrteback rate set
-	 * to each dc->writeback_rate.rate.
-	 * In order to avoid extra locking cost for counting exact dirty cached
-	 * devices number, c->attached_dev_nr is used to calculate the idle
-	 * throushold. It might be bigger if not all cached device are in write-
-	 * back mode, but it still works well with limited extra rounds of
-	 * update_writeback_rate().
-	 */
-	if (atomic_inc_return(&c->idle_counter) <
-	    atomic_read(&c->attached_dev_nr) * 6)
+
+	if (!idle_counter_exceeded(c))
 		return false;
 
 	if (atomic_read(&c->at_max_writeback_rate) != 1)
@@ -153,13 +187,10 @@ static bool set_at_max_writeback_rate(struct cache_set *c,
 	dc->writeback_rate_change = 0;
 
 	/*
-	 * Check c->idle_counter and c->at_max_writeback_rate agagain in case
-	 * new I/O arrives during before set_at_max_writeback_rate() returns.
-	 * Then the writeback rate is set to 1, and its new value should be
-	 * decided via __update_writeback_rate().
+	 * In case new I/O arrives during before
+	 * set_at_max_writeback_rate() returns.
 	 */
-	if ((atomic_read(&c->idle_counter) <
-	     atomic_read(&c->attached_dev_nr) * 6) ||
+	if (!idle_counter_exceeded(c) ||
 	    !atomic_read(&c->at_max_writeback_rate))
 		return false;
 
Signature · 1 line
2.35.1
author maintainer
#14 lore raw

PATCH 12/27 hid: topre: Add driver fixing report descriptor

In reply to Sasha Levin

From: Harry Stern <[email protected]>

[ Upstream commit a109d5c45b3d6728b9430716b915afbe16eef27c ]

The Topre REALFORCE R2 firmware incorrectly reports that interface descriptor number 1, input report descriptor 2's events are array events rather than variable events. That particular report descriptor is used to report keypresses when there are more than 6 keys held at a time. This bug prevents events from this interface from being registered properly, so only 6 keypresses (from a different interface) can be registered at once, rather than full n-key rollover.

This commit fixes the bug by setting the correct value in a report_fixup function.

The original bug report can be found here:

Thanks to Benjamin Tissoires for diagnosing the issue with the report descriptor.

---
 drivers/hid/Kconfig     |  6 +++++
 drivers/hid/Makefile    |  1 +
 drivers/hid/hid-ids.h   |  3 +++
 drivers/hid/hid-topre.c | 49 +++++++++++++++++++++++++++++++++++++++++
 4 files changed, 59 insertions(+)
 create mode 100644 drivers/hid/hid-topre.c
drivers/hid/Kconfig +6 -0
index 5169a38ee47a..56315a0be78a 100644
@@ -992,6 +992,12 @@ config HID_TOPSEED
 	Say Y if you have a TopSeed Cyberlink or BTC Emprex or Conceptronic
 	CLLRCMCE remote control.
 
+config HID_TOPRE
+	tristate "Topre REALFORCE keyboards"
+	depends on HID
+	help
+	  Say Y for N-key rollover support on Topre REALFORCE R2 108 key keyboards.
+
 config HID_THINGM
 	tristate "ThingM blink(1) USB RGB LED"
 	depends on HID
drivers/hid/Makefile +1 -0
index 0c03308cfb08..bb618c74c442 100644
@@ -110,6 +110,7 @@ obj-$(CONFIG_HID_GREENASIA)	+= hid-gaff.o
 obj-$(CONFIG_HID_THRUSTMASTER)	+= hid-tmff.o
 obj-$(CONFIG_HID_TIVO)		+= hid-tivo.o
 obj-$(CONFIG_HID_TOPSEED)	+= hid-topseed.o
+obj-$(CONFIG_HID_TOPRE)	+= hid-topre.o
 obj-$(CONFIG_HID_TWINHAN)	+= hid-twinhan.o
 obj-$(CONFIG_HID_U2FZERO)	+= hid-u2fzero.o
 hid-uclogic-objs		:= hid-uclogic-core.o \
drivers/hid/hid-ids.h +3 -0
index c587a77d493c..fbceead99b3c 100644
@@ -1156,6 +1156,9 @@
 #define USB_DEVICE_ID_TIVO_SLIDE	0x1201
 #define USB_DEVICE_ID_TIVO_SLIDE_PRO	0x1203
 
+#define USB_VENDOR_ID_TOPRE			0x0853
+#define USB_DEVICE_ID_TOPRE_REALFORCE_R2_108			0x0148
+
 #define USB_VENDOR_ID_TOPSEED		0x0766
 #define USB_DEVICE_ID_TOPSEED_CYBERLINK	0x0204
 
drivers/hid/hid-topre.c new file +49 -0

49 lines not on this page · raw message

Signature · 1 line
2.35.1
author maintainer
#15 lore raw

PATCH 13/27 HID: roccat: Fix use-after-free in roccat_read()

In reply to Sasha Levin

From: Hyunwoo Kim <[email protected]>

[ Upstream commit cacdb14b1c8d3804a3a7d31773bc7569837b71a4 ]

roccat_report_event() is responsible for registering roccat-related reports in struct roccat_device.

int roccat_report_event(int minor, u8 const *data)

{
	struct roccat_device *device;
	struct roccat_reader *reader;
	struct roccat_report *report;
	uint8_t *new_value;

	device = devices[minor];

	new_value = kmemdup(data, device->report_size, GFP_ATOMIC);
	if (!new_value)
		return -ENOMEM;

	report = &device->cbuf[device->cbuf_end];

	/* passing NULL is safe */
	kfree(report->value);
	...

The registered report is stored in the struct roccat_device member "struct roccat_report cbuf[ROCCAT_CBUF_SIZE];".
If more reports are received than the "ROCCAT_CBUF_SIZE" value, kfree() the saved report from cbuf[0] and allocates a new reprot. Since there is no lock when this kfree() is performed, kfree() can be performed even while reading the saved report.

static ssize_t roccat_read(struct file *file, char __user *buffer,
size_t count, loff_t *ppos)

{
	struct roccat_reader *reader = file->private_data;
	struct roccat_device *device = reader->device;
	struct roccat_report *report;
	ssize_t retval = 0, len;
	DECLARE_WAITQUEUE(wait, current);

	mutex_lock(&device->cbuf_lock);

	...

	report = &device->cbuf[reader->cbuf_start];
	/*
	 * If report is larger than requested amount of data, rest of report
	 * is lost!
	 */
	len = device->report_size > count ? count : device->report_size;

	if (copy_to_user(buffer, report->value, len)) {
		retval = -EFAULT;
		goto exit_unlock;
	}
	...

The roccat_read() function receives the device->cbuf report and delivers it to the user through copy_to_user().
If the N+ROCCAT_CBUF_SIZE th report is received while copying of the Nth report->value is in progress, the pointer that copy_to_user() is working on is kfree()ed and UAF read may occur. (race condition)

Since the device node of this driver does not set separate permissions, this is not a security vulnerability, but because it is used for requesting screen display of profile or dpi settings,
a user using the roccat device can apply udev to this device node or There is a possibility to use it by giving.

---
 drivers/hid/hid-roccat.c | 4 ++++
 1 file changed, 4 insertions(+)
drivers/hid/hid-roccat.c +4 -0
index 26373b82fe81..6da80e442fdd 100644
@@ -257,6 +257,8 @@ int roccat_report_event(int minor, u8 const *data)
 	if (!new_value)
 		return -ENOMEM;
 
+	mutex_lock(&device->cbuf_lock);
+
 	report = &device->cbuf[device->cbuf_end];
 
 	/* passing NULL is safe */
@@ -276,6 +278,8 @@ int roccat_report_event(int minor, u8 const *data)
 			reader->cbuf_start = (reader->cbuf_start + 1) % ROCCAT_CBUF_SIZE;
 	}
 
+	mutex_unlock(&device->cbuf_lock);
+
 	wake_up_interruptible(&device->wait);
 	return 0;
 }
Signature · 1 line
2.35.1
author maintainer
#16 lore raw

PATCH 14/27 HSI: ssi_protocol: fix potential resource leak in ssip_pn_open()

In reply to Sasha Levin

From: Jianglei Nie <[email protected]>

[ Upstream commit b28dbcb379e6a7f80262c2732a57681b1ee548ca ]

ssip_pn_open() claims the HSI client's port with hsi_claim_port(). When hsi_register_port_event() gets some error and returns a negetive value, the HSI client's port should be released with hsi_release_port().

Fix it by calling hsi_release_port() when hsi_register_port_event() fails.

---
 drivers/hsi/clients/ssi_protocol.c | 1 +
 1 file changed, 1 insertion(+)
drivers/hsi/clients/ssi_protocol.c +1 -0
index 365b5d5967ac..01a2e861c399 100644
@@ -931,6 +931,7 @@ static int ssip_pn_open(struct net_device *dev)
 	if (err < 0) {
 		dev_err(&cl->device, "Register HSI port event failed (%d)\n",
 			err);
+		hsi_release_port(cl);
 		return err;
 	}
 	dev_dbg(&cl->device, "Configuring SSI port\n");
Signature · 1 line
2.35.1
author maintainer
#17 lore raw

PATCH 15/27 md/raid5: Wait for MD_SB_CHANGE_PENDING in raid5d

In reply to Sasha Levin

From: Logan Gunthorpe <[email protected]>

[ Upstream commit 5e2cf333b7bd5d3e62595a44d598a254c697cd74 ]

A complicated deadlock exists when using the journal and an elevated group_thrtead_cnt. It was found with loop devices, but its not clear whether it can be seen with real disks. The deadlock can occur simply by writing data with an fio script.

When the deadlock occurs, multiple threads will hang in different ways:

1) The group threads will hang in the blk-wbt code with bios waiting to

    be submitted to the block layer:

        io_schedule+0x70/0xb0
        rq_qos_wait+0x153/0x210
        wbt_wait+0x115/0x1b0
        io_schedule+0x70/0xb0
        rq_qos_wait+0x153/0x210
        wbt_wait+0x115/0x1b0
        __rq_qos_throttle+0x38/0x60
        blk_mq_submit_bio+0x589/0xcd0
        wbt_wait+0x115/0x1b0
        __rq_qos_throttle+0x38/0x60
        blk_mq_submit_bio+0x589/0xcd0
        __submit_bio+0xe6/0x100
        submit_bio_noacct_nocheck+0x42e/0x470
        submit_bio_noacct+0x4c2/0xbb0
        ops_run_io+0x46b/0x1a30
        handle_stripe+0xcd3/0x36b0
        handle_active_stripes.constprop.0+0x6f6/0xa60
        raid5_do_work+0x177/0x330

    Or:
        io_schedule+0x70/0xb0
        rq_qos_wait+0x153/0x210
        wbt_wait+0x115/0x1b0
        __rq_qos_throttle+0x38/0x60
        blk_mq_submit_bio+0x589/0xcd0
        __submit_bio+0xe6/0x100
        submit_bio_noacct_nocheck+0x42e/0x470
        submit_bio_noacct+0x4c2/0xbb0
        flush_deferred_bios+0x136/0x170
        raid5_do_work+0x262/0x330

2) The r5l_reclaim thread will hang in the same way, submitting a

    bio to the block layer:

        io_schedule+0x70/0xb0
        rq_qos_wait+0x153/0x210
        wbt_wait+0x115/0x1b0
        __rq_qos_throttle+0x38/0x60
        blk_mq_submit_bio+0x589/0xcd0
        __submit_bio+0xe6/0x100
        submit_bio_noacct_nocheck+0x42e/0x470
        submit_bio_noacct+0x4c2/0xbb0
        submit_bio+0x3f/0xf0
        md_super_write+0x12f/0x1b0
        md_update_sb.part.0+0x7c6/0xff0
        md_update_sb+0x30/0x60
        r5l_do_reclaim+0x4f9/0x5e0
        r5l_reclaim_thread+0x69/0x30b

    However, before hanging, the MD_SB_CHANGE_PENDING flag will be
    set for sb_flags in r5l_write_super_and_discard_space(). This
    flag will never be cleared because the submit_bio() call never
    returns.

3) Due to the MD_SB_CHANGE_PENDING flag being set, handle_stripe() will do no processing on any pending stripes and re-set STRIPE_HANDLE. This will cause the raid5d thread to enter an infinite loop, constantly trying to handle the same stripes stuck in the queue.

The raid5d thread has a blk_plug that holds a number of bios that are also stuck waiting seeing the thread is in a loop that never schedules. These bios have been accounted for by blk-wbt thus preventing the other threads above from continuing when they try to submit bios. --Deadlock.

To fix this, add the same wait_event() that is used in raid5_do_work() to raid5d() such that if MD_SB_CHANGE_PENDING is set, the thread will schedule and wait until the flag is cleared. The schedule action will flush the plug which will allow the r5l_reclaim thread to continue, thus preventing the deadlock.

However, md_check_recovery() calls can also clear MD_SB_CHANGE_PENDING from the same thread and can thus deadlock if the thread is put to sleep. So avoid waiting if md_check_recovery() is being called in the loop.

It's not clear when the deadlock was introduced, but the similar wait_event() call in raid5_do_work() was added in 2017 by this commit:

16d997b78b15 ("md/raid5: simplfy delaying of writes while metadata
is updated.")

---
 drivers/md/raid5.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)
drivers/md/raid5.c +12 -0
index fe99e8cdc026..24d91db73503 100644
@@ -36,6 +36,7 @@
  */
 
 #include <linux/blkdev.h>
+#include <linux/delay.h>
 #include <linux/kthread.h>
 #include <linux/raid/pq.h>
 #include <linux/async_tx.h>
@@ -6334,7 +6335,18 @@ static void raid5d(struct md_thread *thread)
 			spin_unlock_irq(&conf->device_lock);
 			md_check_recovery(mddev);
 			spin_lock_irq(&conf->device_lock);
+
+			/*
+			 * Waiting on MD_SB_CHANGE_PENDING below may deadlock
+			 * seeing md_check_recovery() is needed to clear
+			 * the flag when using mdmon.
+			 */
+			continue;
 		}
+
+		wait_event_lock_irq(mddev->sb_wait,
+			!test_bit(MD_SB_CHANGE_PENDING, &mddev->sb_flags),
+			conf->device_lock);
 	}
 	pr_debug("%d stripes handled\n", handled);
 
Signature · 1 line
2.35.1
author maintainer
#18 lore raw

PATCH 16/27 usb: host: xhci: Fix potential memory leak in xhci_alloc_stream_info()

In reply to Sasha Levin

From: Jianglei Nie <[email protected]>

[ Upstream commit 7e271f42a5cc3768cd2622b929ba66859ae21f97 ]

xhci_alloc_stream_info() allocates stream context array for stream_info ->stream_ctx_array with xhci_alloc_stream_ctx(). When some error occurs, stream_info->stream_ctx_array is not released, which will lead to a memory leak.

We can fix it by releasing the stream_info->stream_ctx_array with xhci_free_stream_ctx() on the error path to avoid the potential memory leak.

---
 drivers/usb/host/xhci-mem.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/usb/host/xhci-mem.c +6 -1
index ef23a69c6553..6125a98ffbf5 100644
@@ -650,7 +650,7 @@ struct xhci_stream_info *xhci_alloc_stream_info(struct xhci_hcd *xhci,
 			num_stream_ctxs, &stream_info->ctx_array_dma,
 			mem_flags);
 	if (!stream_info->stream_ctx_array)
-		goto cleanup_ctx;
+		goto cleanup_ring_array;
 	memset(stream_info->stream_ctx_array, 0,
 			sizeof(struct xhci_stream_ctx)*num_stream_ctxs);
 
@@ -711,6 +711,11 @@ struct xhci_stream_info *xhci_alloc_stream_info(struct xhci_hcd *xhci,
 	}
 	xhci_free_command(xhci, stream_info->free_streams_command);
 cleanup_ctx:
+	xhci_free_stream_ctx(xhci,
+		stream_info->num_stream_ctxs,
+		stream_info->stream_ctx_array,
+		stream_info->ctx_array_dma);
+cleanup_ring_array:
 	kfree(stream_info->stream_rings);
 cleanup_info:
 	kfree(stream_info);
Signature · 1 line
2.35.1
author maintainer
#19 lore raw

PATCH 17/27 usb: musb: Fix musb_gadget.c rxstate overflow bug

In reply to Sasha Levin

From: Robin Guo <[email protected]>

[ Upstream commit eea4c860c3b366369eff0489d94ee4f0571d467d ]

The usb function device call musb_gadget_queue() adds the passed request to musb_ep::req_list,If the (request->length > musb_ep->packet_sz) and (is_buffer_mapped(req) return false),the rxstate() will copy all data in fifo to request->buf which may cause request->buf out of bounds.

Fix it by add the length check :
fifocnt = min_t(unsigned, request->length - request->actual, fifocnt);

---
 drivers/usb/musb/musb_gadget.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/usb/musb/musb_gadget.c +3 -0
index 4622400ba4dd..8e83995fc3bd 100644
@@ -760,6 +760,9 @@ static void rxstate(struct musb *musb, struct musb_request *req)
 			musb_writew(epio, MUSB_RXCSR, csr);
 
 buffer_aint_mapped:
+			fifo_count = min_t(unsigned int,
+					request->length - request->actual,
+					(unsigned int)fifo_count);
 			musb_read_fifo(musb_ep->hw_ep, fifo_count, (u8 *)
 					(request->buf + request->actual));
 			request->actual += fifo_count;
Signature · 1 line
2.35.1
author maintainer
#20 lore raw

PATCH 18/27 Revert "usb: storage: Add quirk for Samsung Fit flash"

In reply to Sasha Levin

From: sunghwan jung <[email protected]>

[ Upstream commit ad5dbfc123e6ffbbde194e2a4603323e09f741ee ]

This reverts commit 86d92f5465958752481269348d474414dccb1552, which fix the timeout issue for "Samsung Fit Flash".

But the commit affects not only "Samsung Fit Flash" but also other usb storages that use the same controller and causes severe performance regression.

 # hdparm -t /dev/sda (without the quirk)
 Timing buffered disk reads: 622 MB in  3.01 seconds = 206.66 MB/sec
 # hdparm -t /dev/sda (with the quirk)
 Timing buffered disk reads: 220 MB in  3.00 seconds =  73.32 MB/sec

The commit author mentioned that "Issue was reproduced after device has bad block", so this quirk should be applied when we have the timeout issue with a device that has bad blocks.

We revert the commit so that we apply this quirk by adding kernel paramters using a bootloader or other ways when we really need it, without the performance regression with devices that don't have the issue.

---
 drivers/usb/storage/unusual_devs.h | 6 ------
 1 file changed, 6 deletions(-)
drivers/usb/storage/unusual_devs.h +0 -6
index 6a59950a63a0..b270be141b8e 100644
@@ -1275,12 +1275,6 @@ UNUSUAL_DEV( 0x090a, 0x1200, 0x0000, 0x9999,
 		USB_SC_RBC, USB_PR_BULK, NULL,
 		0 ),
 
-UNUSUAL_DEV(0x090c, 0x1000, 0x1100, 0x1100,
-		"Samsung",
-		"Flash Drive FIT",
-		USB_SC_DEVICE, USB_PR_DEVICE, NULL,
-		US_FL_MAX_SECTORS_64),
-
 /* aeb */
 UNUSUAL_DEV( 0x090c, 0x1132, 0x0000, 0xffff,
 		"Feiya",
Signature · 1 line
2.35.1
author maintainer
#21 lore raw

PATCH 19/27 staging: rtl8723bs: fix a potential memory leak in rtw_init_cmd_priv()

In reply to Sasha Levin

From: Xiaoke Wang <[email protected]>

[ Upstream commit 708056fba733a73d926772ea4ce9a42d240345da ]

In rtw_init_cmd_priv(), if pcmdpriv->rsp_allocated_buf is allocated in failure, then pcmdpriv->cmd_allocated_buf will be not properly released. Besides, considering there are only two error paths and the first one can directly return, so we do not need implicitly jump to the exit tag to execute the error handler.

So this patch added kfree(pcmdpriv->cmd_allocated_buf); on the error path to release the resource and simplified the return logic of rtw_init_cmd_priv(). As there is no proper device to test with, no runtime testing was performed.

---
 drivers/staging/rtl8723bs/core/rtw_cmd.c | 16 ++++++----------
 1 file changed, 6 insertions(+), 10 deletions(-)
drivers/staging/rtl8723bs/core/rtw_cmd.c +6 -10
index 8d93c2f26890..a82114de21a7 100644
@@ -165,8 +165,6 @@ No irqsave is necessary.
 
 int rtw_init_cmd_priv(struct	cmd_priv *pcmdpriv)
 {
-	int res = 0;
-
 	init_completion(&pcmdpriv->cmd_queue_comp);
 	init_completion(&pcmdpriv->terminate_cmdthread_comp);
 
@@ -178,18 +176,16 @@ int rtw_init_cmd_priv(struct	cmd_priv *pcmdpriv)
 
 	pcmdpriv->cmd_allocated_buf = rtw_zmalloc(MAX_CMDSZ + CMDBUFF_ALIGN_SZ);
 
-	if (!pcmdpriv->cmd_allocated_buf) {
-		res = -ENOMEM;
-		goto exit;
-	}
+	if (!pcmdpriv->cmd_allocated_buf)
+		return -ENOMEM;
 
 	pcmdpriv->cmd_buf = pcmdpriv->cmd_allocated_buf  +  CMDBUFF_ALIGN_SZ - ((SIZE_PTR)(pcmdpriv->cmd_allocated_buf) & (CMDBUFF_ALIGN_SZ-1));
 
 	pcmdpriv->rsp_allocated_buf = rtw_zmalloc(MAX_RSPSZ + 4);
 
 	if (!pcmdpriv->rsp_allocated_buf) {
-		res = -ENOMEM;
-		goto exit;
+		kfree(pcmdpriv->cmd_allocated_buf);
+		return -ENOMEM;
 	}
 
 	pcmdpriv->rsp_buf = pcmdpriv->rsp_allocated_buf  +  4 - ((SIZE_PTR)(pcmdpriv->rsp_allocated_buf) & 3);
@@ -197,8 +193,8 @@ int rtw_init_cmd_priv(struct	cmd_priv *pcmdpriv)
 	pcmdpriv->cmd_issued_cnt = pcmdpriv->cmd_done_cnt = pcmdpriv->rsp_cnt = 0;
 
 	mutex_init(&pcmdpriv->sctx_mutex);
-exit:
-	return res;
+
+	return 0;
 }
 
 static void c2h_wk_callback(_workitem *work);
Signature · 1 line
2.35.1
author maintainer
#22 lore raw

PATCH 20/27 staging: rtl8192u: Fix return type of ieee80211_xmit

In reply to Sasha Levin

From: Nathan Huckleberry <[email protected]>

[ Upstream commit 2851349ac351010a2649e0ff86a1e3d68fe5d683 ]

The ndo_start_xmit field in net_device_ops is expected to be of type netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb, struct net_device *dev).

The mismatched return type breaks forward edge kCFI since the underlying function definition does not match the function hook definition.

The return type of ieee80211_xmit should be changed from int to netdev_tx_t.

---
 drivers/staging/rtl8192u/ieee80211/ieee80211.h    | 2 +-
 drivers/staging/rtl8192u/ieee80211/ieee80211_tx.c | 6 +++---
 2 files changed, 4 insertions(+), 4 deletions(-)
drivers/staging/rtl8192u/ieee80211/ieee80211.h +1 -1
index 9576b647f6b1..34f9a23feddb 100644
@@ -2178,7 +2178,7 @@ int ieee80211_set_encryption(struct ieee80211_device *ieee);
 int ieee80211_encrypt_fragment(struct ieee80211_device *ieee,
 			       struct sk_buff *frag, int hdr_len);
 
-int ieee80211_xmit(struct sk_buff *skb, struct net_device *dev);
+netdev_tx_t ieee80211_xmit(struct sk_buff *skb, struct net_device *dev);
 void ieee80211_txb_free(struct ieee80211_txb *txb);
 
 
drivers/staging/rtl8192u/ieee80211/ieee80211_tx.c +3 -3
index f0b6b8372f91..035b40a3e5e5 100644
@@ -527,7 +527,7 @@ static void ieee80211_query_seqnum(struct ieee80211_device *ieee,
 	}
 }
 
-int ieee80211_xmit(struct sk_buff *skb, struct net_device *dev)
+netdev_tx_t ieee80211_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	struct ieee80211_device *ieee = netdev_priv(dev);
 	struct ieee80211_txb *txb = NULL;
@@ -826,13 +826,13 @@ int ieee80211_xmit(struct sk_buff *skb, struct net_device *dev)
 			if ((*ieee->hard_start_xmit)(txb, dev) == 0) {
 				stats->tx_packets++;
 				stats->tx_bytes += __le16_to_cpu(txb->payload_size);
-				return 0;
+				return NETDEV_TX_OK;
 			}
 			ieee80211_txb_free(txb);
 		}
 	}
 
-	return 0;
+	return NETDEV_TX_OK;
 
  failed:
 	spin_unlock_irqrestore(&ieee->lock, flags);
Signature · 1 line
2.35.1
author maintainer
#23 lore raw

PATCH 21/27 staging: octeon: Fix return type of cvm_oct_xmit and cvm_oct_xmit_pow

In reply to Sasha Levin

From: Nathan Huckleberry <[email protected]>

[ Upstream commit b77599043f00fce9253d0f22522c5d5b521555ce ]

The ndo_start_xmit field in net_device_ops is expected to be of type netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb, struct net_device *dev).

The mismatched return type breaks forward edge kCFI since the underlying function definition does not match the function hook definition.

The return type of cvm_oct_xmit and cvm_oct_xmit_pow should be changed from int to netdev_tx_t.

---
 drivers/staging/octeon/ethernet-tx.c | 4 ++--
 drivers/staging/octeon/ethernet-tx.h | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)
drivers/staging/octeon/ethernet-tx.c +2 -2
index fe6e1ae73460..15568bf2ca56 100644
@@ -125,7 +125,7 @@ static void cvm_oct_free_tx_skbs(struct net_device *dev)
  *
  * Returns Always returns NETDEV_TX_OK
  */
-int cvm_oct_xmit(struct sk_buff *skb, struct net_device *dev)
+netdev_tx_t cvm_oct_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	cvmx_pko_command_word0_t pko_command;
 	union cvmx_buf_ptr hw_buffer;
@@ -507,7 +507,7 @@ int cvm_oct_xmit(struct sk_buff *skb, struct net_device *dev)
 
  * Returns Always returns zero
  */
-int cvm_oct_xmit_pow(struct sk_buff *skb, struct net_device *dev)
+netdev_tx_t cvm_oct_xmit_pow(struct sk_buff *skb, struct net_device *dev)
 {
 	struct octeon_ethernet *priv = netdev_priv(dev);
 	void *packet_buffer;
drivers/staging/octeon/ethernet-tx.h +2 -2
index 78936e9b33b0..6c524668f65a 100644
@@ -5,8 +5,8 @@
  * Copyright (c) 2003-2007 Cavium Networks
  */
 
-int cvm_oct_xmit(struct sk_buff *skb, struct net_device *dev);
-int cvm_oct_xmit_pow(struct sk_buff *skb, struct net_device *dev);
+netdev_tx_t cvm_oct_xmit(struct sk_buff *skb, struct net_device *dev);
+netdev_tx_t cvm_oct_xmit_pow(struct sk_buff *skb, struct net_device *dev);
 int cvm_oct_transmit_qos(struct net_device *dev, void *work_queue_entry,
 			 int do_free, int qos);
 void cvm_oct_tx_initialize(void);
Signature · 1 line
2.35.1
author maintainer
#24 lore raw

PATCH 22/27 nvme: copy firmware_rev on each init

In reply to Sasha Levin

From: Keith Busch <[email protected]>

[ Upstream commit a8eb6c1ba48bddea82e8d74cbe6e119f006be97d ]

The firmware revision can change on after a reset so copy the most recent info each time instead of just the first time, otherwise the sysfs firmware_rev entry may contain stale data.

---
 drivers/nvme/host/core.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/nvme/host/core.c +2 -1
index f1717f34b2f3..6627fb531f33 100644
@@ -2671,7 +2671,6 @@ static int nvme_init_subsystem(struct nvme_ctrl *ctrl, struct nvme_id_ctrl *id)
 	nvme_init_subnqn(subsys, ctrl, id);
 	memcpy(subsys->serial, id->sn, sizeof(subsys->serial));
 	memcpy(subsys->model, id->mn, sizeof(subsys->model));
-	memcpy(subsys->firmware_rev, id->fr, sizeof(subsys->firmware_rev));
 	subsys->vendor_id = le16_to_cpu(id->vid);
 	subsys->cmic = id->cmic;
 	subsys->awupf = le16_to_cpu(id->awupf);
@@ -2824,6 +2823,8 @@ int nvme_init_identify(struct nvme_ctrl *ctrl)
 				ctrl->quirks |= core_quirks[i].quirks;
 		}
 	}
+	memcpy(ctrl->subsys->firmware_rev, id->fr,
+	       sizeof(ctrl->subsys->firmware_rev));
 
 	if (force_apst && (ctrl->quirks & NVME_QUIRK_NO_DEEPEST_PS)) {
 		dev_warn(ctrl->device, "forcibly allowing all power states due to nvme_core.force_apst -- use at your own risk\n");
Signature · 1 line
2.35.1
author maintainer
#25 lore raw

PATCH 23/27 nvmet-tcp: add bounds check on Transfer Tag

In reply to Sasha Levin

From: Varun Prakash <[email protected]>

[ Upstream commit b6a545ffa2c192b1e6da4a7924edac5ba9f4ea2b ]

ttag is used as an index to get cmd in nvmet_tcp_handle_h2c_data_pdu(), add a bounds check to avoid out-of-bounds access.

---
 drivers/nvme/target/tcp.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)
drivers/nvme/target/tcp.c +9 -2
index eb5b39c2bba8..df7a911d303f 100644
@@ -858,10 +858,17 @@ static int nvmet_tcp_handle_h2c_data_pdu(struct nvmet_tcp_queue *queue)
 	struct nvme_tcp_data_pdu *data = &queue->pdu.data;
 	struct nvmet_tcp_cmd *cmd;
 
-	if (likely(queue->nr_cmds))
+	if (likely(queue->nr_cmds)) {
+		if (unlikely(data->ttag >= queue->nr_cmds)) {
+			pr_err("queue %d: received out of bound ttag %u, nr_cmds %u\n",
+				queue->idx, data->ttag, queue->nr_cmds);
+			nvmet_tcp_fatal_error(queue);
+			return -EPROTO;
+		}
 		cmd = &queue->cmds[data->ttag];
-	else
+	} else {
 		cmd = &queue->connect;
+	}
 
 	if (le32_to_cpu(data->data_offset) != cmd->rbytes_done) {
 		pr_err("ttag %u unexpected data offset %u (expected %u)\n",
Signature · 1 line
2.35.1
author maintainer
#26 lore raw

PATCH 24/27 usb: idmouse: fix an uninit-value in idmouse_open

In reply to Sasha Levin

From: Dongliang Mu <[email protected]>

[ Upstream commit bce2b0539933e485d22d6f6f076c0fcd6f185c4c ]

In idmouse_create_image, if any ftip_command fails, it will go to the reset label. However, this leads to the data in bulk_in_buffer[HEADER..IMGSIZE] uninitialized. And the check for valid image incurs an uninitialized dereference.

Fix this by moving the check before reset label since this check only be valid if the data after bulk_in_buffer[HEADER] has concrete data.

Note that this is found by KMSAN, so only kernel compilation is tested.

---
 drivers/usb/misc/idmouse.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/usb/misc/idmouse.c +4 -4
index bb24527f3c70..ba2b6fbab9b8 100644
@@ -178,10 +178,6 @@ static int idmouse_create_image(struct usb_idmouse *dev)
 		bytes_read += bulk_read;
 	}
 
-	/* reset the device */
-reset:
-	ftip_command(dev, FTIP_RELEASE, 0, 0);
-
 	/* check for valid image */
 	/* right border should be black (0x00) */
 	for (bytes_read = sizeof(HEADER)-1 + WIDTH-1; bytes_read < IMGSIZE; bytes_read += WIDTH)
@@ -193,6 +189,10 @@ static int idmouse_create_image(struct usb_idmouse *dev)
 		if (dev->bulk_in_buffer[bytes_read] != 0xFF)
 			return -EAGAIN;
 
+	/* reset the device */
+reset:
+	ftip_command(dev, FTIP_RELEASE, 0, 0);
+
 	/* should be IMGSIZE == 65040 */
 	dev_dbg(&dev->interface->dev, "read %d bytes fingerprint data\n",
 		bytes_read);
Signature · 1 line
2.35.1
author maintainer
#27 lore raw

PATCH 25/27 fsi: master-ast-cf: Fix missing of_node_put in fsi_master_acf_probe

In reply to Sasha Levin

From: Lv Ruyi <[email protected]>

[ Upstream commit 182d98e00e4745fe253cb0c24c63bbac253464a2 ]

of_parse_phandle returns node pointer with refcount incremented, use of_node_put() on it when done.

---
 drivers/fsi/fsi-master-ast-cf.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/fsi/fsi-master-ast-cf.c +2 -0
index 04d10ea8d343..1d39e435bfd1 100644
@@ -1324,12 +1324,14 @@ static int fsi_master_acf_probe(struct platform_device *pdev)
 		}
 		master->cvic = devm_of_iomap(&pdev->dev, np, 0, NULL);
 		if (IS_ERR(master->cvic)) {
+			of_node_put(np);
 			rc = PTR_ERR(master->cvic);
 			dev_err(&pdev->dev, "Error %d mapping CVIC\n", rc);
 			goto err_free;
 		}
 		rc = of_property_read_u32(np, "copro-sw-interrupts",
 					  &master->cvic_sw_irq);
+		of_node_put(np);
 		if (rc) {
 			dev_err(&pdev->dev, "Can't find coprocessor SW interrupt\n");
 			goto err_free;
Signature · 1 line
2.35.1
author maintainer
#28 lore raw

PATCH 26/27 sbitmap: fix lockup while swapping

In reply to Sasha Levin

From: Hugh Dickins <[email protected]>

[ Upstream commit 30514bd2dd4e86a3ecfd6a93a3eadf7b9ea164a0 ]

Commit 4acb83417cad ("sbitmap: fix batched wait_cnt accounting") is a big improvement: without it, I had to revert to before commit 040b83fcecfb ("sbitmap: fix possible io hung due to lost wakeup") to avoid the high system time and freezes which that had introduced.

Now okay on the NVME laptop, but 4acb83417cad is a disaster for heavy swapping (kernel builds in low memory) on another: soon locking up in sbitmap_queue_wake_up() (into which __sbq_wake_up() is inlined), cycling around with waitqueue_active() but wait_cnt 0 . Here is a backtrace, showing the common pattern of outer sbitmap_queue_wake_up() interrupted before setting wait_cnt 0 back to wake_batch (in some cases other CPUs are idle, in other cases they're spinning for a lock in dd_bio_merge()):

sbitmap_queue_wake_up < sbitmap_queue_clear < blk_mq_put_tag < __blk_mq_free_request < blk_mq_free_request < __blk_mq_end_request < scsi_end_request < scsi_io_completion < scsi_finish_command < scsi_complete < blk_complete_reqs < blk_done_softirq < __do_softirq < __irq_exit_rcu < irq_exit_rcu < common_interrupt < asm_common_interrupt < _raw_spin_unlock_irqrestore < __wake_up_common_lock < __wake_up < sbitmap_queue_wake_up < sbitmap_queue_clear < blk_mq_put_tag < __blk_mq_free_request < blk_mq_free_request < dd_bio_merge < blk_mq_sched_bio_merge < blk_mq_attempt_bio_merge < blk_mq_submit_bio < __submit_bio < submit_bio_noacct_nocheck < submit_bio_noacct < submit_bio < __swap_writepage < swap_writepage < pageout < shrink_folio_list < evict_folios < lru_gen_shrink_lruvec < shrink_lruvec < shrink_node < do_try_to_free_pages < try_to_free_pages < __alloc_pages_slowpath < __alloc_pages < folio_alloc < vma_alloc_folio < do_anonymous_page < __handle_mm_fault < handle_mm_fault < do_user_addr_fault < exc_page_fault < asm_exc_page_fault

See how the process-context sbitmap_queue_wake_up() has been interrupted, after bringing wait_cnt down to 0 (and in this example, after doing its wakeups), before advancing wake_index and refilling wake_cnt: an interrupt-context sbitmap_queue_wake_up() of the same sbq gets stuck.

I have almost no grasp of all the possible sbitmap races, and their
consequences: but __sbq_wake_up() can do nothing useful while wait_cnt 0, so it is better if sbq_wake_ptr() skips on to the next ws in that case:
which fixes the lockup and shows no adverse consequence for me.

The check for wait_cnt being 0 is obviously racy, and ultimately can lead to lost wakeups: for example, when there is only a single waitqueue with waiters. However, lost wakeups are unlikely to matter in these cases, and a proper fix requires redesign (and benchmarking) of the batched wakeup code: so let's plug the hole with this bandaid for now.

---
 lib/sbitmap.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
lib/sbitmap.c +1 -1
index ee3ce1494568..3e277a57254e 100644
@@ -513,7 +513,7 @@ static struct sbq_wait_state *sbq_wake_ptr(struct sbitmap_queue *sbq)
 	for (i = 0; i < SBQ_WAIT_QUEUES; i++) {
 		struct sbq_wait_state *ws = &sbq->ws[wake_index];
 
-		if (waitqueue_active(&ws->wait)) {
+		if (waitqueue_active(&ws->wait) && atomic_read(&ws->wait_cnt)) {
 			if (wake_index != atomic_read(&sbq->wake_index))
 				atomic_set(&sbq->wake_index, wake_index);
 			return ws;
Signature · 1 line
2.35.1
author maintainer
#29 lore raw

PATCH 27/27 clk: bcm2835: Make peripheral PLLC critical

In reply to Sasha Levin

From: Maxime Ripard <[email protected]>

[ Upstream commit 6c5422851d8be8c7451e968fd2e6da41b6109e17 ]

When testing for a series affecting the VEC, it was discovered that turning off and on the VEC clock is crashing the system.

It turns out that, when disabling the VEC clock, it's the only child of the PLLC-per clock which will also get disabled. The source of the crash is PLLC-per being disabled.

It's likely that some other device might not take a clock reference that it actually needs, but it's unclear which at this point. Let's make PLLC-per critical so that we don't have that crash.

---
 drivers/clk/bcm/clk-bcm2835.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/clk/bcm/clk-bcm2835.c +1 -1
index e637bd6b295b..5a4d13365399 100644
@@ -1756,7 +1756,7 @@ static const struct bcm2835_clk_desc clk_desc_array[] = {
 		.load_mask = CM_PLLC_LOADPER,
 		.hold_mask = CM_PLLC_HOLDPER,
 		.fixed_divider = 1,
-		.flags = CLK_SET_RATE_PARENT),
+		.flags = CLK_IS_CRITICAL | CLK_SET_RATE_PARENT),
 
 	/*
 	 * PLLD is the display PLL, used to drive DSI display panels.
Signature · 1 line
2.35.1