Skip to contentSkip to conversation

fs: jfs: fix possible NULL pointer dereference in dbFree()

PATCH In review Sasha Levin 2 participants 13 messages Jun 1, 2022 – Jun 5, 2022

Subsystems (4)
lore
Share & tools
lore
author maintainer
#12 lore raw

PATCH 11/11 drm: fix EDID struct for old ARM OABI format

In reply to Sasha Levin

From: Linus Torvalds <[email protected]>

[ Upstream commit 47f15561b69e226bfc034e94ff6dbec51a4662af ]

When building the kernel for arm with the "-mabi=apcs-gnu" option, gcc will force alignment of all structures and unions to a word boundary (see also STRUCTURE_SIZE_BOUNDARY and the "-mstructure-size-boundary=XX" option if you're a gcc person), even when the members of said structures do not want or need said alignment.

This completely messes up the structure alignment of 'struct edid' on those targets, because even though all the embedded structures are marked with "__attribute__((packed))", the unions that contain them are not.

This was exposed by commit f1e4c916f97f ("drm/edid: add EDID block count and size helpers"), but the bug is pre-existing. That commit just made the structure layout problem cause a build failure due to the addition of the

        BUILD_BUG_ON(sizeof(*edid) != EDID_LENGTH);

sanity check in drivers/gpu/drm/drm_edid.c:edid_block_data().

This legacy union alignment should probably not be used in the first place, but we can fix the layout by adding the packed attribute to the union entries even when each member is already packed and it shouldn't matter in a sane build environment.

You can see this issue with a trivial test program:

  union {
	struct {
		char c[5];
	};
	struct {
		char d;
		unsigned e;
	} __attribute__((packed));

} a = { "1234" };

where building this with a normal "gcc -S" will result in the expected 5-byte size of said union:

	.type	a, @object
	.size	a, 5

but with an ARM compiler and the old ABI:

arm-linux-gnu-gcc -mabi=apcs-gnu -mfloat-abi=soft -S t.c

you get

	.type	a, %object
	.size	a, 8

instead, because even though each member of the union is packed, the union itself still gets aligned.

This was reported by Sudip for the spear3xx_defconfig target.

---
 include/drm/drm_edid.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
include/drm/drm_edid.h +3 -3
index c3a7d440bc11..514a02095983 100644
@@ -114,7 +114,7 @@ struct detailed_data_monitor_range {
 			u8 supported_scalings;
 			u8 preferred_refresh;
 		} __attribute__((packed)) cvt;
-	} formula;
+	} __attribute__((packed)) formula;
 } __attribute__((packed));
 
 struct detailed_data_wpindex {
@@ -147,7 +147,7 @@ struct detailed_non_pixel {
 		struct detailed_data_wpindex color;
 		struct std_timing timings[6];
 		struct cvt_timing cvt[4];
-	} data;
+	} __attribute__((packed)) data;
 } __attribute__((packed));
 
 #define EDID_DETAIL_EST_TIMINGS 0xf7
@@ -165,7 +165,7 @@ struct detailed_timing {
 	union {
 		struct detailed_pixel_timing pixel_data;
 		struct detailed_non_pixel other_data;
-	} data;
+	} __attribute__((packed)) data;
 } __attribute__((packed));
 
 #define DRM_EDID_INPUT_SERRATION_VSYNC (1 << 0)
Signature · 1 line
2.35.1
author maintainer
#1 lore raw

PATCH 2/11 ARM: OMAP1: clock: Fix UART rate reporting algorithm

In reply to Sasha Levin

From: Janusz Krzysztofik <[email protected]>

[ Upstream commit 338d5d476cde853dfd97378d20496baabc2ce3c0 ]

Since its introduction to the mainline kernel, omap1_uart_recalc() helper makes incorrect use of clk->enable_bit as a ready to use bitmap mask while it only provides the bit number. Fix it.

---
 arch/arm/mach-omap1/clock.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm/mach-omap1/clock.c +1 -1
index 034b89499bd7..a3599696e7cc 100644
@@ -44,7 +44,7 @@ static DEFINE_SPINLOCK(clockfw_lock);
 unsigned long omap1_uart_recalc(struct clk *clk)
 {
 	unsigned int val = __raw_readl(clk->enable_reg);
-	return val & clk->enable_bit ? 48000000 : 12000000;
+	return val & 1 << clk->enable_bit ? 48000000 : 12000000;
 }
 
 unsigned long omap1_sossi_recalc(struct clk *clk)
Signature · 1 line
2.35.1

Sasha Levin

author maintainer
#0 lore raw

PATCH 1/11 fs: jfs: fix possible NULL pointer dereference in dbFree()

From: Zixuan Fu <[email protected]>

[ Upstream commit 0d4837fdb796f99369cf7691d33de1b856bcaf1f ]

In our fault-injection testing, the variable "nblocks" in dbFree() can be zero when kmalloc_array() fails in dtSearch(). In this case, the variable
"mp" in dbFree() would be NULL and then it is dereferenced in
"write_metapage(mp)".

The failure log is listed as follows:

[   13.824137] BUG: kernel NULL pointer dereference, address: 0000000000000020
...
[   13.827416] RIP: 0010:dbFree+0x5f7/0x910 [jfs]
[   13.834341] Call Trace:
[   13.834540]  <TASK>
[   13.834713]  txFreeMap+0x7b4/0xb10 [jfs]
[   13.835038]  txUpdateMap+0x311/0x650 [jfs]
[   13.835375]  jfs_lazycommit+0x5f2/0xc70 [jfs]
[   13.835726]  ? sched_dynamic_update+0x1b0/0x1b0
[   13.836092]  kthread+0x3c2/0x4a0
[   13.836355]  ? txLockFree+0x160/0x160 [jfs]
[   13.836763]  ? kthread_unuse_mm+0x160/0x160
[   13.837106]  ret_from_fork+0x1f/0x30
[   13.837402]  </TASK>

...

This patch adds a NULL check of "mp" before "write_metapage(mp)" is called.

---
 fs/jfs/jfs_dmap.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
fs/jfs/jfs_dmap.c +2 -1
index 6dac48e29d28..a07fbb60ac3c 100644
@@ -398,7 +398,8 @@ int dbFree(struct inode *ip, s64 blkno, s64 nblocks)
 	}
 
 	/* write the last buffer. */
-	write_metapage(mp);
+	if (mp)
+		write_metapage(mp);
 
 	IREAD_UNLOCK(ipbmap);
 
Signature · 1 line
2.35.1
author maintainer
#2 lore raw

PATCH 3/11 selftests/resctrl: Change the default limited time to 120 seconds

From: Shaopeng Tan <[email protected]>

[ Upstream commit e2e3fb6ef0d6548defbe0be6e092397aaa92f3a1 ]

When testing on a Intel(R) Xeon(R) Gold 6254 CPU @ 3.10GHz the resctrl selftests fail due to timeout after exceeding the default time limit of 45 seconds. On this system the test takes about 68 seconds. Since the failing test by default accesses a fixed size of memory, the execution time should not vary significantly between different environment. A new default of 120 seconds should be sufficient yet easy to customize with the introduction of the "settings" file for reference.

---
 tools/testing/selftests/resctrl/settings | 3 +++
 1 file changed, 3 insertions(+)
 create mode 100644 tools/testing/selftests/resctrl/settings
tools/testing/selftests/resctrl/settings new file +3 -0
new file mode 100644
index 000000000000..a383f3d4565b
@@ -0,0 +1,3 @@
+# If running time is longer than 120 seconds when new tests are added in
+# the future, increase timeout here.
+timeout=120
Signature · 1 line
2.35.1
author maintainer
#7 lore raw

PATCH 6/11 ARM: dts: exynos: add atmel,24c128 fallback to Samsung EEPROM

In reply to Sasha Levin

From: Krzysztof Kozlowski <[email protected]>

[ Upstream commit f038e8186fbc5723d7d38c6fa1d342945107347e ]

The Samsung s524ad0xd1 EEPROM should use atmel,24c128 fallback, according to the AT24 EEPROM bindings.

---
 arch/arm/boot/dts/exynos5250-smdk5250.dts | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
arch/arm/boot/dts/exynos5250-smdk5250.dts +2 -2
index 54e79f6887ff..3dda0569f86a 100644
@@ -129,7 +129,7 @@ &i2c_0 {
 	samsung,i2c-max-bus-freq = <20000>;
 
 	eeprom@50 {
-		compatible = "samsung,s524ad0xd1";
+		compatible = "samsung,s524ad0xd1", "atmel,24c128";
 		reg = <0x50>;
 	};
 
@@ -288,7 +288,7 @@ &i2c_1 {
 	samsung,i2c-max-bus-freq = <20000>;
 
 	eeprom@51 {
-		compatible = "samsung,s524ad0xd1";
+		compatible = "samsung,s524ad0xd1", "atmel,24c128";
 		reg = <0x51>;
 	};
 
Signature · 1 line
2.35.1
author maintainer
#8 lore raw

PATCH 7/11 ARM: hisi: Add missing of_node_put after of_find_compatible_node

In reply to Sasha Levin

From: Peng Wu <[email protected]>

[ Upstream commit 9bc72e47d4630d58a840a66a869c56b29554cfe4 ]

of_find_compatible_node will increment the refcount of the returned device_node. Calling of_node_put() to avoid the refcount leak

---
 arch/arm/mach-hisi/platsmp.c | 4 ++++
 1 file changed, 4 insertions(+)
arch/arm/mach-hisi/platsmp.c +4 -0
index e1d67648d5d0..fccceab33325 100644
@@ -70,14 +70,17 @@ static void __init hi3xxx_smp_prepare_cpus(unsigned int max_cpus)
 		}
 		ctrl_base = of_iomap(np, 0);
 		if (!ctrl_base) {
+			of_node_put(np);
 			pr_err("failed to map address\n");
 			return;
 		}
 		if (of_property_read_u32(np, "smp-offset", &offset) < 0) {
+			of_node_put(np);
 			pr_err("failed to find smp-offset property\n");
 			return;
 		}
 		ctrl_base += offset;
+		of_node_put(np);
 	}
 }
 
@@ -163,6 +166,7 @@ static int hip01_boot_secondary(unsigned int cpu, struct task_struct *idle)
 	if (WARN_ON(!node))
 		return -1;
 	ctrl_base = of_iomap(node, 0);
+	of_node_put(node);
 
 	/* set the secondary core boot from DDR */
 	remap_reg_value = readl_relaxed(ctrl_base + REG_SC_CTRL);
Signature · 1 line
2.35.1
author maintainer
#6 lore raw

PATCH 5/11 ARM: versatile: Add missing of_node_put in dcscb_init

In reply to Sasha Levin

From: Peng Wu <[email protected]>

[ Upstream commit 23b44f9c649bbef10b45fa33080cd8b4166800ae ]

The device_node pointer is returned by of_find_compatible_node with refcount incremented. We should use of_node_put() to avoid the refcount leak.

---
 arch/arm/mach-vexpress/dcscb.c | 1 +
 1 file changed, 1 insertion(+)
arch/arm/mach-vexpress/dcscb.c +1 -0
index 5cedcf572104..3e86cff1d4d3 100644
@@ -146,6 +146,7 @@ static int __init dcscb_init(void)
 	if (!node)
 		return -ENODEV;
 	dcscb_base = of_iomap(node, 0);
+	of_node_put(node);
 	if (!dcscb_base)
 		return -EADDRNOTAVAIL;
 	cfg = readl_relaxed(dcscb_base + DCS_CFG_R);
Signature · 1 line
2.35.1
author maintainer
#5 lore raw

PATCH 4/11 fat: add ratelimit to fat*_ent_bread()

In reply to Sasha Levin

From: OGAWA Hirofumi <[email protected]>

[ Upstream commit 183c3237c928109d2008c0456dff508baf692b20 ]

fat*_ent_bread() can be the cause of too many report on I/O error path. So use fat_msg_ratelimit() instead.

---
 fs/fat/fatent.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
fs/fat/fatent.c +4 -3
index 0129d4d07a54..b0b1a71c07b7 100644
@@ -92,7 +92,8 @@ static int fat12_ent_bread(struct super_block *sb, struct fat_entry *fatent,
 err_brelse:
 	brelse(bhs[0]);
 err:
-	fat_msg(sb, KERN_ERR, "FAT read failed (blocknr %llu)", (llu)blocknr);
+	fat_msg_ratelimit(sb, KERN_ERR, "FAT read failed (blocknr %llu)",
+			  (llu)blocknr);
 	return -EIO;
 }
 
@@ -105,8 +106,8 @@ static int fat_ent_bread(struct super_block *sb, struct fat_entry *fatent,
 	fatent->fat_inode = MSDOS_SB(sb)->fat_inode;
 	fatent->bhs[0] = sb_bread(sb, blocknr);
 	if (!fatent->bhs[0]) {
-		fat_msg(sb, KERN_ERR, "FAT read failed (blocknr %llu)",
-		       (llu)blocknr);
+		fat_msg_ratelimit(sb, KERN_ERR, "FAT read failed (blocknr %llu)",
+				  (llu)blocknr);
 		return -EIO;
 	}
 	fatent->nr_bhs = 1;
Signature · 1 line
2.35.1
author maintainer
#9 lore raw

PATCH 8/11 PCI: Avoid pci_dev_lock() AB/BA deadlock with sriov_numvfs_store()

In reply to Sasha Levin

From: Yicong Yang <[email protected]>

[ Upstream commit a91ee0e9fca9d7501286cfbced9b30a33e52740a ]

The sysfs sriov_numvfs_store() path acquires the device lock before the config space access lock:

sriov_numvfs_store
device_lock # A (1) acquire device lock
sriov_configure
vfio_pci_sriov_configure # (for example)
vfio_pci_core_sriov_configure
pci_disable_sriov
sriov_disable
pci_cfg_access_lock
pci_wait_cfg # B (4) wait for dev->block_cfg_access == 0

Previously, pci_dev_lock() acquired the config space access lock before the device lock:

pci_dev_lock
pci_cfg_access_lock
dev->block_cfg_access = 1 # B (2) set dev->block_cfg_access = 1
device_lock # A (3) wait for device lock

Any path that uses pci_dev_lock(), e.g., pci_reset_function(), may deadlock with sriov_numvfs_store() if the operations occur in the sequence
(1) (2) (3) (4).

Avoid the deadlock by reversing the order in pci_dev_lock() so it acquires the device lock before the config space access lock, the same as the sriov_numvfs_store() path.

[bhelgaas: combined and adapted commit log from Jay Zhou's independent subsequent posting:
https://lore.kernel.org/r/[email protected]]

---
 drivers/pci/pci.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
drivers/pci/pci.c +5 -5
index 2cf13578fe75..e6e0012269cd 100644
@@ -4079,18 +4079,18 @@ static int __pci_dev_reset(struct pci_dev *dev, int probe)
 
 static void pci_dev_lock(struct pci_dev *dev)
 {
-	pci_cfg_access_lock(dev);
 	/* block PM suspend, driver probe, etc. */
 	device_lock(&dev->dev);
+	pci_cfg_access_lock(dev);
 }
 
 /* Return 1 on successful lock, 0 on contention */
 static int pci_dev_trylock(struct pci_dev *dev)
 {
-	if (pci_cfg_access_trylock(dev)) {
-		if (device_trylock(&dev->dev))
+	if (device_trylock(&dev->dev)) {
+		if (pci_cfg_access_trylock(dev))
 			return 1;
-		pci_cfg_access_unlock(dev);
+		device_unlock(&dev->dev);
 	}
 
 	return 0;
@@ -4098,8 +4098,8 @@ static int pci_dev_trylock(struct pci_dev *dev)
 
 static void pci_dev_unlock(struct pci_dev *dev)
 {
-	device_unlock(&dev->dev);
 	pci_cfg_access_unlock(dev);
+	device_unlock(&dev->dev);
 }
 
 /**
Signature · 1 line
2.35.1
author maintainer
#11 lore raw

PATCH 10/11 macintosh/via-pmu: Fix build failure when CONFIG_INPUT is disabled

In reply to Sasha Levin

From: Finn Thain <[email protected]>

[ Upstream commit 86ce436e30d86327c9f5260f718104ae7b21f506 ]

drivers/macintosh/via-pmu-event.o: In function `via_pmu_event':
via-pmu-event.c:(.text+0x44): undefined reference to input_event' via-pmu-event.c:(.text+0x68): undefined reference to input_event' via-pmu-event.c:(.text+0x94): undefined reference to input_event' via-pmu-event.c:(.text+0xb8): undefined reference to input_event' drivers/macintosh/via-pmu-event.o: In function `via_pmu_event_init':
via-pmu-event.c:(.init.text+0x20): undefined reference to input_allocate_device' via-pmu-event.c:(.init.text+0xc4): undefined reference to input_register_device' via-pmu-event.c:(.init.text+0xd4): undefined reference to `input_free_device' make[1]: *** [Makefile:1155: vmlinux] Error 1
make: *** [Makefile:350: __build_one_by_one] Error 2

Don't call into the input subsystem unless CONFIG_INPUT is built-in.

---
 drivers/macintosh/Kconfig   | 4 ++++
 drivers/macintosh/Makefile  | 3 ++-
 drivers/macintosh/via-pmu.c | 2 +-
 3 files changed, 7 insertions(+), 2 deletions(-)
drivers/macintosh/Kconfig +4 -0
index d28690f6e262..9e226e143473 100644
@@ -87,6 +87,10 @@ config ADB_PMU
 	  this device; you should do so if your machine is one of those
 	  mentioned above.
 
+config ADB_PMU_EVENT
+	def_bool y
+	depends on ADB_PMU && INPUT=y
+
 config ADB_PMU_LED
 	bool "Support for the Power/iBook front LED"
 	depends on ADB_PMU
drivers/macintosh/Makefile +2 -1
index 383ba920085b..8513c8aa2faf 100644
@@ -11,7 +11,8 @@ obj-$(CONFIG_MAC_EMUMOUSEBTN)	+= mac_hid.o
 obj-$(CONFIG_INPUT_ADBHID)	+= adbhid.o
 obj-$(CONFIG_ANSLCD)		+= ans-lcd.o
 
-obj-$(CONFIG_ADB_PMU)		+= via-pmu.o via-pmu-event.o
+obj-$(CONFIG_ADB_PMU)		+= via-pmu.o
+obj-$(CONFIG_ADB_PMU_EVENT)	+= via-pmu-event.o
 obj-$(CONFIG_ADB_PMU_LED)	+= via-pmu-led.o
 obj-$(CONFIG_PMAC_BACKLIGHT)	+= via-pmu-backlight.o
 obj-$(CONFIG_ADB_CUDA)		+= via-cuda.o
drivers/macintosh/via-pmu.c +1 -1
index 32c696799300..9bdb7d2055b1 100644
@@ -1439,7 +1439,7 @@ pmu_handle_data(unsigned char *data, int len)
 		pmu_pass_intr(data, len);
 		/* len == 6 is probably a bad check. But how do I
 		 * know what PMU versions send what events here? */
-		if (len == 6) {
+		if (IS_ENABLED(CONFIG_ADB_PMU_EVENT) && len == 6) {
 			via_pmu_event(PMU_EVT_POWER, !!(data[1]&8));
 			via_pmu_event(PMU_EVT_LID, data[1]&1);
 		}
Signature · 1 line
2.35.1
author maintainer
#10 lore raw

PATCH 9/11 powerpc/xics: fix refcount leak in icp_opal_init()

In reply to Sasha Levin

From: Lv Ruyi <[email protected]>

[ Upstream commit 5dd9e27ea4a39f7edd4bf81e9e70208e7ac0b7c9 ]

The of_find_compatible_node() function returns a node pointer with refcount incremented, use of_node_put() on it when done.

---
 arch/powerpc/sysdev/xics/icp-opal.c | 1 +
 1 file changed, 1 insertion(+)
arch/powerpc/sysdev/xics/icp-opal.c +1 -0
index b53f80f0b4d8..80a4fa6dcc55 100644
@@ -199,6 +199,7 @@ int icp_opal_init(void)
 
 	printk("XICS: Using OPAL ICP fallbacks\n");
 
+	of_node_put(np);
 	return 0;
 }
 
Signature · 1 line
2.35.1
#3 lore raw

Re: [PATCH AUTOSEL 4.9 03/11] selftests/resctrl: Change the default limited time to 120 seconds

In reply to Sasha Levin

Dear Stable Team,

I received the below notice about this commit being considered for 4.9 as well as other notices for it being considered for 4.14, 4.19, 5.4, 5.10, 5.15, 5.17, and 5.18.

I do not believe this commit is appropriate for stable because:
- It forms part of a series and without the other accompanying patches from that series it does not do anything. Series is at:
https://lore.kernel.org/lkml/[email protected]/
- The original series was not not submitted for inclusion to stable and none of its patches have a Fixes: tag.
- The series this patch forms part of aims to port resctrl_tests to the kselftest framework and I do not believe such a change matches stable criteria.

Reinette

Quoted 31 lines · Sasha Levin

From: Shaopeng Tan <[email protected]>

[ Upstream commit e2e3fb6ef0d6548defbe0be6e092397aaa92f3a1 ]

When testing on a Intel(R) Xeon(R) Gold 6254 CPU @ 3.10GHz the resctrl selftests fail due to timeout after exceeding the default time limit of 45 seconds. On this system the test takes about 68 seconds. Since the failing test by default accesses a fixed size of memory, the execution time should not vary significantly between different environment. A new default of 120 seconds should be sufficient yet easy to customize with the introduction of the "settings" file for reference.

---
 tools/testing/selftests/resctrl/settings | 3 +++
 1 file changed, 3 insertions(+)
 create mode 100644 tools/testing/selftests/resctrl/settings
tools/testing/selftests/resctrl/settings new file +3 -0
new file mode 100644
index 000000000000..a383f3d4565b
@@ -0,0 +1,3 @@
+# If running time is longer than 120 seconds when new tests are added in
+# the future, increase timeout here.
+timeout=120
author maintainer
#4 lore raw

Re: [PATCH AUTOSEL 4.9 03/11] selftests/resctrl: Change the default limited time to 120 seconds

Quoted 6 lines · Reinette Chatre

Dear Stable Team,

I received the below notice about this commit being considered for 4.9 as well as other notices for it being considered for 4.14, 4.19, 5.4, 5.10, 5.15, 5.17, and 5.18.

Quoted 9 lines

I do not believe this commit is appropriate for stable because:
- It forms part of a series and without the other accompanying patches from that series it does not do anything. Series is at:
https://lore.kernel.org/lkml/[email protected]/
- The original series was not not submitted for inclusion to stable and none of its patches have a Fixes: tag.
- The series this patch forms part of aims to port resctrl_tests to the kselftest framework and I do not believe such a change matches stable criteria.

Sure, I'll drop it, thanks!

Signature · 2 lines
Thanks,
Sasha