- Linux-stable-mirror - lists.linaro.org

[PATCH net v4 1/3] mptcp: disallow MPTCP subflows from sockmap

by Jiayuan Chen

The sockmap feature allows bpf syscall from userspace using , or based on bpf sockops, replacing the sk_prot of sockets during protocol stack processing with sockmap's custom read/write interfaces. ''' tcp_rcv_state_process() subflow_syn_recv_sock() tcp_init_transfer(BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB) bpf_skops_established <== sockops bpf_sock_map_update(sk) <== call bpf helper tcp_bpf_update_proto() <== update sk_prot ''' Consider two scenarios: 1. When the server has MPTCP enabled and the client also requests MPTCP, the sk passed to the BPF program is a subflow sk. Since subflows only handle partial data, replacing their sk_prot is meaningless and will cause traffic disruption. 2. When the server has MPTCP enabled but the client sends a TCP SYN without MPTCP, subflow_syn_recv_sock() performs a fallback on the subflow, replacing the subflow sk's sk_prot with the native sk_prot. ''' subflow_ulp_fallback() subflow_drop_ctx() mptcp_subflow_ops_undo_override() ''' Subsequently, accept::mptcp_stream_accept::mptcp_fallback_tcp_ops() converts the subflow to plain TCP. For the first case, we should prevent it from being combined with sockmap by setting sk_prot->psock_update_sk_prot to NULL, which will be blocked by sockmap's own flow. For the second case, since subflow_syn_recv_sock() has already restored sk_prot to native tcp_prot/tcpv6_prot, no further action is needed. Fixes: 0b4f33def7bb ("mptcp: fix tcp fallback crash") Cc: <stable(a)vger.kernel.org> Signed-off-by: Jiayuan Chen <jiayuan.chen(a)linux.dev> --- net/mptcp/subflow.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c index 30961b3d1702..ddd0fc6fcf45 100644 --- a/net/mptcp/subflow.c +++ b/net/mptcp/subflow.c @@ -2144,6 +2144,10 @@ void __init mptcp_subflow_init(void) tcp_prot_override = tcp_prot; tcp_prot_override.release_cb = tcp_release_cb_override; tcp_prot_override.diag_destroy = tcp_abort_override; +#ifdef CONFIG_BPF_SYSCALL + /* Disable sockmap processing for subflows */ + tcp_prot_override.psock_update_sk_prot = NULL; +#endif #if IS_ENABLED(CONFIG_MPTCP_IPV6) /* In struct mptcp_subflow_request_sock, we assume the TCP request sock @@ -2180,6 +2184,10 @@ void __init mptcp_subflow_init(void) tcpv6_prot_override = tcpv6_prot; tcpv6_prot_override.release_cb = tcp_release_cb_override; tcpv6_prot_override.diag_destroy = tcp_abort_override; +#ifdef CONFIG_BPF_SYSCALL + /* Disable sockmap processing for subflows */ + tcpv6_prot_override.psock_update_sk_prot = NULL; +#endif #endif mptcp_diag_subflow_init(&subflow_ulp_ops); -- 2.43.0

6 days, 5 hours

2
1
0 0

[PATCH] drm/xe: allow request_irq on GSC interrupt

by Simon Richter

The default flags for freshly allocated interrupts are platform dependent, and apparently powerpc and arm set IRQ_NOREQUEST by default, and resets it once setup is complete. Most do this in the IRQ domain's "map" function. The xe driver does not define a domain, so clear the NOREQUEST and NOAUTOEN flags as part of the initialization. Also set NOPROBE -- it is doubtful this will ever be relevant, but it seems correct for what is effectively a softirq. Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/6468 Fixes: 87a4c85d3a3ed579c86fd2612715ccb94c4001ff Cc: <stable(a)vger.kernel.org> # v6.7+ Signed-off-by: Simon Richter <Simon.Richter(a)hogyros.de> --- drivers/gpu/drm/xe/xe_heci_gsc.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/xe/xe_heci_gsc.c b/drivers/gpu/drm/xe/xe_heci_gsc.c index 2b3d49dd394c..b0e38bd2e6f8 100644 --- a/drivers/gpu/drm/xe/xe_heci_gsc.c +++ b/drivers/gpu/drm/xe/xe_heci_gsc.c @@ -40,6 +40,7 @@ static int heci_gsc_irq_init(int irq) irq_set_chip_and_handler_name(irq, &heci_gsc_irq_chip, handle_simple_irq, "heci_gsc_irq_handler"); + irq_modify_status(irq, IRQ_NOREQUEST | IRQ_NOAUTOEN, IRQ_NOPROBE); return irq_set_chip_data(irq, NULL); } -- 2.47.3

6 days, 5 hours

1
0
0 0

[PATCH 5.10] net: dlink: handle copy_thresh allocation failure

by Andrey Troshin

From: Yeounsu Moon <yyyynoom(a)gmail.com> [ Upstream commit 8169a6011c5fecc6cb1c3654c541c567d3318de8 ] The driver did not handle failure of `netdev_alloc_skb_ip_align()`. If the allocation failed, dereferencing `skb->protocol` could lead to a NULL pointer dereference. This patch tries to allocate `skb`. If the allocation fails, it falls back to the normal path. Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Suggested-by: Jakub Kicinski <kuba(a)kernel.org> Tested-on: D-Link DGE-550T Rev-A3 Signed-off-by: Yeounsu Moon <yyyynoom(a)gmail.com> Reviewed-by: Andrew Lunn <andrew(a)lunn.ch> Link: https://patch.msgid.link/20250928190124.1156-1-yyyynoom@gmail.com Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> Signed-off-by: Sasha Levin <sashal(a)kernel.org> [Andrey Troshin: fixe merge conflicts] Signed-off-by: Andrey Troshin <drtrosh(a)yandex-team.ru> --- Backport fix for CVE-2025-40053 Link: https://nvd.nist.gov/vuln/detail/CVE-2025-40053 --- drivers/net/ethernet/dlink/dl2k.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/dlink/dl2k.c b/drivers/net/ethernet/dlink/dl2k.c index af1e96e0209f..0af58c4dcebc 100644 --- a/drivers/net/ethernet/dlink/dl2k.c +++ b/drivers/net/ethernet/dlink/dl2k.c @@ -957,15 +957,18 @@ receive_packet (struct net_device *dev) } else { struct sk_buff *skb; + skb = NULL; /* Small skbuffs for short packets */ - if (pkt_len > copy_thresh) { + if (pkt_len <= copy_thresh) + skb = netdev_alloc_skb_ip_align(dev, pkt_len); + if (!skb) { dma_unmap_single(&np->pdev->dev, desc_to_dma(desc), np->rx_buf_sz, DMA_FROM_DEVICE); skb_put (skb = np->rx_skbuff[entry], pkt_len); np->rx_skbuff[entry] = NULL; - } else if ((skb = netdev_alloc_skb_ip_align(dev, pkt_len))) { + } else { dma_sync_single_for_cpu(&np->pdev->dev, desc_to_dma(desc), np->rx_buf_sz, -- 2.34.1

6 days, 7 hours

1
0
0 0

[PATCH 0/5] platform/x86: alienware-wmi-wmax: Add AWCC support for most models

by Kurt Borja

Hi all, This patchset adds support for almost all models listed as supported by the AWCC windows tool. This is important because the "old" interface, which this driver defaults, is supported by very few and old models, while most Dell gaming laptops support the newer AWCC interface. Thanks! Signed-off-by: Kurt Borja <kuurtb(a)gmail.com> --- Kurt Borja (5): platform/x86: alienware-wmi-wmax: Fix "Alienware m16 R1 AMD" quirk order platform/x86: alienware-wmi-wmax: Drop redundant DMI entries platform/x86: alienware-wmi-wmax: Add support for the whole "M" family platform/x86: alienware-wmi-wmax: Add support for the whole "X" family platform/x86: alienware-wmi-wmax: Add support for the whole "G" family drivers/platform/x86/dell/alienware-wmi-wmax.c | 104 +++++-------------------- 1 file changed, 20 insertions(+), 84 deletions(-) --- base-commit: bd34bf518a5ffeb8eb7c8b9907ba97b606166f7b change-id: 20251013-family-supp-a3aa8d3bb27a -- ~ Kurt

6 days, 7 hours

2
5
0 0

[PATCH] nvmem: zynqmp_nvmem: fix DMA buffer size

by Ivan Vera

The efuse data buffer was allocated/freed with sizeof(bytes) instead of the requested length, resulting in an undersized DMA buffer and possible memory corruption. Allocate and free using the actual 'bytes' length. Fixes: 737c0c8d07b5 ("nvmem: zynqmp_nvmem: Add support to access efuse") Cc: stable(a)vger.kernel.org Signed-off-by: Ivan Vera <ivan.vera(a)enclustra.com> --- drivers/nvmem/zynqmp_nvmem.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/nvmem/zynqmp_nvmem.c b/drivers/nvmem/zynqmp_nvmem.c index 7da717d6c7fa..d909c8da747e 100644 --- a/drivers/nvmem/zynqmp_nvmem.c +++ b/drivers/nvmem/zynqmp_nvmem.c @@ -100,7 +100,7 @@ static int zynqmp_efuse_access(void *context, unsigned int offset, if (!efuse) return -ENOMEM; - data = dma_alloc_coherent(dev, sizeof(bytes), + data = dma_alloc_coherent(dev, bytes, &dma_buf, GFP_KERNEL); if (!data) { ret = -ENOMEM; @@ -134,7 +134,7 @@ static int zynqmp_efuse_access(void *context, unsigned int offset, if (flag == EFUSE_READ) memcpy(val, data, bytes); efuse_access_err: - dma_free_coherent(dev, sizeof(bytes), + dma_free_coherent(dev, bytes, data, dma_buf); efuse_data_fail: dma_free_coherent(dev, sizeof(struct xilinx_efuse), -- 2.25.1

6 days, 7 hours

1
0
0 0

[PATCH] firewire: nosy: break circular locking with misc_mtx in ->open()

by Guangshuo Li

Lockdep reports a circular dependency between card_mutex and misc_mtx: misc_open() holds misc_mtx and calls ->open(): misc_mtx -> nosy_open() add_card()/remove_card() hold card_mutex and (de)register the miscdev: card_mutex -> misc_register()/misc_deregister() -> misc_mtx nosy_open() only used card_mutex to map the minor to the pcilynx instance by scanning card_list. However, misc_open() already sets file->private_data to the struct miscdevice of the opened minor, so we can obtain the pcilynx pointer via container_of() and take a kref, without taking card_mutex in the open path. This removes the misc_mtx -> card_mutex edge and breaks the cycle. The crash info is below： ====================================================== WARNING: possible circular locking dependency detected 5.18.0-rc1 #1 Not tainted ------------------------------------------------------ syz-executor.1/1374 is trying to acquire lock: ffffffff8f8bb4e8 (card_mutex#2){+.+.}-{3:3}, at: nosy_open+0x55/0x480 but task is already holding lock: ffffffff8ef78a88 (misc_mtx){+.+.}-{3:3}, at: misc_open+0x5a/0x3f0 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #1 (misc_mtx){+.+.}-{3:3}: __mutex_lock_common+0x138/0x2450 mutex_lock_nested+0x17/0x20 misc_register+0x95/0x490 foo_misc_register+0x3a/0x50 add_card+0xd38/0x1250 local_pci_probe+0x140/0x200 pci_device_probe+0x345/0x770 really_probe+0x2d1/0x990 __driver_probe_device+0x1a7/0x270 driver_probe_device+0x54/0x370 __driver_attach+0x430/0x590 bus_for_each_dev+0x125/0x190 bus_add_driver+0x32c/0x530 driver_register+0x309/0x410 do_one_initcall+0x104/0x250 do_initcall_level+0x102/0x132 do_initcalls+0x46/0x74 kernel_init_freeable+0x28f/0x393 kernel_init+0x14/0x1a0 ret_from_fork+0x22/0x30 -> #0 (card_mutex#2){+.+.}-{3:3}: __lock_acquire+0x3607/0x7930 lock_acquire+0xff/0x2d0 __mutex_lock_common+0x138/0x2450 mutex_lock_nested+0x17/0x20 nosy_open+0x55/0x480 misc_open+0x363/0x3f0 chrdev_open+0x407/0x490 do_dentry_open+0x5b4/0xc20 path_openat+0x1d7a/0x2300 do_filp_open+0x1cb/0x3e0 do_sys_openat2+0x96/0x350 __x64_sys_openat+0x186/0x1b0 do_syscall_64+0x43/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae other info that might help us debug this: Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(misc_mtx); lock(card_mutex#2); lock(misc_mtx); lock(card_mutex#2); *** DEADLOCK *** 1 lock held by syz-executor.1/1374: #0: ffffffff8ef78a88 (misc_mtx){+.+.}-{3:3}, at: misc_open+0x5a/0x3f0 stack backtrace: CPU: 0 PID: 1374 Comm: syz-executor.1 Not tainted 5.18.0-rc1 #1 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.12.0-59-gc9ba5276e321-prebuilt.qemu.org 04/01/2014 Call Trace: <TASK> dump_stack_lvl+0x5a/0x74 check_noncircular+0x223/0x2d0 __lock_acquire+0x3607/0x7930 lock_acquire+0xff/0x2d0 __mutex_lock_common+0x138/0x2450 mutex_lock_nested+0x17/0x20 nosy_open+0x55/0x480 misc_open+0x363/0x3f0 chrdev_open+0x407/0x490 do_dentry_open+0x5b4/0xc20 path_openat+0x1d7a/0x2300 do_filp_open+0x1cb/0x3e0 do_sys_openat2+0x96/0x350 __x64_sys_openat+0x186/0x1b0 do_syscall_64+0x43/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7fbaab4abbcd Code: 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007fbaaac1cbe8 EFLAGS: 00000246 ORIG_RAX: 0000000000000101 RAX: ffffffffffffffda RBX: 00007fbaab5c9f80 RCX: 00007fbaab4abbcd RDX: 0000000000062803 RSI: 0000000020003080 RDI: ffffffffffffff9c RBP: 00007fbaab519edb R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000 R13: 00007ffe62540a6f R14: 00007ffe62540c10 R15: 00007fbaaac1cd80 </TASK> ====================================================== Fixes: 424d66cedae8 ("firewire: nosy: fix device shutdown with active client") Cc: stable(a)vger.kernel.org Signed-off-by: Guangshuo Li <lgs201920130244(a)gmail.com> --- drivers/firewire/nosy.c | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-) diff --git a/drivers/firewire/nosy.c b/drivers/firewire/nosy.c index b0d671db178a..726d5f15ff08 100644 --- a/drivers/firewire/nosy.c +++ b/drivers/firewire/nosy.c @@ -263,19 +263,13 @@ set_phy_reg(struct pcilynx *lynx, int addr, int val) static int nosy_open(struct inode *inode, struct file *file) { - int minor = iminor(inode); + struct miscdevice *misc = file->private_data; struct client *client; - struct pcilynx *tmp, *lynx = NULL; + struct pcilynx *lynx; - mutex_lock(&card_mutex); - list_for_each_entry(tmp, &card_list, link) - if (tmp->misc.minor == minor) { - lynx = lynx_get(tmp); - break; - } - mutex_unlock(&card_mutex); - if (lynx == NULL) - return -ENODEV; + /* misc_open() set file->private_data to the miscdevice already. */ + lynx = container_of(misc, struct pcilynx, misc); + lynx_get(lynx); client = kmalloc(sizeof *client, GFP_KERNEL); if (client == NULL) -- 2.43.0

6 days, 7 hours

1
0
0 0

[PATCH] mount: fix duplicate mounts using the new mount API

by GuangFei Luo

When using the new mount API, invoking the same mount command multiple times such as: mount /dev/mapper/mydevice /mnt/mydisk2 results in multiple identical mount entries being created: /dev/mapper/mydevice on /mnt/mydisk2 type ext4 (rw,relatime) /dev/mapper/mydevice on /mnt/mydisk2 type ext4 (rw,relatime) /dev/mapper/mydevice on /mnt/mydisk2 type ext4 (rw,relatime) ... Fixes: 2db154b3ea8e ("vfs: syscall: Add move_mount(2) to move mounts around") Signed-off-by: GuangFei Luo <luogf2025(a)163.com> Cc: <stable(a)vger.kernel.org> # 5.2+ --- fs/namespace.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/fs/namespace.c b/fs/namespace.c index d82910f33dc4..9436471955ce 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -4427,6 +4427,7 @@ SYSCALL_DEFINE5(move_mount, { struct path to_path __free(path_put) = {}; struct path from_path __free(path_put) = {}; + struct path path __free(path_put) = {}; struct filename *to_name __free(putname) = NULL; struct filename *from_name __free(putname) = NULL; unsigned int lflags, uflags; @@ -4472,6 +4473,14 @@ SYSCALL_DEFINE5(move_mount, return ret; } + ret = user_path_at(AT_FDCWD, to_pathname, LOOKUP_FOLLOW, &path); + if (ret) + return ret; + + /* Refuse the same filesystem on the same mount point */ + if (path.mnt->mnt_sb == to_path.mnt->mnt_sb && path_mounted(&path)) + return -EBUSY; + uflags = 0; if (flags & MOVE_MOUNT_F_EMPTY_PATH) uflags = AT_EMPTY_PATH; -- 2.43.0

6 days, 7 hours

4
6
0 0

[PATCH v2] ASoC: da7213: Use component driver suspend/resume

by Claudiu

From: Claudiu Beznea <claudiu.beznea.uj(a)bp.renesas.com> Since snd_soc_suspend() is invoked through snd_soc_pm_ops->suspend(), and snd_soc_pm_ops is associated with the soc_driver (defined in sound/soc/soc-core.c), and there is no parent-child relationship between the soc_driver and the DA7213 codec driver, the power management subsystem does not enforce a specific suspend/resume order between the DA7213 driver and the soc_driver. Because of this, the different codec component functionalities, called from snd_soc_resume() to reconfigure various functions, can race with the DA7213 struct dev_pm_ops::resume function, leading to misapplied configuration. This occasionally results in clipped sound. Fix this by dropping the struct dev_pm_ops::{suspend, resume} and use instead struct snd_soc_component_driver::{suspend, resume}. This ensures the proper configuration sequence is handled by the ASoC subsystem. Cc: stable(a)vger.kernel.org Fixes: 431e040065c8 ("ASoC: da7213: Add suspend to RAM support") Signed-off-by: Claudiu Beznea <claudiu.beznea.uj(a)bp.renesas.com> --- Changes in v2: - kept runtime PM ops unchanged and re-use them on struct snd_soc_component_driver::{suspend, resume} - updated the patch description to reflect the new approach - fixed the patch title - dropped patch 2/2 from v1 along with the cover letter sound/soc/codecs/da7213.c | 69 +++++++++++++++++++++++++-------------- sound/soc/codecs/da7213.h | 1 + 2 files changed, 45 insertions(+), 25 deletions(-) diff --git a/sound/soc/codecs/da7213.c b/sound/soc/codecs/da7213.c index c1657f348ad9..81bd5b03e2b6 100644 --- a/sound/soc/codecs/da7213.c +++ b/sound/soc/codecs/da7213.c @@ -2124,11 +2124,50 @@ static int da7213_probe(struct snd_soc_component *component) return 0; } +static int da7213_runtime_suspend(struct device *dev) +{ + struct da7213_priv *da7213 = dev_get_drvdata(dev); + + regcache_cache_only(da7213->regmap, true); + regcache_mark_dirty(da7213->regmap); + regulator_bulk_disable(DA7213_NUM_SUPPLIES, da7213->supplies); + + return 0; +} + +static int da7213_runtime_resume(struct device *dev) +{ + struct da7213_priv *da7213 = dev_get_drvdata(dev); + int ret; + + ret = regulator_bulk_enable(DA7213_NUM_SUPPLIES, da7213->supplies); + if (ret < 0) + return ret; + regcache_cache_only(da7213->regmap, false); + return regcache_sync(da7213->regmap); +} + +static int da7213_suspend(struct snd_soc_component *component) +{ + struct da7213_priv *da7213 = snd_soc_component_get_drvdata(component); + + return da7213_runtime_suspend(da7213->dev); +} + +static int da7213_resume(struct snd_soc_component *component) +{ + struct da7213_priv *da7213 = snd_soc_component_get_drvdata(component); + + return da7213_runtime_resume(da7213->dev); +} + static const struct snd_soc_component_driver soc_component_dev_da7213 = { .probe = da7213_probe, .set_bias_level = da7213_set_bias_level, .controls = da7213_snd_controls, .num_controls = ARRAY_SIZE(da7213_snd_controls), + .suspend = da7213_suspend, + .resume = da7213_resume, .dapm_widgets = da7213_dapm_widgets, .num_dapm_widgets = ARRAY_SIZE(da7213_dapm_widgets), .dapm_routes = da7213_audio_map, @@ -2175,6 +2214,8 @@ static int da7213_i2c_probe(struct i2c_client *i2c) if (!da7213->fin_min_rate) return -EINVAL; + da7213->dev = &i2c->dev; + i2c_set_clientdata(i2c, da7213); /* Get required supplies */ @@ -2224,31 +2265,9 @@ static void da7213_i2c_remove(struct i2c_client *i2c) pm_runtime_disable(&i2c->dev); } -static int da7213_runtime_suspend(struct device *dev) -{ - struct da7213_priv *da7213 = dev_get_drvdata(dev); - - regcache_cache_only(da7213->regmap, true); - regcache_mark_dirty(da7213->regmap); - regulator_bulk_disable(DA7213_NUM_SUPPLIES, da7213->supplies); - - return 0; -} - -static int da7213_runtime_resume(struct device *dev) -{ - struct da7213_priv *da7213 = dev_get_drvdata(dev); - int ret; - - ret = regulator_bulk_enable(DA7213_NUM_SUPPLIES, da7213->supplies); - if (ret < 0) - return ret; - regcache_cache_only(da7213->regmap, false); - return regcache_sync(da7213->regmap); -} - -static DEFINE_RUNTIME_DEV_PM_OPS(da7213_pm, da7213_runtime_suspend, - da7213_runtime_resume, NULL); +static const struct dev_pm_ops da7213_pm = { + RUNTIME_PM_OPS(da7213_runtime_suspend, da7213_runtime_resume, NULL) +}; static const struct i2c_device_id da7213_i2c_id[] = { { "da7213" }, diff --git a/sound/soc/codecs/da7213.h b/sound/soc/codecs/da7213.h index b9ab791d6b88..29cbf0eb6124 100644 --- a/sound/soc/codecs/da7213.h +++ b/sound/soc/codecs/da7213.h @@ -595,6 +595,7 @@ enum da7213_supplies { /* Codec private data */ struct da7213_priv { struct regmap *regmap; + struct device *dev; struct mutex ctrl_lock; struct regulator_bulk_data supplies[DA7213_NUM_SUPPLIES]; struct clk *mclk; -- 2.43.0

6 days, 7 hours

2
1
0 0

[PATCH v6 1/5] vfat: fix missing sb_min_blocksize() return value checks

by Yongpeng Yang

From: Yongpeng Yang <yangyongpeng(a)xiaomi.com> When emulating an nvme device on qemu with both logical_block_size and physical_block_size set to 8 KiB, but without format, a kernel panic was triggered during the early boot stage while attempting to mount a vfat filesystem. [95553.682035] EXT4-fs (nvme0n1): unable to set blocksize [95553.684326] EXT4-fs (nvme0n1): unable to set blocksize [95553.686501] EXT4-fs (nvme0n1): unable to set blocksize [95553.696448] ISOFS: unsupported/invalid hardware sector size 8192 [95553.697117] ------------[ cut here ]------------ [95553.697567] kernel BUG at fs/buffer.c:1582! [95553.697984] Oops: invalid opcode: 0000 [#1] SMP NOPTI [95553.698602] CPU: 0 UID: 0 PID: 7212 Comm: mount Kdump: loaded Not tainted 6.18.0-rc2+ #38 PREEMPT(voluntary) [95553.699511] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/2014 [95553.700534] RIP: 0010:folio_alloc_buffers+0x1bb/0x1c0 [95553.701018] Code: 48 8b 15 e8 93 18 02 65 48 89 35 e0 93 18 02 48 83 c4 10 5b 41 5c 41 5d 41 5e 41 5f 5d 31 d2 31 c9 31 f6 31 ff c3 cc cc cc cc <0f> 0b 90 66 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 0f [95553.702648] RSP: 0018:ffffd1b0c676f990 EFLAGS: 00010246 [95553.703132] RAX: ffff8cfc4176d820 RBX: 0000000000508c48 RCX: 0000000000000001 [95553.703805] RDX: 0000000000002000 RSI: 0000000000000000 RDI: 0000000000000000 [95553.704481] RBP: ffffd1b0c676f9c8 R08: 0000000000000000 R09: 0000000000000000 [95553.705148] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000001 [95553.705816] R13: 0000000000002000 R14: fffff8bc8257e800 R15: 0000000000000000 [95553.706483] FS: 000072ee77315840(0000) GS:ffff8cfdd2c8d000(0000) knlGS:0000000000000000 [95553.707248] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [95553.707782] CR2: 00007d8f2a9e5a20 CR3: 0000000039d0c006 CR4: 0000000000772ef0 [95553.708439] PKRU: 55555554 [95553.708734] Call Trace: [95553.709015] <TASK> [95553.709266] __getblk_slow+0xd2/0x230 [95553.709641] ? find_get_block_common+0x8b/0x530 [95553.710084] bdev_getblk+0x77/0xa0 [95553.710449] __bread_gfp+0x22/0x140 [95553.710810] fat_fill_super+0x23a/0xfc0 [95553.711216] ? __pfx_setup+0x10/0x10 [95553.711580] ? __pfx_vfat_fill_super+0x10/0x10 [95553.712014] vfat_fill_super+0x15/0x30 [95553.712401] get_tree_bdev_flags+0x141/0x1e0 [95553.712817] get_tree_bdev+0x10/0x20 [95553.713177] vfat_get_tree+0x15/0x20 [95553.713550] vfs_get_tree+0x2a/0x100 [95553.713910] vfs_cmd_create+0x62/0xf0 [95553.714273] __do_sys_fsconfig+0x4e7/0x660 [95553.714669] __x64_sys_fsconfig+0x20/0x40 [95553.715062] x64_sys_call+0x21ee/0x26a0 [95553.715453] do_syscall_64+0x80/0x670 [95553.715816] ? __fs_parse+0x65/0x1e0 [95553.716172] ? fat_parse_param+0x103/0x4b0 [95553.716587] ? vfs_parse_fs_param_source+0x21/0xa0 [95553.717034] ? __do_sys_fsconfig+0x3d9/0x660 [95553.717548] ? __x64_sys_fsconfig+0x20/0x40 [95553.717957] ? x64_sys_call+0x21ee/0x26a0 [95553.718360] ? do_syscall_64+0xb8/0x670 [95553.718734] ? __x64_sys_fsconfig+0x20/0x40 [95553.719141] ? x64_sys_call+0x21ee/0x26a0 [95553.719545] ? do_syscall_64+0xb8/0x670 [95553.719922] ? x64_sys_call+0x1405/0x26a0 [95553.720317] ? do_syscall_64+0xb8/0x670 [95553.720702] ? __x64_sys_close+0x3e/0x90 [95553.721080] ? x64_sys_call+0x1b5e/0x26a0 [95553.721478] ? do_syscall_64+0xb8/0x670 [95553.721841] ? irqentry_exit+0x43/0x50 [95553.722211] ? exc_page_fault+0x90/0x1b0 [95553.722681] entry_SYSCALL_64_after_hwframe+0x76/0x7e [95553.723166] RIP: 0033:0x72ee774f3afe [95553.723562] Code: 73 01 c3 48 8b 0d 0a 33 0f 00 f7 d8 64 89 01 48 83 c8 ff c3 0f 1f 84 00 00 00 00 00 f3 0f 1e fa 49 89 ca b8 af 01 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d da 32 0f 00 f7 d8 64 89 01 48 [95553.725188] RSP: 002b:00007ffe97148978 EFLAGS: 00000246 ORIG_RAX: 00000000000001af [95553.725892] RAX: ffffffffffffffda RBX: 00005dcfe53d0080 RCX: 000072ee774f3afe [95553.726526] RDX: 0000000000000000 RSI: 0000000000000006 RDI: 0000000000000003 [95553.727176] RBP: 00007ffe97148ac0 R08: 0000000000000000 R09: 000072ee775e7ac0 [95553.727818] R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000 [95553.728459] R13: 00005dcfe53d04b0 R14: 000072ee77670b00 R15: 00005dcfe53d1a28 [95553.729086] </TASK> The panic occurs as follows: 1. logical_block_size is 8KiB, causing {struct super_block *sb}->s_blocksize is initialized to 0. vfat_fill_super - fat_fill_super - sb_min_blocksize - sb_set_blocksize //return 0 when size is 8KiB. 2. __bread_gfp is called with size == 0, causing folio_alloc_buffers() to compute an offset equal to folio_size(folio), which triggers a BUG_ON. fat_fill_super - sb_bread - __bread_gfp // size == {struct super_block *sb}->s_blocksize == 0 - bdev_getblk - __getblk_slow - grow_buffers - grow_dev_folio - folio_alloc_buffers // size == 0 - folio_set_bh //offset == folio_size(folio) and panic To fix this issue, add proper return value checks for sb_min_blocksize(). Cc: <stable(a)vger.kernel.org> # v6.15 Fixes: a64e5a596067bd ("bdev: add back PAGE_SIZE block size validation for sb_set_blocksize()") Reviewed-by: Matthew Wilcox <willy(a)infradead.org> Reviewed-by: Darrick J. Wong <djwong(a)kernel.org> Reviewed-by: Jan Kara <jack(a)suse.cz> Reviewed-by: OGAWA Hirofumi <hirofumi(a)mail.parknet.co.jp> Reviewed-by: Christoph Hellwig <hch(a)lst.de> Signed-off-by: Yongpeng Yang <yangyongpeng(a)xiaomi.com> --- v6: - fix 'Fixes tag' format error - drop the pointless extern and spell out the parameter of sb_set_blocksize and sb_set_blocksize v5: - add cc tag for 5th patch v4: - split the changes into 5 patches v3: - remove the unnecessary blocksize variable definition v2: - add the __must_check mark to sb_min_blocksize() and include the Fixes tag --- fs/fat/inode.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/fs/fat/inode.c b/fs/fat/inode.c index 9648ed097816..9cfe20a3daaf 100644 --- a/fs/fat/inode.c +++ b/fs/fat/inode.c @@ -1595,8 +1595,12 @@ int fat_fill_super(struct super_block *sb, struct fs_context *fc, setup(sb); /* flavour-specific stuff that needs options */ + error = -EINVAL; + if (!sb_min_blocksize(sb, 512)) { + fat_msg(sb, KERN_ERR, "unable to set blocksize"); + goto out_fail; + } error = -EIO; - sb_min_blocksize(sb, 512); bh = sb_bread(sb, 0); if (bh == NULL) { fat_msg(sb, KERN_ERR, "unable to read boot sector"); -- 2.43.0

6 days, 8 hours

4
8
0 0

[PATCH v1 2/2] kasan: Unpoison vms[area] addresses with a common tag

by Maciej Wieczor-Retman

From: Maciej Wieczor-Retman <maciej.wieczor-retman(a)intel.com> A KASAN tag mismatch, possibly causing a kernel panic, can be observed on systems with a tag-based KASAN enabled and with multiple NUMA nodes. It was reported on arm64 and reproduced on x86. It can be explained in the following points: 1. There can be more than one virtual memory chunk. 2. Chunk's base address has a tag. 3. The base address points at the first chunk and thus inherits the tag of the first chunk. 4. The subsequent chunks will be accessed with the tag from the first chunk. 5. Thus, the subsequent chunks need to have their tag set to match that of the first chunk. Unpoison all vm_structs after allocating them for the percpu allocator. Use the same tag to resolve the pcpu chunk address mismatch. Fixes: 1d96320f8d53 ("kasan, vmalloc: add vmalloc tagging for SW_TAGS") Cc: <stable(a)vger.kernel.org> # 6.1+ Signed-off-by: Maciej Wieczor-Retman <maciej.wieczor-retman(a)intel.com> Tested-by: Baoquan He <bhe(a)redhat.com> --- Changelog v1 (after splitting of from the KASAN series): - Rewrite the patch message to point at the user impact of the issue. - Move helper to common.c so it can be compiled in all KASAN modes. mm/kasan/common.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/mm/kasan/common.c b/mm/kasan/common.c index c63544a98c24..a6bbc68984cd 100644 --- a/mm/kasan/common.c +++ b/mm/kasan/common.c @@ -584,12 +584,20 @@ bool __kasan_check_byte(const void *address, unsigned long ip) return true; } +/* + * A tag mismatch happens when calculating per-cpu chunk addresses, because + * they all inherit the tag from vms[0]->addr, even when nr_vms is bigger + * than 1. This is a problem because all the vms[]->addr come from separate + * allocations and have different tags so while the calculated address is + * correct the tag isn't. + */ void __kasan_unpoison_vmap_areas(struct vm_struct **vms, int nr_vms) { int area; for (area = 0 ; area < nr_vms ; area++) { kasan_poison(vms[area]->addr, vms[area]->size, - arch_kasan_get_tag(vms[area]->addr), false); + arch_kasan_get_tag(vms[0]->addr), false); + arch_kasan_set_tag(vms[area]->addr, arch_kasan_get_tag(vms[0]->addr)); } } -- 2.51.0

6 days, 8 hours

2
2
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror