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

[REGRESSION FIX resend 1/1] gpiolib: acpi: Make set debounce errors non fatal

by Hans de Goede

Commit 16c07342b542 ("gpiolib: acpi: Program debounce when finding GPIO") adds a gpio_set_debounce_timeout() call to acpi_find_gpio() and makes acpi_find_gpio() fail if this fails. But gpio_set_debounce_timeout() failing is a somewhat normal occurrence, since not all debounce values are supported on all GPIO/pinctrl chips. Making this an error for example break getting the card-detect GPIO for the micro-sd slot found on many Bay Trail tablets, breaking support for the micro-sd slot on these tablets. acpi_request_own_gpiod() already treats gpio_set_debounce_timeout() failures as non-fatal, just warning about them. Add a acpi_gpio_set_debounce_timeout() helper which wraps gpio_set_debounce_timeout() and warns on failures and replace both existing gpio_set_debounce_timeout() calls with the helper. Since the helper only warns on failures this fixes the card-detect issue. Fixes: 16c07342b542 ("gpiolib: acpi: Program debounce when finding GPIO") Cc: stable(a)vger.kernel.org Cc: Mario Limonciello <superm1(a)kernel.org> Signed-off-by: Hans de Goede <hansg(a)kernel.org> --- drivers/gpio/gpiolib-acpi-core.c | 27 +++++++++++++++------------ 1 file changed, 15 insertions(+), 12 deletions(-) diff --git a/drivers/gpio/gpiolib-acpi-core.c b/drivers/gpio/gpiolib-acpi-core.c index 284e762d92c4..67c4c38afb86 100644 --- a/drivers/gpio/gpiolib-acpi-core.c +++ b/drivers/gpio/gpiolib-acpi-core.c @@ -291,6 +291,19 @@ acpi_gpio_to_gpiod_flags(const struct acpi_resource_gpio *agpio, int polarity) return GPIOD_ASIS; } +static void acpi_gpio_set_debounce_timeout(struct gpio_desc *desc, + unsigned int acpi_debounce) +{ + int ret; + + /* ACPI uses hundredths of milliseconds units */ + acpi_debounce *= 10; + ret = gpio_set_debounce_timeout(desc, acpi_debounce); + if (ret) + gpiod_warn(desc, "Failed to set debounce-timeout %u: %d\n", + acpi_debounce, ret); +} + static struct gpio_desc *acpi_request_own_gpiod(struct gpio_chip *chip, struct acpi_resource_gpio *agpio, unsigned int index, @@ -300,18 +313,12 @@ static struct gpio_desc *acpi_request_own_gpiod(struct gpio_chip *chip, enum gpiod_flags flags = acpi_gpio_to_gpiod_flags(agpio, polarity); unsigned int pin = agpio->pin_table[index]; struct gpio_desc *desc; - int ret; desc = gpiochip_request_own_desc(chip, pin, label, polarity, flags); if (IS_ERR(desc)) return desc; - /* ACPI uses hundredths of milliseconds units */ - ret = gpio_set_debounce_timeout(desc, agpio->debounce_timeout * 10); - if (ret) - dev_warn(chip->parent, - "Failed to set debounce-timeout for pin 0x%04X, err %d\n", - pin, ret); + acpi_gpio_set_debounce_timeout(desc, agpio->debounce_timeout); return desc; } @@ -944,7 +951,6 @@ struct gpio_desc *acpi_find_gpio(struct fwnode_handle *fwnode, bool can_fallback = acpi_can_fallback_to_crs(adev, con_id); struct acpi_gpio_info info = {}; struct gpio_desc *desc; - int ret; desc = __acpi_find_gpio(fwnode, con_id, idx, can_fallback, &info); if (IS_ERR(desc)) @@ -959,10 +965,7 @@ struct gpio_desc *acpi_find_gpio(struct fwnode_handle *fwnode, acpi_gpio_update_gpiod_flags(dflags, &info); acpi_gpio_update_gpiod_lookup_flags(lookupflags, &info); - /* ACPI uses hundredths of milliseconds units */ - ret = gpio_set_debounce_timeout(desc, info.debounce * 10); - if (ret) - return ERR_PTR(ret); + acpi_gpio_set_debounce_timeout(desc, info.debounce); return desc; } -- 2.51.0

6 days, 8 hours

3
2
0 0

[PATCH v2] mm/shmem: fix THP allocation and fallback loop

by Kairui Song

From: Kairui Song <kasong(a)tencent.com> The order check and fallback loop is updating the index value on every loop, this will cause the index to be aligned by a larger value while the loop shrinks the order. This may result in inserting and returning a folio of the wrong index and cause data corruption with some userspace workloads [1]. Cc: stable(a)vger.kernel.org Link: https://lore.kernel.org/linux-mm/CAMgjq7DqgAmj25nDUwwu1U2cSGSn8n4-Hqpgotted… [1] Fixes: e7a2ab7b3bb5d ("mm: shmem: add mTHP support for anonymous shmem") Signed-off-by: Kairui Song <kasong(a)tencent.com> --- Changes from V1: - Link to V1: https://lore.kernel.org/linux-mm/20251021190436.81682-1-ryncsn@gmail.com/ - Remove unnecessary cleanup and simplify the commit message. mm/shmem.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/mm/shmem.c b/mm/shmem.c index b50ce7dbc84a..7559773ebb30 100644 --- a/mm/shmem.c +++ b/mm/shmem.c @@ -1895,10 +1895,11 @@ static struct folio *shmem_alloc_and_add_folio(struct vm_fault *vmf, order = highest_order(suitable_orders); while (suitable_orders) { pages = 1UL << order; - index = round_down(index, pages); - folio = shmem_alloc_folio(gfp, order, info, index); - if (folio) + folio = shmem_alloc_folio(gfp, order, info, round_down(index, pages)); + if (folio) { + index = round_down(index, pages); goto allocated; + } if (pages == HPAGE_PMD_NR) count_vm_event(THP_FILE_FALLBACK); -- 2.51.0

6 days, 9 hours

2
3
0 0

[PATCH v3] usb: gadget: udc: fix use-after-free in usb_gadget_state_work

by Jimmy Hu

A race condition during gadget teardown can lead to a use-after-free in usb_gadget_state_work(), as reported by KASAN: BUG: KASAN: invalid-access in sysfs_notify+0x2c/0xd0 Workqueue: events usb_gadget_state_work The fundamental race occurs because a concurrent event (e.g., an interrupt) can call usb_gadget_set_state() and schedule gadget->work at any time during the cleanup process in usb_del_gadget(). Commit 399a45e5237c ("usb: gadget: core: flush gadget workqueue after device removal") attempted to fix this by moving flush_work() to after device_del(). However, this does not fully solve the race, as a new work item can still be scheduled *after* flush_work() completes but before the gadget's memory is freed, leading to the same use-after-free. This patch fixes the race condition robustly by introducing a 'teardown' flag and a 'state_lock' spinlock to the usb_gadget struct. The flag is set during cleanup in usb_del_gadget() *before* calling flush_work() to prevent any new work from being scheduled once cleanup has commenced. The scheduling site, usb_gadget_set_state(), now checks this flag under the lock before queueing the work, thus safely closing the race window. Fixes: 5702f75375aa9 ("usb: gadget: udc-core: move sysfs_notify() to a workqueue") Cc: stable(a)vger.kernel.org Signed-off-by: Jimmy Hu <hhhuuu(a)google.com> --- Changes in v3: - Updated patch title to more accurately describe the bug. - Moved changelog below the '---' line as requested by Greg KH. - Rebased on usb-linus branch as requested by Greg KH. Changes in v2: - Removed redundant inline comments as suggested by Alan Stern. drivers/usb/gadget/udc/core.c | 17 ++++++++++++++++- include/linux/usb/gadget.h | 5 +++++ 2 files changed, 21 insertions(+), 1 deletion(-) diff --git a/drivers/usb/gadget/udc/core.c b/drivers/usb/gadget/udc/core.c index 694653761c44..8dbe79bdc0f9 100644 --- a/drivers/usb/gadget/udc/core.c +++ b/drivers/usb/gadget/udc/core.c @@ -1126,8 +1126,13 @@ static void usb_gadget_state_work(struct work_struct *work) void usb_gadget_set_state(struct usb_gadget *gadget, enum usb_device_state state) { + unsigned long flags; + + spin_lock_irqsave(&gadget->state_lock, flags); gadget->state = state; - schedule_work(&gadget->work); + if (!gadget->teardown) + schedule_work(&gadget->work); + spin_unlock_irqrestore(&gadget->state_lock, flags); trace_usb_gadget_set_state(gadget, 0); } EXPORT_SYMBOL_GPL(usb_gadget_set_state); @@ -1361,6 +1366,8 @@ static void usb_udc_nop_release(struct device *dev) void usb_initialize_gadget(struct device *parent, struct usb_gadget *gadget, void (*release)(struct device *dev)) { + spin_lock_init(&gadget->state_lock); + gadget->teardown = false; INIT_WORK(&gadget->work, usb_gadget_state_work); gadget->dev.parent = parent; @@ -1535,6 +1542,7 @@ EXPORT_SYMBOL_GPL(usb_add_gadget_udc); void usb_del_gadget(struct usb_gadget *gadget) { struct usb_udc *udc = gadget->udc; + unsigned long flags; if (!udc) return; @@ -1548,6 +1556,13 @@ void usb_del_gadget(struct usb_gadget *gadget) kobject_uevent(&udc->dev.kobj, KOBJ_REMOVE); sysfs_remove_link(&udc->dev.kobj, "gadget"); device_del(&gadget->dev); + /* + * Set the teardown flag before flushing the work to prevent new work + * from being scheduled while we are cleaning up. + */ + spin_lock_irqsave(&gadget->state_lock, flags); + gadget->teardown = true; + spin_unlock_irqrestore(&gadget->state_lock, flags); flush_work(&gadget->work); ida_free(&gadget_id_numbers, gadget->id_number); cancel_work_sync(&udc->vbus_work); diff --git a/include/linux/usb/gadget.h b/include/linux/usb/gadget.h index 3aaf19e77558..8285b19a25e0 100644 --- a/include/linux/usb/gadget.h +++ b/include/linux/usb/gadget.h @@ -376,6 +376,9 @@ struct usb_gadget_ops { * can handle. The UDC must support this and all slower speeds and lower * number of lanes. * @state: the state we are now (attached, suspended, configured, etc) + * @state_lock: Spinlock protecting the `state` and `teardown` members. + * @teardown: True if the device is undergoing teardown, used to prevent + * new work from being scheduled during cleanup. * @name: Identifies the controller hardware type. Used in diagnostics * and sometimes configuration. * @dev: Driver model state for this abstract device. @@ -451,6 +454,8 @@ struct usb_gadget { enum usb_ssp_rate max_ssp_rate; enum usb_device_state state; + spinlock_t state_lock; + bool teardown; const char *name; struct device dev; unsigned isoch_delay; -- 2.51.1.814.gb8fa24458f-goog

6 days, 9 hours

1
0
0 0

[PATCH RESEND] drm/tegra: dc: fix reference leak in tegra_dc_couple()

by Ma Ke

driver_find_device() calls get_device() to increment the reference count once a matching device is found, but there is no put_device() to balance the reference count. To avoid reference count leakage, add put_device() to decrease the reference count. Found by code review. Cc: stable(a)vger.kernel.org Fixes: a31500fe7055 ("drm/tegra: dc: Restore coupling of display controllers") Signed-off-by: Ma Ke <make24(a)iscas.ac.cn> --- drivers/gpu/drm/tegra/dc.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/tegra/dc.c b/drivers/gpu/drm/tegra/dc.c index 59d5c1ba145a..6c84bd69b11f 100644 --- a/drivers/gpu/drm/tegra/dc.c +++ b/drivers/gpu/drm/tegra/dc.c @@ -3148,6 +3148,7 @@ static int tegra_dc_couple(struct tegra_dc *dc) dc->client.parent = &parent->client; dev_dbg(dc->dev, "coupled to %s\n", dev_name(companion)); + put_device(companion); } return 0; -- 2.17.1

6 days, 11 hours

2
1
0 0

[PATCH] Revert "sched/core: Tweak wait_task_inactive() to force dequeue sched_delayed tasks"

by Matt Fleming

From: Matt Fleming <mfleming(a)cloudflare.com> This reverts commit b7ca5743a2604156d6083b88cefacef983f3a3a6. If we dequeue a task (task B) that was sched delayed then that task is definitely no longer on the rq and not tracked in the rbtree. Unfortunately, task_on_rq_queued(B) will still return true because dequeue_task() doesn't update p->on_rq. This inconsistency can lead to tasks (task A) spinning indefinitely in wait_task_inactive(), e.g. when delivering a fatal signal to a thread group, because it thinks the task B is still queued (it's not) and waits forever for it to unschedule. Task A Task B arch_do_signal_or_restart() get_signal() do_coredump() coredump_wait() zap_threads() arch_do_signal_or_restart() wait_task_inactive() <-- SPIN get_signal() do_group_exit() do_exit() coredump_task_exit() schedule() <--- never comes back Not only will task A spin forever in wait_task_inactive(), but task B will also trigger RCU stalls: INFO: rcu_tasks detected stalls on tasks: 00000000a973a4d8: .. nvcsw: 2/2 holdout: 1 idle_cpu: -1/79 task:ffmpeg state:I stack:0 pid:665601 tgid:665155 ppid:668691 task_flags:0x400448 flags:0x00004006 Call Trace: <TASK> __schedule+0x4fb/0xbf0 ? srso_return_thunk+0x5/0x5f schedule+0x27/0xf0 do_exit+0xdd/0xaa0 ? __pfx_futex_wake_mark+0x10/0x10 do_group_exit+0x30/0x80 get_signal+0x81e/0x860 ? srso_return_thunk+0x5/0x5f ? futex_wake+0x177/0x1a0 arch_do_signal_or_restart+0x2e/0x1f0 ? srso_return_thunk+0x5/0x5f ? srso_return_thunk+0x5/0x5f ? __x64_sys_futex+0x10c/0x1d0 syscall_exit_to_user_mode+0xa5/0x130 do_syscall_64+0x57/0x110 entry_SYSCALL_64_after_hwframe+0x76/0x7e RIP: 0033:0x7f22d05b0f16 RSP: 002b:00007f2265761cf0 EFLAGS: 00000246 ORIG_RAX: 00000000000000ca RAX: fffffffffffffe00 RBX: 0000000000000000 RCX: 00007f22d05b0f16 RDX: 0000000000000000 RSI: 0000000000000189 RDI: 00005629e320d97c RBP: 0000000000000000 R08: 0000000000000000 R09: 00000000ffffffff R10: 0000000000000000 R11: 0000000000000246 R12: 00005629e320d928 R13: 0000000000000000 R14: 0000000000000001 R15: 00005629e320d97c </TASK> Fixes: b7ca5743a260 ("sched/core: Tweak wait_task_inactive() to force dequeue sched_delayed tasks") Cc: Peter Zijlstra <peterz(a)infradead.org> Cc: Oleg Nesterov <oleg(a)redhat.com> Cc: John Stultz <jstultz(a)google.com> Cc: Chris Arges <carges(a)cloudflare.com> Cc: stable(a)vger.kernel.org # v6.12 Signed-off-by: Matt Fleming <mfleming(a)cloudflare.com> --- kernel/sched/core.c | 6 ------ 1 file changed, 6 deletions(-) diff --git a/kernel/sched/core.c b/kernel/sched/core.c index ccba6fc3c3fe..2dfc3977920d 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -2293,12 +2293,6 @@ unsigned long wait_task_inactive(struct task_struct *p, unsigned int match_state * just go back and repeat. */ rq = task_rq_lock(p, &rf); - /* - * If task is sched_delayed, force dequeue it, to avoid always - * hitting the tick timeout in the queued case - */ - if (p->se.sched_delayed) - dequeue_task(rq, p, DEQUEUE_SLEEP | DEQUEUE_DELAYED); trace_sched_wait_task(p); running = task_on_cpu(rq, p); queued = task_on_rq_queued(p); -- 2.34.1

6 days, 11 hours

5
13
0 0

[PATCH 6.1.y v2] Revert "selftests: mm: fix map_hugetlb failure on 64K page size systems"

by Leon Hwang

This reverts commit a584c7734a4dd050451fcdd65c66317e15660e81 to avoid the build error: map_hugetlb.c: In function 'main': map_hugetlb.c:79:25: warning: implicit declaration of function 'default_huge_page_size' [-Wimplicit-function-declaration] 79 | hugepage_size = default_huge_page_size(); Signed-off-by: Leon Hwang <leon.hwang(a)linux.dev> --- v1 -> v2: - Revert the commit instead of fixing the build error. - https://lore.kernel.org/linux-mm/20251022055138.375042-1-leon.hwang@linux.d… tools/testing/selftests/vm/map_hugetlb.c | 7 ------- 1 file changed, 7 deletions(-) diff --git a/tools/testing/selftests/vm/map_hugetlb.c b/tools/testing/selftests/vm/map_hugetlb.c index c65c55b7a789..312889edb84a 100644 --- a/tools/testing/selftests/vm/map_hugetlb.c +++ b/tools/testing/selftests/vm/map_hugetlb.c @@ -15,7 +15,6 @@ #include <unistd.h> #include <sys/mman.h> #include <fcntl.h> -#include "vm_util.h" #define LENGTH (256UL*1024*1024) #define PROTECTION (PROT_READ | PROT_WRITE) @@ -71,16 +70,10 @@ int main(int argc, char **argv) { void *addr; int ret; - size_t hugepage_size; size_t length = LENGTH; int flags = FLAGS; int shift = 0; - hugepage_size = default_huge_page_size(); - /* munmap with fail if the length is not page aligned */ - if (hugepage_size > length) - length = hugepage_size; - if (argc > 1) length = atol(argv[1]) << 20; if (argc > 2) { -- 2.51.0

6 days, 13 hours

1
0
0 0

[PATCH net 1/1] octeontx2-af: CGX: fix bitmap leaks

by Bo Sun

The RX/TX flow-control bitmaps (rx_fc_pfvf_bmap and tx_fc_pfvf_bmap) are allocated by cgx_lmac_init() but never freed in cgx_lmac_exit(). Unbinding and rebinding the driver therefore triggers kmemleak: unreferenced object (size 16): backtrace: rvu_alloc_bitmap cgx_probe Free both bitmaps during teardown. Fixes: e740003874ed ("octeontx2-af: Flow control resource management") Cc: stable(a)vger.kernel.org Signed-off-by: Bo Sun <bo(a)mboxify.com> --- drivers/net/ethernet/marvell/octeontx2/af/cgx.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/net/ethernet/marvell/octeontx2/af/cgx.c b/drivers/net/ethernet/marvell/octeontx2/af/cgx.c index ec0e11c77cbf..f56e6782c4de 100644 --- a/drivers/net/ethernet/marvell/octeontx2/af/cgx.c +++ b/drivers/net/ethernet/marvell/octeontx2/af/cgx.c @@ -1823,6 +1823,8 @@ static int cgx_lmac_exit(struct cgx *cgx) cgx->mac_ops->mac_pause_frm_config(cgx, lmac->lmac_id, false); cgx_configure_interrupt(cgx, lmac, lmac->lmac_id, true); kfree(lmac->mac_to_index_bmap.bmap); + kfree(lmac->rx_fc_pfvf_bmap.bmap); + kfree(lmac->tx_fc_pfvf_bmap.bmap); kfree(lmac->name); kfree(lmac); }

6 days, 14 hours

3
2
0 0

Re: [PATCH] ksmbd: transport_ipc: validate payload size before reading handle

by Namjae Jeon

On Wed, Oct 22, 2025 at 7:45 PM くさあさ <pioooooooooip(a)gmail.com> wrote: > > Hi Namjae, Steve, Hi, > > Thanks for updating the patch. I’ve reviewed the changes and they look good to me. Okay. > > Minor impact note: this patch prevents a 4-byte out-of-bounds read in ksmbd’s handle_response() when the declared Generic Netlink payload size is < 4. > If a remote client can influence ksmbd.mountd to emit a truncated payload, this could be remotely triggerable (info-leak/DoS potential). I don't understand how this is possible. Could you please explain it to me via private email? > If you consider this security-impacting, I’m happy to request a CVE via the kernel.org CNA. > > Thanks!! > Qianchang Zhao

6 days, 16 hours

1
0
0 0

[PATCH 6.1.y] net: mctp: Don't access ifa_index when missing

by Vasiliy Kovalev

From: Matt Johnston <matt(a)codeconstruct.com.au> commit f11cf946c0a92c560a890d68e4775723353599e1 upstream. In mctp_dump_addrinfo, ifa_index can be used to filter interfaces, but only when the struct ifaddrmsg is provided. Otherwise it will be comparing to uninitialised memory - reproducible in the syzkaller case from dhcpd, or busybox "ip addr show". The kernel MCTP implementation has always filtered by ifa_index, so existing userspace programs expecting to dump MCTP addresses must already be passing a valid ifa_index value (either 0 or a real index). BUG: KMSAN: uninit-value in mctp_dump_addrinfo+0x208/0xac0 net/mctp/device.c:128 mctp_dump_addrinfo+0x208/0xac0 net/mctp/device.c:128 rtnl_dump_all+0x3ec/0x5b0 net/core/rtnetlink.c:4380 rtnl_dumpit+0xd5/0x2f0 net/core/rtnetlink.c:6824 netlink_dump+0x97b/0x1690 net/netlink/af_netlink.c:2309 Fixes: 583be982d934 ("mctp: Add device handling and netlink interface") Reported-by: syzbot+e76d52dadc089b9d197f(a)syzkaller.appspotmail.com Closes: https://lore.kernel.org/all/68135815.050a0220.3a872c.000e.GAE@google.com/ Reported-by: syzbot+1065a199625a388fce60(a)syzkaller.appspotmail.com Closes: https://lore.kernel.org/all/681357d6.050a0220.14dd7d.000d.GAE@google.com/ Signed-off-by: Matt Johnston <matt(a)codeconstruct.com.au> Link: https://patch.msgid.link/20250508-mctp-addr-dump-v2-1-c8a53fd2dd66@codecons… Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> [ kovalev: bp to fix CVE-2025-38006 ] Signed-off-by: Vasiliy Kovalev <kovalev(a)altlinux.org> --- net/mctp/device.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) diff --git a/net/mctp/device.c b/net/mctp/device.c index 85cc5f31f1e7..27aee8b04055 100644 --- a/net/mctp/device.c +++ b/net/mctp/device.c @@ -119,12 +119,19 @@ static int mctp_dump_addrinfo(struct sk_buff *skb, struct netlink_callback *cb) struct net_device *dev; struct ifaddrmsg *hdr; struct mctp_dev *mdev; - int ifindex; + int ifindex = 0; int idx = 0, rc; - hdr = nlmsg_data(cb->nlh); - // filter by ifindex if requested - ifindex = hdr->ifa_index; + /* Filter by ifindex if a header is provided */ + if (cb->nlh->nlmsg_len >= nlmsg_msg_size(sizeof(*hdr))) { + hdr = nlmsg_data(cb->nlh); + ifindex = hdr->ifa_index; + } else { + if (cb->strict_check) { + NL_SET_ERR_MSG(cb->extack, "mctp: Invalid header for addr dump request"); + return -EINVAL; + } + } rcu_read_lock(); for (; mcb->h < NETDEV_HASHENTRIES; mcb->h++, mcb->idx = 0) { -- 2.50.1

6 days, 17 hours

1
0
0 0

+ mm-shmem-fix-thp-allocation-and-fallback-loop.patch added to mm-hotfixes-unstable branch

by Andrew Morton

The patch titled Subject: mm/shmem: fix THP allocation and fallback loop has been added to the -mm mm-hotfixes-unstable branch. Its filename is mm-shmem-fix-thp-allocation-and-fallback-loop.patch This patch will shortly appear at https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patche… This patch will later appear in the mm-hotfixes-unstable branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/process/submit-checklist.rst when testing your code *** The -mm tree is included into linux-next via the mm-everything branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm and is updated there every 2-3 working days ------------------------------------------------------ From: Kairui Song <kasong(a)tencent.com> Subject: mm/shmem: fix THP allocation and fallback loop Date: Wed, 22 Oct 2025 18:57:19 +0800 The order check and fallback loop is updating the index value on every loop, this will cause the index to be aligned by a larger value while the loop shrinks the order. This may result in inserting and returning a folio of the wrong index and cause data corruption with some userspace workloads [1]. Link: https://lkml.kernel.org/r/20251022105719.18321-1-ryncsn@gmail.com Link: https://lore.kernel.org/linux-mm/CAMgjq7DqgAmj25nDUwwu1U2cSGSn8n4-Hqpgotted… [1] Fixes: e7a2ab7b3bb5d ("mm: shmem: add mTHP support for anonymous shmem") Signed-off-by: Kairui Song <kasong(a)tencent.com> Cc: Baolin Wang <baolin.wang(a)linux.alibaba.com> Cc: Barry Song <baohua(a)kernel.org> Cc: David Hildenbrand <david(a)redhat.com> Cc: Dev Jain <dev.jain(a)arm.com> Cc: Hugh Dickins <hughd(a)google.com> Cc: Liam Howlett <liam.howlett(a)oracle.com> Cc: Lorenzo Stoakes <lorenzo.stoakes(a)oracle.com> Cc: Matthew Wilcox (Oracle) <willy(a)infradead.org> Cc: Nico Pache <npache(a)redhat.com> Cc: Ryan Roberts <ryan.roberts(a)arm.com> Cc: Zi Yan <ziy(a)nvidia.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- mm/shmem.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) --- a/mm/shmem.c~mm-shmem-fix-thp-allocation-and-fallback-loop +++ a/mm/shmem.c @@ -1895,10 +1895,11 @@ static struct folio *shmem_alloc_and_add order = highest_order(suitable_orders); while (suitable_orders) { pages = 1UL << order; - index = round_down(index, pages); - folio = shmem_alloc_folio(gfp, order, info, index); - if (folio) + folio = shmem_alloc_folio(gfp, order, info, round_down(index, pages)); + if (folio) { + index = round_down(index, pages); goto allocated; + } if (pages == HPAGE_PMD_NR) count_vm_event(THP_FILE_FALLBACK); _ Patches currently in -mm which might be from kasong(a)tencent.com are mm-shmem-fix-thp-allocation-and-fallback-loop.patch

6 days, 18 hours

1
0
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror