December 2025 - Linux-stable-mirror

[PATCH] leds: led-class: Only Add LED to leds_list when it is fully ready

by Hans de Goede

Before this change the LED was added to leds_list before led_init_core() gets called adding it the list before led_classdev.set_brightness_work gets initialized. This leaves a window where led_trigger_register() of a LED's default trigger will call led_trigger_set() which calls led_set_brightness() which in turn will end up queueing the *uninitialized* led_classdev.set_brightness_work. This race gets hit by the lenovo-thinkpad-t14s EC driver which registers 2 LEDs with a default trigger provided by snd_ctl_led.ko in quick succession. The first led_classdev_register() causes an async modprobe of snd_ctl_led to run and that async modprobe manages to exactly hit the window where the second LED is on the leds_list without led_init_core() being called for it, resulting in: ------------[ cut here ]------------ WARNING: CPU: 11 PID: 5608 at kernel/workqueue.c:4234 __flush_work+0x344/0x390 Hardware name: LENOVO 21N2S01F0B/21N2S01F0B, BIOS N42ET93W (2.23 ) 09/01/2025 ... Call trace: __flush_work+0x344/0x390 (P) flush_work+0x2c/0x50 led_trigger_set+0x1c8/0x340 led_trigger_register+0x17c/0x1c0 led_trigger_register_simple+0x84/0xe8 snd_ctl_led_init+0x40/0xf88 [snd_ctl_led] do_one_initcall+0x5c/0x318 do_init_module+0x9c/0x2b8 load_module+0x7e0/0x998 Close the race window by moving the adding of the LED to leds_list to after the led_init_core() call. Cc: Sebastian Reichel <sre(a)kernel.org> Cc: stable(a)vger.kernel.org Signed-off-by: Hans de Goede <johannes.goede(a)oss.qualcomm.com> --- Note no Fixes tag as this problem has been around for a long long time, so I could not really find a good commit for the Fixes tag. --- drivers/leds/led-class.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/drivers/leds/led-class.c b/drivers/leds/led-class.c index f3faf37f9a08..6b9fa060c3a1 100644 --- a/drivers/leds/led-class.c +++ b/drivers/leds/led-class.c @@ -560,11 +560,6 @@ int led_classdev_register_ext(struct device *parent, #ifdef CONFIG_LEDS_BRIGHTNESS_HW_CHANGED led_cdev->brightness_hw_changed = -1; #endif - /* add to the list of leds */ - down_write(&leds_list_lock); - list_add_tail(&led_cdev->node, &leds_list); - up_write(&leds_list_lock); - if (!led_cdev->max_brightness) led_cdev->max_brightness = LED_FULL; @@ -574,6 +569,11 @@ int led_classdev_register_ext(struct device *parent, led_init_core(led_cdev); + /* add to the list of leds */ + down_write(&leds_list_lock); + list_add_tail(&led_cdev->node, &leds_list); + up_write(&leds_list_lock); + #ifdef CONFIG_LEDS_TRIGGERS led_trigger_set_default(led_cdev); #endif -- 2.52.0

2 weeks, 5 days

2
2
0 0

[PATCH] powerpc/pasemi: Fix error handling in pasemi_dma_init

by Miaoqian Lin

Add missing pci_dev_put(iob_pdev) to drop the reference count obtained by pci_get_device() in case of error. Found via static analysis and code review. Fixes: 8ee9d8577935 ("pasemi: DMA engine management library") Cc: stable(a)vger.kernel.org Signed-off-by: Miaoqian Lin <linmq006(a)gmail.com> --- arch/powerpc/platforms/pasemi/dma_lib.c | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/powerpc/platforms/pasemi/dma_lib.c b/arch/powerpc/platforms/pasemi/dma_lib.c index 1be1f18f6f09..b824bfe97ce8 100644 --- a/arch/powerpc/platforms/pasemi/dma_lib.c +++ b/arch/powerpc/platforms/pasemi/dma_lib.c @@ -530,6 +530,7 @@ int pasemi_dma_init(void) BUG(); pr_warn("Can't find DMA controller\n"); err = -ENODEV; + pci_dev_put(iob_pdev); goto out; } dma_regs = map_onedev(dma_pdev, 0); -- 2.25.1

2 weeks, 5 days

1
0
0 0

[PATCH] input: lkkbd: cancel pending work before freeing device

by Minseong Kim

lkkbd_interrupt() schedules lk->tq with schedule_work(), and the work handler lkkbd_reinit() dereferences the lkkbd structure and its serio/input_dev fields. lkkbd_disconnect() frees the lkkbd structure without cancelling this work, so the work can run after the structure has been freed, leading to a potential use-after-free. Cancel the pending work in lkkbd_disconnect() before unregistering and freeing the device, following the same pattern as sunkbd. Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Reported-by: Minseong Kim <ii4gsp(a)gmail.com> Cc: stable(a)vger.kernel.org Signed-off-by: Minseong Kim <ii4gsp(a)gmail.com> --- drivers/input/keyboard/lkkbd.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/input/keyboard/lkkbd.c b/drivers/input/keyboard/lkkbd.c index c035216dd27c..72c477aab1fc 100644 --- a/drivers/input/keyboard/lkkbd.c +++ b/drivers/input/keyboard/lkkbd.c @@ -684,6 +684,8 @@ static void lkkbd_disconnect(struct serio *serio) { struct lkkbd *lk = serio_get_drvdata(serio); + cancel_work_sync(&lk->tq); + input_get_device(lk->dev); input_unregister_device(lk->dev); serio_close(serio); -- 2.39.5

2 weeks, 5 days

2
5
0 0

[tip: x86/urgent] x86/acpi/boot: Correct acpi_is_processor_usable() check again

by tip-bot2 for Yazen Ghannam

The following commit has been merged into the x86/urgent branch of tip: Commit-ID: 4368a3b96c427ea3299be8cedb28684ba4157529 Gitweb: https://git.kernel.org/tip/4368a3b96c427ea3299be8cedb28684ba4157529 Author: Yazen Ghannam <yazen.ghannam(a)amd.com> AuthorDate: Tue, 11 Nov 2025 14:53:57 Committer: Ingo Molnar <mingo(a)kernel.org> CommitterDate: Fri, 12 Dec 2025 09:38:30 +01:00 x86/acpi/boot: Correct acpi_is_processor_usable() check again ACPI v6.3 defined a new "Online Capable" MADT LAPIC flag. This bit is used in conjunction with the "Enabled" MADT LAPIC flag to determine if a CPU can be enabled/hotplugged by the OS after boot. Before the new bit was defined, the "Enabled" bit was explicitly described like this (ACPI v6.0 wording provided): "If zero, this processor is unusable, and the operating system support will not attempt to use it" This means that CPU hotplug (based on MADT) is not possible. Many BIOS implementations follow this guidance. They may include LAPIC entries in MADT for unavailable CPUs, but since these entries are marked with "Enabled=0" it is expected that the OS will completely ignore these entries. However, QEMU will do the same (include entries with "Enabled=0") for the purpose of allowing CPU hotplug within the guest. Comment from QEMU function pc_madt_cpu_entry(): /* ACPI spec says that LAPIC entry for non present * CPU may be omitted from MADT or it must be marked * as disabled. However omitting non present CPU from * MADT breaks hotplug on linux. So possible CPUs * should be put in MADT but kept disabled. */ Recent Linux topology changes broke the QEMU use case. A following fix for the QEMU use case broke bare metal topology enumeration. Rework the Linux MADT LAPIC flags check to allow the QEMU use case only for guests and to maintain the ACPI spec behavior for bare metal. Remove an unnecessary check added to fix a bare metal case introduced by the QEMU "fix". [ bp: Change logic as Michal suggested. ] Fixes: fed8d8773b8e ("x86/acpi/boot: Correct acpi_is_processor_usable() check") Fixes: f0551af02130 ("x86/topology: Ignore non-present APIC IDs in a present package") Closes: https://lore.kernel.org/r/20251024204658.3da9bf3f.michal.pecio@gmail.com Reported-by: Michal Pecio <michal.pecio(a)gmail.com> Signed-off-by: Yazen Ghannam <yazen.ghannam(a)amd.com> Signed-off-by: Borislav Petkov (AMD) <bp(a)alien8.de> Signed-off-by: Ingo Molnar <mingo(a)kernel.org> Tested-by: Michal Pecio <michal.pecio(a)gmail.com> Tested-by: Ricardo Neri <ricardo.neri-calderon(a)linux.intel.com> Cc: stable(a)vger.kernel.org Link: https://lore.kernel.org/20251111145357.4031846-1-yazen.ghannam@amd.com --- arch/x86/kernel/acpi/boot.c | 12 ++++++++---- arch/x86/kernel/cpu/topology.c | 15 --------------- 2 files changed, 8 insertions(+), 19 deletions(-) diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index 9fa321a..d6138b2 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -35,6 +35,7 @@ #include <asm/smp.h> #include <asm/i8259.h> #include <asm/setup.h> +#include <asm/hypervisor.h> #include "sleep.h" /* To include x86_acpi_suspend_lowlevel */ static int __initdata acpi_force = 0; @@ -164,11 +165,14 @@ static bool __init acpi_is_processor_usable(u32 lapic_flags) if (lapic_flags & ACPI_MADT_ENABLED) return true; - if (!acpi_support_online_capable || - (lapic_flags & ACPI_MADT_ONLINE_CAPABLE)) - return true; + if (acpi_support_online_capable) + return lapic_flags & ACPI_MADT_ONLINE_CAPABLE; - return false; + /* + * QEMU expects legacy "Enabled=0" LAPIC entries to be counted as usable + * in order to support CPU hotplug in guests. + */ + return !hypervisor_is_type(X86_HYPER_NATIVE); } static int __init diff --git a/arch/x86/kernel/cpu/topology.c b/arch/x86/kernel/cpu/topology.c index f55ea3c..23190a7 100644 --- a/arch/x86/kernel/cpu/topology.c +++ b/arch/x86/kernel/cpu/topology.c @@ -27,7 +27,6 @@ #include <xen/xen.h> #include <asm/apic.h> -#include <asm/hypervisor.h> #include <asm/io_apic.h> #include <asm/mpspec.h> #include <asm/msr.h> @@ -236,20 +235,6 @@ static __init void topo_register_apic(u32 apic_id, u32 acpi_id, bool present) cpuid_to_apicid[cpu] = apic_id; topo_set_cpuids(cpu, apic_id, acpi_id); } else { - u32 pkgid = topo_apicid(apic_id, TOPO_PKG_DOMAIN); - - /* - * Check for present APICs in the same package when running - * on bare metal. Allow the bogosity in a guest. - */ - if (hypervisor_is_type(X86_HYPER_NATIVE) && - topo_unit_count(pkgid, TOPO_PKG_DOMAIN, phys_cpu_present_map)) { - pr_info_once("Ignoring hot-pluggable APIC ID %x in present package.\n", - apic_id); - topo_info.nr_rejected_cpus++; - return; - } - topo_info.nr_disabled_cpus++; }

2 weeks, 5 days

1
0
0 0

[PATCH v4 0/2] Enable 1GHz OPP am335x-bonegreen-eco

by Kory Maincent (TI.com)

The vdd_mpu regulator maximum voltage was previously limited to 1.2985V, which prevented the CPU from reaching the 1GHz operating point. This limitation was put in place because voltage changes were not working correctly, causing the board to stall when attempting higher frequencies. Increase the maximum voltage to 1.3515V to allow the full 1GHz OPP to be used. Add a TPS65219 PMIC driver fixes that properly implement the LOCK register handling, to make voltage transitions work reliably. Changes in v4: - Move the registers unlock in the probe instead of a custom regmap write operation. - Link to v3: https://lore.kernel.org/r/20251112-fix_tps65219-v3-0-e49bab4c01ce@bootlin.c… Changes in v3: - Remove an unused variable - Link to v2: https://lore.kernel.org/r/20251106-fix_tps65219-v2-0-a7d608c4272f@bootlin.c… Changes in v2: - Setup a custom regmap_bus only for the TPS65214 instead of checking the chip_id every time reg_write is called. - Add the am335x-bonegreen-eco devicetree change in the same patch series. Signed-off-by: Kory Maincent (TI.com) <kory.maincent(a)bootlin.com> --- Kory Maincent (TI.com) (2): mfd: tps65219: Implement LOCK register handling for TPS65214 ARM: dts: am335x-bonegreen-eco: Enable 1GHz OPP by increasing vdd_mpu voltage arch/arm/boot/dts/ti/omap/am335x-bonegreen-eco.dts | 2 +- drivers/mfd/tps65219.c | 7 +++++++ include/linux/mfd/tps65219.h | 2 ++ 3 files changed, 10 insertions(+), 1 deletion(-) --- base-commit: 1c353dc8d962de652bc7ad2ba2e63f553331391c change-id: 20251106-fix_tps65219-dd62141d22cf Best regards, -- Köry Maincent, Bootlin Embedded Linux and kernel engineering https://bootlin.com

2 weeks, 5 days

5
7
0 0

[PATCH] spi: cadence-quadspi: Fix clock enable underflows due to runtime PM

by Mark Brown

The recent refactoring of where runtime PM is enabled done in commit f1eb4e792bb1 ("spi: spi-cadence-quadspi: Enable pm runtime earlier to avoid imbalance") made the fact that when we do a pm_runtime_disable() in the error paths of probe() we can trigger a runtime disable which in turn results in duplicate clock disables. Early on in the probe function we do a pm_runtime_get_noresume() since the probe function leaves the device in a powered up state but in the error path we can't assume that PM is enabled so we also manually disable everything, including clocks. This means that when runtime PM is active both it and the probe function release the same reference to the main clock for the IP, triggering warnings from the clock subsystem: [ 8.693719] clk:75:7 already disabled [ 8.693791] WARNING: CPU: 1 PID: 185 at /usr/src/kernel/drivers/clk/clk.c:1188 clk_core_disable+0xa0/0xb ... [ 8.694261] clk_core_disable+0xa0/0xb4 (P) [ 8.694272] clk_disable+0x38/0x60 [ 8.694283] cqspi_probe+0x7c8/0xc5c [spi_cadence_quadspi] [ 8.694309] platform_probe+0x5c/0xa4 Avoid this confused ownership by moving the pm_runtime_get_noresume() to after the last point at which the probe() function can fail. Reported-by: Francesco Dolcini <francesco(a)dolcini.it> Closes: https://lore.kernel.org/r/20251201072844.GA6785@francesco-nb Signed-off-by: Mark Brown <broonie(a)kernel.org> Cc: stable(a)vger.kernel.org --- drivers/spi/spi-cadence-quadspi.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/spi/spi-cadence-quadspi.c b/drivers/spi/spi-cadence-quadspi.c index af6d050da1c8..0833b6f666d0 100644 --- a/drivers/spi/spi-cadence-quadspi.c +++ b/drivers/spi/spi-cadence-quadspi.c @@ -1985,7 +1985,6 @@ static int cqspi_probe(struct platform_device *pdev) pm_runtime_enable(dev); pm_runtime_set_autosuspend_delay(dev, CQSPI_AUTOSUSPEND_TIMEOUT); pm_runtime_use_autosuspend(dev); - pm_runtime_get_noresume(dev); } ret = cqspi_setup_flash(cqspi); @@ -2012,6 +2011,7 @@ static int cqspi_probe(struct platform_device *pdev) } if (!(ddata && (ddata->quirks & CQSPI_DISABLE_RUNTIME_PM))) { + pm_runtime_get_noresume(dev); pm_runtime_mark_last_busy(dev); pm_runtime_put_autosuspend(dev); } --- base-commit: 7d0a66e4bb9081d75c82ec4957c50034cb0ea449 change-id: 20251202-spi-cadence-qspi-runtime-pm-imbalance-657740cf7eae Best regards, -- Mark Brown <broonie(a)kernel.org>

2 weeks, 5 days

5
18
0 0

[PATCH v2 1/4] mm/hugetlb: fix hugetlb_pmd_shared()

by David Hildenbrand (Red Hat)

We switched from (wrongly) using the page count to an independent shared count. Now, shared page tables have a refcount of 1 (excluding speculative references) and instead use ptdesc->pt_share_count to identify sharing. We didn't convert hugetlb_pmd_shared(), so right now, we would never detect a shared PMD table as such, because sharing/unsharing no longer touches the refcount of a PMD table. Page migration, like mbind() or migrate_pages() would allow for migrating folios mapped into such shared PMD tables, even though the folios are not exclusive. In smaps we would account them as "private" although they are "shared", and we would be wrongly setting the PM_MMAP_EXCLUSIVE in the pagemap interface. Fix it by properly using ptdesc_pmd_is_shared() in hugetlb_pmd_shared(). Fixes: 59d9094df3d7 ("mm: hugetlb: independent PMD page table shared count") Reviewed-by: Rik van Riel <riel(a)surriel.com> Reviewed-by: Lance Yang <lance.yang(a)linux.dev> Tested-by: Lance Yang <lance.yang(a)linux.dev> Reviewed-by: Harry Yoo <harry.yoo(a)oracle.com> Tested-by: Laurence Oberman <loberman(a)redhat.com> Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes(a)oracle.com> Acked-by: Oscar Salvador <osalvador(a)suse.de> Cc: <stable(a)vger.kernel.org> Cc: Liu Shixin <liushixin2(a)huawei.com> Signed-off-by: David Hildenbrand (Red Hat) <david(a)kernel.org> --- include/linux/hugetlb.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h index 019a1c5281e4e..03c8725efa289 100644 --- a/include/linux/hugetlb.h +++ b/include/linux/hugetlb.h @@ -1326,7 +1326,7 @@ static inline __init void hugetlb_cma_reserve(int order) #ifdef CONFIG_HUGETLB_PMD_PAGE_TABLE_SHARING static inline bool hugetlb_pmd_shared(pte_t *pte) { - return page_count(virt_to_page(pte)) > 1; + return ptdesc_pmd_is_shared(virt_to_ptdesc(pte)); } #else static inline bool hugetlb_pmd_shared(pte_t *pte) -- 2.52.0

2 weeks, 5 days

1
0
0 0

[PATCH AUTOSEL 6.18-5.10] alpha: don't reference obsolete termio struct for TC* constants

by Sasha Levin

From: Sam James <sam(a)gentoo.org> [ Upstream commit 9aeed9041929812a10a6d693af050846942a1d16 ] Similar in nature to ab107276607af90b13a5994997e19b7b9731e251. glibc-2.42 drops the legacy termio struct, but the ioctls.h header still defines some TC* constants in terms of termio (via sizeof). Hardcode the values instead. This fixes building Python for example, which falls over like: ./Modules/termios.c:1119:16: error: invalid application of 'sizeof' to incomplete type 'struct termio' Link: https://bugs.gentoo.org/961769 Link: https://bugs.gentoo.org/962600 Signed-off-by: Sam James <sam(a)gentoo.org> Reviewed-by: Magnus Lindholm <linmag7(a)gmail.com> Link: https://lore.kernel.org/r/6ebd3451908785cad53b50ca6bc46cfe9d6bc03c.17649224… Signed-off-by: Magnus Lindholm <linmag7(a)gmail.com> Signed-off-by: Sasha Levin <sashal(a)kernel.org> --- LLM Generated explanations, may be completely bogus: ## Analysis Summary ### 1. COMMIT MESSAGE ANALYSIS This commit addresses a **userspace build failure** caused by glibc-2.42 removing the legacy `struct termio` definition. The kernel's uapi header `arch/alpha/include/uapi/asm/ioctls.h` uses `sizeof(struct termio)` in macro expansions, which fails when that struct is undefined. **Key signals:** - Links to two real bug reports (Gentoo bugs #961769 and #962600) - References a prior fix (ab107276607af) for powerpc with the exact same issue - Has `Reviewed-by:` tag - Demonstrates real-world impact: Python build failure ### 2. CODE CHANGE ANALYSIS **What the change does:** ```c // Before: Uses sizeof(struct termio) in macro expansion #define TCGETA _IOR('t', 23, struct termio) // After: Pre-computed constant #define TCGETA 0x40127417 ``` **Verification of the hardcoded values:** Looking at `arch/alpha/include/uapi/asm/ioctl.h`, the ioctl encoding on alpha is: - `_IOC_SIZESHIFT` = 16, `_IOC_DIRSHIFT` = 29 - `_IOC_READ` = 2, `_IOC_WRITE` = 4 For `TCGETA = _IOR('t', 23, struct termio)`: - dir=2, type=0x74, nr=0x17, size=18(0x12) - Result: `(2<<29)|(0x74<<8)|(0x17)|(0x12<<16)` = **0x40127417** ✓ The hardcoded values are mathematically correct. ### 3. CLASSIFICATION **Category: BUILD FIX** This falls under the **build fixes** exception category - it's critical for users who need to build userspace software with modern glibc. Other architectures (powerpc, sh, xtensa) already have identical fixes in the tree. ### 4. SCOPE AND RISK ASSESSMENT - **Lines changed:** 4 - **Files touched:** 1 (alpha-specific uapi header) - **Risk: EXTREMELY LOW** - No runtime behavior change - only affects compilation - The ioctl numbers are identical (pre-computed vs macro-computed) - Pattern already established in other architectures ### 5. USER IMPACT - **Affected users:** Anyone on alpha using glibc 2.42+ (e.g., Gentoo users) - **Severity:** Critical for affected users - cannot build Python, and potentially many other programs that use termios - **Scope:** Limited to alpha architecture, but complete blocker for those users ### 6. STABILITY INDICATORS - Has `Reviewed-by:` tag - Same pattern accepted for powerpc (commit ab107276607af) - Documented with actual bug reports showing real users affected - No runtime changes - purely compile-time fix ### 7. DEPENDENCY CHECK - **No dependencies** - standalone fix - The affected file exists in all stable trees (uapi header) - Clean application expected ### Final Assessment **Pros:** - Fixes a real, documented build breakage with modern glibc - Extremely small and localized (4 lines, 1 file) - Zero runtime risk - mathematically equivalent values - Established precedent with powerpc fix - Has review tag and bug report links **Cons:** - No explicit `Cc: stable(a)vger.kernel.org` tag - Alpha is a niche architecture **Verdict:** This is textbook stable material. It's a build fix that: 1. Is obviously correct (values are mathematically equivalent) 2. Fixes a real bug (complete build failure) 3. Is small and contained (4 lines) 4. Has no runtime risk whatsoever 5. Follows established precedent from other architectures The lack of a stable tag is not disqualifying - many valid stable patches don't have it. Build compatibility with modern toolchains is exactly what stable trees need to maintain. **YES** arch/alpha/include/uapi/asm/ioctls.h | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/arch/alpha/include/uapi/asm/ioctls.h b/arch/alpha/include/uapi/asm/ioctls.h index 971311605288f..a09d04b49cc65 100644 --- a/arch/alpha/include/uapi/asm/ioctls.h +++ b/arch/alpha/include/uapi/asm/ioctls.h @@ -23,10 +23,10 @@ #define TCSETSW _IOW('t', 21, struct termios) #define TCSETSF _IOW('t', 22, struct termios) -#define TCGETA _IOR('t', 23, struct termio) -#define TCSETA _IOW('t', 24, struct termio) -#define TCSETAW _IOW('t', 25, struct termio) -#define TCSETAF _IOW('t', 28, struct termio) +#define TCGETA 0x40127417 +#define TCSETA 0x80127418 +#define TCSETAW 0x80127419 +#define TCSETAF 0x8012741c #define TCSBRK _IO('t', 29) #define TCXONC _IO('t', 30) -- 2.51.0

2 weeks, 5 days

1
4
0 0

[PATCH 5.15] drm/i915/selftests: fix subtraction overflow bug

by Rajani Kantha

From: Andrzej Hajda <andrzej.hajda(a)intel.com> [ Upstream commit ab3edc679c552a466e4bf0b11af3666008bd65a2 ] On some machines hole_end can be small enough to cause subtraction overflow. On the other side (addr + 2 * min_alignment) can overflow in case of mock tests. This patch should handle both cases. Fixes: e1c5f754067b59 ("drm/i915: Avoid overflow in computing pot_hole loop termination") Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/3674 Signed-off-by: Andrzej Hajda <andrzej.hajda(a)intel.com> Reviewed-by: Andi Shyti <andi.shyti(a)linux.intel.com> Signed-off-by: Andi Shyti <andi.shyti(a)linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20220624113528.2159210-1-andr… Signed-off-by: Rodrigo Vivi <rodrigo.vivi(a)intel.com> [ Using I915_GTT_PAGE_SIZE instead of min_alignment due to 5.15 missing commit:87bd701ee268 ("drm/i915: enforce min GTT alignment for discrete cards")] Signed-off-by: Rajani Kantha <681739313(a)139.com> --- drivers/gpu/drm/i915/selftests/i915_gem_gtt.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c index df3934a990d0..cd0494b84959 100644 --- a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c +++ b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c @@ -715,7 +715,7 @@ static int pot_hole(struct i915_address_space *vm, u64 addr; for (addr = round_up(hole_start + I915_GTT_PAGE_SIZE, step) - I915_GTT_PAGE_SIZE; - addr <= round_down(hole_end - 2*I915_GTT_PAGE_SIZE, step) - I915_GTT_PAGE_SIZE; + hole_end > addr && hole_end - addr >= 2 * I915_GTT_PAGE_SIZE; addr += step) { err = i915_vma_pin(vma, 0, 0, addr | flags); if (err) { -- 2.17.1

2 weeks, 5 days

1
0
0 0

[PATCH v5.10.y] bpf, sockmap: Don't let sock_map_{close,destroy,unhash} call itself

by HarinadhD

From: Jakub Sitnicki <jakub(a)cloudflare.com> [ Upstream commit 5b4a79ba65a1ab479903fff2e604865d229b70a9 ] sock_map proto callbacks should never call themselves by design. Protect against bugs like [1] and break out of the recursive loop to avoid a stack overflow in favor of a resource leak. [1] https://lore.kernel.org/all/00000000000073b14905ef2e7401@google.com/ Suggested-by: Eric Dumazet <edumazet(a)google.com> Signed-off-by: Jakub Sitnicki <jakub(a)cloudflare.com> Acked-by: John Fastabend <john.fastabend(a)gmail.com> Link: https://lore.kernel.org/r/20230113-sockmap-fix-v2-1-1e0ee7ac2f90@cloudflare… Signed-off-by: Alexei Starovoitov <ast(a)kernel.org> Signed-off-by: Sasha Levin <sashal(a)kernel.org> [ Harinadh: Modified to apply on v5.10.y ] Signed-off-by: HarinadhD <Harinadh.Dommaraju(a)broadcom.com> --- net/core/sock_map.c | 53 +++++++++++++++++++++++++-------------------- 1 file changed, 30 insertions(+), 23 deletions(-) diff --git a/net/core/sock_map.c b/net/core/sock_map.c index 3a9e0046a780..438bbef5ff75 100644 --- a/net/core/sock_map.c +++ b/net/core/sock_map.c @@ -1558,15 +1558,16 @@ void sock_map_unhash(struct sock *sk) psock = sk_psock(sk); if (unlikely(!psock)) { rcu_read_unlock(); - if (sk->sk_prot->unhash) - sk->sk_prot->unhash(sk); - return; + saved_unhash = READ_ONCE(sk->sk_prot)->unhash; + } else { + saved_unhash = psock->saved_unhash; + sock_map_remove_links(sk, psock); + rcu_read_unlock(); } - - saved_unhash = psock->saved_unhash; - sock_map_remove_links(sk, psock); - rcu_read_unlock(); - saved_unhash(sk); + if (WARN_ON_ONCE(saved_unhash == sock_map_unhash)) + return; + if (saved_unhash) + saved_unhash(sk); } void sock_map_destroy(struct sock *sk) @@ -1578,16 +1579,17 @@ void sock_map_destroy(struct sock *sk) psock = sk_psock_get(sk); if (unlikely(!psock)) { rcu_read_unlock(); - if (sk->sk_prot->destroy) - sk->sk_prot->destroy(sk); - return; + saved_destroy = READ_ONCE(sk->sk_prot)->destroy; + } else { + saved_destroy = psock->saved_destroy; + sock_map_remove_links(sk, psock); + rcu_read_unlock(); + sk_psock_put(sk, psock); } - - saved_destroy = psock->saved_destroy; - sock_map_remove_links(sk, psock); - rcu_read_unlock(); - sk_psock_put(sk, psock); - saved_destroy(sk); + if (WARN_ON_ONCE(saved_destroy == sock_map_destroy)) + return; + if (saved_destroy) + saved_destroy(sk); } EXPORT_SYMBOL_GPL(sock_map_destroy); @@ -1602,13 +1604,18 @@ void sock_map_close(struct sock *sk, long timeout) if (unlikely(!psock)) { rcu_read_unlock(); release_sock(sk); - return sk->sk_prot->close(sk, timeout); + saved_close = READ_ONCE(sk->sk_prot)->close; + } else { + saved_close = psock->saved_close; + sock_map_remove_links(sk, psock); + rcu_read_unlock(); + release_sock(sk); } - - saved_close = psock->saved_close; - sock_map_remove_links(sk, psock); - rcu_read_unlock(); - release_sock(sk); + /* Make sure we do not recurse. This is a bug. + * Leak the socket instead of crashing on a stack overflow. + */ + if (WARN_ON_ONCE(saved_close == sock_map_close)) + return; saved_close(sk, timeout); } -- 2.43.7

2 weeks, 5 days

1
0
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror December 2025