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

[PATCH 12/15] dmaengine: ti: dma-crossbar: fix device leak on dra7x route allocation

by Johan Hovold

Make sure to drop the reference taken when looking up the crossbar platform device during dra7x route allocation. Note that commit 615a4bfc426e ("dmaengine: ti: Add missing put_device in ti_dra7_xbar_route_allocate") fixed the leak in the error paths but the reference is still leaking on successful allocation. Fixes: a074ae38f859 ("dmaengine: Add driver for TI DMA crossbar on DRA7x") Fixes: 615a4bfc426e ("dmaengine: ti: Add missing put_device in ti_dra7_xbar_route_allocate") Cc: stable(a)vger.kernel.org # 4.2: 615a4bfc426e Cc: Peter Ujfalusi <peter.ujfalusi(a)ti.com> Cc: Miaoqian Lin <linmq006(a)gmail.com> Signed-off-by: Johan Hovold <johan(a)kernel.org> --- drivers/dma/ti/dma-crossbar.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/dma/ti/dma-crossbar.c b/drivers/dma/ti/dma-crossbar.c index 7f17ee87a6dc..e52b0e139900 100644 --- a/drivers/dma/ti/dma-crossbar.c +++ b/drivers/dma/ti/dma-crossbar.c @@ -288,6 +288,8 @@ static void *ti_dra7_xbar_route_allocate(struct of_phandle_args *dma_spec, ti_dra7_xbar_write(xbar->iomem, map->xbar_out, map->xbar_in); + put_device(&pdev->dev); + return map; } -- 2.51.0

3 days, 23 hours

1
0
0 0

[PATCH 03/15] dmaengine: cv1800b-dmamux: fix device leak on route allocation

by Johan Hovold

Make sure to drop the reference taken when looking up the DMA mux platform device during route allocation. Note that holding a reference to a device does not prevent its driver data from going away so there is no point in keeping the reference. Fixes: db7d07b5add4 ("dmaengine: add driver for Sophgo CV18XX/SG200X dmamux") Cc: stable(a)vger.kernel.org # 6.17 Cc: Inochi Amaoto <inochiama(a)gmail.com> Signed-off-by: Johan Hovold <johan(a)kernel.org> --- drivers/dma/cv1800b-dmamux.c | 17 ++++++++++------- 1 file changed, 10 insertions(+), 7 deletions(-) diff --git a/drivers/dma/cv1800b-dmamux.c b/drivers/dma/cv1800b-dmamux.c index e900d6595617..f7a952fcbc7d 100644 --- a/drivers/dma/cv1800b-dmamux.c +++ b/drivers/dma/cv1800b-dmamux.c @@ -102,11 +102,11 @@ static void *cv1800_dmamux_route_allocate(struct of_phandle_args *dma_spec, struct llist_node *node; unsigned long flags; unsigned int chid, devid, cpuid; - int ret; + int ret = -EINVAL; if (dma_spec->args_count != DMAMUX_NCELLS) { dev_err(&pdev->dev, "invalid number of dma mux args\n"); - return ERR_PTR(-EINVAL); + goto err_put_pdev; } devid = dma_spec->args[0]; @@ -115,18 +115,18 @@ static void *cv1800_dmamux_route_allocate(struct of_phandle_args *dma_spec, if (devid > MAX_DMA_MAPPING_ID) { dev_err(&pdev->dev, "invalid device id: %u\n", devid); - return ERR_PTR(-EINVAL); + goto err_put_pdev; } if (cpuid > MAX_DMA_CPU_ID) { dev_err(&pdev->dev, "invalid cpu id: %u\n", cpuid); - return ERR_PTR(-EINVAL); + goto err_put_pdev; } dma_spec->np = of_parse_phandle(ofdma->of_node, "dma-masters", 0); if (!dma_spec->np) { dev_err(&pdev->dev, "can't get dma master\n"); - return ERR_PTR(-EINVAL); + goto err_put_pdev; } spin_lock_irqsave(&dmamux->lock, flags); @@ -136,8 +136,6 @@ static void *cv1800_dmamux_route_allocate(struct of_phandle_args *dma_spec, if (map->peripheral == devid && map->cpu == cpuid) goto found; } - - ret = -EINVAL; goto failed; } else { node = llist_del_first(&dmamux->free_maps); @@ -171,12 +169,17 @@ static void *cv1800_dmamux_route_allocate(struct of_phandle_args *dma_spec, dev_dbg(&pdev->dev, "register channel %u for req %u (cpu %u)\n", chid, devid, cpuid); + put_device(&pdev->dev); + return map; failed: spin_unlock_irqrestore(&dmamux->lock, flags); of_node_put(dma_spec->np); dev_err(&pdev->dev, "errno %d\n", ret); +err_put_pdev: + put_device(&pdev->dev); + return ERR_PTR(ret); } -- 2.51.0

3 days, 23 hours

1
0
0 0

[PATCH 02/15] dmaengine: bcm-sba-raid: fix device leak on probe

by Johan Hovold

Make sure to drop the reference taken when looking up the mailbox device during probe on probe failures and on driver unbind. Fixes: 743e1c8ffe4e ("dmaengine: Add Broadcom SBA RAID driver") Cc: stable(a)vger.kernel.org # 4.13 Signed-off-by: Johan Hovold <johan(a)kernel.org> --- drivers/dma/bcm-sba-raid.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/drivers/dma/bcm-sba-raid.c b/drivers/dma/bcm-sba-raid.c index 7f0e76439ce5..ed037fa883f6 100644 --- a/drivers/dma/bcm-sba-raid.c +++ b/drivers/dma/bcm-sba-raid.c @@ -1699,7 +1699,7 @@ static int sba_probe(struct platform_device *pdev) /* Prealloc channel resource */ ret = sba_prealloc_channel_resources(sba); if (ret) - goto fail_free_mchan; + goto fail_put_mbox; /* Check availability of debugfs */ if (!debugfs_initialized()) @@ -1729,6 +1729,8 @@ static int sba_probe(struct platform_device *pdev) fail_free_resources: debugfs_remove_recursive(sba->root); sba_freeup_channel_resources(sba); +fail_put_mbox: + put_device(sba->mbox_dev); fail_free_mchan: mbox_free_channel(sba->mchan); return ret; @@ -1744,6 +1746,8 @@ static void sba_remove(struct platform_device *pdev) sba_freeup_channel_resources(sba); + put_device(sba->mbox_dev); + mbox_free_channel(sba->mchan); } -- 2.51.0

3 days, 23 hours

1
0
0 0

[PATCH] usb: dwc3: gadget: Prevent EPs resource conflict during StartTransfer

by Selvarasu Ganesan

This commit fixes the below warning that occurs when a StartTransfer command is issued for bulk or interrupt endpoints in `dwc3_gadget_ep_enable` while a previous transfer on the same endpoint is still in progress. The gadget functions drivers can invoke usb_ep_enable (which triggers a new StartTransfer command) before the earlier transfer has completed. Because the previous StartTransfer is still active, `dwc3_gadget_ep_disable` can skip the required `EndTransfer` due to `DWC3_EP_DELAY_STOP`, leading to the endpoint resources are busy for previous StartTransfer and warning ("No resource for ep") from gadget driver. To resolve this, a check is added to `dwc3_gadget_ep_enable` that checks the `DWC3_EP_TRANSFER_STARTED` flag before issuing a new StartTransfer. By preventing a second StartTransfer on an already busy endpoint, the resource conflict is eliminated, the warning disappears, and potential kernel panics caused by `panic_on_warn` are avoided. ------------[ cut here ]------------ dwc3 13200000.dwc3: No resource for ep1out WARNING: CPU: 0 PID: 700 at drivers/usb/dwc3/gadget.c:398 dwc3_send_gadget_ep_cmd+0x2f8/0x76c Call trace: dwc3_send_gadget_ep_cmd+0x2f8/0x76c __dwc3_gadget_ep_enable+0x490/0x7c0 dwc3_gadget_ep_enable+0x6c/0xe4 usb_ep_enable+0x5c/0x15c mp_eth_stop+0xd4/0x11c __dev_close_many+0x160/0x1c8 __dev_change_flags+0xfc/0x220 dev_change_flags+0x24/0x70 devinet_ioctl+0x434/0x524 inet_ioctl+0xa8/0x224 sock_do_ioctl+0x74/0x128 sock_ioctl+0x3bc/0x468 __arm64_sys_ioctl+0xa8/0xe4 invoke_syscall+0x58/0x10c el0_svc_common+0xa8/0xdc do_el0_svc+0x1c/0x28 el0_svc+0x38/0x88 el0t_64_sync_handler+0x70/0xbc el0t_64_sync+0x1a8/0x1ac Change-Id: Id292265a34448e566ef1ea882e313856423342dc Fixes: a97ea994605e ("usb: dwc3: gadget: offset Start Transfer latency for bulk EPs") Cc: stable(a)vger.kernel.org Signed-off-by: Selvarasu Ganesan <selvarasu.g(a)samsung.com> --- drivers/usb/dwc3/gadget.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c index f95d1369bbc6..23e5c111da7c 100644 --- a/drivers/usb/dwc3/gadget.c +++ b/drivers/usb/dwc3/gadget.c @@ -951,8 +951,9 @@ static int __dwc3_gadget_ep_enable(struct dwc3_ep *dep, unsigned int action) * Issue StartTransfer here with no-op TRB so we can always rely on No * Response Update Transfer command. */ - if (usb_endpoint_xfer_bulk(desc) || - usb_endpoint_xfer_int(desc)) { + if ((usb_endpoint_xfer_bulk(desc) || + usb_endpoint_xfer_int(desc)) && + !(dep->flags & DWC3_EP_TRANSFER_STARTED)) { struct dwc3_gadget_ep_cmd_params params; struct dwc3_trb *trb; dma_addr_t trb_dma; -- 2.34.1

3 days, 23 hours

1
1
0 0

[PATCH] drm/kmb: Fix error handling in kmb_probe

by Ma Ke

kmb_probe() obtain a reference to a platform device by of_find_device_by_node(). This call increases the reference count of the returned device, which should be dropped by calling put_device() when the device is no longer needed. However, the code fails to call put_device() in several error handling paths and the normal device removal path. This could result in reference count leaks that prevent the proper cleanup of the platform device when the driver is unloaded or during error recovery. Add put_device() in all code paths where dsi_pdev is no longer needed, including error paths and the normal removal path. Found by code review. Cc: stable(a)vger.kernel.org Fixes: 7f7b96a8a0a1 ("drm/kmb: Add support for KeemBay Display") Signed-off-by: Ma Ke <make24(a)iscas.ac.cn> --- drivers/gpu/drm/kmb/kmb_drv.c | 16 +++++++++++----- 1 file changed, 11 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/drm/kmb/kmb_drv.c b/drivers/gpu/drm/kmb/kmb_drv.c index 32cda134ae3e..4fc9fdf92118 100644 --- a/drivers/gpu/drm/kmb/kmb_drv.c +++ b/drivers/gpu/drm/kmb/kmb_drv.c @@ -473,6 +473,8 @@ static void kmb_remove(struct platform_device *pdev) /* Unregister DSI host */ kmb_dsi_host_unregister(kmb->kmb_dsi); + if (kmb->kmb_dsi && kmb->kmb_dsi->pdev) + put_device(&kmb->kmb_dsi->pdev->dev); drm_atomic_helper_shutdown(drm); } @@ -517,17 +519,20 @@ static int kmb_probe(struct platform_device *pdev) ret = kmb_dsi_host_bridge_init(get_device(&dsi_pdev->dev)); if (ret == -EPROBE_DEFER) { - return -EPROBE_DEFER; + ret = -EPROBE_DEFER; + goto err_free2; } else if (ret) { DRM_ERROR("probe failed to initialize DSI host bridge\n"); - return ret; + goto err_free2; } /* Create DRM device */ kmb = devm_drm_dev_alloc(dev, &kmb_driver, struct kmb_drm_private, drm); - if (IS_ERR(kmb)) - return PTR_ERR(kmb); + if (IS_ERR(kmb)) { + ret = PTR_ERR(kmb); + goto err_free2; + } dev_set_drvdata(dev, &kmb->drm); @@ -576,7 +581,8 @@ static int kmb_probe(struct platform_device *pdev) err_free1: dev_set_drvdata(dev, NULL); kmb_dsi_host_unregister(kmb->kmb_dsi); - + err_free2: + put_device(&dsi_pdev->dev); return ret; } -- 2.17.1

4 days

2
1
0 0

[PATCH 5.4.y] mm/ksm: fix flag-dropping behavior in ksm_madvise

by Jakub Acs

[ Upstream commit f04aad36a07cc17b7a5d5b9a2d386ce6fae63e93 ] syzkaller discovered the following crash: (kernel BUG) [ 44.607039] ------------[ cut here ]------------ [ 44.607422] kernel BUG at mm/userfaultfd.c:2067! [ 44.608148] Oops: invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN NOPTI [ 44.608814] CPU: 1 UID: 0 PID: 2475 Comm: reproducer Not tainted 6.16.0-rc6 #1 PREEMPT(none) [ 44.609635] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/2014 [ 44.610695] RIP: 0010:userfaultfd_release_all+0x3a8/0x460 <snip other registers, drop unreliable trace> [ 44.617726] Call Trace: [ 44.617926] <TASK> [ 44.619284] userfaultfd_release+0xef/0x1b0 [ 44.620976] __fput+0x3f9/0xb60 [ 44.621240] fput_close_sync+0x110/0x210 [ 44.622222] __x64_sys_close+0x8f/0x120 [ 44.622530] do_syscall_64+0x5b/0x2f0 [ 44.622840] entry_SYSCALL_64_after_hwframe+0x76/0x7e [ 44.623244] RIP: 0033:0x7f365bb3f227 Kernel panics because it detects UFFD inconsistency during userfaultfd_release_all(). Specifically, a VMA which has a valid pointer to vma->vm_userfaultfd_ctx, but no UFFD flags in vma->vm_flags. The inconsistency is caused in ksm_madvise(): when user calls madvise() with MADV_UNMEARGEABLE on a VMA that is registered for UFFD in MINOR mode, it accidentally clears all flags stored in the upper 32 bits of vma->vm_flags. Assuming x86_64 kernel build, unsigned long is 64-bit and unsigned int and int are 32-bit wide. This setup causes the following mishap during the &= ~VM_MERGEABLE assignment. VM_MERGEABLE is a 32-bit constant of type unsigned int, 0x8000'0000. After ~ is applied, it becomes 0x7fff'ffff unsigned int, which is then promoted to unsigned long before the & operation. This promotion fills upper 32 bits with leading 0s, as we're doing unsigned conversion (and even for a signed conversion, this wouldn't help as the leading bit is 0). & operation thus ends up AND-ing vm_flags with 0x0000'0000'7fff'ffff instead of intended 0xffff'ffff'7fff'ffff and hence accidentally clears the upper 32-bits of its value. Fix it by changing `VM_MERGEABLE` constant to unsigned long, using the BIT() macro. Note: other VM_* flags are not affected: This only happens to the VM_MERGEABLE flag, as the other VM_* flags are all constants of type int and after ~ operation, they end up with leading 1 and are thus converted to unsigned long with leading 1s. Note 2: After commit 31defc3b01d9 ("userfaultfd: remove (VM_)BUG_ON()s"), this is no longer a kernel BUG, but a WARNING at the same place: [ 45.595973] WARNING: CPU: 1 PID: 2474 at mm/userfaultfd.c:2067 but the root-cause (flag-drop) remains the same. [akpm(a)linux-foundation.org: rust bindgen wasn't able to handle BIT(), from Miguel] Link: https://lore.kernel.org/oe-kbuild-all/202510030449.VfSaAjvd-lkp@intel.com/ Link: https://lkml.kernel.org/r/20251001090353.57523-2-acsjakub@amazon.de Fixes: 7677f7fd8be7 ("userfaultfd: add minor fault registration mode") Signed-off-by: Jakub Acs <acsjakub(a)amazon.de> Signed-off-by: Miguel Ojeda <miguel.ojeda.sandonis(a)gmail.com> Acked-by: David Hildenbrand <david(a)redhat.com> Acked-by: SeongJae Park <sj(a)kernel.org> Tested-by: Alice Ryhl <aliceryhl(a)google.com> Tested-by: Miguel Ojeda <miguel.ojeda.sandonis(a)gmail.com> Cc: Xu Xin <xu.xin16(a)zte.com.cn> Cc: Chengming Zhou <chengming.zhou(a)linux.dev> Cc: Peter Xu <peterx(a)redhat.com> Cc: Axel Rasmussen <axelrasmussen(a)google.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> [ acsjakub: drop rust-compatibility change (no rust in 5.4) ] Signed-off-by: Jakub Acs <acsjakub(a)amazon.de> --- Why sending to stable version from before "fixes"? In the original patch, I set fixes tag to the change that allows the panic to manifest, not to the one that is real root-cause of the problem. The change that introduced the root-cause of the problem is: f8af4da3b4c1 ("ksm: the mm interface to ksm"), as pointed out by Vlastimil in [1]. Hence, as the older kernels can be affected by the flag-drop as well, backport to older kernels. [1]: https://lore.kernel.org/all/13c7242e-3a40-469b-9e99-8a65a21449bb@suse.cz/ include/linux/mm.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/include/linux/mm.h b/include/linux/mm.h index 57cba6e4fdcd..be8c793233d3 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -293,7 +293,7 @@ extern unsigned int kobjsize(const void *objp); #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */ #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */ -#define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ +#define VM_MERGEABLE BIT(31) /* KSM may merge identical pages */ #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS #define VM_HIGH_ARCH_BIT_0 32 /* bit only usable on 64-bit architectures */ -- 2.47.3 Amazon Web Services Development Center Germany GmbH Tamara-Danz-Str. 13 10243 Berlin Geschaeftsfuehrung: Christian Schlaeger, Christof Hellmis Eingetragen am Amtsgericht Charlottenburg unter HRB 257764 B Sitz: Berlin Ust-ID: DE 365 538 597

4 days

2
1
0 0

[PATCH 5.10] lib/crypto: arm/curve25519: Disable on CPU_BIG_ENDIAN

by Eric Biggers

commit 44e8241c51f762aafa50ed116da68fd6ecdcc954 upstream. On big endian arm kernels, the arm optimized Curve25519 code produces incorrect outputs and fails the Curve25519 test. This has been true ever since this code was added. It seems that hardly anyone (or even no one?) actually uses big endian arm kernels. But as long as they're ostensibly supported, we should disable this code on them so that it's not accidentally used. Note: for future-proofing, use !CPU_BIG_ENDIAN instead of CPU_LITTLE_ENDIAN. Both of these are arch-specific options that could get removed in the future if big endian support gets dropped. Fixes: d8f1308a025f ("crypto: arm/curve25519 - wire up NEON implementation") Cc: stable(a)vger.kernel.org Acked-by: Ard Biesheuvel <ardb(a)kernel.org> Link: https://lore.kernel.org/r/20251104054906.716914-1-ebiggers@kernel.org Signed-off-by: Eric Biggers <ebiggers(a)kernel.org> --- arch/arm/crypto/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/arm/crypto/Kconfig b/arch/arm/crypto/Kconfig index c46c05548080..c5d676e7f16b 100644 --- a/arch/arm/crypto/Kconfig +++ b/arch/arm/crypto/Kconfig @@ -145,10 +145,10 @@ config CRYPTO_NHPOLY1305_NEON depends on KERNEL_MODE_NEON select CRYPTO_NHPOLY1305 config CRYPTO_CURVE25519_NEON tristate "NEON accelerated Curve25519 scalar multiplication library" - depends on KERNEL_MODE_NEON + depends on KERNEL_MODE_NEON && !CPU_BIG_ENDIAN select CRYPTO_LIB_CURVE25519_GENERIC select CRYPTO_ARCH_HAVE_LIB_CURVE25519 endif base-commit: df70e44fa05b01476a78d0f6a210354784ff0992 -- 2.51.2

4 days

2
1
0 0

[PATCH] dmaengine: ti-dma-crossbar: Fix error handling in ti_am335x_xbar_route_allocate

by Ma Ke

ti_am335x_xbar_route_allocate() calls of_find_device_by_node() which increments the reference count of the platform device, but fails to call put_device() to decrement the reference count before returning. This could cause a reference count leak each time the function is called, preventing the platform device from being properly cleaned up and leading to memory leakage. Add proper put_device() calls in all exit paths to fix the reference count imbalance. Found by code review. Cc: stable(a)vger.kernel.org Fixes: 42dbdcc6bf96 ("dmaengine: ti-dma-crossbar: Add support for crossbar on AM33xx/AM43xx") Signed-off-by: Ma Ke <make24(a)iscas.ac.cn> --- drivers/dma/ti/dma-crossbar.c | 24 ++++++++++++++++++------ 1 file changed, 18 insertions(+), 6 deletions(-) diff --git a/drivers/dma/ti/dma-crossbar.c b/drivers/dma/ti/dma-crossbar.c index 7f17ee87a6dc..58bc515ec8b3 100644 --- a/drivers/dma/ti/dma-crossbar.c +++ b/drivers/dma/ti/dma-crossbar.c @@ -80,33 +80,40 @@ static void *ti_am335x_xbar_route_allocate(struct of_phandle_args *dma_spec, struct platform_device *pdev = of_find_device_by_node(ofdma->of_node); struct ti_am335x_xbar_data *xbar = platform_get_drvdata(pdev); struct ti_am335x_xbar_map *map; + int ret; - if (dma_spec->args_count != 3) - return ERR_PTR(-EINVAL); + if (dma_spec->args_count != 3) { + ret = -EINVAL; + goto err_put_device; + } if (dma_spec->args[2] >= xbar->xbar_events) { dev_err(&pdev->dev, "Invalid XBAR event number: %d\n", dma_spec->args[2]); - return ERR_PTR(-EINVAL); + ret = -EINVAL; + goto err_put_device; } if (dma_spec->args[0] >= xbar->dma_requests) { dev_err(&pdev->dev, "Invalid DMA request line number: %d\n", dma_spec->args[0]); - return ERR_PTR(-EINVAL); + ret = -EINVAL; + goto err_put_device; } /* The of_node_put() will be done in the core for the node */ dma_spec->np = of_parse_phandle(ofdma->of_node, "dma-masters", 0); if (!dma_spec->np) { dev_err(&pdev->dev, "Can't get DMA master\n"); - return ERR_PTR(-EINVAL); + ret = -EINVAL; + goto err_put_device; } map = kzalloc(sizeof(*map), GFP_KERNEL); if (!map) { of_node_put(dma_spec->np); - return ERR_PTR(-ENOMEM); + ret = -ENOMEM; + goto err_put_device; } map->dma_line = (u16)dma_spec->args[0]; @@ -120,7 +127,12 @@ static void *ti_am335x_xbar_route_allocate(struct of_phandle_args *dma_spec, ti_am335x_xbar_write(xbar->iomem, map->dma_line, map->mux_val); + put_device(&pdev->dev); return map; + +err_put_device: + put_device(&pdev->dev); + return ERR_PTR(ret); } static const struct of_device_id ti_am335x_master_match[] __maybe_unused = { -- 2.17.1

4 days

2
1
0 0

[PATCH 5.10.y] mm/ksm: fix flag-dropping behavior in ksm_madvise

by Jakub Acs

[ Upstream commit f04aad36a07cc17b7a5d5b9a2d386ce6fae63e93 ] syzkaller discovered the following crash: (kernel BUG) [ 44.607039] ------------[ cut here ]------------ [ 44.607422] kernel BUG at mm/userfaultfd.c:2067! [ 44.608148] Oops: invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN NOPTI [ 44.608814] CPU: 1 UID: 0 PID: 2475 Comm: reproducer Not tainted 6.16.0-rc6 #1 PREEMPT(none) [ 44.609635] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/2014 [ 44.610695] RIP: 0010:userfaultfd_release_all+0x3a8/0x460 <snip other registers, drop unreliable trace> [ 44.617726] Call Trace: [ 44.617926] <TASK> [ 44.619284] userfaultfd_release+0xef/0x1b0 [ 44.620976] __fput+0x3f9/0xb60 [ 44.621240] fput_close_sync+0x110/0x210 [ 44.622222] __x64_sys_close+0x8f/0x120 [ 44.622530] do_syscall_64+0x5b/0x2f0 [ 44.622840] entry_SYSCALL_64_after_hwframe+0x76/0x7e [ 44.623244] RIP: 0033:0x7f365bb3f227 Kernel panics because it detects UFFD inconsistency during userfaultfd_release_all(). Specifically, a VMA which has a valid pointer to vma->vm_userfaultfd_ctx, but no UFFD flags in vma->vm_flags. The inconsistency is caused in ksm_madvise(): when user calls madvise() with MADV_UNMEARGEABLE on a VMA that is registered for UFFD in MINOR mode, it accidentally clears all flags stored in the upper 32 bits of vma->vm_flags. Assuming x86_64 kernel build, unsigned long is 64-bit and unsigned int and int are 32-bit wide. This setup causes the following mishap during the &= ~VM_MERGEABLE assignment. VM_MERGEABLE is a 32-bit constant of type unsigned int, 0x8000'0000. After ~ is applied, it becomes 0x7fff'ffff unsigned int, which is then promoted to unsigned long before the & operation. This promotion fills upper 32 bits with leading 0s, as we're doing unsigned conversion (and even for a signed conversion, this wouldn't help as the leading bit is 0). & operation thus ends up AND-ing vm_flags with 0x0000'0000'7fff'ffff instead of intended 0xffff'ffff'7fff'ffff and hence accidentally clears the upper 32-bits of its value. Fix it by changing `VM_MERGEABLE` constant to unsigned long, using the BIT() macro. Note: other VM_* flags are not affected: This only happens to the VM_MERGEABLE flag, as the other VM_* flags are all constants of type int and after ~ operation, they end up with leading 1 and are thus converted to unsigned long with leading 1s. Note 2: After commit 31defc3b01d9 ("userfaultfd: remove (VM_)BUG_ON()s"), this is no longer a kernel BUG, but a WARNING at the same place: [ 45.595973] WARNING: CPU: 1 PID: 2474 at mm/userfaultfd.c:2067 but the root-cause (flag-drop) remains the same. [akpm(a)linux-foundation.org: rust bindgen wasn't able to handle BIT(), from Miguel] Link: https://lore.kernel.org/oe-kbuild-all/202510030449.VfSaAjvd-lkp@intel.com/ Link: https://lkml.kernel.org/r/20251001090353.57523-2-acsjakub@amazon.de Fixes: 7677f7fd8be7 ("userfaultfd: add minor fault registration mode") Signed-off-by: Jakub Acs <acsjakub(a)amazon.de> Signed-off-by: Miguel Ojeda <miguel.ojeda.sandonis(a)gmail.com> Acked-by: David Hildenbrand <david(a)redhat.com> Acked-by: SeongJae Park <sj(a)kernel.org> Tested-by: Alice Ryhl <aliceryhl(a)google.com> Tested-by: Miguel Ojeda <miguel.ojeda.sandonis(a)gmail.com> Cc: Xu Xin <xu.xin16(a)zte.com.cn> Cc: Chengming Zhou <chengming.zhou(a)linux.dev> Cc: Peter Xu <peterx(a)redhat.com> Cc: Axel Rasmussen <axelrasmussen(a)google.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> [ acsjakub: drop rust-compatibility change (no rust in 5.10) ] Signed-off-by: Jakub Acs <acsjakub(a)amazon.de> --- Why sending to stable version from before "fixes"? In the original patch, I set fixes tag to the change that allows the panic to manifest, not to the one that is real root-cause of the problem. The change that introduced the root-cause of the problem is: f8af4da3b4c1 ("ksm: the mm interface to ksm"), as pointed out by Vlastimil in [1]. Hence, as the older kernels can be affected by the flag-drop as well, backport to older kernels. [1]: https://lore.kernel.org/all/13c7242e-3a40-469b-9e99-8a65a21449bb@suse.cz/ include/linux/mm.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/include/linux/mm.h b/include/linux/mm.h index e168d87d6f2e..4787d39bbad4 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -296,7 +296,7 @@ extern unsigned int kobjsize(const void *objp); #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */ #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */ -#define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ +#define VM_MERGEABLE BIT(31) /* KSM may merge identical pages */ #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS #define VM_HIGH_ARCH_BIT_0 32 /* bit only usable on 64-bit architectures */ -- 2.47.3 Amazon Web Services Development Center Germany GmbH Tamara-Danz-Str. 13 10243 Berlin Geschaeftsfuehrung: Christian Schlaeger, Christof Hellmis Eingetragen am Amtsgericht Charlottenburg unter HRB 257764 B Sitz: Berlin Ust-ID: DE 365 538 597

4 days

2
1
0 0

[PATCH v2 5.10.y] fsdax: mark the iomap argument to dax_iomap_sector as const

by Eliav Farber

From: Christoph Hellwig <hch(a)lst.de> [ Upstream commit 7e4f4b2d689d959b03cb07dfbdb97b9696cb1076 ] Signed-off-by: Christoph Hellwig <hch(a)lst.de> Reviewed-by: Darrick J. Wong <djwong(a)kernel.org> Signed-off-by: Darrick J. Wong <djwong(a)kernel.org> Signed-off-by: Eliav Farber <farbere(a)amazon.com> --- V1 -> V2: Added missing 'Signed-off-by' Fixes: fs/dax.c: In function 'dax_iomap_iter': fs/dax.c:1147:44: error: passing argument 1 of 'dax_iomap_sector' discards 'const' qualifier from pointer target type [-Werror=discarded-qualifiers] const sector_t sector = dax_iomap_sector(iomap, pos); ^~~~~ fs/dax.c:1009:17: note: expected 'struct iomap *' but argument is of type 'const struct iomap *' static sector_t dax_iomap_sector(struct iomap *iomap, loff_t pos) ^~~~~~~~~~~~~~~~ The issue was introduced by the cherry-pick of commit 8df4919cb921 ("fsdax: switch dax_iomap_rw to use iomap_iter") https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/fs/… The upstream change made callers pass a const struct iomap *: const struct iomap *iomap = &iomi->iomap; but dax_iomap_sector() still expected a mutable pointer: static sector_t dax_iomap_sector(struct iomap *iomap, loff_t pos) fs/dax.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/dax.c b/fs/dax.c index 91820b9b50b7..2ca33ef5d519 100644 --- a/fs/dax.c +++ b/fs/dax.c @@ -1006,7 +1006,7 @@ int dax_writeback_mapping_range(struct address_space *mapping, } EXPORT_SYMBOL_GPL(dax_writeback_mapping_range); -static sector_t dax_iomap_sector(struct iomap *iomap, loff_t pos) +static sector_t dax_iomap_sector(const struct iomap *iomap, loff_t pos) { return (iomap->addr + (pos & PAGE_MASK) - iomap->offset) >> 9; } -- 2.47.3

4 days, 1 hour

2
1
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror