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

FAILED: patch "[PATCH] KVM: x86: Reject Hyper-V's SEND_IPI hypercalls if local APIC" failed to apply to 5.10-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.10-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. To reproduce the conflict and resubmit, you may use the following commands: git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-5.10.y git checkout FETCH_HEAD git cherry-pick -x a8de7f100bb5989d9c3627d3a223ee1c863f3b69 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025021850-exes-cabana-e868@gregkh' --subject-prefix 'PATCH 5.10.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From a8de7f100bb5989d9c3627d3a223ee1c863f3b69 Mon Sep 17 00:00:00 2001 From: Sean Christopherson <seanjc(a)google.com> Date: Fri, 17 Jan 2025 16:34:51 -0800 Subject: [PATCH] KVM: x86: Reject Hyper-V's SEND_IPI hypercalls if local APIC isn't in-kernel Advertise support for Hyper-V's SEND_IPI and SEND_IPI_EX hypercalls if and only if the local API is emulated/virtualized by KVM, and explicitly reject said hypercalls if the local APIC is emulated in userspace, i.e. don't rely on userspace to opt-in to KVM_CAP_HYPERV_ENFORCE_CPUID. Rejecting SEND_IPI and SEND_IPI_EX fixes a NULL-pointer dereference if Hyper-V enlightenments are exposed to the guest without an in-kernel local APIC: dump_stack+0xbe/0xfd __kasan_report.cold+0x34/0x84 kasan_report+0x3a/0x50 __apic_accept_irq+0x3a/0x5c0 kvm_hv_send_ipi.isra.0+0x34e/0x820 kvm_hv_hypercall+0x8d9/0x9d0 kvm_emulate_hypercall+0x506/0x7e0 __vmx_handle_exit+0x283/0xb60 vmx_handle_exit+0x1d/0xd0 vcpu_enter_guest+0x16b0/0x24c0 vcpu_run+0xc0/0x550 kvm_arch_vcpu_ioctl_run+0x170/0x6d0 kvm_vcpu_ioctl+0x413/0xb20 __se_sys_ioctl+0x111/0x160 do_syscal1_64+0x30/0x40 entry_SYSCALL_64_after_hwframe+0x67/0xd1 Note, checking the sending vCPU is sufficient, as the per-VM irqchip_mode can't be modified after vCPUs are created, i.e. if one vCPU has an in-kernel local APIC, then all vCPUs have an in-kernel local APIC. Reported-by: Dongjie Zou <zoudongjie(a)huawei.com> Fixes: 214ff83d4473 ("KVM: x86: hyperv: implement PV IPI send hypercalls") Fixes: 2bc39970e932 ("x86/kvm/hyper-v: Introduce KVM_GET_SUPPORTED_HV_CPUID") Cc: stable(a)vger.kernel.org Reviewed-by: Vitaly Kuznetsov <vkuznets(a)redhat.com> Link: https://lore.kernel.org/r/20250118003454.2619573-2-seanjc@google.com Signed-off-by: Sean Christopherson <seanjc(a)google.com> diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c index 6a6dd5a84f22..6ebeb6cea6c0 100644 --- a/arch/x86/kvm/hyperv.c +++ b/arch/x86/kvm/hyperv.c @@ -2226,6 +2226,9 @@ static u64 kvm_hv_send_ipi(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc) u32 vector; bool all_cpus; + if (!lapic_in_kernel(vcpu)) + return HV_STATUS_INVALID_HYPERCALL_INPUT; + if (hc->code == HVCALL_SEND_IPI) { if (!hc->fast) { if (unlikely(kvm_read_guest(kvm, hc->ingpa, &send_ipi, @@ -2852,7 +2855,8 @@ int kvm_get_hv_cpuid(struct kvm_vcpu *vcpu, struct kvm_cpuid2 *cpuid, ent->eax |= HV_X64_REMOTE_TLB_FLUSH_RECOMMENDED; ent->eax |= HV_X64_APIC_ACCESS_RECOMMENDED; ent->eax |= HV_X64_RELAXED_TIMING_RECOMMENDED; - ent->eax |= HV_X64_CLUSTER_IPI_RECOMMENDED; + if (!vcpu || lapic_in_kernel(vcpu)) + ent->eax |= HV_X64_CLUSTER_IPI_RECOMMENDED; ent->eax |= HV_X64_EX_PROCESSOR_MASKS_RECOMMENDED; if (evmcs_ver) ent->eax |= HV_X64_ENLIGHTENED_VMCS_RECOMMENDED;

8 months, 4 weeks

1
0
0 0

FAILED: patch "[PATCH] drm/xe: Carve out wopcm portion from the stolen memory" failed to apply to 6.12-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.12-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. To reproduce the conflict and resubmit, you may use the following commands: git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.12.y git checkout FETCH_HEAD git cherry-pick -x e977499820782ab1c69f354d9f41b6d9ad1f43d9 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025021815-sublease-unneeded-0077@gregkh' --subject-prefix 'PATCH 6.12.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From e977499820782ab1c69f354d9f41b6d9ad1f43d9 Mon Sep 17 00:00:00 2001 From: Nirmoy Das <nirmoy.das(a)intel.com> Date: Mon, 10 Feb 2025 15:36:54 +0100 Subject: [PATCH] drm/xe: Carve out wopcm portion from the stolen memory The top of stolen memory is WOPCM, which shouldn't be accessed. Remove this portion from the stolen memory region for discrete platforms. This was already done for integrated, but was missing for discrete platforms. This also moves get_wopcm_size() so detect_bar2_dgfx() and detect_bar2_integrated can use the same function. v2: Improve commit message and suitable stable version tag(Lucas) Fixes: d8b52a02cb40 ("drm/xe: Implement stolen memory.") Cc: Maarten Lankhorst <maarten.lankhorst(a)linux.intel.com> Cc: Matthew Auld <matthew.auld(a)intel.com> Cc: Lucas De Marchi <lucas.demarchi(a)intel.com> Cc: stable(a)vger.kernel.org # v6.11+ Reviewed-by: Lucas De Marchi <lucas.demarchi(a)intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20250210143654.2076747-1-nirm… Signed-off-by: Nirmoy Das <nirmoy.das(a)intel.com> (cherry picked from commit 2c7f45cc7e197a792ce5c693e56ea48f60b312da) Signed-off-by: Rodrigo Vivi <rodrigo.vivi(a)intel.com> diff --git a/drivers/gpu/drm/xe/xe_ttm_stolen_mgr.c b/drivers/gpu/drm/xe/xe_ttm_stolen_mgr.c index 423856cc18d4..d414421f8c13 100644 --- a/drivers/gpu/drm/xe/xe_ttm_stolen_mgr.c +++ b/drivers/gpu/drm/xe/xe_ttm_stolen_mgr.c @@ -57,38 +57,6 @@ bool xe_ttm_stolen_cpu_access_needs_ggtt(struct xe_device *xe) return GRAPHICS_VERx100(xe) < 1270 && !IS_DGFX(xe); } -static s64 detect_bar2_dgfx(struct xe_device *xe, struct xe_ttm_stolen_mgr *mgr) -{ - struct xe_tile *tile = xe_device_get_root_tile(xe); - struct xe_mmio *mmio = xe_root_tile_mmio(xe); - struct pci_dev *pdev = to_pci_dev(xe->drm.dev); - u64 stolen_size; - u64 tile_offset; - u64 tile_size; - - tile_offset = tile->mem.vram.io_start - xe->mem.vram.io_start; - tile_size = tile->mem.vram.actual_physical_size; - - /* Use DSM base address instead for stolen memory */ - mgr->stolen_base = (xe_mmio_read64_2x32(mmio, DSMBASE) & BDSM_MASK) - tile_offset; - if (drm_WARN_ON(&xe->drm, tile_size < mgr->stolen_base)) - return 0; - - stolen_size = tile_size - mgr->stolen_base; - - /* Verify usage fits in the actual resource available */ - if (mgr->stolen_base + stolen_size <= pci_resource_len(pdev, LMEM_BAR)) - mgr->io_base = tile->mem.vram.io_start + mgr->stolen_base; - - /* - * There may be few KB of platform dependent reserved memory at the end - * of vram which is not part of the DSM. Such reserved memory portion is - * always less then DSM granularity so align down the stolen_size to DSM - * granularity to accommodate such reserve vram portion. - */ - return ALIGN_DOWN(stolen_size, SZ_1M); -} - static u32 get_wopcm_size(struct xe_device *xe) { u32 wopcm_size; @@ -112,6 +80,44 @@ static u32 get_wopcm_size(struct xe_device *xe) return wopcm_size; } +static s64 detect_bar2_dgfx(struct xe_device *xe, struct xe_ttm_stolen_mgr *mgr) +{ + struct xe_tile *tile = xe_device_get_root_tile(xe); + struct xe_mmio *mmio = xe_root_tile_mmio(xe); + struct pci_dev *pdev = to_pci_dev(xe->drm.dev); + u64 stolen_size, wopcm_size; + u64 tile_offset; + u64 tile_size; + + tile_offset = tile->mem.vram.io_start - xe->mem.vram.io_start; + tile_size = tile->mem.vram.actual_physical_size; + + /* Use DSM base address instead for stolen memory */ + mgr->stolen_base = (xe_mmio_read64_2x32(mmio, DSMBASE) & BDSM_MASK) - tile_offset; + if (drm_WARN_ON(&xe->drm, tile_size < mgr->stolen_base)) + return 0; + + /* Carve out the top of DSM as it contains the reserved WOPCM region */ + wopcm_size = get_wopcm_size(xe); + if (drm_WARN_ON(&xe->drm, !wopcm_size)) + return 0; + + stolen_size = tile_size - mgr->stolen_base; + stolen_size -= wopcm_size; + + /* Verify usage fits in the actual resource available */ + if (mgr->stolen_base + stolen_size <= pci_resource_len(pdev, LMEM_BAR)) + mgr->io_base = tile->mem.vram.io_start + mgr->stolen_base; + + /* + * There may be few KB of platform dependent reserved memory at the end + * of vram which is not part of the DSM. Such reserved memory portion is + * always less then DSM granularity so align down the stolen_size to DSM + * granularity to accommodate such reserve vram portion. + */ + return ALIGN_DOWN(stolen_size, SZ_1M); +} + static u32 detect_bar2_integrated(struct xe_device *xe, struct xe_ttm_stolen_mgr *mgr) { struct pci_dev *pdev = to_pci_dev(xe->drm.dev);

8 months, 4 weeks

1
0
0 0

[PATCH v2] ufs: core: bsg: Fix memory crash in case arpmb command failed

by Arthur Simchaev

In case the device doesn't support arpmb, the kernel get memory crash due to copy user data in bsg_transport_sg_io_fn level. So in case ufs_bsg_exec_advanced_rpmb_req returned error, do not set the job's reply_len. Memory crash backtrace: 3,1290,531166405,-;ufshcd 0000:00:12.5: ARPMB OP failed: error code -22 4,1308,531166555,-;Call Trace: 4,1309,531166559,-; <TASK> 4,1310,531166565,-; ? show_regs+0x6d/0x80 4,1311,531166575,-; ? die+0x37/0xa0 4,1312,531166583,-; ? do_trap+0xd4/0xf0 4,1313,531166593,-; ? do_error_trap+0x71/0xb0 4,1314,531166601,-; ? usercopy_abort+0x6c/0x80 4,1315,531166610,-; ? exc_invalid_op+0x52/0x80 4,1316,531166622,-; ? usercopy_abort+0x6c/0x80 4,1317,531166630,-; ? asm_exc_invalid_op+0x1b/0x20 4,1318,531166643,-; ? usercopy_abort+0x6c/0x80 4,1319,531166652,-; __check_heap_object+0xe3/0x120 4,1320,531166661,-; check_heap_object+0x185/0x1d0 4,1321,531166670,-; __check_object_size.part.0+0x72/0x150 4,1322,531166679,-; __check_object_size+0x23/0x30 4,1323,531166688,-; bsg_transport_sg_io_fn+0x314/0x3b0 Fixes: 6ff265fc5ef6 ("scsi: ufs: core: bsg: Add advanced RPMB support in ufs_bsg") Cc: stable(a)vger.kernel.org Signed-off-by: Arthur Simchaev <arthur.simchaev(a)sandisk.com> --- Changes in v2: - Add Fixes tag - Elaborate commit log Signed-off-by: Arthur Simchaev <arthur.simchaev(a)sandisk.com> --- drivers/ufs/core/ufs_bsg.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/drivers/ufs/core/ufs_bsg.c b/drivers/ufs/core/ufs_bsg.c index 8d4ad0a3f2cf..a8ed9bc6e4f1 100644 --- a/drivers/ufs/core/ufs_bsg.c +++ b/drivers/ufs/core/ufs_bsg.c @@ -194,10 +194,12 @@ static int ufs_bsg_request(struct bsg_job *job) ufshcd_rpm_put_sync(hba); kfree(buff); bsg_reply->result = ret; - job->reply_len = !rpmb ? sizeof(struct ufs_bsg_reply) : sizeof(struct ufs_rpmb_reply); /* complete the job here only if no error */ - if (ret == 0) + if (ret == 0) { + job->reply_len = !rpmb ? sizeof(struct ufs_bsg_reply) : + sizeof(struct ufs_rpmb_reply); bsg_job_done(job, ret, bsg_reply->reply_payload_rcv_len); + } return ret; } -- 2.34.1

8 months, 4 weeks

1
0
0 0

[PATCH v2] gpiolib: protect gpio_chip with SRCU in array_info paths in multi get/set

by Bartosz Golaszewski

From: Bartosz Golaszewski <bartosz.golaszewski(a)linaro.org> During the locking rework in GPIOLIB, we omitted one important use-case, namely: setting and getting values for GPIO descriptor arrays with array_info present. This patch does two things: first it makes struct gpio_array store the address of the underlying GPIO device and not chip. Next: it protects the chip with SRCU from removal in gpiod_get_array_value_complex() and gpiod_set_array_value_complex(). Cc: stable(a)vger.kernel.org Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski(a)linaro.org> --- Changes in v2: - fix sparse warning about differing address spaces by checking for open-drain/source flags directly in the descriptor drivers/gpio/gpiolib.c | 48 +++++++++++++++++++++++++++++------------- drivers/gpio/gpiolib.h | 4 ++-- 2 files changed, 35 insertions(+), 17 deletions(-) diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c index f261f7893f85..fddbcf8a360e 100644 --- a/drivers/gpio/gpiolib.c +++ b/drivers/gpio/gpiolib.c @@ -3128,6 +3128,8 @@ static int gpiod_get_raw_value_commit(const struct gpio_desc *desc) static int gpio_chip_get_multiple(struct gpio_chip *gc, unsigned long *mask, unsigned long *bits) { + lockdep_assert_held(&gc->gpiodev->srcu); + if (gc->get_multiple) return gc->get_multiple(gc, mask, bits); if (gc->get) { @@ -3158,6 +3160,7 @@ int gpiod_get_array_value_complex(bool raw, bool can_sleep, struct gpio_array *array_info, unsigned long *value_bitmap) { + struct gpio_chip *gc; int ret, i = 0; /* @@ -3169,10 +3172,15 @@ int gpiod_get_array_value_complex(bool raw, bool can_sleep, array_size <= array_info->size && (void *)array_info == desc_array + array_info->size) { if (!can_sleep) - WARN_ON(array_info->chip->can_sleep); + WARN_ON(array_info->gdev->can_sleep); - ret = gpio_chip_get_multiple(array_info->chip, - array_info->get_mask, + guard(srcu)(&array_info->gdev->srcu); + gc = srcu_dereference(array_info->gdev->chip, + &array_info->gdev->srcu); + if (!gc) + return -ENODEV; + + ret = gpio_chip_get_multiple(gc, array_info->get_mask, value_bitmap); if (ret) return ret; @@ -3453,6 +3461,8 @@ static void gpiod_set_raw_value_commit(struct gpio_desc *desc, bool value) static void gpio_chip_set_multiple(struct gpio_chip *gc, unsigned long *mask, unsigned long *bits) { + lockdep_assert_held(&gc->gpiodev->srcu); + if (gc->set_multiple) { gc->set_multiple(gc, mask, bits); } else { @@ -3470,6 +3480,7 @@ int gpiod_set_array_value_complex(bool raw, bool can_sleep, struct gpio_array *array_info, unsigned long *value_bitmap) { + struct gpio_chip *gc; int i = 0; /* @@ -3481,14 +3492,19 @@ int gpiod_set_array_value_complex(bool raw, bool can_sleep, array_size <= array_info->size && (void *)array_info == desc_array + array_info->size) { if (!can_sleep) - WARN_ON(array_info->chip->can_sleep); + WARN_ON(array_info->gdev->can_sleep); + + guard(srcu)(&array_info->gdev->srcu); + gc = srcu_dereference(array_info->gdev->chip, + &array_info->gdev->srcu); + if (!gc) + return -ENODEV; if (!raw && !bitmap_empty(array_info->invert_mask, array_size)) bitmap_xor(value_bitmap, value_bitmap, array_info->invert_mask, array_size); - gpio_chip_set_multiple(array_info->chip, array_info->set_mask, - value_bitmap); + gpio_chip_set_multiple(gc, array_info->set_mask, value_bitmap); i = find_first_zero_bit(array_info->set_mask, array_size); if (i == array_size) @@ -4750,9 +4766,10 @@ struct gpio_descs *__must_check gpiod_get_array(struct device *dev, { struct gpio_desc *desc; struct gpio_descs *descs; + struct gpio_device *gdev; struct gpio_array *array_info = NULL; - struct gpio_chip *gc; int count, bitmap_size; + unsigned long dflags; size_t descs_size; count = gpiod_count(dev, con_id); @@ -4773,7 +4790,7 @@ struct gpio_descs *__must_check gpiod_get_array(struct device *dev, descs->desc[descs->ndescs] = desc; - gc = gpiod_to_chip(desc); + gdev = gpiod_to_gpio_device(desc); /* * If pin hardware number of array member 0 is also 0, select * its chip as a candidate for fast bitmap processing path. @@ -4781,8 +4798,8 @@ struct gpio_descs *__must_check gpiod_get_array(struct device *dev, if (descs->ndescs == 0 && gpio_chip_hwgpio(desc) == 0) { struct gpio_descs *array; - bitmap_size = BITS_TO_LONGS(gc->ngpio > count ? - gc->ngpio : count); + bitmap_size = BITS_TO_LONGS(gdev->ngpio > count ? + gdev->ngpio : count); array = krealloc(descs, descs_size + struct_size(array_info, invert_mask, 3 * bitmap_size), @@ -4802,7 +4819,7 @@ struct gpio_descs *__must_check gpiod_get_array(struct device *dev, array_info->desc = descs->desc; array_info->size = count; - array_info->chip = gc; + array_info->gdev = gdev; bitmap_set(array_info->get_mask, descs->ndescs, count - descs->ndescs); bitmap_set(array_info->set_mask, descs->ndescs, @@ -4815,7 +4832,7 @@ struct gpio_descs *__must_check gpiod_get_array(struct device *dev, continue; /* Unmark array members which don't belong to the 'fast' chip */ - if (array_info->chip != gc) { + if (array_info->gdev != gdev) { __clear_bit(descs->ndescs, array_info->get_mask); __clear_bit(descs->ndescs, array_info->set_mask); } @@ -4838,9 +4855,10 @@ struct gpio_descs *__must_check gpiod_get_array(struct device *dev, array_info->set_mask); } } else { + dflags = READ_ONCE(desc->flags); /* Exclude open drain or open source from fast output */ - if (gpiochip_line_is_open_drain(gc, descs->ndescs) || - gpiochip_line_is_open_source(gc, descs->ndescs)) + if (test_bit(FLAG_OPEN_DRAIN, &dflags) || + test_bit(FLAG_OPEN_SOURCE, &dflags)) __clear_bit(descs->ndescs, array_info->set_mask); /* Identify 'fast' pins which require invertion */ @@ -4852,7 +4870,7 @@ struct gpio_descs *__must_check gpiod_get_array(struct device *dev, if (array_info) dev_dbg(dev, "GPIO array info: chip=%s, size=%d, get_mask=%lx, set_mask=%lx, invert_mask=%lx\n", - array_info->chip->label, array_info->size, + array_info->gdev->label, array_info->size, *array_info->get_mask, *array_info->set_mask, *array_info->invert_mask); return descs; diff --git a/drivers/gpio/gpiolib.h b/drivers/gpio/gpiolib.h index 83690f72f7e5..147156ec502b 100644 --- a/drivers/gpio/gpiolib.h +++ b/drivers/gpio/gpiolib.h @@ -114,7 +114,7 @@ extern const char *const gpio_suffixes[]; * * @desc: Array of pointers to the GPIO descriptors * @size: Number of elements in desc - * @chip: Parent GPIO chip + * @gdev: Parent GPIO device * @get_mask: Get mask used in fastpath * @set_mask: Set mask used in fastpath * @invert_mask: Invert mask used in fastpath @@ -126,7 +126,7 @@ extern const char *const gpio_suffixes[]; struct gpio_array { struct gpio_desc **desc; unsigned int size; - struct gpio_chip *chip; + struct gpio_device *gdev; unsigned long *get_mask; unsigned long *set_mask; unsigned long invert_mask[]; -- 2.45.2

8 months, 4 weeks

1
1
0 0

[PATCH net 0/8] net: enetc: fix some known issues

by Wei Fang

There are some issues with the enetc driver, some of which are specific to the LS1028A platform, and some of which were introduced recently when i.MX95 ENETC support was added, so this patch set aims to clean up those issues. Wei Fang (8): net: enetc: fix the off-by-one issue in enetc_map_tx_buffs() net: enetc: correct the tx_swbd statistics net: enetc: correct the xdp_tx statistics net: enetc: VFs do not support HWTSTAMP_TX_ONESTEP_SYNC net: enetc: update UDP checksum when updating originTimestamp field net: enetc: add missing enetc4_link_deinit() net: enetc: remove the mm_lock from the ENETC v4 driver net: enetc: correct the EMDIO base offset for ENETC v4 drivers/net/ethernet/freescale/enetc/enetc.c | 53 +++++++++++++++---- .../net/ethernet/freescale/enetc/enetc4_hw.h | 3 ++ .../net/ethernet/freescale/enetc/enetc4_pf.c | 2 +- .../ethernet/freescale/enetc/enetc_ethtool.c | 7 ++- .../freescale/enetc/enetc_pf_common.c | 10 +++- 5 files changed, 60 insertions(+), 15 deletions(-) -- 2.34.1

8 months, 4 weeks

4
20
0 0

[PATCH] mm/hugetlb: wait for hugepage folios to be freed

by yangge1116＠126.com

From: Ge Yang <yangge1116(a)126.com> Since the introduction of commit b65d4adbc0f0 ("mm: hugetlb: defer freeing of HugeTLB pages"), which supports deferring the freeing of HugeTLB pages, the allocation of contiguous memory through cma_alloc() may fail probabilistically. In the CMA allocation process, if it is found that the CMA area is occupied by in-use hugepage folios, these in-use hugepage folios need to be migrated to another location. When there are no available hugepage folios in the free HugeTLB pool during the migration of in-use HugeTLB pages, new folios are allocated from the buddy system. A temporary state is set on the newly allocated folio. Upon completion of the hugepage folio migration, the temporary state is transferred from the new folios to the old folios. Normally, when the old folios with the temporary state are freed, it is directly released back to the buddy system. However, due to the deferred freeing of HugeTLB pages, the PageBuddy() check fails, ultimately leading to the failure of cma_alloc(). Here is a simplified call trace illustrating the process: cma_alloc() ->__alloc_contig_migrate_range() // Migrate in-use hugepage ->unmap_and_move_huge_page() ->folio_putback_hugetlb() // Free old folios ->test_pages_isolated() ->__test_page_isolated_in_pageblock() ->PageBuddy(page) // Check if the page is in buddy To resolve this issue, we have implemented a function named wait_for_hugepage_folios_freed(). This function ensures that the hugepage folios are properly released back to the buddy system after their migration is completed. By invoking wait_for_hugepage_folios_freed() following the migration process, we guarantee that when test_pages_isolated() is executed, it will successfully pass. Fixes: b65d4adbc0f0 ("mm: hugetlb: defer freeing of HugeTLB pages") Signed-off-by: Ge Yang <yangge1116(a)126.com> --- include/linux/hugetlb.h | 5 +++++ mm/hugetlb.c | 7 +++++++ mm/migrate.c | 16 ++++++++++++++-- 3 files changed, 26 insertions(+), 2 deletions(-) diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h index 6c6546b..c39e0d5 100644 --- a/include/linux/hugetlb.h +++ b/include/linux/hugetlb.h @@ -697,6 +697,7 @@ bool hugetlb_bootmem_page_zones_valid(int nid, struct huge_bootmem_page *m); int isolate_or_dissolve_huge_page(struct page *page, struct list_head *list); int replace_free_hugepage_folios(unsigned long start_pfn, unsigned long end_pfn); +void wait_for_hugepage_folios_freed(struct hstate *h); struct folio *alloc_hugetlb_folio(struct vm_area_struct *vma, unsigned long addr, bool cow_from_owner); struct folio *alloc_hugetlb_folio_nodemask(struct hstate *h, int preferred_nid, @@ -1092,6 +1093,10 @@ static inline int replace_free_hugepage_folios(unsigned long start_pfn, return 0; } +static inline void wait_for_hugepage_folios_freed(struct hstate *h) +{ +} + static inline struct folio *alloc_hugetlb_folio(struct vm_area_struct *vma, unsigned long addr, bool cow_from_owner) diff --git a/mm/hugetlb.c b/mm/hugetlb.c index 30bc34d..64cae39 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -2955,6 +2955,13 @@ int replace_free_hugepage_folios(unsigned long start_pfn, unsigned long end_pfn) return ret; } +void wait_for_hugepage_folios_freed(struct hstate *h) +{ + WARN_ON(!h); + + flush_free_hpage_work(h); +} + typedef enum { /* * For either 0/1: we checked the per-vma resv map, and one resv diff --git a/mm/migrate.c b/mm/migrate.c index fb19a18..5dd1851 100644 --- a/mm/migrate.c +++ b/mm/migrate.c @@ -1448,6 +1448,7 @@ static int unmap_and_move_huge_page(new_folio_t get_new_folio, int page_was_mapped = 0; struct anon_vma *anon_vma = NULL; struct address_space *mapping = NULL; + unsigned long size; if (folio_ref_count(src) == 1) { /* page was freed from under us. So we are done. */ @@ -1533,9 +1534,20 @@ static int unmap_and_move_huge_page(new_folio_t get_new_folio, out_unlock: folio_unlock(src); out: - if (rc == MIGRATEPAGE_SUCCESS) + if (rc == MIGRATEPAGE_SUCCESS) { + size = folio_size(src); folio_putback_hugetlb(src); - else if (rc != -EAGAIN) + + /* + * Due to the deferred freeing of HugeTLB folios, the hugepage 'src' may + * not immediately release to the buddy system. This can lead to failure + * in allocating memory through the cma_alloc() function. To ensure that + * the hugepage folios are properly released back to the buddy system, + * we invoke the wait_for_hugepage_folios_freed() function to wait for + * the release to complete. + */ + wait_for_hugepage_folios_freed(size_to_hstate(size)); + } else if (rc != -EAGAIN) list_move_tail(&src->lru, ret); /* -- 2.7.4

8 months, 4 weeks

4
7
0 0

[PATCH v1] ARM: dts: imx6qdl-apalis: Fix poweroff on Apalis iMX6

by Stefan Eichenberger

From: Stefan Eichenberger <stefan.eichenberger(a)toradex.com> The current solution for powering off the Apalis iMX6 is not functioning as intended. To resolve this, it is necessary to power off the vgen2_reg, which will also set the POWER_ENABLE_MOCI signal to a low state. This ensures the carrier board is properly informed to initiate its power-off sequence. The new solution uses the regulator-poweroff driver, which will power off the regulator during a system shutdown. CC: stable(a)vger.kernel.org Fixes: 4eb56e26f92e ("ARM: dts: imx6q-apalis: Command pmic to standby for poweroff") Signed-off-by: Stefan Eichenberger <stefan.eichenberger(a)toradex.com> --- arch/arm/boot/dts/nxp/imx/imx6qdl-apalis.dtsi | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/arch/arm/boot/dts/nxp/imx/imx6qdl-apalis.dtsi b/arch/arm/boot/dts/nxp/imx/imx6qdl-apalis.dtsi index 1c72da417011..614b65821995 100644 --- a/arch/arm/boot/dts/nxp/imx/imx6qdl-apalis.dtsi +++ b/arch/arm/boot/dts/nxp/imx/imx6qdl-apalis.dtsi @@ -108,6 +108,11 @@ lvds_panel_in: endpoint { }; }; + poweroff { + compatible = "regulator-poweroff"; + cpu-supply = <&vgen2_reg>; + }; + reg_module_3v3: regulator-module-3v3 { compatible = "regulator-fixed"; regulator-always-on; @@ -236,10 +241,6 @@ &can2 { status = "disabled"; }; -&clks { - fsl,pmic-stby-poweroff; -}; - /* Apalis SPI1 */ &ecspi1 { cs-gpios = <&gpio5 25 GPIO_ACTIVE_LOW>; @@ -527,7 +528,6 @@ &i2c2 { pmic: pmic@8 { compatible = "fsl,pfuze100"; - fsl,pmic-stby-poweroff; reg = <0x08>; regulators { -- 2.45.2

8 months, 4 weeks

2
1
0 0

[PATCH v2] media: verisilicon: Fix AV1 decoder clock frequency

by Nicolas Dufresne

The desired clock frequency was correctly set to 400MHz in the device tree but was lowered by the driver to 300MHz breaking 4K 60Hz content playback. Fix the issue by removing the driver call to clk_set_rate(), which reduce the amount of board specific code. Fixes: 003afda97c65 ("media: verisilicon: Enable AV1 decoder on rk3588") Cc: stable(a)vger.kernel.org Signed-off-by: Nicolas Dufresne <nicolas.dufresne(a)collabora.com> --- This patch fixes user report of AV1 4K60 decoder not being fast enough on RK3588 based SoC. This is a break from Hantro original authors habbit of coding the frequencies in the driver instead of specifying this frequency in the device tree. The other calls to clk_set_rate() are left since this would require modifying many dtsi files, which would then be unsuitable for backport. --- Changes in v2: - Completely remove the unused init function, the driver is null-safe - Link to v1: https://lore.kernel.org/r/20250217-b4-hantro-av1-clock-rate-v1-1-65b91132c5… --- drivers/media/platform/verisilicon/rockchip_vpu_hw.c | 9 --------- 1 file changed, 9 deletions(-) diff --git a/drivers/media/platform/verisilicon/rockchip_vpu_hw.c b/drivers/media/platform/verisilicon/rockchip_vpu_hw.c index 964122e7c355934cd80eb442219f6ba51bba8b71..842040f713c15e6ff295771bc9fa5a7b66e584b2 100644 --- a/drivers/media/platform/verisilicon/rockchip_vpu_hw.c +++ b/drivers/media/platform/verisilicon/rockchip_vpu_hw.c @@ -17,7 +17,6 @@ #define RK3066_ACLK_MAX_FREQ (300 * 1000 * 1000) #define RK3288_ACLK_MAX_FREQ (400 * 1000 * 1000) -#define RK3588_ACLK_MAX_FREQ (300 * 1000 * 1000) #define ROCKCHIP_VPU981_MIN_SIZE 64 @@ -440,13 +439,6 @@ static int rk3066_vpu_hw_init(struct hantro_dev *vpu) return 0; } -static int rk3588_vpu981_hw_init(struct hantro_dev *vpu) -{ - /* Bump ACLKs to max. possible freq. to improve performance. */ - clk_set_rate(vpu->clocks[0].clk, RK3588_ACLK_MAX_FREQ); - return 0; -} - static int rockchip_vpu_hw_init(struct hantro_dev *vpu) { /* Bump ACLK to max. possible freq. to improve performance. */ @@ -807,7 +799,6 @@ const struct hantro_variant rk3588_vpu981_variant = { .codec_ops = rk3588_vpu981_codec_ops, .irqs = rk3588_vpu981_irqs, .num_irqs = ARRAY_SIZE(rk3588_vpu981_irqs), - .init = rk3588_vpu981_hw_init, .clk_names = rk3588_vpu981_vpu_clk_names, .num_clocks = ARRAY_SIZE(rk3588_vpu981_vpu_clk_names) }; --- base-commit: 2014c95afecee3e76ca4a56956a936e23283f05b change-id: 20250217-b4-hantro-av1-clock-rate-e5497f1499df Best regards, -- Nicolas Dufresne <nicolas.dufresne(a)collabora.com>

8 months, 4 weeks

3
2
0 0

regression on 6.1.129-rc with perf

by Max Krummenacher

Our CI built linux-stable-rc.git queue/6.1 at commit eef4a8a45ba1 ("btrfs: output the reason for open_ctree() failure"). (built for arm and arm64, albeit I don't think it matters.) Building perf produced 2 build errors which I wanted to report even before the RC1 is out. | ...tools/perf/util/namespaces.c:247:27: error: invalid type argument of '->' (have 'int') | 247 | RC_CHK_ACCESS(nsi)->in_pidns = true; | | ^~ introduced by commit 93520bacf784 ("perf namespaces: Introduce nsinfo__set_in_pidns()"). The RC_CK_ACCSS macro was introduced in 6.4. Removing the macro made this go away ( nsi->in_pidns = true; ). Second perf build error: | ld: ...tools/perf/util/machine.c:1176: undefined reference to `kallsyms__get_symbol_start' Introduced by commits: 710c2e913aa9 perf machine: Don't ignore _etext when not a text symbol 69a87a32f5cd perf machine: Include data symbols in the kernel map These two use the function kallsyms__get_symbol_start added with: f9dd531c5b82 perf symbols: Add kallsyms__get_symbol_start() So f9dd531c5b82 would additionally be needed. The kernel itself built fine, due to the perf error we don't have runtime testresults. Best regards Max

8 months, 4 weeks

2
1
0 0

Linux 6.12.15

by Greg Kroah-Hartman

I'm announcing the release of the 6.12.15 kernel. This is ONLY needed for users of the XFS filesystem. If you do not use XFS, no need to upgrade. If you do use XFS, please upgrade. The updated 6.12.y git tree can be found at: git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git linux-6.12.y and can be browsed at the normal kernel.org git web browser: https://git.kernel.org/?p=linux/kernel/git/stable/linux-stable.git;a=summary thanks, greg k-h ------------ Makefile | 2 +- fs/xfs/xfs_quota.h | 5 +++-- fs/xfs/xfs_trans.c | 10 +++------- fs/xfs/xfs_trans_dquot.c | 31 ++++++++++++++++++++++++++----- 4 files changed, 33 insertions(+), 15 deletions(-) Darrick J. Wong (1): xfs: don't lose solo dquot update transactions Greg Kroah-Hartman (1): Linux 6.12.15

8 months, 4 weeks

1
1
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror