April 2024 - Kernel - mailweb.openeuler.org

[PATCH OLK-5.10 4/4] gicv3: add lpi support for cvm guest
by hejingxian 08 Apr '24

08 Apr '24

From: Jingxian He hejingxian(a)huawei.com<mailto:hejingxian@huawei.com> hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I9CC0X CVE: N/A ------------ Add lpi support for cvm guest os: The gic-its driver mapped memory must shared with qemu/kvm. The cvm guest gic-its driver alloc memory from bounce buffer to share with qemu/kvm. Signed-off-by: Jingxian He <hejingxian(a)huawei.com> Signed-off-by: wuweinan <wuweinan(a)huawei.com> --- drivers/irqchip/irq-gic-v3-its.c | 228 ++++++++++++++++++++++++++++--- 1 file changed, 207 insertions(+), 21 deletions(-) diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c index 632444f86..a09cad8a0 100644 --- a/drivers/irqchip/irq-gic-v3-its.c +++ b/drivers/irqchip/irq-gic-v3-its.c @@ -29,6 +29,10 @@ #include <linux/percpu.h> #include <linux/slab.h> #include <linux/syscore_ops.h> +#ifdef CONFIG_CVM_GUEST +#include <linux/swiotlb.h> +#include <asm/cvm_guest.h> +#endif #include <linux/irqchip.h> #include <linux/irqchip/arm-gic-v3.h> @@ -182,6 +186,90 @@ struct its_baser { struct its_device; +#ifdef CONFIG_CVM_GUEST +static struct device cvm_alloc_device; +static LIST_HEAD(cvm_its_nodes); +static raw_spinlock_t cvm_its_lock; + +struct its_device_order { + struct its_device *dev; + struct list_head entry; + int itt_order; +}; + +static inline struct page *its_alloc_shared_pages_node(int node, gfp_t gfp, + unsigned int order) +{ + return swiotlb_alloc(&cvm_alloc_device, (1 << order) * PAGE_SIZE); +} + +static inline struct page *its_alloc_shared_pages(gfp_t gfp, unsigned int order) +{ + return its_alloc_shared_pages_node(NUMA_NO_NODE, gfp, order); +} + +static void its_free_shared_pages(void *addr, int order) +{ + if (order < 0) + return; + + swiotlb_free(&cvm_alloc_device, (struct page*)addr, (1 << order) * PAGE_SIZE); +} + +static int add_its_device_order(struct its_device *dev, int itt_order) +{ + struct its_device_order *new; + unsigned long flags; + + new = kmalloc(sizeof(struct its_device_order), GFP_KERNEL); + if (!new) + return -ENOMEM; + new->dev = dev; + new->itt_order = itt_order; + raw_spin_lock_irqsave(&cvm_its_lock, flags); + list_add_tail(&new->entry, &cvm_its_nodes); + raw_spin_unlock_irqrestore(&cvm_its_lock, flags); + return 0; +} + +/* get its device order and then free its device order */ +static int get_its_device_order(struct its_device *dev) +{ + struct its_device_order *pos, *tmp; + unsigned long flags; + int itt_order = -1; + + raw_spin_lock_irqsave(&cvm_its_lock, flags); + list_for_each_entry_safe(pos, tmp, &cvm_its_nodes, entry) { + if (pos->dev == dev) { + itt_order = pos->itt_order; + list_del(&pos->entry); + kfree(pos); + goto found; + } + } +found: + raw_spin_unlock_irqrestore(&cvm_its_lock, flags); + return itt_order; +} + +static void *its_alloc_shared_page_address(struct its_device *dev, struct its_node *its, int sz) +{ + struct page *page; + int itt_order; + + itt_order = get_order(sz); + if (add_its_device_order(dev, itt_order)) + return NULL; + + page = its_alloc_shared_pages_node(its->numa_node, GFP_KERNEL | __GFP_ZERO, + itt_order); + if (!page) + return NULL; + return (void *)page_address(page); +} +#endif + /* * The ITS structure - contains most of the infrastructure, with the * top-level MSI domain, the command queue, the collections, and the @@ -2447,7 +2535,13 @@ static struct page *its_allocate_prop_table(gfp_t gfp_flags) { struct page *prop_page; - prop_page = alloc_pages(gfp_flags, get_order(LPI_PROPBASE_SZ)); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + prop_page = its_alloc_shared_pages(gfp_flags, + get_order(LPI_PROPBASE_SZ)); + else +#endif + prop_page = alloc_pages(gfp_flags, get_order(LPI_PROPBASE_SZ)); if (!prop_page) return NULL; @@ -2458,8 +2552,14 @@ static struct page *its_allocate_prop_table(gfp_t gfp_flags) static void its_free_prop_table(struct page *prop_page) { - free_pages((unsigned long)page_address(prop_page), - get_order(LPI_PROPBASE_SZ)); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + its_free_shared_pages(page_address(prop_page), + get_order(LPI_PROPBASE_SZ)); + else +#endif + free_pages((unsigned long)page_address(prop_page), + get_order(LPI_PROPBASE_SZ)); } static bool gic_check_reserved_range(phys_addr_t addr, unsigned long size) @@ -2581,7 +2681,13 @@ static int its_setup_baser(struct its_node *its, struct its_baser *baser, order = get_order(GITS_BASER_PAGES_MAX * psz); } - page = alloc_pages_node(its->numa_node, GFP_KERNEL | __GFP_ZERO, order); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + page = its_alloc_shared_pages_node(its->numa_node, + GFP_KERNEL | __GFP_ZERO, order); + else +#endif + page = alloc_pages_node(its->numa_node, GFP_KERNEL | __GFP_ZERO, order); if (!page) return -ENOMEM; @@ -2594,7 +2700,12 @@ static int its_setup_baser(struct its_node *its, struct its_baser *baser, /* 52bit PA is supported only when PageSize=64K */ if (psz != SZ_64K) { pr_err("ITS: no 52bit PA support when psz=%d\n", psz); - free_pages((unsigned long)base, order); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + its_free_shared_pages(base, order); + else +#endif + free_pages((unsigned long)base, order); return -ENXIO; } @@ -2648,7 +2759,12 @@ static int its_setup_baser(struct its_node *its, struct its_baser *baser, pr_err("ITS@%pa: %s doesn't stick: %llx %llx\n", &its->phys_base, its_base_type_string[type], val, tmp); - free_pages((unsigned long)base, order); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + its_free_shared_pages(base, order); + else +#endif + free_pages((unsigned long)base, order); return -ENXIO; } @@ -2787,8 +2903,14 @@ static void its_free_tables(struct its_node *its) for (i = 0; i < GITS_BASER_NR_REGS; i++) { if (its->tables[i].base) { - free_pages((unsigned long)its->tables[i].base, - its->tables[i].order); +#ifdef CONFIG_CVM_GUEST + if (!is_cvm_world()) + its_free_shared_pages(its->tables[i].base, + its->tables[i].order); + else +#endif + free_pages((unsigned long)its->tables[i].base, + its->tables[i].order); its->tables[i].base = NULL; } } @@ -3051,7 +3173,13 @@ static bool allocate_vpe_l2_table(int cpu, u32 id) /* Allocate memory for 2nd level table */ if (!table[idx]) { - page = alloc_pages(GFP_KERNEL | __GFP_ZERO, get_order(psz)); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + page = its_alloc_shared_pages(GFP_KERNEL | __GFP_ZERO, + get_order(psz)); + else +#endif + page = alloc_pages(GFP_KERNEL | __GFP_ZERO, get_order(psz)); if (!page) return false; @@ -3170,7 +3298,13 @@ static int allocate_vpe_l1_table(void) pr_debug("np = %d, npg = %lld, psz = %d, epp = %d, esz = %d\n", np, npg, psz, epp, esz); - page = alloc_pages(GFP_ATOMIC | __GFP_ZERO, get_order(np * PAGE_SIZE)); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + page = its_alloc_shared_pages(GFP_ATOMIC | __GFP_ZERO, + get_order(np * PAGE_SIZE)); + else +#endif + page = alloc_pages(GFP_ATOMIC | __GFP_ZERO, get_order(np * PAGE_SIZE)); if (!page) return -ENOMEM; @@ -3218,8 +3352,14 @@ static struct page *its_allocate_pending_table(gfp_t gfp_flags) { struct page *pend_page; - pend_page = alloc_pages(gfp_flags | __GFP_ZERO, - get_order(LPI_PENDBASE_SZ)); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + pend_page = its_alloc_shared_pages(gfp_flags | __GFP_ZERO, + get_order(LPI_PENDBASE_SZ)); + else +#endif + pend_page = alloc_pages(gfp_flags | __GFP_ZERO, + get_order(LPI_PENDBASE_SZ)); if (!pend_page) return NULL; @@ -3231,7 +3371,13 @@ static struct page *its_allocate_pending_table(gfp_t gfp_flags) static void its_free_pending_table(struct page *pt) { - free_pages((unsigned long)page_address(pt), get_order(LPI_PENDBASE_SZ)); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + its_free_shared_pages(page_address(pt), + get_order(LPI_PENDBASE_SZ)); + else +#endif + free_pages((unsigned long)page_address(pt), get_order(LPI_PENDBASE_SZ)); } /* @@ -3768,8 +3914,15 @@ static bool its_alloc_table_entry(struct its_node *its, /* Allocate memory for 2nd level table */ if (!table[idx]) { - page = alloc_pages_node(its->numa_node, GFP_KERNEL | __GFP_ZERO, - get_order(baser->psz)); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + page = its_alloc_shared_pages_node(its->numa_node, + GFP_KERNEL | __GFP_ZERO, + get_order(baser->psz)); + else +#endif + page = alloc_pages_node(its->numa_node, GFP_KERNEL | __GFP_ZERO, + get_order(baser->psz)); if (!page) return false; @@ -3872,7 +4025,12 @@ static struct its_device *its_create_device(struct its_node *its, u32 dev_id, nr_ites = max(2, nvecs); sz = nr_ites * (FIELD_GET(GITS_TYPER_ITT_ENTRY_SIZE, its->typer) + 1); sz = max(sz, ITS_ITT_ALIGN) + ITS_ITT_ALIGN - 1; - itt = kzalloc_node(sz, GFP_KERNEL, its->numa_node); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + itt = its_alloc_shared_page_address(dev, its, sz); + else +#endif + itt = kzalloc_node(sz, GFP_KERNEL, its->numa_node); if (alloc_lpis) { lpi_map = its_lpi_alloc(nvecs, &lpi_base, &nr_lpis); if (lpi_map) @@ -3886,7 +4044,12 @@ static struct its_device *its_create_device(struct its_node *its, u32 dev_id, if (!dev || !itt || !col_map || (!lpi_map && alloc_lpis)) { kfree(dev); - kfree(itt); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + its_free_shared_pages(itt, get_order(sz)); + else +#endif + kfree(itt); kfree(lpi_map); kfree(col_map); return NULL; @@ -3923,7 +4086,12 @@ static void its_free_device(struct its_device *its_dev) list_del(&its_dev->entry); raw_spin_unlock_irqrestore(&its_dev->its->lock, flags); kfree(its_dev->event_map.col_map); - kfree(its_dev->itt); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + its_free_shared_pages(its_dev->itt, get_its_device_order(its_dev)); + else +#endif + kfree(its_dev->itt); if (its_dev->is_vdev) { WARN_ON(!rsv_devid_pool_cap); @@ -5594,8 +5762,15 @@ static int __init its_probe_one(struct resource *res, its->numa_node = numa_node; - page = alloc_pages_node(its->numa_node, GFP_KERNEL | __GFP_ZERO, - get_order(ITS_CMD_QUEUE_SZ)); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + page = its_alloc_shared_pages_node(its->numa_node, + GFP_KERNEL | __GFP_ZERO, + get_order(ITS_CMD_QUEUE_SZ)); + else +#endif + page = alloc_pages_node(its->numa_node, GFP_KERNEL | __GFP_ZERO, + get_order(ITS_CMD_QUEUE_SZ)); if (!page) { err = -ENOMEM; goto out_unmap_sgir; @@ -5661,7 +5836,12 @@ static int __init its_probe_one(struct resource *res, out_free_tables: its_free_tables(its); out_free_cmd: - free_pages((unsigned long)its->cmd_base, get_order(ITS_CMD_QUEUE_SZ)); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + its_free_shared_pages(its->cmd_base, get_order(ITS_CMD_QUEUE_SZ)); + else +#endif + free_pages((unsigned long)its->cmd_base, get_order(ITS_CMD_QUEUE_SZ)); out_unmap_sgir: if (its->sgir_base) iounmap(its->sgir_base); @@ -5957,6 +6137,12 @@ int __init its_init(struct fwnode_handle *handle, struct rdists *rdists, bool has_vtimer_irqbypass = false; int err; +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) { + device_initialize(&cvm_alloc_device); + raw_spin_lock_init(&cvm_its_lock); + } +#endif gic_rdists = rdists; its_parent = parent_domain; -- 2.33.0

1 0

[PATCH OLK-5.10 3/4] dma: add bounce buffer feature for cvm guest
by hejingxian 08 Apr '24

08 Apr '24

From: Jingxian He <hejingxian(a)huawei.com> hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I9CC0X CVE: N/A ------------ Add bounce buffer feature for cvm guest os: 1) Cvm guest mapped memory is secure memory. 2) Qemu/kvm cannot access the secure memory. 3) Use bounce buffer as memory shared by cvm guest and qemu/kvm. Signed-off-by: Jingxian He <hejingxian(a)huawei.com> --- arch/arm64/configs/defconfig | 1 + arch/arm64/configs/openeuler_defconfig | 1 + arch/arm64/include/asm/cvm_guest.h | 21 ++++++ arch/arm64/kvm/Kconfig | 8 +++ arch/arm64/kvm/Makefile | 1 + arch/arm64/kvm/cvm_guest.c | 90 ++++++++++++++++++++++++++ arch/arm64/mm/mmu.c | 11 ++++ arch/arm64/mm/pageattr.c | 9 ++- include/linux/swiotlb.h | 13 ++++ kernel/dma/direct.c | 39 +++++++++++ kernel/dma/swiotlb.c | 86 +++++++++++++++++++++++- 11 files changed, 278 insertions(+), 2 deletions(-) create mode 100644 arch/arm64/include/asm/cvm_guest.h create mode 100644 arch/arm64/kvm/cvm_guest.c diff --git a/arch/arm64/configs/defconfig b/arch/arm64/configs/defconfig index ace2bf4ad..0ba4538d9 100644 --- a/arch/arm64/configs/defconfig +++ b/arch/arm64/configs/defconfig @@ -111,6 +111,7 @@ CONFIG_ACPI_APEI_EINJ=y CONFIG_VIRTUALIZATION=y CONFIG_KVM=y CONFIG_CVM_HOST=y +CONFIG_CVM_GUEST=y CONFIG_ARM64_CRYPTO=y CONFIG_CRYPTO_SHA1_ARM64_CE=y CONFIG_CRYPTO_SHA2_ARM64_CE=y diff --git a/arch/arm64/configs/openeuler_defconfig b/arch/arm64/configs/openeuler_defconfig index e298ca7e5..25a5fa5c7 100644 --- a/arch/arm64/configs/openeuler_defconfig +++ b/arch/arm64/configs/openeuler_defconfig @@ -763,6 +763,7 @@ CONFIG_IRQ_BYPASS_MANAGER=y CONFIG_VIRTUALIZATION=y CONFIG_KVM=y CONFIG_CVM_HOST=y +CONFIG_CVM_GUEST=y CONFIG_HAVE_KVM_IRQCHIP=y CONFIG_HAVE_KVM_IRQFD=y CONFIG_HAVE_KVM_IRQ_ROUTING=y diff --git a/arch/arm64/include/asm/cvm_guest.h b/arch/arm64/include/asm/cvm_guest.h new file mode 100644 index 000000000..3c5bda7ca --- /dev/null +++ b/arch/arm64/include/asm/cvm_guest.h @@ -0,0 +1,21 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Copyright (c) 2024, The Linux Foundation. All rights reserved. + */ +#ifndef __ASM_CVM_GUEST_H +#define __ASM_CVM_GUEST_H + +#ifdef CONFIG_CVM_GUEST +static inline bool cvm_mem_encrypt_active(void) +{ + return false; +} + +int set_cvm_memory_encrypted(unsigned long addr, int numpages); + +int set_cvm_memory_decrypted(unsigned long addr, int numpages); + +bool is_cvm_world(void); + +#endif +#endif diff --git a/arch/arm64/kvm/Kconfig b/arch/arm64/kvm/Kconfig index 7c24a4d33..d21e27f74 100644 --- a/arch/arm64/kvm/Kconfig +++ b/arch/arm64/kvm/Kconfig @@ -54,6 +54,14 @@ config CVM_HOST If unsure, say N. +config CVM_GUEST + bool "CVM guest enable" + depends on KVM && SWIOTLB && ARM64 + help + Support CVM guest based on S-EL2 + + If unsure, say N. + if KVM source "virt/kvm/Kconfig" diff --git a/arch/arm64/kvm/Makefile b/arch/arm64/kvm/Makefile index 3b92eaa4f..61dce3ab4 100644 --- a/arch/arm64/kvm/Makefile +++ b/arch/arm64/kvm/Makefile @@ -30,3 +30,4 @@ kvm-$(CONFIG_CVM_HOST) += cvm_exit.o kvm-$(CONFIG_KVM_ARM_PMU) += pmu-emul.o obj-$(CONFIG_KVM_HISI_VIRT) += hisilicon/ +obj-$(CONFIG_CVM_GUEST) += cvm_guest.o diff --git a/arch/arm64/kvm/cvm_guest.c b/arch/arm64/kvm/cvm_guest.c new file mode 100644 index 000000000..9df24af46 --- /dev/null +++ b/arch/arm64/kvm/cvm_guest.c @@ -0,0 +1,90 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Copyright (c) 2024, The Linux Foundation. All rights reserved. + */ +#include <linux/kernel.h> +#include <linux/mm.h> +#include <linux/module.h> +#include <linux/sched.h> +#include <linux/vmalloc.h> + +#include <asm/cacheflush.h> +#include <asm/set_memory.h> +#include <asm/tlbflush.h> + +static bool cvm_guest_enable __read_mostly; +static int __init setup_cvm_guest(char *str) +{ + int ret; + unsigned int val; + + if (!str) + return 0; + + cvm_guest_enable = false; + ret = kstrtouint(str, 10, &val); + if (ret) { + pr_warn("Unable to parse cvm_guest.\n"); + } else { + if (val) + cvm_guest_enable = true; + } + return ret; +} +early_param("cvm_guest", setup_cvm_guest); + +bool is_cvm_world(void) +{ + return cvm_guest_enable; +} + +static int change_page_range_cvm(pte_t *ptep, unsigned long addr, void *data) +{ + bool encrypt = (bool)data; + pte_t pte = READ_ONCE(*ptep); + if (encrypt) { + if(!(pte.pte & 0x20)) { + return 0; + } + pte.pte = pte.pte & (~0x20); + } else { + if (pte.pte & 0x20) { + return 0; + } + /* Set NS BIT */ + pte.pte = pte.pte | 0x20; + } + set_pte(ptep, pte); + + return 0; +} + +static int __change_memory_common_cvm(unsigned long start, unsigned long size, bool encrypt) +{ + int ret; + ret = apply_to_page_range(&init_mm, start, size, change_page_range_cvm, (void *)encrypt); + flush_tlb_kernel_range(start, start + size); + return ret; +} + +static int __set_memory_encrypted(unsigned long addr, + int numpages, + bool encrypt) +{ + if (!is_cvm_world()) { + return 0; + } + WARN_ON(!__is_lm_address(addr)); + return __change_memory_common_cvm(addr, PAGE_SIZE * numpages, encrypt); +} + +int set_cvm_memory_encrypted(unsigned long addr, int numpages) +{ + return __set_memory_encrypted(addr, numpages, true); +} + +int set_cvm_memory_decrypted(unsigned long addr, int numpages) +{ + return __set_memory_encrypted(addr, numpages, false); +} + diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c index 804d5197c..b6eb82f6d 100644 --- a/arch/arm64/mm/mmu.c +++ b/arch/arm64/mm/mmu.c @@ -38,6 +38,9 @@ #include <asm/ptdump.h> #include <asm/tlbflush.h> #include <asm/pgalloc.h> +#ifdef CONFIG_CVM_GUEST +#include <asm/cvm_guest.h> +#endif #define NO_BLOCK_MAPPINGS BIT(0) #define NO_CONT_MAPPINGS BIT(1) @@ -494,7 +497,11 @@ static void __init map_mem(pgd_t *pgdp) int flags = 0, eflags = 0; u64 i; +#ifdef CONFIG_CVM_GUEST + if (rodata_full || debug_pagealloc_enabled() || is_cvm_world()) +#else if (rodata_full || debug_pagealloc_enabled()) +#endif flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS; #ifdef CONFIG_KFENCE @@ -1514,7 +1521,11 @@ int arch_add_memory(int nid, u64 start, u64 size, } +#ifdef CONFIG_CVM_GUEST + if (rodata_full || debug_pagealloc_enabled() || is_cvm_world()) +#else if (rodata_full || debug_pagealloc_enabled()) +#endif flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS; __create_pgd_mapping(swapper_pg_dir, start, __phys_to_virt(start), diff --git a/arch/arm64/mm/pageattr.c b/arch/arm64/mm/pageattr.c index 0bc12dbf2..fe0650386 100644 --- a/arch/arm64/mm/pageattr.c +++ b/arch/arm64/mm/pageattr.c @@ -11,6 +11,9 @@ #include <asm/cacheflush.h> #include <asm/set_memory.h> #include <asm/tlbflush.h> +#ifdef CONFIG_CVM_GUEST +#include <asm/cvm_guest.h> +#endif struct page_change_data { pgprot_t set_mask; @@ -188,7 +191,11 @@ int set_direct_map_default_noflush(struct page *page) void __kernel_map_pages(struct page *page, int numpages, int enable) { - if (!debug_pagealloc_enabled() && !rodata_full) +#ifdef CONFIG_CVM_GUEST + if ((!debug_pagealloc_enabled() && !rodata_full) || is_cvm_world()) +#else + if ((!debug_pagealloc_enabled() && !rodata_full)) +#endif return; set_memory_valid((unsigned long)page_address(page), numpages, enable); diff --git a/include/linux/swiotlb.h b/include/linux/swiotlb.h index 5d2dbe7e0..b4ca6622a 100644 --- a/include/linux/swiotlb.h +++ b/include/linux/swiotlb.h @@ -6,6 +6,9 @@ #include <linux/init.h> #include <linux/types.h> #include <linux/limits.h> +#ifdef CONFIG_CVM_GUEST +#include <asm/cvm_guest.h> +#endif struct device; struct page; @@ -75,6 +78,16 @@ static inline bool is_swiotlb_buffer(phys_addr_t paddr) return paddr >= io_tlb_start && paddr < io_tlb_end; } +#ifdef CONFIG_CVM_GUEST +static inline bool is_swiotlb_for_alloc(struct device *dev) +{ + return is_cvm_world(); +} + +struct page *swiotlb_alloc(struct device *dev, size_t size); +bool swiotlb_free(struct device *dev, struct page *page, size_t size); +#endif + void __init swiotlb_exit(void); unsigned int swiotlb_max_segment(void); size_t swiotlb_max_mapping_size(struct device *dev); diff --git a/kernel/dma/direct.c b/kernel/dma/direct.c index 2922250f9..075e85cfb 100644 --- a/kernel/dma/direct.c +++ b/kernel/dma/direct.c @@ -75,6 +75,20 @@ static bool dma_coherent_ok(struct device *dev, phys_addr_t phys, size_t size) min_not_zero(dev->coherent_dma_mask, dev->bus_dma_limit); } +#ifdef CONFIG_CVM_GUEST +static struct page *dma_direct_alloc_swiotlb(struct device *dev, size_t size) +{ + struct page *page = swiotlb_alloc(dev, size); + + if (page && !dma_coherent_ok(dev, page_to_phys(page), size)) { + swiotlb_free(dev, page, size); + return NULL; + } + + return page; +} +#endif + static struct page *__dma_direct_alloc_pages(struct device *dev, size_t size, gfp_t gfp) { @@ -84,6 +98,11 @@ static struct page *__dma_direct_alloc_pages(struct device *dev, size_t size, WARN_ON_ONCE(!PAGE_ALIGNED(size)); +#ifdef CONFIG_CVM_GUEST + if (is_swiotlb_for_alloc(dev)) + return dma_direct_alloc_swiotlb(dev, size); +#endif + gfp |= dma_direct_optimal_gfp_mask(dev, dev->coherent_dma_mask, &phys_limit); page = dma_alloc_contiguous(dev, size, gfp); @@ -237,6 +256,11 @@ void *dma_direct_alloc(struct device *dev, size_t size, return NULL; } out_free_pages: +#ifdef CONFIG_CVM_GUEST + if (is_swiotlb_for_alloc(dev) && + swiotlb_free(dev, page, size)) + return NULL; +#endif dma_free_contiguous(dev, page, size); return NULL; } @@ -271,6 +295,11 @@ void dma_direct_free(struct device *dev, size_t size, else if (IS_ENABLED(CONFIG_ARCH_HAS_DMA_CLEAR_UNCACHED)) arch_dma_clear_uncached(cpu_addr, size); +#ifdef CONFIG_CVM_GUEST + if (is_swiotlb_for_alloc(dev) && + swiotlb_free(dev, dma_direct_to_page(dev, dma_addr), size)) + return; +#endif dma_free_contiguous(dev, dma_direct_to_page(dev, dma_addr), size); } @@ -307,6 +336,11 @@ struct page *dma_direct_alloc_pages(struct device *dev, size_t size, *dma_handle = phys_to_dma_direct(dev, page_to_phys(page)); return page; out_free_pages: +#ifdef CONFIG_CVM_GUEST + if (is_swiotlb_for_alloc(dev) && + swiotlb_free(dev, page, size)) + return NULL; +#endif dma_free_contiguous(dev, page, size); return NULL; } @@ -325,6 +359,11 @@ void dma_direct_free_pages(struct device *dev, size_t size, if (force_dma_unencrypted(dev)) set_memory_encrypted((unsigned long)vaddr, PFN_UP(size)); +#ifdef CONFIG_CVM_GUEST + if (is_swiotlb_for_alloc(dev) && + swiotlb_free(dev, page, size)) + return; +#endif dma_free_contiguous(dev, page, size); } diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c index d897d1613..579d3cb50 100644 --- a/kernel/dma/swiotlb.c +++ b/kernel/dma/swiotlb.c @@ -47,6 +47,10 @@ #include <linux/memblock.h> #include <linux/iommu-helper.h> +#ifdef CONFIG_CVM_GUEST +#include <asm/cvm_guest.h> +#endif + #define CREATE_TRACE_POINTS #include <trace/events/swiotlb.h> @@ -194,12 +198,20 @@ void __init swiotlb_update_mem_attributes(void) void *vaddr; unsigned long bytes; +#ifdef CONFIG_CVM_GUEST + if (!is_cvm_world() && (no_iotlb_memory || late_alloc)) +#else if (no_iotlb_memory || late_alloc) +#endif return; vaddr = phys_to_virt(io_tlb_start); bytes = PAGE_ALIGN(io_tlb_nslabs << IO_TLB_SHIFT); set_memory_decrypted((unsigned long)vaddr, bytes >> PAGE_SHIFT); +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + set_cvm_memory_decrypted((unsigned long)vaddr, bytes >> PAGE_SHIFT); +#endif memset(vaddr, 0, bytes); } @@ -265,8 +277,13 @@ swiotlb_init(int verbose) /* Get IO TLB memory from the low pages */ vstart = memblock_alloc_low(PAGE_ALIGN(bytes), PAGE_SIZE); - if (vstart && !swiotlb_init_with_tbl(vstart, io_tlb_nslabs, verbose)) + if (vstart && !swiotlb_init_with_tbl(vstart, io_tlb_nslabs, verbose)) { +#ifdef CONFIG_CVM_GUEST + if (is_cvm_world()) + swiotlb_update_mem_attributes(); +#endif return; + } if (io_tlb_start) { memblock_free_early(io_tlb_start, @@ -772,3 +789,70 @@ static int __init swiotlb_create_debugfs(void) late_initcall(swiotlb_create_debugfs); #endif + +#ifdef CONFIG_CVM_GUEST +struct page *swiotlb_alloc(struct device *dev, size_t size) +{ + phys_addr_t tlb_addr; + int index; + + index = find_slots(dev, 0, size); + if (index == -1) + return NULL; + + tlb_addr = slot_addr(io_tlb_start, index); + return pfn_to_page(PFN_DOWN(tlb_addr)); +} + +static void swiotlb_release_slots(struct device *hwdev, phys_addr_t tlb_addr, + size_t alloc_size) +{ + unsigned long flags; + unsigned int offset = swiotlb_align_offset(hwdev, tlb_addr); + int i, count, nslots = nr_slots(alloc_size + offset); + int index = (tlb_addr - offset - io_tlb_start) >> IO_TLB_SHIFT; + + /* + * Return the buffer to the free list by setting the corresponding + * entries to indicate the number of contiguous entries available. + * While returning the entries to the free list, we merge the entries + * with slots below and above the pool being returned. + */ + spin_lock_irqsave(&io_tlb_lock, flags); + if (index + nslots < ALIGN(index + 1, IO_TLB_SEGSIZE)) + count = io_tlb_list[index + nslots]; + else + count = 0; + + /* + * Step 1: return the slots to the free list, merging the slots with + * superceeding slots + */ + for (i = index + nslots - 1; i >= index; i--) { + io_tlb_list[i] = ++count; + io_tlb_orig_addr[i] = INVALID_PHYS_ADDR; + } + + /* + * Step 2: merge the returned slots with the preceding slots, if + * available (non zero) + */ + for (i = index - 1; + io_tlb_offset(i) != IO_TLB_SEGSIZE - 1 && io_tlb_list[i]; + i--) + io_tlb_list[i] = ++count; + io_tlb_used -= nslots; + spin_unlock_irqrestore(&io_tlb_lock, flags); +} + +bool swiotlb_free(struct device *dev, struct page *page, size_t size) +{ + phys_addr_t tlb_addr = page_to_phys(page); + + if (!is_swiotlb_buffer(tlb_addr)) + return false; + + swiotlb_release_slots(dev, tlb_addr, size); + return true; +} +#endif -- 2.33.0

1 0

[PATCH OLK-5.10 2/4] pmu: enable pmu phys irq inject for cvm
by hejingxian 08 Apr '24

08 Apr '24

From: Jingxian He <hejingxian(a)huawei.com> hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I9CC0X CVE: N/A ------------ Enable pmu phys irq inject to Confidential VMs. Signed-off-by: Jingxian He <hejingxian(a)huawei.com> --- arch/arm64/include/asm/kvm_tmi.h | 5 ++++- arch/arm64/kvm/arm.c | 24 ++++++++++++++++++++++++ arch/arm64/kvm/pmu-emul.c | 9 +++++++++ drivers/perf/arm_pmu.c | 17 +++++++++++++++++ include/linux/perf/arm_pmu.h | 3 +++ 5 files changed, 57 insertions(+), 1 deletion(-) diff --git a/arch/arm64/include/asm/kvm_tmi.h b/arch/arm64/include/asm/kvm_tmi.h index 554b3e439..49ae4f77c 100644 --- a/arch/arm64/include/asm/kvm_tmi.h +++ b/arch/arm64/include/asm/kvm_tmi.h @@ -4,6 +4,7 @@ */ #ifndef __TMM_TMI_H #define __TMM_TMI_H +#ifdef CONFIG_CVM_HOST #include <linux/kvm_host.h> #include <asm/kvm_asm.h> #include <asm/kvm_pgtable.h> @@ -144,6 +145,7 @@ struct tmi_tec_exit { uint64_t cntp_ctl; uint64_t cntp_cval; uint64_t imm; + uint64_t pmu_ovf_status; }; struct tmi_tec_run { @@ -370,4 +372,5 @@ unsigned long cvm_psci_vcpu_affinity_info(struct kvm_vcpu *vcpu, int kvm_cvm_vcpu_set_events(struct kvm_vcpu *vcpu, bool serror_pending, bool ext_dabt_pending); -#endif +#endif +#endif diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c index 32974a10e..6790b06f9 100644 --- a/arch/arm64/kvm/arm.c +++ b/arch/arm64/kvm/arm.c @@ -40,6 +40,7 @@ #include <asm/sections.h> #ifdef CONFIG_CVM_HOST #include <asm/kvm_tmi.h> +#include <linux/perf/arm_pmu.h> #endif #include <kvm/arm_hypercalls.h> @@ -890,6 +891,18 @@ static bool kvm_vcpu_exit_request(struct kvm_vcpu *vcpu, int *ret) xfer_to_guest_mode_work_pending(); } +#ifdef CONFIG_CVM_HOST +static inline void update_pmu_phys_irq(struct kvm_vcpu *vcpu, bool *pmu_stopped) +{ + struct kvm_pmu *pmu = &vcpu->arch.pmu; + + if (pmu->irq_level) { + *pmu_stopped = true; + arm_pmu_set_phys_irq(false); + } +} +#endif + /** * kvm_arch_vcpu_ioctl_run - the main VCPU run function to execute guest code * @vcpu: The VCPU pointer @@ -934,6 +947,9 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu) ret = 1; run->exit_reason = KVM_EXIT_UNKNOWN; while (ret > 0) { +#ifdef CONFIG_CVM_HOST + bool pmu_stopped = false; +#endif /* * Check conditions before entering the guest */ @@ -953,6 +969,10 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu) preempt_disable(); kvm_pmu_flush_hwstate(vcpu); +#ifdef CONFIG_CVM_HOST + if (vcpu_is_tec(vcpu)) + update_pmu_phys_irq(vcpu, &pmu_stopped); +#endif local_irq_disable(); @@ -1063,6 +1083,10 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu) } #endif preempt_enable(); +#ifdef CONFIG_CVM_HOST + if (pmu_stopped) + arm_pmu_set_phys_irq(true); +#endif /* * The ARMv8 architecture doesn't give the hypervisor diff --git a/arch/arm64/kvm/pmu-emul.c b/arch/arm64/kvm/pmu-emul.c index 9fdc76c6d..00aa9ebe6 100644 --- a/arch/arm64/kvm/pmu-emul.c +++ b/arch/arm64/kvm/pmu-emul.c @@ -13,6 +13,7 @@ #include <asm/kvm_emulate.h> #include <kvm/arm_pmu.h> #include <kvm/arm_vgic.h> +#include <asm/kvm_tmi.h> static void kvm_pmu_create_perf_event(struct kvm_vcpu *vcpu, u64 select_idx); static void kvm_pmu_update_pmc_chained(struct kvm_vcpu *vcpu, u64 select_idx); @@ -370,6 +371,14 @@ static u64 kvm_pmu_overflow_status(struct kvm_vcpu *vcpu) { u64 reg = 0; +#ifdef CONFIG_CVM_HOST + if (vcpu_is_tec(vcpu)) { + struct tmi_tec_run *run = vcpu->arch.tec.tec_run; + reg = run->tec_exit.pmu_ovf_status; + return reg; + } +#endif + if ((__vcpu_sys_reg(vcpu, PMCR_EL0) & ARMV8_PMU_PMCR_E)) { reg = __vcpu_sys_reg(vcpu, PMOVSSET_EL0); reg &= __vcpu_sys_reg(vcpu, PMCNTENSET_EL0); diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c index 4ef8aee84..743f52d94 100644 --- a/drivers/perf/arm_pmu.c +++ b/drivers/perf/arm_pmu.c @@ -797,6 +797,23 @@ static int arm_perf_teardown_cpu(unsigned int cpu, struct hlist_node *node) return 0; } +#ifdef CONFIG_CVM_HOST +void arm_pmu_set_phys_irq(bool enable) +{ + int cpu = get_cpu(); + struct arm_pmu *pmu = per_cpu(cpu_armpmu, cpu); + int irq; + + irq = armpmu_get_cpu_irq(pmu, cpu); + if (irq && !enable) + per_cpu(cpu_irq_ops, cpu)->disable_pmuirq(irq); + else if (irq && enable) + per_cpu(cpu_irq_ops, cpu)->enable_pmuirq(irq); + + put_cpu(); +} +#endif + #ifdef CONFIG_CPU_PM static void cpu_pm_pmu_setup(struct arm_pmu *armpmu, unsigned long cmd) { diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h index 6fd58c8f9..c7a35d321 100644 --- a/include/linux/perf/arm_pmu.h +++ b/include/linux/perf/arm_pmu.h @@ -189,6 +189,9 @@ int arm_pmu_acpi_probe(armpmu_init_fn init_fn); static inline int arm_pmu_acpi_probe(armpmu_init_fn init_fn) { return 0; } #endif +#ifdef CONFIG_CVM_HOST +void arm_pmu_set_phys_irq(bool enable); +#endif /* Internal functions only for core arm_pmu code */ struct arm_pmu *armpmu_alloc(void); struct arm_pmu *armpmu_alloc_atomic(void); -- 2.33.0

1 0

[PATCH OLK-5.10 v1 0/4] add cvm feature patches
by hejingxian 08 Apr '24

08 Apr '24

From: Jingxian He <hejingxian(a)huawei.com> Add cvm feature patches: 1. add cvm host feature 2. enable pmu phys irq inject for cvm 3. add bounce buffer feature for cvm guest 4. add lpi support for cvm guest arch/arm64/configs/defconfig | 2 + arch/arm64/configs/openeuler_defconfig | 2 + arch/arm64/include/asm/cvm_guest.h | 21 + arch/arm64/include/asm/kvm_emulate.h | 14 + arch/arm64/include/asm/kvm_host.h | 12 + arch/arm64/include/asm/kvm_tmi.h | 376 +++++++++++ arch/arm64/include/asm/kvm_tmm.h | 72 +++ arch/arm64/kvm/Kconfig | 16 + arch/arm64/kvm/Makefile | 5 + arch/arm64/kvm/arch_timer.c | 104 +++- arch/arm64/kvm/arm.c | 155 ++++- arch/arm64/kvm/cvm.c | 824 +++++++++++++++++++++++++ arch/arm64/kvm/cvm_exit.c | 229 +++++++ arch/arm64/kvm/cvm_guest.c | 90 +++ arch/arm64/kvm/guest.c | 8 + arch/arm64/kvm/hyp/vgic-v3-sr.c | 19 + arch/arm64/kvm/mmio.c | 17 +- arch/arm64/kvm/mmu.c | 7 + arch/arm64/kvm/pmu-emul.c | 9 + arch/arm64/kvm/psci.c | 12 +- arch/arm64/kvm/reset.c | 10 + arch/arm64/kvm/tmi.c | 148 +++++ arch/arm64/kvm/vgic/vgic-v3.c | 16 +- arch/arm64/kvm/vgic/vgic.c | 52 +- arch/arm64/mm/mmu.c | 11 + arch/arm64/mm/pageattr.c | 9 +- drivers/irqchip/irq-gic-v3-its.c | 228 ++++++- drivers/perf/arm_pmu.c | 17 + include/kvm/arm_arch_timer.h | 4 + include/linux/kvm_host.h | 21 + include/linux/perf/arm_pmu.h | 3 + include/linux/swiotlb.h | 13 + include/uapi/linux/kvm.h | 29 + kernel/dma/direct.c | 39 ++ kernel/dma/swiotlb.c | 86 ++- virt/kvm/kvm_main.c | 7 +- 36 files changed, 2646 insertions(+), 41 deletions(-) create mode 100644 arch/arm64/include/asm/cvm_guest.h create mode 100644 arch/arm64/include/asm/kvm_tmi.h create mode 100644 arch/arm64/include/asm/kvm_tmm.h create mode 100644 arch/arm64/kvm/cvm.c create mode 100644 arch/arm64/kvm/cvm_exit.c create mode 100644 arch/arm64/kvm/cvm_guest.c create mode 100644 arch/arm64/kvm/tmi.c -- 2.33.0

1 0

[PATCH OLK-5.10] nilfs2: fix hang in nilfs_lookup_dirty_data_buffers()
by Yifan Qiao 08 Apr '24

08 Apr '24

From: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> stable inclusion from stable-v5.10.210 commit 98a4026b22ff440c7f47056481bcbbe442f607d6 category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/I9E2EQ CVE: CVE-2024-26696 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- commit 38296afe3c6ee07319e01bb249aa4bb47c07b534 upstream. Syzbot reported a hang issue in migrate_pages_batch() called by mbind() and nilfs_lookup_dirty_data_buffers() called in the log writer of nilfs2. While migrate_pages_batch() locks a folio and waits for the writeback to complete, the log writer thread that should bring the writeback to completion picks up the folio being written back in nilfs_lookup_dirty_data_buffers() that it calls for subsequent log creation and was trying to lock the folio. Thus causing a deadlock. In the first place, it is unexpected that folios/pages in the middle of writeback will be updated and become dirty. Nilfs2 adds a checksum to verify the validity of the log being written and uses it for recovery at mount, so data changes during writeback are suppressed. Since this is broken, an unclean shutdown could potentially cause recovery to fail. Investigation revealed that the root cause is that the wait for writeback completion in nilfs_page_mkwrite() is conditional, and if the backing device does not require stable writes, data may be modified without waiting. Fix these issues by making nilfs_page_mkwrite() wait for writeback to finish regardless of the stable write requirement of the backing device. Link: https://lkml.kernel.org/r/20240131145657.4209-1-konishi.ryusuke@gmail.com Fixes: 1d1d1a767206 ("mm: only enforce stable page writes if the backing device requires it") Signed-off-by: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> Reported-by: syzbot+ee2ae68da3b22d04cd8d(a)syzkaller.appspotmail.com Closes: https://lkml.kernel.org/r/00000000000047d819061004ad6c@google.com Tested-by: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> Signed-off-by: Lin Yujun <linyujun809(a)huawei.com> Signed-off-by: Yifan Qiao <qiaoyifan4(a)huawei.com> --- fs/nilfs2/file.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/fs/nilfs2/file.c b/fs/nilfs2/file.c index 64bc81363c6c..3802b42e1cb4 100644 --- a/fs/nilfs2/file.c +++ b/fs/nilfs2/file.c @@ -105,7 +105,13 @@ static vm_fault_t nilfs_page_mkwrite(struct vm_fault *vmf) nilfs_transaction_commit(inode->i_sb); mapped: - wait_for_stable_page(page); + /* + * Since checksumming including data blocks is performed to determine + * the validity of the log to be written and used for recovery, it is + * necessary to wait for writeback to finish here, regardless of the + * stable write requirement of the backing device. + */ + wait_on_page_writeback(page); out: sb_end_pagefault(inode->i_sb); return block_page_mkwrite_return(ret); -- 2.39.2

2 1

[PATCH openEuler-1.0-LTS] nilfs2: fix hang in nilfs_lookup_dirty_data_buffers()
by Yifan Qiao 08 Apr '24

08 Apr '24

From: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> stable inclusion from stable-v5.10.210 commit 98a4026b22ff440c7f47056481bcbbe442f607d6 category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/I9E2EQ CVE: CVE-2024-26696 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- commit 38296afe3c6ee07319e01bb249aa4bb47c07b534 upstream. Syzbot reported a hang issue in migrate_pages_batch() called by mbind() and nilfs_lookup_dirty_data_buffers() called in the log writer of nilfs2. While migrate_pages_batch() locks a folio and waits for the writeback to complete, the log writer thread that should bring the writeback to completion picks up the folio being written back in nilfs_lookup_dirty_data_buffers() that it calls for subsequent log creation and was trying to lock the folio. Thus causing a deadlock. In the first place, it is unexpected that folios/pages in the middle of writeback will be updated and become dirty. Nilfs2 adds a checksum to verify the validity of the log being written and uses it for recovery at mount, so data changes during writeback are suppressed. Since this is broken, an unclean shutdown could potentially cause recovery to fail. Investigation revealed that the root cause is that the wait for writeback completion in nilfs_page_mkwrite() is conditional, and if the backing device does not require stable writes, data may be modified without waiting. Fix these issues by making nilfs_page_mkwrite() wait for writeback to finish regardless of the stable write requirement of the backing device. Link: https://lkml.kernel.org/r/20240131145657.4209-1-konishi.ryusuke@gmail.com Fixes: 1d1d1a767206 ("mm: only enforce stable page writes if the backing device requires it") Signed-off-by: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> Reported-by: syzbot+ee2ae68da3b22d04cd8d(a)syzkaller.appspotmail.com Closes: https://lkml.kernel.org/r/00000000000047d819061004ad6c@google.com Tested-by: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> Signed-off-by: Lin Yujun <linyujun809(a)huawei.com> Signed-off-by: Yifan Qiao <qiaoyifan4(a)huawei.com> --- fs/nilfs2/file.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/fs/nilfs2/file.c b/fs/nilfs2/file.c index 64bc81363c6c..3802b42e1cb4 100644 --- a/fs/nilfs2/file.c +++ b/fs/nilfs2/file.c @@ -105,7 +105,13 @@ static vm_fault_t nilfs_page_mkwrite(struct vm_fault *vmf) nilfs_transaction_commit(inode->i_sb); mapped: - wait_for_stable_page(page); + /* + * Since checksumming including data blocks is performed to determine + * the validity of the log to be written and used for recovery, it is + * necessary to wait for writeback to finish here, regardless of the + * stable write requirement of the backing device. + */ + wait_on_page_writeback(page); out: sb_end_pagefault(inode->i_sb); return block_page_mkwrite_return(ret); -- 2.39.2

2 2

[PATCH OLK-5.10] nilfs2: fix hang in nilfs_lookup_dirty_data_buffers()
by Yifan Qiao 08 Apr '24

08 Apr '24

From: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> stable inclusion from stable-v5.10.210 commit 98a4026b22ff440c7f47056481bcbbe442f607d6 category: bugfix bugzilla: 189665 CVE: CVE-2024-26696 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- commit 38296afe3c6ee07319e01bb249aa4bb47c07b534 upstream. Syzbot reported a hang issue in migrate_pages_batch() called by mbind() and nilfs_lookup_dirty_data_buffers() called in the log writer of nilfs2. While migrate_pages_batch() locks a folio and waits for the writeback to complete, the log writer thread that should bring the writeback to completion picks up the folio being written back in nilfs_lookup_dirty_data_buffers() that it calls for subsequent log creation and was trying to lock the folio. Thus causing a deadlock. In the first place, it is unexpected that folios/pages in the middle of writeback will be updated and become dirty. Nilfs2 adds a checksum to verify the validity of the log being written and uses it for recovery at mount, so data changes during writeback are suppressed. Since this is broken, an unclean shutdown could potentially cause recovery to fail. Investigation revealed that the root cause is that the wait for writeback completion in nilfs_page_mkwrite() is conditional, and if the backing device does not require stable writes, data may be modified without waiting. Fix these issues by making nilfs_page_mkwrite() wait for writeback to finish regardless of the stable write requirement of the backing device. Link: https://lkml.kernel.org/r/20240131145657.4209-1-konishi.ryusuke@gmail.com Fixes: 1d1d1a767206 ("mm: only enforce stable page writes if the backing device requires it") Signed-off-by: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> Reported-by: syzbot+ee2ae68da3b22d04cd8d(a)syzkaller.appspotmail.com Closes: https://lkml.kernel.org/r/00000000000047d819061004ad6c@google.com Tested-by: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> Signed-off-by: Lin Yujun <linyujun809(a)huawei.com> Signed-off-by: Yifan Qiao <qiaoyifan4(a)huawei.com> --- fs/nilfs2/file.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/fs/nilfs2/file.c b/fs/nilfs2/file.c index 64bc81363c6c..3802b42e1cb4 100644 --- a/fs/nilfs2/file.c +++ b/fs/nilfs2/file.c @@ -105,7 +105,13 @@ static vm_fault_t nilfs_page_mkwrite(struct vm_fault *vmf) nilfs_transaction_commit(inode->i_sb); mapped: - wait_for_stable_page(page); + /* + * Since checksumming including data blocks is performed to determine + * the validity of the log to be written and used for recovery, it is + * necessary to wait for writeback to finish here, regardless of the + * stable write requirement of the backing device. + */ + wait_on_page_writeback(page); out: sb_end_pagefault(inode->i_sb); return block_page_mkwrite_return(ret); -- 2.39.2

2 1

[PATCH openEuler-1.0-LTS] nilfs2: fix hang in nilfs_lookup_dirty_data_buffers()
by Yifan Qiao 08 Apr '24

08 Apr '24

From: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> stable inclusion from stable-v5.10.210 commit 98a4026b22ff440c7f47056481bcbbe442f607d6 category: bugfix bugzilla: 189665 CVE: CVE-2024-26696 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- commit 38296afe3c6ee07319e01bb249aa4bb47c07b534 upstream. Syzbot reported a hang issue in migrate_pages_batch() called by mbind() and nilfs_lookup_dirty_data_buffers() called in the log writer of nilfs2. While migrate_pages_batch() locks a folio and waits for the writeback to complete, the log writer thread that should bring the writeback to completion picks up the folio being written back in nilfs_lookup_dirty_data_buffers() that it calls for subsequent log creation and was trying to lock the folio. Thus causing a deadlock. In the first place, it is unexpected that folios/pages in the middle of writeback will be updated and become dirty. Nilfs2 adds a checksum to verify the validity of the log being written and uses it for recovery at mount, so data changes during writeback are suppressed. Since this is broken, an unclean shutdown could potentially cause recovery to fail. Investigation revealed that the root cause is that the wait for writeback completion in nilfs_page_mkwrite() is conditional, and if the backing device does not require stable writes, data may be modified without waiting. Fix these issues by making nilfs_page_mkwrite() wait for writeback to finish regardless of the stable write requirement of the backing device. Link: https://lkml.kernel.org/r/20240131145657.4209-1-konishi.ryusuke@gmail.com Fixes: 1d1d1a767206 ("mm: only enforce stable page writes if the backing device requires it") Signed-off-by: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> Reported-by: syzbot+ee2ae68da3b22d04cd8d(a)syzkaller.appspotmail.com Closes: https://lkml.kernel.org/r/00000000000047d819061004ad6c@google.com Tested-by: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> Signed-off-by: Lin Yujun <linyujun809(a)huawei.com> Signed-off-by: Yifan Qiao <qiaoyifan4(a)huawei.com> --- fs/nilfs2/file.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/fs/nilfs2/file.c b/fs/nilfs2/file.c index 64bc81363c6c..3802b42e1cb4 100644 --- a/fs/nilfs2/file.c +++ b/fs/nilfs2/file.c @@ -105,7 +105,13 @@ static vm_fault_t nilfs_page_mkwrite(struct vm_fault *vmf) nilfs_transaction_commit(inode->i_sb); mapped: - wait_for_stable_page(page); + /* + * Since checksumming including data blocks is performed to determine + * the validity of the log to be written and used for recovery, it is + * necessary to wait for writeback to finish here, regardless of the + * stable write requirement of the backing device. + */ + wait_on_page_writeback(page); out: sb_end_pagefault(inode->i_sb); return block_page_mkwrite_return(ret); -- 2.39.2

2 1

[PATCH openEuler-1.0-LTS] nilfs2: fix hang in nilfs_lookup_dirty_data_buffers()
by Yifan Qiao 08 Apr '24

08 Apr '24

From: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> stable inclusion from stable-v5.10.210 commit 98a4026b22ff440c7f47056481bcbbe442f607d6 category: bugfix bugzilla: 189665 CVE: CVE-2024-26696 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- commit 38296afe3c6ee07319e01bb249aa4bb47c07b534 upstream. Syzbot reported a hang issue in migrate_pages_batch() called by mbind() and nilfs_lookup_dirty_data_buffers() called in the log writer of nilfs2. While migrate_pages_batch() locks a folio and waits for the writeback to complete, the log writer thread that should bring the writeback to completion picks up the folio being written back in nilfs_lookup_dirty_data_buffers() that it calls for subsequent log creation and was trying to lock the folio. Thus causing a deadlock. In the first place, it is unexpected that folios/pages in the middle of writeback will be updated and become dirty. Nilfs2 adds a checksum to verify the validity of the log being written and uses it for recovery at mount, so data changes during writeback are suppressed. Since this is broken, an unclean shutdown could potentially cause recovery to fail. Investigation revealed that the root cause is that the wait for writeback completion in nilfs_page_mkwrite() is conditional, and if the backing device does not require stable writes, data may be modified without waiting. Fix these issues by making nilfs_page_mkwrite() wait for writeback to finish regardless of the stable write requirement of the backing device. Link: https://lkml.kernel.org/r/20240131145657.4209-1-konishi.ryusuke@gmail.com Fixes: 1d1d1a767206 ("mm: only enforce stable page writes if the backing device requires it") Signed-off-by: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> Reported-by: syzbot+ee2ae68da3b22d04cd8d(a)syzkaller.appspotmail.com Closes: https://lkml.kernel.org/r/00000000000047d819061004ad6c@google.com Tested-by: Ryusuke Konishi <konishi.ryusuke(a)gmail.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> Signed-off-by: Lin Yujun <linyujun809(a)huawei.com> Signed-off-by: Yifan Qiao <qiaoyifan4(a)huawei.com> --- fs/nilfs2/file.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/fs/nilfs2/file.c b/fs/nilfs2/file.c index 64bc81363c6c..3802b42e1cb4 100644 --- a/fs/nilfs2/file.c +++ b/fs/nilfs2/file.c @@ -105,7 +105,13 @@ static vm_fault_t nilfs_page_mkwrite(struct vm_fault *vmf) nilfs_transaction_commit(inode->i_sb); mapped: - wait_for_stable_page(page); + /* + * Since checksumming including data blocks is performed to determine + * the validity of the log to be written and used for recovery, it is + * necessary to wait for writeback to finish here, regardless of the + * stable write requirement of the backing device. + */ + wait_on_page_writeback(page); out: sb_end_pagefault(inode->i_sb); return block_page_mkwrite_return(ret); -- 2.39.2

2 1

[PATCH OLK-5.10] fs/ntfs3: Fix an NULL dereference bug
by Long Li 08 Apr '24

08 Apr '24

From: Dan Carpenter <dan.carpenter(a)linaro.org> stable inclusion from stable-v5.15.81 commit ae4acad41b0f93f1c26cc0fc9135bb79d8282d0b category: bugfix bugzilla: https://gitee.com/src-openeuler/kernel/issues/I9DNXE CVE: CVE-2023-52631 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=… -------------------------------- [ Upstream commit b2dd7b953c25ffd5912dda17e980e7168bebcf6c ] The issue here is when this is called from ntfs_load_attr_list(). The "size" comes from le32_to_cpu(attr->res.data_size) so it can't overflow on a 64bit systems but on 32bit systems the "+ 1023" can overflow and the result is zero. This means that the kmalloc will succeed by returning the ZERO_SIZE_PTR and then the memcpy() will crash with an Oops on the next line. Fixes: be71b5cba2e6 ("fs/ntfs3: Add attrib operations") Signed-off-by: Dan Carpenter <dan.carpenter(a)linaro.org> Signed-off-by: Konstantin Komarov <almaz.alexandrovich(a)paragon-software.com> Signed-off-by: Sasha Levin <sashal(a)kernel.org> Signed-off-by: Long Li <leo.lilong(a)huawei.com> --- fs/ntfs3/ntfs_fs.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/ntfs3/ntfs_fs.h b/fs/ntfs3/ntfs_fs.h index 7e84f0060133..e0c8f59d62ab 100644 --- a/fs/ntfs3/ntfs_fs.h +++ b/fs/ntfs3/ntfs_fs.h @@ -466,7 +466,7 @@ bool al_delete_le(struct ntfs_inode *ni, enum ATTR_TYPE type, CLST vcn, int al_update(struct ntfs_inode *ni, int sync); static inline size_t al_aligned(size_t size) { - return (size + 1023) & ~(size_t)1023; + return size_add(size, 1023) & ~(size_t)1023; } /* Globals from bitfunc.c */ -- 2.31.1

2 1

2024

2023

2022

2021

2020

2019

Kernel April 2024