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

[PATCH] powerpc/64s/radix/kfence: map __kfence_pool at page granularity

by Aboorva Devarajan

From: Hari Bathini <hbathini(a)linux.ibm.com> When KFENCE is enabled, total system memory is mapped at page level granularity. But in radix MMU mode, ~3GB additional memory is needed to map 100GB of system memory at page level granularity when compared to using 2MB direct mapping.This is not desired considering KFENCE is designed to be enabled in production kernels [1]. Mapping only the memory allocated for KFENCE pool at page granularity is sufficient to enable KFENCE support. So, allocate __kfence_pool during bootup and map it at page granularity instead of mapping all system memory at page granularity. Without patch: # cat /proc/meminfo MemTotal: 101201920 kB With patch: # cat /proc/meminfo MemTotal: 104483904 kB Note that enabling KFENCE at runtime is disabled for radix MMU for now, as it depends on the ability to split page table mappings and such APIs are not currently implemented for radix MMU. All kfence_test.c testcases passed with this patch. [1] https://lore.kernel.org/all/20201103175841.3495947-2-elver@google.com/ Fixes: a5edf9815dd7 ("powerpc/64s: Enable KFENCE on book3s64") Cc: stable(a)vger.kernel.org Signed-off-by: Hari Bathini <hbathini(a)linux.ibm.com> Signed-off-by: Michael Ellerman <mpe(a)ellerman.id.au> Signed-off-by: Aboorva Devarajan <aboorvad(a)linux.ibm.com> Link: https://msgid.link/20240701130021.578240-1-hbathini@linux.ibm.com --- Upstream commit 353d7a84c214 ("powerpc/64s/radix/kfence: map __kfence_pool at page granularity") This has already been merged upstream and is required in stable kernels as well. --- arch/powerpc/include/asm/kfence.h | 11 +++- arch/powerpc/mm/book3s64/radix_pgtable.c | 84 ++++++++++++++++++++++-- arch/powerpc/mm/init-common.c | 3 + 3 files changed, 93 insertions(+), 5 deletions(-) diff --git a/arch/powerpc/include/asm/kfence.h b/arch/powerpc/include/asm/kfence.h index 424ceef82ae615..fab124ada1c7f2 100644 --- a/arch/powerpc/include/asm/kfence.h +++ b/arch/powerpc/include/asm/kfence.h @@ -15,10 +15,19 @@ #define ARCH_FUNC_PREFIX "." #endif +#ifdef CONFIG_KFENCE +extern bool kfence_disabled; + +static inline void disable_kfence(void) +{ + kfence_disabled = true; +} + static inline bool arch_kfence_init_pool(void) { - return true; + return !kfence_disabled; } +#endif #ifdef CONFIG_PPC64 static inline bool kfence_protect_page(unsigned long addr, bool protect) diff --git a/arch/powerpc/mm/book3s64/radix_pgtable.c b/arch/powerpc/mm/book3s64/radix_pgtable.c index 15e88f1439ec20..b0d927009af83c 100644 --- a/arch/powerpc/mm/book3s64/radix_pgtable.c +++ b/arch/powerpc/mm/book3s64/radix_pgtable.c @@ -17,6 +17,7 @@ #include <linux/hugetlb.h> #include <linux/string_helpers.h> #include <linux/memory.h> +#include <linux/kfence.h> #include <asm/pgalloc.h> #include <asm/mmu_context.h> @@ -31,6 +32,7 @@ #include <asm/uaccess.h> #include <asm/ultravisor.h> #include <asm/set_memory.h> +#include <asm/kfence.h> #include <trace/events/thp.h> @@ -293,7 +295,8 @@ static unsigned long next_boundary(unsigned long addr, unsigned long end) static int __meminit create_physical_mapping(unsigned long start, unsigned long end, - int nid, pgprot_t _prot) + int nid, pgprot_t _prot, + unsigned long mapping_sz_limit) { unsigned long vaddr, addr, mapping_size = 0; bool prev_exec, exec = false; @@ -301,7 +304,10 @@ static int __meminit create_physical_mapping(unsigned long start, int psize; unsigned long max_mapping_size = memory_block_size; - if (debug_pagealloc_enabled_or_kfence()) + if (mapping_sz_limit < max_mapping_size) + max_mapping_size = mapping_sz_limit; + + if (debug_pagealloc_enabled()) max_mapping_size = PAGE_SIZE; start = ALIGN(start, PAGE_SIZE); @@ -356,8 +362,74 @@ static int __meminit create_physical_mapping(unsigned long start, return 0; } +#ifdef CONFIG_KFENCE +static bool __ro_after_init kfence_early_init = !!CONFIG_KFENCE_SAMPLE_INTERVAL; + +static int __init parse_kfence_early_init(char *arg) +{ + int val; + + if (get_option(&arg, &val)) + kfence_early_init = !!val; + return 0; +} +early_param("kfence.sample_interval", parse_kfence_early_init); + +static inline phys_addr_t alloc_kfence_pool(void) +{ + phys_addr_t kfence_pool; + + /* + * TODO: Support to enable KFENCE after bootup depends on the ability to + * split page table mappings. As such support is not currently + * implemented for radix pagetables, support enabling KFENCE + * only at system startup for now. + * + * After support for splitting mappings is available on radix, + * alloc_kfence_pool() & map_kfence_pool() can be dropped and + * mapping for __kfence_pool memory can be + * split during arch_kfence_init_pool(). + */ + if (!kfence_early_init) + goto no_kfence; + + kfence_pool = memblock_phys_alloc(KFENCE_POOL_SIZE, PAGE_SIZE); + if (!kfence_pool) + goto no_kfence; + + memblock_mark_nomap(kfence_pool, KFENCE_POOL_SIZE); + return kfence_pool; + +no_kfence: + disable_kfence(); + return 0; +} + +static inline void map_kfence_pool(phys_addr_t kfence_pool) +{ + if (!kfence_pool) + return; + + if (create_physical_mapping(kfence_pool, kfence_pool + KFENCE_POOL_SIZE, + -1, PAGE_KERNEL, PAGE_SIZE)) + goto err; + + memblock_clear_nomap(kfence_pool, KFENCE_POOL_SIZE); + __kfence_pool = __va(kfence_pool); + return; + +err: + memblock_phys_free(kfence_pool, KFENCE_POOL_SIZE); + disable_kfence(); +} +#else +static inline phys_addr_t alloc_kfence_pool(void) { return 0; } +static inline void map_kfence_pool(phys_addr_t kfence_pool) { } +#endif + static void __init radix_init_pgtable(void) { + phys_addr_t kfence_pool; unsigned long rts_field; phys_addr_t start, end; u64 i; @@ -365,6 +437,8 @@ static void __init radix_init_pgtable(void) /* We don't support slb for radix */ slb_set_size(0); + kfence_pool = alloc_kfence_pool(); + /* * Create the linear mapping */ @@ -381,9 +455,11 @@ static void __init radix_init_pgtable(void) } WARN_ON(create_physical_mapping(start, end, - -1, PAGE_KERNEL)); + -1, PAGE_KERNEL, ~0UL)); } + map_kfence_pool(kfence_pool); + if (!cpu_has_feature(CPU_FTR_HVMODE) && cpu_has_feature(CPU_FTR_P9_RADIX_PREFETCH_BUG)) { /* @@ -875,7 +951,7 @@ int __meminit radix__create_section_mapping(unsigned long start, } return create_physical_mapping(__pa(start), __pa(end), - nid, prot); + nid, prot, ~0UL); } int __meminit radix__remove_section_mapping(unsigned long start, unsigned long end) diff --git a/arch/powerpc/mm/init-common.c b/arch/powerpc/mm/init-common.c index d3a7726ecf512c..21131b96d20901 100644 --- a/arch/powerpc/mm/init-common.c +++ b/arch/powerpc/mm/init-common.c @@ -31,6 +31,9 @@ EXPORT_SYMBOL_GPL(kernstart_virt_addr); bool disable_kuep = !IS_ENABLED(CONFIG_PPC_KUEP); bool disable_kuap = !IS_ENABLED(CONFIG_PPC_KUAP); +#ifdef CONFIG_KFENCE +bool __ro_after_init kfence_disabled; +#endif static int __init parse_nosmep(char *p) {

4 days, 14 hours

2
4
0 0

Good News About Grant

by Charlotte Phoenix

4 days, 15 hours

1
0
0 0

Backport 353d7a84c214f18 ("powerpc/64s/radix/kfence: map __kfence_pool at page granularity") to stable versions 6.1 and 6.6

by Aboorva Devarajan

Hi stable maintainers, Please consider backporting upstream commit 353d7a84c214f184d5a6b62acdec8b4424159b7c ("powerpc/64s/radix/kfence: map __kfence_pool at page granularity") to the stable 6.1 and 6.6 kernels. This should be backported because it restores the intended KFENCE behaviour on powerpc64 radix MMU by limiting page-granular mappings to the KFENCE pool, rather than all system memory. This commit appeared upstream in v6.11-rc1, and fixes an issue introduced by a5edf9815dd7 ("powerpc/64s: Enable KFENCE on book3s64") but it is not backported to stable versions. The patch applies cleanly to both 6.1 and 6.6 stable versions. Thanks, Aboorva

4 days, 15 hours

1
0
0 0

[PATCH v1 1/1] scsi: ufs : core: Add ufshcd_update_evt_hist for ufs suspend error.

by Seunghwan Baek

If the ufs resume fails, the event history is updated in ufshcd_resume, but there is no code anywhere to record ufs suspend. Therefore, add code to record ufs suspend error event history. Fixes: dd11376b9f1b ("scsi: ufs: Split the drivers/scsi/ufs directory") Cc: stable(a)vger.kernel.org Signed-off-by: Seunghwan Baek <sh8267.baek(a)samsung.com> --- drivers/ufs/core/ufshcd.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c index 040a0ceb170a..6bb2781aefc7 100644 --- a/drivers/ufs/core/ufshcd.c +++ b/drivers/ufs/core/ufshcd.c @@ -10337,7 +10337,7 @@ static int ufshcd_suspend(struct ufs_hba *hba) ret = ufshcd_setup_clocks(hba, false); if (ret) { ufshcd_enable_irq(hba); - return ret; + goto out; } if (ufshcd_is_clkgating_allowed(hba)) { hba->clk_gating.state = CLKS_OFF; @@ -10349,6 +10349,9 @@ static int ufshcd_suspend(struct ufs_hba *hba) /* Put the host controller in low power mode if possible */ ufshcd_hba_vreg_set_lpm(hba); ufshcd_pm_qos_update(hba, false); +out: + if (ret) + ufshcd_update_evt_hist(hba, UFS_EVT_SUSPEND_ERR, (u32)ret); return ret; } -- 2.43.0

4 days, 15 hours

3
2
0 0

[PATCH] mpi3mr: Read missing IOCFacts flag for reply queue full overflow

by Chandrakanth Patil

The driver was not reading the MAX_REQ_PER_REPLY_QUEUE_LIMIT IOCFacts flag, so the reply-queue-full handling was never enabled even on firmware that supports it. Reading this flag enables the feature and prevents reply queue overflow Fixes: f08b24d82749 ("scsi: mpi3mr: Avoid reply queue full condition") Cc: stable(a)vger.kernel.org Signed-off-by: Chandrakanth Patil <chandrakanth.patil(a)broadcom.com> --- drivers/scsi/mpi3mr/mpi/mpi30_ioc.h | 1 + drivers/scsi/mpi3mr/mpi3mr_fw.c | 2 ++ 2 files changed, 3 insertions(+) diff --git a/drivers/scsi/mpi3mr/mpi/mpi30_ioc.h b/drivers/scsi/mpi3mr/mpi/mpi30_ioc.h index b42933fcd423..6561f98c3cb2 100644 --- a/drivers/scsi/mpi3mr/mpi/mpi30_ioc.h +++ b/drivers/scsi/mpi3mr/mpi/mpi30_ioc.h @@ -166,6 +166,7 @@ struct mpi3_ioc_facts_data { #define MPI3_IOCFACTS_FLAGS_SIGNED_NVDATA_REQUIRED (0x00010000) #define MPI3_IOCFACTS_FLAGS_DMA_ADDRESS_WIDTH_MASK (0x0000ff00) #define MPI3_IOCFACTS_FLAGS_DMA_ADDRESS_WIDTH_SHIFT (8) +#define MPI3_IOCFACTS_FLAGS_MAX_REQ_PER_REPLY_QUEUE_LIMIT (0x00000040) #define MPI3_IOCFACTS_FLAGS_INITIAL_PORT_ENABLE_MASK (0x00000030) #define MPI3_IOCFACTS_FLAGS_INITIAL_PORT_ENABLE_SHIFT (4) #define MPI3_IOCFACTS_FLAGS_INITIAL_PORT_ENABLE_NOT_STARTED (0x00000000) diff --git a/drivers/scsi/mpi3mr/mpi3mr_fw.c b/drivers/scsi/mpi3mr/mpi3mr_fw.c index 8fe6e0bf342e..8c4bb7169a87 100644 --- a/drivers/scsi/mpi3mr/mpi3mr_fw.c +++ b/drivers/scsi/mpi3mr/mpi3mr_fw.c @@ -3158,6 +3158,8 @@ static void mpi3mr_process_factsdata(struct mpi3mr_ioc *mrioc, mrioc->facts.dma_mask = (facts_flags & MPI3_IOCFACTS_FLAGS_DMA_ADDRESS_WIDTH_MASK) >> MPI3_IOCFACTS_FLAGS_DMA_ADDRESS_WIDTH_SHIFT; + mrioc->facts.max_req_limit = (facts_flags & + MPI3_IOCFACTS_FLAGS_MAX_REQ_PER_REPLY_QUEUE_LIMIT); mrioc->facts.protocol_flags = facts_data->protocol_flags; mrioc->facts.mpi_version = le32_to_cpu(facts_data->mpi_version.word); mrioc->facts.max_reqs = le16_to_cpu(facts_data->max_outstanding_requests); -- 2.47.1

4 days, 15 hours

2
1
0 0

[PATCH] RDMA/cm: Fix leaking the multicast GID table reference

by Jason Gunthorpe

If the CM ID is destroyed while the CM event for multicast creating is still queued the cancel_work_sync() will prevent the work from running which also prevents destroying the ah_attr. This leaks a refcount and triggers a WARN: GID entry ref leak for dev syz1 index 2 ref=573 WARNING: CPU: 1 PID: 655 at drivers/infiniband/core/cache.c:809 release_gid_table drivers/infiniband/core/cache.c:806 [inline] WARNING: CPU: 1 PID: 655 at drivers/infiniband/core/cache.c:809 gid_table_release_one+0x284/0x3cc drivers/infiniband/core/cache.c:886 Destroy the ah_attr after canceling the work, it is safe to call this twice. Cc: stable(a)vger.kernel.org Fixes: fe454dc31e84 ("RDMA/ucma: Fix use-after-free bug in ucma_create_uevent") Reported-by: syzbot+b0da83a6c0e2e2bddbd4(a)syzkaller.appspotmail.com Closes: https://lore.kernel.org/all/68232e7b.050a0220.f2294.09f6.GAE@google.com Signed-off-by: Jason Gunthorpe <jgg(a)nvidia.com> --- drivers/infiniband/core/cma.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index 95e89f5c147c2c..4f5fd47086ab90 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -2031,6 +2031,8 @@ static void destroy_mc(struct rdma_id_private *id_priv, dev_put(ndev); cancel_work_sync(&mc->iboe_join.work); + if (event->event == RDMA_CM_EVENT_MULTICAST_JOIN) + rdma_destroy_ah_attr(&event->param.ud.ah_attr); } kfree(mc); } base-commit: 3fbaef0942719187f3396bfd0c780d55d35e0980 -- 2.43.0

4 days, 17 hours

1
2
0 0

[PATCH] RDMA/core: Check for the presence of LS_NLA_TYPE_DGID correctly

by Jason Gunthorpe

The netlink response for RDMA_NL_LS_OP_IP_RESOLVE should always have a LS_NLA_TYPE_DGID attribute, it is invalid if it does not. Use the nl parsing logic properly and call nla_parse_deprecated() to fill the nlattrs array and then directly index that array to get the data for the DGID. Just fail if it is NULL. Remove the for loop searching for the nla, and squash the validation and parsing into one function. Fixes an uninitialized read from the stack triggered by userspace if it does not provide the DGID to a kernel initiated RDMA_NL_LS_OP_IP_RESOLVE query. BUG: KMSAN: uninit-value in hex_byte_pack include/linux/hex.h:13 [inline] BUG: KMSAN: uninit-value in ip6_string+0xef4/0x13a0 lib/vsprintf.c:1490 hex_byte_pack include/linux/hex.h:13 [inline] ip6_string+0xef4/0x13a0 lib/vsprintf.c:1490 ip6_addr_string+0x18a/0x3e0 lib/vsprintf.c:1509 ip_addr_string+0x245/0xee0 lib/vsprintf.c:1633 pointer+0xc09/0x1bd0 lib/vsprintf.c:2542 vsnprintf+0xf8a/0x1bd0 lib/vsprintf.c:2930 vprintk_store+0x3ae/0x1530 kernel/printk/printk.c:2279 vprintk_emit+0x307/0xcd0 kernel/printk/printk.c:2426 vprintk_default+0x3f/0x50 kernel/printk/printk.c:2465 vprintk+0x36/0x50 kernel/printk/printk_safe.c:82 _printk+0x17e/0x1b0 kernel/printk/printk.c:2475 ib_nl_process_good_ip_rsep drivers/infiniband/core/addr.c:128 [inline] ib_nl_handle_ip_res_resp+0x963/0x9d0 drivers/infiniband/core/addr.c:141 rdma_nl_rcv_msg drivers/infiniband/core/netlink.c:-1 [inline] rdma_nl_rcv_skb drivers/infiniband/core/netlink.c:239 [inline] rdma_nl_rcv+0xefa/0x11c0 drivers/infiniband/core/netlink.c:259 netlink_unicast_kernel net/netlink/af_netlink.c:1320 [inline] netlink_unicast+0xf04/0x12b0 net/netlink/af_netlink.c:1346 netlink_sendmsg+0x10b3/0x1250 net/netlink/af_netlink.c:1896 sock_sendmsg_nosec net/socket.c:714 [inline] __sock_sendmsg+0x333/0x3d0 net/socket.c:729 ____sys_sendmsg+0x7e0/0xd80 net/socket.c:2617 ___sys_sendmsg+0x271/0x3b0 net/socket.c:2671 __sys_sendmsg+0x1aa/0x300 net/socket.c:2703 __compat_sys_sendmsg net/compat.c:346 [inline] __do_compat_sys_sendmsg net/compat.c:353 [inline] __se_compat_sys_sendmsg net/compat.c:350 [inline] __ia32_compat_sys_sendmsg+0xa4/0x100 net/compat.c:350 ia32_sys_call+0x3f6c/0x4310 arch/x86/include/generated/asm/syscalls_32.h:371 do_syscall_32_irqs_on arch/x86/entry/syscall_32.c:83 [inline] __do_fast_syscall_32+0xb0/0x150 arch/x86/entry/syscall_32.c:306 do_fast_syscall_32+0x38/0x80 arch/x86/entry/syscall_32.c:331 do_SYSENTER_32+0x1f/0x30 arch/x86/entry/syscall_32.c:3 Cc: stable(a)vger.kernel.org Fixes: ae43f8286730 ("IB/core: Add IP to GID netlink offload") Reported-by: syzbot+938fcd548c303fe33c1a(a)syzkaller.appspotmail.com Closes: https://lore.kernel.org/r/68dc3dac.a00a0220.102ee.004f.GAE@google.com Signed-off-by: Jason Gunthorpe <jgg(a)nvidia.com> --- drivers/infiniband/core/addr.c | 33 ++++++++++----------------------- 1 file changed, 10 insertions(+), 23 deletions(-) diff --git a/drivers/infiniband/core/addr.c b/drivers/infiniband/core/addr.c index 61596cda2b65f3..35ba852a172aad 100644 --- a/drivers/infiniband/core/addr.c +++ b/drivers/infiniband/core/addr.c @@ -80,37 +80,25 @@ static const struct nla_policy ib_nl_addr_policy[LS_NLA_TYPE_MAX] = { .min = sizeof(struct rdma_nla_ls_gid)}, }; -static inline bool ib_nl_is_good_ip_resp(const struct nlmsghdr *nlh) +static void ib_nl_process_ip_rsep(const struct nlmsghdr *nlh) { struct nlattr *tb[LS_NLA_TYPE_MAX] = {}; + union ib_gid gid; + struct addr_req *req; + int found = 0; int ret; if (nlh->nlmsg_flags & RDMA_NL_LS_F_ERR) - return false; + return; ret = nla_parse_deprecated(tb, LS_NLA_TYPE_MAX - 1, nlmsg_data(nlh), nlmsg_len(nlh), ib_nl_addr_policy, NULL); if (ret) - return false; + return; - return true; -} - -static void ib_nl_process_good_ip_rsep(const struct nlmsghdr *nlh) -{ - const struct nlattr *head, *curr; - union ib_gid gid; - struct addr_req *req; - int len, rem; - int found = 0; - - head = (const struct nlattr *)nlmsg_data(nlh); - len = nlmsg_len(nlh); - - nla_for_each_attr(curr, head, len, rem) { - if (curr->nla_type == LS_NLA_TYPE_DGID) - memcpy(&gid, nla_data(curr), nla_len(curr)); - } + if (!tb[LS_NLA_TYPE_DGID]) + return; + memcpy(&gid, nla_data(tb[LS_NLA_TYPE_DGID]), sizeof(gid)); spin_lock_bh(&lock); list_for_each_entry(req, &req_list, list) { @@ -137,8 +125,7 @@ int ib_nl_handle_ip_res_resp(struct sk_buff *skb, !(NETLINK_CB(skb).sk)) return -EPERM; - if (ib_nl_is_good_ip_resp(nlh)) - ib_nl_process_good_ip_rsep(nlh); + ib_nl_process_ip_rsep(nlh); return 0; } base-commit: 80a85a771deb113cfe2e295fb9e84467a43ebfe4 -- 2.43.0

4 days, 17 hours

1
1
0 0

[PATCH] usb: dwc3: of-simple: fix clock resource leak in dwc3_of_simple_probe

by Miaoqian Lin

When clk_bulk_prepare_enable() fails, the error path jumps to err_resetc_assert, skipping clk_bulk_put_all() and leaking the clock references acquired by clk_bulk_get_all(). Add err_clk_put_all label to properly release clock resources in all error paths. Found via static analysis and code review. Fixes: c0c61471ef86 ("usb: dwc3: of-simple: Convert to bulk clk API") Cc: stable(a)vger.kernel.org Signed-off-by: Miaoqian Lin <linmq006(a)gmail.com> --- drivers/usb/dwc3/dwc3-of-simple.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/drivers/usb/dwc3/dwc3-of-simple.c b/drivers/usb/dwc3/dwc3-of-simple.c index a4954a21be93..c116143335d9 100644 --- a/drivers/usb/dwc3/dwc3-of-simple.c +++ b/drivers/usb/dwc3/dwc3-of-simple.c @@ -70,11 +70,11 @@ static int dwc3_of_simple_probe(struct platform_device *pdev) simple->num_clocks = ret; ret = clk_bulk_prepare_enable(simple->num_clocks, simple->clks); if (ret) - goto err_resetc_assert; + goto err_clk_put_all; ret = of_platform_populate(np, NULL, NULL, dev); if (ret) - goto err_clk_put; + goto err_clk_disable; pm_runtime_set_active(dev); pm_runtime_enable(dev); @@ -82,8 +82,9 @@ static int dwc3_of_simple_probe(struct platform_device *pdev) return 0; -err_clk_put: +err_clk_disable: clk_bulk_disable_unprepare(simple->num_clocks, simple->clks); +err_clk_put_all: clk_bulk_put_all(simple->num_clocks, simple->clks); err_resetc_assert: -- 2.25.1

4 days, 20 hours

2
1
0 0

[PATCH v5] drm/i915/gem: Zero-initialize the eb.vma array in i915_gem_do_execbuffer

by Krzysztof Niemiec

Initialize the eb.vma array with values of 0 when the eb structure is first set up. In particular, this sets the eb->vma[i].vma pointers to NULL, simplifying cleanup and getting rid of the bug described below. During the execution of eb_lookup_vmas(), the eb->vma array is successively filled up with struct eb_vma objects. This process includes calling eb_add_vma(), which might fail; however, even in the event of failure, eb->vma[i].vma is set for the currently processed buffer. If eb_add_vma() fails, eb_lookup_vmas() returns with an error, which prompts a call to eb_release_vmas() to clean up the mess. Since eb_lookup_vmas() might fail during processing any (possibly not first) buffer, eb_release_vmas() checks whether a buffer's vma is NULL to know at what point did the lookup function fail. In eb_lookup_vmas(), eb->vma[i].vma is set to NULL if either the helper function eb_lookup_vma() or eb_validate_vma() fails. eb->vma[i+1].vma is set to NULL in case i915_gem_object_userptr_submit_init() fails; the current one needs to be cleaned up by eb_release_vmas() at this point, so the next one is set. If eb_add_vma() fails, neither the current nor the next vma is set to NULL, which is a source of a NULL deref bug described in the issue linked in the Closes tag. When entering eb_lookup_vmas(), the vma pointers are set to the slab poison value, instead of NULL. This doesn't matter for the actual lookup, since it gets overwritten anyway, however the eb_release_vmas() function only recognizes NULL as the stopping value, hence the pointers are being set to NULL as they go in case of intermediate failure. This patch changes the approach to filling them all with NULL at the start instead, rather than handling that manually during failure. Closes: https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/15062 Fixes: 544460c33821 ("drm/i915: Multi-BB execbuf") Reported-by: Gangmin Kim <km.kim1503(a)gmail.com> Cc: <stable(a)vger.kernel.org> # 5.16.x Signed-off-by: Krzysztof Niemiec <krzysztof.niemiec(a)intel.com> Reviewed-by: Janusz Krzysztofik <janusz.krzysztofik(a)linux.intel.com> Reviewed-by: Krzysztof Karas <krzysztof.karas(a)intel.com> Reviewed-by: Andi Shyti <andi.shyti(a)linux.intel.com> --- I messed up the continuity in previous revisions; the original patch was sent as [1], and the first revision (which I didn't mark as v2 due to the title change) was sent as [2]. This is the full current changelog: v5: - improve style and fix nits in commit log (Andi) - fix typos and style in the code and comments (Andi) - set args->buffer_count + 1 values to 0 instead of just args->buffer_count (Andi) v4: - delete an empty line (Janusz), reword the comment a bit (Krzysztof, Janusz) v3: - use memset() to fill the entire eb.vma array with zeros instead of looping through the elements (Janusz) - add a comment clarifying the mechanism of the initial allocation (Janusz) - change the commit log again, including title - rearrange the tags to keep checkpatch happy v2: - set the eb->vma[i].vma pointers to NULL during setup instead of ad-hoc at failure (Janusz) - romanize the reporter's name (Andi, offline) - change the commit log, including title [1] https://patchwork.freedesktop.org/series/156832/ [2] https://patchwork.freedesktop.org/series/158036/ .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 37 +++++++++---------- 1 file changed, 17 insertions(+), 20 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index b057c2fa03a4..d49e96f9be51 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -951,13 +951,13 @@ static int eb_lookup_vmas(struct i915_execbuffer *eb) vma = eb_lookup_vma(eb, eb->exec[i].handle); if (IS_ERR(vma)) { err = PTR_ERR(vma); - goto err; + return err; } err = eb_validate_vma(eb, &eb->exec[i], vma); if (unlikely(err)) { i915_vma_put(vma); - goto err; + return err; } err = eb_add_vma(eb, &current_batch, i, vma); @@ -966,19 +966,8 @@ static int eb_lookup_vmas(struct i915_execbuffer *eb) if (i915_gem_object_is_userptr(vma->obj)) { err = i915_gem_object_userptr_submit_init(vma->obj); - if (err) { - if (i + 1 < eb->buffer_count) { - /* - * Execbuffer code expects last vma entry to be NULL, - * since we already initialized this entry, - * set the next value to NULL or we mess up - * cleanup handling. - */ - eb->vma[i + 1].vma = NULL; - } - + if (err) return err; - } eb->vma[i].flags |= __EXEC_OBJECT_USERPTR_INIT; eb->args->flags |= __EXEC_USERPTR_USED; @@ -986,10 +975,6 @@ static int eb_lookup_vmas(struct i915_execbuffer *eb) } return 0; - -err: - eb->vma[i].vma = NULL; - return err; } static int eb_lock_vmas(struct i915_execbuffer *eb) @@ -3375,7 +3360,8 @@ i915_gem_do_execbuffer(struct drm_device *dev, eb.exec = exec; eb.vma = (struct eb_vma *)(exec + args->buffer_count + 1); - eb.vma[0].vma = NULL; + memset(eb.vma, 0, (args->buffer_count + 1) * sizeof(struct eb_vma)); + eb.batch_pool = NULL; eb.invalid_flags = __EXEC_OBJECT_UNKNOWN_FLAGS; @@ -3584,7 +3570,18 @@ i915_gem_execbuffer2_ioctl(struct drm_device *dev, void *data, if (err) return err; - /* Allocate extra slots for use by the command parser */ + /* + * Allocate extra slots for use by the command parser. + * + * Note that this allocation handles two different arrays (the + * exec2_list array, and the eventual eb.vma array introduced in + * i915_gem_do_execbuffer()), that reside in virtually contiguous + * memory. Also note that the allocation intentionally doesn't fill the + * area with zeros, because the exec2_list part doesn't need to be, as + * it's immediately overwritten by user data a few lines below. + * However, the eb.vma part is explicitly zeroed later in + * i915_gem_do_execbuffer(). + */ exec2_list = kvmalloc_array(count + 2, eb_element_size(), __GFP_NOWARN | GFP_KERNEL); if (exec2_list == NULL) { -- 2.45.2

4 days, 20 hours

2
1
0 0

+ mm-memory-failure-fix-missing-mf_stats-count-in-hugetlb-poison.patch added to mm-hotfixes-unstable branch

by Andrew Morton

The patch titled Subject: mm/memory-failure: fix missing ->mf_stats count in hugetlb poison has been added to the -mm mm-hotfixes-unstable branch. Its filename is mm-memory-failure-fix-missing-mf_stats-count-in-hugetlb-poison.patch This patch will shortly appear at https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patche… This patch will later appear in the mm-hotfixes-unstable branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/process/submit-checklist.rst when testing your code *** The -mm tree is included into linux-next via various branches at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm and is updated there most days ------------------------------------------------------ From: Jane Chu <jane.chu(a)oracle.com> Subject: mm/memory-failure: fix missing ->mf_stats count in hugetlb poison Date: Tue, 16 Dec 2025 14:56:21 -0700 When a newly poisoned subpage ends up in an already poisoned hugetlb folio, 'num_poisoned_pages' is incremented, but the per node ->mf_stats is not. Fix the inconsistency by designating action_result() to update them both. Link: https://lkml.kernel.org/r/20251216215621.920093-1-jane.chu@oracle.com Fixes: 18f41fa616ee4 ("mm: memory-failure: bump memory failure stats to pglist_data") Signed-off-by: Jane Chu <jane.chu(a)oracle.com> Cc: David Rientjes <rientjes(a)google.com> Cc: Jiaqi Yan <jiaqiyan(a)google.com> Cc: Liam Howlett <liam.howlett(a)oracle.com> Cc: Lorenzo Stoakes <lorenzo.stoakes(a)oracle.com> Cc: Miaohe Lin <linmiaohe(a)huawei.com> Cc: Michal Hocko <mhocko(a)suse.com> Cc: Mike Rapoport <rppt(a)kernel.org> Cc: Muchun Song <muchun.song(a)linux.dev> Cc: Oscar Salvador <osalvador(a)suse.de> Cc: Suren Baghdasaryan <surenb(a)google.com> Cc: William Roche <william.roche(a)oracle.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- include/linux/hugetlb.h | 4 ++-- include/linux/mm.h | 4 ++-- mm/hugetlb.c | 4 ++-- mm/memory-failure.c | 22 +++++++++++++--------- 4 files changed, 19 insertions(+), 15 deletions(-) --- a/include/linux/hugetlb.h~mm-memory-failure-fix-missing-mf_stats-count-in-hugetlb-poison +++ a/include/linux/hugetlb.h @@ -156,7 +156,7 @@ long hugetlb_unreserve_pages(struct inod bool folio_isolate_hugetlb(struct folio *folio, struct list_head *list); int get_hwpoison_hugetlb_folio(struct folio *folio, bool *hugetlb, bool unpoison); int get_huge_page_for_hwpoison(unsigned long pfn, int flags, - bool *migratable_cleared); + bool *migratable_cleared, bool *samepg); void folio_putback_hugetlb(struct folio *folio); void move_hugetlb_state(struct folio *old_folio, struct folio *new_folio, int reason); void hugetlb_fix_reserve_counts(struct inode *inode); @@ -418,7 +418,7 @@ static inline int get_hwpoison_hugetlb_f } static inline int get_huge_page_for_hwpoison(unsigned long pfn, int flags, - bool *migratable_cleared) + bool *migratable_cleared, bool *samepg) { return 0; } --- a/include/linux/mm.h~mm-memory-failure-fix-missing-mf_stats-count-in-hugetlb-poison +++ a/include/linux/mm.h @@ -4351,7 +4351,7 @@ extern int soft_offline_page(unsigned lo extern const struct attribute_group memory_failure_attr_group; extern void memory_failure_queue(unsigned long pfn, int flags); extern int __get_huge_page_for_hwpoison(unsigned long pfn, int flags, - bool *migratable_cleared); + bool *migratable_cleared, bool *samepg); void num_poisoned_pages_inc(unsigned long pfn); void num_poisoned_pages_sub(unsigned long pfn, long i); #else @@ -4360,7 +4360,7 @@ static inline void memory_failure_queue( } static inline int __get_huge_page_for_hwpoison(unsigned long pfn, int flags, - bool *migratable_cleared) + bool *migratable_cleared, bool *samepg) { return 0; } --- a/mm/hugetlb.c~mm-memory-failure-fix-missing-mf_stats-count-in-hugetlb-poison +++ a/mm/hugetlb.c @@ -7132,12 +7132,12 @@ int get_hwpoison_hugetlb_folio(struct fo } int get_huge_page_for_hwpoison(unsigned long pfn, int flags, - bool *migratable_cleared) + bool *migratable_cleared, bool *samepg) { int ret; spin_lock_irq(&hugetlb_lock); - ret = __get_huge_page_for_hwpoison(pfn, flags, migratable_cleared); + ret = __get_huge_page_for_hwpoison(pfn, flags, migratable_cleared, samepg); spin_unlock_irq(&hugetlb_lock); return ret; } --- a/mm/memory-failure.c~mm-memory-failure-fix-missing-mf_stats-count-in-hugetlb-poison +++ a/mm/memory-failure.c @@ -1883,7 +1883,8 @@ static unsigned long __folio_free_raw_hw return count; } -static int folio_set_hugetlb_hwpoison(struct folio *folio, struct page *page) +static int folio_set_hugetlb_hwpoison(struct folio *folio, struct page *page, + bool *samepg) { struct llist_head *head; struct raw_hwp_page *raw_hwp; @@ -1899,17 +1900,16 @@ static int folio_set_hugetlb_hwpoison(st return -EHWPOISON; head = raw_hwp_list_head(folio); llist_for_each_entry(p, head->first, node) { - if (p->page == page) + if (p->page == page) { + *samepg = true; return -EHWPOISON; + } } raw_hwp = kmalloc(sizeof(struct raw_hwp_page), GFP_ATOMIC); if (raw_hwp) { raw_hwp->page = page; llist_add(&raw_hwp->node, head); - /* the first error event will be counted in action_result(). */ - if (ret) - num_poisoned_pages_inc(page_to_pfn(page)); } else { /* * Failed to save raw error info. We no longer trace all @@ -1966,7 +1966,7 @@ void folio_clear_hugetlb_hwpoison(struct * -EHWPOISON - the hugepage is already hwpoisoned */ int __get_huge_page_for_hwpoison(unsigned long pfn, int flags, - bool *migratable_cleared) + bool *migratable_cleared, bool *samepg) { struct page *page = pfn_to_page(pfn); struct folio *folio = page_folio(page); @@ -1991,7 +1991,7 @@ int __get_huge_page_for_hwpoison(unsigne goto out; } - if (folio_set_hugetlb_hwpoison(folio, page)) { + if (folio_set_hugetlb_hwpoison(folio, page, samepg)) { ret = -EHWPOISON; goto out; } @@ -2024,11 +2024,12 @@ static int try_memory_failure_hugetlb(un struct page *p = pfn_to_page(pfn); struct folio *folio; unsigned long page_flags; + bool samepg = false; bool migratable_cleared = false; *hugetlb = 1; retry: - res = get_huge_page_for_hwpoison(pfn, flags, &migratable_cleared); + res = get_huge_page_for_hwpoison(pfn, flags, &migratable_cleared, &samepg); if (res == 2) { /* fallback to normal page handling */ *hugetlb = 0; return 0; @@ -2037,7 +2038,10 @@ retry: folio = page_folio(p); res = kill_accessing_process(current, folio_pfn(folio), flags); } - action_result(pfn, MF_MSG_ALREADY_POISONED, MF_FAILED); + if (samepg) + action_result(pfn, MF_MSG_ALREADY_POISONED, MF_FAILED); + else + action_result(pfn, MF_MSG_HUGE, MF_FAILED); return res; } else if (res == -EBUSY) { if (!(flags & MF_NO_RETRY)) { _ Patches currently in -mm which might be from jane.chu(a)oracle.com are mm-memory-failure-fix-missing-mf_stats-count-in-hugetlb-poison.patch

4 days, 21 hours

1
0
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror