January 2024 - Kernel - mailweb.openeuler.org

[PATCH OLK-6.6 V1 00/12] infrastructure for scheduler bpf
by Lu Jialin 09 Jan '24

09 Jan '24

From: Guan Jing <guanjing6(a)huawei.com> infrastructure for scheduler bpf Guan Jing (12): sched: programmable: Introduce bpf sched sched: programmable: Add a tag for the task sched: programmable: Add a tag for the task group sched: programmable: Add user interface of task group tag sched: programmable: Add user interface of task tag sched: basic infrastructure for scheduler bpf sched: introduce bpf_sched_enable() libbpf: add support for scheduler bpf programs bpftool: recognize scheduler programs sched: Add helper functions to get cpu statistics sched: programmable: Add hook in select_task_rq_fair() sched: programmable: Add hook in can_migrate_task() arch/arm64/configs/openeuler_defconfig | 1 + arch/x86/configs/openeuler_defconfig | 1 + fs/proc/base.c | 64 ++++++++++++++ include/linux/bpf_sched.h | 50 +++++++++++ include/linux/bpf_types.h | 4 + include/linux/sched.h | 38 ++++++++ include/linux/sched_hook_defs.h | 4 + include/uapi/linux/bpf.h | 9 ++ init/init_task.c | 3 + kernel/bpf/Kconfig | 12 +++ kernel/bpf/btf.c | 1 + kernel/bpf/syscall.c | 27 ++++++ kernel/bpf/trampoline.c | 1 + kernel/bpf/verifier.c | 11 ++- kernel/sched/bpf_sched.c | 102 +++++++++++++++++++++ kernel/sched/build_utility.c | 4 + kernel/sched/core.c | 118 +++++++++++++++++++++++++ kernel/sched/fair.c | 45 ++++++++++ kernel/sched/sched.h | 10 +++ scripts/bpf_doc.py | 4 + tools/include/uapi/linux/bpf.h | 9 ++ tools/lib/bpf/bpf.c | 1 + tools/lib/bpf/libbpf.c | 23 ++++- tools/lib/bpf/libbpf.h | 2 + tools/lib/bpf/libbpf.map | 1 + 25 files changed, 543 insertions(+), 2 deletions(-) create mode 100644 include/linux/bpf_sched.h create mode 100644 include/linux/sched_hook_defs.h create mode 100644 kernel/sched/bpf_sched.c -- 2.34.1

1 12

[PATCH OLK-6.6 V1 00/12] infrastructure for scheduler bpf
by Guan Jing 09 Jan '24

09 Jan '24

infrastructure for scheduler bpf Guan Jing (12): sched: programmable: Introduce bpf sched sched: programmable: Add a tag for the task sched: programmable: Add a tag for the task group sched: programmable: Add user interface of task group tag sched: programmable: Add user interface of task tag sched: basic infrastructure for scheduler bpf sched: introduce bpf_sched_enable() libbpf: add support for scheduler bpf programs bpftool: recognize scheduler programs sched: Add helper functions to get cpu statistics sched: programmable: Add hook in select_task_rq_fair() sched: programmable: Add hook in can_migrate_task() arch/arm64/configs/openeuler_defconfig | 1 + arch/x86/configs/openeuler_defconfig | 1 + fs/proc/base.c | 64 ++++++++++++++ include/linux/bpf_sched.h | 50 +++++++++++ include/linux/bpf_types.h | 4 + include/linux/sched.h | 38 ++++++++ include/linux/sched_hook_defs.h | 4 + include/uapi/linux/bpf.h | 9 ++ init/init_task.c | 3 + kernel/bpf/Kconfig | 12 +++ kernel/bpf/btf.c | 1 + kernel/bpf/syscall.c | 27 ++++++ kernel/bpf/trampoline.c | 1 + kernel/bpf/verifier.c | 11 ++- kernel/sched/bpf_sched.c | 102 +++++++++++++++++++++ kernel/sched/build_utility.c | 4 + kernel/sched/core.c | 118 +++++++++++++++++++++++++ kernel/sched/fair.c | 45 ++++++++++ kernel/sched/sched.h | 10 +++ scripts/bpf_doc.py | 4 + tools/include/uapi/linux/bpf.h | 9 ++ tools/lib/bpf/bpf.c | 1 + tools/lib/bpf/libbpf.c | 23 ++++- tools/lib/bpf/libbpf.h | 2 + tools/lib/bpf/libbpf.map | 1 + 25 files changed, 543 insertions(+), 2 deletions(-) create mode 100644 include/linux/bpf_sched.h create mode 100644 include/linux/sched_hook_defs.h create mode 100644 kernel/sched/bpf_sched.c -- 2.34.1

1 12

[PATCH OLK-6.6 0/1] iommu: set CONFIG_SMMU_BYPASS_DEV=y
by zhangnaichuan 09 Jan '24

09 Jan '24

set CONFIG_SMMU_BYPASS_DEV=y in openeuler_defconfig arch/arm64/configs/openeuler_defconfig | 1 + 1 file changed, 1 insertion(+) -- 2.33.0

2 2

[PATCH V3 OLK-6.6] fs/dirty_pages: dump the number of dirty pages for each inode
by Zizhi Wo 09 Jan '24

09 Jan '24

hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I8OHM4?from=project-issue CVE: NA ---------------------------------------- In order to analyse the IO performance when using buffer IO, it's useful to obtain the number of dirty pages for each inode in the filesystem. This feature depends on 'CONFIG_DIRTY_PAGES'. It creates 3 interfaces by using procfs. /proc/dirty/buffer_size for buffer allocation and release, /proc/dirty/page_threshold to filter result and /proc/dirty/dirty_list to get dirty pages. Signed-off-by: Zizhi Wo <wozizhi(a)huawei.com> --- arch/arm64/configs/openeuler_defconfig | 1 + arch/x86/configs/openeuler_defconfig | 1 + fs/Kconfig | 13 + fs/Makefile | 1 + fs/dirty_pages.c | 355 +++++++++++++++++++++++++ 5 files changed, 371 insertions(+) create mode 100644 fs/dirty_pages.c diff --git a/arch/arm64/configs/openeuler_defconfig b/arch/arm64/configs/openeuler_defconfig index 66737e8c8673..677b4cc4668a 100644 --- a/arch/arm64/configs/openeuler_defconfig +++ b/arch/arm64/configs/openeuler_defconfig @@ -6924,6 +6924,7 @@ CONFIG_PROC_PAGE_MONITOR=y CONFIG_PROC_CHILDREN=y CONFIG_KERNFS=y CONFIG_SYSFS=y +CONFIG_DIRTY_PAGES=y CONFIG_TMPFS=y CONFIG_TMPFS_POSIX_ACL=y CONFIG_TMPFS_XATTR=y diff --git a/arch/x86/configs/openeuler_defconfig b/arch/x86/configs/openeuler_defconfig index fcbef6c06587..c16cdb7b2596 100644 --- a/arch/x86/configs/openeuler_defconfig +++ b/arch/x86/configs/openeuler_defconfig @@ -8120,6 +8120,7 @@ CONFIG_PROC_PID_ARCH_STATUS=y CONFIG_PROC_CPU_RESCTRL=y CONFIG_KERNFS=y CONFIG_SYSFS=y +CONFIG_DIRTY_PAGES=y CONFIG_TMPFS=y CONFIG_TMPFS_POSIX_ACL=y CONFIG_TMPFS_XATTR=y diff --git a/fs/Kconfig b/fs/Kconfig index 7c9d45ba9b0f..6dc83223c61b 100644 --- a/fs/Kconfig +++ b/fs/Kconfig @@ -170,6 +170,19 @@ source "fs/proc/Kconfig" source "fs/kernfs/Kconfig" source "fs/sysfs/Kconfig" +config DIRTY_PAGES + bool "Dumps the number of dirty pages of each file" + depends on PROC_FS + default y + help + This config supports the rendering of dirty page data to the user, + which may be useful to analyze the IO performance when using buffer + IO. + + It create 3 interfaces by using procfs. /proc/dirty/buffer_size for + buffer allocation and release; /proc/dirty/page_threshold to filter + result; /proc/dirty/dirty_list to get dirty pages. + config TMPFS bool "Tmpfs virtual memory file system support (former shm fs)" depends on SHMEM diff --git a/fs/Makefile b/fs/Makefile index bb6164200a39..cc4735c23c32 100644 --- a/fs/Makefile +++ b/fs/Makefile @@ -44,6 +44,7 @@ obj-$(CONFIG_FS_POSIX_ACL) += posix_acl.o obj-$(CONFIG_NFS_COMMON) += nfs_common/ obj-$(CONFIG_COREDUMP) += coredump.o obj-$(CONFIG_SYSCTL) += drop_caches.o sysctls.o +obj-$(CONFIG_DIRTY_PAGES) += dirty_pages.o obj-$(CONFIG_FHANDLE) += fhandle.o obj-$(CONFIG_CGROUP_FILES) += filescontrol.o diff --git a/fs/dirty_pages.c b/fs/dirty_pages.c new file mode 100644 index 000000000000..d0aea022b68b --- /dev/null +++ b/fs/dirty_pages.c @@ -0,0 +1,355 @@ +// SPDX-License-Identifier: GPL-2.0 +#include <linux/proc_fs.h> +#include <linux/seq_file.h> +#include <linux/uaccess.h> +#include <linux/pagemap.h> +#include <linux/pagevec.h> +#include <linux/fs.h> +#include <linux/mm.h> +#include <linux/slab.h> +#include <linux/sched.h> +#include <linux/proc_fs.h> +#include <linux/kdev_t.h> +#include <linux/vmalloc.h> +#include <linux/module.h> +#include <linux/kernel.h> +#include <linux/init.h> +#include "internal.h" + +#define DIRTY_ROOT "dirty" /* proc root directory name */ +#define DIRTY_BUFFER_SIZE "buffer_size" /* proc file to control buffer */ +#define DIRTY_PAGES "dirty_list" /* proc file to obtain diry pages */ +#define DIRTY_LIMIT "page_threshold" /* proc file to filter result */ +#define MAX_BUF_NUM 102400 /* max buffer num in number of pages */ +#define TERM_STR "terminated\n" /* tail string for overflowed buffer */ +#define UNKNOW_STR "unknown" /* for unknown file name or fs type */ + +static char *buf_dirty; /* buffer to store number of dirty pages */ +static unsigned long buf_size; /* size of buffer in bytes */ +static unsigned long buf_num; /* size of buffer in number of pages */ +static unsigned long buf_limit; /* filter threshold of dirty pages */ +unsigned long lock_word; /* for exclusive access to buffer */ + +static bool dirty_pages_lock(void) +{ + if (xchg(&lock_word, 1) == 1) + return false; + return true; +} + +static void dirty_pages_unlock(void) +{ + lock_word = 0; +} + +static unsigned long dump_dirtypages_inode(struct inode *inode) +{ + XA_STATE(xas, &inode->i_mapping->i_pages, 0); + unsigned long pages = 0; + void *page; + + if (!mapping_tagged(inode->i_mapping, PAGECACHE_TAG_DIRTY)) + return 0; + + xas_lock_irq(&xas); + xas_for_each_marked(&xas, page, (pgoff_t)-1, PAGECACHE_TAG_DIRTY) { + if (++pages % XA_CHECK_SCHED) + continue; + + xas_pause(&xas); + xas_unlock_irq(&xas); + cond_resched(); + xas_lock_irq(&xas); + } + xas_unlock_irq(&xas); + + return pages; +} + +static char *inode_filename(struct inode *inode) +{ + char *path; + static char buf[PATH_MAX]; + struct dentry *dentry; + + dentry = d_find_alias(inode); + if (!dentry) + return UNKNOW_STR; + + path = dentry_path_raw(dentry, buf, PATH_MAX); + dput(dentry); + + return IS_ERR_OR_NULL(path) ? UNKNOW_STR : path; +} + +static const char *sb_fstype(struct super_block *sb) +{ + if (sb->s_type && sb->s_type->name) + return sb->s_type->name; + else + return UNKNOW_STR; +} + +static inline bool is_sb_writable(struct super_block *sb) +{ + if (sb_rdonly(sb)) + return false; + + if (sb->s_writers.frozen == SB_FREEZE_COMPLETE) + return false; + + return true; +} + +/* + * dump_dirtypages_sb - dump the dirty pages of each inode in the sb + * @sb the super block + * @m the seq_file witch is initialized in proc_dpages_open + * + * For each inode in the sb, call dump_dirtypages_pages to get the number + * of dirty pages. And use seq_printf to store the result in the buffer + * if it's not less than the threshold. The inode in unusual state will + * be skipped. + */ +static void dump_dirtypages_sb(struct super_block *sb, struct seq_file *m) +{ + struct inode *inode, *toput_inode = NULL; + unsigned long limit; + + if (!is_sb_writable(sb) || !sb->s_bdev) + return; + + limit = READ_ONCE(buf_limit); + spin_lock(&sb->s_inode_list_lock); + list_for_each_entry(inode, &sb->s_inodes, i_sb_list) { + unsigned long dirty_pages; + + spin_lock(&inode->i_lock); + if (inode->i_state & (I_FREEING|I_WILL_FREE|I_NEW)) { + spin_unlock(&inode->i_lock); + continue; + } + + __iget(inode); + spin_unlock(&inode->i_lock); + spin_unlock(&sb->s_inode_list_lock); + + dirty_pages = dump_dirtypages_inode(inode); + if (!dirty_pages || dirty_pages < limit) + goto skip; + + seq_printf(m, "FSType: %s, Dev ID: %u(%u:%u) ino %lu, dirty pages %lu, path %s\n", + sb_fstype(sb), sb->s_dev, MAJOR(sb->s_dev), + MINOR(sb->s_dev), inode->i_ino, dirty_pages, + inode_filename(inode)); + + if (seq_has_overflowed(m)) { + /* keep size > count to avoid overflow in seq_read_iter() */ + m->size += strlen(TERM_STR) + 1; + seq_puts(m, TERM_STR); + iput(inode); + iput(toput_inode); + return; + } +skip: + iput(toput_inode); + toput_inode = inode; + cond_resched(); + spin_lock(&sb->s_inode_list_lock); + } + spin_unlock(&sb->s_inode_list_lock); + iput(toput_inode); +} + +static int proc_dpages_show(struct seq_file *m, void *v) +{ + iterate_supers((void *)dump_dirtypages_sb, (void *)m); + return 0; +} + +static int proc_dpages_open(struct inode *inode, struct file *filp) +{ + int ret; + struct seq_file *m; + + if (!dirty_pages_lock()) + return -EBUSY; + + if (buf_dirty == NULL || buf_size == 0) { + pr_warn("please allocate buffer before getting dirty pages\n"); + dirty_pages_unlock(); + return -ENOMEM; + } + + ret = single_open(filp, proc_dpages_show, NULL); + if (ret) { + dirty_pages_unlock(); + return ret; + } + + m = (struct seq_file *)filp->private_data; + memset(buf_dirty, 0, buf_size); + m->size = buf_size - (strlen(TERM_STR) + 1); + m->buf = buf_dirty; + + return 0; +} + +static int proc_dpages_release(struct inode *inode, struct file *file) +{ + struct seq_file *m = file->private_data; + + /* we don't want to free the buf */ + m->buf = NULL; + dirty_pages_unlock(); + return single_release(inode, file); +} + +static void free_buf_dirty(void) +{ + if (!buf_dirty) + return; + + vfree(buf_dirty); + buf_dirty = NULL; + buf_size = 0; + WRITE_ONCE(buf_num, 0); +} + +static int proc_buffer_show(struct seq_file *m, void *v) +{ + seq_printf(m, "%lu\n", READ_ONCE(buf_num)); + return 0; +} + +static int proc_buffer_open(struct inode *inode, struct file *filp) +{ + int ret; + + if (((filp->f_flags & O_ACCMODE) != O_RDONLY) && !dirty_pages_lock()) + return -EBUSY; + + ret = single_open(filp, proc_buffer_show, NULL); + if (ret) + dirty_pages_unlock(); + + return ret; +} + +static ssize_t proc_buffer_write(struct file *filp, const char __user *buf, + size_t count, loff_t *offp) +{ + int ret; + unsigned long new_buf_num, new_buf_size; + char *new_buf_dirty; + + ret = kstrtoul_from_user(buf, count, 10, &new_buf_num); + if (ret != 0 || new_buf_num > MAX_BUF_NUM) + return -EINVAL; + + if (new_buf_num == buf_num) + return count; + + if (new_buf_num == 0) { + free_buf_dirty(); + return count; + } + + new_buf_size = PAGE_SIZE * new_buf_num; + new_buf_dirty = vzalloc(new_buf_size); + if (!new_buf_dirty) + return -ENOMEM; + + free_buf_dirty(); + buf_dirty = new_buf_dirty; + buf_size = new_buf_size; + WRITE_ONCE(buf_num, new_buf_num); + + return count; +} + +static int proc_buffer_release(struct inode *inode, struct file *filp) +{ + if ((filp->f_flags & O_ACCMODE) != O_RDONLY) + dirty_pages_unlock(); + return single_release(inode, filp); +} + +static int proc_limit_show(struct seq_file *m, void *v) +{ + seq_printf(m, "%lu\n", READ_ONCE(buf_limit)); + return 0; +} + +static int proc_limit_open(struct inode *inode, struct file *filp) +{ + return single_open(filp, proc_limit_show, NULL); +} + +static ssize_t proc_limit_write(struct file *filp, const char __user *buf, + size_t count, loff_t *offp) +{ + int ret; + unsigned long limit; + + ret = kstrtoul_from_user(buf, count, 10, &limit); + if (ret != 0) + return ret; + + WRITE_ONCE(buf_limit, limit); + + return count; +} + +static const struct proc_ops proc_dpages_operations = { + .proc_open = proc_dpages_open, + .proc_read = seq_read, + .proc_release = proc_dpages_release, +}; + +static const struct proc_ops proc_buffer_operations = { + .proc_open = proc_buffer_open, + .proc_read = seq_read, + .proc_write = proc_buffer_write, + .proc_lseek = seq_lseek, + .proc_release = proc_buffer_release, +}; + +static const struct proc_ops proc_limit_operations = { + .proc_open = proc_limit_open, + .proc_read = seq_read, + .proc_write = proc_limit_write, + .proc_lseek = seq_lseek, + .proc_release = single_release, +}; + +static int __init dpages_proc_init(void) +{ + struct proc_dir_entry *dirty_dir; + + dirty_dir = proc_mkdir(DIRTY_ROOT, NULL); + if (!dirty_dir) + return -ENOMEM; + + if (!proc_create(DIRTY_PAGES, 0440, dirty_dir, &proc_dpages_operations)) + goto fail_pages; + + if (!proc_create(DIRTY_BUFFER_SIZE, 0640, dirty_dir, + &proc_buffer_operations)) + goto fail_switch; + + if (!proc_create(DIRTY_LIMIT, 0640, dirty_dir, &proc_limit_operations)) + goto fail_limit; + + return 0; + +fail_limit: + remove_proc_entry(DIRTY_BUFFER_SIZE, dirty_dir); +fail_switch: + remove_proc_entry(DIRTY_PAGES, dirty_dir); +fail_pages: + remove_proc_entry(DIRTY_ROOT, NULL); + return -ENOMEM; +} + +subsys_initcall(dpages_proc_init); -- 2.39.2

1 0

[PATCH OLK-6.6] fs/dcache.c: avoid panic while lockref of dentry overflow
by Zizhi Wo 09 Jan '24

09 Jan '24

From: yangerkun <yangerkun(a)huawei.com> hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I8JDP6 --------------------------- We use lockref for dentry reference without notice that so many negative dentry under one dir can lead to overflow of lockref. This can lead to system crash if we do this under root dir. Since there is not a perfect solution, we just limit max number of dentry count up to INT_MAX / 2. Also, it will cost a lot of time from INT_MAX / 2 to INT_MAX, so we no need to do this under protection of dentry lock. Also, we limit the FILES_MAX to INT_MAX / 2, since a lot open for same file can lead to overflow too. Changelog: v1->v2: add a function to do check / add a Macro to mean INT_MAX / 2 Signed-off-by: yangerkun <yangerkun(a)huawei.com> Signed-off-by: Zizhi Wo <wozizhi(a)huawei.com> --- fs/dcache.c | 32 ++++++++++++++++++++++++++++---- fs/filescontrol.c | 2 +- include/linux/fs.h | 3 +++ 3 files changed, 32 insertions(+), 5 deletions(-) diff --git a/fs/dcache.c b/fs/dcache.c index 25ac74d30bff..7cdd8bbedd1c 100644 --- a/fs/dcache.c +++ b/fs/dcache.c @@ -1834,6 +1834,18 @@ static struct dentry *__d_alloc(struct super_block *sb, const struct qstr *name) return dentry; } +static inline bool d_forbid_overflow(struct dentry *dentry) +{ + if (unlikely(d_count(dentry) >= D_COUNT_MAX)) { + shrink_dcache_parent(dentry); + + if (d_count(dentry) >= D_COUNT_MAX) + return false; + } + + return true; +} + /** * d_alloc - allocate a dcache entry * @parent: parent of entry to allocate @@ -1845,9 +1857,15 @@ static struct dentry *__d_alloc(struct super_block *sb, const struct qstr *name) */ struct dentry *d_alloc(struct dentry * parent, const struct qstr *name) { - struct dentry *dentry = __d_alloc(parent->d_sb, name); + struct dentry *dentry = NULL; + + if (unlikely(!d_forbid_overflow(parent))) + goto out; + + dentry = __d_alloc(parent->d_sb, name); if (!dentry) - return NULL; + goto out; + spin_lock(&parent->d_lock); /* * don't need child lock because it is not subject @@ -1857,7 +1875,7 @@ struct dentry *d_alloc(struct dentry * parent, const struct qstr *name) dentry->d_parent = parent; list_add(&dentry->d_child, &parent->d_subdirs); spin_unlock(&parent->d_lock); - +out: return dentry; } EXPORT_SYMBOL(d_alloc); @@ -1870,11 +1888,17 @@ EXPORT_SYMBOL(d_alloc_anon); struct dentry *d_alloc_cursor(struct dentry * parent) { - struct dentry *dentry = d_alloc_anon(parent->d_sb); + struct dentry *dentry = NULL; + + if (unlikely(!d_forbid_overflow(parent))) + goto out; + + dentry = d_alloc_anon(parent->d_sb); if (dentry) { dentry->d_flags |= DCACHE_DENTRY_CURSOR; dentry->d_parent = dget(parent); } +out: return dentry; } diff --git a/fs/filescontrol.c b/fs/filescontrol.c index 610f13a36962..0947566b9119 100644 --- a/fs/filescontrol.c +++ b/fs/filescontrol.c @@ -27,7 +27,7 @@ #include <linux/sched/signal.h> #include <linux/module.h> -#define FILES_MAX ULONG_MAX +#define FILES_MAX D_COUNT_MAX #define FILES_MAX_STR "max" static bool no_acct; diff --git a/include/linux/fs.h b/include/linux/fs.h index fb5accebdcdf..67c2ac09e700 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -47,6 +47,9 @@ #include <asm/byteorder.h> #include <uapi/linux/fs.h> +#define D_COUNT_MAX (INT_MAX / 2) + + struct backing_dev_info; struct bdi_writeback; struct bio; -- 2.39.2

1 0

[PATCH OLK-6.6] hfs: fix null-ptr-deref in hfs_find_init()
by Long Li 09 Jan '24

09 Jan '24

From: Yang Yingliang <yangyingliang(a)huawei.com> hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I8URA4 CVE: CVE-2018-12928 -------------------------------- It will cause a null-ptr-deref in hfs_find_init() in fuzz test. [ 107.092729] hfs: continuing without an alternate MDB [ 107.097632] general protection fault, probably for non-canonical address 0xdffffc0000000008: 0000 [#1] SMP KASAN PTI [ 107.104679] KASAN: null-ptr-deref in range [0x0000000000000040-0x0000000000000047] [ 107.109100] CPU: 0 PID: 379 Comm: hfs_inject Not tainted 5.7.0-rc7-00001-g24627f5f2973 #897 [ 107.114142] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014 [ 107.121095] RIP: 0010:hfs_find_init+0x72/0x170 [ 107.123609] Code: c1 ea 03 80 3c 02 00 0f 85 e6 00 00 00 4c 8d 65 40 48 c7 43 18 00 00 00 00 48 b8 00 00 00 00 00 fc ff df 4c 89 e2 48 c1 ea 03 <0f> b6 04 02 84 c0 74 08 3c 03 0f 8e a5 00 00 00 8b 45 40 be c0 0c [ 107.134660] RSP: 0018:ffff88810291f3f8 EFLAGS: 00010202 [ 107.137897] RAX: dffffc0000000000 RBX: ffff88810291f468 RCX: 1ffff110175cdf05 [ 107.141874] RDX: 0000000000000008 RSI: ffff88810291f468 RDI: ffff88810291f480 [ 107.145844] RBP: 0000000000000000 R08: 0000000000000000 R09: ffffed1020381013 [ 107.149431] R10: ffff88810291f500 R11: ffffed1020381012 R12: 0000000000000040 [ 107.152315] R13: 0000000000000000 R14: ffff888101c0814a R15: ffff88810291f468 [ 107.155464] FS: 00000000009ea880(0000) GS:ffff88810c600000(0000) knlGS:0000000000000000 [ 107.159795] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 107.162987] CR2: 00005605a19dd284 CR3: 0000000103a0c006 CR4: 0000000000020ef0 [ 107.166665] Call Trace: [ 107.167969] ? find_held_lock+0x33/0x1c0 [ 107.169972] hfs_ext_read_extent+0x16b/0xb00 [ 107.172092] ? create_page_buffers+0x14e/0x1b0 [ 107.174303] ? hfs_free_extents+0x280/0x280 [ 107.176437] ? lock_downgrade+0x730/0x730 [ 107.178272] hfs_get_block+0x496/0x8a0 [ 107.179972] block_read_full_page+0x241/0x8d0 [ 107.181971] ? hfs_extend_file+0xae0/0xae0 [ 107.183814] ? end_buffer_async_read_io+0x10/0x10 [ 107.185954] ? add_to_page_cache_lru+0x13f/0x1f0 [ 107.188006] ? add_to_page_cache_locked+0x10/0x10 [ 107.190175] do_read_cache_page+0xc6a/0x1180 [ 107.192096] ? generic_file_read_iter+0x4c0/0x4c0 [ 107.194234] ? hfs_btree_open+0x408/0x1000 [ 107.196068] ? lock_downgrade+0x730/0x730 [ 107.197926] ? wake_bit_function+0x180/0x180 [ 107.199845] ? lockdep_init_map_waits+0x267/0x7c0 [ 107.201895] hfs_btree_open+0x455/0x1000 [ 107.203479] hfs_mdb_get+0x122c/0x1ae8 [ 107.205065] ? hfs_mdb_put+0x350/0x350 [ 107.206590] ? queue_work_node+0x260/0x260 [ 107.208309] ? rcu_read_lock_sched_held+0xa1/0xd0 [ 107.210227] ? lockdep_init_map_waits+0x267/0x7c0 [ 107.212144] ? lockdep_init_map_waits+0x267/0x7c0 [ 107.213979] hfs_fill_super+0x9ba/0x1280 [ 107.215444] ? bdev_name.isra.9+0xf1/0x2b0 [ 107.217028] ? hfs_remount+0x190/0x190 [ 107.218428] ? pointer+0x5da/0x710 [ 107.219745] ? file_dentry_name+0xf0/0xf0 [ 107.221262] ? mount_bdev+0xd1/0x330 [ 107.222592] ? vsnprintf+0x7bd/0x1250 [ 107.224007] ? pointer+0x710/0x710 [ 107.225332] ? down_write+0xe5/0x160 [ 107.226698] ? hfs_remount+0x190/0x190 [ 107.228120] ? snprintf+0x91/0xc0 [ 107.229388] ? vsprintf+0x10/0x10 [ 107.230628] ? sget+0x3af/0x4a0 [ 107.231848] ? hfs_remount+0x190/0x190 [ 107.233300] mount_bdev+0x26e/0x330 [ 107.234611] ? hfs_statfs+0x540/0x540 [ 107.236015] legacy_get_tree+0x101/0x1f0 [ 107.237431] ? security_capable+0x58/0x90 [ 107.238832] vfs_get_tree+0x89/0x2d0 [ 107.240082] ? ns_capable_common+0x5c/0xd0 [ 107.241521] do_mount+0xd8a/0x1720 [ 107.242727] ? lock_downgrade+0x730/0x730 [ 107.244116] ? copy_mount_string+0x20/0x20 [ 107.245557] ? _copy_from_user+0xbe/0x100 [ 107.246967] ? memdup_user+0x47/0x70 [ 107.248212] __x64_sys_mount+0x162/0x1b0 [ 107.249537] do_syscall_64+0xa5/0x4f0 [ 107.250742] entry_SYSCALL_64_after_hwframe+0x49/0xb3 [ 107.252369] RIP: 0033:0x44e8ea [ 107.253360] Code: 48 c7 c1 c0 ff ff ff f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 49 89 ca b8 a5 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 c0 ff ff ff f7 d8 64 89 01 48 [ 107.259240] RSP: 002b:00007ffd910e4c28 EFLAGS: 00000207 ORIG_RAX: 00000000000000a5 [ 107.261668] RAX: ffffffffffffffda RBX: 0000000000400400 RCX: 000000000044e8ea [ 107.263920] RDX: 000000000049321e RSI: 0000000000493222 RDI: 00007ffd910e4d00 [ 107.266177] RBP: 00007ffd910e5d10 R08: 0000000000000000 R09: 000000000000000a [ 107.268451] R10: 0000000000000001 R11: 0000000000000207 R12: 0000000000401c40 [ 107.270721] R13: 0000000000000000 R14: 00000000006ba018 R15: 0000000000000000 [ 107.273025] Modules linked in: [ 107.274029] Dumping ftrace buffer: [ 107.275121] (ftrace buffer empty) [ 107.276370] ---[ end trace c5e0b9d684f3570e ]--- We need check tree in hfs_find_init(). https://lore.kernel.org/linux-fsdevel/20180419024358.GA5215@bombadil.infrad… https://marc.info/?l=linux-fsdevel&m=152406881024567&w=2 References: CVE-2018-12928 Signed-off-by: Yang Yingliang <yangyingliang(a)huawei.com> Signed-off-by: Long Li <leo.lilong(a)huawei.com> --- fs/hfs/bfind.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/fs/hfs/bfind.c b/fs/hfs/bfind.c index ef9498a6e88a..ae44c8fa86ec 100644 --- a/fs/hfs/bfind.c +++ b/fs/hfs/bfind.c @@ -16,6 +16,8 @@ int hfs_find_init(struct hfs_btree *tree, struct hfs_find_data *fd) { void *ptr; + if (!tree) + return -EINVAL; fd->tree = tree; fd->bnode = NULL; ptr = kmalloc(tree->max_key_len * 2 + 4, GFP_KERNEL); -- 2.31.1

1 0

[PATCH OLK-6.6] sched/smt: fix unbalance sched_smt_present dec/inc
by Xia Fukun 09 Jan '24

09 Jan '24

From: Yang Yingliang <yangyingliang(a)huawei.com> hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I8UR4U CVE: NA -------------------------------- If cpuset_cpu_inactive() fails in sched_cpu_deactivate(), the cpu offline failed, but sched_smt_present is decreased before calling sched_cpu_deactivate, it leads unbalance dec/inc, so fix it by increasing sched_smt_present in the error path. Signed-off-by: Yang Yingliang <yangyingliang(a)huawei.com> Signed-off-by: Yu Liao <liaoyu15(a)huawei.com> Signed-off-by: Xia Fukun <xiafukun(a)huawei.com> --- kernel/sched/core.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 7a0997e7e136..adecea9c607d 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -9792,6 +9792,10 @@ int sched_cpu_deactivate(unsigned int cpu) ret = cpuset_cpu_inactive(cpu); if (ret) { balance_push_set(cpu, false); +#ifdef CONFIG_SCHED_SMT + if (cpumask_weight(cpu_smt_mask(cpu)) == 2) + static_branch_inc_cpuslocked(&sched_smt_present); +#endif set_cpu_active(cpu, true); sched_update_numa(cpu, true); return ret; -- 2.34.1

1 0

[PATCH OLK-6.6 0/1] iommu: set CONFIG_SMMU_BYPASS_DEV=y
by zhangnaichuan 09 Jan '24

09 Jan '24

set CONFIG_SMMU_BYPASS_DEV=y in openeuler_defconfig arch/arm64/configs/openeuler_defconfig | 1 + 1 file changed, 1 insertion(+) -- 2.33.0

1 0

[PATCH OLK-6.6 v5 0/5] block: Add config to show info about opening a mounted device for write
by Li Lingfeng 09 Jan '24

09 Jan '24

v1->v2: 1) Replace BLK_DEV_DUMPINFO with BLK_DEV_WRITE_MOUNTED_QUIET, which is logically opposite; 2) Make BLK_DEV_WRITE_MOUNTED_QUIET dependent on BLK_DEV_WRITE_MOUNTED, and decide whether to set bd_writers by BLK_DEV_WRITE_MOUNTED_QUIET; 3) Move bdev_dump_info() into bdev_may_open(); 4) Add the cmdline interface to control whether to show info; 5) Pass the state of "writes blocked" to part0 when mounting a partition. v2->v3: Add restriction for modifying bd_writers. v3->v4: 1) Count bd_writers regardless of whether bdev_allow_write_mounted is set; 2) Expand the meaning of bdev_allow_write_mounted; 3) Add config option to detect writing to part0 while partitions mounted; 4) Add bd_mounters to record the number of times that part0 or partition is mounted; 5) Show info about opening a lower device for write while upper-layers mounted. v4->v5: Move introduction of bd_mounters to patch1. Li Lingfeng (5): block: Count writers and mounters regardless of whether bdev_allow_write_mounted is set block: Expand the meaning of bdev_allow_write_mounted block: Add config option to detect writing to part0 while partitions mounted block: Add config option to show info about opening a mounted device for write block: Show info about opening a lower device for write while upper-layers mounted block/Kconfig | 19 ++++++++ block/bdev.c | 100 +++++++++++++++++++++++++++++++++----- include/linux/blk_types.h | 1 + 3 files changed, 107 insertions(+), 13 deletions(-) -- 2.31.1

2 8

[PATCH OLK-6.6 v2] block: support to account io_ticks precisely
by Yu Kuai 09 Jan '24

09 Jan '24

hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I8RIME CVE: NA ----------------------------------------------- Currently, io_ticks is accounted based on sampling, specifically update_io_ticks() will always account io_ticks by 1 jiffies from bdev_start_io_acct()/blk_account_io_start(), and the result can be inaccurate, for example(HZ is 250): Test script: fio -filename=/dev/sda -bs=4k -rw=write -direct=1 -name=test -thinktime=4ms Test result: util is about 90%, while the disk is really idle. In order to account io_ticks precisely, update_io_ticks() must know if there are IO inflight already, and this requires overhead slightly, hence precise io accounting is disabled by default, and user can enable it through sysfs entry. Noted that for rq-based devcie, part_stat_local_inc/dec() and part_in_flight() is used to track inflight instead of iterating tags, which is not supposed to be used in fast path because 'tags->lock' is grabbed in blk_mq_find_and_get_req(). Signed-off-by: Yu Kuai <yukuai3(a)huawei.com> --- Changes in v2: - some minor code changes, no functional change involed; Documentation/ABI/stable/sysfs-block | 8 ++++-- block/blk-core.c | 27 ++++++++++++++++-- block/blk-merge.c | 3 ++ block/blk-mq-debugfs.c | 2 ++ block/blk-mq.c | 11 +++++++- block/blk-sysfs.c | 42 ++++++++++++++++++++++++++-- block/blk.h | 1 + block/genhd.c | 2 +- include/linux/blk-mq.h | 1 + include/linux/blkdev.h | 3 ++ 10 files changed, 91 insertions(+), 9 deletions(-) diff --git a/Documentation/ABI/stable/sysfs-block b/Documentation/ABI/stable/sysfs-block index 1fe9a553c37b..79027bf2661a 100644 --- a/Documentation/ABI/stable/sysfs-block +++ b/Documentation/ABI/stable/sysfs-block @@ -358,8 +358,12 @@ What: /sys/block/<disk>/queue/iostats Date: January 2009 Contact: linux-block(a)vger.kernel.org Description: - [RW] This file is used to control (on/off) the iostats - accounting of the disk. + [RW] This file is used to control the iostats accounting of the + disk. If this value is 0, iostats accounting is disabled; If + this value is 1, iostats accounting is enabled, but io_ticks is + accounted by sampling and the result is not accurate; If this + value is 2, iostats accounting is enabled and io_ticks is + accounted precisely, but there will be slightly more overhead. What: /sys/block/<disk>/queue/logical_block_size diff --git a/block/blk-core.c b/block/blk-core.c index fdf25b8d6e78..13a27d1d8540 100644 --- a/block/blk-core.c +++ b/block/blk-core.c @@ -71,6 +71,21 @@ static struct kmem_cache *blk_requestq_cachep; */ static struct workqueue_struct *kblockd_workqueue; +static bool precise_iostat; + +static int __init precise_iostat_setup(char *str) +{ + bool precise; + + if (!strtobool(str, &precise)) { + precise_iostat = precise; + pr_info("precise iostat %d\n", precise_iostat); + } + + return 1; +} +__setup("precise_iostat=", precise_iostat_setup); + /** * blk_queue_flag_set - atomically set a queue flag * @flag: flag to be set @@ -441,6 +456,8 @@ struct request_queue *blk_alloc_queue(int node_id) blk_set_default_limits(&q->limits); q->nr_requests = BLKDEV_DEFAULT_RQ; + if (precise_iostat) + blk_queue_flag_set(QUEUE_FLAG_PRECISE_IO_STAT, q); return q; @@ -938,11 +955,15 @@ EXPORT_SYMBOL_GPL(iocb_bio_iopoll); void update_io_ticks(struct block_device *part, unsigned long now, bool end) { unsigned long stamp; + bool precise = blk_queue_precise_io_stat(part->bd_queue); again: stamp = READ_ONCE(part->bd_stamp); - if (unlikely(time_after(now, stamp))) { - if (likely(try_cmpxchg(&part->bd_stamp, &stamp, now))) - __part_stat_add(part, io_ticks, end ? now - stamp : 1); + if (unlikely(time_after(now, stamp)) && + likely(try_cmpxchg(&part->bd_stamp, &stamp, now))) { + if (end || (precise && part_in_flight(part))) + __part_stat_add(part, io_ticks, now - stamp); + else if (!precise) + __part_stat_add(part, io_ticks, 1); } if (part->bd_partno) { part = bdev_whole(part); diff --git a/block/blk-merge.c b/block/blk-merge.c index 65e75efa9bd3..5db8228c46fc 100644 --- a/block/blk-merge.c +++ b/block/blk-merge.c @@ -783,6 +783,9 @@ static void blk_account_io_merge_request(struct request *req) if (blk_do_io_stat(req)) { part_stat_lock(); part_stat_inc(req->part, merges[op_stat_group(req_op(req))]); + if (req->rq_flags & RQF_PRECISE_IO_STAT) + part_stat_local_dec(req->part, + in_flight[op_is_write(req_op(req))]); part_stat_unlock(); } } diff --git a/block/blk-mq-debugfs.c b/block/blk-mq-debugfs.c index aa7a1357c3e1..9ad108307344 100644 --- a/block/blk-mq-debugfs.c +++ b/block/blk-mq-debugfs.c @@ -86,6 +86,7 @@ static const char *const blk_queue_flag_name[] = { QUEUE_FLAG_NAME(FAIL_IO), QUEUE_FLAG_NAME(NONROT), QUEUE_FLAG_NAME(IO_STAT), + QUEUE_FLAG_NAME(PRECISE_IO_STAT), QUEUE_FLAG_NAME(NOXMERGES), QUEUE_FLAG_NAME(ADD_RANDOM), QUEUE_FLAG_NAME(SYNCHRONOUS), @@ -254,6 +255,7 @@ static const char *const rqf_name[] = { RQF_NAME(FAILED), RQF_NAME(QUIET), RQF_NAME(IO_STAT), + RQF_NAME(PRECISE_IO_STAT), RQF_NAME(PM), RQF_NAME(HASHED), RQF_NAME(STATS), diff --git a/block/blk-mq.c b/block/blk-mq.c index dd7c9e3eca1b..498a6306cbb5 100644 --- a/block/blk-mq.c +++ b/block/blk-mq.c @@ -360,8 +360,11 @@ static struct request *blk_mq_rq_ctx_init(struct blk_mq_alloc_data *data, if (data->flags & BLK_MQ_REQ_PM) data->rq_flags |= RQF_PM; - if (blk_queue_io_stat(q)) + if (blk_queue_io_stat(q)) { data->rq_flags |= RQF_IO_STAT; + if (blk_queue_precise_io_stat(q)) + data->rq_flags |= RQF_PRECISE_IO_STAT; + } rq->rq_flags = data->rq_flags; if (data->rq_flags & RQF_SCHED_TAGS) { @@ -994,6 +997,9 @@ static inline void blk_account_io_done(struct request *req, u64 now) update_io_ticks(req->part, jiffies, true); part_stat_inc(req->part, ios[sgrp]); part_stat_add(req->part, nsecs[sgrp], now - req->start_time_ns); + if (req->rq_flags & RQF_PRECISE_IO_STAT) + part_stat_local_dec(req->part, + in_flight[op_is_write(req_op(req))]); part_stat_unlock(); } } @@ -1016,6 +1022,9 @@ static inline void blk_account_io_start(struct request *req) part_stat_lock(); update_io_ticks(req->part, jiffies, false); + if (req->rq_flags & RQF_PRECISE_IO_STAT) + part_stat_local_inc(req->part, + in_flight[op_is_write(req_op(req))]); part_stat_unlock(); } } diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c index 63e481262336..e58b1574e023 100644 --- a/block/blk-sysfs.c +++ b/block/blk-sysfs.c @@ -303,7 +303,6 @@ queue_##name##_store(struct request_queue *q, const char *page, size_t count) \ QUEUE_SYSFS_BIT_FNS(nonrot, NONROT, 1); QUEUE_SYSFS_BIT_FNS(random, ADD_RANDOM, 0); -QUEUE_SYSFS_BIT_FNS(iostats, IO_STAT, 0); QUEUE_SYSFS_BIT_FNS(stable_writes, STABLE_WRITES, 0); #undef QUEUE_SYSFS_BIT_FNS @@ -473,6 +472,45 @@ static ssize_t queue_dax_show(struct request_queue *q, char *page) return queue_var_show(blk_queue_dax(q), page); } +static ssize_t queue_iostats_show(struct request_queue *q, char *page) +{ + int val = 0; + + if (blk_queue_io_stat(q)) + val = blk_queue_precise_io_stat(q) ? 2 : 1; + + return queue_var_show(val, page); +} + +static ssize_t +queue_iostats_store(struct request_queue *q, const char *page, size_t count) +{ + unsigned long nr; + int ret = queue_var_store(&nr, page, count); + + if (ret < 0) + return ret; + + switch (nr) { + case 0: + blk_queue_flag_clear(QUEUE_FLAG_IO_STAT, q); + blk_queue_flag_clear(QUEUE_FLAG_PRECISE_IO_STAT, q); + break; + case 1: + blk_queue_flag_set(QUEUE_FLAG_IO_STAT, q); + blk_queue_flag_clear(QUEUE_FLAG_PRECISE_IO_STAT, q); + break; + case 2: + blk_queue_flag_set(QUEUE_FLAG_IO_STAT, q); + blk_queue_flag_set(QUEUE_FLAG_PRECISE_IO_STAT, q); + break; + default: + return -EINVAL; + } + + return count; +} + #define QUEUE_RO_ENTRY(_prefix, _name) \ static struct queue_sysfs_entry _prefix##_entry = { \ .attr = { .name = _name, .mode = 0444 }, \ @@ -494,6 +532,7 @@ QUEUE_RO_ENTRY(queue_max_segments, "max_segments"); QUEUE_RO_ENTRY(queue_max_integrity_segments, "max_integrity_segments"); QUEUE_RO_ENTRY(queue_max_segment_size, "max_segment_size"); QUEUE_RW_ENTRY(elv_iosched, "scheduler"); +QUEUE_RW_ENTRY(queue_iostats, "iostats"); QUEUE_RO_ENTRY(queue_logical_block_size, "logical_block_size"); QUEUE_RO_ENTRY(queue_physical_block_size, "physical_block_size"); @@ -539,7 +578,6 @@ static struct queue_sysfs_entry queue_hw_sector_size_entry = { }; QUEUE_RW_ENTRY(queue_nonrot, "rotational"); -QUEUE_RW_ENTRY(queue_iostats, "iostats"); QUEUE_RW_ENTRY(queue_random, "add_random"); QUEUE_RW_ENTRY(queue_stable_writes, "stable_writes"); diff --git a/block/blk.h b/block/blk.h index 08a358bc0919..67915b04b3c1 100644 --- a/block/blk.h +++ b/block/blk.h @@ -344,6 +344,7 @@ static inline bool blk_do_io_stat(struct request *rq) } void update_io_ticks(struct block_device *part, unsigned long now, bool end); +unsigned int part_in_flight(struct block_device *part); static inline void req_set_nomerge(struct request_queue *q, struct request *req) { diff --git a/block/genhd.c b/block/genhd.c index 4a16a424f57d..b3ff653f3e50 100644 --- a/block/genhd.c +++ b/block/genhd.c @@ -118,7 +118,7 @@ static void part_stat_read_all(struct block_device *part, } } -static unsigned int part_in_flight(struct block_device *part) +unsigned int part_in_flight(struct block_device *part) { unsigned int inflight = 0; int cpu; diff --git a/include/linux/blk-mq.h b/include/linux/blk-mq.h index 3d74f3e5b995..9dfee0644f9c 100644 --- a/include/linux/blk-mq.h +++ b/include/linux/blk-mq.h @@ -46,6 +46,7 @@ typedef __u32 __bitwise req_flags_t; #define RQF_QUIET ((__force req_flags_t)(1 << 11)) /* account into disk and partition IO statistics */ #define RQF_IO_STAT ((__force req_flags_t)(1 << 13)) +#define RQF_PRECISE_IO_STAT ((__force req_flags_t)(1 << 14)) /* runtime pm request */ #define RQF_PM ((__force req_flags_t)(1 << 15)) /* on IO scheduler merge hash */ diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 9f3bcbcb156d..bea0b5fdac74 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -536,6 +536,7 @@ struct request_queue { #define QUEUE_FLAG_NONROT 6 /* non-rotational device (SSD) */ #define QUEUE_FLAG_VIRT QUEUE_FLAG_NONROT /* paravirt device */ #define QUEUE_FLAG_IO_STAT 7 /* do disk/partitions IO accounting */ +#define QUEUE_FLAG_PRECISE_IO_STAT 8 /* do disk/partitions IO accounting precisely */ #define QUEUE_FLAG_NOXMERGES 9 /* No extended merges */ #define QUEUE_FLAG_ADD_RANDOM 10 /* Contributes to random pool */ #define QUEUE_FLAG_SYNCHRONOUS 11 /* always completes in submit context */ @@ -576,6 +577,8 @@ bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q); #define blk_queue_stable_writes(q) \ test_bit(QUEUE_FLAG_STABLE_WRITES, &(q)->queue_flags) #define blk_queue_io_stat(q) test_bit(QUEUE_FLAG_IO_STAT, &(q)->queue_flags) +#define blk_queue_precise_io_stat(q) \ + test_bit(QUEUE_FLAG_PRECISE_IO_STAT, &(q)->queue_flags) #define blk_queue_add_random(q) test_bit(QUEUE_FLAG_ADD_RANDOM, &(q)->queue_flags) #define blk_queue_zone_resetall(q) \ test_bit(QUEUE_FLAG_ZONE_RESETALL, &(q)->queue_flags) -- 2.39.2

1 0

2024

2023

2022

2021

2020

2019

Kernel January 2024