- Kernel - mailweb.openeuler.org

[PATCH openEuler-1.0-LTS 01/16] mm,hwpoison: cleanup unused PageHuge() check
by Yang Yingliang 09 Feb '22

09 Feb '22

From: Naoya Horiguchi <naoya.horiguchi(a)nec.com> mainline inclusion from linux-v5.10-rc1 commit 7d9d46ac87f91b8dedad5241d64382b650e26487 category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I4LE22 CVE: NA -------------------------------- Drop the PageHuge check, which is dead code since memory_failure() forks into memory_failure_hugetlb() for hugetlb pages. memory_failure() and memory_failure_hugetlb() shares some functions like hwpoison_user_mappings() and identify_page_state(), so they should properly handle 4kB page, thp, and hugetlb. Signed-off-by: Naoya Horiguchi <naoya.horiguchi(a)nec.com> Signed-off-by: Oscar Salvador <osalvador(a)suse.de> Reviewed-by: Mike Kravetz <mike.kravetz(a)oracle.com> Signed-off-by: Ma Wupeng <mawupeng1(a)huawei.com> Reviewed-by: tong tiangen <tongtiangen(a)huawei.com> Reviewed-by: Kefeng Wang <wangkefeng.wang(a)huawei.com> Signed-off-by: Yang Yingliang <yangyingliang(a)huawei.com> --- mm/memory-failure.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) diff --git a/mm/memory-failure.c b/mm/memory-failure.c index 72e1746e5386f..a432ba37e132a 100644 --- a/mm/memory-failure.c +++ b/mm/memory-failure.c @@ -1373,10 +1373,7 @@ int memory_failure(unsigned long pfn, int flags) * page_remove_rmap() in try_to_unmap_one(). So to determine page status * correctly, we save a copy of the page flags at this time. */ - if (PageHuge(p)) - page_flags = hpage->flags; - else - page_flags = p->flags; + page_flags = p->flags; /* * unpoison always clear PG_hwpoison inside page lock -- 2.25.1

1 15

[PATCH openEuler-1.0-LTS] scsi: Revert "target: iscsi: Wait for all commands to finish before freeing a session"
by Yang Yingliang 09 Feb '22

09 Feb '22

From: Bart Van Assche <bvanassche(a)acm.org> mainline inclusion from mainline-v5.6-rc3 commit 807b9515b7d044cf77df31f1af9d842a76ecd5cb category: bugfix bugzilla: 186215 CVE: NA -------------------------------- Since commit e9d3009cb936 introduced a regression and since the fix for that regression was not perfect, revert this commit. Link: https://marc.info/?l=target-devel&m=158157054906195 Cc: Rahul Kundu <rahul.kundu(a)chelsio.com> Cc: Mike Marciniszyn <mike.marciniszyn(a)intel.com> Cc: Sagi Grimberg <sagi(a)grimberg.me> Reported-by: Dakshaja Uppalapati <dakshaja(a)chelsio.com> Fixes: e9d3009cb936 ("scsi: target: iscsi: Wait for all commands to finish before freeing a session") Signed-off-by: Bart Van Assche <bvanassche(a)acm.org> Signed-off-by: Martin K. Petersen <martin.petersen(a)oracle.com> Signed-off-by: Ye Bin <yebin10(a)huawei.com> Reviewed-by: Jason Yan <yanaijie(a)huawei.com> Signed-off-by: Yang Yingliang <yangyingliang(a)huawei.com> --- drivers/target/iscsi/iscsi_target.c | 10 ++-------- include/scsi/iscsi_proto.h | 1 - 2 files changed, 2 insertions(+), 9 deletions(-) diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c index 92843e8db711b..58ccded1be857 100644 --- a/drivers/target/iscsi/iscsi_target.c +++ b/drivers/target/iscsi/iscsi_target.c @@ -1156,9 +1156,7 @@ int iscsit_setup_scsi_cmd(struct iscsi_conn *conn, struct iscsi_cmd *cmd, hdr->cmdsn, be32_to_cpu(hdr->data_length), payload_length, conn->cid); - if (target_get_sess_cmd(&cmd->se_cmd, true) < 0) - return iscsit_add_reject_cmd(cmd, - ISCSI_REASON_WAITING_FOR_LOGOUT, buf); + target_get_sess_cmd(&cmd->se_cmd, true); cmd->sense_reason = transport_lookup_cmd_lun(&cmd->se_cmd, scsilun_to_int(&hdr->lun)); @@ -2012,9 +2010,7 @@ iscsit_handle_task_mgt_cmd(struct iscsi_conn *conn, struct iscsi_cmd *cmd, conn->sess->se_sess, 0, DMA_NONE, TCM_SIMPLE_TAG, cmd->sense_buffer + 2); - if (target_get_sess_cmd(&cmd->se_cmd, true) < 0) - return iscsit_add_reject_cmd(cmd, - ISCSI_REASON_WAITING_FOR_LOGOUT, buf); + target_get_sess_cmd(&cmd->se_cmd, true); /* * TASK_REASSIGN for ERL=2 / connection stays inside of @@ -4230,8 +4226,6 @@ int iscsit_close_connection( * must wait until they have completed. */ iscsit_check_conn_usage_count(conn); - target_sess_cmd_list_set_waiting(sess->se_sess); - target_wait_for_sess_cmds(sess->se_sess); ahash_request_free(conn->conn_tx_hash); if (conn->conn_rx_hash) { diff --git a/include/scsi/iscsi_proto.h b/include/scsi/iscsi_proto.h index f0a01a54bd153..df156f1d50b2d 100644 --- a/include/scsi/iscsi_proto.h +++ b/include/scsi/iscsi_proto.h @@ -638,7 +638,6 @@ struct iscsi_reject { #define ISCSI_REASON_BOOKMARK_INVALID 9 #define ISCSI_REASON_BOOKMARK_NO_RESOURCES 10 #define ISCSI_REASON_NEGOTIATION_RESET 11 -#define ISCSI_REASON_WAITING_FOR_LOGOUT 12 /* Max. number of Key=Value pairs in a text message */ #define MAX_KEY_VALUE_PAIRS 8192 -- 2.25.1

1 0

[PATCH openEuler-1.0-LTS] mm: Modify sharepool sp_mmap() page_offset
by Yang Yingliang 09 Feb '22

09 Feb '22

From: Guo Mengqi <guomengqi3(a)huawei.com> ascend inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I4SPNL CVE: NA ----------------------------------- In sp_mmap(), if use offset = va - MMAP_BASE/DVPP_BASE, then normal sp_alloc pgoff may have same value with DVPP pgoff, causing DVPP and sp_alloc mapped to overlapped part of file unexpectedly. To fix the problem, pass VA value as mmap offset, for in this scenario, VA value in one task address space will not be same. Signed-off-by: Guo Mengqi <guomengqi3(a)huawei.com> Reviewed-by: Ding Tianhong <dingtianhong(a)huawei.com> Signed-off-by: Yang Yingliang <yangyingliang(a)huawei.com> --- mm/share_pool.c | 21 +++++---------------- 1 file changed, 5 insertions(+), 16 deletions(-) diff --git a/mm/share_pool.c b/mm/share_pool.c index ef74de39053b2..90733b807f12e 100644 --- a/mm/share_pool.c +++ b/mm/share_pool.c @@ -57,6 +57,11 @@ #define spg_valid(spg) ((spg)->is_alive == true) +/* Use spa va address as mmap offset. This can work because spa_file + * is setup with 64-bit address space. So va shall be well covered. + */ +#define addr_offset(spa) ((spa)->va_start) + #define byte2kb(size) ((size) >> 10) #define byte2mb(size) ((size) >> 20) #define page2kb(page_num) ((page_num) << (PAGE_SHIFT - 10)) @@ -950,22 +955,6 @@ static bool is_device_addr(unsigned long addr) return false; } -static loff_t addr_offset(struct sp_area *spa) -{ - unsigned long addr; - - if (unlikely(!spa)) { - WARN(1, "invalid spa when calculate addr offset\n"); - return 0; - } - addr = spa->va_start; - - if (!is_device_addr(addr)) - return (loff_t)(addr - MMAP_SHARE_POOL_START); - - return (loff_t)(addr - sp_dev_va_start[spa->device_id]); -} - static struct sp_group *create_spg(int spg_id) { int ret; -- 2.25.1

1 0

[PATCH openEuler-1.0-LTS 1/2] uce: copy_from_user scenario support kernel recovery
by Yang Yingliang 09 Feb '22

09 Feb '22

From: Tong Tiangen <tongtiangen(a)huawei.com> hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I4SK3S CVE: NA -------------------------------- This patch add uce kernel recovery path support in copy_from_user. Signed-off-by: Tong Tiangen <tongtiangen(a)huawei.com> Reviewed-by: Kefeng Wang <wangkefeng.wang(a)huawei.com> Signed-off-by: Yang Yingliang <yangyingliang(a)huawei.com> --- arch/arm64/include/asm/exception.h | 1 + arch/arm64/lib/copy_from_user.S | 6 ++++++ arch/arm64/mm/fault.c | 17 +++++++++++++++-- kernel/sysctl.c | 3 ++- 4 files changed, 24 insertions(+), 3 deletions(-) diff --git a/arch/arm64/include/asm/exception.h b/arch/arm64/include/asm/exception.h index 559d86ad9e5d5..d0c8a1fda453a 100644 --- a/arch/arm64/include/asm/exception.h +++ b/arch/arm64/include/asm/exception.h @@ -51,6 +51,7 @@ struct uce_kernel_recovery_info { extern int copy_page_cow_sea_fallback(void); extern int copy_generic_read_sea_fallback(void); +extern int copy_from_user_sea_fallback(void); #endif #endif /* __ASM_EXCEPTION_H */ diff --git a/arch/arm64/lib/copy_from_user.S b/arch/arm64/lib/copy_from_user.S index 7cd6eeaa216cf..d1afb61df158b 100644 --- a/arch/arm64/lib/copy_from_user.S +++ b/arch/arm64/lib/copy_from_user.S @@ -72,6 +72,12 @@ ENTRY(__arch_copy_from_user) uaccess_disable_not_uao x3, x4 mov x0, #0 // Nothing to copy ret + + .global copy_from_user_sea_fallback +copy_from_user_sea_fallback: + uaccess_disable_not_uao x3, x4 + mov x0, #-1 + ret ENDPROC(__arch_copy_from_user) .section .fixup,"ax" diff --git a/arch/arm64/mm/fault.c b/arch/arm64/mm/fault.c index 08040fe73199a..50e37f4097ccc 100644 --- a/arch/arm64/mm/fault.c +++ b/arch/arm64/mm/fault.c @@ -660,9 +660,18 @@ static int do_bad(unsigned long addr, unsigned int esr, struct pt_regs *regs) int kernel_access_sea_recovery; #define UCE_KER_REC_NUM ARRAY_SIZE(reco_info) +/* + * One entry corresponds to one scene, and the scene switch is controlled by the + * corresponding bit of kernel_access_sea_recovery + * (the first entry corresponds to bit0, the second entry corresponds to bit1...), + * and the switch is visible to the user, so the order of eatch entry here cannot + * be easily change. Now the maximum entry is limited by the type of variable + * kernel_access_sea_recovery. + */ static struct uce_kernel_recovery_info reco_info[] = { {copy_page_cow_sea_fallback, "copy_page_cow", (unsigned long)copy_page_cow, 0}, {copy_generic_read_sea_fallback, "__arch_copy_to_user_generic_read", (unsigned long)__arch_copy_to_user_generic_read, 0}, + {copy_from_user_sea_fallback, "__arch_copy_from_user", (unsigned long)__arch_copy_from_user, 0}, }; static int __init kernel_access_sea_recovery_init(void) @@ -769,6 +778,9 @@ static int is_in_kernel_recovery(unsigned int esr, struct pt_regs *regs) } for (i = 0; i < UCE_KER_REC_NUM; i++) { + if (!((kernel_access_sea_recovery >> i) & 0x1)) + continue; + info = &reco_info[i]; if (info->fn && regs->pc >= info->addr && regs->pc < (info->addr + info->size)) { @@ -777,7 +789,8 @@ static int is_in_kernel_recovery(unsigned int esr, struct pt_regs *regs) } } - pr_emerg("UCE: symbol is not match.\n"); + pr_emerg("UCE: symbol is not match or switch if off, kernel recovery %d.\n", + kernel_access_sea_recovery); return -EINVAL; } #endif @@ -847,7 +860,7 @@ static int do_sea(unsigned long addr, unsigned int esr, struct pt_regs *regs) "Uncorrected hardware memory use with kernel recovery in kernel-access\n", current); } else { - die("Uncorrected hardware memory error (kernel recovery on but not match idx) in kernel-access\n", + die("Uncorrected hardware memory error (not match idx or sence switch is off) in kernel-access\n", regs, esr); } diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 542611081c610..302cc4c7de4df 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -130,6 +130,7 @@ static int __maybe_unused two = 2; static int __maybe_unused three = 3; static int __maybe_unused four = 4; static int __maybe_unused five = 5; +static int __maybe_unused seven = 7; static unsigned long zero_ul; static unsigned long one_ul = 1; static unsigned long long_max = LONG_MAX; @@ -1280,7 +1281,7 @@ static struct ctl_table kern_table[] = { .mode = 0644, .proc_handler = proc_dointvec_minmax, .extra1 = &zero, - .extra2 = &three, + .extra2 = &seven, }, #endif -- 2.25.1

1 1

[PATCH openEuler-1.0-LTS 1/5] share_pool: Fix flags conflict
by Yang Yingliang 09 Feb '22

09 Feb '22

From: Wang Wensheng <wangwensheng4(a)huawei.com> ascend inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I4SON8 CVE: NA ------------------------------------------------- MAP_SHARE_POOL and MAP_FIXED_NOREPLACE have the same value. Redefine MAP_SHARE_POOL to fix it. Signed-off-by: Wang Wensheng <wangwensheng4(a)huawei.com> Reviewed-by: Weilong Chen <chenweilong(a)huawei.com> Signed-off-by: Yang Yingliang <yangyingliang(a)huawei.com> --- include/linux/share_pool.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/include/linux/share_pool.h b/include/linux/share_pool.h index cd4c305449dd9..88ef96ac0bfb3 100644 --- a/include/linux/share_pool.h +++ b/include/linux/share_pool.h @@ -170,7 +170,7 @@ struct sp_walk_data { pmd_t *pmd; }; -#define MAP_SHARE_POOL 0x100000 +#define MAP_SHARE_POOL 0x200000 #define MMAP_TOP_4G_SIZE 0x100000000UL -- 2.25.1

1 4

[PATCH openEuler-1.0-LTS 01/36] mm/memory_hotplug: drop "online" parameter from add_memory_resource()
by Yang Yingliang 09 Feb '22

09 Feb '22

From: David Hildenbrand <david(a)redhat.com> mainline inclusion from linux-5.0-rc1 commit f29d8e9c0191a2a02500945db505e5c89159c3f4 category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I4SK3S CVE: NA -------------------------------- Userspace should always be in charge of how to online memory and if memory should be onlined automatically in the kernel. Let's drop the parameter to overwrite this - XEN passes memhp_auto_online, just like add_memory(), so we can directly use that instead internally. Link: http://lkml.kernel.org/r/20181123123740.27652-1-david@redhat.com Signed-off-by: David Hildenbrand <david(a)redhat.com> Acked-by: Michal Hocko <mhocko(a)suse.com> Reviewed-by: Oscar Salvador <osalvador(a)suse.de> Acked-by: Juergen Gross <jgross(a)suse.com> Cc: Boris Ostrovsky <boris.ostrovsky(a)oracle.com> Cc: Stefano Stabellini <sstabellini(a)kernel.org> Cc: Dan Williams <dan.j.williams(a)intel.com> Cc: Pavel Tatashin <pasha.tatashin(a)oracle.com> Cc: David Hildenbrand <david(a)redhat.com> Cc: Joonsoo Kim <iamjoonsoo.kim(a)lge.com> Cc: Arun KS <arunks(a)codeaurora.org> Cc: Mathieu Malaterre <malat(a)debian.org> Cc: Stephen Rothwell <sfr(a)canb.auug.org.au> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds(a)linux-foundation.org> Signed-off-by: Ma Wupeng <mawupeng1(a)huawei.com> Reviewed-by: Kefeng Wang <wangkefeng.wang(a)huawei.com> Signed-off-by: Yang Yingliang <yangyingliang(a)huawei.com> --- drivers/xen/balloon.c | 2 +- include/linux/memory_hotplug.h | 2 +- mm/memory_hotplug.c | 6 +++--- 3 files changed, 5 insertions(+), 5 deletions(-) diff --git a/drivers/xen/balloon.c b/drivers/xen/balloon.c index 7703ac47062fa..e692dcf576d66 100644 --- a/drivers/xen/balloon.c +++ b/drivers/xen/balloon.c @@ -352,7 +352,7 @@ static enum bp_state reserve_additional_memory(void) mutex_unlock(&balloon_mutex); /* add_memory_resource() requires the device_hotplug lock */ lock_device_hotplug(); - rc = add_memory_resource(nid, resource, memhp_auto_online); + rc = add_memory_resource(nid, resource); unlock_device_hotplug(); mutex_lock(&balloon_mutex); diff --git a/include/linux/memory_hotplug.h b/include/linux/memory_hotplug.h index 8782f0e993704..9d28fca5bbde8 100644 --- a/include/linux/memory_hotplug.h +++ b/include/linux/memory_hotplug.h @@ -326,7 +326,7 @@ extern int walk_memory_range(unsigned long start_pfn, unsigned long end_pfn, void *arg, int (*func)(struct memory_block *, void *)); extern int __add_memory(int nid, u64 start, u64 size); extern int add_memory(int nid, u64 start, u64 size); -extern int add_memory_resource(int nid, struct resource *resource, bool online); +extern int add_memory_resource(int nid, struct resource *resource); extern int arch_add_memory(int nid, u64 start, u64 size, struct vmem_altmap *altmap, bool want_memblock); extern void move_pfn_range_to_zone(struct zone *zone, unsigned long start_pfn, diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c index 1e20ad038496d..f13144a65b127 100644 --- a/mm/memory_hotplug.c +++ b/mm/memory_hotplug.c @@ -1051,7 +1051,7 @@ static int online_memory_block(struct memory_block *mem, void *arg) * * we are OK calling __meminit stuff here - we have CONFIG_MEMORY_HOTPLUG */ -int __ref add_memory_resource(int nid, struct resource *res, bool online) +int __ref add_memory_resource(int nid, struct resource *res) { u64 start, size; bool new_node = false; @@ -1114,7 +1114,7 @@ int __ref add_memory_resource(int nid, struct resource *res, bool online) mem_hotplug_done(); /* online pages if requested */ - if (online) + if (memhp_auto_online) walk_memory_range(PFN_DOWN(start), PFN_UP(start + size - 1), NULL, online_memory_block); @@ -1138,7 +1138,7 @@ int __ref __add_memory(int nid, u64 start, u64 size) if (IS_ERR(res)) return PTR_ERR(res); - ret = add_memory_resource(nid, res, memhp_auto_online); + ret = add_memory_resource(nid, res); if (ret < 0) release_memory_resource(res); return ret; -- 2.25.1

1 34

Re: [PATCH -next] dm: make sure dm_table is binded before queue request
by weiyongjun (A) 09 Feb '22

09 Feb '22

在 2022/2/9 10:55, Zhang Yi 写道: > We found a NULL pointer dereference problem when using dm-mpath target. > The problem is if we submit IO between loading and binding the table, > we could neither get a valid dm_target nor a valid dm table when > submitting request in dm_mq_queue_rq(). BIO based dm target could > handle this case in dm_submit_bio(). This patch fix this by checking > the valid live table before submitting request. > > Signed-off-by: Zhang Yi <yi.zhang(a)huawei.com> > --- > drivers/md/dm-rq.c | 11 +++++++++-- > 1 file changed, 9 insertions(+), 2 deletions(-) > Looks good to me

1 0

[PATCH openEuler-1.0-LTS v3 0/2]Add copy_from_user and get_user support uce kernel recovery
by Tong Tiangen 08 Feb '22

08 Feb '22

v3 -> v2: make is_get_user_kernel_recovery_enable() inline. v2 -> v1: update commit message. Tong Tiangen (2): uce: copy_from_user scenario support kernel recovery uce: get_user scenario support kernel recovery arch/arm64/include/asm/exception.h | 6 +++ arch/arm64/include/asm/uaccess.h | 67 ++++++++++++++++++++++++++++++ arch/arm64/lib/Makefile | 2 + arch/arm64/lib/copy_from_user.S | 6 +++ arch/arm64/lib/get_user.c | 20 +++++++++ arch/arm64/mm/fault.c | 29 +++++++++++-- include/linux/sched.h | 1 + kernel/sysctl.c | 3 +- mm/internal.h | 1 + 9 files changed, 130 insertions(+), 5 deletions(-) create mode 100644 arch/arm64/lib/get_user.c -- 2.18.0.huawei.25

1 2

[PATCH openEuler-1.0-LTS] ext4: fix e2fsprogs checksum failure for mounted filesystem
by Yang Yingliang 08 Feb '22

08 Feb '22

From: Jan Kara <jack(a)suse.cz> mainline inclusion from mainline-v5.15-rc1 commit b2bbb92f7042e8075fb036bf97043339576330c3 category: bugfix bugzilla: 186203 CVE: NA -------------------------------- Commit 81414b4dd48 ("ext4: remove redundant sb checksum recomputation") removed checksum recalculation after updating superblock free space / inode counters in ext4_fill_super() based on the fact that we will recalculate the checksum on superblock writeout. That is correct assumption but until the writeout happens (which can take a long time) the checksum is incorrect in the buffer cache and if programs such as tune2fs or resize2fs is called shortly after a file system is mounted can fail. So return back the checksum recalculation and add a comment explaining why. Fixes: 81414b4dd48f ("ext4: remove redundant sb checksum recomputation") Cc: stable(a)kernel.org Reported-by: Boyang Xue <bxue(a)redhat.com> Signed-off-by: Jan Kara <jack(a)suse.cz> Signed-off-by: Theodore Ts'o <tytso(a)mit.edu> Link: https://lore.kernel.org/r/20210812124737.21981-1-jack@suse.cz Signed-off-by: Zheng Liang <zhengliang6(a)huawei.com> Reviewed-by: Zhang Yi <yi.zhang(a)huawei.com> Signed-off-by: Yang Yingliang <yangyingliang(a)huawei.com> --- fs/ext4/super.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/fs/ext4/super.c b/fs/ext4/super.c index b89f431ec78d3..59637e70f9e32 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -4693,6 +4693,14 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent) err = percpu_counter_init(&sbi->s_freeinodes_counter, freei, GFP_KERNEL); } + /* + * Update the checksum after updating free space/inode + * counters. Otherwise the superblock can have an incorrect + * checksum in the buffer cache until it is written out and + * e2fsprogs programs trying to open a file system immediately + * after it is mounted can fail. + */ + ext4_superblock_csum_set(sb); if (!err) err = percpu_counter_init(&sbi->s_dirs_counter, ext4_count_dirs(sb), GFP_KERNEL); -- 2.25.1

1 0

[PATCH openEuler-1.0-LTS 1/2] mm/memcg_memfs_info: show files that having pages charged in mem_cgroup
by Yang Yingliang 08 Feb '22

08 Feb '22

From: Liu Shixin <liushixin2(a)huawei.com> hulk inclusion category: feature bugzilla: 186182, https://gitee.com/openeuler/kernel/issues/I4SBQX CVE: NA -------------------------------- Support to print rootfs files and tmpfs files that having pages charged in given memory cgroup. The files infomations can be printed through interface "memory.memfs_files_info" or printed when OOM is triggered. In order not to flush memory logs, we limit the maximum number of files to be printed when oom through interface "max_print_files_in_oom". And in order to filter out small files, we limit the minimum size of files that can be printed through interface "size_threshold". Signed-off-by: Liu Shixin <liushixin2(a)huawei.com> Reviewed-by: Kefeng Wang <wangkefeng.wang(a)huawei.com> Reviewed-by: Zhang Yi <yi.zhang(a)huawei.com> Signed-off-by: Yang Yingliang <yangyingliang(a)huawei.com> --- Documentation/vm/memcg_memfs_info.rst | 40 ++++ include/linux/memcg_memfs_info.h | 21 ++ init/Kconfig | 10 + mm/Makefile | 1 + mm/memcg_memfs_info.c | 321 ++++++++++++++++++++++++++ mm/memcontrol.c | 11 + 6 files changed, 404 insertions(+) create mode 100644 Documentation/vm/memcg_memfs_info.rst create mode 100644 include/linux/memcg_memfs_info.h create mode 100644 mm/memcg_memfs_info.c diff --git a/Documentation/vm/memcg_memfs_info.rst b/Documentation/vm/memcg_memfs_info.rst new file mode 100644 index 0000000000000..aff432d125e52 --- /dev/null +++ b/Documentation/vm/memcg_memfs_info.rst @@ -0,0 +1,40 @@ +.. SPDX-License-Identifier: GPL-2.0+ + +================ +Memcg Memfs Info +================ + +Overview +======== + +Support to print rootfs files and tmpfs files that having pages charged +in given memory cgroup. The files infomations can be printed through +interface "memory.memfs_files_info" or printed when OOM is triggered. + +User control +============ + +1. /sys/kernel/mm/memcg_memfs_info/enable +----------------------------------------- + +Boolean type. The default value is 0, set it to 1 to enable the feature. + +2. /sys/kernel/mm/memcg_memfs_info/max_print_files_in_oom +--------------------------------------------------------- + +Unsigned long type. The default value is 500, indicating that the maximum of +files can be print to console when OOM is triggered. + +3. /sys/kernel/mm/memcg_memfs_info/size_threshold +------------------------------------------------- + +Unsigned long type. The default value is 0, indicating that the minimum size of +files that can be printed. + +4. /sys/fs/cgroup/memory/<memory>/memory.memfs_files_info +--------------------------------------------------------- + +Outputs the files who use memory in this memory cgroup. + +--- +Liu Shixin, Jan 2022 diff --git a/include/linux/memcg_memfs_info.h b/include/linux/memcg_memfs_info.h new file mode 100644 index 0000000000000..658a91e22bd7e --- /dev/null +++ b/include/linux/memcg_memfs_info.h @@ -0,0 +1,21 @@ +/* SPDX-License-Identifier: GPL-2.0+ */ +#ifndef _LINUX_MEMCG_MEMFS_INFO_H +#define _LINUX_MEMCG_MEMFS_INFO_H + +#include <linux/memcontrol.h> +#include <linux/seq_file.h> + +#ifdef CONFIG_MEMCG_MEMFS_INFO +void mem_cgroup_print_memfs_info(struct mem_cgroup *memcg, struct seq_file *m); +int mem_cgroup_memfs_files_show(struct seq_file *m, void *v); +void mem_cgroup_memfs_info_init(void); +#else +static inline void mem_cgroup_print_memfs_info(struct mem_cgroup *memcg, + struct seq_file *m) +{ +} +static inline void mem_cgroup_memfs_info_init(void) +{ +} +#endif +#endif diff --git a/init/Kconfig b/init/Kconfig index a338519692d54..1a0b15c5a82b9 100644 --- a/init/Kconfig +++ b/init/Kconfig @@ -733,6 +733,16 @@ config MEMCG_KMEM depends on MEMCG && !SLOB default y +config MEMCG_MEMFS_INFO + bool "Show memfs files that have pages charged in given memory cgroup" + depends on MEMCG + default n + help + Support to print rootfs files and tmpfs files that having pages + charged in given memory cgroup. The files infomations can be printed + through interface "memory.memfs_files_info" or printed when OOM is + triggered. + config BLK_CGROUP bool "IO controller" depends on BLOCK diff --git a/mm/Makefile b/mm/Makefile index deee05d22a853..8fba091be3868 100644 --- a/mm/Makefile +++ b/mm/Makefile @@ -108,3 +108,4 @@ obj-$(CONFIG_MEMFD_CREATE) += memfd.o obj-$(CONFIG_ASCEND_AUTO_TUNING_HUGEPAGE) += hugepage_tuning.o obj-$(CONFIG_PIN_MEMORY) += pin_mem.o obj-$(CONFIG_ASCEND_SHARE_POOL) += share_pool.o +obj-$(CONFIG_MEMCG_MEMFS_INFO) += memcg_memfs_info.o diff --git a/mm/memcg_memfs_info.c b/mm/memcg_memfs_info.c new file mode 100644 index 0000000000000..346175026cae6 --- /dev/null +++ b/mm/memcg_memfs_info.c @@ -0,0 +1,321 @@ +// SPDX-License-Identifier: GPL-2.0+ + +#include <linux/memcg_memfs_info.h> +#include <linux/fs.h> +#include <linux/sysfs.h> +#include <linux/kobject.h> +#include <linux/slab.h> +#include "../fs/mount.h" + +#define SEQ_printf(m, x...) \ +do { \ + if (m) \ + seq_printf(m, x); \ + else \ + pr_info(x); \ +} while (0) + +struct print_files_control { + struct mem_cgroup *memcg; + struct seq_file *m; + unsigned long size_threshold; + unsigned long max_print_files; + + char *pathbuf; + unsigned long pathbuf_size; + + const char *fs_type_name; + struct vfsmount *vfsmnt; + unsigned long total_print_files; + unsigned long total_files_size; +}; + +static bool memfs_enable; +static unsigned long memfs_size_threshold; +static unsigned long memfs_max_print_files = 500; + +static const char *const fs_type_names[] = { + "rootfs", + "tmpfs", +}; + +static struct vfsmount *memfs_get_vfsmount(struct super_block *sb) +{ + struct mount *mnt; + struct vfsmount *vfsmnt; + + lock_mount_hash(); + list_for_each_entry(mnt, &sb->s_mounts, mnt_instance) { + /* + * There may be multiple mount points for a super_block, + * just need to print one of these mount points to determine + * the file path. + */ + vfsmnt = mntget(&mnt->mnt); + unlock_mount_hash(); + return vfsmnt; + } + unlock_mount_hash(); + + return NULL; +} + +static unsigned long memfs_count_in_mem_cgroup(struct mem_cgroup *memcg, + struct address_space *mapping) +{ + struct radix_tree_iter iter; + unsigned long size = 0; + struct page *page, *head; + void __rcu **slot; + + rcu_read_lock(); + radix_tree_for_each_slot(slot, &mapping->i_pages, &iter, 0) { + page = radix_tree_deref_slot(slot); + + if (unlikely(!page)) + continue; + if (radix_tree_exception(page)) { + if (radix_tree_deref_retry(page)) + slot = radix_tree_iter_retry(&iter); + continue; + } + + head = compound_head(page); + if (memcg == head->mem_cgroup) + size += PAGE_SIZE; + } + rcu_read_unlock(); + return size; +} + +static void memfs_show_file_in_mem_cgroup(void *data, struct inode *inode) +{ + struct print_files_control *pfc = data; + struct dentry *dentry; + unsigned long size; + struct path path; + char *filepath; + + size = memfs_count_in_mem_cgroup(pfc->memcg, inode->i_mapping); + if (!size || size < pfc->size_threshold) + return; + + dentry = d_find_alias(inode); + if (!dentry) + return; + path.mnt = pfc->vfsmnt; + path.dentry = dentry; + filepath = d_absolute_path(&path, pfc->pathbuf, pfc->pathbuf_size); + if (!filepath || IS_ERR(filepath)) + filepath = "(too long)"; + pfc->total_print_files++; + pfc->total_files_size += size; + dput(dentry); + + /* + * To prevent excessive logs, limit the amount of data + * that can be output to logs. + */ + if (!pfc->m && pfc->total_print_files > pfc->max_print_files) + return; + + SEQ_printf(pfc->m, "%lukB %llukB %s\n", + size >> 10, inode->i_size >> 10, filepath); +} + +static void memfs_show_files_in_mem_cgroup(struct super_block *sb, void *data) +{ + struct print_files_control *pfc = data; + struct inode *inode, *toput_inode = NULL; + + if (strncmp(sb->s_type->name, + pfc->fs_type_name, strlen(pfc->fs_type_name))) + return; + + pfc->vfsmnt = memfs_get_vfsmount(sb); + if (!pfc->vfsmnt) + return; + + spin_lock(&sb->s_inode_list_lock); + list_for_each_entry(inode, &sb->s_inodes, i_sb_list) { + spin_lock(&inode->i_lock); + + if ((inode->i_state & (I_FREEING|I_WILL_FREE|I_NEW)) || + (inode->i_mapping->nrpages == 0 && !need_resched())) { + spin_unlock(&inode->i_lock); + continue; + } + __iget(inode); + spin_unlock(&inode->i_lock); + spin_unlock(&sb->s_inode_list_lock); + + memfs_show_file_in_mem_cgroup(pfc, inode); + + 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); + mntput(pfc->vfsmnt); +} + +void mem_cgroup_print_memfs_info(struct mem_cgroup *memcg, struct seq_file *m) +{ + struct print_files_control pfc = { + .memcg = memcg, + .m = m, + .max_print_files = memfs_max_print_files, + .size_threshold = memfs_size_threshold, + }; + char *pathbuf; + int i; + + if (!memfs_enable || !memcg) + return; + + pathbuf = kmalloc(PATH_MAX, GFP_KERNEL); + if (!pathbuf) { + SEQ_printf(m, "Show memfs failed due to OOM\n"); + return; + } + pfc.pathbuf = pathbuf; + pfc.pathbuf_size = PATH_MAX; + + for (i = 0; i < ARRAY_SIZE(fs_type_names); i++) { + pfc.fs_type_name = fs_type_names[i]; + pfc.total_print_files = 0; + pfc.total_files_size = 0; + + SEQ_printf(m, "Show %s files (memory-size > %lukB):\n", + pfc.fs_type_name, pfc.size_threshold >> 10); + SEQ_printf(m, "<memory-size> <file-size> <path>\n"); + iterate_supers(memfs_show_files_in_mem_cgroup, &pfc); + + SEQ_printf(m, "total files: %lu, total memory-size: %lukB\n", + pfc.total_print_files, pfc.total_files_size >> 10); + } + + kfree(pfc.pathbuf); +} + +int mem_cgroup_memfs_files_show(struct seq_file *m, void *v) +{ + struct mem_cgroup *memcg = mem_cgroup_from_css(seq_css(m)); + + mem_cgroup_print_memfs_info(memcg, m); + return 0; +} + +static ssize_t memfs_size_threshold_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *buf) +{ + return sprintf(buf, "%lu\n", memfs_size_threshold); +} + +static ssize_t memfs_size_threshold_store(struct kobject *kobj, + struct kobj_attribute *attr, + const char *buf, size_t len) +{ + unsigned long count; + int err; + + err = kstrtoul(buf, 10, &count); + if (err) + return err; + memfs_size_threshold = count; + return len; +} + +static struct kobj_attribute memfs_size_threshold_attr = { + .attr = {"size_threshold", 0644}, + .show = &memfs_size_threshold_show, + .store = &memfs_size_threshold_store, +}; + +static ssize_t memfs_max_print_files_show(struct kobject *kobj, + struct kobj_attribute *attr, + char *buf) +{ + return sprintf(buf, "%lu\n", memfs_max_print_files); +} + +static ssize_t memfs_max_print_files_store(struct kobject *kobj, + struct kobj_attribute *attr, + const char *buf, size_t len) +{ + unsigned long count; + int err; + + err = kstrtoul(buf, 10, &count); + if (err) + return err; + memfs_max_print_files = count; + return len; +} + +static struct kobj_attribute memfs_max_print_files_attr = { + .attr = {"max_print_files_in_oom", 0644}, + .show = &memfs_max_print_files_show, + .store = &memfs_max_print_files_store, +}; + +static ssize_t memfs_enable_show(struct kobject *kobj, + struct kobj_attribute *attr, char *buf) +{ + return sprintf(buf, "%u\n", memfs_enable); +} + +static ssize_t memfs_enable_store(struct kobject *kobj, + struct kobj_attribute *attr, + const char *buf, size_t len) +{ + bool enable; + int err; + + err = kstrtobool(buf, &enable); + if (err) + return err; + + memfs_enable = enable; + return len; +} + +static struct kobj_attribute memfs_enable_attr = { + .attr = {"enable", 0644}, + .show = &memfs_enable_show, + .store = &memfs_enable_store, +}; + +static struct attribute *memfs_attr[] = { + &memfs_size_threshold_attr.attr, + &memfs_max_print_files_attr.attr, + &memfs_enable_attr.attr, + NULL, +}; + +static struct attribute_group memfs_attr_group = { + .attrs = memfs_attr, +}; + +void mem_cgroup_memfs_info_init(void) +{ + struct kobject *memcg_memfs_kobj; + + if (mem_cgroup_disabled()) + return; + + memcg_memfs_kobj = kobject_create_and_add("memcg_memfs_info", mm_kobj); + if (unlikely(!memcg_memfs_kobj)) { + pr_err("failed to create memcg_memfs kobject\n"); + return; + } + + if (sysfs_create_group(memcg_memfs_kobj, &memfs_attr_group)) { + pr_err("failed to register memcg_memfs group\n"); + kobject_put(memcg_memfs_kobj); + } +} diff --git a/mm/memcontrol.c b/mm/memcontrol.c index 974fc5dc6dc81..18b5660dc2459 100644 --- a/mm/memcontrol.c +++ b/mm/memcontrol.c @@ -66,6 +66,7 @@ #include <linux/lockdep.h> #include <linux/file.h> #include <linux/tracehook.h> +#include <linux/memcg_memfs_info.h> #include "internal.h" #include <net/sock.h> #include <net/ip.h> @@ -1484,6 +1485,8 @@ void mem_cgroup_print_oom_meminfo(struct mem_cgroup *memcg) pr_cont("\n"); } + + mem_cgroup_print_memfs_info(memcg, NULL); } /* @@ -4705,6 +4708,12 @@ static struct cftype mem_cgroup_legacy_files[] = { .write_s64 = memcg_qos_write, }, #endif +#ifdef CONFIG_MEMCG_MEMFS_INFO + { + .name = "memfs_files_info", + .seq_show = mem_cgroup_memfs_files_show, + }, +#endif #ifdef CONFIG_NUMA { .name = "numa_stat", @@ -6916,6 +6925,8 @@ static int __init mem_cgroup_init(void) soft_limit_tree.rb_tree_per_node[node] = rtpn; } + mem_cgroup_memfs_info_init(); + return 0; } subsys_initcall(mem_cgroup_init); -- 2.25.1

1 1