| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <80d54ab2864e4011a9f5e5b198ccfe8e@huawei.com>
Date: Mon, 30 May 2022 13:00:24 +0000
From: "liubo (AW)" <liubo254@...wei.com>
To: "akpm@...ux-foundation.org" <akpm@...ux-foundation.org>,
"linux-mm@...ck.org" <linux-mm@...ck.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
CC: "ying.huang@...el.com" <ying.huang@...el.com>,
"willy@...radead.org" <willy@...radead.org>,
"vbabka@...e.cz" <vbabka@...e.cz>,
"surenb@...gle.com" <surenb@...gle.com>,
"peterx@...hat.com" <peterx@...hat.com>,
"neilb@...e.de" <neilb@...e.de>,
"naoya.horiguchi@....com" <naoya.horiguchi@....com>,
"minchan@...nel.org" <minchan@...nel.org>,
linmiaohe <linmiaohe@...wei.com>,
Louhongxiang <louhongxiang@...wei.com>,
linfeilong <linfeilong@...wei.com>
Subject: 答复: [PATCH] mm/swapfile: release swap info when swap device is unpluged
friendly ping.
-----邮件原件-----
发件人: liubo (AW)
发送时间: 2022年5月28日 16:50
收件人: akpm@...ux-foundation.org; linux-mm@...ck.org; linux-kernel@...r.kernel.org
抄送: ying.huang@...el.com; willy@...radead.org; vbabka@...e.cz; surenb@...gle.com; peterx@...hat.com; neilb@...e.de; naoya.horiguchi@....com; minchan@...nel.org; linmiaohe <linmiaohe@...wei.com>; Louhongxiang <louhongxiang@...wei.com>; linfeilong <linfeilong@...wei.com>; liubo (AW) <liubo254@...wei.com>
主题: [PATCH] mm/swapfile: release swap info when swap device is unpluged
When the swap partition is mounted through the swapon command, the kernel will create the swap_info_struct data structure and initialize it, and save it in the swap_info global array.
When the swap partition is no longer in use, the disk is unloaded through the swapoff command.
However, if the disk is pulled out after swapon, an error will occur when swapoff the disk, causing the swap_info_struct data structure to remain in the kernel and cannot be cleared.
This patch identifies which disks are no longer available by adding a traversal operation for swap_active_head available swap partitions in the swapon and swapoff processes, so as to clear the above data structures and release the corresponding resources.
Example:
[root@...alhost ~]# swapon -s
[root@...alhost ~]# lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sda 8:0 0 1.1T 0 disk
├─sda1 8:1 0 600M 0 part /boot/efi
├─sda2 8:2 0 1G 0 part /boot
└─sda3 8:3 0 1.1T 0 part
├─root 253:0 0 70G 0 lvm /
├─swap 253:1 0 4G 0 lvm
└─home 253:2 0 1T 0 lvm /home
nvme0n1 259:0 0 3.6T 0 disk
└─nvme0n1p1 259:5 0 60G 0 part
[root@...alhost ~]# swapon /dev/nvme0n1p1 [root@...alhost ~]# swapon -s
Filename Type Size Used Priority
/dev/nvme0n1p1 partition 62914556 0 -2
[root@...alhost ~]# echo 1 > /sys/bus/pci/devices/0000:d8:00.0/remove
[root@...alhost ~]# lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sda 8:0 0 1.1T 0 disk
├─sda1 8:1 0 600M 0 part /boot/efi
├─sda2 8:2 0 1G 0 part /boot
└─sda3 8:3 0 1.1T 0 part
├─root 253:0 0 70G 0 lvm /
├─swap 253:1 0 4G 0 lvm
└─home 253:2 0 1T 0 lvm /home
[root@...alhost ~]# swapon -s
Filename Type Size Used Priority
/dev/nvme0n1p1 partition 62914556 0 -2
[root@...alhost ~]# swapoff /dev/nvme0n1p1
swapoff: /dev/nvme0n1p1: swapoff failed: No such file or directory [root@...alhost ~]# swapoff -a [root@...alhost ~]# swapon -s
Filename Type Size Used Priority
/dev/nvme0n1p1 partition 62914556 0 -2
In the swapoff command, the device is acquired in the following ways, but the device has been unplugged at this time, causing the "victim"
acquisition to fail, thus returning an error directly.
And the invalid swap_info_struct cannot be effectively released.
pathname = getname(specialfile);
if (IS_ERR(pathname))
return PTR_ERR(pathname);
victim = file_open_name(pathname, O_RDWR|O_LARGEFILE, 0); err = PTR_ERR(victim); if (IS_ERR(victim))
goto out;
In order to solve the above problems, by adding traversal of swap_avail_heads (available swap partitions) in the swapoff and swapon processes, find the swap_info_struct whose disk partition has been unplugged, and release resources.
The reason why the judgment of unavailable swap information is also added to the swapon process is that the swapoff is executed by the user, and the timing is uncontrollable.
The system supports swapon multiple disks, and the unavailable swap can be deleted at the same time as swapon is mounted.
In order to realize the interface reuse in the swapoff resource release process, some of the operations are abstracted into separate interfaces.
del_useless_swap_info():
Remove specific swap_info_struct from swap_active_head and update total_swap_pages.
release_swap_info_memory():
Clear the corresponding resources of swap_info_struct.
swapoff_invalid_swapinfo():
Traverse the swap_avail_heads list to release the invalid swap area resources.
Signed-off-by: liubo <liubo254@...wei.com>
---
mm/swapfile.c | 262 +++++++++++++++++++++++++++++++++++---------------
1 file changed, 182 insertions(+), 80 deletions(-)
diff --git a/mm/swapfile.c b/mm/swapfile.c index a2e66d855b19..8d2e75891ff4 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -68,7 +68,7 @@ static const char Bad_file[] = "Bad swap file entry "; static const char Unused_file[] = "Unused swap file entry "; static const char Bad_offset[] = "Bad swap offset entry "; static const char Unused_offset[] = "Unused swap offset entry ";
-
+static const char invalid_info[] = "deleted";
/*
* all active swap_info_structs
* protected with swap_lock, and ordered by priority.
@@ -2384,18 +2384,184 @@ bool has_usable_swap(void)
return ret;
}
-SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
+static void release_swap_info_memory(struct swap_info_struct *p)
{
- struct swap_info_struct *p = NULL;
unsigned char *swap_map;
struct swap_cluster_info *cluster_info;
unsigned long *frontswap_map;
- struct file *swap_file, *victim;
+ struct file *swap_file;
struct address_space *mapping;
struct inode *inode;
+ unsigned int old_block_size;
+
+ mutex_lock(&swapon_mutex);
+ spin_lock(&swap_lock);
+ spin_lock(&p->lock);
+ drain_mmlist();
+
+ /* wait for anyone still in scan_swap_map */
+ p->highest_bit = 0; /* cuts scans short */
+ while (p->flags >= SWP_SCANNING) {
+ spin_unlock(&p->lock);
+ spin_unlock(&swap_lock);
+ schedule_timeout_uninterruptible(1);
+ spin_lock(&swap_lock);
+ spin_lock(&p->lock);
+ }
+
+ swap_file = p->swap_file;
+ mapping = p->swap_file->f_mapping;
+ old_block_size = p->old_block_size;
+ p->swap_file = NULL;
+ p->max = 0;
+ swap_map = p->swap_map;
+ p->swap_map = NULL;
+ cluster_info = p->cluster_info;
+ p->cluster_info = NULL;
+ frontswap_map = frontswap_map_get(p);
+ spin_unlock(&p->lock);
+ spin_unlock(&swap_lock);
+ arch_swap_invalidate_area(p->type);
+ frontswap_invalidate_area(p->type);
+ frontswap_map_set(p, NULL);
+ mutex_unlock(&swapon_mutex);
+ free_percpu(p->percpu_cluster);
+ p->percpu_cluster = NULL;
+ free_percpu(p->cluster_next_cpu);
+ p->cluster_next_cpu = NULL;
+ vfree(swap_map);
+ kvfree(cluster_info);
+ kvfree(frontswap_map);
+ /* Destroy swap account information */
+ swap_cgroup_swapoff(p->type);
+ exit_swap_address_space(p->type);
+
+ inode = mapping->host;
+ if (S_ISBLK(inode->i_mode)) {
+ struct block_device *bdev = I_BDEV(inode);
+
+ set_blocksize(bdev, old_block_size);
+ blkdev_put(bdev, FMODE_READ | FMODE_WRITE | FMODE_EXCL);
+ }
+
+ inode_lock(inode);
+ inode->i_flags &= ~S_SWAPFILE;
+ inode_unlock(inode);
+ filp_close(swap_file, NULL);
+}
+
+static void del_useless_swap_info(struct swap_info_struct *p) {
+ del_from_avail_list(p);
+ spin_lock(&p->lock);
+ if (p->prio < 0) {
+ struct swap_info_struct *si = p;
+ int nid;
+
+ plist_for_each_entry_continue(si, &swap_active_head, list) {
+ si->prio++;
+ si->list.prio--;
+ for_each_node(nid) {
+ if (si->avail_lists[nid].prio != 1)
+ si->avail_lists[nid].prio--;
+ }
+ }
+ least_priority++;
+ }
+ plist_del(&p->list, &swap_active_head);
+ atomic_long_sub(p->pages, &nr_swap_pages);
+ total_swap_pages -= p->pages;
+ p->flags &= ~SWP_WRITEOK;
+ spin_unlock(&p->lock);
+}
+
+static int swapoff_invalid_swapinfo(void) {
+ struct swap_info_struct *p = NULL;
+ struct file *swap_file;
+ int err, found = 0;
+
+ char *tmp = NULL;
+ char *swap_name = NULL;
+
+ tmp = kvzalloc(PAGE_SIZE, GFP_KERNEL);
+ if (!tmp)
+ return -ENOMEM;
+rescan:
+ memset(tmp, 0, PAGE_SIZE);
+ spin_lock(&swap_lock);
+ plist_for_each_entry(p, &swap_active_head, list) {
+ if (p->flags & SWP_WRITEOK) {
+ swap_file = p->swap_file;
+ swap_name = d_path(&swap_file->f_path, tmp, PAGE_SIZE);
+
+ if (strstr(swap_name, invalid_info)) {
+ found = 1;
+ break;
+ }
+ }
+ }
+
+ if (!found) {
+ err = 0;
+ spin_unlock(&swap_lock);
+ goto out;
+ }
+
+ del_useless_swap_info(p);
+ spin_unlock(&swap_lock);
+
+ disable_swap_slots_cache_lock();
+ set_current_oom_origin();
+ try_to_unuse(p->type);
+ clear_current_oom_origin();
+
+ reenable_swap_slots_cache_unlock();
+
+ /*
+ * wait for swap operations protected by get/put_swap_device()
+ * to complete
+ */
+ synchronize_rcu();
+
+ flush_work(&p->discard_work);
+
+ destroy_swap_extents(p);
+ if (p->flags & SWP_CONTINUED)
+ free_swap_count_continuations(p);
+
+ if (!p->bdev || !blk_queue_nonrot(bdev_get_queue(p->bdev)))
+ atomic_dec(&nr_rotate_swap);
+
+ release_swap_info_memory(p);
+
+ /*
+ * Clear the SWP_USED flag after all resources are freed so that swapon
+ * can reuse this swap_info in alloc_swap_info() safely. It is ok to
+ * not hold p->lock after we cleared its SWP_WRITEOK.
+ */
+ spin_lock(&swap_lock);
+ p->flags = 0;
+ spin_unlock(&swap_lock);
+
+ err = 0;
+ atomic_inc(&proc_poll_event);
+ wake_up_interruptible(&proc_poll_wait);
+
+ found = 0;
+ goto rescan;
+out:
+ kfree(tmp);
+ return err;
+}
+
+SYSCALL_DEFINE1(swapoff, const char __user *, specialfile) {
+ struct swap_info_struct *p = NULL;
+ struct file *victim;
+ struct address_space *mapping;
struct filename *pathname;
int err, found = 0;
- unsigned int old_block_size;
if (!capable(CAP_SYS_ADMIN))
return -EPERM;
@@ -2408,8 +2574,12 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
victim = file_open_name(pathname, O_RDWR|O_LARGEFILE, 0);
err = PTR_ERR(victim);
- if (IS_ERR(victim))
+ if (IS_ERR(victim)) {
+ /* check if the pathname is a device that has been unpluged */
+ err = swapoff_invalid_swapinfo();
+ err = err < 0 ? err : PTR_ERR(victim);
goto out;
+ }
mapping = victim->f_mapping;
spin_lock(&swap_lock);
@@ -2433,27 +2603,8 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
spin_unlock(&swap_lock);
goto out_dput;
}
- del_from_avail_list(p);
- spin_lock(&p->lock);
- if (p->prio < 0) {
- struct swap_info_struct *si = p;
- int nid;
- plist_for_each_entry_continue(si, &swap_active_head, list) {
- si->prio++;
- si->list.prio--;
- for_each_node(nid) {
- if (si->avail_lists[nid].prio != 1)
- si->avail_lists[nid].prio--;
- }
- }
- least_priority++;
- }
- plist_del(&p->list, &swap_active_head);
- atomic_long_sub(p->pages, &nr_swap_pages);
- total_swap_pages -= p->pages;
- p->flags &= ~SWP_WRITEOK;
- spin_unlock(&p->lock);
+ del_useless_swap_info(p);
spin_unlock(&swap_lock);
disable_swap_slots_cache_lock();
@@ -2491,60 +2642,7 @@ SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
if (!p->bdev || !bdev_nonrot(p->bdev))
atomic_dec(&nr_rotate_swap);
- mutex_lock(&swapon_mutex);
- spin_lock(&swap_lock);
- spin_lock(&p->lock);
- drain_mmlist();
-
- /* wait for anyone still in scan_swap_map_slots */
- p->highest_bit = 0; /* cuts scans short */
- while (p->flags >= SWP_SCANNING) {
- spin_unlock(&p->lock);
- spin_unlock(&swap_lock);
- schedule_timeout_uninterruptible(1);
- spin_lock(&swap_lock);
- spin_lock(&p->lock);
- }
-
- swap_file = p->swap_file;
- old_block_size = p->old_block_size;
- p->swap_file = NULL;
- p->max = 0;
- swap_map = p->swap_map;
- p->swap_map = NULL;
- cluster_info = p->cluster_info;
- p->cluster_info = NULL;
- frontswap_map = frontswap_map_get(p);
- spin_unlock(&p->lock);
- spin_unlock(&swap_lock);
- arch_swap_invalidate_area(p->type);
- frontswap_invalidate_area(p->type);
- frontswap_map_set(p, NULL);
- mutex_unlock(&swapon_mutex);
- free_percpu(p->percpu_cluster);
- p->percpu_cluster = NULL;
- free_percpu(p->cluster_next_cpu);
- p->cluster_next_cpu = NULL;
- vfree(swap_map);
- kvfree(cluster_info);
- kvfree(frontswap_map);
- /* Destroy swap account information */
- swap_cgroup_swapoff(p->type);
- exit_swap_address_space(p->type);
-
- inode = mapping->host;
- if (S_ISBLK(inode->i_mode)) {
- struct block_device *bdev = I_BDEV(inode);
-
- set_blocksize(bdev, old_block_size);
- blkdev_put(bdev, FMODE_READ | FMODE_WRITE | FMODE_EXCL);
- }
-
- inode_lock(inode);
- inode->i_flags &= ~S_SWAPFILE;
- inode_unlock(inode);
- filp_close(swap_file, NULL);
-
+ release_swap_info_memory(p);
/*
* Clear the SWP_USED flag after all resources are freed so that swapon
* can reuse this swap_info in alloc_swap_info() safely. It is ok to @@ -3008,6 +3106,10 @@ SYSCALL_DEFINE2(swapon, const char __user *, specialfile, int, swap_flags)
if (!swap_avail_heads)
return -ENOMEM;
+ error = swapoff_invalid_swapinfo();
+ if (error < 0)
+ return error;
+
p = alloc_swap_info();
if (IS_ERR(p))
return PTR_ERR(p);
--
2.27.0
Powered by blists - more mailing lists