lkml.org 
[lkml]   [2023]   [Jan]   [11]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: [PATCH] uprobes: list all active uprobes in the system
From

在 1/12/2023 5:35 AM, Ian Rogers 写道:
> On Wed, Jan 11, 2023 at 2:49 AM <luolongjun@huawei.com> wrote:
>> From: Longjun Luo <luolongjun@huawei.com>
>>
>> Since uprobes will replace instructions in the process
>> memory, it is necessary to provide one way to list
>> all active uprobes. One can access this file through
>> /sys/kernel/debug/uprobes/list.
>>
>> Output looks like this
>> =====================
>> inode+offset/ref_ctr_offset 4160760670+0x30a10/0x0
>> ref 1
>> register_rwsem (unlocked)
>> consumer_rwsem (unlocked)
>> consumers-1
>> handler 0xffffffffc03ee02e(handler) uprobe
>> ret_handler 0x0000000000000000( )
>> filter 0xffffffffc03ee010(filter) uprobe
>> consumers-2
>> handler 0xffffffffc03e902e(handler_1) uprobe_1
>> ret_handler 0x0000000000000000( )
>> filter 0xffffffffc03e9010(filter_1) uprobe_1
>>
>> Signed-off-by: Longjun Luo <luolongjun@huawei.com>
>> ---
>> kernel/events/uprobes.c | 120 ++++++++++++++++++++++++++++++++++++----
>> 1 file changed, 109 insertions(+), 11 deletions(-)
>>
>> diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c
>> index d9e357b7e17c..2fa9b910abc7 100644
>> --- a/kernel/events/uprobes.c
>> +++ b/kernel/events/uprobes.c
>> @@ -27,6 +27,9 @@
>> #include <linux/task_work.h>
>> #include <linux/shmem_fs.h>
>> #include <linux/khugepaged.h>
>> +#include <linux/rbtree.h>
>> +#include <linux/debugfs.h>
>> +#include <linux/kallsyms.h>
>>
>> #include <linux/uprobes.h>
>>
>> @@ -40,7 +43,7 @@ static struct rb_root uprobes_tree = RB_ROOT;
>> */
>> #define no_uprobe_events() RB_EMPTY_ROOT(&uprobes_tree)
>>
>> -static DEFINE_SPINLOCK(uprobes_treelock); /* serialize rbtree access */
>> +static DECLARE_RWSEM(uprobes_treelock); /* serialize rbtree access */
>>
>> #define UPROBES_HASH_SZ 13
>> /* serialize uprobe->pending_list */
>> @@ -672,9 +675,9 @@ static struct uprobe *find_uprobe(struct inode *inode, loff_t offset)
>> {
>> struct uprobe *uprobe;
>>
>> - spin_lock(&uprobes_treelock);
>> + down_read(&uprobes_treelock);
>> uprobe = __find_uprobe(inode, offset);
>> - spin_unlock(&uprobes_treelock);
>> + up_read(&uprobes_treelock);
>>
>> return uprobe;
>> }
>> @@ -704,9 +707,9 @@ static struct uprobe *insert_uprobe(struct uprobe *uprobe)
>> {
>> struct uprobe *u;
>>
>> - spin_lock(&uprobes_treelock);
>> + down_write(&uprobes_treelock);
>> u = __insert_uprobe(uprobe);
>> - spin_unlock(&uprobes_treelock);
>> + up_write(&uprobes_treelock);
>>
>> return u;
>> }
>> @@ -938,9 +941,9 @@ static void delete_uprobe(struct uprobe *uprobe)
>> if (WARN_ON(!uprobe_is_active(uprobe)))
>> return;
>>
>> - spin_lock(&uprobes_treelock);
>> + down_write(&uprobes_treelock);
>> rb_erase(&uprobe->rb_node, &uprobes_tree);
>> - spin_unlock(&uprobes_treelock);
>> + up_write(&uprobes_treelock);
>> RB_CLEAR_NODE(&uprobe->rb_node); /* for uprobe_is_active() */
>> put_uprobe(uprobe);
>> }
>> @@ -1301,7 +1304,7 @@ static void build_probe_list(struct inode *inode,
>> min = vaddr_to_offset(vma, start);
>> max = min + (end - start) - 1;
>>
>> - spin_lock(&uprobes_treelock);
>> + down_read(&uprobes_treelock);
>> n = find_node_in_range(inode, min, max);
>> if (n) {
>> for (t = n; t; t = rb_prev(t)) {
>> @@ -1319,7 +1322,7 @@ static void build_probe_list(struct inode *inode,
>> get_uprobe(u);
>> }
>> }
>> - spin_unlock(&uprobes_treelock);
>> + up_read(&uprobes_treelock);
>> }
>>
>> /* @vma contains reference counter, not the probed instruction. */
>> @@ -1410,9 +1413,9 @@ vma_has_uprobes(struct vm_area_struct *vma, unsigned long start, unsigned long e
>> min = vaddr_to_offset(vma, start);
>> max = min + (end - start) - 1;
>>
>> - spin_lock(&uprobes_treelock);
>> + down_read(&uprobes_treelock);
>> n = find_node_in_range(inode, min, max);
>> - spin_unlock(&uprobes_treelock);
>> + up_read(&uprobes_treelock);
>>
>> return !!n;
>> }
>> @@ -2357,3 +2360,98 @@ void __init uprobes_init(void)
>>
>> BUG_ON(register_die_notifier(&uprobe_exception_nb));
>> }
>> +
>> +#ifdef CONFIG_DEBUG_FS
>> +
>> +static void *uprobe_seq_start(struct seq_file *m, loff_t *pos)
>> +{
>> + loff_t num = *pos;
>> + struct rb_node *rbp;
>> +
>> + down_read(&uprobes_treelock);
> Is there a missing corresponding up_read?
>
> Thanks,
> Ian
This missing corresponding up_read is in the uprobe_seq_stop.

In my understanding, uprobe_seq_start and uprobe_seq_stop is one pair of
API.

uprobe_seq_start holds the lock and the uprobe_seq_stop releases it.

I am not sure if it is right.

By the way, may I ask if the community still plan to evolve the uprobe
since the uprobe hasn't changed for a long time.
>> + for (rbp = rb_first(&uprobes_tree); rbp; rbp = rb_next(rbp)) {
>> + if (num-- == 0)
>> + return rbp;
>> + }
>> +
>> + return NULL;
>> +}
>> +
>> +static void *uprobe_seq_next(struct seq_file *f, void *v, loff_t *pos)
>> +{
>> + ++*pos;
>> + return (void *)rb_next(v);
>> +}
>> +
>> +static void uprobe_seq_stop(struct seq_file *f, void *v)
>> +{
>> + up_read(&uprobes_treelock);
>> +}
>> +
>> +static int show_uprobe(struct seq_file *m, void *v)
>> +{
>> + struct uprobe *uprobe = rb_entry_safe(v, struct uprobe, rb_node);
>> + struct uprobe_consumer *con;
>> + char *modname, namebuf[KSYM_NAME_LEN];
>> + const char *sym;
>> + unsigned int cnt = 1;
>> +
>> + seq_printf(m, "inode+offset/ref_ctr_offset %lu+0x%llx/0x%llx\n",
>> + uprobe->inode->i_ino, uprobe->offset, uprobe->ref_ctr_offset);
>> + seq_printf(m, " ref %u\n",
>> + refcount_read(&uprobe->ref));
>> + seq_printf(m, " register_rwsem %s\n",
>> + rwsem_is_locked(&uprobe->register_rwsem)?"(locked)":"(unlocked)");
>> + seq_printf(m, " consumer_rwsem %s\n",
>> + rwsem_is_locked(&uprobe->consumer_rwsem)?"(locked)":"(unlocked)");
>> +
>> + /* lock consumer_rwsem when hold uprobes_treelock */
>> + down_read(&uprobe->consumer_rwsem);
>> + for (con = uprobe->consumers; con; con = con->next, cnt++) {
>> + seq_printf(m, " consumers-%u\n", cnt);
>> +
>> + modname = NULL;
>> + sym = kallsyms_lookup((unsigned long)con->handler, NULL, NULL,
>> + &modname, namebuf);
>> + seq_printf(m, " handler 0x%px(%s) %s\n",
>> + con->handler, (sym ? sym : " "), (modname ? modname : " "));
>> +
>> + modname = NULL;
>> + sym = kallsyms_lookup((unsigned long)con->ret_handler, NULL, NULL,
>> + &modname, namebuf);
>> + seq_printf(m, " ret_handler 0x%px(%s) %s\n",
>> + con->ret_handler, (sym ? sym : " "), (modname ? modname : " "));
>> +
>> + modname = NULL;
>> + sym = kallsyms_lookup((unsigned long)con->filter, NULL, NULL,
>> + &modname, namebuf);
>> + seq_printf(m, " filter 0x%px(%s) %s\n",
>> + con->filter, (sym ? sym : " "), (modname ? modname : " "));
>> + }
>> + up_read(&uprobe->consumer_rwsem);
>> +
>> + return 0;
>> +}
>> +
>> +static const struct seq_operations uprobes_sops = {
>> + .start = uprobe_seq_start,
>> + .next = uprobe_seq_next,
>> + .stop = uprobe_seq_stop,
>> + .show = show_uprobe,
>> +};
>> +
>> +DEFINE_SEQ_ATTRIBUTE(uprobes);
>> +
>> +static int __init debugfs_uprobe_init(void)
>> +{
>> + struct dentry *dir;
>> +
>> + dir = debugfs_create_dir("uprobes", NULL);
>> +
>> + debugfs_create_file("list", 0400, dir, NULL, &uprobes_fops);
>> +
>> + return 0;
>> +}
>> +late_initcall(debugfs_uprobe_init);
>> +
>> +#endif /* CONFIG_DEBUG_FS */
>> --
>> 2.38.1
>>

\
 
 \ /
  Last update: 2023-03-26 23:37    [W:0.194 / U:0.184 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site