mirrors/scoutfs
1
0
Fork 0
mirror of https://github.com/versity/scoutfs.git synced 2026-08-15 19:56:36 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix warning about missing cluster lock during readahead

Browse Source 
scoutfs_readahead() will acquire the cluster read lock if the task
doesn't already hold it. But it doesn't set the corresponding
per-task variable, so the scoutfs_get_block() call won't find
it and mistakenly warns that we aren't holding the lock.

We need to set the per-task variable in scoutfs_readahead(). And since
we can get into this path from scoutfs_fadvise(), and that function
acquires the extent_sem, we also need to first acquire the cluster
read lock and set the per-task variable there too. This fixes a
lock ordering issue that we had not yet seen.

Signed-off-by: Chris Kirby <ckirby@versity.com>
This commit is contained in:
Chris Kirby
2026-07-30 09:36:22 -05:00
parent 9054b6fb60
commit c9fd300f7d
1 changed files with 25 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
kmod/src/data.c
+25 -3
View File
@@ -779,6 +779,7 @@ static void scoutfs_readahead(struct readahead_control *rac)
{
struct inode *inode = rac->file->f_inode;
struct scoutfs_inode_info *si = SCOUTFS_I(inode);
SCOUTFS_DECLARE_PER_TASK_ENTRY(pt_data_ent);
struct super_block *sb = inode->i_sb;
struct scoutfs_lock *inode_lock = NULL;
bool found_lock;
@@ -791,6 +792,7 @@ static void scoutfs_readahead(struct readahead_control *rac)
if (ret)
return;
found_lock = false;
scoutfs_per_task_add(&si->pt_data_lock, &pt_data_ent, inode_lock);
} else {
found_lock = true;
}
@@ -799,11 +801,14 @@ static void scoutfs_readahead(struct readahead_control *rac)
readahead_length(rac), SEF_OFFLINE,
SCOUTFS_IOC_DWO_READ, NULL,
inode_lock);
if (ret == 0)
mpage_readahead(rac, scoutfs_get_block_read);
if (!found_lock)
if (!found_lock) {
scoutfs_per_task_del(&si->pt_data_lock, &pt_data_ent);
scoutfs_unlock(sb, inode_lock, SCOUTFS_LOCK_READ);
}
}
static int scoutfs_writepage(struct page *page, struct writeback_control *wbc)
@@ -1177,21 +1182,35 @@ static int scoutfs_fadvise(struct file *file, loff_t start, loff_t end, int advi
{
struct inode *inode = file_inode(file);
struct scoutfs_inode_info *si = SCOUTFS_I(inode);
struct super_block *sb = inode->i_sb;
SCOUTFS_DECLARE_PER_TASK_ENTRY(pt_extent_ent);
SCOUTFS_DECLARE_PER_TASK_ENTRY(pt_data_ent);
struct scoutfs_lock *inode_lock = NULL;
bool locked = false;
int ret;
/*
* We need to get the extent_sem now, or we'll be out of order with the
* mapping.invalidate_lock.
* mapping.invalidate_lock. This also means we need to get the cluster
* read lock now, or we'll be out of order with it in scoutfs_readahead().
*/
if (advice == POSIX_FADV_WILLNEED) {
ret = scoutfs_lock_inode(sb, SCOUTFS_LOCK_READ, SCOUTFS_LKF_REFRESH_INODE,
inode, &inode_lock);
if (ret)
return ret;
if (!scoutfs_per_task_add_excl(&si->pt_data_lock, &pt_data_ent,
inode_lock))
WARN_ON_ONCE(true);
down_read(&si->extent_sem);
locked = true;
if (!scoutfs_per_task_add_excl(&si->pt_extent_sem, &pt_extent_ent,
&pt_extent_ent))
WARN_ON_ONCE(true);
locked = true;
}
ret = generic_fadvise(file, start, end, advice);
@@ -1199,6 +1218,9 @@ static int scoutfs_fadvise(struct file *file, loff_t start, loff_t end, int advi
if (locked) {
scoutfs_per_task_del(&si->pt_extent_sem, &pt_extent_ent);
up_read(&si->extent_sem);
scoutfs_per_task_del(&si->pt_data_lock, &pt_data_ent);
scoutfs_unlock(sb, inode_lock, SCOUTFS_LOCK_READ);
}
return ret;