Skip to contentSkip to conversation

FUSE kmemleak report

DISCUSSION Sitsofe Wheeler 3 participants 4 messages Sep 5, 2011 – Sep 6, 2011 FUSE filesystem core

lore
Share & tools
lore
#0 lore raw

Hi,

kmemleak is reporting that 32 bytes are being leaked by FUSE:

unreferenced object 0xe373b270 (size 32):
comm "fusermount", pid 1207, jiffies 4294707026 (age 2675.187s) hex dump (first 32 bytes):

    01 00 00 00 00 00 00 00 01 00 00 00 00 00 00 00  ................
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  backtrace:
    [<b05517d7>] kmemleak_alloc+0x27/0x50
    [<b0196435>] kmem_cache_alloc+0xc5/0x180
    [<b02455be>] fuse_alloc_forget+0x1e/0x20
    [<b0245670>] fuse_alloc_inode+0xb0/0xd0
    [<b01b1a8c>] alloc_inode+0x1c/0x80
    [<b01b290f>] iget5_locked+0x8f/0x1a0
    [<b0246022>] fuse_iget+0x72/0x1a0
    [<b02461da>] fuse_get_root_inode+0x8a/0x90
    [<b02465cf>] fuse_fill_super+0x3ef/0x590
    [<b019e56f>] mount_nodev+0x3f/0x90
    [<b0244e95>] fuse_mount+0x15/0x20
    [<b019d1bc>] mount_fs+0x1c/0xc0
    [<b01b5811>] vfs_kern_mount+0x41/0x90
    [<b01b5af9>] do_kern_mount+0x39/0xd0
    [<b01b7585>] do_mount+0x2e5/0x660
    [<b01b7966>] sys_mount+0x66/0xa0

This leak report is consistent and happens once per boot on 3.1.0-rc5-dirty.

Signature · 1 line
Sitsofe | http://sucs.org/~sits/
#1 lore raw
Quoted 25 lines · Sitsofe Wheeler

kmemleak is reporting that 32 bytes are being leaked by FUSE:

unreferenced object 0xe373b270 (size 32):
 comm "fusermount", pid 1207, jiffies 4294707026 (age 2675.187s) hex dump (first 32 bytes):
   01 00 00 00 00 00 00 00 01 00 00 00 00 00 00 00  ................ 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................ backtrace:

   [<b05517d7>] kmemleak_alloc+0x27/0x50
   [<b0196435>] kmem_cache_alloc+0xc5/0x180
   [<b02455be>] fuse_alloc_forget+0x1e/0x20
   [<b0245670>] fuse_alloc_inode+0xb0/0xd0
   [<b01b1a8c>] alloc_inode+0x1c/0x80
   [<b01b290f>] iget5_locked+0x8f/0x1a0
   [<b0246022>] fuse_iget+0x72/0x1a0
   [<b02461da>] fuse_get_root_inode+0x8a/0x90
   [<b02465cf>] fuse_fill_super+0x3ef/0x590
   [<b019e56f>] mount_nodev+0x3f/0x90
   [<b0244e95>] fuse_mount+0x15/0x20
   [<b019d1bc>] mount_fs+0x1c/0xc0
   [<b01b5811>] vfs_kern_mount+0x41/0x90
   [<b01b5af9>] do_kern_mount+0x39/0xd0
   [<b01b7585>] do_mount+0x2e5/0x660
   [<b01b7966>] sys_mount+0x66/0xa0
Quoted 2 lines

This leak report is consistent and happens once per boot on 3.1.0-rc5-dirty.

IIUC, kmemleak reports the fuse_forget_link that corresponds to the root inode (allocated via fuse_fill_super). The root inode is later freed (did the mount failed?) but not the forget link. I'm not familiar with this code, so note exactly sure where the forget link should have been freed.

Signature · 1 line
Catalin
maintainer
#2 lore raw
Quoted 26 lines · Catalin Marinas
Quoted 25 lines · Sitsofe Wheeler

kmemleak is reporting that 32 bytes are being leaked by FUSE:

unreferenced object 0xe373b270 (size 32):
 comm "fusermount", pid 1207, jiffies 4294707026 (age 2675.187s) hex dump (first 32 bytes):
   01 00 00 00 00 00 00 00 01 00 00 00 00 00 00 00  ................ 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................ backtrace:

   [<b05517d7>] kmemleak_alloc+0x27/0x50
   [<b0196435>] kmem_cache_alloc+0xc5/0x180
   [<b02455be>] fuse_alloc_forget+0x1e/0x20
   [<b0245670>] fuse_alloc_inode+0xb0/0xd0
   [<b01b1a8c>] alloc_inode+0x1c/0x80
   [<b01b290f>] iget5_locked+0x8f/0x1a0
   [<b0246022>] fuse_iget+0x72/0x1a0
   [<b02461da>] fuse_get_root_inode+0x8a/0x90
   [<b02465cf>] fuse_fill_super+0x3ef/0x590
   [<b019e56f>] mount_nodev+0x3f/0x90
   [<b0244e95>] fuse_mount+0x15/0x20
   [<b019d1bc>] mount_fs+0x1c/0xc0
   [<b01b5811>] vfs_kern_mount+0x41/0x90
   [<b01b5af9>] do_kern_mount+0x39/0xd0
   [<b01b7585>] do_mount+0x2e5/0x660
   [<b01b7966>] sys_mount+0x66/0xa0
Quoted 8 lines
Quoted 2 lines

This leak report is consistent and happens once per boot on 3.1.0-rc5-dirty.

IIUC, kmemleak reports the fuse_forget_link that corresponds to the root inode (allocated via fuse_fill_super). The root inode is later freed (did the mount failed?) but not the forget link. I'm not familiar with this code, so note exactly sure where the forget link should have been freed.

Sitsofe, thanks for the report.

Could you please try the patch below?

Thanks,
Miklos

fs/fuse/dev.c +8 -4
index 168a80f..5cb8614 100644
@@ -258,10 +258,14 @@ void fuse_queue_forget(struct fuse_conn *fc, struct fuse_forget_link *forget,
 	forget->forget_one.nlookup = nlookup;
 
 	spin_lock(&fc->lock);
-	fc->forget_list_tail->next = forget;
-	fc->forget_list_tail = forget;
-	wake_up(&fc->waitq);
-	kill_fasync(&fc->fasync, SIGIO, POLL_IN);
+	if (fc->connected) {
+		fc->forget_list_tail->next = forget;
+		fc->forget_list_tail = forget;
+		wake_up(&fc->waitq);
+		kill_fasync(&fc->fasync, SIGIO, POLL_IN);
+	} else {
+		kfree(forget);
+	}
 	spin_unlock(&fc->lock);
 }
 
author Tested-by
#3 lore raw
Quoted 4 lines · Miklos Szeredi

Sitsofe, thanks for the report.

Could you please try the patch below?

You're welcome. The patch resolves the leak for me on 3.1.0-rc5 on Ubuntu 10.04 using GNOME 2.30.2 (I could previously reproduce it my logging in as one user, logging out, logging in as another user and then as root doing echo scan > /sys/kernel/debug/kmemleak ).

Signature · 1 line
Sitsofe | http://sucs.org/~sits/