Skip to contentSkip to conversation

sysfs_dir_cache growing out of control

DISCUSSION Joel Fuster 4 participants 7 messages Aug 23, 2007 – Sep 5, 2007

lore
Share & tools
lore
#0 lore raw

Hi,

I am running 2.6.22.3. For reasons that escape me, over time (days) the sysfs_dir_cache, dentry, and inode_cache SLUB entries grow until they consume all the memory on my system, requiring a reboot.

Although I did not record objective evidence of this at the time, I suspect I had the same problem with 2.6.18 and 2.6.20...I certainly had the same symptoms.

Here is some hopefully useful information. Let me know what else would be helpful.

Thanks.

P.S. Please keep me CC'd as I do not subscribe to the list.

Linux periphas 2.6.22.3 #1 SMP Mon Aug 20 21:47:51 EDT 2007 x86_64 GNU/Linux

*************************************
18:30:02 up 21:27, 4 users, load average: 0.22, 0.36, 0.35

MemTotal:      1026056 kB
MemFree:          8996 kB
Buffers:           176 kB
Cached:         209524 kB
SwapCached:      32308 kB
Active:         610448 kB
Inactive:       170004 kB
SwapTotal:     1048568 kB
SwapFree:       710260 kB
Dirty:            2444 kB
Writeback:           0 kB
AnonPages:      568988 kB
Mapped:          34360 kB
Slab:           210664 kB
SReclaimable:   179460 kB
SUnreclaim:      31204 kB
PageTables:      10008 kB
NFS_Unstable:        0 kB
Bounce:              0 kB
CommitLimit:   1561596 kB
Committed_AS:  1184628 kB
VmallocTotal: 34359738367 kB
VmallocUsed:      1032 kB
VmallocChunk: 34359737307 kB
Name                   Objects Objsize    Space Slabs/Part/Cpu  O/S O %Fr %Ef Flg
:0000016                  2937      16    65.5K         16/5/2  256 0 31  71 *
:0000024                  1222      24    32.7K          8/0/2  170 0 0  89 *
:0000032                  2095      32    73.7K         18/3/2  128 0 16  90 *
:0000040                  1862      40   172.0K        42/31/2  102 0 73  43 *
:0000064                  4651      64   434.1K       106/46/2   64 0 43  68 *
:0000072                    81      72     8.1K          2/0/2   56 0 0  71 *
:0000096                   937      96   135.1K        33/14/2   42 0 42  66 *
:0000112                   144     112    16.3K          4/0/1   36 0 0  98 *
:0000128                   595     128   118.7K        29/19/2   32 0 65  64 *
:0000192                   807     192   262.1K        64/36/2   21 0 56  59 *
:0000256                  3504     256   897.0K        219/0/2   16 0 0 100 *
:0000320                    47     320    16.3K          4/1/2   12 0 25  91 *A
:0000384                    56     384    28.6K          7/1/2   10 0 14  75 *A
:0000512                   238     512   135.1K         33/7/2    8 0 21  90 *
:0000704                   133     704   114.6K         14/7/2   11 1 50  81 *A
:0000768                   180     744   159.7K         39/9/2    5 0 23  83 *A
:0000832                    90     832   114.6K        14/11/2    9 1 78  65 *A
:0001024                   370    1024   393.2K         96/9/2    4 0 9  96 *
:0002048                   474    2048   974.8K        119/1/2    4 1 0  99 *
:0004096                   114    4096   483.3K         59/2/2    2 1 3  96 *
Acpi-State                 102      80     8.1K          2/0/2   51 0 0  99
anon_vma                  2561      24    90.1K        22/10/2  128 0 45  68
bdev_cache                  53     736    45.0K         11/0/2    5 0 0  86 Aa
blkdev_queue                45    1480    73.7K          9/0/2    5 1 0  90
blkdev_requests             42     280    16.3K          4/0/2   14 0 0  71
buffer_head               2695     104   327.6K        80/25/2   39 0 31  85 a
cfq_io_context             189     152    32.7K          8/3/2   26 0 37  87
cfq_queue                  203     144    32.7K          8/4/2   28 0 50  89
dentry                  228340     200    46.7M      11418/0/2   20 0 0  97 a
ext3_inode_cache            16     736    24.5K          6/4/2    5 0 66  47 a
file_lock_cache             46     176    16.3K          4/2/2   22 0 50  49
idr_layer_cache            135     528    81.9K         20/1/2    7 0 5  87
inode_cache             224448     536   131.3M      32065/0/2    7 0 0  91 a
kmalloc-16384                8   16384   163.8K         10/0/2    1 2 0  80
kmalloc-32768               34   32768     1.1M         34/0/1    1 3 0 100
kmalloc-65536                3   65536   196.6K          3/0/2    1 4 0 100
kmalloc-8                  893       8     8.1K          2/0/2  512 0 0  87
kmalloc-8192                14    8192   131.0K         16/0/2    1 1 0  87
kmalloc_dma-512              8     512     4.0K          1/0/1    8 0 0 100 d
mqueue_inode_cache           9     824     8.1K          1/0/1    9 1 0  90 A
proc_inode_cache            28     568    40.9K         10/8/2    7 0 80  38 a
radix_tree_node           2759     552     2.2M      544/243/2    7 0 44  68
raid5-md2                  259     584   151.5K         37/0/1    7 0 0  99
shmem_inode_cache          524     728   434.1K        106/5/2    5 0 4  87
sighand_cache              135    2080   376.8K         46/2/2    3 1 4  74 A
sigqueue                     8     160     8.1K          2/0/2   25 0 0  15
sock_inode_cache           261     616   192.5K         47/8/2    6 0 17  83 Aa
sysfs_dir_cache         229196      88    20.4M       4984/2/2   46 0 0  98
task_struct                152    1696   327.6K         40/6/2    4 1 15  78
TCP                         27    1496    57.3K          7/2/2    5 1 28  70 A
UNIX                       218     640   155.6K         38/5/2    6 0 13  89 A
vm_area_struct            7107     168     1.2M       300/19/2   24 0 6  97
xfs_acl                     14     304     8.1K          2/0/2   13 0 0  51
xfs_buf_item                30     184     8.1K          2/0/2   22 0 0  67
xfs_da_state                 8     488     8.1K          2/0/2    8 0 0  47
xfs_efd_item                11     360     8.1K          2/0/2   11 0 0  48
xfs_efi_item                12     352     8.1K          2/0/2   11 0 0  51
xfs_inode                 4634     544     2.7M        662/0/2    7 0 0  92 Aa
xfs_vnode                 4638     576     3.1M        773/0/2    6 0 0  84 Aa
#1 lore raw
Quoted 5 lines · Joel Fuster

Hi,

I am running 2.6.22.3. For reasons that escape me, over time (days) the sysfs_dir_cache, dentry, and inode_cache SLUB entries grow until they consume all the memory on my system, requiring a reboot.

Although I did not record objective evidence of this at the time, I suspect I had the same problem with 2.6.18 and 2.6.20...I certainly had the same symptoms.

An update to this...it seems to be related to "scanbuttond" which is an app that uses libusb to poll a scanner for button presses on the scanner itself. The number of objects in sysfs_dir_cache stops growing when I kill the scanbuttond process.

An strace of one poll loop for scanbuttond follows:

nanosleep({0, 333000000}, NULL) = 0
open("/dev/bus/usb", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 1 fstat(1, {st_mode=S_IFDIR|0755, st_size=80, ...}) = 0

fcntl(1, F_SETFD, FD_CLOEXEC)           = 0
getdents(1, /* 4 entries */, 4096)      = 96
getdents(1, /* 0 entries */, 4096)      = 0
close(1)                                = 0
open("/dev/bus/usb/002", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 1
fstat(1, {st_mode=S_IFDIR|0755, st_size=60, ...}) = 0
fcntl(1, F_SETFD, FD_CLOEXEC)           = 0
getdents(1, /* 3 entries */, 4096)      = 72
open("/dev/bus/usb/002/001", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/002/001", O_RDONLY)  = 2

ioctl(2, USBDEVFS_CONNECTINFO, 0x7fffb3a08420) = -1 EPERM (Operation not permitted) read(2, "\22\1\0\2\t\0\1@\0\0\0\0\6\2\3\2\1\1", 18) = 18
read(2, "\t\2\31\0\1\1\0\340", 8) = 8
read(2, "\0\t\4\0\0\1\t\0\0\0\7\5\201\3\4\0\f", 17) = 17

close(2)                                = 0
getdents(1, /* 0 entries */, 4096)      = 0
close(1)                                = 0
open("/dev/bus/usb/002/001", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/002/001", O_RDONLY)  = 1

ioctl(1, USBDEVFS_IOCTL, 0x7fffb3a08420) = -1 EPERM (Operation not permitted)
close(1) = 0
open("/dev/bus/usb/001", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 1 fstat(1, {st_mode=S_IFDIR|0755, st_size=100, ...}) = 0

fcntl(1, F_SETFD, FD_CLOEXEC)           = 0
getdents(1, /* 4 entries */, 4096)      = 96
getdents(1, /* 0 entries */, 4096)      = 0
close(1)                                = 0
open("/dev/bus/usb/002", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 1
fstat(1, {st_mode=S_IFDIR|0755, st_size=60, ...}) = 0
fcntl(1, F_SETFD, FD_CLOEXEC)           = 0
getdents(1, /* 3 entries */, 4096)      = 72
open("/dev/bus/usb/002/001", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/002/001", O_RDONLY)  = 2

ioctl(2, USBDEVFS_CONNECTINFO, 0x7fffb3a08420) = -1 EPERM (Operation not permitted) read(2, "\22\1\0\2\t\0\1@\0\0\0\0\6\2\3\2\1\1", 18) = 18
read(2, "\t\2\31\0\1\1\0\340", 8) = 8
read(2, "\0\t\4\0\0\1\t\0\0\0\7\5\201\3\4\0\f", 17) = 17

close(2)                                = 0
getdents(1, /* 0 entries */, 4096)      = 0
close(1)                                = 0
open("/dev/bus/usb/002/001", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/002/001", O_RDONLY)  = 1

ioctl(1, USBDEVFS_IOCTL, 0x7fffb3a08420) = -1 EPERM (Operation not permitted)
close(1) = 0
open("/dev/bus/usb/001", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 1 fstat(1, {st_mode=S_IFDIR|0755, st_size=100, ...}) = 0

fcntl(1, F_SETFD, FD_CLOEXEC)           = 0
getdents(1, /* 5 entries */, 4096)      = 120
open("/dev/bus/usb/001/004", O_RDWR)    = 2

ioctl(2, USBDEVFS_CONNECTINFO, 0x7fffb3a08420) = 0
read(2, "\22\1\20\1\0\0\0@\330\5\3@\0\1\0\0\0\1", 18) = 18
read(2, "\t\2 \0\1\1\0\240", 8) = 8
read(2, "\372\t\4\0\0\2\377\377\377\0\7\5\201\2@\0\0\7\5\2\2@\0"..., 24) = 24

close(2)                                = 0
open("/dev/bus/usb/001/003", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/001/003", O_RDONLY)  = 2

ioctl(2, USBDEVFS_CONNECTINFO, 0x7fffb3a08420) = -1 EPERM (Operation not permitted) read(2, "\22\1\20\1\0\0\0@{\6\3#\0\3\1\2\0\1", 18) = 18
read(2, "\t\2\'\0\1\1\0\200", 8) = 8
read(2, "2\t\4\0\0\3\377\0\0\0\7\5\201\3\n\0\1\7\5\2\2@\0\0\7\5"..., 31) = 31

close(2)                                = 0
open("/dev/bus/usb/001/001", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/001/001", O_RDONLY)  = 2

ioctl(2, USBDEVFS_CONNECTINFO, 0x7fffb3a08420) = -1 EPERM (Operation not permitted) read(2, "\22\1\20\1\t\0\0@\0\0\0\0\6\2\3\2\1\1", 18) = 18
read(2, "\t\2\31\0\1\1\0\340", 8) = 8
read(2, "\0\t\4\0\0\1\t\0\0\0\7\5\201\3\2\0\377", 17) = 17

close(2)                                = 0
getdents(1, /* 0 entries */, 4096)      = 0
close(1)                                = 0
open("/dev/bus/usb/001/004", O_RDWR)    = 1
ioctl(1, USBDEVFS_IOCTL, 0x7fffb3a08420) = -1 ENOTTY (Inappropriate ioctl for device)
close(1)                                = 0
open("/dev/bus/usb/001/003", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/001/003", O_RDONLY)  = 1
ioctl(1, USBDEVFS_IOCTL, 0x7fffb3a08420) = -1 EPERM (Operation not permitted)
close(1)                                = 0
open("/dev/bus/usb/001/001", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/001/001", O_RDONLY)  = 1

ioctl(1, USBDEVFS_IOCTL, 0x7fffb3a08420) = -1 EPERM (Operation not permitted)
close(1) = 0
open("/dev/bus/usb", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 1 fstat(1, {st_mode=S_IFDIR|0755, st_size=80, ...}) = 0

fcntl(1, F_SETFD, FD_CLOEXEC)           = 0
getdents(1, /* 4 entries */, 4096)      = 96
getdents(1, /* 0 entries */, 4096)      = 0
close(1)                                = 0
open("/dev/bus/usb/002", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 1
fstat(1, {st_mode=S_IFDIR|0755, st_size=60, ...}) = 0
fcntl(1, F_SETFD, FD_CLOEXEC)           = 0
getdents(1, /* 3 entries */, 4096)      = 72
open("/dev/bus/usb/002/001", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/002/001", O_RDONLY)  = 2

ioctl(2, USBDEVFS_CONNECTINFO, 0x7fffb3a08420) = -1 EPERM (Operation not permitted) read(2, "\22\1\0\2\t\0\1@\0\0\0\0\6\2\3\2\1\1", 18) = 18
read(2, "\t\2\31\0\1\1\0\340", 8) = 8
read(2, "\0\t\4\0\0\1\t\0\0\0\7\5\201\3\4\0\f", 17) = 17

close(2)                                = 0
getdents(1, /* 0 entries */, 4096)      = 0
close(1)                                = 0
open("/dev/bus/usb/002/001", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/002/001", O_RDONLY)  = 1

ioctl(1, USBDEVFS_IOCTL, 0x7fffb3a08420) = -1 EPERM (Operation not permitted)
close(1) = 0
open("/dev/bus/usb/001", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 1 fstat(1, {st_mode=S_IFDIR|0755, st_size=100, ...}) = 0

fcntl(1, F_SETFD, FD_CLOEXEC)           = 0
getdents(1, /* 5 entries */, 4096)      = 120
open("/dev/bus/usb/001/004", O_RDWR)    = 2

ioctl(2, USBDEVFS_CONNECTINFO, 0x7fffb3a08420) = 0
read(2, "\22\1\20\1\0\0\0@\330\5\3@\0\1\0\0\0\1", 18) = 18
read(2, "\t\2 \0\1\1\0\240", 8) = 8
read(2, "\372\t\4\0\0\2\377\377\377\0\7\5\201\2@\0\0\7\5\2\2@\0"..., 24) = 24

close(2)                                = 0
open("/dev/bus/usb/001/003", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/001/003", O_RDONLY)  = 2

ioctl(2, USBDEVFS_CONNECTINFO, 0x7fffb3a08420) = -1 EPERM (Operation not permitted) read(2, "\22\1\20\1\0\0\0@{\6\3#\0\3\1\2\0\1", 18) = 18
read(2, "\t\2\'\0\1\1\0\200", 8) = 8
read(2, "2\t\4\0\0\3\377\0\0\0\7\5\201\3\n\0\1\7\5\2\2@\0\0\7\5"..., 31) = 31

close(2)                                = 0
open("/dev/bus/usb/001/001", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/001/001", O_RDONLY)  = 2

ioctl(2, USBDEVFS_CONNECTINFO, 0x7fffb3a08420) = -1 EPERM (Operation not permitted) read(2, "\22\1\20\1\t\0\0@\0\0\0\0\6\2\3\2\1\1", 18) = 18
read(2, "\t\2\31\0\1\1\0\340", 8) = 8
read(2, "\0\t\4\0\0\1\t\0\0\0\7\5\201\3\2\0\377", 17) = 17

close(2)                                = 0
getdents(1, /* 0 entries */, 4096)      = 0
close(1)                                = 0
open("/dev/bus/usb/001/004", O_RDWR)    = 1
ioctl(1, USBDEVFS_IOCTL, 0x7fffb3a08420) = -1 ENOTTY (Inappropriate ioctl for device)
close(1)                                = 0
open("/dev/bus/usb/001/003", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/001/003", O_RDONLY)  = 1
ioctl(1, USBDEVFS_IOCTL, 0x7fffb3a08420) = -1 EPERM (Operation not permitted)
close(1)                                = 0
open("/dev/bus/usb/001/001", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/001/001", O_RDONLY)  = 1

ioctl(1, USBDEVFS_IOCTL, 0x7fffb3a08420) = -1 EPERM (Operation not permitted)
close(1) = 0
open("/dev/bus/usb/001/004", O_RDWR) = 1
ioctl(1, USBDEVFS_CLAIMINTERFACE, 0x7fffb3a0848c) = 0
ioctl(1, USBDEVFS_CONTROL, 0x7fffb3a08420) = 64
ioctl(1, USBDEVFS_CONTROL, 0x7fffb3a08420) = 8
ioctl(1, USBDEVFS_RELEASEINTERFACE, 0x7fffb3a08494) = 0
close(1) = 0
nanosleep({0, 333000000}, <unfinished ...>

#2 lore raw
Quoted 5 lines · Joel Fuster
Quoted 4 lines · Joel Fuster

Hi,
I am running 2.6.22.3. For reasons that escape me, over time (days) the sysfs_dir_cache, dentry, and inode_cache SLUB entries grow until they consume all the memory on my system, requiring a reboot.

Hm, those items should consume all the memory, but it should be freed if you have memory pressure from other places. Does it cause the machine to lock up, or you just got scared when seeing them?

Oh, and does the same thing happen if you do not use SLUB, but rather the older SLAB?

Quoted 25 lines
Quoted 3 lines

Although I did not record objective evidence of this at the time, I suspect I had the same problem with 2.6.18 and 2.6.20...I certainly had the same symptoms.

An update to this...it seems to be related to "scanbuttond" which is an app that uses libusb to poll a scanner for button presses on the scanner itself. The number of objects in sysfs_dir_cache stops growing when I kill the scanbuttond process.

An strace of one poll loop for scanbuttond follows:

nanosleep({0, 333000000}, NULL) = 0
open("/dev/bus/usb", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 1 fstat(1, {st_mode=S_IFDIR|0755, st_size=80, ...}) = 0

fcntl(1, F_SETFD, FD_CLOEXEC)           = 0
getdents(1, /* 4 entries */, 4096)      = 96
getdents(1, /* 0 entries */, 4096)      = 0
close(1)                                = 0
open("/dev/bus/usb/002", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 1
fstat(1, {st_mode=S_IFDIR|0755, st_size=60, ...}) = 0
fcntl(1, F_SETFD, FD_CLOEXEC)           = 0
getdents(1, /* 3 entries */, 4096)      = 72
open("/dev/bus/usb/002/001", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/002/001", O_RDONLY)  = 2
Quoted 2 lines

ioctl(2, USBDEVFS_CONNECTINFO, 0x7fffb3a08420) = -1 EPERM (Operation not permitted)

<snip>

I don't see any sysfs accesses there, only usbfs accesses.

And it looks like a very dumb program, iterating over all usb devices all the time? You would think that it would not do that, otherwise your power consumption is going to be very high, and the odds that a new usb device is going to show up that quickly is pretty low.

thanks,

greg k-h

#3 lore raw
Quoted 9 lines · Greg KH
Quoted 3 lines · Joel Fuster
Quoted 2 lines · Joel Fuster

Hi,
I am running 2.6.22.3. For reasons that escape me, over time (days) the sysfs_dir_cache, dentry, and inode_cache SLUB entries grow until they consume all the memory on my system, requiring a reboot.

Hm, those items should consume all the memory, but it should be freed if you have memory pressure from other places. Does it cause the machine to lock up, or you just got scared when seeing them?

Right. The problem is that the memory never seems to get freed no matter what I do. I've tried setting /proc/sys/vm/vfs_cache_pressure to 10000, but after a few days all my programs are running out of swap and I have to reboot to get things back to a usable state.

Quoted 2 lines

Oh, and does the same thing happen if you do not use SLUB, but rather the older SLAB?

OK I just rebuilt 2.6.22.3 with SLAB and I seem to be getting the same result..obviously I haven't waited several days, but sysfs_dir_cache/dentry/inode_cache grow continuously when scanbuttond is running, and stop growing when it isn't.

Quoted 19 lines
Quoted 17 lines

An strace of one poll loop for scanbuttond follows:

nanosleep({0, 333000000}, NULL) = 0
open("/dev/bus/usb", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 1 fstat(1, {st_mode=S_IFDIR|0755, st_size=80, ...}) = 0

fcntl(1, F_SETFD, FD_CLOEXEC)           = 0
getdents(1, /* 4 entries */, 4096)      = 96
getdents(1, /* 0 entries */, 4096)      = 0
close(1)                                = 0
open("/dev/bus/usb/002", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 1
fstat(1, {st_mode=S_IFDIR|0755, st_size=60, ...}) = 0
fcntl(1, F_SETFD, FD_CLOEXEC)           = 0
getdents(1, /* 3 entries */, 4096)      = 72
open("/dev/bus/usb/002/001", O_RDWR)    = -1 EACCES (Permission denied)
open("/dev/bus/usb/002/001", O_RDONLY)  = 2
Quoted 1 line

ioctl(2, USBDEVFS_CONNECTINFO, 0x7fffb3a08420) = -1 EPERM (Operation not permitted)

Quoted 3 lines

<snip>

I don't see any sysfs accesses there, only usbfs accesses.

Yes, I don't know enough to understand why this would affect sysfs_dir_cache, but there definitely seems to be a connection.

Thanks for the help,

Joel

#4 lore raw
Quoted 15 lines · Joel Fuster
Quoted 9 lines · Greg KH
Quoted 5 lines · Joel Fuster
Quoted 4 lines · Joel Fuster

Hi,
I am running 2.6.22.3. For reasons that escape me, over time (days) the sysfs_dir_cache, dentry, and inode_cache SLUB entries grow until they consume all the memory on my system, requiring a reboot.

Hm, those items should consume all the memory, but it should be freed if you have memory pressure from other places. Does it cause the machine to lock up, or you just got scared when seeing them?

Right. The problem is that the memory never seems to get freed no matter what I do. I've tried setting /proc/sys/vm/vfs_cache_pressure to 10000, but after a few days all my programs are running out of swap and I have to reboot to get things back to a usable state.

Quoted 7 lines
Quoted 2 lines

Oh, and does the same thing happen if you do not use SLUB, but rather the older SLAB?

OK I just rebuilt 2.6.22.3 with SLAB and I seem to be getting the same result..obviously I haven't waited several days, but sysfs_dir_cache/dentry/inode_cache grow continuously when scanbuttond is running, and stop growing when it isn't.

Do you have a pointer to the scanbuttond source code? I'll try to take a look at this tomorrow.

thanks,

greg k-h

#5 lore raw
Quoted 18 lines · Greg KH
Quoted 17 lines · Joel Fuster
Quoted 9 lines · Greg KH
Quoted 5 lines · Joel Fuster
Quoted 4 lines · Joel Fuster

Hi,
I am running 2.6.22.3. For reasons that escape me, over time (days) the sysfs_dir_cache, dentry, and inode_cache SLUB entries grow until they consume all the memory on my system, requiring a reboot.

Hm, those items should consume all the memory, but it should be freed if you have memory pressure from other places. Does it cause the machine to lock up, or you just got scared when seeing them?

Right. The problem is that the memory never seems to get freed no matter what I do. I've tried setting /proc/sys/vm/vfs_cache_pressure to 10000, but after a few days all my programs are running out of swap and I have to reboot to get things back to a usable state.

Quoted 2 lines

Oh, and does the same thing happen if you do not use SLUB, but rather the older SLAB?

Quoted 8 lines
Quoted 4 lines

OK I just rebuilt 2.6.22.3 with SLAB and I seem to be getting the same result..obviously I haven't waited several days, but sysfs_dir_cache/dentry/inode_cache grow continuously when scanbuttond is running, and stop growing when it isn't.

Do you have a pointer to the scanbuttond source code? I'll try to take a look at this tomorrow.

I guess this one :

https://sourceforge.net/projects/scanbuttond/

Quoted 4 lines

thanks,

greg k-h
-

Gabriel

#6 lore raw
Quoted 20 lines · On Fri, 24 Aug 2007 03:26:50 +0200 Gabriel C <[email protected]> wrote:
Quoted 18 lines · Greg KH
Quoted 17 lines · Joel Fuster
Quoted 9 lines · Greg KH
Quoted 5 lines · Joel Fuster
Quoted 4 lines · Joel Fuster

Hi,
I am running 2.6.22.3. For reasons that escape me, over time (days) the sysfs_dir_cache, dentry, and inode_cache SLUB entries grow until they consume all the memory on my system, requiring a reboot.

Hm, those items should consume all the memory, but it should be freed if you have memory pressure from other places. Does it cause the machine to lock up, or you just got scared when seeing them?

Right. The problem is that the memory never seems to get freed no matter what I do. I've tried setting /proc/sys/vm/vfs_cache_pressure to 10000, but after a few days all my programs are running out of swap and I have to reboot to get things back to a usable state.

Quoted 2 lines

Oh, and does the same thing happen if you do not use SLUB, but rather the older SLAB?

Quoted 8 lines
Quoted 8 lines
Quoted 4 lines

OK I just rebuilt 2.6.22.3 with SLAB and I seem to be getting the same result..obviously I haven't waited several days, but sysfs_dir_cache/dentry/inode_cache grow continuously when scanbuttond is running, and stop growing when it isn't.

Do you have a pointer to the scanbuttond source code? I'll try to take a look at this tomorrow.

that was a long day?

Quoted 3 lines

I guess this one :

https://sourceforge.net/projects/scanbuttond/

I needed setenv LD_LIBRARY_PATH /usr/local/lib' to be able to run scanbuttond -p 1000'

and after a few minutes I'm seeing zero growth of /proc/spabinfo:sysfs_dir_cache over a few minutes, with 2.6.23-rc4. Perhaps it's a specific device driver?