Subject : 2.6.34-rc2: "ima_dec_counts: open/free imbalance"? Submitter : Thomas Meyer <thomas@m3y3r.de> Date : 2010-03-28 11:31 Message-ID : 1269775909.5301.4.camel@localhost.localdomain References : http://marc.info/?l=linux-kernel&m=126977593326800&w=2 This entry is being used for tracking a regression from 2.6.33. Please don't close it until the problem is fixed in the mainline.
Created attachment 26881 [details] config for 2.6.34
I don't see the warning anymore with above config and plain 2.6.34.
Created attachment 26952 [details] dmesg with warning Stop! I still see this warning with 2.6.34. See attached dmesg. this warning is seen with also attached config.
On Saturday, July 10, 2010, Thomas Meyer wrote: > Am Samstag, den 10.07.2010, 02:33 +0200 schrieb Rafael J. Wysocki: > > This message has been generated automatically as a part of a report > > of regressions introduced between 2.6.33 and 2.6.34. > > > > The following bug entry is on the current list of known regressions > > introduced between 2.6.33 and 2.6.34. Please verify if it still should > > be listed and let the tracking team know (either way). > > > > > > Bug-Entry : http://bugzilla.kernel.org/show_bug.cgi?id=15673 > > Subject : 2.6.34-rc2: "ima_dec_counts: open/free imbalance"? > > Submitter : Thomas Meyer <thomas@m3y3r.de> > > Date : 2010-03-28 11:31 (104 days old) > > Still see the warning in 2.6.34.1: > > [ 1466.091349] IMA: unmeasured files on fsmagic: 1021994 > [ 1466.091356] ima_dec_counts: open/free imbalance (r:0 w:-1 o:-1) > [ 1466.091360] Pid: 2357, comm: async/10 Tainted: G W 2.6.34.1 #3 > [ 1466.091362] Call Trace: > [ 1466.091371] [<c08865c7>] ? printk+0x1d/0x26 > [ 1466.091377] [<c05f994f>] ima_file_free+0x16f/0x210 > [ 1466.091384] [<c04d7742>] __fput+0xf2/0x1f0 > [ 1466.091388] [<c04d785d>] fput+0x1d/0x30 > [ 1466.091394] [<c06cad8f>] drm_gem_object_free_common+0x1f/0x40 > [ 1466.091398] [<c06cae30>] ? drm_gem_object_free+0x0/0x40 > [ 1466.091402] [<c06cae61>] drm_gem_object_free+0x31/0x40 > [ 1466.091407] [<c061d59c>] kref_put+0x2c/0x60 > [ 1466.091412] [<c06e7308>] i915_gem_cleanup_ringbuffer+0x48/0x70 > [ 1466.091416] [<c06e8a6c>] i915_gem_idle+0x9c/0x120 > [ 1466.091420] [<c06dbe1d>] i915_drm_freeze+0x3d/0xa0 > [ 1466.091423] [<c06dbfde>] i915_pm_suspend+0x2e/0x80 > [ 1466.091427] [<c0886d0a>] ? wait_for_common+0x1a/0xf0 > [ 1466.091432] [<c0636629>] pci_pm_suspend+0x49/0x110 > [ 1466.091436] [<c06365e0>] ? pci_pm_suspend+0x0/0x110 > [ 1466.091441] [<c0718be1>] pm_op+0x181/0x1d0 > [ 1466.091445] [<c0711d34>] ? device_for_each_child+0x54/0x60 > [ 1466.091449] [<c071959f>] __device_suspend+0xbf/0x110 > [ 1466.091453] [<c07199e3>] async_suspend+0x23/0x60 > [ 1466.091457] [<c0450025>] async_thread+0xc5/0x210 > [ 1466.091461] [<c0886931>] ? schedule+0x1e1/0x450 > [ 1466.091472] [<c042bd60>] ? default_wake_function+0x0/0x20 > [ 1466.091476] [<c044ff60>] ? async_thread+0x0/0x210 > [ 1466.091480] [<c04493c4>] kthread+0x74/0x80 > [ 1466.091483] [<c0449350>] ? kthread+0x0/0x80 > [ 1466.091488] [<c040347e>] kernel_thread_helper+0x6/0x10 > [ 1466.096531] iint_free: writecount: -1 > [ 1466.096533] iint_free: opencount: -1 > [ 1466.096535] iint_free: writecount: -1 > [ 1466.096537] iint_free: opencount: -1 > [ 1466.103256] ehci_hcd 0000:00:1d.7: PCI INT A disabled
Created attachment 27237 [details] the warning can also be seen in 2.6.35-rc6
Created attachment 27238 [details] config for 2.6.35-rc6
still in 2.6.35: [ 3123.733374] IMA: unmeasured files on fsmagic: 1021994 [ 3123.733377] ima_dec_counts: open/free imbalance (r:0 w:-1 o:-1) [ 3123.733382] Pid: 2886, comm: async/10 Tainted: G I 2.6.35 #2 [ 3123.733384] Call Trace: [ 3123.733394] [<c05cd4e3>] ? ima_file_free+0x163/0x1c0 [ 3123.733405] [<c04b5a99>] ? fput+0x119/0x220 [ 3123.733411] [<c06956a7>] ? drm_gem_object_release+0x17/0x40 [ 3123.733416] [<c06b2430>] ? i915_gem_free_object+0x50/0xc0 [ 3123.733420] [<c0695660>] ? drm_gem_object_free+0x0/0x30 [ 3123.733424] [<c069567d>] ? drm_gem_object_free+0x1d/0x30 [ 3123.733428] [<c05edcd9>] ? kref_put+0x29/0x60 [ 3123.733433] [<c06cb49b>] ? intel_cleanup_ring_buffer+0x3b/0x70 [ 3123.733437] [<c06b0be1>] ? i915_gem_cleanup_ringbuffer+0x11/0x30 [ 3123.733440] [<c06b2236>] ? i915_gem_idle+0x86/0xf0 [ 3123.733444] [<c06a5ce8>] ? i915_drm_freeze+0x38/0xb0 [ 3123.733448] [<c06a5ea8>] ? i915_pm_suspend+0x28/0x80 [ 3123.733452] [<c0606305>] ? pci_pm_suspend+0x45/0x110 [ 3123.733456] [<c06062c0>] ? pci_pm_suspend+0x0/0x110 [ 3123.733460] [<c06df5b3>] ? pm_op+0x193/0x1e0 [ 3123.733464] [<c06df240>] ? dpm_wait_fn+0x0/0x10 [ 3123.733469] [<c06d8e13>] ? device_for_each_child+0x43/0x50 [ 3123.733473] [<c06dffbf>] ? __device_suspend+0xbf/0x110 [ 3123.733477] [<c06e03fd>] ? async_suspend+0x1d/0x50 [ 3123.733482] [<c0444b25>] ? async_thread+0xc5/0x210 [ 3123.733486] [<c0427b25>] ? finish_task_switch.clone.0+0x25/0x90 [ 3123.733490] [<c0428c90>] ? default_wake_function+0x0/0x10 [ 3123.733494] [<c0444a60>] ? async_thread+0x0/0x210 [ 3123.733498] [<c043eb14>] ? kthread+0x74/0x80 [ 3123.733501] [<c043eaa0>] ? kthread+0x0/0x80 [ 3123.733505] [<c04032be>] ? kernel_thread_helper+0x6/0x18 [ 3123.739698] iint_free: writecount: -1 [ 3123.739700] iint_free: opencount: -1 [ 3123.739702] iint_free: writecount: -1 [ 3123.739704] iint_free: opencount: -1
The IMA counters are incremented/decremented only after IMA has been initialized. It seems the 'imbalance' message occurs only when the i915 is builtin, not as a module. It seems when i915 is builtin, the ring_buffer is initialized before IMA. I've posted a patch which should resolve this bug on the LSM mailing list, copying stable. Mimi
Still there in 2.6.36 or current 2.6.37-rc7?
Fixed by commit e950598d43dce8d97e7d5270808393425d1e5cbd (ima: always maintain counters), that went in 2.6.36-rc4
commit e950598d43dce8d97e7d5270808393425d1e5cbd Author: Mimi Zohar <zohar@linux.vnet.ibm.com> Date: Tue Aug 31 09:38:51 2010 -0400 ima: always maintain counters commit 8262bb85da allocated the inode integrity struct (iint) before any inodes were created. Only after IMA was initialized in late_initcall were the counters updated. This patch updates the counters, whether or not IMA has been initialized, to resolve 'imbalance' messages. This patch fixes the bug as reported in bugzilla: 15673. When the i915 is builtin, the ring_buffer is initialized before IMA, causing the imbalance message on suspend. Reported-by: Thomas Meyer <thomas@m3y3r.de> Signed-off-by: Mimi Zohar <zohar@linux.vnet.ibm.com> Tested-by: Thomas Meyer <thomas@m3y3r.de> Tested-by: David Safford<safford@watson.ibm.com> Cc: Stable Kernel <stable@kernel.org> Signed-off-by: James Morris <jmorris@namei.org>