Bug#607879: System hangs up with mmap.c:873!

[Ronny Standtke]

 Anyway, it shouldn't be necessary to rebuild the header packages as there is 
 no ABI change and the previous version should be compatible.  Does 
 live-build require an exact version match?

As I already said, the package linux-headers-2.6.32-5-686_2.6.32-30a~test 
depends on linux-headers-2.6.32-5-common_2.6.32-30a~test. Therefore we need to 
have the common package falling out of the customized kernel package build 
process, too. And therefore I was such a pain in the neck to get an official 
way to build this package...

 OK, we'll include these changes in an update to squeeze.

Great! Thank you very much.

Ronny



-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Ben Hutchings]

On Tue, 2011-02-08 at 08:38 +0100, Ronny Standtke wrote:
  Fortunately, after some trial and error, I successfully built a 
  linux-headers-2.6.32-5-common_2.6.32-30a~test package with the following 
  two 
  commands:
  
  
  export UPSTREAMVERSION=2.6.32-5
  fakeroot make -f debian/rules.real binary-arch-featureset
  
  
  Is this the right way?
 
 To partially answer my own question: No, the resulting package is not useable.
 
 I had to use the following list of commands to build a package that worked 
 for 
 me:
 
 
 export VERSION=2.6.32
 export UPSTREAMVERSION=2.6.32-5
 export KERNEL_ARCH=x86
 fakeroot make -f debian/rules.real binary-arch-featureset
 
 
 I still would like to know the official and supported way to build the
 linux-headers-x.y.z-a-common package. Ben?

You can use:

fakeroot make -f debian/rules.gen binary-arch_i386_none

but this is subject to change at any time.  I would not say it is
official or supported.  Anyway, it shouldn't be necessary to rebuild the
header packages as there is no ABI change and the previous version
should be compatible.  Does live-build require an exact version match?

 The good thing is that the patch series Ben provided seems to fix the 
 problem. 
 At least I did no longer run into this bug for more than a week now.

OK, we'll include these changes in an update to squeeze.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.


signature.asc
Description: This is a digitally signed message part

Bug#607879: System hangs up with mmap.c:873!

[Ronny Standtke]

 Fortunately, after some trial and error, I successfully built a 
 linux-headers-2.6.32-5-common_2.6.32-30a~test package with the following two 
 commands:
 
 
 export UPSTREAMVERSION=2.6.32-5
 fakeroot make -f debian/rules.real binary-arch-featureset
 
 
 Is this the right way?

To partially answer my own question: No, the resulting package is not useable.

I had to use the following list of commands to build a package that worked for 
me:


export VERSION=2.6.32
export UPSTREAMVERSION=2.6.32-5
export KERNEL_ARCH=x86
fakeroot make -f debian/rules.real binary-arch-featureset


I still would like to know the official and supported way to build the
linux-headers-x.y.z-a-common package. Ben?

The good thing is that the patch series Ben provided seems to fix the problem. 
At least I did no longer run into this bug for more than a week now.

Thanks

Ronny



-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Ronny Standtke]

 This *is* explained on the page I pointed you at.

Sorry for being a *blind* newbie...

After building the linux-base package I was able to install the patched linux-
image package.

The system I am trying to build with live-build also needs the linux-headers 
package because of some dkms packages. Unfortunately, there is again a 
dependency problem:


# dpkg -i linux-headers-2.6.32-5-686_2.6.32-30a~test_i386.deb 
(Reading database ... 166017 files and directories currently installed.)
Preparing to replace linux-headers-2.6.32-5-686 2.6.32-30a~test (using linux-
headers-2.6.32-5-686_2.6.32-30a~test_i386.deb) ...
Unpacking replacement linux-headers-2.6.32-5-686 ...
dpkg: dependency problems prevent configuration of linux-headers-2.6.32-5-686:
 linux-headers-2.6.32-5-686 depends on linux-headers-2.6.32-5-common (= 
2.6.32-30a~test); however:
  Version of linux-headers-2.6.32-5-common on system is 2.6.32-30.
dpkg: error processing linux-headers-2.6.32-5-686 (--install):
 dependency problems - leaving unconfigured
Errors were encountered while processing:
 linux-headers-2.6.32-5-686


Unfortunately, I could not find any information about this issue on the page 
you pointet me at. Fortunately, after some trial and error, I successfully 
built a linux-headers-2.6.32-5-common_2.6.32-30a~test package with the 
following two commands:


export UPSTREAMVERSION=2.6.32-5
fakeroot make -f debian/rules.real binary-arch-featureset


Is this the right way?

Best regards

Ronny



-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Ronny Standtke]

 Not sure how I did that, but try this version of patch 3 instead.

Thanks, building the package works now. Unfortunately, the resulting package 
is not installable because its dependencies seem to be broken:

---
# dpkg -i linux-image-2.6.32-5-686_2.6.32-30a~test_i386.deb 
(Reading database ... 166017 files and directories currently installed.)
Preparing to replace linux-image-2.6.32-5-686 2.6.32-30a~test (using linux-
image-2.6.32-5-686_2.6.32-30a~test_i386.deb) ...
Unpacking replacement linux-image-2.6.32-5-686 ...
Examining /etc/kernel/postrm.d .
run-parts: executing /etc/kernel/postrm.d/initramfs-tools 2.6.32-5-686 
/boot/vmlinuz-2.6.32-5-686
run-parts: executing /etc/kernel/postrm.d/zz-update-grub 2.6.32-5-686 
/boot/vmlinuz-2.6.32-5-686
dpkg: dependency problems prevent configuration of linux-image-2.6.32-5-686:
 linux-image-2.6.32-5-686 depends on linux-base (= 2.6.32-30a~test); however:
  Version of linux-base on system is 2.6.32-30.
dpkg: error processing linux-image-2.6.32-5-686 (--install):
 dependency problems - leaving unconfigured
Errors were encountered while processing:
 linux-image-2.6.32-5-686
---

The dependency on linux-base (= 2.6.32-30a~test) looks like the build script 
is too eagerly replacing version strings. Can you give another hint or an 
updated build script, please?

Ronny



-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Ben Hutchings]

On Tue, 2011-01-25 at 21:34 +0100, Ronny Standtke wrote:
  Not sure how I did that, but try this version of patch 3 instead.
 
 Thanks, building the package works now. Unfortunately, the resulting package 
 is not installable because its dependencies seem to be broken:
[...]

This *is* explained on the page I pointed you at.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.


signature.asc
Description: This is a digitally signed message part

Bug#607879: System hangs up with mmap.c:873!

[Ronny Standtke]

 See
 http://kernel-handbook.alioth.debian.org/ch-common-tasks.html#s-common-off
 icial - in particular, section 4.2.5, Simplified patching and building.

OK, here is what I tried now:
-
apt-get source linux-image-2.6.32-5-686
cd linux-2.6-2.6.32
bash debian/bin/test-patches ../000*
-

Unfortunately, this fails with the following error message:
...
-- 30 fully applied.
-- Try to apply 30a~test.
  (+) OK   test/0001-Revert-aufs-narrow-down-the-BKL-region.patch
  (+) OK   test/0002-Revert-aufs-bugfix-unlock-mmap_sem-temporary-using-
B.patch
patch:  malformed patch at line 182: @@ -630,25 +698,22 @@ static int 
aufs_mmap(struct file *file, struct vm_area_struct *vma)

  (+) FAIL test/0003-aufs-bugfix-another-approach-to-keep-the-lock-
order-.patch
Error: Patch failed
make[2]: *** [debian/stamps/source] Error 1
make[2]: Leaving directory `/root/linux-2.6-2.6.32'
make[1]: *** [source_i386_none_real] Error 2
make[1]: Leaving directory `/root/linux-2.6-2.6.32'
make: *** [debian/stamps/source-base] Error 2

Is there anything else that I need to do or is the patch really malformed?

Ronny



-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Ben Hutchings]

On Sun, 2011-01-23 at 09:18 +0100, Ronny Standtke wrote:
  See
  http://kernel-handbook.alioth.debian.org/ch-common-tasks.html#s-common-off
  icial - in particular, section 4.2.5, Simplified patching and building.
 
 OK, here is what I tried now:
 -
 apt-get source linux-image-2.6.32-5-686
 cd linux-2.6-2.6.32
 bash debian/bin/test-patches ../000*
 -
 
 Unfortunately, this fails with the following error message:
 ...
 -- 30 fully applied.
 -- Try to apply 30a~test.
   (+) OK   test/0001-Revert-aufs-narrow-down-the-BKL-region.patch
   (+) OK   test/0002-Revert-aufs-bugfix-unlock-mmap_sem-temporary-using-
 B.patch
 patch:  malformed patch at line 182: @@ -630,25 +698,22 @@ static int 
 aufs_mmap(struct file *file, struct vm_area_struct *vma)
 
   (+) FAIL test/0003-aufs-bugfix-another-approach-to-keep-the-lock-
[...]

Not sure how I did that, but try this version of patch 3 instead.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.
From: J. R. Okajima hooano...@yahoo.co.jp
Date: Mon, 8 Mar 2010 23:45:56 +0900
Subject: [PATCH 3/4] aufs: bugfix, another approach to keep the lock order of mmap_sem

commit d986fa5a8557f6861fcac4106b6d75301bf5d118 in aufs2-2.6

The previous approach
	4b70e6f aufs: bugfix, unlock mmap_sem temporary using BKL
was bad and already reverted.
This approach is ugly too, but it works.

- split aufs_mmap() into two parts.
- the first part is for copy-ing up which requires rwsem and executed by
  aufsd workqueue.
- the second part is generic_file_mmap() and customizing vm_ops, and
  executed by the original context.
- to protect customizing vm_ops from race between two mmaps, introduce a
  new mutex in au_finfo. lock in the first phase, and release it in the
  second. this is the most ugly part of this approach. if we could use
  fi_rwsem for this use, we would use it. but there is no 'set_owner'
  method for rwsem, but mutex has.

Signed-off-by: J. R. Okajima hooano...@yahoo.co.jp
[bwh: Adjust for 2.6.32]
---
 fs/aufs/f_op.c  |  113 +++
 fs/aufs/file.h  |3 +
 fs/aufs/finfo.c |   17 
 3 files changed, 109 insertions(+), 24 deletions(-)

diff --git a/fs/aufs/f_op.c b/fs/aufs/f_op.c
index 6f89992..32cc36f 100644
--- a/fs/aufs/f_op.c
+++ b/fs/aufs/f_op.c
@@ -77,6 +77,7 @@ int au_do_open_nondir(struct file *file, int flags)
 	finfo = au_fi(file);
 	finfo-fi_h_vm_ops = NULL;
 	finfo-fi_vm_ops = NULL;
+	mutex_init(finfo-fi_mmap); /* regular file only? */
 	bindex = au_dbstart(dentry);
 	/* O_TRUNC is processed already */
 	BUG_ON(au_test_ro(dentry-d_sb, bindex, dentry-d_inode)
@@ -544,7 +545,7 @@ static int au_custom_vm_ops(struct au_finfo *finfo, struct vm_area_struct *vma)
 	int err;
 	struct vm_operations_struct *h_ops;
 
-	AuRwMustAnyLock(finfo-fi_rwsem);
+	MtxMustLock(finfo-fi_mmap);
 
 	err = 0;
 	h_ops = finfo-fi_h_vm_ops;
@@ -570,49 +571,115 @@ static int au_custom_vm_ops(struct au_finfo *finfo, struct vm_area_struct *vma)
 	return err;
 }
 
-static int aufs_mmap(struct file *file, struct vm_area_struct *vma)
+/*
+ * This is another ugly approach to keep the lock order, particularly
+ * mm-mmap_sem and aufs rwsem. The previous approach was reverted and you can
+ * find it in git-log, if you want.
+ *
+ * native readdir: i_mutex, copy_to_user, mmap_sem
+ * aufs readdir: i_mutex, rwsem, nested-i_mutex, copy_to_user, mmap_sem
+ *
+ * Before aufs_mmap() mmap_sem is acquired already, but aufs_mmap() has to
+ * acquire aufs rwsem. It introduces a circular locking dependency.
+ * To address this problem, aufs_mmap() delegates the part which requires aufs
+ * rwsem to its internal workqueue.
+ */
+
+/* very ugly approach */
+#ifdef CONFIG_DEBUG_MUTEXES
+#include ../kernel/mutex-debug.h
+#else
+#include ../kernel/mutex.h
+#endif
+
+struct au_mmap_pre_args {
+	/* input */
+	struct file *file;
+	struct vm_area_struct *vma;
+
+	/* output */
+	int *errp;
+	struct file *h_file;
+	int mmapped;
+};
+
+static int au_mmap_pre(struct file *file, struct vm_area_struct *vma,
+		   struct file **h_file, int *mmapped)
 {
 	int err;
-	unsigned char wlock, mmapped;
+	const unsigned char wlock
+		= !!(file-f_mode  FMODE_WRITE)  (vma-vm_flags  VM_SHARED);
 	struct dentry *dentry;
 	struct super_block *sb;
-	struct file *h_file;
-	struct vm_operations_struct *vm_ops;
 
 	dentry = file-f_dentry;
-	wlock = !!(file-f_mode  FMODE_WRITE)  (vma-vm_flags  VM_SHARED);
 	sb = dentry-d_sb;
-	si_read_lock(sb, AuLock_FLUSH);
+	si_read_lock(sb, !AuLock_FLUSH);
 	err = au_reval_and_lock_fdi(file, au_reopen_nondir, /*wlock*/1);
 	if (unlikely(err))
 		goto out;
 
-	mmapped = !!au_test_mmapped(file);
+	*mmapped = !!au_test_mmapped(file);
 	if (wlock) {
 		struct au_pin pin;
 
 		err = au_ready_to_write(file, -1, pin);
-		di_downgrade_lock(dentry, AuLock_IR);
+		di_write_unlock(dentry);
 		if (unlikely(err))
 			goto out_unlock;
 		au_unpin(pin);
 	} else
-		di_downgrade_lock(dentry, 

Bug#607879: System hangs up with mmap.c:873!

[Ronny Standtke]

 The attached patch series might deal with this bug.  These patches are
 simply cherry-picked from aufs upstream and modified as necessary to
 compile.  I leave any further building and testing to you.

Thank you very much for the patch series. Sorry for being such a newbie but 
how do I deal with this patch series? Here is what I tried:
---
apt-get source linux-image-2.6.32-5-686
cd linux-2.6-2.6.32/
# copying your patch series to the (hopefully) correct place
cp ../000* debian/patches/features/all/aufs2/
# adding your patch series at the end of the aufs block
vi debian/patches/series/base
fakeroot make -f debian/rules.gen binary-arch_i386_none_686
---

This failed with the following error output:
...
-- Try to apply base.
  (+) OK   debian/version.patch
  (+) OK   debian/kernelvariables.patch
  (+) OK   debian/doc-build-parallel.patch
  (+) OK   debian/scripts-kconfig-reportoldconfig.patch
  (+) OK   debian/arch-mips-not-embedded.patch
  (+) OK   debian/drivers-ata-ata_piix-postpone-pata.patch
  (+) OK   debian/drivers-ata-pata_sis-postpone-pata.patch
  (+) OK   features/all/drivers-infiniband-hw-ipath-iba7220-use-
request_firmware.patch
  (+) OK   features/all/drivers-media-dvb-usb-af9005-request_firmware.patch
  (+) OK   features/all/drivers-staging-rt28x0sta-request_firmware.patch
  (+) OK   features/all/export-unionfs-symbols.patch
  (+) OK   features/all/lgs8gxx-lgs8g75-request_firmware.patch
  (+) OK   features/all/r8169-init-phy-return-error.patch
  (+) OK   features/all/r8169-rtl8168d-1-2-request_firmware.patch
  (+) OK   features/all/sound-pci-cs46xx-request_firmware.patch
  (+) OK   features/all/aufs2/aufs2-base.patch
  (+) OK   features/all/aufs2/aufs2-standalone.patch
  (+) OK   features/all/aufs2/aufs2-kbuild.patch
  (+) OK   features/all/aufs2/aufs2-add.patch
  (+) OK   features/all/aufs2/mark-as-staging.patch
2 out of 3 hunks FAILED -- saving rejects to file fs/aufs/f_op.c.rej
  (+) FAIL features/all/aufs2/0001-Revert-aufs-narrow-down-the-BKL-
region.patch
Error: Patch failed
make[1]: *** [debian/stamps/source] Error 1
make[1]: Leaving directory `/root/linux-2.6-2.6.32'
make: *** [binary-arch_i386_none_686_real] Error 2

Can you please give some simple instructions how to create a new kernel 
package that includes your patch series?

Thank you very much!

Ronny



-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Ben Hutchings]

On Sat, 2011-01-22 at 22:40 +0100, Ronny Standtke wrote:
[...]
 Can you please give some simple instructions how to create a new kernel 
 package that includes your patch series?
[...]

See
http://kernel-handbook.alioth.debian.org/ch-common-tasks.html#s-common-official
 - in particular, section 4.2.5, Simplified patching and building.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.


signature.asc
Description: This is a digitally signed message part

Bug#607879: System hangs up with mmap.c:873!

[Ronny Standtke]

I contacted the Debian Live maintainers about this bug, who in turn asked me 
to contact upstream. The aufs developers stated that the aufs version in 
Debian Live is too old and obsoleted and should be updated, see here:

https://sourceforge.net/mailarchive/message.php?msg_id=26901616

I informed the Debian Live maintainers about that, who in turn again pointed 
to the Debian kernel team, see here:

http://lists.debian.org/debian-live/2011/01/msg00100.html

This pushing around gets somewhat exhausting. I would be pleased if all 
involved parties (Debian kernel team, Debian Live maintainers, aufs 
developers) could cooperate in fixing this bug instead of working against or 
ignoring each other.

Thanks a lot

Ronny



-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Ben Hutchings]

On Sat, 2011-01-15 at 14:26 +0100, Ronny Standtke wrote:
 I contacted the Debian Live maintainers about this bug, who in turn asked me 
 to contact upstream. The aufs developers stated that the aufs version in 
 Debian Live is too old and obsoleted and should be updated, see here:
 
 https://sourceforge.net/mailarchive/message.php?msg_id=26901616
 
 I informed the Debian Live maintainers about that, who in turn again pointed 
 to the Debian kernel team, see here:
 
 http://lists.debian.org/debian-live/2011/01/msg00100.html
 
 This pushing around gets somewhat exhausting. I would be pleased if all 
 involved parties (Debian kernel team, Debian Live maintainers, aufs 
 developers) could cooperate in fixing this bug instead of working against or 
 ignoring each other.

Sorry about this.

As I said before, aufs is included in the kernel package for use in
Debian Live.  The kernel team does not support aufs beyond this.  I
consider it the responsibility of the Debian Live maintainers to let us
know when bug fixes are needed for aufs.  So, Daniel, please identify
the bug fix we need.  If you think we need to make a wholesale update to
aufs, please clear this with the release team.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.


signature.asc
Description: This is a digitally signed message part

Bug#607879: System hangs up with mmap.c:873!

[Daniel Baumann]

On 01/15/2011 03:29 PM, Ben Hutchings wrote:
 Daniel, please identify the bug fix we need.

http://lists.debian.org/debian-live/2011/01/msg00093.html

-- 
Address:Daniel Baumann, Burgunderstrasse 3, CH-4562 Biberist
Email:  daniel.baum...@progress-technologies.net
Internet:   http://people.progress-technologies.net/~daniel.baumann/



-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Ben Hutchings]

On Sat, 2011-01-15 at 15:36 +0100, Daniel Baumann wrote:
 On 01/15/2011 03:29 PM, Ben Hutchings wrote:
  Daniel, please identify the bug fix we need.
 
 http://lists.debian.org/debian-live/2011/01/msg00093.html

Well, we might as well just remove aufs then.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.


signature.asc
Description: This is a digitally signed message part

Bug#607879: System hangs up with mmap.c:873!

[Ben Hutchings]

On Sat, 2011-01-15 at 14:42 +, Ben Hutchings wrote:
 On Sat, 2011-01-15 at 15:36 +0100, Daniel Baumann wrote:
  On 01/15/2011 03:29 PM, Ben Hutchings wrote:
   Daniel, please identify the bug fix we need.
  
  http://lists.debian.org/debian-live/2011/01/msg00093.html
 
 Well, we might as well just remove aufs then.

(No, that's not serious.)

The attached patch series might deal with this bug.  These patches are
simply cherry-picked from aufs upstream and modified as necessary to
compile.  I leave any further building and testing to you.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.
From: J. R. Okajima hooano...@yahoo.co.jp
Date: Mon, 1 Mar 2010 17:52:50 +0900
Subject: [PATCH 1/4] Revert aufs: narrow down the BKL region

commit e62ca9737674cf9b70a961cb8d1efed4a7cff976 in aufs2-2.6

This reverts commit d84deeb079e09b33c2339bc7a54cf7d15c3b8a85.
BKL doesn't help the multi threaded application.
Lockdep says mmap_sem is circular here. It may be correct, but I am not
sure whether it is false positive or not in real world.

Reported-by: James . jazzr...@gmail.com
Signed-off-by: J. R. Okajima hooano...@yahoo.co.jp
---
 fs/aufs/f_op.c |   14 --
 1 files changed, 4 insertions(+), 10 deletions(-)

diff --git a/fs/aufs/f_op.c b/fs/aufs/f_op.c
index 3e9e47d..4c4ef82 100644
--- a/fs/aufs/f_op.c
+++ b/fs/aufs/f_op.c
@@ -597,11 +597,8 @@ static int aufs_mmap(struct file *file, struct vm_area_struct *vma)
 	up_write(current-mm-mmap_sem);
 	si_read_lock(sb, AuLock_FLUSH);
 	err = au_reval_and_lock_fdi(file, au_reopen_nondir, /*wlock*/1);
-	if (unlikely(err)) {
-		down_write(current-mm-mmap_sem);
-		unlock_kernel();
+	if (unlikely(err))
 		goto out;
-	}
 
 	mmapped = !!au_test_mmapped(file);
 	if (wlock) {
@@ -609,16 +606,11 @@ static int aufs_mmap(struct file *file, struct vm_area_struct *vma)
 
 		err = au_ready_to_write(file, -1, pin);
 		di_downgrade_lock(dentry, AuLock_IR);
-		if (unlikely(err)) {
-			down_write(current-mm-mmap_sem);
-			unlock_kernel();
+		if (unlikely(err))
 			goto out_unlock;
-		}
 		au_unpin(pin);
 	} else
 		di_downgrade_lock(dentry, AuLock_IR);
-	down_write(current-mm-mmap_sem);
-	unlock_kernel();
 
 	h_file = au_h_fptr(file, au_fbstart(file));
 	if (!mmapped  au_test_fs_bad_mapping(h_file-f_dentry-d_sb)) {
@@ -670,6 +662,8 @@ static int aufs_mmap(struct file *file, struct vm_area_struct *vma)
 	fi_write_unlock(file);
  out:
 	si_read_unlock(sb);
+	down_write(current-mm-mmap_sem);
+	unlock_kernel();
 	return err;
 }
 
-- 
1.7.2.3

From: J. R. Okajima hooano...@yahoo.co.jp
Date: Mon, 1 Mar 2010 23:13:34 +0900
Subject: [PATCH 2/4] Revert aufs: bugfix, unlock mmap_sem temporary using BKL

commit 639e607997502dfe7dbe140c8de5d81ba99d4240 in aufs2-2.6

This reverts commit 4b70e6f04d4292d8b5ce6cd7ac7371e68eab9175.
BKL doesn't help the multi threaded application.
Lockdep says mmap_sem is circular here. It may be correct, but I am not
sure whether it is false positive or not in real world.

Reported-by: James . jazzr...@gmail.com
---
 fs/aufs/f_op.c |   15 ---
 1 files changed, 0 insertions(+), 15 deletions(-)

diff --git a/fs/aufs/f_op.c b/fs/aufs/f_op.c
index 4c4ef82..6f89992 100644
--- a/fs/aufs/f_op.c
+++ b/fs/aufs/f_op.c
@@ -25,7 +25,6 @@
 #include linux/mman.h
 #include linux/mm.h
 #include linux/security.h
-#include linux/smp_lock.h
 #include aufs.h
 
 /* common function to regular file and dir */
@@ -583,18 +582,6 @@ static int aufs_mmap(struct file *file, struct vm_area_struct *vma)
 	dentry = file-f_dentry;
 	wlock = !!(file-f_mode  FMODE_WRITE)  (vma-vm_flags  VM_SHARED);
 	sb = dentry-d_sb;
-	/*
-	 * Very ugly BKL approach to keep the order of locks.
-	 * Here mm-mmap_sem is acquired by our caller.
-	 *
-	 * native readdir, i_mutex, copy_to_user, mmap_sem
-	 * aufs readdir, i_mutex, rwsem, nested-i_mutex, copy_to_user, mmap_sem
-	 * aufs mmap, mmap_sem, rwsem
-	 *
-	 * Unlock it temporary.
-	 */
-	lock_kernel();
-	up_write(current-mm-mmap_sem);
 	si_read_lock(sb, AuLock_FLUSH);
 	err = au_reval_and_lock_fdi(file, au_reopen_nondir, /*wlock*/1);
 	if (unlikely(err))
@@ -662,8 +649,6 @@ static int aufs_mmap(struct file *file, struct vm_area_struct *vma)
 	fi_write_unlock(file);
  out:
 	si_read_unlock(sb);
-	down_write(current-mm-mmap_sem);
-	unlock_kernel();
 	return err;
 }
 
-- 
1.7.2.3

From: J. R. Okajima hooano...@yahoo.co.jp
Date: Mon, 8 Mar 2010 23:45:56 +0900
Subject: [PATCH 3/4] aufs: bugfix, another approach to keep the lock order of mmap_sem

commit d986fa5a8557f6861fcac4106b6d75301bf5d118 in aufs2-2.6

The previous approach
	4b70e6f aufs: bugfix, unlock mmap_sem temporary using BKL
was bad and already reverted.
This approach is ugly too, but it works.

- split aufs_mmap() into two parts.
- the first part is for copy-ing up which requires rwsem and executed by
  aufsd workqueue.
- the second part is generic_file_mmap() and customizing vm_ops, and
  executed by the original context.
- to protect customizing 

Bug#607879: System hangs up with mmap.c:873!

[Timo Juhani Lindfors]

Ronny Standtke ronny.stand...@gmx.net writes:
 Just boot the DVD in your home machine, the installer starts up automatically 
 and makes producing a Debian Live USB flash drive a breeze. Then you can test 
 on any machine that can boot from USB.

Well I don't have any DVDs and the cdrom drive is not easily
accessible. I tried booting

http://www.imedias.ch/dateien/lernstick-testversion/lernstick_debian5_2010-11-09.iso

with xen configuration

name = 'lernstick'
vcpus = '1'
memory = '1024'
kernel = '/usr/lib/xen/boot/hvmloader'
builder = 'hvm'
device_model='/usr/lib/xen/bin/qemu-dm'
disk = [ 
'file:/local/xen/lernstick/lernstick_debian5_2010-11-09.iso,hdc:cdrom,r',
 'file:/local/xen/lernstick/usb.img,hda,w' ]
vif = [ 'bridge=eth0,mac=52:54:00:12:35:11' ]
vnc=1
vnclisten='0.0.0.0'
vncdisplay=22
vncpasswd='secret'
keymap='sv'
usbdevice='tablet'
localtime=1
boot = 'd'

but it gives me an error message in german and does not let me
copypaste it :-) It seems to not recognize my empty usb.img as an usb
stick.

I tried creating partition with fdisk and filesystem with mkfs.vfat
but it still can't recognize it as an usb stick.




-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Ronny Standtke]

 I tried booting
 http://www.imedias.ch/dateien/lernstick-testversion/lernstick_debian5_2010-
 11-09.iso

The bug does not happen with Debian 5. Please use the latest Debian 6 version, 
e.g.
http://www.imedias.ch/dateien/lernstick-
testversion/lernstick_debian6_2011-01-11.iso

 but it gives me an error message in german and does not let me
 copypaste it :-)

In the Debian 5 version you used, you could have switched the system locale by 
pressing the F1 button in the syslinux gfxboot screen.
Unfortunately, the gfxboot screen is not yet available in the Debian 6 
version. You can switch the system locale by pressing TAB in the syslinux menu 
and replace de_CH with en_US to get an English system.

 It seems to not recognize my empty usb.img as an usb stick.

Unfortunately, I do not know anything about xen...



-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Ben Hutchings]

The Debian kernel team really doesn't like to support aufs; we include
it only to support Debian Live.  Therefore I'm passing this bug over to
the Debian Live maintainers.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.


signature.asc
Description: This is a digitally signed message part

Bug#607879: System hangs up with mmap.c:873!

[Ronny Standtke]

 ok, does it happen if you boot the livecd under qemu?

I just tested it. Booting the DVD ISO in qemu took more than 20 minutes to get 
to a console prompt. My usual setting is Debian Live on a USB flash drive with 
a live-rw persistency partition. This setup boots from the USB flash drive 
in around 50 seconds on my machine.

Starting up KDE did not finish in the qemu test (I stopped it after 40 
minutes). Shutting down in qemu took almost three minutes, in my usual setting 
it takes around ten seconds.

I did not experience the system hang in this one single qemu test run (I also 
do not run into the system hangup on real hardware all the time) but I fail to 
see the value in retrying with quemu as it is slow beyond repair. We better 
test with real hardware and real scenarios.

Best regards

Ronny



-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Timo Juhani Lindfors]

Ronny Standtke ronny.stand...@gmx.net writes:
 I just tested it. Booting the DVD ISO in qemu took more than 20 minutes to 
 get 

How much RAM did you assign to the virtualized machine? I think the
default is 128MB.

 I did not experience the system hang in this one single qemu test run (I also 
 do not run into the system hangup on real hardware all the time) but I fail 
 to 
 see the value in retrying with quemu as it is slow beyond repair. We better 
 test with real hardware and real scenarios.

Sure, I was just asking if there would be something I could easily
test here. I don't have a machine with cdrom drive easily accessible
(my work machine is a geode, my laptop does not have cdrom and my home
machine is in a closet to make it more quiet...)

-Timo





-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Ronny Standtke]

 How much RAM did you assign to the virtualized machine? I think the
 default is 128MB.

I gave the virtual machine 800 MB RAM on a host with 2 GB RAM.

 Sure, I was just asking if there would be something I could easily
 test here. I don't have a machine with cdrom drive easily accessible
 (my work machine is a geode, my laptop does not have cdrom and my home
 machine is in a closet to make it more quiet...)

If your machines can boot from USB flash drives you can download the Debian 
Live version we created for Swiss schools (follow the download link on the 
webpage to download the Debian 6 version):
http://www.imedias.ch/lernstick/lernstick_en

Just boot the DVD in your home machine, the installer starts up automatically 
and makes producing a Debian Live USB flash drive a breeze. Then you can test 
on any machine that can boot from USB.

Regards

Ronny



-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Ronny Standtke]

I can confirm this bug.
I created a Debian Live system based on Squeeze and on many machines the 
system just hangs when shutting down.

My latest test was on a Dell Optiplex SX280. I will attach the entries in 
kern.log and the output of lspci on this system.
00:00.0 Host bridge: Intel Corporation 82915G/P/GV/GL/PL/910GL Memory 
Controller Hub (rev 04)
00:01.0 PCI bridge: Intel Corporation 82915G/P/GV/GL/PL/910GL PCI Express Root 
Port (rev 04)
00:02.0 VGA compatible controller: Intel Corporation 82915G/GV/910GL Integrated 
Graphics Controller (rev 04)
00:02.1 Display controller: Intel Corporation 82915G Integrated Graphics 
Controller (rev 04)
00:1c.0 PCI bridge: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) PCI 
Express Port 1 (rev 03)
00:1d.0 USB Controller: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) 
USB UHCI #1 (rev 03)
00:1d.1 USB Controller: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) 
USB UHCI #2 (rev 03)
00:1d.2 USB Controller: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) 
USB UHCI #3 (rev 03)
00:1d.3 USB Controller: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) 
USB UHCI #4 (rev 03)
00:1d.7 USB Controller: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) 
USB2 EHCI Controller (rev 03)
00:1e.0 PCI bridge: Intel Corporation 82801 PCI Bridge (rev d3)
00:1e.2 Multimedia audio controller: Intel Corporation 82801FB/FBM/FR/FW/FRW 
(ICH6 Family) AC'97 Audio Controller (rev 03)
00:1f.0 ISA bridge: Intel Corporation 82801FB/FR (ICH6/ICH6R) LPC Interface 
Bridge (rev 03)
00:1f.1 IDE interface: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) 
IDE Controller (rev 03)
00:1f.2 IDE interface: Intel Corporation 82801FB/FW (ICH6/ICH6W) SATA 
Controller (rev 03)
00:1f.3 SMBus: Intel Corporation 82801FB/FBM/FR/FW/FRW (ICH6 Family) SMBus 
Controller (rev 03)
02:00.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5751 Gigabit 
Ethernet PCI Express (rev 01)
Dec 28 10:44:58 debian kernel: [   84.124117] [ cut here ]
Dec 28 10:44:58 debian kernel: [   84.124191] kernel BUG at /build/buildd-linux-2.6_2.6.32-29-i386-Of6Yt1/linux-2.6-2.6.32/debian/build/source_i386_none/mm/mmap.c:873!
Dec 28 10:44:58 debian kernel: [   84.124346] invalid opcode:  [#1] SMP 
Dec 28 10:44:58 debian kernel: [   84.124413] last sysfs file: /sys/devices/virtual/sound/timer/uevent
Dec 28 10:44:58 debian kernel: [   84.124496] Modules linked in: ppdev lp sco bridge stp bnep l2cap bluetooth rfkill vboxnetadp vboxnetflt vboxdrv uinput fuse dm_crypt dm_mod snd_intel8x0 snd_ac97_codec ac97_bus snd_pcm_oss snd_mixer_oss snd_pcm snd_seq_midi snd_rawmidi snd_seq_midi_event snd_seq snd_timer snd_seq_device snd parport_pc dcdbas soundcore parport psmouse pcspkr snd_page_alloc i2c_i801 rng_core serio_raw evdev processor ext4 jbd2 crc16 ohci_hcd squashfs loop aufs(C) nls_utf8 nls_cp437 vfat fat ext2 mbcache ide_generic ide_core sg i915 sd_mod crc_t10dif drm_kms_helper sr_mod cdrom drm i2c_algo_bit usbhid hid usb_storage ata_generic tg3 i2c_core ata_piix video libata thermal uhci_hcd libphy output ehci_hcd scsi_mod button thermal_sys usbcore nls_base [last unloaded: scsi_wait_scan]
Dec 28 10:44:58 debian kernel: [   84.125836] 
Dec 28 10:44:58 debian kernel: [   84.125864] Pid: 2475, comm: console-kit-dae Tainted: G C (2.6.32-5-686 #1) OptiPlex SX280   
Dec 28 10:44:58 debian kernel: [   84.125993] EIP: 0060:[c109dd1c] EFLAGS: 00010202 CPU: 0
Dec 28 10:44:58 debian kernel: [   84.126071] EIP is at find_mergeable_anon_vma+0xbd/0x14b
Dec 28 10:44:58 debian kernel: [   84.126141] EAX: dbdab478 EBX: dbdab490 ECX: dbdab478 EDX: dc39bdc0
Dec 28 10:44:58 debian kernel: [   84.126224] ESI: dc39b0b0 EDI: dc39bdc0 EBP: 00100073 ESP: db99be80
Dec 28 10:44:58 debian kernel: [   84.126306]  DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
Dec 28 10:44:58 debian kernel: [   84.126378] Process console-kit-dae (pid: 2475, ti=db99a000 task=dcdfb740 task.ti=db99a000)
Dec 28 10:44:58 debian kernel: [   84.126484] Stack:
Dec 28 10:44:58 debian kernel: [   84.126514]  dbb56200 e00e6a3f db9b06e0 b66d6000 dc39b0b0  dc39b0b0 
Dec 28 10:44:58 debian kernel: [   84.126654] 0 dc3bba00 c10a22bd 0001 c1856c60 dd780b64 b66d6218 c1099709 b66d6218
Dec 28 10:44:58 debian kernel: [   84.126808] 0 dc39b0b0 dc3bba00 0001 b66b9000 0200 c126d2bf df1e6d00 e00e6593
Dec 28 10:44:58 debian kernel: [   84.126967] Call Trace:
Dec 28 10:44:58 debian kernel: [   84.127019]  [e00e6a3f] ? aufs_fault+0xf1/0xfb [aufs]
Dec 28 10:44:58 debian kernel: [   84.128004]  [c10a22bd] ? anon_vma_prepare+0x1d/0x8c
Dec 28 10:44:58 debian kernel: [   84.136836]  [c1099709] ? __do_fault+0xb2/0x3b1
Dec 28 10:44:58 debian kernel: [   84.136836]  [c126d2bf] ? down_read+0x8/0x16
Dec 28 10:44:58 debian kernel: [   84.136836]  [e00e6593] ? si_read_lock+0x7b/0x83 [aufs]
Dec 28 10:44:58 debian kernel: [   84.136836]  [c109b62b] ? handle_mm_fault+0x48f/0x959
Dec 28 10:44:58 debian kernel: [   

Bug#607879: System hangs up with mmap.c:873!

[Timo Juhani Lindfors]

Ronny Standtke ronny.stand...@gmx.net writes:
 I can confirm this bug.
 I created a Debian Live system based on Squeeze and on many machines the 
 system just hangs when shutting down.

 My latest test was on a Dell Optiplex SX280. I will attach the entries in 
 kern.log and the output of lspci on this system.

ok, does it happen if you boot the livecd under qemu?




-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Lukasz Szotek]

I can confirm this bug too.
It happens under VirtualBox and 2 testing PCs.

But I think I managed to solve this problem for me.
After I removed package 'console-kit-dae' it looks stable.
So I think this is problem with console-kit-dae or console-kit-dae+aufs.

kern.log
Description: Binary data

Bug#607879: Fwd: Bug#607879: System hangs up with mmap.c:873!

[Grzegorz Wyrobek]

 Are you creating them with live-helper?

No.

 After I removed package 'console-kit-dae' it looks stable.

So it will be checked.

Bug#607879: System hangs up with mmap.c:873!

[Grzegorz Wyrobek]

So it appears that the aufs is the problem ?
We need to use aufs because it is Live system (squashfs).

When the Linux 2.6.36-2.dmz.5-liquorix-686 #1 was used the problem doesn't
exists ! Unfortunelly we cannont use that kernel.
So I'm looking forward to fix this bug.
We are going to make test ISO tomorrow for you.

Bug#607879: System hangs up with mmap.c:873!

[Timo Juhani Lindfors]

Grzegorz Wyrobek w.gri...@gmail.com writes:
 So I'm looking forward to fix this bug.
 We are going to make test ISO tomorrow for you.

If the ISO includes non-free nvidia drivers then I don't think I want
to test it either. Are you creating them with live-helper?




-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[Ben Hutchings]

On Thu, 2010-12-23 at 22:41 +0100, Grzegorz Wyrobek wrote:
 
  Ps. Last test was to unload all nvidia drivers from
 customized cd and
  testing on intel graphics card. So far looks good but what
 about
  nvidia gpu ?
 
 nouveau seems to provide 2D functionality.  For 3D
 acceleration, you are
 out of luck.  Blame Nvidia for not working with the Linux
 community.
 
 Thank you for fast reply.
 As I mentioned on the begginning the problem shows up on different
 pc's - even on that with intel card.

You originally said it *didn't* show up with an Intel card.

 The last test was prepared on pc with intel card to make sure that
 this is not a nvidia problem and after 2 hours of promising tests it
 shows the same bad behavior !
 Aufs is needed because it is live system so it must be used or maybe
 there is a better alternative ?
[...]

aufs is a horrible hack; we added it to the kernel package to let Debian
Live work but we don't encourage anyone to use it.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.


signature.asc
Description: This is a digitally signed message part

Bug#607879: System hangs up with mmap.c:873!

[Timo Juhani Lindfors]

gring...@poczta.onet.pl writes:
 Dec  7 12:16:26 wokanda1 kernel: [   24.817488]  [f08bda3f] ? 
 aufs_fault+0xf1/0xfb [aufs]

Hmm, how are you using aufs? (I have never used it so I don't even
know how to set it up to test anything...)






--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org

Bug#607879: System hangs up with mmap.c:873!

[gringo_w]

Package: Kernel
Version: 2.6
I have Debian Linux SQUEEZE version 2.6.32-3-686 (Debian 2.6.32-9)  (gcc 
version 4.3.4 (Debian 4.3.4-8) ) #1 SMP Thu Feb 25 06:14:20 UTC 2010 custimized 
on my own cd.
I had also the made the same cd on debian lenny version  and there wasn't any 
problem at all.
 
In Squeeze I've tested on a few pc's and had a same problem after few reboots 
system hangs up.Here is how to reproduce problem:
I've noticed that sometimes (very randomly after reboot)  when i log into 
console after typing password and confirm it , sometimes I must wait a few 
seconds to get the acces to console which is obviously abnormal, then when I 
look into var/log/kern.log file it gives a bug info  (every time the same on 
different pc's)  but i dont'know were to search the problem. 
Also after doing the reboot command when there is an error in kern.log file the 
systems stops few services and hangs up and only hard reset can help.Here is 
tail of the kern.log file:
 
Dec  7 12:16:26 wokanda1 kernel: [   24.817139] [ cut here 
]
Dec  7 12:16:26 wokanda1 kernel: [   24.817154] kernel BUG at 
/build/buildd-linux-2.6_2.6.32-28-i386-zIFx2J/linux-2.6-2.6.32/debian/build/source_i386_none/mm/mmap.c:873!
Dec  7 12:16:26 wokanda1 kernel: [   24.817166] invalid opcode:  [#1] SMP
Dec  7 12:16:26 wokanda1 kernel: [   24.817175] last sysfs file: 
/sys/devices/virtual/vc/vcsa7/uevent
Dec  7 12:16:26 wokanda1 kernel: [   24.817182] Modules linked in: rfcomm sco 
l2cap crc16 bluetooth rfkill fuse snd_hda_codec_nvhdmi snd_hda_codec_realtek 
snd_hda_intel sg snd_hda_codec snd_hwdep snd_pcm snd_seq snd_timer 
snd_seq_device snd sr_mod shpchp cdrom wmi evdev joydev nvidia(P) pci_hotplug 
soundcore psmouse i2c_nforce2 pcspkr serio_raw snd_page_alloc i2c_core button 
processor usbhid hid squashfs loop aufs(C) nls_utf8 nls_cp437 vfat fat 
ide_generic ide_core usb_storage sd_mod crc_t10dif ahci ohci_hcd libata 
ehci_hcd thermal forcedeth thermal_sys scsi_mod usbcore nls_base [last 
unloaded: scsi_wait_scan]
Dec  7 12:16:26 wokanda1 kernel: [   24.817305]
Dec  7 12:16:26 wokanda1 kernel: [   24.817316] Pid: 2158, comm: 
console-kit-dae Tainted: P C (2.6.32-5-686 #1) MCP79
Dec  7 12:16:26 wokanda1 kernel: [   24.817326] EIP: 0060:[c109dc2c] EFLAGS: 
00010202 CPU: 0
Dec  7 12:16:26 wokanda1 kernel: [   24.817341] EIP is at 
find_mergeable_anon_vma+0xbd/0x14b
Dec  7 12:16:26 wokanda1 kernel: [   24.817349] EAX: ede63e18 EBX: ede63e30 
ECX: ede63e18 EDX: ed03e9f8
Dec  7 12:16:26 wokanda1 kernel: [   24.817358] ESI: ed03e7e8 EDI: ed03e9f8 
EBP: 00100073 ESP: eda39e80
Dec  7 12:16:26 wokanda1 kernel: [   24.817367]  DS: 007b ES: 007b FS: 00d8 GS: 
00e0 SS: 0068
Dec  7 12:16:26 wokanda1 kernel: [   24.817377] Process console-kit-dae (pid: 
2158, ti=eda38000 task=ee823fc0 task.ti=eda38000)
Dec  7 12:16:26 wokanda1 kernel: [   24.817384] Stack:
Dec  7 12:16:26 wokanda1 kernel: [   24.817389]  ede76f00 f08bda3f ecc88ea0 
b65f ed03e7e8  ed03e7e8 
Dec  7 12:16:26 wokanda1 kernel: [   24.817408] 0 ef96c700 c10a21d1 0001 
c1970e20 ed177b64 b65f0218 c1099619 b65f0218
Dec  7 12:16:26 wokanda1 kernel: [   24.817429] 0 ed03e7e8 ef96c700 0001 
b65d3000 0200 c126ceff ed1b6b00 f08bd593
Dec  7 12:16:26 wokanda1 kernel: [   24.817451] Call Trace:
Dec  7 12:16:26 wokanda1 kernel: [   24.817488]  [f08bda3f] ? 
aufs_fault+0xf1/0xfb [aufs]
Dec  7 12:16:26 wokanda1 kernel: [   24.817503]  [c10a21d1] ? 
anon_vma_prepare+0x1d/0x8c
Dec  7 12:16:26 wokanda1 kernel: [   24.817514]  [c1099619] ? 
__do_fault+0xb2/0x3b1
Dec  7 12:16:26 wokanda1 kernel: [   24.817528]  [c126ceff] ? 
down_read+0x8/0x16
Dec  7 12:16:26 wokanda1 kernel: [   24.817558]  [f08bd593] ? 
si_read_lock+0x7b/0x83 [aufs]
Dec  7 12:16:26 wokanda1 kernel: [   24.817570]  [c109b53b] ? 
handle_mm_fault+0x48f/0x959
Dec  7 12:16:26 wokanda1 kernel: [   24.817578]  [c109e792] ? 
vma_link+0x4e/0x68
Dec  7 12:16:26 wokanda1 kernel: [   24.817586]  [c109fa8e] ? 
mmap_region+0x328/0x444
Dec  7 12:16:26 wokanda1 kernel: [   24.817595]  [c126f2a0] ? 
do_page_fault+0x2f1/0x307
Dec  7 12:16:26 wokanda1 kernel: [   24.817603]  [c126efaf] ? 
do_page_fault+0x0/0x307
Dec  7 12:16:26 wokanda1 kernel: [   24.817610]  [c126d803] ? 
error_code+0x73/0x78
Dec  7 12:16:26 wokanda1 kernel: [   24.817615] Code: c8 eb 13 8b 50 0c 85 d2 
74 14 8b 4c 24 0c 3b 4a 08 72 0b 8b 5b 04 89 c1 85 db 75 d4 eb 02 89 c1 85 c9 
74 03 8b 79 0c 39 f7 74 04 0f 0b eb fe 85 c9 74 49 8b 41 3c 8b 5f 14 8b 69 14 
85 c0 89 44
Dec  7 12:16:26 wokanda1 kernel: [   24.817694] EIP: [c109dc2c] 
find_mergeable_anon_vma+0xbd/0x14b SS:ESP 0068:eda39e80
Dec  7 12:16:26 wokanda1 kernel: [   24.817707] ---[ end trace 25e4f973f029f5bf 
]---
Dec  7 12:16:27 wokanda1 kernel: [   25.700076] [drm] Initialized drm 1.1.0 
20060810
Dec  7 12:25:00 wokanda1 kernel: [  538.368945] eth0: link up.
Dec  7 12:25:00 wokanda1 kernel: [  538.370087] ADDRCONF(NETDEV_CHANGE): eth0: 
link becomes ready
Dec  7 

Bug#607879: System hangs up with mmap.c:873!

[Ben Hutchings]

On Thu, 2010-12-23 at 15:24 +0100, gring...@poczta.onet.pl wrote:
 Package: Kernel
 Version: 2.6
 
 I have Debian Linux SQUEEZE version 2.6.32-3-686 (Debian 2.6.32-9)
  (gcc version 4.3.4 (Debian 4.3.4-8) ) #1 SMP Thu Feb 25 06:14:20 UTC
 2010 custimized on my own cd.
 I had also the made the same cd on debian lenny version  and there
 wasn't any problem at all.

2.6.32-9 is way outdated and you should no longer use it.  However, the
messages below indicate you were actually running 2.6.32-28 (almost the
latest).

 In Squeeze I've tested on a few pc's and had a same problem after few
 reboots system hangs up.Here is how to reproduce problem:
 I've noticed that sometimes (very randomly after reboot)  when i log
 into console after typing password and confirm it , sometimes I must
 wait a few seconds to get the acces to console which is obviously
 abnormal, then when I look into var/log/kern.log file it gives a bug
 info  (every time the same on different pc's)  but i dont'know were to
 search the problem. Also after doing the reboot command when there is
 an error in kern.log file the systems stops few services and hangs up
 and only hard reset can help.Here is tail of the kern.log file:
  
 Dec  7 12:16:26 wokanda1 kernel: [   24.817139] [ cut
 here ]
 Dec  7 12:16:26 wokanda1 kernel: [   24.817154] kernel BUG
 at 
 /build/buildd-linux-2.6_2.6.32-28-i386-zIFx2J/linux-2.6-2.6.32/debian/build/source_i386_none/mm/mmap.c:873!
 Dec  7 12:16:26 wokanda1 kernel: [   24.817166] invalid opcode: 
 [#1] SMP 
 Dec  7 12:16:26 wokanda1 kernel: [   24.817175] last sysfs
 file: /sys/devices/virtual/vc/vcsa7/uevent
 Dec  7 12:16:26 wokanda1 kernel: [   24.817182] Modules linked in:
 rfcomm sco l2cap crc16 bluetooth rfkill fuse snd_hda_codec_nvhdmi
 snd_hda_codec_realtek snd_hda_intel sg snd_hda_codec snd_hwdep snd_pcm
 snd_seq snd_timer snd_seq_device snd sr_mod shpchp cdrom wmi evdev
 joydev nvidia(P) pci_hotplug soundcore psmouse i2c_nforce2 pcspkr
 serio_raw snd_page_alloc i2c_core button processor usbhid hid squashfs
 loop aufs(C) nls_utf8 nls_cp437 vfat fat ide_generic ide_core
 usb_storage sd_mod crc_t10dif ahci ohci_hcd libata ehci_hcd thermal
 forcedeth thermal_sys scsi_mod usbcore nls_base [last unloaded:
 scsi_wait_scan]

I won't investigate this unless it is reproducible without nvidia and
aufs.

[...]
 Ps. Last test was to unload all nvidia drivers from customized cd and
 testing on intel graphics card. So far looks good but what about
 nvidia gpu ?

nouveau seems to provide 2D functionality.  For 3D acceleration, you are
out of luck.  Blame Nvidia for not working with the Linux community.

Ben.

-- 
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.


signature.asc
Description: This is a digitally signed message part