Mimi Zohar [Thu, 12 May 2011 22:33:20 +0000 (18:33 -0400)]
evm: permit only valid security.evm xattrs to be updated
In addition to requiring CAP_SYS_ADMIN permission to modify/delete
security.evm, prohibit invalid security.evm xattrs from changing,
unless in fixmode. This patch prevents inadvertent 'fixing' of
security.evm to reflect offline modifications.
Changelog v7:
- rename boot paramater 'evm_mode' to 'evm'
Dmitry Kasatkin [Fri, 6 May 2011 08:34:14 +0000 (11:34 +0300)]
evm: evm_verify_hmac must not return INTEGRITY_UNKNOWN
If EVM is not supported or enabled, evm_verify_hmac() returns
INTEGRITY_UNKNOWN, which ima_appraise_measurement() ignores and sets
the appraisal status based solely on the security.ima verification.
evm_verify_hmac() also returns INTEGRITY_UNKNOWN for other failures, such
as temporary failures like -ENOMEM, resulting in possible attack vectors.
This patch changes the default return code for temporary/unexpected
failures, like -ENOMEM, from INTEGRITY_UNKNOWN to INTEGRITY_FAIL, making
evm_verify_hmac() fail safe.
As a result, failures need to be re-evaluated in order to catch both
temporary errors, such as the -ENOMEM, as well as errors that have been
resolved in fix mode.
Dmitry Kasatkin [Wed, 9 Mar 2011 20:07:36 +0000 (15:07 -0500)]
evm: crypto hash replaced by shash
Using shash is more efficient, because the algorithm is allocated only
once. Only the descriptor to store the hash state needs to be allocated
for every operation.
Changelog v6:
- check for crypto_shash_setkey failure
Mimi Zohar [Thu, 16 Jun 2011 01:19:10 +0000 (21:19 -0400)]
evm: call evm_inode_init_security from security_inode_init_security
Changelog v7:
- moved the initialization call to security_inode_init_security,
renaming evm_inode_post_init_security to evm_inode_init_security
- increase size of xattr array for EVM xattr
Mimi Zohar [Wed, 9 Mar 2011 19:40:44 +0000 (14:40 -0500)]
evm: add evm_inode_init_security to initialize new files
Initialize 'security.evm' for new files.
Changelog v7:
- renamed evm_inode_post_init_security to evm_inode_init_security
- moved struct xattr definition to earlier patch
- allocate xattr name
Changelog v6:
- Use 'struct evm_ima_xattr_data'
Mimi Zohar [Wed, 9 Mar 2011 19:38:26 +0000 (14:38 -0500)]
security: imbed evm calls in security hooks
Imbed the evm calls evm_inode_setxattr(), evm_inode_post_setxattr(),
evm_inode_removexattr() in the security hooks. evm_inode_setxattr()
protects security.evm xattr. evm_inode_post_setxattr() and
evm_inode_removexattr() updates the hmac associated with an inode.
(Assumes an LSM module protects the setting/removing of xattr.)
Changelog:
- Don't define evm_verifyxattr(), unless CONFIG_INTEGRITY is enabled.
- xattr_name is a 'const', value is 'void *'
Dmitry Kasatkin [Wed, 9 Mar 2011 19:28:20 +0000 (14:28 -0500)]
evm: add support for different security.evm data types
EVM protects a file's security extended attributes(xattrs) against integrity
attacks. The current patchset maintains an HMAC-sha1 value across the security
xattrs, storing the value as the extended attribute 'security.evm'. We
anticipate other methods for protecting the security extended attributes.
This patch reserves the first byte of 'security.evm' as a place holder for
the type of method.
Changelog v6:
- move evm_ima_xattr_type definition to security/integrity/integrity.h
- defined a structure for the EVM xattr called evm_ima_xattr_data
(based on Serge Hallyn's suggestion)
- removed unnecessary memset
Mimi Zohar [Tue, 15 Mar 2011 20:12:09 +0000 (16:12 -0400)]
evm: re-release
EVM protects a file's security extended attributes(xattrs) against integrity
attacks. This patchset provides the framework and an initial method. The
initial method maintains an HMAC-sha1 value across the security extended
attributes, storing the HMAC value as the extended attribute 'security.evm'.
Other methods of validating the integrity of a file's metadata will be posted
separately (eg. EVM-digital-signatures).
While this patchset does authenticate the security xattrs, and
cryptographically binds them to the inode, coming extensions will bind other
directory and inode metadata for more complete protection. To help simplify
the review and upstreaming process, each extension will be posted separately
(eg. IMA-appraisal, IMA-appraisal-directory). For a general overview of the
proposed Linux integrity subsystem, refer to Dave Safford's whitepaper:
http://downloads.sf.net/project/linux-ima/linux-ima/Integrity_overview.pdf.
EVM depends on the Kernel Key Retention System to provide it with a
trusted/encrypted key for the HMAC-sha1 operation. The key is loaded onto the
root's keyring using keyctl. Until EVM receives notification that the key has
been successfully loaded onto the keyring (echo 1 > <securityfs>/evm), EVM can
not create or validate the 'security.evm' xattr, but returns INTEGRITY_UNKNOWN.
Loading the key and signaling EVM should be done as early as possible. Normally
this is done in the initramfs, which has already been measured as part of the
trusted boot. For more information on creating and loading existing
trusted/encrypted keys, refer to Documentation/keys-trusted-encrypted.txt. A
sample dracut patch, which loads the trusted/encrypted key and enables EVM, is
available from http://linux-ima.sourceforge.net/#EVM.
Based on the LSMs enabled, the set of EVM protected security xattrs is defined
at compile. EVM adds the following three calls to the existing security hooks:
evm_inode_setxattr(), evm_inode_post_setxattr(), and evm_inode_removexattr. To
initialize and update the 'security.evm' extended attribute, EVM defines three
calls: evm_inode_post_init(), evm_inode_post_setattr() and
evm_inode_post_removexattr() hooks. To verify the integrity of a security
xattr, EVM exports evm_verifyxattr().
Changelog v7:
- Fixed URL in EVM ABI documentation
Changelog v6: (based on Serge Hallyn's review)
- fix URL in patch description
- remove evm_hmac_size definition
- use SHA1_DIGEST_SIZE (removed both MAX_DIGEST_SIZE and evm_hmac_size)
- moved linux include before other includes
- test for crypto_hash_setkey failure
- fail earlier for invalid key
- clear entire encrypted key, even on failure
- check xattr name length before comparing xattr names
Changelog:
- locking based on i_mutex, remove evm_mutex
- using trusted/encrypted keys for storing the EVM key used in the HMAC-sha1
operation.
- replaced crypto hash with shash (Dmitry Kasatkin)
- support for additional methods of verifying the security xattrs
(Dmitry Kasatkin)
- iint not allocated for all regular files, but only for those appraised
- Use cap_sys_admin in lieu of cap_mac_admin
- Use __vfs_setxattr_noperm(), without permission checks, from EVM
Mimi Zohar [Wed, 9 Mar 2011 19:23:34 +0000 (14:23 -0500)]
xattr: define vfs_getxattr_alloc and vfs_xattr_cmp
vfs_getxattr_alloc() and vfs_xattr_cmp() are two new kernel xattr helper
functions. vfs_getxattr_alloc() first allocates memory for the requested
xattr and then retrieves it. vfs_xattr_cmp() compares a given value with
the contents of an extended attribute.
Mimi Zohar [Wed, 9 Mar 2011 19:13:22 +0000 (14:13 -0500)]
integrity: move ima inode integrity data management
Move the inode integrity data(iint) management up to the integrity directory
in order to share the iint among the different integrity models.
Changelog:
- don't define MAX_DIGEST_SIZE
- rename several globally visible 'ima_' prefixed functions, structs,
locks, etc to 'integrity_'
- replace '20' with SHA1_DIGEST_SIZE
- reflect location change in appropriate Kconfig and Makefiles
- remove unnecessary initialization of iint_initialized to 0
- rebased on current ima_iint.c
- define integrity_iint_store/lock as static
Mimi Zohar [Mon, 6 Jun 2011 19:29:25 +0000 (15:29 -0400)]
security: new security_inode_init_security API adds function callback
This patch changes the security_inode_init_security API by adding a
filesystem specific callback to write security extended attributes.
This change is in preparation for supporting the initialization of
multiple LSM xattrs and the EVM xattr. Initially the callback function
walks an array of xattrs, writing each xattr separately, but could be
optimized to write multiple xattrs at once.
For existing security_inode_init_security() calls, which have not yet
been converted to use the new callback function, such as those in
reiserfs and ocfs2, this patch defines security_old_inode_init_security().
Stefan Berger [Wed, 30 Mar 2011 16:13:33 +0000 (12:13 -0400)]
tpm_tis: Probing function for Intel iTPM bug
This patch introduces a function for automatic probing for the Intel iTPM
STS_DATA_EXPECT flaw.
The patch splits the current tpm_tis_send function into 2 parts where the 1st
part is now called tpm_tis_send_data() and merely sends the data to the TPM.
This function is then used for probing. The new tpm_tis_send function now
first calls tpm_tis_send_data and if that succeeds has the TPM process the
command and waits until the response is there.
The probing for the Intel iTPM is only invoked if the user has not passed
itpm=1 as parameter for the module *or* if such a TPM was detected via ACPI.
Previously it was necessary to pass itpm=1 when also passing force=1 to the
module when doing a 'modprobe'. This function is more general than the ACPI
test function and the function relying on ACPI could probably be removed.
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com> Signed-off-by: Rajiv Andrade <srajiv@linux.vnet.ibm.com>
Stefan Berger [Wed, 30 Mar 2011 16:13:32 +0000 (12:13 -0400)]
tpm_tis: Fix the probing for interrupts
This patch fixes several aspects of the probing for interrupts.
This patch reads the TPM's timeouts before probing for the interrupts. The
tpm_get_timeouts() function is invoked in polling mode and gets the proper
timeouts from the TPM so that we don't need to fall back to 2 minutes timeouts
for short duration commands while the interrupt probing is happening.
This patch introduces a variable probed_irq into the vendor structure that gets
the irq number if an interrupt is received while the the tpm_gen_interrupt()
function is run in polling mode during interrupt probing. Previously some
parts of tpm_gen_interrupt() were run in polling mode, then the irq variable
was set in the interrupt handler when an interrupt was received and execution
of tpm_gen_interrupt() ended up switching over to interrupt mode.
tpm_gen_interrupt() execution ended up on an event queue where it eventually
timed out since the probing handler doesn't wake any queues.
Before calling into free_irq() clear all interrupt flags that may have
been set by the TPM. The reason is that free_irq() will call into the probing
interrupt handler and may otherwise fool us into thinking that a real interrupt
happened (because we see the flags as being set) while the TPM's interrupt line
is not even connected to anything on the motherboard. This solves a problem
on one machine I did testing on (Thinkpad T60).
If a TPM claims to use a specifc interrupt, the probing is done as well
to verify that the interrupt is actually working. If a TPM indicates
that it does not use a specific interrupt (returns '0'), probe all interrupts
from 3 to 15.
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com> Signed-off-by: Rajiv Andrade <srajiv@linux.vnet.ibm.com>
Stefan Berger [Wed, 30 Mar 2011 16:13:31 +0000 (12:13 -0400)]
tpm_tis: Delay ACPI S3 suspend while the TPM is busy
This patch delays the (ACPI S3) suspend while the TPM is busy processing a
command and the TPM TIS driver is run in interrupt mode. This is the same
behavior as we already have it for the TPM TIS driver in polling mode.
Reasoning: Some of the TPM's commands advance the internal state of the TPM.
An example would be the extending of one of its PCR registers. Upper layers,
such as IMA or TSS (TrouSerS), would certainly want to be sure that the
command succeeded rather than getting an error code (-62 = -ETIME) that may
not give a conclusive answer as for what reason the command failed. Reissuing
such a command would put the TPM into the wrong state, so waiting for it to
finish is really the only option.
The downside is that some commands (key creation) can take a long time and
actually prevent the machine from entering S3 at all before the 20 second
timeout of the power management subsystem arrives.
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com> Signed-off-by: Rajiv Andrade <srajiv@linux.vnet.ibm.com>
Stefan Berger [Wed, 30 Mar 2011 16:13:30 +0000 (12:13 -0400)]
tpm_tis: Re-enable interrupts upon (S3) resume
This patch makes sure that if the TPM TIS interface is run in interrupt mode
(rather than polling mode) that all interrupts are enabled in the TPM's
interrupt enable register after a resume from ACPI S3 suspend. The registers
may either have been cleared by the TPM loosing its state during device sleep
or by the BIOS leaving the TPM in polling mode (after sending a command to
the TPM for starting it up again)
You may want to check if your TPM runs with interrupts by doing
cat /proc/interrupts | grep -i tpm
and see whether there is an entry or otherwise for it to use interrupts:
modprobe tpm_tis interrupts=1 [add 'itpm=1' for Intel TPM ]
v2:
- the patch was adapted to work with the pnp and platform driver
implementations in tpm_tis.c
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com> Signed-off-by: Rajiv Andrade <srajiv@linux.vnet.ibm.com>
Stefan Berger [Wed, 30 Mar 2011 16:13:29 +0000 (12:13 -0400)]
tpm: Fix display of data in pubek sysfs entry
This patch fixes the TPM's pubek sysfs entry that is accessible as long
as the TPM doesn't have an owner. It was necessary to shift the access to the
data by -10 -- the first byte immediately follows the 10 byte header. The
line
data = tpm_cmd.params.readpubek_out_buffer;
sets it at the offset '10' in the packet, so we can read the data array
starting at offset '0'.
Before:
Algorithm: 00 0C 00 00
Encscheme: 08 00
Sigscheme: 00 00
Parameters: 00 00 00 00 01 00 AC E2 5E 3C A0 78
Modulus length: -563306801
Modulus:
28 21 08 0F 82 CD F2 B1 E7 49 F7 74 70 BE 59 8C
43 78 B1 24 EA 52 E2 FE 52 5C 3A 12 3B DC 61 71
[...]
After:
Algorithm: 00 00 00 01
Encscheme: 00 03
Sigscheme: 00 01
Parameters: 00 00 08 00 00 00 00 02 00 00 00 00
Modulus length: 256
Modulus:
AC E2 5E 3C A0 78 DE 6C 9E CF 28 21 08 0F 82 CD
F2 B1 E7 49 F7 74 70 BE 59 8C 43 78 B1 24 EA 52
[...]
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com> Signed-off-by: Rajiv Andrade <srajiv@linux.vnet.ibm.com>
Stefan Berger [Wed, 30 Mar 2011 16:13:28 +0000 (12:13 -0400)]
tpm_tis: Add timeouts sysfs entry
Display the TPM's interface timeouts in a 'timeouts' sysfs entry. Display
the entries as having been adjusted when they were scaled due to their values
being reported in milliseconds rather than microseconds.
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com> Signed-off-by: Rajiv Andrade <srajiv@linux.vnet.ibm.com>
Stefan Berger [Wed, 30 Mar 2011 16:13:27 +0000 (12:13 -0400)]
tpm: Adjust interface timeouts if they are too small
Adjust the interface timeouts if they are found to be too small, i.e., if
they are returned in milliseconds rather than microseconds as we heared
from Infineon that some (old) Infineon TPMs do.
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com> Signed-off-by: Rajiv Andrade <srajiv@linux.vnet.ibm.com>
Stefan Berger [Wed, 30 Mar 2011 16:13:26 +0000 (12:13 -0400)]
tpm: Use interface timeouts returned from the TPM
The TPM driver currently discards the interface timeout values returned
from the TPM. The check of the response packet needs to consider that
the return_code field is 0 on success and the size of the expected
packet is equivalent to the header size + u32 length indicator for the
TPM_GetCapability() result + 4 interface timeout indicators of type u32.
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com> Signed-off-by: Rajiv Andrade <srajiv@linux.vnet.ibm.com>
Stefan Berger [Wed, 30 Mar 2011 16:13:25 +0000 (12:13 -0400)]
tpm_tis: Introduce durations sysfs entry
Display the TPM's command timeouts in a 'durations' sysfs entry. Display
the entries as having been adjusted when they were scaled due to their values
being reported in milliseconds rather than microseconds.
Stefan Berger [Wed, 30 Mar 2011 16:13:24 +0000 (12:13 -0400)]
tpm: Adjust the durations if they are too small
Adjust the durations if they are found to be too small, i.e., if they are
returned in milliseconds rather than microseconds as some Infineon TPMs are
reported to do.
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com> Signed-off-by: Rajiv Andrade <srajiv@linux.vnet.ibm.com>
Stefan Berger [Wed, 30 Mar 2011 16:13:23 +0000 (12:13 -0400)]
tpm: Use durations returned from TPM
The TPM driver currently discards the durations values returned
from the TPM. The check of the response packet needs to consider that
the return_code field is 0 on success and the size of the expected
packet is equivalent to the header size + u32 length indicator for the
TPM_GetCapability() result + 3 timeout indicators of type u32.
v4:
- sysfs entry 'durations' is now a patch of its own
- the work-around for TPMs reporting durations in milliseconds is now in a
patch of its own
v3:
- sysfs entry now called 'durations' to resemble TPM-speak (previously
was called 'timeouts')
v2:
- adjusting all timeouts for TPM devices reporting timeouts in msec rather
than usec
- also displaying in sysfs whether the timeouts are 'original' or 'adjusted'
TOMOYO: Allow using executable's realpath and symlink's target as conditions.
This patch adds support for permission checks using executable file's realpath
upon execve() and symlink's target upon symlink(). Hooks are in the last patch
of this pathset.
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: James Morris <jmorris@namei.org>
TOMOYO: Allow using owner/group etc. of file objects as conditions.
This patch adds support for permission checks using file object's DAC
attributes (e.g. owner/group) when checking file's pathnames. Hooks for passing
file object's pointers are in the last patch of this pathset.
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: James Morris <jmorris@namei.org>
/sys/kernel/security/tomoyo/.domain_status can be easily emulated using
/sys/kernel/security/tomoyo/domain_policy . We can remove this interface by
updating /usr/sbin/tomoyo-setprofile utility.
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: James Morris <jmorris@namei.org>
Linus Torvalds [Wed, 29 Jun 2011 18:07:02 +0000 (11:07 -0700)]
Merge branch 'i2c-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelvare/staging
* 'i2c-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelvare/staging:
i2c/pca954x: Initialize the mux to disconnected state
i2c-taos-evm: Fix log messages
Linus Torvalds [Wed, 29 Jun 2011 18:03:27 +0000 (11:03 -0700)]
Merge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc
* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc:
arch/powerpc: use printk_ratelimited instead of printk_ratelimit
powerpc/rtas-rtc: remove sideeffects of printk_ratelimit
powerpc/pseries: remove duplicate SCSI_BNX2_ISCSI in pseries_defconfig
powerpc/e500: fix breakage with fsl_rio_mcheck_exception
powerpc/p1022ds: fix audio-related properties in the device tree
powerpc/85xx: fix NAND_CMD_READID read bytes number
Linus Torvalds [Wed, 29 Jun 2011 17:10:07 +0000 (10:10 -0700)]
ahci: change 'masking port_map' printk to KERN_WARNING level
It's not so much an error as a warning about normal Marvell crazines.
So don't use KERN_ERR that ends up spamming the console even in quiet
mode, it's not _that_ critical.
Explained by Jeff:
"Long explanation, it's a mess:
Marvell took standard AHCI, and bastardized it to include a weird mode
whereby PATA devices appear inside the AHCI DMA and interrupt
infrastructure you're familiar with.
So, PATA devices appear via pata_marvell driver, using basic legacy
IDE programming interface. But SATA devices, which might also be
attached to this chip, either work in under-performing mode or
simply don't work at all (e.g. newer 6 Gbps devices or port
multiplier attachments, NCQ, ...)
On the other hand, 'ahci' driver loads and works with the chip's
attached SATA devices quite beautifully, but is completely unable to
drive any attached PATA devices, due to the Marvell-specific
PATA-under-AHCI interface.
The "masking port_map 0x7 -> 0x3" message is the ahci driver "hiding"
the PATA port(s) from itself, making sure it will only drive the SATA
ports it knows how to drive."
Acked-by: Jeff Garzik <jgarzik@redhat.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Petri Gynther [Wed, 29 Jun 2011 09:36:11 +0000 (11:36 +0200)]
i2c/pca954x: Initialize the mux to disconnected state
pca954x power-on default is channel 0 connected. If multiple pca954x
muxes are connected to the same physical I2C bus, the parent bus will
see channel 0 devices behind both muxes by default. This is bad.
1. Load I2C bus driver: creates I2C-bus-1
2. Load pca954x driver: creates virtual I2C-bus-101 and I2C-bus-111
3. Load eeprom driver
4. Try to read EEPROM @ 0x50 on I2C-bus-101. The transaction will also bleed
onto I2C-bus-111 because pca954x @ 0x71 channel 0 is connected by default.
Fix: Initialize pca954x to disconnected state in pca954x_probe()
Signed-off-by: Petri Gynther <pgynther@google.com> Signed-off-by: Jean Delvare <khali@linux-fr.org> Cc: stable@kernel.org
powerpc/rtas-rtc: remove sideeffects of printk_ratelimit
Don't use printk_ratelimit() as an additional condition for returning
on an error. Because when the ratelimit is reached, printk_ratelimit
will return 0 and e.g. in rtas_get_boot_time won't check for an error
condition.
Signed-off-by: Christian Dietrich <christian.dietrich@informatik.uni-erlangen.de> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
John Johansen [Sat, 25 Jun 2011 15:57:07 +0000 (16:57 +0100)]
AppArmor: Fix masking of capabilities in complain mode
AppArmor is masking the capabilities returned by capget against the
capabilities mask in the profile. This is wrong, in complain mode the
profile has effectively all capabilities, as the profile restrictions are
not being enforced, merely tested against to determine if an access is
known by the profile.
This can result in the wrong behavior of security conscience applications
like sshd which examine their capability set, and change their behavior
accordingly. In this case because of the masked capability set being
returned sshd fails due to DAC checks, even when the profile is in complain
mode.
Kernels affected: 2.6.36 - 3.0.
Signed-off-by: John Johansen <john.johansen@canonical.com>
John Johansen [Tue, 28 Jun 2011 14:06:38 +0000 (15:06 +0100)]
AppArmor: Fix reference to rcu protected pointer outside of rcu_read_lock
The pointer returned from tracehook_tracer_task() is only valid inside
the rcu_read_lock. However the tracer pointer obtained is being passed
to aa_may_ptrace outside of the rcu_read_lock critical section.
Mover the aa_may_ptrace test into the rcu_read_lock critical section, to
fix this.
Kernels affected: 2.6.36 - 3.0
Reported-by: Oleg Nesterov <oleg@redhat.com> Cc: stable@kernel.org Signed-off-by: John Johansen <john.johansen@canonical.com>
Tetsuo Handa [Sun, 26 Jun 2011 14:22:59 +0000 (23:22 +0900)]
TOMOYO: Make several options configurable.
To be able to start using enforcing mode from the early stage of boot sequence,
this patch adds support for activating access control without calling external
policy loader program. This will be useful for systems where operations which
can lead to the hijacking of the boot sequence are needed before loading the
policy. For example, you can activate immediately after loading the fixed part
of policy which will allow only operations needed for mounting a partition
which contains the variant part of policy and verifying (e.g. running GPG
check) and loading the variant part of policy. Since you can start using
enforcing mode from the beginning, you can reduce the possibility of hijacking
the boot sequence.
This patch makes several variables configurable on build time. This patch also
adds TOMOYO_loader= and TOMOYO_trigger= kernel command line option to boot the
same kernel in two different init systems (BSD-style init and systemd).
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: James Morris <jmorris@namei.org>
Tetsuo Handa [Sun, 26 Jun 2011 14:22:18 +0000 (23:22 +0900)]
TOMOYO: Add built-in policy support.
To be able to start using enforcing mode from the early stage of boot sequence,
this patch adds support for built-in policy configuration (and next patch adds
support for activating access control without calling external policy loader
program).
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: James Morris <jmorris@namei.org>
Tetsuo Handa [Sun, 26 Jun 2011 14:20:55 +0000 (23:20 +0900)]
TOMOYO: Fix lockdep warning.
Currently TOMOYO holds SRCU lock upon open() and releases it upon close()
because list elements stored in the "struct tomoyo_io_buffer" instances are
accessed until close() is called. However, such SRCU usage causes lockdep to
complain about leaving the kernel with SRCU lock held.
This patch solves the warning by holding/releasing SRCU upon each
read()/write(). This patch is doing something similar to calling kfree()
without calling synchronize_srcu(), by selectively deferring kfree() by keeping
track of the "struct tomoyo_io_buffer" instances.
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: James Morris <jmorris@namei.org>
Tetsuo Handa [Sun, 26 Jun 2011 14:20:23 +0000 (23:20 +0900)]
TOMOYO: Change pathname for non-rename()able filesystems.
TOMOYO wants to use /proc/self/ rather than /proc/$PID/ if $PID matches current
thread's process ID in order to prevent current thread from accessing other
process's information unless needed.
But since procfs can be mounted on various locations (e.g. /proc/ /proc2/ /p/
/tmp/foo/100/p/ ), TOMOYO cannot tell that whether the numeric part in the
string returned by __d_path() represents process ID or not.
Therefore, to be able to convert from $PID to self no matter where procfs is
mounted, this patch changes pathname representations for filesystems which do
not support rename() operation (e.g. proc, sysfs, securityfs).
Tetsuo Handa [Sun, 26 Jun 2011 14:19:52 +0000 (23:19 +0900)]
TOMOYO: Add policy namespace support.
Mauras Olivier reported that it is difficult to use TOMOYO in LXC environments,
for TOMOYO cannot distinguish between environments outside the container and
environments inside the container since LXC environments are created using
pivot_root(). To address this problem, this patch introduces policy namespace.
Each policy namespace has its own set of domain policy, exception policy and
profiles, which are all independent of other namespaces. This independency
allows users to develop policy without worrying interference among namespaces.
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: James Morris <jmorris@namei.org>
Tetsuo Handa [Sun, 26 Jun 2011 14:18:58 +0000 (23:18 +0900)]
TOMOYO: Add auditing interface.
Add /sys/kernel/security/tomoyo/audit interface. This interface generates audit
logs in the form of domain policy so that /usr/sbin/tomoyo-auditd can reuse
audit logs for appending to /sys/kernel/security/tomoyo/domain_policy
interface.
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: James Morris <jmorris@namei.org>
Tetsuo Handa [Sun, 26 Jun 2011 14:18:21 +0000 (23:18 +0900)]
TOMOYO: Simplify profile structure.
Remove global preference from profile structure in order to make code simpler.
Due to this structure change, printk() warnings upon policy violation are
temporarily disabled. They will be replaced by
/sys/kernel/security/tomoyo/audit by next patch.
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: James Morris <jmorris@namei.org>
Tetsuo Handa [Sun, 26 Jun 2011 14:17:46 +0000 (23:17 +0900)]
TOMOYO: Rename directives.
Convert "allow_..." style directives to "file ..." style directives.
By converting to the latter style, we can pack policy like
"file read/write/execute /path/to/file".
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: James Morris <jmorris@namei.org>
Tetsuo Handa [Sun, 26 Jun 2011 14:16:36 +0000 (23:16 +0900)]
TOMOYO: Cleanup part 3.
Use common structure for ACL with "struct list_head" + "atomic_t".
Use array/struct where possible.
Remove is_group from "struct tomoyo_name_union"/"struct tomoyo_number_union".
Pass "struct file"->private_data rather than "struct file".
Update some of comments.
Bring tomoyo_same_acl_head() from common.h to domain.c .
Bring tomoyo_invalid()/tomoyo_valid() from common.h to util.c .
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: James Morris <jmorris@namei.org>
Tetsuo Handa [Sun, 26 Jun 2011 14:15:31 +0000 (23:15 +0900)]
TOMOYO: Cleanup part 1.
In order to synchronize with TOMOYO 1.8's syntax,
(1) Remove special handling for allow_read/write permission.
(2) Replace deny_rewrite/allow_rewrite permission with allow_append permission.
(3) Remove file_pattern keyword.
(4) Remove allow_read permission from exception policy.
(5) Allow creating domains in enforcing mode without calling supervisor.
(6) Add permission check for opening directory for reading.
(7) Add permission check for stat() operation.
(8) Make "cat < /sys/kernel/security/tomoyo/self_domain" behave as if
"cat /sys/kernel/security/tomoyo/self_domain".
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: James Morris <jmorris@namei.org>
Michael Neuling [Mon, 27 Jun 2011 19:55:30 +0000 (19:55 +0000)]
powerpc/pseries: remove duplicate SCSI_BNX2_ISCSI in pseries_defconfig
Remove duplicate assignment of SCSI_BNX2_ISCSI in pseries_defconfig
introduced by: 37e0c21e powerpc/pseries: Enable iSCSI support for a number of cards
causes warning:
arch/powerpc/configs/pseries_defconfig:151:warning: override: reassigning to symbol SCSI_BNX2_ISCSI
Signed-off-by: Michael Neuling <mikey@neuling.org> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Linus Torvalds [Tue, 28 Jun 2011 18:15:36 +0000 (11:15 -0700)]
Merge branch 'driver-core-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6
* 'driver-core-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6:
Connector: Correctly set the error code in case of success when dispatching receive callbacks
Connector: Set the CN_NETLINK_USERS correctly
pti: PTI semantics fix in pti_tty_cleanup.
pti: ENXIO error case memory leak PTI fix.
pti: double-free security PTI fix
drivers:misc: ti-st: fix skipping of change remote baud
drivers/base/platform.c: don't mark platform_device_register_resndata() as __init_or_module
st_kim: Handle case of no device found for ID 0
firmware: fix GOOGLE_SMI kconfig dependency warning
Linus Torvalds [Tue, 28 Jun 2011 18:15:17 +0000 (11:15 -0700)]
Merge branch 'usb-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6
* 'usb-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6:
MAINTAINERS: add myself as maintainer of USB/IP
usb: r8a66597-hcd: fix cannot detect low/full speed device
USB: ehci-ath79: fix a NULL pointer dereference
USB: Add new FT232H chip to drivers/usb/serial/ftdi_sio.c
usb/isp1760: Fix bug preventing the unlinking of control urbs
USB: Fix up URB error codes to reflect implementation.
xhci: Always set urb->status to zero for isoc endpoints.
xhci: Add reset on resume quirk for asrock p67 host
xHCI 1.0: Incompatible Device Error
USB: don't let errors prevent system sleep
USB: don't let the hub driver prevent system sleep
USB: change maintainership of ohci-hcd and ehci-hcd
xHCI 1.0: Force Stopped Event(FSE)
xhci: Don't warn about zeroed bMaxBurst descriptor field.
USB: Free bandwidth when usb_disable_device is called.
xhci: Reject double add of active endpoints.
USB: TI 3410/5052 USB Serial Driver: Fix mem leak when firmware is too big.
usb: musb: gadget: clear TXPKTRDY flag when set FLUSHFIFO
usb: musb: host: compare status for negative error values
Linus Torvalds [Tue, 28 Jun 2011 18:14:55 +0000 (11:14 -0700)]
Merge branch 'tty-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty-2.6
* 'tty-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty-2.6:
serial: bcm63xx_uart: fix irq storm after rx fifo overrun.
amba pl011: platform data for reg lockup and glitch v2
amba pl011: workaround for uart registers lockup
tty: n_gsm: improper skb_pull() use was leaking framed data
tty: n_gsm: Fixed logic to decode break signal from modem status
TTY: ntty, add one more sanity check
TTY: ldisc, do not close until there are readers
8250: Fix capabilities when changing the port type
8250_pci: Fix missing const from merges
ARM: SAMSUNG: serial: Fix on handling of one clock source for UART
serial: ioremap warning fix for jsm driver.
8250_pci: add -ENODEV code for Intel EG20T PCH
Linus Torvalds [Tue, 28 Jun 2011 18:14:22 +0000 (11:14 -0700)]
Merge branch 'staging-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging-2.6
* 'staging-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging-2.6:
Staging: comedi: fix build breakages on some platforms
Staging: brcm80211: disable drivers except for X86 or MIPS platforms
Staging: brcm80211: disable drivers for PPC platforms
Staging: iio: Make IIO depend on GENERIC_HARDIRQS
Staging: mei: fix suspend failure
Staging: fix iio builds when IIO_RING_BUFFER is not enabled
Staging: Comedi: Build only on arches providing PAGE_KERNEL_NOCACHE
Staging: fix more iio builds when IIO_RING_BUFFER is not enabled
Chris Wilson [Tue, 28 Jun 2011 10:48:51 +0000 (11:48 +0100)]
drm/i915: Use chipset-specific irq installers
Konstantin Belousov pointed out that 4697995b98417 replaced the generic
i915_driver_irq_*install() functions with chipset specific routines
accessible only through driver->irq_*install(). So update the sanity
check in i915_request_wait() to match.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Keith Packard <keithp@keithp.com>
Ben Widawsky [Fri, 24 Jun 2011 21:31:47 +0000 (14:31 -0700)]
drm/i915: forcewake fix after reset
The failure is as follows:
1. Userspace gets forcewake lock, lock count >=1
2. GPU hang/reset occurs (forcewake bit is reset)
3. count is now incorrect
The failure can only occur when using the forcewake userspace lock.
This has the unfortunate consequence of messing up the driver as well as
userspace, unless userspace closes the debugfs file, the kernel will
never end up waking the GT since the refcount will be > 1.
The solution is to try to recover the correct forcewake state based on
the refcount. There is a period of time where userspace reads/writes may
occur after the reset, before the GT has been forcewaked. The interface
was never designed to be a perfect solution for userspace reads/writes,
and the kernel portion is fixed by this patch.
Suggested-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Ben Widawsky <ben@bwidawsk.net> Signed-off-by: Keith Packard <keithp@keithp.com>
Vasiliy Kulikov [Fri, 24 Jun 2011 12:08:38 +0000 (16:08 +0400)]
proc: restrict access to /proc/PID/io
/proc/PID/io may be used for gathering private information. E.g. for
openssh and vsftpd daemons wchars/rchars may be used to learn the
precise password length. Restrict it to processes being able to ptrace
the target process.
ptrace_may_access() is needed to prevent keeping open file descriptor of
"io" file, executing setuid binary and gathering io information of the
setuid'ed process.
NeilBrown [Tue, 28 Jun 2011 06:59:42 +0000 (16:59 +1000)]
md: avoid endless recovery loop when waiting for fail device to complete.
If a device fails in a way that causes pending request to take a while
to complete, md will not be able to immediately remove it from the
array in remove_and_add_spares.
It will then incorrectly look like a spare device and md will try to
recover it even though it is failed.
This leads to a recovery process starting and instantly aborting over
and over again.
We should check if the device is faulty before considering it to be a
spare. This will avoid trying to start a recovery that cannot
proceed.
This bug was introduced in 2.6.26 so that patch is suitable for any
kernel since then.
Cc: stable@kernel.org Reported-by: Jim Paradis <james.paradis@stratus.com> Signed-off-by: NeilBrown <neilb@suse.de>
Hugh Dickins [Mon, 27 Jun 2011 23:18:20 +0000 (16:18 -0700)]
drm/i915: more struct_mutex locking
When auditing the locking in i915_gem.c (for a prospective change which
I then abandoned), I noticed two places where struct_mutex is not held
across GEM object manipulations that would usually require it.
Since one is in initial setup and the other in driver unload, I'm
guessing the mutex is not required for either; but post a patch in case
it is.
Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Keith Packard <keithp@keithp.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Hugh Dickins [Mon, 27 Jun 2011 23:18:19 +0000 (16:18 -0700)]
drm/i915: use shmem_truncate_range
The interface to ->truncate_range is changing very slightly: once "tmpfs:
take control of its truncate_range" has been applied, this can be applied.
For now there is only a slight inefficiency while this remains unapplied,
but it will soon become essential for managing shmem's use of swap.
Change i915_gem_object_truncate() to use shmem_truncate_range() directly:
which should also spare i915 later change if we switch from
inode_operations->truncate_range to file_operations->fallocate.
Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Keith Packard <keithp@keithp.com> Cc: Dave Airlie <airlied@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Hugh Dickins [Mon, 27 Jun 2011 23:18:18 +0000 (16:18 -0700)]
drm/i915: use shmem_read_mapping_page
Soon tmpfs will stop supporting ->readpage and read_cache_page_gfp(): once
"tmpfs: add shmem_read_mapping_page_gfp" has been applied, this patch can
be applied to ease the transition.
Make i915_gem_object_get_pages_gtt() use shmem_read_mapping_page_gfp() in
the one place it's needed; elsewhere use shmem_read_mapping_page(), with
the mapping's gfp_mask properly initialized.
Forget about __GFP_COLD: since tmpfs initializes its pages with memset,
asking for a cold page is counter-productive.
Include linux/shmem_fs.h also in drm_gem.c: with shmem_file_setup() now
declared there too, we shall remove the prototype from linux/mm.h later.
Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Keith Packard <keithp@keithp.com> Cc: Dave Airlie <airlied@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Hugh Dickins [Mon, 27 Jun 2011 23:18:17 +0000 (16:18 -0700)]
drm/ttm: use shmem_read_mapping_page
Soon tmpfs will stop supporting ->readpage and read_mapping_page(): once
"tmpfs: add shmem_read_mapping_page_gfp" has been applied, this patch can
be applied to ease the transition.
ttm_tt_swapin() and ttm_tt_swapout() use shmem_read_mapping_page() in
place of read_mapping_page(), since their swap_space has been created with
shmem_file_setup().
Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: Thomas Hellstrom <thellstrom@vmware.com> Cc: Dave Airlie <airlied@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
WARNING: drivers/misc/ioc4.o(.data+0x144): Section mismatch in reference from the variable ioc4_load_modules_work to the function .devinit.text:ioc4_load_modules()
The variable ioc4_load_modules_work references
the function __devinit ioc4_load_modules()
If the reference is valid then annotate the
variable with __init* or __refdata (see linux/init.h) or name the variable:
*driver, *_template, *_timer, *_sht, *_ops, *_probe, *_probe_one, *_console
This one is potentially fatal; by the time ioc4_load_modules is invoked
it may already have been freed. For that reason ioc4_load_modules_work
can't be turned to __devinitdata but also because it's referenced in
ioc4_exit.
WARNING: drivers/leds/leds-lp5523.o(.text+0x12f4): Section mismatch in reference from the function lp5523_probe() to the function .init.text:lp5523_init_led()
The function lp5523_probe() references
the function __init lp5523_init_led().
This is often because lp5523_probe lacks a __init
annotation or the annotation of lp5523_init_led is wrong.
Fixing this one triggers one more mismatch, fix that one as well.
Signed-off-by: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Purdie <rpurdie@rpsys.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
WARNING: drivers/leds/leds-lp5521.o(.text+0xf2c): Section mismatch in reference from the function lp5521_probe() to the function .init.text:lp5521_init_led()
The function lp5521_probe() references
the function __init lp5521_init_led().
This is often because lp5521_probe lacks a __init
annotation or the annotation of lp5521_init_led is wrong.
Fixing this mismatch triggers one more mismatch, fix that one as well.
Signed-off-by: Ralf Baechle <ralf@linux-mips.org> Cc: Richard Purdie <rpurdie@rpsys.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
memcg: fix direct softlimit reclaim to be called in limit path
Commit d149e3b25d7c ("memcg: add the soft_limit reclaim in global direct
reclaim") adds a softlimit hook to shrink_zones(). By this, soft limit
is called as
This will cause a global reclaim when a memcg hits limit.
This is bug. soft_limit_reclaim() should be called when
scanning_global_lru(sc) == true.
And the commit adds a variable "total_scanned" for counting softlimit
scanned pages....it's not "total". This patch removes the variable and
update sc->nr_scanned instead of it. This will affect shrink_slab()'s
scan condition but, global LRU is scanned by softlimit and I think this
change makes sense.
TODO: avoid too much scanning of a zone when softlimit did enough work.
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp> Cc: Ying Han <yinghan@google.com> Cc: Michal Hocko <mhocko@suse.cz> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Vasiliy Kulikov [Mon, 27 Jun 2011 23:18:11 +0000 (16:18 -0700)]
taskstats: don't allow duplicate entries in listener mode
Currently a single process may register exit handlers unlimited times.
It may lead to a bloated listeners chain and very slow process
terminations.
Eg after 10KK sent TASKSTATS_CMD_ATTR_REGISTER_CPUMASKs ~300 Mb of
kernel memory is stolen for the handlers chain and "time id" shows 2-7
seconds instead of normal 0.003. It makes it possible to exhaust all
kernel memory and to eat much of CPU time by triggerring numerous exits
on a single CPU.
The patch limits the number of times a single process may register
itself on a single CPU to one.
One little issue is kept unfixed - as taskstats_exit() is called before
exit_files() in do_exit(), the orphaned listener entry (if it was not
explicitly deregistered) is kept until the next someone's exit() and
implicit deregistration in send_cpu_listeners(). So, if a process
registered itself as a listener exits and the next spawned process gets
the same pid, it would inherit taskstats attributes.
Jan Kara [Mon, 27 Jun 2011 23:18:10 +0000 (16:18 -0700)]
mm: fix assertion mapping->nrpages == 0 in end_writeback()
Under heavy memory and filesystem load, users observe the assertion
mapping->nrpages == 0 in end_writeback() trigger. This can be caused by
page reclaim reclaiming the last page from a mapping in the following
race:
Josh Hunt [Mon, 27 Jun 2011 23:18:08 +0000 (16:18 -0700)]
drivers/misc/lkdtm.c: fix race when crashpoint is hit multiple times before checking count
We observed the crash point count going negative in cases where the
crash point is hit multiple times before the check of "count == 0" is
done. Because of this we never call lkdtm_do_action(). This patch just
adds a spinlock to protect count.
romfs_get_unmapped_area() checks argument `len' without considering
PAGE_ALIGN which will cause do_mmap_pgoff() return -EINVAL error after
commit f67d9b1576c ("nommu: add page_align to mmap").
Fix the check by changing it in same way ramfs_nommu_get_unmapped_area()
was changed in ramfs/file-nommu.c.
Signed-off-by: Bob Liu <lliubbo@gmail.com> Cc: David Howells <dhowells@redhat.com> Cc: Paul Mundt <lethal@linux-sh.org> Acked-by: Greg Ungerer <gerg@snapgear.com> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
It has 56 bytes of nonvolatile RAM. Its register set is same as that of
rtc device: DS1307.
Signed-off-by: Priyanka Jain <Priyanka.Jain@freescale.com> Acked-by: Timur Tabi <timur@freescale.com> Reviewed-by: Wolfram Sang <w.sang@pengutronix.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Hugh Dickins [Mon, 27 Jun 2011 23:18:04 +0000 (16:18 -0700)]
tmpfs: add shmem_read_mapping_page_gfp
Although it is used (by i915) on nothing but tmpfs, read_cache_page_gfp()
is unsuited to tmpfs, because it inserts a page into pagecache before
calling the filesystem's ->readpage: tmpfs may have pages in swapcache
which only it knows how to locate and switch to filecache.
At present tmpfs provides a ->readpage method, and copes with this by
copying pages; but soon we can simplify it by removing its ->readpage.
Provide shmem_read_mapping_page_gfp() now, ready for that transition,
Export shmem_read_mapping_page_gfp() and add it to list in shmem_fs.h,
with shmem_read_mapping_page() inline for the common mapping_gfp case.
(shmem_read_mapping_page_gfp or shmem_read_cache_page_gfp? Generally the
read_mapping_page functions use the mapping's ->readpage, and the
read_cache_page functions use the supplied filler, so I think
read_cache_page_gfp was slightly misnamed.)
Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Christoph Hellwig <hch@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Hugh Dickins [Mon, 27 Jun 2011 23:18:03 +0000 (16:18 -0700)]
tmpfs: take control of its truncate_range
2.6.35's new truncate convention gave tmpfs the opportunity to control
its file truncation, no longer enforced from outside by vmtruncate().
We shall want to build upon that, to handle pagecache and swap together.
Slightly redefine the ->truncate_range interface: let it now be called
between the unmap_mapping_range()s, with the filesystem responsible for
doing the truncate_inode_pages_range() from it - just as the filesystem
is nowadays responsible for doing that from its ->setattr.
Let's rename shmem_notify_change() to shmem_setattr(). Instead of
calling the generic truncate_setsize(), bring that code in so we can
call shmem_truncate_range() - which will later be updated to perform its
own variant of truncate_inode_pages_range().
Remove the punch_hole unmap_mapping_range() from shmem_truncate_range():
now that the COW's unmap_mapping_range() comes after ->truncate_range,
there is no need to call it a third time.
Export shmem_truncate_range() and add it to the list in shmem_fs.h, so
that i915_gem_object_truncate() can call it explicitly in future; get
this patch in first, then update drm/i915 once this is available (until
then, i915 will just be doing the truncate_inode_pages() twice).
Though introduced five years ago, no other filesystem is implementing
->truncate_range, and its only other user is madvise(,,MADV_REMOVE): we
expect to convert it to fallocate(,FALLOC_FL_PUNCH_HOLE,,) shortly,
whereupon ->truncate_range can be removed from inode_operations -
shmem_truncate_range() will help i915 across that transition too.
Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Christoph Hellwig <hch@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Hugh Dickins [Mon, 27 Jun 2011 23:18:02 +0000 (16:18 -0700)]
mm: move shmem prototypes to shmem_fs.h
Before adding any more global entry points into shmem.c, gather such
prototypes into shmem_fs.h. Remove mm's own declarations from swap.h,
but for now leave the ones in mm.h: because shmem_file_setup() and
shmem_zero_setup() are called from various places, and we should not
force other subsystems to update immediately.
Signed-off-by: Hugh Dickins <hughd@google.com> Cc: Christoph Hellwig <hch@infradead.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
projects / mv-sheeva.git / log