Eric Dumazet [Wed, 14 Dec 2005 07:13:48 +0000 (23:13 -0800)]
[NETFILTER] ip_tables: NUMA-aware allocation
Part of a performance problem with ip_tables is that memory allocation
is not NUMA aware, but 'only' SMP aware (ie each CPU normally touch
separate cache lines)
Even with small iptables rules, the cost of this misplacement can be
high on common workloads. Instead of using one vmalloc() area
(located in the node of the iptables process), we now allocate an area
for each possible CPU, using vmalloc_node() so that memory should be
allocated in the CPU's node if possible.
Port to arp_tables and ip6_tables by Harald Welte.
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Replace existing BIC version 1.1 with new version 2.0.
The main change is to replace the window growth function
with a cubic function as described in:
http://www.csc.ncsu.edu/faculty/rhee/export/bitcp/cubic-paper.pdf
Signed-off-by: Stephen Hemminger <shemminger@osdl.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Trent Jaeger [Wed, 14 Dec 2005 07:12:40 +0000 (23:12 -0800)]
[LSM-IPSec]: Per-packet access control.
This patch series implements per packet access control via the
extension of the Linux Security Modules (LSM) interface by hooks in
the XFRM and pfkey subsystems that leverage IPSec security
associations to label packets. Extensions to the SELinux LSM are
included that leverage the patch for this purpose.
This patch implements the changes necessary to the SELinux LSM to
create, deallocate, and use security contexts for policies
(xfrm_policy) and security associations (xfrm_state) that enable
control of a socket's ability to send and receive packets.
Patch purpose:
The patch is designed to enable the SELinux LSM to implement access
control on individual packets based on the strongly authenticated
IPSec security association. Such access controls augment the existing
ones in SELinux based on network interface and IP address. The former
are very coarse-grained, and the latter can be spoofed. By using
IPSec, the SELinux can control access to remote hosts based on
cryptographic keys generated using the IPSec mechanism. This enables
access control on a per-machine basis or per-application if the remote
machine is running the same mechanism and trusted to enforce the
access control policy.
Patch design approach:
The patch's main function is to authorize a socket's access to a IPSec
policy based on their security contexts. Since the communication is
implemented by a security association, the patch ensures that the
security association's negotiated and used have the same security
context. The patch enables allocation and deallocation of such
security contexts for policies and security associations. It also
enables copying of the security context when policies are cloned.
Lastly, the patch ensures that packets that are sent without using a
IPSec security assocation with a security context are allowed to be
sent in that manner.
A presentation available at
www.selinux-symposium.org/2005/presentations/session2/2-3-jaeger.pdf
from the SELinux symposium describes the overall approach.
Patch implementation details:
The function which authorizes a socket to perform a requested
operation (send/receive) on a IPSec policy (xfrm_policy) is
selinux_xfrm_policy_lookup. The Netfilter and rcv_skb hooks ensure
that if a IPSec SA with a securit y association has not been used,
then the socket is allowed to send or receive the packet,
respectively.
The patch implements SELinux function for allocating security contexts
when policies (xfrm_policy) are created via the pfkey or xfrm_user
interfaces via selinux_xfrm_policy_alloc. When a security association
is built, SELinux allocates the security context designated by the
XFRM subsystem which is based on that of the authorized policy via
selinux_xfrm_state_alloc.
When a xfrm_policy is cloned, the security context of that policy, if
any, is copied to the clone via selinux_xfrm_policy_clone.
When a xfrm_policy or xfrm_state is freed, its security context, if
any is also freed at selinux_xfrm_policy_free or
selinux_xfrm_state_free.
Testing:
The SELinux authorization function is tested using ipsec-tools. We
created policies and security associations with particular security
contexts and added SELinux access control policy entries to verify the
authorization decision. We also made sure that packets for which no
security context was supplied (which either did or did not use
security associations) were authorized using an unlabelled context.
Signed-off-by: Trent Jaeger <tjaeger@cse.psu.edu> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Trent Jaeger [Wed, 14 Dec 2005 07:12:27 +0000 (23:12 -0800)]
[LSM-IPSec]: Security association restriction.
This patch series implements per packet access control via the
extension of the Linux Security Modules (LSM) interface by hooks in
the XFRM and pfkey subsystems that leverage IPSec security
associations to label packets. Extensions to the SELinux LSM are
included that leverage the patch for this purpose.
This patch implements the changes necessary to the XFRM subsystem,
pfkey interface, ipv4/ipv6, and xfrm_user interface to restrict a
socket to use only authorized security associations (or no security
association) to send/receive network packets.
Patch purpose:
The patch is designed to enable access control per packets based on
the strongly authenticated IPSec security association. Such access
controls augment the existing ones based on network interface and IP
address. The former are very coarse-grained, and the latter can be
spoofed. By using IPSec, the system can control access to remote
hosts based on cryptographic keys generated using the IPSec mechanism.
This enables access control on a per-machine basis or per-application
if the remote machine is running the same mechanism and trusted to
enforce the access control policy.
Patch design approach:
The overall approach is that policy (xfrm_policy) entries set by
user-level programs (e.g., setkey for ipsec-tools) are extended with a
security context that is used at policy selection time in the XFRM
subsystem to restrict the sockets that can send/receive packets via
security associations (xfrm_states) that are built from those
policies.
A presentation available at
www.selinux-symposium.org/2005/presentations/session2/2-3-jaeger.pdf
from the SELinux symposium describes the overall approach.
Patch implementation details:
On output, the policy retrieved (via xfrm_policy_lookup or
xfrm_sk_policy_lookup) must be authorized for the security context of
the socket and the same security context is required for resultant
security association (retrieved or negotiated via racoon in
ipsec-tools). This is enforced in xfrm_state_find.
On input, the policy retrieved must also be authorized for the socket
(at __xfrm_policy_check), and the security context of the policy must
also match the security association being used.
The patch has virtually no impact on packets that do not use IPSec.
The existing Netfilter (outgoing) and LSM rcv_skb hooks are used as
before.
Also, if IPSec is used without security contexts, the impact is
minimal. The LSM must allow such policies to be selected for the
combination of socket and remote machine, but subsequent IPSec
processing proceeds as in the original case.
Testing:
The pfkey interface is tested using the ipsec-tools. ipsec-tools have
been modified (a separate ipsec-tools patch is available for version
0.5) that supports assignment of xfrm_policy entries and security
associations with security contexts via setkey and the negotiation
using the security contexts via racoon.
The xfrm_user interface is tested via ad hoc programs that set
security contexts. These programs are also available from me, and
contain programs for setting, getting, and deleting policy for testing
this interface. Testing of sa functions was done by tracing kernel
behavior.
Signed-off-by: Trent Jaeger <tjaeger@cse.psu.edu> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
However, asm/param.h is sometimes used in userland (it is included
indirectly from <sys/param.h>), so your commit pollutes the userland
namespace with tons of CONFIG_FOO macros. This greatly confuses
software packages (such as BusyBox) which use CONFIG_FOO macros
themselves to control the inclusion of optional features.
After a short exchange, Christoph approved this patch
Some G5s still occasionally experience shutdowns due to overtemp
conditions despite the recent fix. After analyzing logs from such
machines, it appears that the overtemp code is a bit too quick at
shutting the machine down when reaching the critical temperature (tmax +
8) and doesn't leave the fan enough time to actually cool it down. This
happens if the temperature of a CPU suddenly rises too high in a very
short period of time, or occasionally on boot (that is the CPUs are
already overtemp by the time the driver loads).
This patches makes the code a bit more relaxed, leaving a few seconds to
the fans to do their job before kicking the machine shutown.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Linus Torvalds [Sun, 1 Jan 2006 01:00:29 +0000 (17:00 -0800)]
sysctl: make sure to terminate strings with a NUL
This is a slightly more complete fix for the previous minimal sysctl
string fix. It always terminates the returned string with a NUL, even
if the full result wouldn't fit in the user-supplied buffer.
The returned length is the full untruncated length, so that you can
tell when truncation has occurred.
Yi Yang [Fri, 30 Dec 2005 08:37:10 +0000 (16:37 +0800)]
[PATCH] Fix false old value return of sysctl
For the sysctl syscall, if the user wants to get the old value of a
sysctl entry and set a new value for it in the same syscall, the old
value is always overwritten by the new value if the sysctl entry is of
string type and if the user sets its strategy to sysctl_string. This
issue lies in the strategy being run twice if the strategy is set to
sysctl_string, the general strategy sysctl_string always returns 0 if
success.
Such strategy routines as sysctl_jiffies and sysctl_jiffies_ms return 1
because they do read and write for the sysctl entry.
The strategy routine sysctl_string return 0 although it actually read
and write the sysctl entry.
According to my analysis, if a strategy routine do read and write, it
should return 1, if it just does some necessary check but not read and
write, it should return 0, for example sysctl_intvec.
Signed-off-by: Yi Yang <yang.y.yi@gmail.com> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Linus Torvalds [Sat, 31 Dec 2005 01:18:53 +0000 (17:18 -0800)]
sysctl: don't overflow the user-supplied buffer with '\0'
If the string was too long to fit in the user-supplied buffer,
the sysctl layer would zero-terminate it by writing past the
end of the buffer. Don't do that.
Linus Torvalds [Fri, 30 Dec 2005 16:39:10 +0000 (08:39 -0800)]
Insanity avoidance in /proc
The old /proc interfaces were never updated to use loff_t, and are just
generally broken. Now, we should be using the seq_file interface for
all of the proc files, but converting the legacy functions is more work
than most people care for and has little upside..
But at least we can make the non-LFS rules explicit, rather than just
insanely wrapping the offset or something.
Erik Hovland [Fri, 30 Dec 2005 15:57:35 +0000 (15:57 +0000)]
[ARM] 3216/1: indent and typo in drivers/serial/pxa.c
Patch from Erik Hovland
This patch provides two changes. An indent is supplied for an if/else clause so that it is more readable. An acronym is incorrectly typed as UER when it should be IER.
Signed-off-by: Erik Hovland <erik@hovland.org> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Jean Delvare [Thu, 29 Dec 2005 21:07:30 +0000 (22:07 +0100)]
[PATCH] Simplify the VIDEO_SAA7134_OSS Kconfig dependency line
Thanks to Roman Zippel for the suggestion.
Signed-off-by: Jean Delvare <khali@linux-fr.org>
[ Short explanation: Kconfig uses ternary math: n/m/y, and !m is m ] Signed-off-by: Linus Torvalds <torvalds@osdl.org>
that changed the GART VM start offset. It fixed some machines, but
seems to continually interact badly with some X versions.
Quoth Ben Herrenschmidt:
"So I think at this point, the best is that we keep the old bogus code
that at least is consistent with the bug in the server. I'm working on a
big patch to X that reworks the memory map stuff completely and fixes
those issues on the server side, I'll do a DRM patch matching this X fix
as well so that the memory map is only ever set in one place and with
what I hope is a correct algorithm..."
Anton Blanchard [Wed, 28 Dec 2005 23:46:29 +0000 (10:46 +1100)]
[PATCH] ppc64: htab_initialize_secondary cannot be marked __init
Sonny has noticed hotplug CPU on ppc64 is broken in 2.6.15-*. One of the
problems is that htab_initialize_secondary is called when a cpu is being
brought up, but it is marked __init.
Signed-off-by: Anton Blanchard <anton@samba.org> Acked-by: Paul Mackerras <paulus@samba.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Dave Jones [Thu, 29 Dec 2005 01:01:04 +0000 (20:01 -0500)]
[PATCH] fix ia64 compile failure with gcc4.1
__get_unaligned creates a typeof the var its passed, and writes to it,
which on gcc4.1, spits out the following error:
drivers/char/vc_screen.c: In function 'vcs_write':
drivers/char/vc_screen.c:422: error: assignment of read-only variable 'val'
Signed-off-by: Dave Jones <davej@redhat.com>
[ The "right" fix would be to try to fix <asm-generic/unaligned.h>
but that's hard to do with the tools gcc gives us. So this
simpler patch is preferable -- Linus ] Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] uml: fix compilation with CONFIG_MODE_TT disabled
Fix UML compilation when SKAS mode is disabled. Indeed, we were compiling
SKAS-only object files, which failed due to some SKAS-only headers being
excluded from the search path.
Thanks to the bug report from Pekka J Enberg.
Acked-by: Pekka J Enberg <penberg (at) cs ! helsinki ! fi> Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] Hostfs: update for new glibc - add missing symbol exports
Today, when compiling UML, I got warnings for two used unexported symbols:
readdir64 and truncate64. Indeed, my glibc headers are aliasing readdir to
readdir64 and truncate to truncate64 (and so on).
I'm then adding additional exports. Since I've no idea if the symbols where
always provided in the supported glibc's, I've added weak definitions too.
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[PATCH] uml: hostfs - fix possible PAGE_CACHE_SHIFT overflows
Prevent page->index << PAGE_CACHE_SHIFT from overflowing.
There is a casting there, but was added without care, so it's at the wrong
place. Note the extra parens around the shift - "+" is higher precedence than
"<<", leading to a GCC warning which saved all us.
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Trivial removal of unused variable from this file - doesn't even change the
generated assembly code, in fact (gcc should trigger a warning for unused value
here).
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Don't use printk() where "current_thread_info()" is crap.
Until when we switch to running on init_stack, current_thread_info() evaluates
to crap. Printk uses "current" at times (in detail, ¤t is evaluated with
CONFIG_DEBUG_SPINLOCK to check the spinlock owner task).
And this leads to random segmentation faults.
Exactly, what happens is that ¤t = *(current_thread_info()), i.e. round
down $esp and dereference the value. I.e. access the stack below $esp, which
causes SIGSEGV on a VM_GROWSDOWN vma (see arch/i386/mm/fault.c).
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
David S. Miller [Wed, 28 Dec 2005 21:27:04 +0000 (13:27 -0800)]
[SERMOUSE]: Sun mice speak 5-byte protocol too.
Noticed by Christophe Zimmerman, this explains the slow mouse movement
with 2.6.x kernels.
And checking the 2.4.x drivers/sbus/char/sunmouse.c driver shows we
always used a 5-byte protocol with Sun mice in the past. I have no
idea how the 3-byte thing got into the 2.6.x driver, but it's surely
wrong.
Signed-off-by: David S. Miller <davem@davemloft.net>
Chris Elmquist [Tue, 20 Dec 2005 21:25:19 +0000 (13:25 -0800)]
[TG3]: ethtool -d hangs PCIe systems
Resubmitting after recommendation to use GET_REG32_1() instead of
GET_REG32_LOOP(..., 1). Retested. Problem remains fixed.
Prevent tg3_get_regs() from reading reserved and undocumented registers
at RX_CPU_BASE and TX_CPU_BASE offsets which caused hostile behavior
on PCIe platforms.
Acked-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
[PATCH] Fix more radeon GART start calculation cases
As reported by Jules Villard <jvillard@ens-lyon.fr> and some others, the
recent GART aperture start reconfiguration causes problems on some
setups.
What I _think_ might be happening is that the X server is also trying to
muck around with the card memory map and is forcing it back into a wrong
setting that also happens to no longer match what the DRM wants to do
and blows up. There are bugs all over the place in that code (and still
some bugs in the DRM as well anyway).
This patch attempts to avoid that by using the largest of the 2 values,
which I think will cause it to behave as it used to for you and will
still fix the problem with machines that have an aperture size smaller
than the video memory.
David L Stevens [Tue, 27 Dec 2005 22:03:00 +0000 (14:03 -0800)]
[IPV6] mcast: Fix multiple issues in MLDv2 reports.
The below "jumbo" patch fixes the following problems in MLDv2.
1) Add necessary "ntohs" to recent "pskb_may_pull" check [breaks
all nonzero source queries on little-endian (!)]
2) Add locking to source filter list [resend of prior patch]
3) fix "mld_marksources()" to
a) send nothing when all queried sources are excluded
b) send full exclude report when source queried sources are
not excluded
c) don't schedule a timer when there's nothing to report
NOTE: RFC 3810 specifies the source list should be saved and each
source reported individually as an IS_IN. This is an obvious DOS
path, requiring the host to store and then multicast as many sources
as are queried (e.g., millions...). This alternative sends a full,
relevant report that's limited to number of sources present on the
machine.
4) fix "add_grec()" to send empty-source records when it should
The original check doesn't account for a non-empty source
list with all sources inactive; the new code keeps that
short-circuit case, and also generates the group header
with an empty list if needed.
5) fix mca_crcount decrement to be after add_grec(), which needs
its original value
These issues (other than item #1 ;-) ) were all found by Yan Zheng,
much thanks!
Signed-off-by: David L Stevens <dlstevens@us.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
David Kimdon [Tue, 27 Dec 2005 01:27:10 +0000 (17:27 -0800)]
[BR_NETFILTER]: Fix leak if skb traverses > 1 bridge
Call nf_bridge_put() before allocating a new nf_bridge structure and
potentially overwriting the pointer to a previously allocated one.
This fixes a memory leak which can occur when the bridge topology
allows for an skb to traverse more than one bridge.
Signed-off-by: David Kimdon <david.kimdon@devicescape.com> Signed-off-by: David S. Miller <davem@davemloft.net>
James Bottomley [Mon, 26 Dec 2005 15:58:49 +0000 (09:58 -0600)]
[PATCH] Fix Fibre Channel boot oops
The oops is characteristic of the underlying device being removed from
visibility before the class device, and sure enough we do device_del()
before transport_unregister() in the scsi_target_reap() routines. I've
no idea why this is suddenly showing up, since the code has been in
there since that function was first invented. However, I've confirmed
this fixes Andrew Vasquez's boot oops.
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Kurt Huwig [Sat, 24 Dec 2005 23:13:08 +0000 (00:13 +0100)]
[PATCH] n_r3964: fixed usage of HZ; removed bad include
Fix n_r3964 timeouts (hardcoded for 100Hz)
Also the include of <asm/termios.h> in 'n_r3964.h' is unnecessary and
prevents using the header file in any application that has to include
<termios.h> due to duplicate definition of 'struct termio'.
[PATCH] x86_64/ia64 : Fix compilation error for node_to_first_cpu
Fixes a compiler error in node_to_first_cpu, __ffs expects unsigned long as
a parameter; instead cpumask_t was being passed. The macro
node_to_first_cpu was not yet used in x86_64 and ia64 arches, and so we never
hit this. This patch replaces __ffs with first_cpu macro, similar to other
arches.
Signed-off-by: Alok N Kataria <alokk@calsoftinc.com> Signed-off-by: Ravikiran G Thirumalai <kiran@scalex86.org> Signed-off-by: Shai Fultheim <shai@scalex86.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Andrew Morton [Sat, 24 Dec 2005 03:54:46 +0000 (19:54 -0800)]
[PATCH] Fix memory ordering problem in wake_futex()
Fix a memory ordering problem that occurs on IA64. The "store" to q->lock_ptr
in wake_futex() can become visible before wake_up_all() clears the lock in the
futex_q.
Manfred Spraul [Sat, 24 Dec 2005 13:19:24 +0000 (14:19 +0100)]
[PATCH] forcedeth: fix random memory scribbling bug
Two critical bugs were found in forcedeth 0.47:
- TSO doesn't work.
- pci_map_single() for the rx buffers is called with size==0. This bug
is critical, it causes random memory corruptions on systems with an
iommu.
Below is a minimal fix for both bugs, for 2.6.15.
TSO will be fixed properly in the next version. Tested on x86-64.
Frank Pavlic [Tue, 13 Dec 2005 07:22:30 +0000 (08:22 +0100)]
[PATCH] s390: minor qeth network driver fixes
[patch 2/3] s390: minor qeth network driver fixes
From: Frank Pavlic <pavlic@de.ibm.com>
- use netif_carrier_on/off calls to tell network stack
link carrier state
- fix possible kfree on NULL
- PDU_LEN2 is at offset 0x29 otherwise OSN chpid won't initialize
Frank Pavlic [Tue, 13 Dec 2005 07:21:47 +0000 (08:21 +0100)]
[PATCH] s390: some minor qeth driver fixes
[patch 1/3] s390: some minor qeth driver fixes
From: Frank Pavlic <pavlic@de.ibm.com>
- let's have just one function for both ,input and output queue
to check qdio errors
- add /proc/s390dbf/qeth_qerr entries for outbound processing
- check removed for layer2 device in qeth_add_multicast_ipv6
- NULL pointer dereference with bonding and VLAN device fixed
- minimum length check for portname fixed
Tony Battersby [Wed, 21 Dec 2005 21:35:44 +0000 (16:35 -0500)]
[PATCH] fix libata inquiry VPD for ATAPI devices
The following patch prevents libata from incorrectly modifying inquiry
VPD pages and command support data from ATAPI devices. I have tested
the patch with a SATA ATAPI tape drive on an AHCI controller.
Patch is against kernel 2.4.32 with 2.4.32-libata1.patch applied.
David S. Miller [Fri, 23 Dec 2005 07:04:39 +0000 (23:04 -0800)]
[SPARC]: Kill CHILD_MAX.
It's definition is wrong (-1 means "no limit" not 999),
only the Sparc SunOS/Solaris compat code uses it, so
let's just kill it off completely from limits.h and
all referencing code.
Noticed by Ulrich Drepper.
Signed-off-by: David S. Miller <davem@davemloft.net>
Thomas Renninger [Wed, 21 Dec 2005 06:29:00 +0000 (01:29 -0500)]
[ACPI] fix passive cooling regression
Return logic was inverted.
Going for changing the return value to not return zero as it is makes
more sense regarding the naming of the function (cpu_has_cpufreq()).
http://bugzilla.kernel.org/show_bug.cgi?id=3410
Signed-off-by: Thomas Renninger <trenn@suse.de> Signed-off-by: Len Brown <len.brown@intel.com>
Russell King [Thu, 22 Dec 2005 23:21:38 +0000 (23:21 +0000)]
[MMC] Set correct capacity for 1024-byte block cards
We were passing set_capacity() the capacity we calculated in terms of
the number of blocks on the card, which happened to be the right units
for 512-byte block cards. However, with 1024-byte block cards, we
end up setting the capacity to half the number of blocks. Fix this
by shifting by the appropriate amount.
Thanks to Todd Blumer for pointing this out.
Use get_capacity() to report the card capacity, rather than
recalculating it from the CSD information.
Finally, use our chosen IO block size for the SET_BLOCKLEN command
rather than the CSD read block size. Currently these are equivalent,
but will not be in the future.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The problem is that the TIF_NEED_RESCHED flag is loaded _before_ the
preemption count is stored back, hence any interrupt coming within that
3 instruction window causing TIF_NEED_RESCHED to be set won't be
seen and scheduling won't happen as it should.
Nothing currently prevents gcc from performing that reordering. There
is already a barrier() before the decrement of the preemption count, but
another one is needed between this and the TIF_NEED_RESCHED flag test
for proper code ordering.
Signed-off-by: Nicolas Pitre <nico@cam.org> Acked-by: Nick Piggin <nickpiggin@yahoo.com.au> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
David S. Miller [Thu, 22 Dec 2005 15:39:48 +0000 (07:39 -0800)]
[IPSEC]: Fix policy updates missed by sockets
The problem is that when new policies are inserted, sockets do not see
the update (but all new route lookups do).
This bug is related to the SA insertion stale route issue solved
recently, and this policy visibility problem can be fixed in a similar
way.
The fix is to flush out the bundles of all policies deeper than the
policy being inserted. Consider beginning state of "outgoing"
direction policy list:
policy A --> policy B --> policy C --> policy D
First, realize that inserting a policy into a list only potentially
changes IPSEC routes for that direction. Therefore we need not bother
considering the policies for other directions. We need only consider
the existing policies in the list we are doing the inserting.
Consider new policy "B'", inserted after B.
policy A --> policy B --> policy B' --> policy C --> policy D
Two rules:
1) If policy A or policy B matched before the insertion, they
appear before B' and thus would still match after inserting
B'
2) Policy C and D, now "shadowed" and after policy B', potentially
contain stale routes because policy B' might be selected
instead of them.
Therefore we only need flush routes assosciated with policies
appearing after a newly inserted policy, if any.
Signed-off-by: David S. Miller <davem@davemloft.net>
Paul Mackerras [Thu, 22 Dec 2005 10:55:37 +0000 (21:55 +1100)]
powerpc: Fix i8259 cascade on pSeries with XICS interrupt controller
It turns out that commit f9bd170a87948a9e077149b70fb192c563770fdf
broke the cascade from XICS to i8259 on pSeries machines; specifically
we ended up not ever doing the EOI on the XICS for the cascade. The
result was that interrupts from the serial ports (and presumably any
other devices using ISA interrupts) didn't get through. This fixes
it and also simplifies the code, by doing the EOI on the XICS in the
xics_get_irq routine after reading and acking the interrupt on the
i8259.
- When ALSA or OSS are loaded, check if the other is present
Fixed hotplug notifiers cleanup on module removal
- The saa7134 DMA sound modules now have their own Kconfig entries, and
if built statically enforce exclusivity
- SND_PCM_OSS isn't necessary for the OSS driver
Kristian Slavov [Thu, 22 Dec 2005 02:47:24 +0000 (18:47 -0800)]
[IPV6]: Fix address deletion
If you add more than one IPv6 address belonging to the same prefix and
delete the address that was last added, routing table entry for that
prefix is also deleted.
Tested on 2.6.14.4
To reproduce:
ip addr add 3ffe::1/64 dev eth0
ip addr add 3ffe::2/64 dev eth0
/* wait DAD */
sleep 1
ip addr del 3ffe::2/64 dev eth0
ip -6 route
(route to 3ffe::/64 should be gone)
In ipv6_del_addr(), if ifa == ifp, we set ifa->if_next to NULL, and later
assign ifap = &ifa->if_next, effectively terminating the for-loop.
This prevents us from checking if there are other addresses using the same
prefix that are valid, and thus resulting in deletion of the prefix.
This applies only if the first entry in idev->addr_list is the address to
be deleted.
Signed-off-by: Kristian Slavov <kristian.slavov@nomadiclab.com> Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Linus Torvalds [Wed, 21 Dec 2005 22:52:32 +0000 (14:52 -0800)]
Initialize drivers/media/video/saa7134 late
When compiled-in, make sure the sound system has initialized
before these drivers do.
Reported by Adrian Bunk <bunk@stusta.de>
(The right fix would be to make the sound core use "subsys_initcall()"
and thus initialize before all normal drivers, but this is the quick
and limited safe fix for 2.6.15).
Alan Stern [Wed, 21 Dec 2005 22:28:11 +0000 (14:28 -0800)]
[PATCH] usbcore: allow suspend/resume even if drivers don't support it
This patch (as618) changes usbcore to prevent derailing the
suspend/resume sequence when a USB driver doesn't include support for
it. This is a workaround rather than a true fix; the core needs to be
changed so that URB submissions from suspended drivers can be refused
and outstanding URBs cancelled.
Paul Walmsley [Wed, 21 Dec 2005 22:28:06 +0000 (14:28 -0800)]
[PATCH] USB Storage: Force starget->scsi_level in usb-storage scsiglue.c
When the usb-storage module forces sdev->scsi_level to SCSI_2, it should
also force starget->scsi_level to the same value. Otherwise, the SCSI
layer may attempt to issue SCSI-3 commands to the device, such as REPORT
LUNS, which it cannot handle. This can prevent the device from working
with Linux.
The AMS Venus DS3 DS2316SU2S SATA-to-SATA+USB enclosure, based on the
Oxford Semiconductor OXU921S chip, requires this patch to function
correctly on Linux. The enclosure reports a SCSI-3 SPC-2 command set
level, but does not correctly handle the REPORT LUNS SCSI command -
probably due to a bug in its firmware.
It seems likely that other USB storage enclosures with similar bugs will
also benefit from this patch.
Tony Lindgren <tony@atomide.com> collaborated in the development of this
patch.
Jan's crosscompile page [1] shows, that one regression in 2.6.15-rc is
that the v850 defconfig does no longer compile.
The compile error is:
<-- snip -->
...
CC arch/v850/kernel/setup.o
In file included from /usr/src/ctest/rc/kernel/arch/v850/kernel/setup.c:17:
/usr/src/ctest/rc/kernel/include/linux/irq.h:13:43: asm/smp.h: No such file or directory
make[2]: *** [arch/v850/kernel/setup.o] Error 1
<-- snip -->
The #include <asm/smp.h> in irq.h was intruduced in 2.6.15-rc.
Since include/linux/irq.h needs code from asm/smp.h only in the
CONFIG_SMP=y case and linux/smp.h #include's asm/smp.h only in the
CONFIG_SMP=y case, I'm suggesting this patch to #include <linux/smp.h>
in irq.h.
I've tested the compilation with both CONFIG_SMP=y and CONFIG_SMP=n
on i386.
Signed-off-by: Adrian Bunk <bunk@stusta.de> Acked-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
[IPV6]: Don't select a tentative address as a source address.
A tentative address is not considered "assigned to an interface"
in the traditional sense (RFC2462 Section 4).
Don't try to select such an address for the source address.
projects / karo-tx-linux.git / log