This patch removes all the old code which handled the nested
selective cr0 write intercepts. This code was only in place
as a work-around until the instruction emulator is capable
of doing the same. This is the case with this patch-set and
so the code can be removed.
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com> Signed-off-by: Avi Kivity <avi@redhat.com>
KVM: SVM: Add intercept checks for SVM instructions
This patch adds the necessary code changes in the
instruction emulator and the extensions to svm.c to
implement intercept checks for the svm instructions.
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com> Signed-off-by: Avi Kivity <avi@redhat.com>
This patch adds a check_perm callback for each opcode into
the instruction emulator. This will be used to do all
necessary permission checks on instructions before checking
whether they are intercepted or not.
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com> Signed-off-by: Avi Kivity <avi@redhat.com>
Avi Kivity [Mon, 4 Apr 2011 10:39:22 +0000 (12:39 +0200)]
KVM: x86 emulator: add framework for instruction intercepts
When running in guest mode, certain instructions can be intercepted by
hardware. This also holds for nested guests running on emulated
virtualization hardware, in particular instructions emulated by kvm
itself.
This patch adds a framework for intercepting instructions. If an
instruction is marked for interception, and if we're running in guest
mode, a callback is called to check whether an intercept is needed or
not. The callback is called at three points in time: immediately after
beginning execution, after checking privilge exceptions, and after
checking memory exception. This suits the different interception points
defined for different instructions and for the various virtualization
instruction sets.
In addition, a new X86EMUL_INTERCEPT is defined, which any callback or
memory access may define, allowing the more complicated intercepts to be
implemented in existing callbacks.
Signed-off-by: Avi Kivity <avi@redhat.com> Signed-off-by: Joerg Roedel <joerg.roedel@amd.com> Signed-off-by: Avi Kivity <avi@redhat.com>
Avi Kivity [Tue, 29 Mar 2011 09:34:38 +0000 (11:34 +0200)]
KVM: x86 emulator: Specialize decoding for insns with 66/f2/f3 prefixes
Most SIMD instructions use the 66/f2/f3 prefixes to distinguish between
different variants of the same instruction. Usually the encoding is quite
regular, but in some cases (including non-SIMD instructions) the prefixes
generate very different instructions. Examples include XCHG/PAUSE,
MOVQ/MOVDQA/MOVDQU, and MOVBE/CRC32.
Allow the emulator to handle these special cases by splitting such opcodes
into groups, with different decode flags and execution functions for different
prefixes.
Avi Kivity [Wed, 20 Jan 2010 14:00:35 +0000 (16:00 +0200)]
KVM: x86 emulator: do not munge rep prefix
Currently we store a rep prefix as 1 or 2 depending on whether it is a REPE or
REPNE. Since sse instructions depend on the prefix value, store it as the
original opcode to simplify things further on.
Avi Kivity [Wed, 20 Jan 2010 10:01:20 +0000 (12:01 +0200)]
KVM: 16-byte mmio support
Since sse instructions can issue 16-byte mmios, we need to support them. We
can't increase the kvm_run mmio buffer size to 16 bytes without breaking
compatibility, so instead we break the large mmios into two smaller 8-byte
ones. Since the bus is 64-bit we aren't breaking any atomicity guarantees.
Glauber Costa [Wed, 23 Mar 2011 16:40:42 +0000 (13:40 -0300)]
KVM: expose async pf through our standard mechanism
As Avi recently mentioned, the new standard mechanism for exposing features
is KVM_GET_SUPPORTED_CPUID, not spamming CAPs. For some reason async pf
missed that.
So expose async_pf here.
Signed-off-by: Glauber Costa <glommer@redhat.com> CC: Gleb Natapov <gleb@redhat.com> CC: Avi Kivity <avi@redhat.com> Signed-off-by: Avi Kivity <avi@redhat.com>
Avi Kivity [Wed, 23 Mar 2011 13:02:47 +0000 (15:02 +0200)]
KVM: VMX: simplify NMI mask management
Use vmx_set_nmi_mask() instead of open-coding management of
the hardware bit and the software hint (nmi_known_unmasked).
There's a slight change of behaviour when running without
hardware virtual NMI support - we now clear the NMI mask if
NMI delivery faulted in that case as well. This improves
emulation accuracy.
Avi Kivity [Mon, 7 Mar 2011 14:52:07 +0000 (16:52 +0200)]
KVM: VMX: Avoid vmx_recover_nmi_blocking() when unneeded
When we haven't injected an interrupt, we don't need to recover
the nmi blocking state (since the guest can't set it by itself).
This allows us to avoid a VMREAD later on.
Avi Kivity [Mon, 7 Mar 2011 13:26:44 +0000 (15:26 +0200)]
KVM: VMX: Cache cpl
We may read the cpl quite often in the same vmexit (instruction privilege
check, memory access checks for instruction and operands), so we gain
a bit if we cache the value.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (27 commits)
slcan: fix ldisc->open retval
net/usb: mark LG VL600 LTE modem ethernet interface as WWAN
xfrm: Don't allow esn with disabled anti replay detection
xfrm: Assign the inner mode output function to the dst entry
net: dev_close() should check IFF_UP
vlan: fix GVRP at dismantle time
netfilter: revert a2361c8735e07322023aedc36e4938b35af31eb0
netfilter: IPv6: fix DSCP mangle code
netfilter: IPv6: initialize TOS field in REJECT target module
IPVS: init and cleanup restructuring
IPVS: Change of socket usage to enable name space exit.
netfilter: ebtables: only call xt_compat_add_offset once per rule
netfilter: fix ebtables compat support
netfilter: ctnetlink: fix timestamp support for new conntracks
pch_gbe: support ML7223 IOH
PCH_GbE : Fixed the issue of checksum judgment
PCH_GbE : Fixed the issue of collision detection
NET: slip, fix ldisc->open retval
be2net: Fixed bugs related to PVID.
ehea: fix wrongly reported speed and port
...
David Rientjes [Wed, 11 May 2011 00:08:54 +0000 (17:08 -0700)]
slub: Revert "[PARISC] slub: fix panic with DISCONTIGMEM"
This reverts commit 4a5fa3590f09, which did not allow SLUB to be used
on architectures that use DISCONTIGMEM without compiling NUMA support
without CONFIG_BROKEN also set.
The slub panic that it was intended to prevent is addressed by d9b41e0b54fd ("[PARISC] set memory ranges in N_NORMAL_MEMORY when
onlined") on parisc so there is no further slub issues with such a
configuration.
The reverts allows SLUB now to be used on such architectures since
there haven't been any reports of additional errors.
Cc: James Bottomley <James.Bottomley@suse.de> Signed-off-by: David Rientjes <rientjes@google.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Oliver Hartkopp [Tue, 10 May 2011 20:12:30 +0000 (13:12 -0700)]
slcan: fix ldisc->open retval
TTY layer expects 0 if the ldisc->open operation succeeded.
Reported-by: Matvejchikov Ilya <matvejchikov@gmail.com> Signed-off-by: Oliver Hartkopp <socketcan@hartkopp.net> Signed-off-by: David S. Miller <davem@davemloft.net>
Dan Williams [Mon, 9 May 2011 07:43:20 +0000 (07:43 +0000)]
net/usb: mark LG VL600 LTE modem ethernet interface as WWAN
Like other mobile broadband device ethernet interfaces, mark the LG
VL600 with the 'wwan' devtype so userspace knows it needs additional
configuration via the AT port before the interface can be used.
Signed-off-by: Dan Williams <dcbw@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
xfrm: Don't allow esn with disabled anti replay detection
Unlike the standard case, disabled anti replay detection needs some
nontrivial extra treatment on ESN. RFC 4303 states:
Note: If a receiver chooses to not enable anti-replay for an SA, then
the receiver SHOULD NOT negotiate ESN in an SA management protocol.
Use of ESN creates a need for the receiver to manage the anti-replay
window (in order to determine the correct value for the high-order
bits of the ESN, which are employed in the ICV computation), which is
generally contrary to the notion of disabling anti-replay for an SA.
So return an error if an ESN state with disabled anti replay detection
is inserted for now and add the extra treatment later if we need it.
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
xfrm: Assign the inner mode output function to the dst entry
As it is, we assign the outer modes output function to the dst entry
when we create the xfrm bundle. This leads to two problems on interfamily
scenarios. We might insert ipv4 packets into ip6_fragment when called
from xfrm6_output. The system crashes if we try to fragment an ipv4
packet with ip6_fragment. This issue was introduced with git commit ad0081e4 (ipv6: Fragment locally generated tunnel-mode IPSec6 packets
as needed). The second issue is, that we might insert ipv4 packets in
netfilter6 and vice versa on interfamily scenarios.
With this patch we assign the inner mode output function to the dst entry
when we create the xfrm bundle. So xfrm4_output/xfrm6_output from the inner
mode is used and the right fragmentation and netfilter functions are called.
We switch then to outer mode with the output_finish functions.
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Reported-by: Frank Blaschka <blaschka@linux.vnet.ibm.com> Reported-by: Einar EL Lueck <ELELUECK@de.ibm.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
We should call vlan_gvrp_request_leave() from unregister_vlan_dev(),
not from vlan_dev_stop(), because vlan_gvrp_uninit_applicant()
is called right after unregister_netdevice_queue(). In batch mode,
unregister_netdevice_queue() doesn’t immediately call vlan_dev_stop().
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Linus Torvalds [Tue, 10 May 2011 18:56:35 +0000 (11:56 -0700)]
Merge branch 'for-linus' of git://oss.sgi.com/xfs/xfs
* 'for-linus' of git://oss.sgi.com/xfs/xfs:
xfs: fix race condition in AIL push trigger
xfs: make AIL target updates and compares 32bit safe.
xfs: always push the AIL to the target
xfs: exit AIL push work correctly when AIL is empty
xfs: ensure reclaim cursor is reset correctly at end of AG
Manuel Lauss [Sat, 7 May 2011 11:55:19 +0000 (13:55 +0200)]
MIPS: Alchemy: fix xxs1500 build error
This fixes:
alchemy/xxs1500/init.c: In function 'prom_init':
alchemy/xxs1500/init.c:57:17: error: ignoring return value of 'kstrtoul', declared with attribute warn_unused_result
MIPS: JZ4740: Set one-shot feature flag for the clockevent
The code for supporting one-shot mode for the clockevent is already there,
only the feature flag was not set. Setting the one-shot flag allows the
kernel to run in tickless mode.
CC arch/mips/jz4740/dma.o
arch/mips/jz4740/dma.c: In function 'jz4740_dma_chan_irq':
arch/mips/jz4740/dma.c:245:11: error: variable 'status' set but not used [-Werro
r=unused-but-set-variable]
HOSTCC arch/mips/boot/compressed/calc_vmlinuz_load_addr
arch/mips/boot/compressed/calc_vmlinuz_load_addr.c: In function 'main':
arch/mips/boot/compressed/calc_vmlinuz_load_addr.c:35:2: warning: format '%llx' expects type 'long long unsigned int *', but argument 3 has type 'uint64_t *'
arch/mips/boot/compressed/calc_vmlinuz_load_addr.c:54:2: warning: format '%llx' expects type 'long long unsigned int', but argument 2 has type 'uint64_t'
CC arch/mips/alchemy/devboards/db1x00/board_setup.o
arch/mips/alchemy/devboards/db1x00/board_setup.c: In function 'board_setup':
arch/mips/alchemy/devboards/db1x00/board_setup.c:130:6: error: variable 'pin_func' set but not used [-Werror=unused-but-set-variable]
CC arch/mips/sgi-ip27/ip27-hubio.o
arch/mips/sgi-ip27/ip27-hubio.c: In function 'hub_pio_map':
arch/mips/sgi-ip27/ip27-hubio.c:32:20: error: variable 'junk' set but not used [-Werror=unused-but-set-variable]
cc1: all warnings being treated as errors
CC arch/mips/sgi-ip27/ip27-hubio.o
arch/mips/sgi-ip27/ip27-hubio.c: In function 'hub_pio_map':
arch/mips/sgi-ip27/ip27-hubio.c:32:20: error: variable 'junk' set but not used [-Werror=unused-but-set-variable]
cc1: all warnings being treated as errors
David Daney [Fri, 18 Feb 2011 02:23:32 +0000 (18:23 -0800)]
MIPS: Octeon: Guard the Kconfig body with CPU_CAVIUM_OCTEON
Instead of making each Octeon specific option depend on
CPU_CAVIUM_OCTEON, gate the body of the entire file with
CPU_CAVIUM_OCTEON. With this change, CAVIUM_OCTEON_SPECIFIC_OPTIONS
becomes useless, so get rid of it as well.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
To: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/2091/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Ralf Baechle [Tue, 29 Mar 2011 10:32:55 +0000 (12:32 +0200)]
MIPS: Loongson: Fix GCC 2.6.0 build error.
CC arch/mips/loongson/common/env.o
arch/mips/loongson/common/env.c: In function 'prom_init_env':
arch/mips/loongson/common/env.c:50:12: error: variable 'ret' set but not used [-Werror=unused-but-set-variable]
arch/mips/loongson/common/env.c:51:12: error: variable 'ret' set but not used [-Werror=unused-but-set-variable]
arch/mips/loongson/common/env.c:52:12: error: variable 'ret' set but not used [-Werror=unused-but-set-variable]
arch/mips/loongson/common/env.c:53:12: error: variable 'ret' set but not used [-Werror=unused-but-set-variable]
cc1: all warnings being treated as errors
Ralf Baechle [Tue, 29 Mar 2011 10:09:51 +0000 (12:09 +0200)]
MIPS: Jazz: Fix GCC 4.6.0 build error
CC arch/mips/jazz/jazzdma.o
arch/mips/jazz/jazzdma.c: In function 'vdma_remap':
arch/mips/jazz/jazzdma.c:214:20: error: variable 'npages' set but not used [-Werror=unused-but-set-variable]
cc1: all warnings being treated as errors
Ralf Baechle [Tue, 29 Mar 2011 09:57:11 +0000 (11:57 +0200)]
MIPS: SNI: Fix GCC 4.6.0 build error
CC arch/mips/sni/time.o
arch/mips/sni/time.c: In function 'dosample':
arch/mips/sni/time.c:98:19: error: variable 'lsb' set but not used [-Werror=unused-but-set-variable]
cc1: all warnings being treated as errors
Ralf Baechle [Tue, 29 Mar 2011 09:48:22 +0000 (11:48 +0200)]
MIPS: Malta: Fix GCC 4.6.0 build error
CC arch/mips/mti-malta/malta-int.o
arch/mips/mti-malta/malta-int.c: In function 'mips_pcibios_iack':
arch/mips/mti-malta/malta-int.c:59:6: error: variable 'dummy' set but not used [-Werror=unused-but-set-variable]
cc1: all warnings being treated as errors
Ralf Baechle [Tue, 29 Mar 2011 09:43:19 +0000 (11:43 +0200)]
MIPS: Malta: Fix GCC 4.6.0 build error
CC arch/mips/mti-malta/malta-init.o
arch/mips/mti-malta/malta-init.c: In function 'prom_init':
arch/mips/mti-malta/malta-init.c:196:6: error: variable 'result' set but not used [-Werror=unused-but-set-variable]
cc1: all warnings being treated as errors
Ralf Baechle [Tue, 29 Mar 2011 09:06:49 +0000 (11:06 +0200)]
MIPS: IP22: Fix GCC 4.6.0 build error
CC arch/mips/sgi-ip22/ip22-platform.o
arch/mips/sgi-ip22/ip22-platform.c: In function 'sgiseeq_devinit':
arch/mips/sgi-ip22/ip22-platform.c:135:15: error: variable 'tmp' set but not used [-Werror=unused-but-set-variable]
cc1: all warnings being treated as errors
While at it rename the variable to pbdma for readability; there is a
local variable tmp of different type being used in two nested blocks.
Ralf Baechle [Tue, 29 Mar 2011 09:00:44 +0000 (11:00 +0200)]
MIPS: IP22: Fix GCC 4.6.0 build error
CC arch/mips/sgi-ip22/ip22-time.o
arch/mips/sgi-ip22/ip22-time.c: In function 'dosample':
arch/mips/sgi-ip22/ip22-time.c:35:10: error: variable 'lsb' set but not used [-Werror=unused-but-set-variable]
cc1: all warnings being treated as errors
Ralf Baechle [Tue, 29 Mar 2011 08:54:54 +0000 (10:54 +0200)]
MIPS: tlbex: Fix GCC 4.6.0 build error
CC arch/mips/mm/tlbex.o
arch/mips/mm/tlbex.c: In function 'build_r4000_tlb_refill_handler':
arch/mips/mm/tlbex.c:1155:22: error: variable 'vmalloc_mode' set but not used [-Werror=unused-but-set-variable]
arch/mips/mm/tlbex.c:1154:28: error: variable 'htlb_info' set but not used [-Werror=unused-but-set-variable]
cc1: all warnings being treated as errors
Ralf Baechle [Tue, 29 Mar 2011 08:50:38 +0000 (10:50 +0200)]
MIPS: c-r4k: Fix GCC 4.6.0 build error
CC arch/mips/mm/c-r4k.o
arch/mips/mm/c-r4k.c: In function 'probe_scache':
arch/mips/mm/c-r4k.c:1078:6: error: variable 'tmp' set but not used [-Werror=unused-but-set-variable]
cc1: all warnings being treated as errors
Older GCC versions didn't warn about the unused variable tmp because it was
getting initialized.
David Daney [Tue, 28 Dec 2010 21:21:37 +0000 (13:21 -0800)]
MIPS: Mask jump target in ftrace_dyn_arch_init_insns().
The current code is abusing the uasm interface by passing jump target
addresses with high bits set. Mask the addresses to avoid annoying
messages at boot time.
Signed-off-by: David Daney <ddaney@caviumnetworks.com> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Wu Zhangjin <wuzhangjin@gmail.com>
Patchwork: https://patchwork.linux-mips.org/patch/1922/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Ryusuke Konishi [Tue, 10 May 2011 11:59:34 +0000 (20:59 +0900)]
nilfs2: fix infinite loop in nilfs_palloc_freev function
After having applied commit 9954e7af14868b8b ("nilfs2: add free
entries count only if clear bit operation succeeded"), a free routine
of nilfs came to fall into an infinite loop, outputting the same
message endlessly:
nilfs_palloc_freev: entry number 29497 already freed
nilfs_palloc_freev: entry number 29497 already freed
nilfs_palloc_freev: entry number 29497 already freed
nilfs_palloc_freev: entry number 29497 already freed
nilfs_palloc_freev: entry number 29497 already freed ...
That patch broke the routine so that a loop counter is never updated
in an abnormal state. This fixes the regression.
Florian Wesphal says:
"... when the packet was sent from the local machine the skb
already has ->nfct attached, and -m conntrack seems to do
the right thing."
Acked-by: Jan Engelhardt <jengelh@medozas.de> Reported-by: Florian Wesphal <fw@strlen.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
netfilter: IPv6: initialize TOS field in REJECT target module
The IPv6 header is not zeroed out in alloc_skb so we must initialize
it properly unless we want to see IPv6 packets with random TOS fields
floating around. The current implementation resets the flow label
but this could be changed if deemed necessary.
We stumbled upon this issue when trying to apply a mangle rule to
the RST packet generated by the REJECT target module.
Signed-off-by: Fernando Luis Vazquez Cao <fernando@oss.ntt.co.jp> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
DESCRIPTION
This patch tries to restore the initial init and cleanup
sequences that was before namspace patch.
Netns also requires action when net devices unregister
which has never been implemented. I.e this patch also
covers when a device moves into a network namespace,
and has to be released.
IMPLEMENTATION
The number of calls to register_pernet_device have been
reduced to one for the ip_vs.ko
Schedulers still have their own calls.
This patch adds a function __ip_vs_service_cleanup()
and an enable flag for the netfilter hooks.
The nf hooks will be enabled when the first service is loaded
and never disabled again, except when a namespace exit starts.
Signed-off-by: Hans Schillstrom <hans@schillstrom.com> Acked-by: Julian Anastasov <ja@ssi.bg>
[horms@verge.net.au: minor edit to changelog] Signed-off-by: Simon Horman <horms@verge.net.au>
IPVS: Change of socket usage to enable name space exit.
If the sync daemons run in a name space while it crashes
or get killed, there is no way to stop them except for a reboot.
When all patches are there, ip_vs_core will handle register_pernet_(),
i.e. ip_vs_sync_init() and ip_vs_sync_cleanup() will be removed.
Kernel threads should not increment the use count of a socket.
By calling sk_change_net() after creating a socket this is avoided.
sock_release cant be used intead sk_release_kernel() should be used.
Thanks Eric W Biederman for your advices.
Signed-off-by: Hans Schillstrom <hans@schillstrom.com>
[horms@verge.net.au: minor edit to changelog] Signed-off-by: Simon Horman <horms@verge.net.au>
netfilter: ebtables: only call xt_compat_add_offset once per rule
The optimizations in commit 255d0dc34068a976
(netfilter: x_table: speedup compat operations) assume that
xt_compat_add_offset is called once per rule.
ebtables however called it for each match/target found in a rule.
The match/watcher/target parser already returns the needed delta, so it
is sufficient to move the xt_compat_add_offset call to a more reasonable
location.
While at it, also get rid of the unused COMPAT iterator macros.
Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Patrick McHardy <kaber@trash.net>
Eric Dumazet [Thu, 21 Apr 2011 08:57:21 +0000 (10:57 +0200)]
netfilter: fix ebtables compat support
commit 255d0dc34068a976 (netfilter: x_table: speedup compat operations)
made ebtables not working anymore.
1) xt_compat_calc_jump() is not an exact match lookup
2) compat_table_info() has a typo in xt_compat_init_offsets() call
3) compat_do_replace() misses a xt_compat_init_offsets() call
Reported-by: dann frazier <dannf@dannf.org> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: Patrick McHardy <kaber@trash.net>
Use proper data type to handle get_user_pages_fast error condition. Also
do not treat EFAULT error as fatal.
Signed-off-by: M. Mohan Kumar <mohan@in.ibm.com> Signed-off-by: Venkateswararao Jujjuri <jvrao@linux.vnet.ibm.com> Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
Hugh Dickins [Tue, 10 May 2011 00:44:42 +0000 (17:44 -0700)]
vm: fix vm_pgoff wrap in upward expansion
Commit a626ca6a6564 ("vm: fix vm_pgoff wrap in stack expansion") fixed
the case of an expanding mapping causing vm_pgoff wrapping when you had
downward stack expansion. But there was another case where IA64 and
PA-RISC expand mappings: upward expansion.
Linus Torvalds [Mon, 9 May 2011 23:59:51 +0000 (16:59 -0700)]
Merge branch 'drm-intel-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/keithp/linux-2.6
* 'drm-intel-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/keithp/linux-2.6:
drm/i915/lvds: Only act on lid notify when the device is on
drm/i915: fix intel_crtc_clock_get pipe reads after "cleanup cleanup"
drm/i915: Only enable the plane after setting the fb base (pre-ILK)
drm/i915/dp: Be paranoid in case we disable a DP before it is attached
drm/i915: Release object along create user fb error path
Dave Chinner [Fri, 6 May 2011 02:54:08 +0000 (02:54 +0000)]
xfs: fix race condition in AIL push trigger
The recent conversion of the xfsaild functionality to a work queue
introduced a hard-to-hit log space grant hang. One is caused by a
race condition in determining whether there is a psh in progress or
not.
The XFS_AIL_PUSHING_BIT is used to determine whether a push is
currently in progress. When the AIL push work completes, it checked
whether the target changed and cleared the PUSHING bit to allow a
new push to be requeued. The race condition is as follows:
Thread 1 push work
smp_wmb()
smp_rmb()
check ailp->xa_target unchanged
update ailp->xa_target
test/set PUSHING bit
does not queue
clear PUSHING bit
does not requeue
Now that the push target is updated, new attempts to push the AIL
will not trigger as the push target will be the same, and hence
despite trying to push the AIL we won't ever wake it again.
The fix is to ensure that the AIL push work clears the PUSHING bit
before it checks if the target is unchanged.
As a result, both push triggers operate on the same test/set bit
criteria, so even if we race in the push work and miss the target
update, the thread requesting the push will still set the PUSHING
bit and queue the push work to occur. For safety sake, the same
queue check is done if the push work detects the target change,
though only one of the two will will queue new work due to the use
of test_and_set_bit() checks.
Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Alex Elder <aelder@sgi.com>
(cherry picked from commit e4d3c4a43b595d5124ae824d300626e6489ae857)
Dave Chinner [Fri, 6 May 2011 02:54:07 +0000 (02:54 +0000)]
xfs: make AIL target updates and compares 32bit safe.
The recent conversion of the xfsaild functionality to a work queue
introduced a hard-to-hit log space grant hang. One of the problems
noticed was that updates of the push target are not 32 bit safe as
the target is a 64 bit value.
We cannot copy a 64 bit LSN without the possibility of corrupting
the result when racing with another updating thread. We have
function to do this update safely without needing to care about
32/64 bit issues - xfs_trans_ail_copy_lsn() - so use that when
updating the AIL push target.
Also move the reading of the target in the push work inside the AIL
lock, and use XFS_LSN_CMP() for the unlocked comparison during work
termination to close read holes as well.
Signed-off-by: Dave Chinner <david@fromorbit.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Alex Elder <aelder@sgi.com>
(cherry picked from commit fd5670f22fce247754243cf2ed41941e5762d990)
Dave Chinner [Fri, 6 May 2011 02:54:06 +0000 (02:54 +0000)]
xfs: always push the AIL to the target
The recent conversion of the xfsaild functionality to a work queue
introduced a hard-to-hit log space grant hang. One of the problems
discovered is a target mismatch between the item pushing loop and
the target itself.
The push trigger checks for the target increasing (i.e. new target >
current) while the push loop only pushes items that have a LSN <
current. As a result, we can get the situation where the push target
is X, the items at the tail of the AIL have LSN X and they don't get
pushed. The push work then completes thinking it is done, and cannot
be restarted until the push target increases to >= X + 1. If the
push target then never increases (because the tail is not moving),
then we never run the push work again and we stall.
Fix it by making sure log items with a LSN that matches the target
exactly are pushed during the loop.
Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Alex Elder <aelder@sgi.com>
(cherry picked from commit cb64026b6e8af50db598ec7c3f59d504259b00bb)
Dave Chinner [Fri, 6 May 2011 02:54:05 +0000 (02:54 +0000)]
xfs: exit AIL push work correctly when AIL is empty
The recent conversion of the xfsaild functionality to a work queue
introduced a hard-to-hit log space grant hang. The main cause is a
regression where a work exit path fails to clear the PUSHING state
and recheck the target correctly.
Make both exit paths do the same PUSHING bit clearing and target
checking when the "no more work to be done" condition is hit.
Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Alex Elder <aelder@sgi.com>
(cherry picked from commit ea35a20021f8497390d05b93271b4d675516c654)
Dave Chinner [Fri, 6 May 2011 02:54:04 +0000 (02:54 +0000)]
xfs: ensure reclaim cursor is reset correctly at end of AG
On a 32 bit highmem PowerPC machine, the XFS inode cache was growing
without bound and exhausting low memory causing the OOM killer to be
triggered. After some effort, the problem was reproduced on a 32 bit
x86 highmem machine.
The problem is that the per-ag inode reclaim index cursor was not
getting reset to the start of the AG if the radix tree tag lookup
found no more reclaimable inodes. Hence every further reclaim
attempt started at the same index beyond where any reclaimable
inodes lay, and no further background reclaim ever occurred from the
AG.
Without background inode reclaim the VM driven cache shrinker
simply cannot keep up with cache growth, and OOM is the result.
While the change that exposed the problem was the conversion of the
inode reclaim to use work queues for background reclaim, it was not
the cause of the bug. The bug was introduced when the cursor code
was added, just waiting for some weird configuration to strike....
Signed-off-by: Dave Chinner <dchinner@redhat.com> Tested-By: Christian Kujau <lists@nerdbynature.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Alex Elder <aelder@sgi.com>
(cherry picked from commit b223221956675ce8a7b436d198ced974bb388571)
projects / karo-tx-linux.git / log