When looking at a performance problem on PowerPC, I noticed some awful code
generation:
c00000000051fc98: 3b 60 00 01 li r27,1
...
c00000000051fca0: 3b 80 00 00 li r28,0
...
c00000000051fcdc: 93 61 00 70 stw r27,112(r1)
c00000000051fce0: 93 81 00 74 stw r28,116(r1)
c00000000051fce4: 81 21 00 70 lwz r9,112(r1)
c00000000051fce8: 80 01 00 74 lwz r0,116(r1)
c00000000051fcec: 7d 29 07 b4 extsw r9,r9
c00000000051fcf0: 7c 00 07 b4 extsw r0,r0
c00000000051fcf4: 7c 20 04 ac lwsync
c00000000051fcf8: 7d 60 f8 28 lwarx r11,0,r31
c00000000051fcfc: 7c 0b 48 00 cmpw r11,r9
c00000000051fd00: 40 c2 00 10 bne- c00000000051fd10
c00000000051fd04: 7c 00 f9 2d stwcx. r0,0,r31
c00000000051fd08: 40 c2 ff f0 bne+ c00000000051fcf8
c00000000051fd0c: 4c 00 01 2c isync
We create two constants, write them out to the stack, read them straight back
in and sign extend them. What a mess.
It turns out this bad code is a result of us defining atomic_t as a
volatile int.
We removed the volatile attribute from the powerpc atomic_t definition years
ago, but commit ea435467500612636f8f4fb639ff6e76b2496e4b (atomic_t: unify all
arch definitions) added it back in.
To dig up an old quote from Linus:
> The fact is, volatile on data structures is a bug. It's a wart in the C
> language. It shouldn't be used.
>
> Volatile accesses in *code* can be ok, and if we have "atomic_read()"
> expand to a "*(volatile int *)&(x)->value", then I'd be ok with that.
>
> But marking data structures volatile just makes the compiler screw up
> totally, and makes code for initialization sequences etc much worse.
And screw up it does :)
With the volatile removed, we see much more reasonable code generation:
c00000000051f5b8: 3b 60 00 01 li r27,1
...
c00000000051f5c0: 3b 80 00 00 li r28,0
...
c00000000051fc7c: 7c 20 04 ac lwsync
c00000000051fc80: 7c 00 f8 28 lwarx r0,0,r31
c00000000051fc84: 7c 00 d8 00 cmpw r0,r27
c00000000051fc88: 40 c2 00 10 bne- c00000000051fc98
c00000000051fc8c: 7f 80 f9 2d stwcx. r28,0,r31
c00000000051fc90: 40 c2 ff f0 bne+ c00000000051fc80
c00000000051fc94: 4c 00 01 2c isync
Six instructions less.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In preparation for removing volatile from the atomic_t definition, this
patch adds a volatile cast to all the atomic read functions.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When umount calls sync_filesystem(), we first do a WB_SYNC_NONE
writeback to kick off writeback of pending dirty inodes, then follow
that up with a WB_SYNC_ALL to wait for it. Since umount already holds
the sb s_umount mutex, WB_SYNC_NONE ends up doing nothing and all
writeback happens as WB_SYNC_ALL. This can greatly slow down umount,
since WB_SYNC_ALL writeback is a data integrity operation and thus
a bigger hammer than simple WB_SYNC_NONE. For barrier aware file systems
it's a lot slower.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
This patch limits the number of pages per memory slot to make
us free from extra care about type issues.
Signed-off-by: Takuya Yoshikawa <yoshikawa.takuya@oss.ntt.co.jp>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
MOL uses its own hypercall interface to call back into userspace when
the guest wants to do something.
So let's implement that as an exit reason, specify it with a CAP and
only really use it when userspace wants us to.
The only user of it so far is MOL.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
Some times we don't want all capabilities to be available to all
our vcpus. One example for that is the OSI interface, implemented
in the next patch.
In order to have a generic mechanism in how to enable capabilities
individually, this patch introduces a new ioctl that can be used
for this purpose. That way features we don't want in all guests or
userspace configurations can just not be enabled and we're good.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
Userspace can tell us that it wants to trigger an interrupt. But
so far it can't tell us that it wants to stop triggering one.
So let's interpret the parameter to the ioctl that we have anyways
to tell us if we want to raise or lower the interrupt line.
Signed-off-by: Alexander Graf <agraf@suse.de>
v2 -> v3:
- Add CAP for unset irq
Signed-off-by: Avi Kivity <avi@redhat.com>
This patch does:
- no need call tracepoint_synchronize_unregister() when kvm module
is unloaded since ftrace can handle it
- cleanup ftrace's macro
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Copy the last breaking event address from the lowcore to a new
field in the thread_struct on each system entry. Add a new
ptrace request PTRACE_GET_LAST_BREAK and a new utrace regset
REGSET_LAST_BREAK to query the last breaking event.
This is useful for debugging wild branches in user space code.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
This patch creates a port_id member in struct fc_lport.
This allows libfc to just deal with fc_lport instances
instead of calling into the fc_host to get the port_id.
This change helps in only using symbols necessary for
operation from the libfc structures. libfc still needs
to change the fc_host_port_id() if the port_id changes
so the presentation layer (scsi_transport_fc) can provide
the user with the correct value, but libfc shouldn't
rely on the presentation layer for operational values.
Signed-off-by: Robert Love <robert.w.love@intel.com>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
rtnetlink: make SR-IOV VF interface symmetric
sctp: delete active ICMP proto unreachable timer when free transport
tcp: fix MD5 (RFC2385) support
Now we have a set of nested attributes:
IFLA_VFINFO_LIST (NESTED)
IFLA_VF_INFO (NESTED)
IFLA_VF_MAC
IFLA_VF_VLAN
IFLA_VF_TX_RATE
This allows a single set to operate on multiple attributes if desired.
Among other things, it means a dump can be replayed to set state.
The current interface has yet to be released, so this seems like
something to consider for 2.6.34.
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
TCP-MD5 sessions have intermittent failures, when route cache is
invalidated. ip_queue_xmit() has to find a new route, calls
sk_setup_caps(sk, &rt->u.dst), destroying the
sk->sk_route_caps &= ~NETIF_F_GSO_MASK
that MD5 desperately try to make all over its way (from
tcp_transmit_skb() for example)
So we send few bad packets, and everything is fine when
tcp_transmit_skb() is called again for this socket.
Since ip_queue_xmit() is at a lower level than TCP-MD5, I chose to use a
socket field, sk_route_nocaps, containing bits to mask on sk_route_caps.
Reported-by: Bhaskar Dutta <bhaskie@gmail.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
TCP MD5 support uses percpu data for temporary storage. It currently
disables preemption so that same storage cannot be reclaimed by another
thread on same cpu.
We also have to make sure a softirq handler wont try to use also same
context. Various bug reports demonstrated corruptions.
Fix is to disable preemption and BH.
Reported-by: Bhaskar Dutta <bhaskie@gmail.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With RPS inclusion, skb timestamping is not consistent in RX path.
If netif_receive_skb() is used, its deferred after RPS dispatch.
If netif_rx() is used, its done before RPS dispatch.
This can give strange tcpdump timestamps results.
I think timestamping should be done as soon as possible in the receive
path, to get meaningful values (ie timestamps taken at the time packet
was delivered by NIC driver to our stack), even if NAPI already can
defer timestamping a bit (RPS can help to reduce the gap)
Tom Herbert prefer to sample timestamps after RPS dispatch. In case
sampling is expensive (HPET/acpi_pm on x86), this makes sense.
Let admins switch from one mode to another, using a new
sysctl, /proc/sys/net/core/netdev_tstamp_prequeue
Its default value (1), means timestamps are taken as soon as possible,
before backlog queueing, giving accurate timestamps.
Setting a 0 value permits to sample timestamps when processing backlog,
after RPS dispatch, to lower the load of the pre-RPS cpu.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now there's null check here and also again in the hook. Looking at bridge bits
which are simmilar, port structure is rcu_dereferenced right away in
handle_bridge and passed to hook. Looks nicer.
Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Acked-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
(Dropped the infiniband part, because Tetsuo modified the related code,
I will send a separate patch for it once this is accepted.)
This patch introduces /proc/sys/net/ipv4/ip_local_reserved_ports which
allows users to reserve ports for third-party applications.
The reserved ports will not be used by automatic port assignments
(e.g. when calling connect() or bind() with port number 0). Explicit
port allocation behavior is unchanged.
Signed-off-by: Octavian Purdila <opurdila@ixiacom.com>
Signed-off-by: WANG Cong <amwang@redhat.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The new function can be used to read/write large bitmaps via /proc. A
comma separated range format is used for compact output and input
(e.g. 1,3-4,10-10).
Writing into the file will first reset the bitmap then update it
based on the given input.
Signed-off-by: Octavian Purdila <opurdila@ixiacom.com>
Signed-off-by: WANG Cong <amwang@redhat.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
1) i_flags simply doesn't work for mount/unlink race prevention;
we may have many links to file and rm on one of those obviously
shouldn't prevent bind on top of another later on. To fix it
right way we need to mark _dentry_ as unsuitable for mounting
upon; new flag (DCACHE_CANT_MOUNT) is protected by d_flags and
i_mutex on the inode in question. Set it (with dont_mount(dentry))
in unlink/rmdir/etc., check (with cant_mount(dentry)) in places
in namespace.c that used to check for S_DEAD. Setting S_DEAD
is still needed in places where we used to set it (for directories
getting killed), since we rely on it for readdir/rmdir race
prevention.
2) rename()/mount() protection has another bogosity - we unhash
the target before we'd checked that it's not a mountpoint. Fixed.
3) ancient bogosity in pivot_root() - we locked i_mutex on the
right directory, but checked S_DEAD on the different (and wrong)
one. Noticed and fixed.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Now, with the introduction of the sff_set_devctl() method, we can
use it in sff_irq_on() method too -- that way its implementations
in 'pata_bf54x' and 'pata_scc' become virtually identical to
ata_sff_irq_on(). The sff_irq_on() method now becomes quite
superfluous, and the only reason not to remove it completely is
the existence of the 'pata_octeon_cf' driver which implements it
as an empty function. Just make the method optional then, with
ata_sff_irq_on() becoming generic taskfile-bound function, still
global for the 'pata_bf54x' driver to be able to call it from its
thaw() and postreset() methods.
While at it, make the sff_irq_on() method and ata_sff_irq_on() return
'void' as the result is always ignored anyway.
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
The set of libata's taskfile access methods is clearly incomplete as
it lacks a method to write to the device control register -- which
forces drivers like 'pata_bf54x' and 'pata_scc' to implement more
"high level" (and more weighty) methods like freeze() and postreset().
So, introduce the optional sff_set_devctl() method which the drivers
only have to implement if the standard iowrite8() can't be used (just
like the existing sff_check_altstatus() method) and make use of it
in the freeze() and postreset() method implementations (I could also
have used it in softreset() method but it also reads other taskfile
registers without using tf_read() making that quite pointless);
this makes freeze() method implementations in the 'pata_bf54x' and
'pata_scc' methods virtually identical to ata_sff_freeze(), so we
can get rid of them completely.
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
There are some SATA devices which take relatively long to get out of
0xff status after reset. In libata, this timeout is determined by
ATA_TMOUT_FF_WAIT. Quantum GoVault is the worst requring about 2s for
reliable detection. However, because 2s 0xff timeout can introduce
rather long spurious delay during boot, libata has been compromising
at the next longest timeout of 800ms for HHD424020F7SV00 iVDR drive.
Now that parallel scan is in place for common drivers, libata can
afford 2s 0xff timeout. Use 2s 0xff timeout if parallel scan is
enabled.
Please note that the chance of spurious wait is pretty slim w/ working
SCR access so this will only affect SATA controllers w/o SCR access
which isn't too common these days.
Please read the following thread for more information on the GoVault
drive.
http://thread.gmane.org/gmane.linux.ide/14545/focus=14663
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Gary Hade <garyhade@us.ibm.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
This can be used for AHCI-compatible interfaces implemented inside
System-On-Chip solutions, or AHCI devices connected via localbus.
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
Now that the rpc.gssd daemon can explicitly tell us that the key expired,
we should cache that information to avoid spamming gssd.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
This improves the packing of the rpc_task, and ensures that on 64-bit
platforms the size reduces to 216 bytes.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
It seems strange to maintain stats for bytes_sent in one structure, and
bytes received in another. Try to assemble all the RPC request-related
stats in struct rpc_rqst
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Again, we can deadlock if the memory reclaim triggers a writeback that
requires a rpcsec_gss credential lookup.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Currently RPC performance metrics that tabulate elapsed time use
jiffies time values. This is problematic on systems that use slow
jiffies (for instance 100HZ systems built for paravirtualized
environments). It is also a problem for computing precise latency
statistics for advanced network transports, such as InfiniBand,
that can have round-trip latencies significanly faster than a single
clock tick.
For the RPC client, adopt the high resolution time stamp mechanism
already used by the network layer and blktrace: ktime.
We use ktime format time stamps for all internal computations, and
convert to milliseconds for presentation. As a result, we need only
addition operations in the performance critical paths; multiply/divide
is required only for presentation.
We could report RTT metrics in microseconds. In fact the mountstats
format is versioned to accomodate exactly this kind of interface
improvement.
For now, however, we'll stay with millisecond precision for
presentation to maintain backwards compatibility with the handful of
currently deployed user space tools. At a later point, we'll move to
an API such as BDI_STATS where a finer timestamp precision can be
reported.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
To report ktime statistics to user space in milliseconds, a new helper
is required.
When considering how to do this conversion, I didn't immediately see
why the extra step of converting ktime to a timeval was needed. To
make that more clear, introduce a couple of large comments.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Compute an RPC request's RTT once, and use that value both for reporting
RPC metrics, and for adjusting the RTT context used by the RPC client's RTT
estimator algorithm.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Clean up: Update the documenting comment, and fix some minor white
space issues.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
We should not allow soft tasks to wait for longer than the major timeout
period when waiting for a reconnect to occur.
Remove the field xprt->connect_timeout since it has been obsoleted by
xprt->reestablish_timeout.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
NFS Filehandles and struct fattr are really too large to be allocated on
the stack. This patch adds in a couple of helper functions to allocate them
dynamically instead.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Add necessary changes to add kernel support for the rc4-hmac Kerberos
encryption type used by Microsoft and described in rfc4757.
Signed-off-by: Kevin Coffman <kwc@citi.umich.edu>
Signed-off-by: Steve Dickson <steved@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
All encryption types use a confounder at the beginning of the
wrap token. In all encryption types except arcfour-hmac, the
confounder is the same as the blocksize. arcfour-hmac has a
blocksize of one, but uses an eight byte confounder.
Add an entry to the crypto framework definitions for the
confounder length and change the wrap/unwrap code to use
the confounder length rather than assuming it is always
the blocksize.
Signed-off-by: Kevin Coffman <kwc@citi.umich.edu>
Signed-off-by: Steve Dickson <steved@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
For the arcfour-hmac support, the make_seq_num and get_seq_num
functions need access to the kerberos context structure.
This will be used in a later patch.
Signed-off-by: Kevin Coffman <kwc@citi.umich.edu>
Signed-off-by: Steve Dickson <steved@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
This is needed for deriving arcfour-hmac keys "on the fly"
using the sequence number or checksu
Signed-off-by: Kevin Coffman <kwc@citi.umich.edu>
Signed-off-by: Steve Dickson <steved@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
For arcfour-hmac support, the make_checksum function needs a usage
field to correctly calculate the checksum differently for MIC and
WRAP tokens.
Signed-off-by: Kevin Coffman <kwc@citi.umich.edu>
Signed-off-by: Steve Dickson <steved@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Add the remaining pieces to enable support for Kerberos AES
encryption types.
Signed-off-by: Kevin Coffman <kwc@citi.umich.edu>
Signed-off-by: Steve Dickson <steved@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
This is a step toward support for AES encryption types which are
required to use the new token formats defined in rfc4121.
Signed-off-by: Kevin Coffman <kwc@citi.umich.edu>
[SteveD: Fixed a typo in gss_verify_mic_v2()]
Signed-off-by: Steve Dickson <steved@redhat.com>
[Trond: Got rid of the TEST_ROTATE/TEST_EXTRA_COUNT crap]
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
