No need to bring in dm-mapper.h and along with it a dependency on BLOCK I/O
just to use dm_div_up(). Just use the existing DIV_ROUND_UP().
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andy King <acking@vmware.com>
Signed-off-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add PCI as a dependency to our build, since we always compile in the guest-side
PCI device support.
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andy King <acking@vmware.com>
Signed-off-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It doesn't seem this spinlock was properly initialized. This bug was
introduced by commit 7a410e8d4d.
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Smatch complains because the call to
pci_set_drvdata(dev, &sockets[i].socket);
is reading one step beyond the end of the sockets[] array. It will
crash when we use it later.
The only place which uses pci_get_drvdata() is i82092aa_pci_remove().
That function should loop through all the sockets and unregister them.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The code to print the FIFO size in tc574_config computes it as:
8 << config & Ram_size
which evaluates the '<<' first, but the actual intent is to evaluate the
'&' first. Add parentheses to enforce desired evaluation order.
Signed-off-by: Nickolai Zeldovich <nickolai@csail.mit.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
The send buffer was being leaked; fix it.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Reported-by: Jason Wang <jasowang@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There is bug in the definition of struct dm_info_msg. This patch fixes
the definition of this structure and makes the corresponding adjustments.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
bnx2x does an internal GRO pass but doesn't provide gso_segs, thus
breaking qdisc_pkt_len_init() in case ingress qdisc is used.
We store gso_segs in NAPI_GRO_CB(skb)->count, where tcp_gro_complete()
expects to find the number of aggregated segments.
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
John W. Linville says:
====================
This batch of fixes is intended for 3.8...
Included is a Bluetooth pull. Gustavo says:
"A few fixes for 3.8. Five of them are just new devices ids addition.
Apart from the that there is fix to a kernel memory leak to userspace from
Anderson Lizardo, two interoperability fixes from Jaganath Kanakkassery and
Szymon Janc. And a crash fix by me."
Along with that, Amitkumar Karwar brings a pair of mwifiex fixes for
problems related to handling of band information within the driver.
These problems can lead to association failures.
Sujith Manoharan fixes a memory leak in the ath9k_htc code (originally
reported by Larry Finger).
The big hero this round is Felix Fietkau. Felix gives us seven
ath9k fixes, including a fix for a race condition, the removal of a
double-free, and a fix for a deadlock, among others.
These have all been in linux-next for at least a couple of days,
with no complaints so far. Please let me know if there are problems!
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix for the daemon code and for hv_set_ifconfig.sh script, so
that the created ifcfg-* file is consistent with initscripts
documentation.
Signed-off-by: Tomas Hozza <thozza@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The variable userlen is initialized but never used
otherwise, so remove the unused variable.
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Now, cleanup and consolidate reporting of host and vmbus version numbers.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Capture the host build information so it can be presented along with the
negotiated vmbus version information.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Implement flow management on the send side. When the sender is blocked, the reader
can potentially signal the sender to indicate there is now room to send.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Now that we have implemented all of the Win8 (WS2012) functionality, negotiate
Win8 protocol with the host.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We establish the handler before we have fully initialized the VMBUS state.
Deal with spurious interrupts.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Vmbus interrupts are unique in that while the interrupt is delivered on a
given vector, these can be handled concurrently on different CPUs. Handle the
vmbus interrupts concurrently on all the CPUs.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Now that we can potentially take vmbus interrupts on any CPU, make the
tasklets per-CPU.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This call to seek offers is not necessary and just adds unnecessary delay.
Get rid of it.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Now that we have implemented a per-connection signaling mechanism, get rid
of the global signaling state. For hosts that don't support per-connection
signaling handle, we have moved the global state to be a per-channel state.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Implement a simple policy for distributing incoming interrupt load.
We classify channels as (a) performance critical and (b) not
performance critical. All non-performance critical channels will
be bound to the boot cpu. Performance critical channels will be
bound to the remaining available CPUs on a round-robin basis.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Starting with Win8 (WS2012), the event page can be used to directly get the
channel ID that needs servicing. Modify the channel event handling code
to take advantage of this feature.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add state to bind a channel to a specific VCPU. This will help us better
distribute incoming interrupt load.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On win8 (ws2012), incoming vmbus interrupt load can be spread across all
available VCPUs in the guest. On a per-channel basis, the interrupts can
be bound to specific CPUs. The Linux notion of cpu ID may be different
from that of the hypervisor's. Setup a mapping structure.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
On win7 (ws2008 R2) and beyond, we have the notion of having dedicated interrupts on
a per-channel basis. When a channel has a dedicated interrupt assigned, there is no need
to set the interrupt bit in the shared page. Implement this optimization.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The current code has a global handle for supporting signaling of the host
from guest. Make this a per-channel attribute as on some versions of the
host we can signal on per-channel handle.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To support version specific optimization in various vmbus drivers,
move the vmbus definitions to the public header file.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In preparation for supporting a per-connection signaling mechanism,
change the signature of vmbus_set_event(). This change is also
needed to implement other aspects of the signaling optimization.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In preparation for implementing a per-connection signaling framework,
change the signature of the function hv_signal_event(). The current
code uses a global handle for signaling the host.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Export the negotiated vmbus version as this may be useful for
individual drivers.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The "offfer" message sent by the host has been extended in win7 (ws2008 R2).
Add/modify state to reflect this extension. All these changes are backward
compatible.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Update the ringbuffer structure to support win8 functionality.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The current code hard coded the vmbus version independent of the host
it was running on. Add code to dynamically negotiate the most appropriate
version.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This is meant to remove a compiler warning. It should not make any
functional change.
payload_len should be initialized when it is passed to
write_tag_64_packet() as a pointer. If that call fails, this function
should return early, and payload_len won't be used.
Signed-off-by: Simon Que <sque@chromium.org>
Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Fix the 64bit optimized version of ipv6_prefix_equal to convert the
bitmask to network byte order only after the bit-shift.
The bug was introduced in:
3867517 ipv6: 64bit version of ipv6_prefix_equal().
Signed-off-by: Fabio Baltieri <fabio.baltieri@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Increase the amount of memory usage limits for incomplete
IP fragments.
Arguing for new thresh high/low values:
High threshold = 4 MBytes
Low threshold = 3 MBytes
The fragmentation memory accounting code, tries to account for the
real memory usage, by measuring both the size of frag queue struct
(inet_frag_queue (ipv4:ipq/ipv6:frag_queue)) and the SKB's truesize.
We want to be able to handle/hold-on-to enough fragments, to ensure
good performance, without causing incomplete fragments to hurt
scalability, by causing the number of inet_frag_queue to grow too much
(resulting longer searches for frag queues).
For IPv4, how much memory does the largest frag consume.
Maximum size fragment is 64K, which is approx 44 fragments with
MTU(1500) sized packets. Sizeof(struct ipq) is 200. A 1500 byte
packet results in a truesize of 2944 (not 2048 as I first assumed)
(44*2944)+200 = 129736 bytes
The current default high thresh of 262144 bytes, is obviously
problematic, as only two 64K fragments can fit in the queue at the
same time.
How many 64K fragment can we fit into 4 MBytes:
4*2^20/((44*2944)+200) = 32.34 fragment in queues
An attacker could send a separate/distinct fake fragment packets per
queue, causing us to allocate one inet_frag_queue per packet, and thus
attacking the hash table and its lists.
How many frag queue do we need to store, and given a current hash size
of 64, what is the average list length.
Using one MTU sized fragment per inet_frag_queue, each consuming
(2944+200) 3144 bytes.
4*2^20/(2944+200) = 1334 frag queues -> 21 avg list length
An attack could send small fragments, the smallest packet I could send
resulted in a truesize of 896 bytes (I'm a little surprised by this).
4*2^20/(896+200) = 3827 frag queues -> 59 avg list length
When increasing these number, we also need to followup with
improvements, that is going to help scalability. Simply increasing
the hash size, is not enough as the current implementation does not
have a per hash bucket locking.
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This function is no longer used; get rid of it.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The host has already implemented the "read" side optimizations.
Leverage that to optimize "write" side signaling.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Now that we have the infratructure for correctly determining when we
should signal the host; optimize the signaling on the read side -
signaling the guest from the host.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Util driver is not a performance critical driver and furthermore some
util services such as KVP can only handle one outstanding message
from the host. Turn off batched reading for util drivers.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For the "read" side signaling optimization, the reader has to completely
drain the queue before exiting. Add state to manage this "batched"
reading.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Implement functions that will support read-side signaling optimization.
By having the reader indicate the start of the "read" operation and the
"end" of the read operation we can more efficiently handle the signaling
protocol: while the read is in progress, there is no need for the "writer"
to signal the "reader" as new items are put on the read queue.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This saves us some cycles, but does not affect the placement result at
all.
This corresponds to ceph.git commit 4abb53d4f.
Signed-off-by: Sage Weil <sage@inktank.com>
Add libceph support for a new CRUSH tunable recently added to Ceph servers.
Consider the CRUSH rule
step chooseleaf firstn 0 type <node_type>
This rule means that <n> replicas will be chosen in a manner such that
each chosen leaf's branch will contain a unique instance of <node_type>.
When an object is re-replicated after a leaf failure, if the CRUSH map uses
a chooseleaf rule the remapped replica ends up under the <node_type> bucket
that held the failed leaf. This causes uneven data distribution across the
storage cluster, to the point that when all the leaves but one fail under a
particular <node_type> bucket, that remaining leaf holds all the data from
its failed peers.
This behavior also limits the number of peers that can participate in the
re-replication of the data held by the failed leaf, which increases the
time required to re-replicate after a failure.
For a chooseleaf CRUSH rule, the tree descent has two steps: call them the
inner and outer descents.
If the tree descent down to <node_type> is the outer descent, and the descent
from <node_type> down to a leaf is the inner descent, the issue is that a
down leaf is detected on the inner descent, so only the inner descent is
retried.
In order to disperse re-replicated data as widely as possible across a
storage cluster after a failure, we want to retry the outer descent. So,
fix up crush_choose() to allow the inner descent to return immediately on
choosing a failed leaf. Wire this up as a new CRUSH tunable.
Note that after this change, for a chooseleaf rule, if the primary OSD
in a placement group has failed, choosing a replacement may result in
one of the other OSDs in the PG colliding with the new primary. This
requires that OSD's data for that PG to need moving as well. This
seems unavoidable but should be relatively rare.
This corresponds to ceph.git commit 88f218181a9e6d2292e2697fc93797d0f6d6e5dc.
Signed-off-by: Jim Schutt <jaschut@sandia.gov>
Reviewed-by: Sage Weil <sage@inktank.com>
The MDS may have incorrect wanted caps after importing caps. So the
client should check the value mds has and send cap update if necessary.
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Reviewed-by: Sage Weil <sage@inktank.com>
When client wants to release an imported cap, it's possible there
is no reserved cap_release message in corresponding mds session.
so __queue_cap_release causes kernel panic.
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Reviewed-by: Sage Weil <sage@inktank.com>
