17 Apr, 2015
1 commit
-
Pull xen features and fixes from David Vrabel:
- use a single source list of hypercalls, generating other tables etc.
at build time.- add a "Xen PV" APIC driver to support >255 VCPUs in PV guests.
- significant performance improve to guest save/restore/migration.
- scsiback/front save/restore support.
- infrastructure for multi-page xenbus rings.
- misc fixes.
* tag 'stable/for-linus-4.1-rc0-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
xen/pci: Try harder to get PXM information for Xen
xenbus_client: Extend interface to support multi-page ring
xen-pciback: also support disabling of bus-mastering and memory-write-invalidate
xen: support suspend/resume in pvscsi frontend
xen: scsiback: add LUN of restored domain
xen-scsiback: define a pr_fmt macro with xen-pvscsi
xen/mce: fix up xen_late_init_mcelog() error handling
xen/privcmd: improve performance of MMAPBATCH_V2
xen: unify foreign GFN map/unmap for auto-xlated physmap guests
x86/xen/apic: WARN with details.
x86/xen: Provide a "Xen PV" APIC driver to support >255 VCPUs
xen/pciback: Don't print scary messages when unsupported by hypervisor.
xen: use generated hypercall symbols in arch/x86/xen/xen-head.S
xen: use generated hypervisor symbols in arch/x86/xen/trace.c
xen: synchronize include/xen/interface/xen.h with xen
xen: build infrastructure for generating hypercall depending symbols
xen: balloon: Use static attribute groups for sysfs entries
xen: pcpu: Use static attribute groups for sysfs entry
15 Apr, 2015
1 commit
-
Originally Xen PV drivers only use single-page ring to pass along
information. This might limit the throughput between frontend and
backend.The patch extends Xenbus driver to support multi-page ring, which in
general should improve throughput if ring is the bottleneck. Changes to
various frontend / backend to adapt to the new interface are also
included.Affected Xen drivers:
* blkfront/back
* netfront/back
* pcifront/back
* scsifront/back
* vtpmfrontThe interface is documented, as before, in xenbus_client.c.
Signed-off-by: Wei Liu
Signed-off-by: Paul Durrant
Signed-off-by: Bob Liu
Cc: Konrad Wilk
Cc: Boris Ostrovsky
Signed-off-by: David Vrabel
21 Mar, 2015
2 commits
-
Conflicts:
drivers/net/ethernet/emulex/benet/be_main.c
net/core/sysctl_net_core.c
net/ipv4/inet_diag.cThe be_main.c conflict resolution was really tricky. The conflict
hunks generated by GIT were very unhelpful, to say the least. It
split functions in half and moved them around, when the real actual
conflict only existed solely inside of one function, that being
be_map_pci_bars().So instead, to resolve this, I checked out be_main.c from the top
of net-next, then I applied the be_main.c changes from 'net' since
the last time I merged. And this worked beautifully.The inet_diag.c and sysctl_net_core.c conflicts were simple
overlapping changes, and were easily to resolve.Signed-off-by: David S. Miller
-
With the current netback, the bandwidth limiter's parameters are only
settable during vif setup time. This patch register a watch on them, and
thus makes them runtime changeable.When the watch fires, the timer is reset. The timer's mutex is used for
fencing the change.Cc: Anthony Liguori
Signed-off-by: Imre Palik
Acked-by: Wei Liu
Signed-off-by: David S. Miller
12 Mar, 2015
1 commit
-
This fixes a performance regression introduced by
7fbb9d8415d4a51cf542e87cf3a717a9f7e6aedc (xen-netback: release pending
index before pushing Tx responses)Moving the notify outside of the spin locks means it can be delayed a
long time (if the dealloc thread is descheduled or there is an
interrupt or softirq).Signed-off-by: David Vrabel
Reviewed-by: Zoltan Kiss
Acked-by: Wei Liu
Signed-off-by: David S. Miller
10 Mar, 2015
1 commit
-
Conflicts:
drivers/net/ethernet/cadence/macb.cOverlapping changes in macb driver, mostly fixes and cleanups
in 'net' overlapping with the integration of at91_ether into
macb in 'net-next'.Signed-off-by: David S. Miller
06 Mar, 2015
3 commits
-
When handling a from-guest frag list, xenvif_handle_frag_list()
replaces the frags before calling the destructor to clean up the
original (foreign) frags. Whilst this is safe (the destructor doesn't
actually use the frags), it looks odd.Reorder the function to be less confusing.
Signed-off-by: David Vrabel
Signed-off-by: David S. Miller -
Every time a VIF is destroyed up to 256 pages may be leaked if packets
with more than MAX_SKB_FRAGS frags were transmitted from the guest.
Even worse, if another user of ballooned pages allocated one of these
ballooned pages it would not handle the unexpectedly >1 page count
(e.g., gntdev would deadlock when unmapping a grant because the page
count would never reach 1).When handling a from-guest skb with a frag list, unref the frags
before releasing them so they are freed correctly when the VIF is
destroyed.Signed-off-by: David Vrabel
Signed-off-by: David S. Miller -
Use correct pointer arithmetic to get the pointer to each stat.
Signed-off-by: David Vrabel
Signed-off-by: David S. Miller
04 Mar, 2015
2 commits
-
Conflicts:
drivers/net/ethernet/rocker/rocker.cThe rocker commit was two overlapping changes, one to rename
the ->vport member to ->pport, and another making the bitmask
expression use '1ULL' instead of plain '1'.Signed-off-by: David S. Miller
-
Use the built-in function instead of memset.
Signed-off-by: Joe Perches
Acked-by: Ian Campbell
Acked-by: Wei Liu
Signed-off-by: David S. Miller
25 Feb, 2015
1 commit
-
If the pending indexes are released /after/ pushing the Tx response
then a stale pending index may be used if a new Tx request is
immediately pushed by the frontend. The may cause various WARNINGs or
BUGs if the stale pending index is actually still in use.Fix this by releasing the pending index before pushing the Tx
response.The full barrier for the pending ring update is not required since the
the Tx response push already has a suitable write barrier.Signed-off-by: David Vrabel
Reviewed-by: Wei Liu
Signed-off-by: David S. Miller
11 Feb, 2015
2 commits
-
Pull networking updates from David Miller:
1) More iov_iter conversion work from Al Viro.
[ The "crypto: switch af_alg_make_sg() to iov_iter" commit was
wrong, and this pull actually adds an extra commit on top of the
branch I'm pulling to fix that up, so that the pre-merge state is
ok. - Linus ]2) Various optimizations to the ipv4 forwarding information base trie
lookup implementation. From Alexander Duyck.3) Remove sock_iocb altogether, from CHristoph Hellwig.
4) Allow congestion control algorithm selection via routing metrics.
From Daniel Borkmann.5) Make ipv4 uncached route list per-cpu, from Eric Dumazet.
6) Handle rfs hash collisions more gracefully, also from Eric Dumazet.
7) Add xmit_more support to r8169, e1000, and e1000e drivers. From
Florian Westphal.8) Transparent Ethernet Bridging support for GRO, from Jesse Gross.
9) Add BPF packet actions to packet scheduler, from Jiri Pirko.
10) Add support for uniqu flow IDs to openvswitch, from Joe Stringer.
11) New NetCP ethernet driver, from Muralidharan Karicheri and Wingman
Kwok.12) More sanely handle out-of-window dupacks, which can result in
serious ACK storms. From Neal Cardwell.13) Various rhashtable bug fixes and enhancements, from Herbert Xu,
Patrick McHardy, and Thomas Graf.14) Support xmit_more in be2net, from Sathya Perla.
15) Group Policy extensions for vxlan, from Thomas Graf.
16) Remove Checksum Offload support for vxlan, from Tom Herbert.
17) Like ipv4, support lockless transmit over ipv6 UDP sockets. From
Vlad Yasevich.* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1494+1 commits)
crypto: fix af_alg_make_sg() conversion to iov_iter
ipv4: Namespecify TCP PMTU mechanism
i40e: Fix for stats init function call in Rx setup
tcp: don't include Fast Open option in SYN-ACK on pure SYN-data
openvswitch: Only set TUNNEL_VXLAN_OPT if VXLAN-GBP metadata is set
ipv6: Make __ipv6_select_ident static
ipv6: Fix fragment id assignment on LE arches.
bridge: Fix inability to add non-vlan fdb entry
net: Mellanox: Delete unnecessary checks before the function call "vunmap"
cxgb4: Add support in cxgb4 to get expansion rom version via ethtool
ethtool: rename reserved1 memeber in ethtool_drvinfo for expansion ROM version
net: dsa: Remove redundant phy_attach()
IB/mlx4: Reset flow support for IB kernel ULPs
IB/mlx4: Always use the correct port for mirrored multicast attachments
net/bonding: Fix potential bad memory access during bonding events
tipc: remove tipc_snprintf
tipc: nl compat add noop and remove legacy nl framework
tipc: convert legacy nl stats show to nl compat
tipc: convert legacy nl net id get to nl compat
tipc: convert legacy nl net id set to nl compat
... -
Pull xen features and fixes from David Vrabel:
- Reworked handling for foreign (grant mapped) pages to simplify the
code, enable a number of additional use cases and fix a number of
long-standing bugs.- Prefer the TSC over the Xen PV clock when dom0 (and the TSC is
stable).- Assorted other cleanup and minor bug fixes.
* tag 'stable/for-linus-3.20-rc0-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip: (25 commits)
xen/manage: Fix USB interaction issues when resuming
xenbus: Add proper handling of XS_ERROR from Xenbus for transactions.
xen/gntdev: provide find_special_page VMA operation
xen/gntdev: mark userspace PTEs as special on x86 PV guests
xen-blkback: safely unmap grants in case they are still in use
xen/gntdev: safely unmap grants in case they are still in use
xen/gntdev: convert priv->lock to a mutex
xen/grant-table: add a mechanism to safely unmap pages that are in use
xen-netback: use foreign page information from the pages themselves
xen: mark grant mapped pages as foreign
xen/grant-table: add helpers for allocating pages
x86/xen: require ballooned pages for grant maps
xen: remove scratch frames for ballooned pages and m2p override
xen/grant-table: pre-populate kernel unmap ops for xen_gnttab_unmap_refs()
mm: add 'foreign' alias for the 'pinned' page flag
mm: provide a find_special_page vma operation
x86/xen: cleanup arch/x86/xen/mmu.c
x86/xen: add some __init annotations in arch/x86/xen/mmu.c
x86/xen: add some __init and static annotations in arch/x86/xen/setup.c
x86/xen: use correct types for addresses in arch/x86/xen/setup.c
...
06 Feb, 2015
2 commits
-
this patch fixes following sparse warning:
interface.c:83:5: warning: symbol 'xenvif_poll' was not declared. Should it be static?
Signed-off-by: Lad, Prabhakar
Acked-by: Wei Liu
Signed-off-by: David S. Miller -
Conflicts:
drivers/net/vxlan.c
drivers/vhost/net.c
include/linux/if_vlan.h
net/core/dev.cThe net/core/dev.c conflict was the overlap of one commit marking an
existing function static whilst another was adding a new function.In the include/linux/if_vlan.h case, the type used for a local
variable was changed in 'net', whereas the function got rewritten
to fix a stacked vlan bug in 'net-next'.In drivers/vhost/net.c, Al Viro's iov_iter conversions in 'net-next'
overlapped with an endainness fix for VHOST 1.0 in 'net'.In drivers/net/vxlan.c, vxlan_find_vni() added a 'flags' parameter
in 'net-next' whereas in 'net' there was a bug fix to pass in the
correct network namespace pointer in calls to this function.Signed-off-by: David S. Miller
03 Feb, 2015
1 commit
-
After commit e9d8b2c2968499c1f96563e6522c56958d5a1d0d (xen-netback:
disable rogue vif in kthread context), a fatal (protocol) error would
leave the guest Rx thread spinning, wasting CPU time. Commit
ecf08d2dbb96d5a4b4bcc53a39e8d29cc8fef02e (xen-netback: reintroduce
guest Rx stall detection) made this even worse by removing a
cond_resched() from this path.Since a fatal error is non-recoverable, just allow the guest Rx thread
to exit. This requires taking additional refs to the task so the
thread exiting early is handled safely.Signed-off-by: David Vrabel
Reported-by: Julien Grall
Tested-by: Julien Grall
Acked-by: Wei Liu
Signed-off-by: David S. Miller
28 Jan, 2015
3 commits
-
Use the foreign page flag in netback to get the domid and grant ref
needed for the grant copy. This signficiantly simplifies the netback
code and makes netback work with foreign pages from other backends
(e.g., blkback).This allows blkback to use iSCSI disks provided by domUs running on
the same host.Signed-off-by: Jennifer Herbert
Acked-by: Ian Campbell
Acked-by: David S. Miller
Signed-off-by: David Vrabel -
Add gnttab_alloc_pages() and gnttab_free_pages() to allocate/free pages
suitable to for granted maps.Signed-off-by: David Vrabel
Reviewed-by: Stefano Stabellini -
Ballooned pages are always used for grant maps which means the
original frame does not need to be saved in page->index nor restored
after the grant unmap.This allows the workaround in netback for the conflicting use of the
(unionized) page->index and page->pfmemalloc to be removed.Signed-off-by: Jennifer Herbert
Reviewed-by: Stefano Stabellini
Signed-off-by: David Vrabel
24 Jan, 2015
1 commit
-
Always fully coalesce guest Rx packets into the minimum number of ring
slots. Reducing the number of slots per packet has significant
performance benefits when receiving off-host traffic.Results from XenServer's performance benchmarks:
Baseline Full coalesce
Interhost VM receive 7.2 Gb/s 11 Gb/s
Interhost aggregate 24 Gb/s 24 Gb/s
Intrahost single stream 14 Gb/s 14 Gb/s
Intrahost aggregate 34 Gb/s 34 Gb/sHowever, this can increase the number of grant ops per packet which
decreases performance of backend (dom0) to VM traffic (by ~10%)
/unless/ grant copy has been optimized for adjacent ops with the same
source or destination (see "grant-table: defer releasing pages
acquired in a grant copy"[1] expected in Xen 4.6).[1] http://lists.xen.org/archives/html/xen-devel/2015-01/msg01118.html
Signed-off-by: David Vrabel
Acked-by: Ian Campbell
Signed-off-by: David S. Miller
07 Jan, 2015
1 commit
-
Since e9ce7cb6b107 ("xen-netback: Factor queue-specific data into queue struct"),
the transimt shaper timeout is always set to 0. The value the user sets via
xenbus is never propagated to the transmit shaper.This patch fixes the issue.
Cc: Anthony Liguori
Signed-off-by: Imre Palik
Acked-by: Ian Campbell
Signed-off-by: David S. Miller
19 Dec, 2014
1 commit
-
Commit bc96f648df1bbc2729abbb84513cf4f64273a1f1 (xen-netback: make
feature-rx-notify mandatory) incorrectly assumed that there were no
frontends in use that did not support this feature. But the frontend
driver in MiniOS does not and since this is used by (qemu) stubdoms,
these stopped working.Netback sort of works as-is in this mode except:
- If there are no Rx requests and the internal Rx queue fills, only
the drain timeout will wake the thread. The default drain timeout
of 10 s would give unacceptable pauses.- If an Rx stall was detected and the internal Rx queue is drained,
then the Rx thread would never wake.Handle these two cases (when feature-rx-notify is disabled) by:
- Reducing the drain timeout to 30 ms.
- Disabling Rx stall detection.
Reported-by: John
Tested-by: John
Signed-off-by: David Vrabel
Reviewed-by: Wei Liu
Signed-off-by: David S. Miller
11 Dec, 2014
1 commit
-
Conflicts:
drivers/net/ethernet/amd/xgbe/xgbe-desc.c
drivers/net/ethernet/renesas/sh_eth.cOverlapping changes in both conflict cases.
Signed-off-by: David S. Miller
10 Dec, 2014
1 commit
-
When xenvif_alloc() fails, it returns a non-NULL error indicator. To
avoid eventual races, we shouldn't store that into struct backend_info
as readers of it only check for NULL.Signed-off-by: Jan Beulich
Acked-by: Ian Campbell
Signed-off-by: David S. Miller
30 Nov, 2014
1 commit
25 Nov, 2014
1 commit
-
If xenvif_alloc() or xenbus_scanf() fail in backend_create_xenvif(),
xenbus is left in offline mode but netback_probe() reports success.The patch implements propagation of error code for backend_create_xenvif().
Found by Linux Driver Verification project (linuxtesting.org).
Signed-off-by: Alexey Khoroshilov
Acked-by: Wei Liu
Signed-off-by: David S. Miller
07 Nov, 2014
1 commit
-
Unconditionally pulling 128 bytes into the linear area is not required
for:- security: Every protocol demux starts with pskb_may_pull() to pull
frag data into the linear area, if necessary, before looking at
headers.- performance: Netback has already grant copied up-to 128 bytes from
the first slot of a packet into the linear area. The first slot
normally contain all the IPv4/IPv6 and TCP/UDP headers.The unconditional pull would often copy frag data unnecessarily. This
is a performance problem when running on a version of Xen where grant
unmap avoids TLB flushes for pages which are not accessed. TLB
flushes can now be avoided for > 99% of unmaps (it was 0% before).Grant unmap TLB flush avoidance will be available in a future version
of Xen (probably 4.6).Signed-off-by: Malcolm Crossley
Signed-off-by: David Vrabel
Acked-by: Ian Campbell
Signed-off-by: David S. Miller
02 Nov, 2014
1 commit
-
Conflicts:
drivers/net/phy/marvell.cSimple overlapping changes in drivers/net/phy/marvell.c
Signed-off-by: David S. Miller
30 Oct, 2014
2 commits
-
This flag is unnecessary, it came from some old code.
Suggested-by: Eric Dumazet
Signed-off-by: Zoltan Kiss
Signed-off-by: David Vrabel
Acked-by: Wei Liu
Signed-off-by: David S. Miller -
Otherwise the interrupt handler still calls napi_complete. Although it
won't schedule NAPI again as either NAPI_STATE_DISABLE or
NAPI_STATE_SCHED is set, it is just unnecessary, and it makes more
sense to do this way.Signed-off-by: Zoltan Kiss
Signed-off-by: David Vrabel
Acked-by: Wei Liu
Signed-off-by: David S. Miller
26 Oct, 2014
3 commits
-
If a frontend not receiving packets it is useful to detect this and
turn off the carrier so packets are dropped early instead of being
queued and drained when they expire.A to-guest queue is stalled if it doesn't have enough free slots for a
an extended period of time (default 60 s).If at least one queue is stalled, the carrier is turned off (in the
expectation that the other queues will soon stall as well). The
carrier is only turned on once all queues are ready.When the frontend connects, all the queues start in the stalled state
and only become ready once the frontend queues enough Rx requests.Signed-off-by: David Vrabel
Reviewed-by: Wei Liu
Signed-off-by: David S. Miller -
Netback needs to discard old to-guest skb's (guest Rx queue drain) and
it needs detect guest Rx stalls (to disable the carrier so packets are
discarded earlier), but the current implementation is very broken.1. The check in hard_start_xmit of the slot availability did not
consider the number of packets that were already in the guest Rx
queue. This could allow the queue to grow without bound.The guest stops consuming packets and the ring was allowed to fill
leaving S slot free. Netback queues a packet requiring more than S
slots (ensuring that the ring stays with S slots free). Netback
queue indefinately packets provided that then require S or fewer
slots.2. The Rx stall detection is not triggered in this case since the
(host) Tx queue is not stopped.3. If the Tx queue is stopped and a guest Rx interrupt occurs, netback
will consider this an Rx purge event which may result in it taking
the carrier down unnecessarily. It also considers a queue with
only 1 slot free as unstalled (even though the next packet might
not fit in this).The internal guest Rx queue is limited by a byte length (to 512 Kib,
enough for half the ring). The (host) Tx queue is stopped and started
based on this limit. This sets an upper bound on the amount of memory
used by packets on the internal queue.This allows the estimatation of the number of slots for an skb to be
removed (it wasn't a very good estimate anyway). Instead, the guest
Rx thread just waits for enough free slots for a maximum sized packet.skbs queued on the internal queue have an 'expires' time (set to the
current time plus the drain timeout). The guest Rx thread will detect
when the skb at the head of the queue has expired and discard expired
skbs. This sets a clear upper bound on the length of time an skb can
be queued for. For a guest being destroyed the maximum time needed to
wait for all the packets it sent to be dropped is still the drain
timeout (10 s) since it will not be sending new packets.Rx stall detection is reintroduced in a later commit.
Signed-off-by: David Vrabel
Reviewed-by: Wei Liu
Signed-off-by: David S. Miller -
Frontends that do not provide feature-rx-notify may stall because
netback depends on the notification from frontend to wake the guest Rx
thread (even if can_queue is false).This could be fixed but feature-rx-notify was introduced in 2006 and I
am not aware of any frontends that do not implement this.Signed-off-by: David Vrabel
Acked-by: Wei Liu
Signed-off-by: David S. Miller
06 Oct, 2014
1 commit
-
The DEFINE_XENBUS_DRIVER() macro looks a bit weird and causes sparse
errors.Replace the uses with standard structure definitions instead. This is
similar to pci and usb device registration.Signed-off-by: David Vrabel
26 Aug, 2014
1 commit
-
Interrupt is enabled when bind_interdomain_evtchn_to_irqhandler returns.
If there's interrupt pending interrupt handler is invoked.NAPI needs to be initialised before binding interrupt otherwise the
interrupt handler will try to scheduling a NAPI instance that is not
initialised yet, resulting in kernel OOPS.This fixes a regression introduced in ea2c5e13 ("xen-netback: move NAPI
add/remove calls").Ideally function calls to create kthreads should also be moved before
binding but I intent to fix this regression with minimal changes and
refactor the code with another patch.Reported-by: Thomas Leonard
Signed-off-by: Wei Liu
Cc: Ian Campbell
Signed-off-by: David S. Miller
14 Aug, 2014
4 commits
-
The original implementation relies on a loop to check if all inflight
packets are freed. Now we have proper reference counting, there's no
need to use loop anymore.Signed-off-by: Wei Liu
Cc: Ian Campbell
Cc: Zoltan Kiss
Signed-off-by: David S. Miller -
Reference count the number of packets in host stack, so that we don't
stop the deallocation thread too early. If not, we can end up with
xenvif_free permanently waiting for deallocation thread to unmap grefs.Reported-by: Thomas Leonard
Signed-off-by: Wei Liu
Cc: Ian Campbell
Cc: Zoltan Kiss
Signed-off-by: David S. Miller -
Originally netif_napi_add was in xenvif_init_queue and netif_napi_del
was in xenvif_deinit_queue, while kthreads were handled in
xenvif_connect and xenvif_disconnect. Move netif_napi_add and
netif_napi_del to xenvif_connect and xenvif_disconnect so that they
reside together with kthread operations.Signed-off-by: Wei Liu
Cc: Ian Campbell
Cc: Zoltan Kiss
Signed-off-by: David S. Miller -
The original code is bogus. The function gets called in a loop which
leaks entries created in previous rounds.Signed-off-by: Wei Liu
Cc: Zoltan Kiss
Cc: Ian Campbell
Signed-off-by: David S. Miller