- 29 Apr, 2015 9 commits
-
-
Jonathan Davies authored
[ Upstream commit 0c36820e ] xen-netfront limits transmitted skbs to be at most 44 segments in size. However, GSO permits up to 65536 bytes, which means a maximum of 45 segments of 1448 bytes each. This slight reduction in the size of packets means a slight loss in efficiency. Since c/s 9ecd1a75, xen-netfront sets gso_max_size to XEN_NETIF_MAX_TX_SIZE - MAX_TCP_HEADER, where XEN_NETIF_MAX_TX_SIZE is 65535 bytes. The calculation used by tcp_tso_autosize (and also tcp_xmit_size_goal since c/s 6c09fa09) in determining when to split an skb into two is sk->sk_gso_max_size - 1 - MAX_TCP_HEADER. So the maximum permitted size of an skb is calculated to be (XEN_NETIF_MAX_TX_SIZE - MAX_TCP_HEADER) - 1 - MAX_TCP_HEADER. Intuitively, this looks like the wrong formula -- we don't need two TCP headers. Instead, there is no need to deviate from the default gso_max_size of 65536 as this already accommodates the size of the header. Currently, the largest skb transmitted by netfront is 63712 bytes (44 segments of 1448 bytes each), as observed via tcpdump. This patch makes netfront send skbs of up to 65160 bytes (45 segments of 1448 bytes each). Similarly, the maximum allowable mtu does not need to subtract MAX_TCP_HEADER as it relates to the size of the whole packet, including the header. Fixes: 9ecd1a75 ("xen-netfront: reduce gso_max_size to account for max TCP header") Signed-off-by:
Jonathan Davies <jonathan.davies@citrix.com> Signed-off-by:
David S. Miller <davem@davemloft.net> Signed-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
Thomas Graf authored
[ Upstream commit fa2d8ff4 ] Return module reference before invoking the respective vport ->destroy() function. This is needed as ovs_vport_del() is not invoked inside an RCU read side critical section so the kfree can occur immediately before returning to ovs_vport_del(). Returning the module reference before ->destroy() is safe because the module unregistration is blocked on ovs_lock which we hold while destroying the datapath. Fixes: 62b9c8d0 ("ovs: Turn vports with dependencies into separate modules") Reported-by:
Pravin Shelar <pshelar@nicira.com> Signed-off-by:
Thomas Graf <tgraf@suug.ch> Acked-by:
-
Anton Nayshtut authored
[ Upstream commit f5e2dc5d ] Before commit 3900f290 ("bonding: slight optimizztion for bond_slave_override()") the override logic was to send packets with non-zero queue_id through the slave with corresponding queue_id, under two conditions only - if the slave can transmit and it's up. The above mentioned commit changed this logic by introducing an additional condition - whether the bond is active (indirectly, using the slave_can_tx and later - bond_is_active_slave), that prevents the user from implementing more complex policies according to the Documentation/networking/bonding.txt. Signed-off-by:
Anton Nayshtut <anton@swortex.com> Signed-off-by:
Alexey Bogoslavsky <alexey@swortex.com> Signed-off-by:
Andy Gospodarek <gospo@cumulusnetworks.com> Signed-off-by:
-
Alexey Kodanev authored
[ Upstream commit 4ad19de8 ] tcp_v6_fill_cb() will be called twice if socket's state changes from TCP_TIME_WAIT to TCP_LISTEN. That can result in control buffer data corruption because in the second tcp_v6_fill_cb() call it's not copying IP6CB(skb) anymore, but 'seq', 'end_seq', etc., so we can get weird and unpredictable results. Performance loss of up to 1200% has been observed in LTP/vxlan03 test. This can be fixed by copying inet6_skb_parm to the beginning of 'cb' only if xfrm6_policy_check() and tcp_v6_fill_cb() are going to be called again. Fixes: 2dc49d16 ("tcp6: don't move IP6CB before xfrm6_policy_check()") Signed-off-by:
Alexey Kodanev <alexey.kodanev@oracle.com> Acked-by:
Eric Dumazet <edumazet@google.com> Signed-off-by:
-
Alex Gartrell authored
[ Upstream commit 957f094f ] Instead of -1 with EAGAIN, read on a O_NONBLOCK tun fd will return 0. This fixes this by properly returning the error code from __skb_recv_datagram. Signed-off-by:
Alex Gartrell <agartrell@fb.com> Signed-off-by:
-
D.S. Ljungmark authored
[ Upstream commit 6fd99094 ] A local route may have a lower hop_limit set than global routes do. RFC 3756, Section 4.2.7, "Parameter Spoofing" > 1. The attacker includes a Current Hop Limit of one or another small > number which the attacker knows will cause legitimate packets to > be dropped before they reach their destination. > As an example, one possible approach to mitigate this threat is to > ignore very small hop limits. The nodes could implement a > configurable minimum hop limit, and ignore attempts to set it below > said limit. Signed-off-by:
D.S. Ljungmark <ljungmark@modio.se> Acked-by:
Hannes Frederic Sowa <hannes@stressinduktion.org> Signed-off-by:
-
Ido Shamay authored
[ Upstream commit e5eda89d ] Netdevice registration should be performed a the end of the driver initialization flow. If we don't do that, after calling register_netdevice, device callbacks may be issued by higher layers of the stack before final configuration of the device is done. For example (VXLAN configuration race), mlx4_SET_PORT_VXLAN was issued after the register_netdev command. System network scripts may configure the interface (UP) right after the registration, which also attach unicast VXLAN steering rule, before mlx4_SET_PORT_VXLAN was called, causing the firmware to fail the rule attachment. Fixes: 837052d0 ("net/mlx4_en: Add netdev support for TCP/IP offloads of vxlan tunneling") Signed-off-by:
Ido Shamay <idos@mellanox.com> Signed-off-by:
Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by:
-
Simon Horman authored
[ Upstream commit a6e95cc7 ] Master change notifications may occur other than when joining or leaving a bridge, for example when being added to or removed from a bond or Open vSwitch. Previously in those cases rocker_port_bridge_leave() was called which results in a null-pointer dereference as rocker_port->bridge_dev is NULL because there is no bridge device. This patch makes provision for doing nothing in such cases. Fixes: 6c707945 ("rocker: implement L2 bridge offloading") Acked-by:
Jiri Pirko <jiri@resnulli.us> Acked-by:
Scott Feldman <sfeldma@gmail.com> Signed-off-by:
Simon Horman <simon.horman@netronome.com> Signed-off-by:
-
Michal Kubeček authored
[ Upstream commit d0c294c5 ] On s390x, gcc 4.8 compiles this part of tcp_v6_early_demux() struct dst_entry *dst = sk->sk_rx_dst; if (dst) dst = dst_check(dst, inet6_sk(sk)->rx_dst_cookie); to code reading sk->sk_rx_dst twice, once for the test and once for the argument of ip6_dst_check() (dst_check() is inline). This allows ip6_dst_check() to be called with null first argument, causing a crash. Protect sk->sk_rx_dst access by READ_ONCE() both in IPv4 and IPv6 TCP early demux code. Fixes: 41063e9d ("ipv4: Early TCP socket demux.") Fixes: c7109986 ("ipv6: Early TCP socket demux") Signed-off-by:
Michal Kubecek <mkubecek@suse.cz> Acked-by:
-
- 19 Apr, 2015 31 commits
-
-
Greg Kroah-Hartman authored
-
Igor Mammedov authored
commit 74496134 upstream. KVM guest can fail to startup with following trace on host: qemu-system-x86: page allocation failure: order:4, mode:0x40d0 Call Trace: dump_stack+0x47/0x67 warn_alloc_failed+0xee/0x150 __alloc_pages_direct_compact+0x14a/0x150 __alloc_pages_nodemask+0x776/0xb80 alloc_kmem_pages+0x3a/0x110 kmalloc_order+0x13/0x50 kmemdup+0x1b/0x40 __kvm_set_memory_region+0x24a/0x9f0 [kvm] kvm_set_ioapic+0x130/0x130 [kvm] kvm_set_memory_region+0x21/0x40 [kvm] kvm_vm_ioctl+0x43f/0x750 [kvm] Failure happens when attempting to allocate pages for 'struct kvm_memslots', however it doesn't have to be present in physically contiguous (kmalloc-ed) address space, change allocation to kvm_kvzalloc() so that it will be vmalloc-ed when its size is more then a page. Signed-off-by:
Igor Mammedov <imammedo@redhat.com> Signed-off-by:
Marcelo Tosatti <mtosatti@redhat.com> Signed-off-by:
-
Daniel Vetter authored
commit f9b61ff6 upstream. It is platform/output depenedent when exactly the pipe will start running. Sometimes we just need the (cpu) pipe enabled, in other cases the pch transcoder is enough and in yet other cases the (DP) port is sending the frame start signal. In a perfect world we'd put the drm_crtc_vblank_on call exactly where the pipe starts running, but due to cloning and similar things this will get messy. And the current approach of picking the most conservative place for all combinations also doesn't work since that results in legit vblank waits (in encoder->enable hooks, e.g. the 2 vblank waits for sdvo) failing. Completely going back to the old world before commit 51e31d49 Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Mon Sep 15 12:36:02 2014 +0200 drm/i915: Use generic vblank wait isn't great either since screaming when the vblank wait work because the pipe is off is kinda nice. Pick a compromise and move the drm_crtc_vblank_on right before the encoder->enable call. This is a lie on some outputs/platforms, but after the ->enable callback the pipe is guaranteed to run everywhere. So not that bad really. Suggested by Ville. v2: Same treatment for drm_crtc_vblank_off and encoder->disable: I've missed the ibx pipe B select w/a, which also has a vblank wait in the disable function (while the pipe is obviously still running). Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Acked-by:
Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by:
Daniel Vetter <daniel.vetter@intel.com> Cc: Jani Nikula <jani.nikula@intel.com> Signed-off-by:
-
Dave Chinner authored
commit 5885ebda upstream. A new fsync vs power fail test in xfstests indicated that XFS can have unreliable data consistency when doing extending truncates that require block zeroing. The blocks beyond EOF get zeroed in memory, but we never force those changes to disk before we run the transaction that extends the file size and exposes those blocks to userspace. This can result in the blocks not being correctly zeroed after a crash. Because in-memory behaviour is correct, tools like fsx don't pick up any coherency problems - it's not until the filesystem is shutdown or the system crashes after writing the truncate transaction to the journal but before the zeroed data in the page cache is flushed that the issue is exposed. Fix this by also flushing the dirty data in memory region between the old size and new size when we've found blocks that need zeroing in the truncate process. Reported-by:
Liu Bo <bo.li.liu@oracle.com> Signed-off-by:
Dave Chinner <dchinner@redhat.com> Reviewed-by:
Brian Foster <bfoster@redhat.com> Signed-off-by:
Dave Chinner <david@fromorbit.com> Signed-off-by:
-
Omar Sandoval authored
commit 6f30b7e3 upstream. Commit 4f579ae7 (ext4: fix punch hole on files with indirect mapping) rewrote FALLOC_FL_PUNCH_HOLE for ext4 files with indirect mapping. However, there are bugs in several corner cases. This fixes 5 distinct bugs: 1. When there is at least one entire level of indirection between the start and end of the punch range and the end of the punch range is the first block of its level, we can't return early; we have to free the intervening levels. 2. When the end is at a higher level of indirection than the start and ext4_find_shared returns a top branch for the end, we still need to free the rest of the shared branch it returns; we can't decrement partial2. 3. When a punch happens within one level of indirection, we need to converge on an indirect block that contains the start and end. However, because the branches returned from ext4_find_shared do not necessarily start at the same level (e.g., the partial2 chain will be shallower if the last block occurs at the beginning of an indirect group), the walk of the two chains can end up "missing" each other and freeing a bunch of extra blocks in the process. This mismatch can be handled by first making sure that the chains are at the same level, then walking them together until they converge. 4. When the punch happens within one level of indirection and ext4_find_shared returns a top branch for the start, we must free it, but only if the end does not occur within that branch. 5. When the punch happens within one level of indirection and ext4_find_shared returns a top branch for the end, then we shouldn't free the block referenced by the end of the returned chain (this mirrors the different levels case). Signed-off-by:
Omar Sandoval <osandov@osandov.com> Cc: Chris J Arges <chris.j.arges@canonical.com> Signed-off-by:
-
Preeti U Murthy authored
commit a127d2bc upstream. The hrtimer mode of broadcast queues hrtimers in the idle entry path so as to wakeup cpus in deep idle states. The associated call graph is : cpuidle_idle_call() |____ clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_ENTER, ....)) |_____tick_broadcast_set_event() |____clockevents_program_event() |____bc_set_next() The hrtimer_{start/cancel} functions call into tracing which uses RCU. But it is not legal to call into RCU in cpuidle because it is one of the quiescent states. Hence protect this region with RCU_NONIDLE which informs RCU that the cpu is momentarily non-idle. As an aside it is helpful to point out that the clock event device that is programmed here is not a per-cpu clock device; it is a pseudo clock device, used by the broadcast framework alone. The per-cpu clock device programming never goes through bc_set_next(). Signed-off-by:
Preeti U Murthy <preeti@linux.vnet.ibm.com> Signed-off-by:
Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by:
Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: linuxppc-dev@ozlabs.org Cc: mpe@ellerman.id.au Cc: tglx@linutronix.de Link: http://lkml.kernel.org/r/20150318104705.17763.56668.stgit@preeti.in.ibm.comSigned-off-by:
Ingo Molnar <mingo@kernel.org> Signed-off-by:
-
Majd Dibbiny authored
commit 61a3855b upstream. For RoCE ports, we set the u32 PMA values based on u64 HCA counters. In case of overflow, according to the IB spec, we have to saturate a counter to its max value, do that. Fixes: c3779134 ('IB/mlx4: Support PMA counters for IBoE') Signed-off-by:
Majd Dibbiny <majd@mellanox.com> Signed-off-by:
Eran Ben Elisha <eranbe@mellanox.com> Signed-off-by:
Hadar Hen Zion <hadarh@mellanox.com> Signed-off-by:
-
Uwe Kleine-König authored
commit da321133 upstream. The rate provided at the output of a clk-divider is calculated as: DIV_ROUND_UP(parent_rate, div) since commit b11d282d (clk: divider: fix rate calculation for fractional rates). So to yield a rate not bigger than r parent_rate must be <= r * div. The effect of choosing a parent rate that is too big as was done before this patch results in wrongly ruling out good dividers. Note that this is not a complete fix as __clk_round_rate might return a value >= its 2nd parameter. Also for dividers with CLK_DIVIDER_ROUND_CLOSEST set the calculation is not accurate. But this fixes the test case by Sascha Hauer that uses a chain of three dividers under a fixed clock. Fixes: b11d282d (clk: divider: fix rate calculation for fractional rates) Suggested-by:
Sascha Hauer <s.hauer@pengutronix.de> Signed-off-by:
Uwe Kleine-König <u.kleine-koenig@pengutronix.de> Acked-by:
Michael Turquette <mturquette@linaro.org> Signed-off-by:
-
Uwe Kleine-König authored
commit 26bac95a upstream. It's an invalid approach to assume that among two divider values the one nearer the exact divider is the better one. Assume a parent rate of 1000 Hz, a divider with CLK_DIVIDER_POWER_OF_TWO and a target rate of 89 Hz. The exact divider is ~ 11.236 so 8 and 16 are the candidates to choose from yielding rates 125 Hz and 62.5 Hz respectivly. While 8 is nearer to 11.236 than 16 is, the latter is still the better divider as 62.5 is nearer to 89 than 125 is. Fixes: 774b5143 (clk: divider: Add round to closest divider) Signed-off-by:
Maxime Coquelin <maxime.coquelin@st.com> Signed-off-by:
-
Hans Verkuil authored
commit 0e661006 upstream. Stopping the vb2 thread (as used by several DVB devices) can result in an 'UNBALANCED' warning such as this: vb2: counters for queue ffff880407ee9828: UNBALANCED! vb2: setup: 1 start_streaming: 1 stop_streaming: 1 vb2: wait_prepare: 249333 wait_finish: 249334 This is due to a race condition between stopping the thread and calling vb2_internal_streamoff(). While I have not been able to deduce the exact mechanism how this race condition can produce this warning, I can see that the way the stream is stopped is likely to lead to a race somewhere. This patch simplifies how this is done by first ensuring that the thread is completely stopped before cleaning up the vb2 queue. It does that by setting threadio->stop to true, followed by a call to vb2_queue_error() which will wake up the thread. The thread sees that 'stop' is true and it will exit. The call to kthread_stop() waits until the thread has exited, and only then is the queue cleaned up by calling __vb2_cleanup_fileio(). This is a much cleaner sequence and the warning has now disappeared. Reported-by:
Jurgen Kramer <gtmkramer@xs4all.nl> Tested-by:
Hans Verkuil <hans.verkuil@cisco.com> Signed-off-by:
Mauro Carvalho Chehab <mchehab@osg.samsung.com> Signed-off-by:
-
Sakari Ailus authored
commit 4879785e upstream. The last argument of vb2_dc_get_user_pages() is of type enum dma_data_direction, but the caller, vb2_dc_get_userptr() passes a value which is the result of comparison dma_dir == DMA_FROM_DEVICE. This results in the write parameter to get_user_pages() being zero in all cases, i.e. that the caller has no intent to write there. This was broken by patch "vb2: replace 'write' by 'dma_dir'". Signed-off-by:
Sakari Ailus <sakari.ailus@linux.intel.com> Acked-by:
-
Geert Uytterhoeven authored
commit 8e48a2d5 upstream. Unlike scan_async_group(), soc_of_bind() doesn't allocate its soc_camera_async_client structure using devm_kzalloc(), but has it embedded inside the soc_of_info structure. Hence on failure, it must free the whole soc_of_info structure, and not just the embedded soc_camera_async_client structure, as the latter causes a warning, and may cause slab corruption: soc-camera-pdrv soc-camera-pdrv.0: Probing soc-camera-pdrv.0 ------------[ cut here ]------------ WARNING: CPU: 0 PID: 1 at drivers/base/devres.c:887 devm_kfree+0x30/0x40() CPU: 0 PID: 1 Comm: swapper/0 Not tainted 3.19.0-shmobile-08386-g37feb0d093cb2d8e #128 Hardware name: Generic R8A7791 (Flattened Device Tree) Backtrace: [<c0011e7c>] (dump_backtrace) from [<c0012024>] (show_stack+0x18/0x1c) r6:c05a923b r5:00000009 r4:00000000 r3:00204140 [<c001200c>] (show_stack) from [<c048ed30>] (dump_stack+0x78/0x94) [<c048ecb8>] (dump_stack) from [<c002687c>] (warn_slowpath_common+0x8c/0xb8) r4:00000000 r3:00000000 [<c00267f0>] (warn_slowpath_common) from [<c0026980>] (warn_slowpath_null+0x24/0x2c) r8:ee7d8214 r7:ed83b810 r6:ed83bc20 r5:fffffffa r4:ed83e510 [<c002695c>] (warn_slowpath_null) from [<c025e0cc>] (devm_kfree+0x30/0x40) [<c025e09c>] (devm_kfree) from [<c032bbf4>] (soc_of_bind.isra.14+0x194/0x1d4) [<c032ba60>] (soc_of_bind.isra.14) from [<c032c6b8>] (soc_camera_host_register+0x208/0x31c) r9:00000070 r8:ee7e05d0 r7:ee153210 r6:00000000 r5:ee7e0218 r4:ed83bc20 [<c032c4b0>] (soc_camera_host_register) from [<c032e80c>] (rcar_vin_probe+0x1f4/0x238) r8:ee153200 r7:00000008 r6:ee153210 r5:ed83bc10 r4:c066319c r3:000000c0 [<c032e618>] (rcar_vin_probe) from [<c025c334>] (platform_drv_probe+0x50/0xa0) r10:00000000 r9:c0662fa8 r8:00000000 r7:c06a3700 r6:c0662fa8 r5:ee153210 r4:00000000 [<c025c2e4>] (platform_drv_probe) from [<c025af08>] (driver_probe_device+0xc4/0x208) r6:c06a36f4 r5:00000000 r4:ee153210 r3:c025c2e4 [<c025ae44>] (driver_probe_device) from [<c025b108>] (__driver_attach+0x70/0x94) r9:c066f9c0 r8:c0624a98 r7:c065b790 r6:c0662fa8 r5:ee153244 r4:ee153210 [<c025b098>] (__driver_attach) from [<c025984c>] (bus_for_each_dev+0x74/0x98) r6:c025b098 r5:c0662fa8 r4:00000000 r3:00000001 [<c02597d8>] (bus_for_each_dev) from [<c025b1dc>] (driver_attach+0x20/0x28) r6:ed83c200 r5:00000000 r4:c0662fa8 [<c025b1bc>] (driver_attach) from [<c025a00c>] (bus_add_driver+0xdc/0x1c4) [<c0259f30>] (bus_add_driver) from [<c025b8f4>] (driver_register+0xa4/0xe8) r7:c0624a98 r6:00000000 r5:c060b010 r4:c0662fa8 [<c025b850>] (driver_register) from [<c025ccd0>] (__platform_driver_register+0x50/0x64) r5:c060b010 r4:ed8394c0 [<c025cc80>] (__platform_driver_register) from [<c060b028>] (rcar_vin_driver_init+0x18/0x20) [<c060b010>] (rcar_vin_driver_init) from [<c05edde8>] (do_one_initcall+0x108/0x1b8) [<c05edce0>] (do_one_initcall) from [<c05edfb4>] (kernel_init_freeable+0x11c/0x1e4) r9:c066f9c0 r8:c066f9c0 r7:c062eab0 r6:c06252c4 r5:000000ad r4:00000006 [<c05ede98>] (kernel_init_freeable) from [<c048c3d0>] (kernel_init+0x10/0xec) r9:00000000 r8:00000000 r7:00000000 r6:00000000 r5:c048c3c0 r4:00000000 [<c048c3c0>] (kernel_init) from [<c000eba0>] (ret_from_fork+0x14/0x34) r4:00000000 r3:ee04e000 ---[ end trace e3a984cc0335c8a0 ]--- rcar_vin e6ef1000.video: group probe failed: -6 Fixes: 1ddc6a6c ("[media] soc_camera: add support for dt binding soc_camera drivers") Signed-off-by:
Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by:
Guennadi Liakhovetski <g.liakhovetski@gmx.de> Signed-off-by:
-
Hans Verkuil authored
commit 6b46211f upstream. cap->device_caps wasn't set in cx23885-417.c causing a warning from the v4l2-core. Reported-by:
Joseph Jasi <joe.yasi@gmail.com> Signed-off-by:
-
Marek Szyprowski authored
commit 05b676ab upstream. TASK_SIZE is depends on the systems architecture (32 or 64 bits) and it should not be used for defining offset boundary for mmaping buffers for CAPTURE and OUTPUT queues. This patch fixes support for MMAP calls on the CAPTURE queue on 64bit architectures (like ARM64). Signed-off-by:
Marek Szyprowski <m.szyprowski@samsung.com> Signed-off-by:
Kamil Debski <k.debski@samsung.com> Signed-off-by:
-
Hans Verkuil authored
commit ab312030 upstream. The v4l2_dev field of struct video_device must be set correctly. This was never done for this driver, so no video nodes were created anymore. Signed-off-by:
-
Mike Christie authored
commit b815fc12 upstream. This fixes a oops due to a double list add when adding a reject PDU for iscsit_allocate_iovecs allocation failures. The cmd has already been added to the conn_cmd_list in iscsit_setup_scsi_cmd, so this has us call iscsit_reject_cmd. Note that for ERL0 the reject PDU is not actually sent, so this patch is not completely tested. Just verified we do not oops. The problem is the add reject functions return -1 which is returned all the way up to iscsi_target_rx_thread which for ERL0 will drop the connection. Signed-off-by:
Mike Christie <michaelc@cs.wisc.edu> Signed-off-by:
Nicholas Bellinger <nab@linux-iscsi.org> Signed-off-by:
-
Al Viro authored
commit deeb8525 upstream. If we fail past the aio_setup_ring(), we need to destroy the mapping. We don't need to care about anybody having found ctx, or added requests to it, since the last failure exit is exactly the failure to make ctx visible to lookups. Reproducer (based on one by Joe Mario <jmario@redhat.com>): void count(char *p) { char s[80]; printf("%s: ", p); fflush(stdout); sprintf(s, "/bin/cat /proc/%d/maps|/bin/fgrep -c '/[aio] (deleted)'", getpid()); system(s); } int main() { io_context_t *ctx; int created, limit, i, destroyed; FILE *f; count("before"); if ((f = fopen("/proc/sys/fs/aio-max-nr", "r")) == NULL) perror("opening aio-max-nr"); else if (fscanf(f, "%d", &limit) != 1) fprintf(stderr, "can't parse aio-max-nr\n"); else if ((ctx = calloc(limit, sizeof(io_context_t))) == NULL) perror("allocating aio_context_t array"); else { for (i = 0, created = 0; i < limit; i++) { if (io_setup(1000, ctx + created) == 0) created++; } for (i = 0, destroyed = 0; i < created; i++) if (io_destroy(ctx[i]) == 0) destroyed++; printf("created %d, failed %d, destroyed %d\n", created, limit - created, destroyed); count("after"); } } Found-by:
Joe Mario <jmario@redhat.com> Signed-off-by:
Al Viro <viro@zeniv.linux.org.uk> Signed-off-by:
-
Al Viro authored
commit 64b4e252 upstream. "ocfs2 syncs the wrong range" had been broken; prior to it the code was doing the wrong thing in case of O_APPEND, all right, but _after_ it we were syncing the wrong range in 100% cases. *ppos, aka iocb->ki_pos is incremented prior to that point, so we are always doing sync on the area _after_ the one we'd written to. Spotted by Joseph Qi <joseph.qi@huawei.com> back in January; unfortunately, I'd missed his mail back then ;-/ Signed-off-by:
-
John Soni Jose authored
commit 2e7cee02 upstream. Kernel panic was happening as iscsi_host_remove() was called on a host which was not yet added. Signed-off-by:
John Soni Jose <sony.john-n@emulex.com> Reviewed-by:
James Bottomley <JBottomley@Odin.com> Signed-off-by:
-
Ilya Dryomov authored
commit 6d7fdb0a upstream. This reverts commit 89baaa57. Dirty page throttling should be sufficient for us in the general case so there is no need to use __GFP_MEMALLOC - it would be needed only in the swap-over-rbd case, which we currently don't support. (It would probably take approximately the commit that is being reverted to add that support, but we would also need the "swap" option to distinguish from the general case and make sure swap ceph_client-s aren't shared with anything else.) See ceph-devel threads [1] and [2] for the details of why enabling pfmemalloc reserves for all cases is a bad thing. On top of potential system lockups related to drained emergency reserves, this turned out to cause ceph lockups in case peers are on the same host and communicating via loopback due to sk_filter() dropping pfmemalloc skbs on the receiving side because the receiving loopback socket is not tagged with SOCK_MEMALLOC. [1] "SOCK_MEMALLOC vs loopback" http://www.spinics.net/lists/ceph-devel/msg22998.html [2] "[PATCH] libceph: don't set memalloc flags in loopback case" http://www.spinics.net/lists/ceph-devel/msg23392.html Conflicts: net/ceph/messenger.c [ context: tcp_nodelay option ] Cc: Mike Christie <michaelc@cs.wisc.edu> Cc: Mel Gorman <mgorman@suse.de> Cc: Sage Weil <sage@redhat.com> Signed-off-by:
Ilya Dryomov <idryomov@gmail.com> Acked-by:
Mel Gorman <mgorman@suse.de> Signed-off-by:
-
Rafael J. Wysocki authored
commit f82daee4 upstream. Commit 84c91b7a (PM / hibernate: avoid unsafe pages in e820 reserved regions) is reported to make resume from hibernation on Lenovo x230 unreliable, so revert it. We will revisit the issue the commit in question was supposed to fix in the future. Link: https://bugzilla.kernel.org/show_bug.cgi?id=96111Reported-by:
rhn <kebuac.rhn@porcupinefactory.org> Signed-off-by:
Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by:
-
Benjamin Herrenschmidt authored
commit 41d94893 upstream. The "sdc" node is missing the ranges property, it needs to be treated as having an empty one otherwise translation fails for its children. Fixes 746c9e9f, "of/base: Fix PowerPC address parsing hack" Tested-by:
Steven Rostedt <rostedt@goodmis.org> Signed-off-by:
Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by:
Grant Likely <grant.likely@linaro.org> Signed-off-by:
-
Larry Finger authored
commit be0b5e63 upstream. Transmission of an AP beacon does not call the TX interrupt service routine, which usually does the cleanup. Instead, cleanup is handled in a tasklet completion routine. Unfortunately, this routine has a serious bug in that it does not release the DMA mapping before it frees the skb, thus one IOMMU mapping is leaked for each beacon. The test system failed with no free IOMMU mapping slots approximately one hour after hostapd was used to start an AP. This issue was reported and tested at https://github.com/lwfinger/rtlwifi_new/issues/30. Reported-and-tested-by:
Kevin Mullican <kevin@mullican.com> Cc: Kevin Mullican <kevin@mullican.com> Signed-off-by:
Shao Fu <shaofu@realtek.com> Signed-off-by:
Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by:
Kalle Valo <kvalo@codeaurora.org> Signed-off-by:
-
Alex Williamson authored
commit 71684406 upstream. Device domains never span IOMMU hardware units, which allows the domain ID space for each IOMMU to be an independent address space. Therefore we can have multiple, independent domains, each with the same domain->id, but attached to different hardware units. This is also why we need to do a heavy-weight search for VM domains since they can span multiple IOMMUs hardware units and we don't require a single global ID to use for all hardware units. Therefore, if we call iommu_detach_domain() across all active IOMMU hardware units for a non-VM domain, the result is that we clear domain IDs that are not associated with our domain, allowing them to be re-allocated and causing apparent coherency issues when the device cannot access IOVAs for the intended domain. This bug was introduced in commit fb170fb4 ("iommu/vt-d: Introduce helper functions to make code symmetric for readability"), but is significantly exacerbated by the more recent commit 62c22167 ("iommu/vt-d: Fix dmar_domain leak in iommu_attach_device") which calls domain_exit() more frequently to resolve a domain leak. Fixes: fb170fb4 ("iommu/vt-d: Introduce helper functions to make code symmetric for readability") Signed-off-by:
Alex Williamson <alex.williamson@redhat.com> Cc: Jiang Liu <jiang.liu@linux.intel.com> Signed-off-by:
Joerg Roedel <jroedel@suse.de> Signed-off-by:
-
Arend van Spriel authored
commit f93a25b3 upstream. The BCM43362 firmware falsely reports it is capable of providing MBSS. As a result AP mode no longer works for this device. Therefor disable MBSS in the driver for this chipset. Reported-by:
Jorg Krause <jkrause@posteo.de> Reviewed-by:
Hante Meuleman <meuleman@broadcom.com> Reviewed-by:
Pieter-Paul Giesberts <pieterpg@broadcom.com> Signed-off-by:
Arend van Spriel <arend@broadcom.com> Signed-off-by:
-
David Disseldorp authored
commit e1e9bda2 upstream. Under intermittent network outages, find_writable_file() is susceptible to the following race condition, which results in a user-after-free in the cifs_writepages code-path: Thread 1 Thread 2 ======== ======== inv_file = NULL refind = 0 spin_lock(&cifs_file_list_lock) // invalidHandle found on openFileList inv_file = open_file // inv_file->count currently 1 cifsFileInfo_get(inv_file) // inv_file->count = 2 spin_unlock(&cifs_file_list_lock); cifs_reopen_file() cifs_close() // fails (rc != 0) ->cifsFileInfo_put() spin_lock(&cifs_file_list_lock) // inv_file->count = 1 spin_unlock(&cifs_file_list_lock) spin_lock(&cifs_file_list_lock); list_move_tail(&inv_file->flist, &cifs_inode->openFileList); spin_unlock(&cifs_file_list_lock); cifsFileInfo_put(inv_file); ->spin_lock(&cifs_file_list_lock) // inv_file->count = 0 list_del(&cifs_file->flist); // cleanup!! kfree(cifs_file); spin_unlock(&cifs_file_list_lock); spin_lock(&cifs_file_list_lock); ++refind; // refind = 1 goto refind_writable; At this point we loop back through with an invalid inv_file pointer and a refind value of 1. On second pass, inv_file is not overwritten on openFileList traversal, and is subsequently dereferenced. Signed-off-by:
David Disseldorp <ddiss@suse.de> Reviewed-by:
Jeff Layton <jlayton@samba.org> Signed-off-by:
Steve French <smfrench@gmail.com> Signed-off-by:
-
Sachin Prabhu authored
commit 2477bc58 upstream. While attempting to clone a file on a samba server, we receive a STATUS_INVALID_DEVICE_REQUEST. This is mapped to -EOPNOTSUPP which isn't handled in smb2_clone_range(). We end up looping in the while loop making same call to the samba server over and over again. The proposed fix is to exit and return the error value when encountered with an unhandled error. Signed-off-by:
Sachin Prabhu <sprabhu@redhat.com> Signed-off-by:
Steve French <steve.french@primarydata.com> Signed-off-by:
-
Peter Hurley authored
commit fb5ef9e7 upstream. In canon mode, the read buffer head will advance over the buffer tail if the input > 4095 bytes without receiving a line termination char. Discard additional input until a line termination is received. Before evaluating for overflow, the 'room' value is normalized for I_PARMRK and 1 byte is reserved for line termination (even in !icanon mode, in case the mode is switched). The following table shows the transform: actual buffer | 'room' value before overflow calc space avail | !I_PARMRK | I_PARMRK -------------------------------------------------- 0 | -1 | -1 1 | 0 | 0 2 | 1 | 0 3 | 2 | 0 4+ | 3 | 1 When !icanon or when icanon and the read buffer contains newlines, normalized 'room' values of -1 and 0 are clamped to 0, and 'overflow' is 0, so read_head is not adjusted and the input i/o loop exits (setting no_room if called from flush_to_ldisc()). No input is discarded since the reader does have input available to read which ensures forward progress. When icanon and the read buffer does not contain newlines and the normalized 'room' value is 0, then overflow and room are reset to 1, so that the i/o loop will process the next input char normally (except for parity errors which are ignored). Thus, erasures, signalling chars, 7-bit mode, etc. will continue to be handled properly. If the input char processed was not a line termination char, then the canon_head index will not have advanced, so the normalized 'room' value will now be -1 and 'overflow' will be set, which indicates the read_head can safely be reset, effectively erasing the last char processed. If the input char processed was a line termination, then the canon_head index will have advanced, so 'overflow' is cleared to 0, the read_head is not reset, and 'room' is cleared to 0, which exits the i/o loop (because the reader now have input available to read which ensures forward progress). Note that it is possible for a line termination to be received, and for the reader to copy the line to the user buffer before the input i/o loop is ready to process the next input char. This is why the i/o loop recomputes the room/overflow state with every input char while handling overflow. Finally, if the input data was processed without receiving a line termination (so that overflow is still set), the pty driver must receive a write wakeup. A pty writer may be waiting to write more data in n_tty_write() but without unthrottling here that wakeup will not arrive, and forward progress will halt. (Normally, the pty writer is woken when the reader reads data out of the buffer and more space become available). Signed-off-by:
Peter Hurley <peter@hurleysoftware.com> Signed-off-by:
Joseph Salisbury <joseph.salisbury@canonical.com>
-
Stefan Agner authored
commit 8e4934c6 upstream. When the receiver was enabled during startup, a character could have been in the FIFO when the UART get initially used. The driver configures the (receive) watermark level, and flushes the FIFO. However, the receive flag (RDRF) could still be set at that stage (as mentioned in the register description of UARTx_RWFIFO). This leads to an interrupt which won't be handled properly in interrupt mode: The receive interrupt function lpuart_rxint checks the FIFO count, which is 0 at that point (due to the flush during initialization). The problem does not manifest when using DMA to receive characters. Fix this situation by explicitly read the status register, which leads to clearing of the RDRF flag. Due to the flush just after the status flag read, a explicit data read is not to required. Signed-off-by:
Stefan Agner <stefan@agner.ch> Signed-off-by:
-
Stefan Agner authored
commit 4e8f2459 upstream. Specify transmit FIFO size which might be different depending on LPUART instance. This makes sure uart_wait_until_sent in serial core getting called, which in turn waits and checks if the FIFO is really empty on shutdown by using the tx_empty callback. Without the call of this callback, the last several characters might not yet be transmitted when closing the serial port. This can be reproduced by simply using echo and redirect the output to a ttyLP device. Signed-off-by:
-
Lu Baolu authored
commit 227a4fd8 upstream. When a device with an isochronous endpoint is plugged into the Intel xHCI host controller, and the driver submits multiple frames per URB, the xHCI driver will set the Block Event Interrupt (BEI) flag on all but the last TD for the URB. This causes the host controller to place an event on the event ring, but not send an interrupt. When the last TD for the URB completes, BEI is cleared, and we get an interrupt for the whole URB. However, under Intel xHCI host controllers, if the event ring is full of events from transfers with BEI set, an "Event Ring is Full" event will be posted to the last entry of the event ring, but no interrupt is generated. Host will cease all transfer and command executions and wait until software completes handling the pending events in the event ring. That means xHC stops, but event of "event ring is full" is not notified. As the result, the xHC looks like dead to user. This patch is to apply XHCI_AVOID_BEI quirk to Intel xHC devices. And it should be backported to kernels as old as 3.0, that contains the commit 69e848c2 ("Intel xhci: Support EHCI/xHCI port switching."). Signed-off-by:
Lu Baolu <baolu.lu@linux.intel.com> Tested-by:
Alistair Grant <akgrant0710@gmail.com> Signed-off-by:
Mathias Nyman <mathias.nyman@linux.intel.com> Signed-off-by:
-
