linux-toradex.git/arch/sparc, branch v3.12.18

bpf: do not use reciprocal divide

2014-02-06T19:22:20+00:00

[ Upstream commit aee636c4809fa54848ff07a899b326eb1f9987a2 ]

At first Jakub Zawadzki noticed that some divisions by reciprocal_divide
were not correct. (off by one in some cases)
http://www.wireshark.org/~darkjames/reciprocal-buggy.c

He could also show this with BPF:
http://www.wireshark.org/~darkjames/set-and-dump-filter-k-bug.c

The reciprocal divide in linux kernel is not generic enough,
lets remove its use in BPF, as it is not worth the pain with
current cpus.

Signed-off-by: Eric Dumazet 
Reported-by: Jakub Zawadzki 
Cc: Mircea Gherzan 
Cc: Daniel Borkmann 
Cc: Hannes Frederic Sowa 
Cc: Matt Evans 
Cc: Martin Schwidefsky 
Cc: Heiko Carstens 
Cc: David S. Miller 
Signed-off-by: David S. Miller 
Signed-off-by: Greg Kroah-Hartman

mm: fix TLB flush race between migration, and change_protection_range

2014-01-09T20:25:14+00:00

commit 20841405940e7be0617612d521e206e4b6b325db upstream.

There are a few subtle races, between change_protection_range (used by
mprotect and change_prot_numa) on one side, and NUMA page migration and
compaction on the other side.

The basic race is that there is a time window between when the PTE gets
made non-present (PROT_NONE or NUMA), and the TLB is flushed.

During that time, a CPU may continue writing to the page.

This is fine most of the time, however compaction or the NUMA migration
code may come in, and migrate the page away.

When that happens, the CPU may continue writing, through the cached
translation, to what is no longer the current memory location of the
process.

This only affects x86, which has a somewhat optimistic pte_accessible.
All other architectures appear to be safe, and will either always flush,
or flush whenever there is a valid mapping, even with no permissions
(SPARC).

The basic race looks like this:

CPU A			CPU B			CPU C

						load TLB entry
make entry PTE/PMD_NUMA
			fault on entry
						read/write old page
			start migrating page
			change PTE/PMD to new page
						read/write old page [*]
flush TLB
						reload TLB from new entry
						read/write new page
						lose data

[*] the old page may belong to a new user at this point!

The obvious fix is to flush remote TLB entries, by making sure that
pte_accessible aware of the fact that PROT_NONE and PROT_NUMA memory may
still be accessible if there is a TLB flush pending for the mm.

This should fix both NUMA migration and compaction.

[mgorman@suse.de: fix build]
Signed-off-by: Rik van Riel 
Signed-off-by: Mel Gorman 
Cc: Alex Thorlton 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net

2013-10-23T06:47:42+00:00

Pull networking fixes from David Miller:
 "Sorry I let so much accumulate, I was in Buffalo and wanted a few
  things to cook in my tree for a while before sending to you.  Anyways,
  it's a lot of little things as usual at this stage in the game"

 1) Make bonding MAINTAINERS entry reflect reality, from Andy
    Gospodarek.

 2) Fix accidental sock_put() on timewait mini sockets, from Eric
    Dumazet.

 3) Fix crashes in l2tp due to mis-handling of ipv4 mapped ipv6
    addresses, from François CACHEREUL.

 4) Fix heap overflow in __audit_sockaddr(), from the eagle eyed Dan
    Carpenter.

 5) tcp_shifted_skb() doesn't take handle FINs properly, from Eric
    Dumazet.

 6) SFC driver bug fixes from Ben Hutchings.

 7) Fix TX packet scheduling wedge after channel change in ath9k driver,
    from Felix Fietkau.

 8) Fix user after free in BPF JIT code, from Alexei Starovoitov.

 9) Source address selection test is reversed in
    __ip_route_output_key(), fix from Jiri Benc.

10) VLAN and CAN layer mis-size netlink attributes, from Marc
    Kleine-Budde.

11) Fix permission checks in sysctls to use current_euid() instead of
    current_uid().  From Eric W Biederman.

12) IPSEC policies can go away while a timer is still pending for them,
    add appropriate ref-counting to fix, from Steffen Klassert.

13) Fix mis-programming of FDR and RMCR registers on R8A7740 sh_eth
    chips, from Nguyen Hong Ky and Simon Horman.

14) MLX4 forgets to DMA unmap pages on RX, fix from Amir Vadai.

15) IPV6 GRE tunnel MTU upper limit is miscalculated, from Oussama
    Ghorbel.

16) Fix typo in fq_change(), we were assigning "initial quantum" to
    "quantum".  From Eric Dumazet.

17) Set a more appropriate sk_pacing_rate for non-TCP sockets, otherwise
    FQ packet scheduler does not pace those flows properly.  Also from
    Eric Dumazet.

18) rtlwifi miscalculates packet pointers, from Mark Cave-Ayland.

19) l2tp_xmit_skb() can be called from process context, not just softirq
    context, so we must always make sure to BH disable around it.  From
    Eric Dumazet.

20) On qdisc reset, we forget to purge the RB tree of SKBs in netem
    packet scheduler.  From Stephen Hemminger.

21) Fix info leak in farsync WAN driver ioctl() handler, from Dan
    Carpenter and Salva Peiró.

22) Fix PHY reset and other issues in dm9000 driver, from Nikita
    Kiryanov and Michael Abbott.

23) When hardware can do SCTP crc32 checksums, we accidently don't
    disable the csum offload when IPSEC transformations have been
    applied.  From Fan Du and Vlad Yasevich.

24) Tail loss probing in TCP leaves the socket in the wrong congestion
    avoidance state.  From Yuchung Cheng.

25) In CPSW driver, enable NAPI before interrupts are turned on, from
    Markus Pargmann.

26) Integer underflow and dual-assignment in YAM hamradio driver, from
    Dan Carpenter.

27) If we are going to mangle a packet in tcp_set_skb_tso_segs() we must
    unclone it.  This fixes various hard to track down crashes in
    drivers where the SKBs ->gso_segs was changing right from underneath
    the driver during TX queueing.  From Eric Dumazet.

28) Fix the handling of VLAN IDs, and in particular the special IDs 0
    and 4095, in the bridging layer.  From Toshiaki Makita.

29) Another info leak, this time in wanxl WAN driver, from Salva Peiró.

30) Fix race in socket credential passing, from Daniel Borkmann.

31) WHen NETLABEL is disabled, we don't validate CIPSO packets properly,
    from Seif Mazareeb.

32) Fix identification of fragmented frames in ipv4/ipv6 UDP
    Fragmentation Offload output paths, from Jiri Pirko.

33) Virtual Function fixes in bnx2x driver from Yuval Mintz and Ariel
    Elior.

34) When we removed the explicit neighbour pointer from ipv6 routes a
    slight regression was introduced for users such as IPVS, xt_TEE, and
    raw sockets.  We mix up the users requested destination address with
    the routes assigned nexthop/gateway.  From Julian Anastasov and
    Simon Horman.

35) Fix stack overruns in rt6_probe(), the issue is that can end up
    doing two full packet xmit paths at the same time when emitting
    neighbour discovery messages.  From Hannes Frederic Sowa.

36) davinci_emac driver doesn't handle IFF_ALLMULTI correctly, from
    Mariusz Ceier.

37) Make sure to set TCP sk_pacing_rate after the first legitimate RTT
    sample, from Neal Cardwell.

38) Wrong netlink attribute passed to xfrm_replay_verify_len(), from
    Steffen Klassert.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (152 commits)
  ax88179_178a: Add VID:DID for Samsung USB Ethernet Adapter
  ax88179_178a: Correct the RX error definition in RX header
  Revert "bridge: only expire the mdb entry when query is received"
  tcp: initialize passive-side sk_pacing_rate after 3WHS
  davinci_emac.c: Fix IFF_ALLMULTI setup
  mac802154: correct a typo in ieee802154_alloc_device() prototype
  ipv6: probe routes asynchronous in rt6_probe
  netfilter: nf_conntrack: fix rt6i_gateway checks for H.323 helper
  ipv6: fill rt6i_gateway with nexthop address
  ipv6: always prefer rt6i_gateway if present
  bnx2x: Set NETIF_F_HIGHDMA unconditionally
  bnx2x: Don't pretend during register dump
  bnx2x: Lock DMAE when used by statistic flow
  bnx2x: Prevent null pointer dereference on error flow
  bnx2x: Fix config when SR-IOV and iSCSI are enabled
  bnx2x: Fix Coalescing configuration
  bnx2x: Unlock VF-PF channel on MAC/VLAN config error
  bnx2x: Prevent an illegal pointer dereference during panic
  bnx2x: Fix Maximum CoS estimation for VFs
  drivers: net: cpsw: fix kernel warn during iperf test with interrupt pacing
  ...

compiler/gcc4: Add quirk for 'asm goto' miscompilation bug

2013-10-11T05:39:14+00:00

Fengguang Wu, Oleg Nesterov and Peter Zijlstra tracked down
a kernel crash to a GCC bug: GCC miscompiles certain 'asm goto'
constructs, as outlined here:

  http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58670

Implement a workaround suggested by Jakub Jelinek.

Reported-and-tested-by: Fengguang Wu 
Reported-by: Oleg Nesterov 
Reported-by: Peter Zijlstra 
Suggested-by: Jakub Jelinek 
Reviewed-by: Richard Henderson 
Cc: Linus Torvalds 
Cc: Andrew Morton 
Cc: 
Signed-off-by: Ingo Molnar

net: fix unsafe set_memory_rw from softirq

2013-10-07T19:16:45+00:00

on x86 system with net.core.bpf_jit_enable = 1

sudo tcpdump -i eth1 'tcp port 22'

causes the warning:
[   56.766097]  Possible unsafe locking scenario:
[   56.766097]
[   56.780146]        CPU0
[   56.786807]        ----
[   56.793188]   lock(&(&vb->lock)->rlock);
[   56.799593]   
[   56.805889]     lock(&(&vb->lock)->rlock);
[   56.812266]
[   56.812266]  *** DEADLOCK ***
[   56.812266]
[   56.830670] 1 lock held by ksoftirqd/1/13:
[   56.836838]  #0:  (rcu_read_lock){.+.+..}, at: [] vm_unmap_aliases+0x8c/0x380
[   56.849757]
[   56.849757] stack backtrace:
[   56.862194] CPU: 1 PID: 13 Comm: ksoftirqd/1 Not tainted 3.12.0-rc3+ #45
[   56.868721] Hardware name: System manufacturer System Product Name/P8Z77 WS, BIOS 3007 07/26/2012
[   56.882004]  ffffffff821944c0 ffff88080bbdb8c8 ffffffff8175a145 0000000000000007
[   56.895630]  ffff88080bbd5f40 ffff88080bbdb928 ffffffff81755b14 0000000000000001
[   56.909313]  ffff880800000001 ffff880800000000 ffffffff8101178f 0000000000000001
[   56.923006] Call Trace:
[   56.929532]  [] dump_stack+0x55/0x76
[   56.936067]  [] print_usage_bug+0x1f7/0x208
[   56.942445]  [] ? save_stack_trace+0x2f/0x50
[   56.948932]  [] ? check_usage_backwards+0x150/0x150
[   56.955470]  [] mark_lock+0x282/0x2c0
[   56.961945]  [] __lock_acquire+0x45d/0x1d50
[   56.968474]  [] ? __lock_acquire+0x2de/0x1d50
[   56.975140]  [] ? cpumask_next_and+0x55/0x90
[   56.981942]  [] lock_acquire+0x92/0x1d0
[   56.988745]  [] ? vm_unmap_aliases+0x16a/0x380
[   56.995619]  [] _raw_spin_lock+0x41/0x50
[   57.002493]  [] ? vm_unmap_aliases+0x16a/0x380
[   57.009447]  [] vm_unmap_aliases+0x16a/0x380
[   57.016477]  [] ? vm_unmap_aliases+0x8c/0x380
[   57.023607]  [] change_page_attr_set_clr+0xc0/0x460
[   57.030818]  [] ? trace_hardirqs_on+0xd/0x10
[   57.037896]  [] ? kmem_cache_free+0xb0/0x2b0
[   57.044789]  [] ? free_object_rcu+0x93/0xa0
[   57.051720]  [] set_memory_rw+0x2f/0x40
[   57.058727]  [] bpf_jit_free+0x2c/0x40
[   57.065577]  [] sk_filter_release_rcu+0x1a/0x30
[   57.072338]  [] rcu_process_callbacks+0x202/0x7c0
[   57.078962]  [] __do_softirq+0xf7/0x3f0
[   57.085373]  [] run_ksoftirqd+0x35/0x70

cannot reuse jited filter memory, since it's readonly,
so use original bpf insns memory to hold work_struct

defer kfree of sk_filter until jit completed freeing

tested on x86_64 and i386

Signed-off-by: Alexei Starovoitov 
Acked-by: Eric Dumazet 
Signed-off-by: David S. Miller

sparc: fix MSI build failure on Sparc32

2013-10-03T00:02:35+00:00

Commit ebd97be635 ('PCI: remove ARCH_SUPPORTS_MSI kconfig option')
removes the ARCH_SUPPORTS_MSI Kconfig option that allowed
architectures to indicate whether they support PCI MSI or not. Now,
PCI MSI support can be compiled in on any architecture thanks to the
use of weak functions thanks to 4287d824f265 ('PCI: use weak functions
for MSI arch-specific functions').

So, architecture specific code is now responsible to ensure that its
PCI MSI code builds in all cases, or be appropriately conditionally
compiled.

On Sparc, the MSI support is only provided for Sparc64, so the
ARCH_SUPPORTS_MSI kconfig option was only selected for SPARC64, and
not for the Sparc architecture as a whole. Therefore, removing
ARCH_SUPPORTS_MSI broke Sparc32 configurations with CONFIG_PCI_MSI=y,
because the Sparc-specific MSI code is not designed to be built on
Sparc32.

To solve this, this commit ensures that the Sparc MSI code is only
built on Sparc64. This is done thanks to a new Kconfig Makefile helper
option SPARC64_PCI_MSI, modeled after the existing SPARC64_PCI. The
SPARC64_PCI_MSI option is an hidden option that is true when both
Sparc64 PCI support is enabled and MSI is enabled. The
arch/sparc/kernel/pci_msi.c file is now only built when
SPARC64_PCI_MSI is true.

Signed-off-by: Thomas Petazzoni 
Reported-by: Guenter Roeck 
Tested-by: Guenter Roeck 
Signed-off-by: David S. Miller

sparc: remove deprecated IRQF_DISABLED

2013-10-03T00:02:35+00:00

This patch proposes to remove the IRQF_DISABLED flag from sparc architecture
code. It's a NOOP since 2.6.35 and it will be removed one day.

Signed-off-by: Michael Opdenacker 
Signed-off-by: David S. Miller

sparc: fix ldom_reboot buffer overflow harder

2013-10-03T00:01:56+00:00

The length argument to strlcpy was still wrong. It could overflow the end of
full_boot_str by 5 bytes. Instead of strcat and strlcpy, just use snprint.

Reported-by: Brad Spengler 
Signed-off-by: Kees Cook 
Signed-off-by: David S. Miller

sparc64: Fix buggy strlcpy() conversion in ldom_reboot().

2013-09-27T20:46:04+00:00

Commit 117a0c5fc9c2d06045bd217385b2b39ea426b5a6 ("sparc: kernel: using
strlcpy() instead of strcpy()") added a bug to ldom_reboot in
arch/sparc/kernel/ds.c

-		strcpy(full_boot_str + strlen("boot "), boot_command);
+				     strlcpy(full_boot_str + strlen("boot "), boot_command,
+				     			     sizeof(full_boot_str + strlen("boot ")));

That last sizeof() expression evaluates to sizeof(size_t) which is
not what was intended.

Also even the corrected:

     sizeof(full_boot_str) + strlen("boot ")

is not right as the destination buffer length is just plain
"sizeof(full_boot_str)" and that's what the final argument
should be.

Signed-off-by: David S. Miller

Remove GENERIC_HARDIRQ config option

2013-09-13T13:09:52+00:00

After the last architecture switched to generic hard irqs the config
options HAVE_GENERIC_HARDIRQS & GENERIC_HARDIRQS and the related code
for !CONFIG_GENERIC_HARDIRQS can be removed.

Signed-off-by: Martin Schwidefsky