Skip to content

L
linux
Switch branch/tag
  1. 23 Sep, 2010 2 commits
  3. 22 Sep, 2010 9 commits
  5. 21 Sep, 2010 4 commits
  7. 20 Sep, 2010 6 commits
  9. 18 Sep, 2010 2 commits
  11. 17 Sep, 2010 9 commits
  13. 16 Sep, 2010 1 commit
  15. 15 Sep, 2010 3 commits
    • Denis Kirjanov's avatar
      3c59x: Remove atomic context inside vortex_{set|get}_wol · 84176b7b
      Denis Kirjanov authored 
      There is no need to use spinlocks in vortex_{set|get}_wol.
This also fixes a bug:
[  254.214993] 3c59x 0000:00:0d.0: PME# enabled
[  254.215021] BUG: sleeping function called from invalid context at kernel/mutex.c:94
[  254.215030] in_atomic(): 0, irqs_disabled(): 1, pid: 4875, name: ethtool
[  254.215042] Pid: 4875, comm: ethtool Tainted: G        W   2.6.36-rc3+ #7
[  254.215049] Call Trace:
[  254.215050]  [] __might_sleep+0xb1/0xb6
[  254.215050]  [] mutex_lock+0x17/0x30
[  254.215050]  [] acpi_enable_wakeup_device_power+0x2b/0xb1
[  254.215050]  [] acpi_pm_device_sleep_wake+0x42/0x7f
[  254.215050]  [] acpi_pci_sleep_wake+0x5d/0x63
[  254.215050]  [] platform_pci_sleep_wake+0x1d/0x20
[  254.215050]  [] __pci_enable_wake+0x90/0xd0
[  254.215050]  [] acpi_set_WOL+0x8e/0xf5 [3c59x]
[  254.215050]  [] vortex_set_wol+0x4e/0x5e [3c59x]
[  254.215050]  [] dev_ethtool+0x1cf/0xb61
[  254.215050]  [] ? debug_mutex_free_waiter+0x45/0x4a
[  254.215050]  [] ? __mutex_lock_common+0x204/0x20e
[  254.215050]  [] ? __mutex_lock_slowpath+0x12/0x15
[  254.215050]  [] ? mutex_lock+0x23/0x30
[  254.215050]  [] dev_ioctl+0x42c/0x533
[  254.215050]  [] ? _cond_resched+0x8/0x1c
[  254.215050]  [] ? lock_page+0x1c/0x30
[  254.215050]  [] ? page_address+0x15/0x7c
[  254.215050]  [] ? filemap_fault+0x187/0x2c4
[  254.215050]  [] sock_ioctl+0x1d4/0x1e0
[  254.215050]  [] ? sock_ioctl+0x0/0x1e0
[  254.215050]  [] vfs_ioctl+0x19/0x33
[  254.215050]  [] do_vfs_ioctl+0x424/0x46f
[  254.215050]  [] ? selinux_file_ioctl+0x3c/0x40
[  254.215050]  [] sys_ioctl+0x40/0x5a
[  254.215050]  [] sysenter_do_call+0x12/0x22

vortex_set_wol protected with a spinlock, but nested  acpi_set_WOL acquires a mutex inside atomic context.
Ethtool operations are already serialized by RTNL mutex, so it is safe to drop the locks.
Signed-off-by: default avatarDenis Kirjanov <dkirjanov@kernel.org>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      84176b7b
    • Alexey Kuznetsov's avatar
      tcp: Prevent overzealous packetization by SWS logic. · 01f83d69
      Alexey Kuznetsov authored 
      If peer uses tiny MSS (say, 75 bytes) and similarly tiny advertised
window, the SWS logic will packetize to half the MSS unnecessarily.

This causes problems with some embedded devices.

However for large MSS devices we do want to half-MSS packetize
otherwise we never get enough packets into the pipe for things
like fast retransmit and recovery to work.

Be careful also to handle the case where MSS > window, otherwise
we'll never send until the probe timer.
Reported-by: default avatarツ Leandro Melo de Sales <leandroal@gmail.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      01f83d69
    • David S. Miller's avatar
      net: RPS needs to depend upon USE_GENERIC_SMP_HELPERS · 6dcbc122
      David S. Miller authored 
      You cannot invoke __smp_call_function_single() unless the
architecture sets this symbol.
Reported-by: default avatarDaniel Hellstrom <daniel@gaisler.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      6dcbc122
  17. 14 Sep, 2010 4 commits
    • Simon Guinot's avatar
      phylib: fix PAL state machine restart on resume · fddd9101
      Simon Guinot authored 
      On resume, before starting the PAL state machine, check if the
adjust_link() method is well supplied. If not, this would lead to a
NULL pointer dereference in the phy_state_machine() function.

This scenario can happen if the Ethernet driver call manually the PHY
functions instead of using the PAL state machine. The mv643xx_eth driver
is a such example.
Signed-off-by: default avatarSimon Guinot <sguinot@lacie.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      fddd9101
    • Eric Dumazet's avatar
      net: use rcu_barrier() in rollback_registered_many · ef885afb
      Eric Dumazet authored 
      netdev_wait_allrefs() waits that all references to a device vanishes.

It currently uses a _very_ pessimistic 250 ms delay between each probe.
Some users reported that no more than 4 devices can be dismantled per
second, this is a pretty serious problem for some setups.

Most of the time, a refcount is about to be released by an RCU callback,
that is still in flight because rollback_registered_many() uses a
synchronize_rcu() call instead of rcu_barrier(). Problem is visible if
number of online cpus is one, because synchronize_rcu() is then a no op.

time to remove 50 ipip tunnels on a UP machine :

before patch : real 11.910s
after patch : real 1.250s
Reported-by: default avatarNicolas Dichtel <nicolas.dichtel@6wind.com>
Reported-by: default avatarOctavian Purdila <opurdila@ixiacom.com>
Reported-by: default avatarBenjamin LaHaise <bcrl@kvack.org>
Signed-off-by: default avatarEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      ef885afb
    • Andy Gospodarek's avatar
      bonding: correctly process non-linear skbs · ab12811c
      Andy Gospodarek authored 
      It was recently brought to my attention that 802.3ad mode bonds would no
longer form when using some network hardware after a driver update.
After snooping around I realized that the particular hardware was using
page-based skbs and found that skb->data did not contain a valid LACPDU
as it was not stored there.  That explained the inability to form an
802.3ad-based bond.  For balance-alb mode bonds this was also an issue
as ARPs would not be properly processed.

This patch fixes the issue in my tests and should be applied to 2.6.36
and as far back as anyone cares to add it to stable.

Thanks to Alexander Duyck <alexander.h.duyck@intel.com> and Jesse
Brandeburg <jesse.brandeburg@intel.com> for the suggestions on this one.
Signed-off-by: default avatarAndy Gospodarek <andy@greyhouse.net>
CC: Alexander Duyck <alexander.h.duyck@intel.com>
CC: Jesse Brandeburg <jesse.brandeburg@intel.com>
CC: stable@kerne.org
Signed-off-by: default avatarJay Vosburgh <fubar@us.ibm.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      ab12811c
    • Michael S. Tsirkin's avatar
      vhost-net: fix range checking in mrg bufs case · ee05d693
      Michael S. Tsirkin authored 
      In mergeable buffer case, we use headcount, log_num
and seg as indexes in same-size arrays, and
we know that headcount <= seg and
log_num equals either 0 or seg.

Therefore, the right thing to do is range-check seg,
not headcount as we do now: these will be different
if guest chains s/g descriptors (this does not
happen now, but we can not trust the guest).

Long term, we should add BUG_ON checks to verify
two other indexes are what we think they should be.
Reported-by: default avatarJason Wang <jasowang@redhat.com>
Signed-off-by: default avatarMichael S. Tsirkin <mst@redhat.com>
      ee05d693
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7