Skip to content

L
linux
Switch branch/tag
  1. 23 Dec, 2004 2 commits
    • Maksim Krasnyanskiy's avatar
      Use random_ether_addr() to generate TAP MAC address. · 49f81c2d
      Maksim Krasnyanskiy authored 
      Signed-off-by: default avatarMark Smith <markzzzsmith@yahoo.com.au>
Signed-off-by: default avatarMax Krasnyansky <maxk@qualcomm.com>
      49f81c2d
    • Maksim Krasnyanskiy's avatar
      TUN/TAP driver packet queuing fixes and improvements · 5ce8b230
      Maksim Krasnyanskiy authored 
      Patch from Harald Roelle <harald.roelle@nm.ifi.lmu.de>

Fixes for the following issues
- "Kicking" packet behavior in case of kernel packet scheduler (! TUN_ONE_QUEUE):
  When the netif queue is stopped because of an overrun of the driver's queue,
  no new packets are delivered to the driver until a new packet arrives, not even
  when in the meantime there's again space in the driver's queue (gained by a
  reading user process). In short, whenever netif queue was stopped, one needs an
  additional packet to "kick" packet delivery to the driver.
  The reason for this is, that the netif queue is started but not woken up, i.e.
  the rest of the kernel is not signaled to resume packet delivery.

- Adjustment of tx queue length by ifconfig has only effect when TUN_ONE_QUEUE
  is set.  Otherwise always constant TUN_READQ_SIZE queue length is used.

- TX queue default length setting is not consistent:
  o TAP dev + TUN_ONE_QUEUE: 1000 (by ether_setup())
  o all other cases: 10

- Default tx queue length is too short in many cases.
  IMHO it should be at least twice as long as the number of fragments needed
  for a maximum sized IP packet at a "typical" MTU size.
  This would ensure that at least one complete IP packet can be delivered
  to the attached user space process, even if the packet's fragments
  are "misaligned" in the buffer and the user process is not scheduled
  in time to read the queue.

Additional modifications:

- To signal that stopping of the queue has occurred, the tx fifo overrun
  counter is increased.

- Implemented ethtool api. Primarily added to have a standard method requesting
  the driver version.
Signed-off-by: default avatarMax Krasnyansky <maxk@qualcomm.com>
      5ce8b230
  3. 22 Dec, 2004 6 commits
  5. 21 Dec, 2004 11 commits
  7. 20 Dec, 2004 5 commits
    • Andi Kleen's avatar
      [PATCH] x86_64: fix syscall/signal restart bug · 0233f53f
      Andi Kleen authored 
      Fix a pretty bad bug that caused sometimes signals on x86-64
to be restarted like system calls. This corrupted the RIP and
in general caused undesirable effects.

The problem happens because orig_rax is unsigned on x86-64,
but it originally was signed when the signal code was written.
And the if (orig_rax >= 0) ended up always true.
And gcc didn't warn about this, because the warning is only in 
-Wextra. 

In 2.4 we still had a cast for it, but somehow it got dropped
in 2.5.

Credit goes to John Slice for tracking it down and Erich Boleyn
for the original fix. All blame to me. I fixed it at another
place too.
Signed-off-by: default avatarAndi Kleen <ak@suse.de>
Signed-off-by: default avatarLinus Torvalds <torvalds@osdl.org>
      0233f53f
    • Thomas Graf's avatar
      [PKT_SCHED]: Provide compat policer stats in action policer · 59b17c27
      Thomas Graf authored 
      This should go in before 2.6.10. It fixes a forgotten case to provide
police backward compatibility statistics for old iproute2 versions
running on a new kernel with actions enabled. Should make distributions
happy with older iproute2 versions and all-included kernel configs
since they probably favour actions over plain policer.

Testing results:
  iproute2-2.4.7 on 2.6.10-rc3-bk8:
  cls-police: police creation succeeded
  cls-police: Sending 10 ICMP echo requests
  cls-police: police dumping succeeded with output:
  filter protocol ip pref 10 u32 
  filter protocol ip pref 10 u32 fh 800: ht divisor 1 
  filter protocol ip pref 10 u32 fh 800::800 order 2048 key ht 800 bkt 0 flowid 10:12 
  police 3 action drop rate 2Kbit burst 10Kb mtu 2Kb 
    match 00010000/00ff0000 at 8
   Sent 420 bytes 10 pkts (dropped 0, overlimits 0)  <-- This would have been missing
  cls-police: police deletion succeeded

 iproute2-2.6.9 on 2.6.10-rc3-bk8:
 ...
  filter protocol ip pref 10 u32 
  filter protocol ip pref 10 u32 fh 800: ht divisor 1 
  filter protocol ip pref 10 u32 fh 800::800 order 2048 key ht 800 bkt 0 flowid 10:12  (rule hit 10 success 10)
    match 00010000/00ff0000 at 8 (success 10 ) 
   police 0x4 rate 2000bit burst 10Kb mtu 2Kb action drop 
  ref 1 bind 1
   Sent 420 bytes 10 pkts (dropped 0, overlimits 0) 
 ...

 (Same results for fw classifier)
Signed-off-by: default avatarThomas Graf <tgraf@suug.ch>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      59b17c27
    • Thomas Graf's avatar
      [PKT_SCHED]: Fix double locking in tcindex destroy path. · ee6fa69b
      Thomas Graf authored 
      tcindex's destroy uses its own delete functions to destroy its
configuration. The delete function (correctly) takes the qdisc_tree_lock
to prevent list walkings from happening while removing from the list.
The qdisc_tree_lock is already held if we're comming via the destroy
path and thus a double locking takes place.
Signed-off-by: default avatarThomas Graf <tgraf@suug.ch>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      ee6fa69b
    • Stephen Hemminger's avatar
      [TCP]: Missing newline character in printk. · 1777c9de
      Stephen Hemminger authored 
      Signed-off-by: default avatarStephen Hemminger <shemminger@osdl.org>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      1777c9de
    • Linus Torvalds's avatar
      Merge bk://linux-scsi.bkbits.net/scsi-for-linus-2.6 · 39509e47
      Linus Torvalds authored 
      into ppc970.osdl.org:/home/torvalds/v2.6/linux
      39509e47
  9. 18 Dec, 2004 3 commits
  11. 17 Dec, 2004 13 commits
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7