- 23 Nov, 2009 22 commits
-
-
Neil Horman authored
Recently had a bug reported to me, in which the user was sending packets with a payload containing a sequence number. The packets were getting delivered in order according the chunk TSN values, but the sequence values in the payload were arriving out of order. At first I thought it must be an application error, but we eventually found it to be a problem on the transmit side in the sctp stack. The conditions for the error are that multihoming must be in use, and it helps if each transport has a different pmtu. The problem occurs in sctp_outq_flush. Basically we dequeue packets from the data queue, and attempt to append them to the orrered packet for a given transport. After we append a data chunk we add the trasport to the end of a list of transports to have their packets sent at the end of sctp_outq_flush. The problem occurs when a data chunks fills up a offered packet on a transport. The function that does the appending (sctp_packet_transmit_chunk), will try to call sctp_packet_transmit on the full packet, and then append the chunk to a new packet. This call to sctp_packet_transmit, sends that packet ahead of the others that may be queued in the transport_list in sctp_outq_flush. The result is that frames that were sent in one order from the user space sending application get re-ordered prior to tsn assignment in sctp_packet_transmit, resulting in mis-sequencing of data payloads, even though tsn ordering is correct. The fix is to change where we assign a tsn. By doing this earlier, we are then free to place chunks in packets, whatever way we see fit and the protocol will make sure to do all the appropriate re-ordering on receive as is needed. Signed-off-by:
Neil Horman <nhorman@tuxdriver.com> Reported-by:
William Reich <reich@ulticom.com> Signed-off-by:
Vlad Yasevich <vladislav.yasevich@hp.com>
-
Vlad Yasevich authored
Current implementation of max.burst ends up limiting new data during cwnd decay period. The decay is happening becuase the connection is idle and we are allowed to fill the congestion window. The point of max.burst is to limit micro-bursts in response to large acks. This still happens, as max.burst is still applied to each transmit opportunity. It will also apply if a very large send is made (greater then allowed by burst). Tested-by:
Florian Niederbacher <florian.niederbacher@student.uibk.ac.at> Signed-off-by:
-
Vlad Yasevich authored
Recent attempt to remove deprecated socket options demonstrated that removing options from the enum space will have severe binary compatibility issues. The reason is that it changes the subsequent enum space and causes option values to be redefined. To solve this, and to get rid of the ugly double statements for every option, we simply convert to the #define scheme. Signed-off-by: -
Vlad Yasevich authored
The transport last_time_used variable is rather useless. It was only used when determining if CWND needs to be updated due to idle transport. However, idle transport detection was based on a Heartbeat timer and last_time_used was not incremented when sending Heartbeats. As a result the check for cwnd reduction was always true. We can get rid of the variable and just base our cwnd manipulation on the HB timer (like the code comment sais). We also have to call into the cwnd manipulation function regardless of whether HBs are enabled or not. That way we will detect idle transports if the user has disabled Heartbeats. Signed-off-by: -
Amerigo Wang authored
SCTP_GET_*_OLD stuffs are schedlued to be removed. Cc: Vlad Yasevich <vladislav.yasevich@hp.com> Signed-off-by:
WANG Cong <amwang@redhat.com> Signed-off-by:
-
Andrei Pelinescu-Onciul authored
Since draft-ietf-tsvwg-sctpsocket-15.txt, setting the SPP_MTUD_ENABLE flag when changing pathmaxrxt via the SCTP_PEER_ADDR_PARAMS setsockopt is not required any longer. Signed-off-by:
Andrei Pelinescu-Onciul <andrei@iptel.org> Signed-off-by:
-
Vlad Yasevich authored
We currently send window update SACKs every time we free up 1 PMTU worth of data. That a lot more SACKs then necessary. Instead, we'll now send back the actuall window every time we send a sack, and do window-update SACKs when a fraction of the receive buffer has been opened. The fraction is controlled with a sysctl. Signed-off-by: -
Vlad Yasevich authored
When sctp_connectx() is used, we pick the first address as primary, even though it may not have worked. This results in excessive retransmits and poor performance. We should select the address that the association was established with. Reported-by:
Thomas Dreibholz <dreibh@iem.uni-due.de> Signed-off-by:
-
Vlad Yasevich authored
The "Invalid Stream Identifier" error has a 16 bit reserved field at the end, thus making the parameter length be 8 bytes. We've never supplied that reserved field making wireshark tag the packet as malformed. Reported-by:
Chris Dischino <cdischino@sonusnet.com> Signed-off-by:
-
Wei Yongjun authored
This patch implement the sender side for SACK-IMMEDIATELY extension. Section 4.1. Sender Side Considerations Whenever the sender of a DATA chunk can benefit from the corresponding SACK chunk being sent back without delay, the sender MAY set the I-bit in the DATA chunk header. Reasons for setting the I-bit include o The sender is in the SHUTDOWN-PENDING state. o The application requests to set the I-bit of the last DATA chunk of a user message when providing the user message to the SCTP implementation. Signed-off-by:Wei Yongjun <yjwei@cn.fujitsu.com> Signed-off-by:
-
Wei Yongjun authored
This patch implement the receiver side for SACK-IMMEDIATELY extension: Section 4.2. Receiver Side Considerations On reception of an SCTP packet containing a DATA chunk with the I-bit set, the receiver SHOULD NOT delay the sending of the corresponding SACK chunk and SHOULD send it back immediately. Signed-off-by:
-
Wei Yongjun authored
This patch implement the definition for SACK-IMMEDIATELY extension. Section 3. The I-bit in the DATA Chunk Header The following Figure 1 shows the extended DATA chunk. 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type = 0 | Res |I|U|B|E| Length | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | TSN | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Stream Identifier | Stream Sequence Number | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Payload Protocol Identifier | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ \ \ / User Data / \ \ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 1 The only difference between the DATA chunk in Figure 1 and the DATA chunk defined in [RFC4960] is the addition of the I-bit in the flags field of the chunk header. Signed-off-by: -
Alexander Duyck authored
There appears to be a stray setting of the VFE bit when registering vlans. This should not be done as vlan filtering should be enabled any time the interface is not in promiscous mode Signed-off-by:
Mallikarjuna R Chilakala <mallikarjuna.chilakala@intel.com> Acked-by:
Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com> Signed-off-by:
Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Mallikarjuna R Chilakala authored
While configuring rscctl use rx buffer length from rx ring structure instead of passing rx_buf_len to ixgbe_configure_rscctl Signed-off-by:
-
Mallikarjuna R Chilakala authored
Divide 82599 HWRSC counters into aggregated and flushed to count number of packets getting coalesced per TCP connection. Signed-off-by:
-
Sathya Perla authored
The driver can run out of mccq wrbs when completions don't arrive due to an unresponsive card. This must not hit a BUG_ON(); instead log a msg and return an error. Signed-off-by:
Sathya Perla <sathyap@serverengines.com> Signed-off-by:
-
Sathya Perla authored
Disabling msix was missing when probe fails after enabling msix. Signed-off-by:
-
Sathya Perla authored
These cmds are issued to the fw in probe/resume and remove/suspend paths to help fw execute some initialization and cleanup code. This change needed the be_hw_up() code to be refactored as be_get_config(). Signed-off-by:
-
Sathya Perla authored
Signed-off-by:
-
Sathya Perla authored
To send upto 64 addresses in the multicast-set cmd, the non-embeeded cmd format that provides for a bigger buffer is used instead of an embedded format. Signed-off-by:
-
Joe Perches authored
On Sun, 2009-11-22 at 16:31 -0800, David Miller wrote: > It should be of the form: > if (x && > y) > > or: > if (x && y) > > Fix patches, rather than complaints, for existing cases where things > do not follow this pattern are certainly welcome. Also collapsed some multiple tabs to single space. Signed-off-by:
Joe Perches <joe@perches.com> Signed-off-by:
-
Jaswinder Singh Rajput authored
Fix the following htmldocs warning: Warning(net/core/dev.c:5378): bad line: Signed-off-by:
Jaswinder Singh Rajput <jaswinderrajput@gmail.com> Signed-off-by:
-
- 21 Nov, 2009 18 commits
-
-
Bruce Allan authored
The Tx unit hang detection code in e1000e detects other hangs caused by hardware components (e.g. Rx, DMA units), but it is not possible to detect exactly which component is hung so it has always assumed a Tx unit hang. When dumping a message to the system log because of a hang, this patch adds more data to help narrow the cause of the issue and makes the message non-Tx-specific. Because this new code reads PHY registers which can sleep, move it off to a workqueue. This and all previously existing work tasks in the driver are now cancelled when the driver is removed. Signed-off-by:
Bruce Allan <bruce.w.allan@intel.com> Signed-off-by:
-
Bruce Allan authored
Signed-off-by:
-
Bruce Allan authored
Now that mutex_lock() calls might_sleep() the driver doesn't have to here. Signed-off-by:
-
Bruce Allan authored
A failure to initialize the identification LED is not a fatal condition and should allow the init path to continue. Signed-off-by:
-
Bruce Allan authored
It was pointed out a pm_qos DMA latency requirement set when the driver is loaded when parts that support early receive of jumbo frames are probed could have that requirement overidden if another part supported by the driver (one that does not support early receive of jumbo frames) is probed later. Change the DMA latency requirement to be per-interface if needed instead of per driver. Signed-off-by:
-
Bruce Allan authored
The e1000_clear_hw_cntrs_*() functions read the registers to clear them. There is no reason to save the register contents so the temp variable can be removed. Signed-off-by:
-
Bruce Allan authored
Signed-off-by:
-
Bruce Allan authored
Set booleans to 'true' or 'false' to make it clear it is a boolean. Also change instances of TRUE/FALSE in comments to lowercase true/false. Signed-off-by:
-
Bruce Allan authored
Signed-off-by:
-
Bruce Allan authored
Signed-off-by:
-
Bruce Allan authored
Signed-off-by:
-
Bruce Allan authored
The phy and nvm operations structures have function pointers that contain "phy" and "nvm" in the pointer names which are redundant since the structures are already obviously in phy and nvm structures. Signed-off-by:
-
Bruce Allan authored
This patch depends on a previous one that cleans up redundant #includes. Signed-off-by:
-
Bruce Allan authored
Signed-off-by:
-
Bruce Allan authored
The driver was only updating MII stats when an LSC up was detected and the interface had not already been reported up to netdev. This meant MII stats returned in response to an SIOCGMIIREG ioctl would always show a link up if it had ever been up. This was misleading the networking daemon guessnet, which uses this ioctl, into making improper network port selections. This fix adds a call to e1000_phy_read_status() to actively read the mii stats before responding to the SIOCGMIIREG ioctl. Signed-off-by:
-
Bruce Allan authored
When performing the ethtool PHY loopback test on PCH-based LOMs (82577 and 82578), disable K1 (a MAC-PHY interconnect low power mode) otherwise packets might get corrupted. Signed-off-by:
-
Bruce Allan authored
The e1000_get_cable_length_82577() should return a negative value upon error. Signed-off-by:
-
Bruce Allan authored
Copper links with WoL or management enabled (any condition which prevents the phy from being powered down when the interface is taken down) were always reporting link-up when the interface had been taken down. This is because when the interface is taken down (ifconfig ethx down), interrupts are disabled. With no interrupts, there is no LSC interrupt, which is normally required to set "get_link_status", which instructs the driver to query the device for link state. The fix is to force get_link_status to true if the interface is not up. Signed-off-by:
-
