- 27 Jun, 2011 13 commits
-
-
jamal authored
Results on dummy device can be seen in my netconf 2011 slides. These results are for a 10Gige IXGBE intel nic - on another i5 machine, very similar specs to the one used in the netconf2011 results. It turns out - this is a hell lot worse than dummy and so this patch is even more beneficial for 10G. Test setup: ---------- System under test sending packets out. Additional box connected directly dropping packets. Installed prio qdisc on the eth device and default netdev default length of 1000 used as is. The 3 prio bands each were set to 100 (didnt factor in the results). 5 packet runs were made and the middle 3 picked. results ------- The "cpu" column indicates the which cpu the sample was taken on, The "Pkt runx" carries the number of packets a cpu dequeued when forced to be in the "dequeuer" role. The "avg" for each run is the number of times each cpu should be a "dequeuer" if the system was fair. 3.0-rc4 (plain) cpu Pkt run1 Pkt run2 Pkt run3 ================================================ cpu0 21853354 21598183 22199900 cpu1 431058 473476 393159 cpu2 481975 477529 458466 cpu3 23261406 23412299 22894315 avg 11506948 11490372 11486460 3.0-rc4 with patch and default weight 64 cpu Pkt run1 Pkt run2 Pkt run3 ================================================ cpu0 13205312 13109359 13132333 cpu1 10189914 10159127 10122270 cpu2 10213871 10124367 10168722 cpu3 13165760 13164767 13096705 avg 11693714 11639405 11630008 As you can see the system is still not perfect but is a lot better than what it was before... At the moment we use the old backlog weight, weight_p which is 64 packets. It seems to be reasonably fine with that value. The system could be made more fair if we reduce the weight_p (as per my presentation), but we are going to affect the shared backlog weight. Unless deemed necessary, I think the default value is fine. If not we could add yet another knob. Signed-off-by:
Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
David S. Miller authored
-
Joe Perches authored
Use pr_fmt, pr_<level> and netdev_<level> as appropriate. Signed-off-by:
Joe Perches <joe@perches.com> Signed-off-by:
-
Joe Perches authored
Use pr_fmt, pr_<level> and netdev_<level> as appropriate. Signed-off-by:
-
Joe Perches authored
Use pr_fmt, pr_<level> and netdev_<level> as appropriate. Signed-off-by:
-
Joe Perches authored
Use pr_fmt, pr_<level> and netdev_<level> as appropriate. Signed-off-by:
-
Joe Perches authored
Use pr_fmt, pr_<level> and netdev_<level> as appropriate. Signed-off-by:
-
Joe Perches authored
Use pr_fmt, pr_<level> and netdev_<level> as appropriate. Signed-off-by:
-
Joe Perches authored
Use pr_fmt, pr_<level> and netdev_<level> as appropriate. Signed-off-by:
-
Joe Perches authored
Use pr_fmt, pr_<level> and netdev_<level> as appropriate. Signed-off-by:
-
Sathya Perla authored
Initialization of this field to "all priorities" must be done before MCC queue creation. As soon as the MCC queue is created, an event modifying this value may be received. Signed-off-by:
Sathya Perla <sathya.perla@emulex.com> Signed-off-by:
-
Sathya Perla authored
Signed-off-by:
-
Sathya Perla authored
Problem initially reproted and fixed by Eric Dumazet <eric.dumazet@gmail.com> netdev_stats_update() resets netdev->stats and then accumulates stats from various rings. This is wrong as stats readers can sometimes catch zero values. Use temporary variables instead for accumulating per-ring values. Signed-off-by:
Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by:
-
- 25 Jun, 2011 19 commits
-
-
John Fastabend authored
Implement DCB ops dcb_ieee_del() and set FCoE to the default priority when no priority exists. Signed-off-by:
John Fastabend <john.r.fastabend@intel.com> Tested-by:
Ross Brattain <ross.b.brattain@intel.com> Signed-off-by:
Jeff Kirsher <jeffrey.t.kirsher@intel.com>
-
John Fastabend authored
The fcoe.tc field is no longer used so remove it. After the field is removed there is no need to keep fcoe_setapp() around so remove it as well. And finally we can get rid of some DCB #ifdef's in the fcoe code. Signed-off-by:
-
John Fastabend authored
Commit, commit c8ca76eb Author: John Fastabend <john.r.fastabend@intel.com> Date: Sat Mar 12 03:50:53 2011 +0000 ixgbe: DCB, further cleanups to app configuration Removed the getapp() routines from ixgbe because they are no longer needed. It also allowed the set hardware routines to use both IEEE 802.1Qaz app types and CEE app types. This added code to do bit shifting in the IEEE case. This patch reverts the checks and handles the IEEE case from the setapp entry point. I prefer this because it keeps the two paths from having to be aware of the DCB mode. This resolves a bug where I missed setting the selector bit in the IEEE spec value and left it in the CEE value. Now that they are separate routines these types of errors should not occur. Signed-off-by:
-
Lior Levy authored
There is a need to configure MMW_SIZE in register RTTBCNRM with a correct value (0x4 for non jumbo frames and 0x14 for jumbo frames support). For 82599 the value is 0x4 and for X540 the value is 0x14. Signed-off-by:
Lior Levy <lior.levy@intel.com> Signed-off-by:
-
Alexander Duyck authored
This patch updates the current methods used for determining if we have enough space to transmit a given skb. The current method is quite wasteful as it has us go through and determine how each page is going to be broken up. That only needs to be done if pages are larger than our maximum data per TXD. As such I have wrapped that in a page size check. Signed-off-by:
Alexander Duyck <alexander.h.duyck@intel.com> Signed-off-by:
-
Alexander Duyck authored
There is a significant amount of shared functionality between the checksum and TSO offload configuration that is shared in regards to how they setup the context descriptors. Since so much of the functionality is shared it makes sense to move the shared functionality into a single function and just call that function from the two context descriptor specific routines. Signed-off-by:
-
Alexander Duyck authored
This change updates all values dealing with count, next_to_use, and next_to_clean so that they stay u16 values. The advantage of this is that there is no re-casting of type during the propagation through the stack. Signed-off-by:
-
Alexander Duyck authored
This change is a minor cleanup that converts the IXGBE_DESC_UNUSED macro into a static inline function just for the case of the code being a bit cleaner. Signed-off-by:
-
Alexander Duyck authored
This change makes it so that we pass the adapter struct instead of the netdev for most of the basic interrupts that are not associated with q_vectors. Signed-off-by:
-
Vitaliy Ivanov authored
Seems like this was not cleaned during the 'rfkill: rewrite' checkin 19d337df. Signed-off-by:
Vitaliy Ivanov <vitalivanov@gmail.com> Acked-by:
Inaky Perez-Gonzalez <inaky.perez-gonzalez@intel.com> Signed-off-by:
-
Shan Wei authored
File net/TUNABLE has never be updated since git age. For some tunable parameters which user can control with proc file-system, They are all in ip-sysctl.txt doc. For tunable parameters that only at compile time, no meaning to note them. Signed-off-by:
Shan Wei <shanwei@cn.fujitsu.com> Signed-off-by:
-
Joe Perches authored
Convert to current logging styles. Move code blocks to eliminate need for prototypes. Use tabs for code indent and standardize spacing. Comment neatening. Signed-off-by:
Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by:
-
Joe Perches authored
Already declared in 8390.h Signed-off-by:
-
Joe Perches authored
Signed-off-by:
-
Joe Perches authored
Make more conformant to normal kernel style. Long line lengths > 80 columns ignored. Signed-off-by:
-
Joe Perches authored
Move the braces around to conform to kernel standard style. Signed-off-by:
-
Joe Perches authored
Use the current logging styles. Signed-off-by:
-
Joe Perches authored
Use current logging styles. Other miscellaneous cleanups: Space removal and additions for checkpatch warnings. Signed-off-by:
-
Joe Perches authored
Convert to current logging styles. Move code blocks to eliminate need for prototypes. Use tabs for code indent and sandardize spacing. Comment neatening. Signed-off-by:
-
- 24 Jun, 2011 8 commits
-
-
-
David S. Miller authored
Merge branch 'for-davem' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless-next-2.6
-
Allan Stephens authored
Simplifies the creation of connection protocol messages by eliminating the passing of information that is no longer required, is constant, or is contained within the port structure that is issuing the message. Signed-off-by:
Allan Stephens <allan.stephens@windriver.com> Signed-off-by:
Paul Gortmaker <paul.gortmaker@windriver.com>
-
Allan Stephens authored
Modifies the logic that creates a connection termination payload message so that it no longer (mis)uses a routine that creates a connection protocol message. The revised code is now more easily understood, and avoids setting several fields that are either not present in payload messages or were being set more than once. Signed-off-by:
-
Allan Stephens authored
Restructures the logic used in tipc_port_recv_proto_msg() to ensure that incoming connection protocol messages are handled properly. The routine now uses a two-stage process that first ensures the message applies on an existing connection and then processes the request. This corrects a loophole that allowed a connection probe request to be processed if it was sent to an unconnected port that had no names bound to it. Signed-off-by:
-
Allan Stephens authored
Speeds up the creation of the FIN message that terminates a TIPC connection. The typical peer termination message is now created by duplicating the terminating port's standard payload message header and adjusting the message size, importance, and error code fields, rather than building all fields of the message from scratch. A FIN message that is directed to the port itself is created the same way. but also requires swapping the origin and destination address fields. In addition to reducing the work required to create FIN messages, these changes eliminate several instances of duplicated code, Signed-off-by:
-
Allan Stephens authored
Performs cosmetic cleanup of the symbolic names used to specify TIPC payload message header sizes. The revised names now more accurately reflect the payload messages in which they can appear. In addition, several places where these payload message symbol names were being used to create non-payload messages have been updated to use the proper internal message symbolic name. No functional changes are introduced by this rework. Signed-off-by:
-
Allan Stephens authored
Gets rid of code that allows tipc_msg_init() to create a short payload message header. This optimization is possible because there are no longer any callers who require this capability. Signed-off-by:
-
