- 17 Apr, 2008 38 commits
-
-
Frank Blaschka authored
Allocate headroom for TR_HLEN but using only ETH_HLEN causes rx performance degradation. Allocate ETH_HLEN for ethernet and TR_HLEN for token ring (layer 3 mode). Signed-off-by:
Frank Blaschka <frank.blaschka@de.ibm.com> Signed-off-by:
Jeff Garzik <jgarzik@redhat.com>
-
Peter Tiedemann authored
Improving debug message handling, moving ipa into messages from kernel to dbf, some cleanups and typo fixes. Signed-off-by:
Peter Tiedemann <ptiedem@de.ibm.com> Signed-off-by:
-
Frank Blaschka authored
hw does not allow to change the mac address. Signed-off-by:
-
Ursula Braun authored
Symptom: slow CCL response time Problem: non-ETH_P_802_2 packets are not delivered to NDH for CCL. But CCL detects missing sequence numbers, which cause a serious performance problem with CCL. Solution: assign sequence numbers only to 802.2 packets. Signed-off-by:Ursula Braun <braunu@de.ibm.com> Signed-off-by:
-
Ursula Braun authored
Problem: A STARTLAN command from the adapter may arrive while a qeth recovery is currently running with a failed qeth STARTLAN. Usually qeth schedules a recovery when receiving a STARTLAN command from the adapter. But another recovery scheduled while a recovery is already running never starts. Thus the qeth-administered lan_online flag remains zero in this scenario, even though the adapter-STARTLAN has happened. Solution: Set lan_online flag for a received STARTLAN from the adapter in case scheduled recovery does not start. Signed-off-by: -
Ursula Braun authored
OSA-adapters do not have an address limitation for the qdio queue structures except the MAX storage level of the current processor. And due to a recent z/VM APAR there is no longer a restriction to allocate qdio structures below 2 GB. Signed-off-by:
-
Ursula Braun authored
1. ip_list handling after deregister failure of multicast address: If error code "MC Address not found" is returned do not re-add multicast address to ip_list. For other error codes readd multicast address at the end of function qeth_delete_all_mc. 2. ip_list handling after deregister failure or normal ip address: If error code "IP Address not found" is returned do not re-add multicast address to ip list. This is especially important in IP address takeover scenarios, to enable re-takeover of a taken over IP address. Signed-off-by:
-
Bruce Allan authored
The register tests should be run with all the proper flags enabled to maximize the test coverage code and make sure we are as close as we can get to testing regular traffic. Reformat the code for readability. Minor cleanups in the descriptor ring setup. Signed-off-by:
Bruce Allan <bruce.w.allan@intel.com> Signed-off-by:
Auke Kok <auke-jan.h.kok@intel.com> Signed-off-by:
-
Jeff Kirsher authored
Several minor cosmetic function renames. Signed-off-by:
Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by:
-
Jeff Kirsher authored
With multiple queues coming into the code these base control registers need to be made into arrays. Signed-off-by:
-
Jeff Kirsher authored
Signed-off-by:
-
Stephen Hemminger authored
Use netdev_alloc_skb since it handles any NUMA node memory localtion issues and sets skb->dev. Since device driver was not setting skb->dev, I bet filter rules based on device would not work. Signed-off-by:
Stephen Hemminger <shemminger@vyatta.com> Signed-off-by:
-
Stephen Hemminger authored
Any negative return value from start_xmit is interpreted as NETDEV_TX_LOCK which is not what this driver wants. It should return 0 (NETDEV_TX_OK) when it consumes a packet. Also, use skb_padto() as the generic way to pad small frames. Signed-off-by:
-
Stephen Hemminger authored
Statistics structure is available for use in net_device structure. Compile tested only. Signed-off-by:
-
Peter Horton authored
This patch works around the MWI bug on the DC21143 rev 65 Tulip by ensuring that the receive buffers don't end on a cache line boundary (as documented in the errata). This patch is required for the MIPS based Cobalt Qube/RaQ as supporting the extra PCI commands seems to reduce the chance of a hard lockup between the Tulip and the PCI bridge. Signed-off-by:
Thomas Bogendoerfer <tsbogend@alpha.franken.de> Signed-off-by:
-
Jeff Garzik authored
winbond-840 shares tulip.h with the tulip driver, because they share many (but not all) of the same register definitions. This is useful for the register definitions, but not helpful when it comes to symbols that are shared among the tulip driver's C modules, but not meant to be shared outside that one driver. Thus, PKT_BUF_SZ is a symbol internal to tulip, but it was intruding upon a similar symbol in winbond-840's namespace. This was not a problem as long as the two symbols had the same value, but upcoming patches result in differing symbol values. Signed-off-by: -
Joe Perches authored
bond_main.c already #includes <linux/seq_file.h> Signed-off-by:
Joe Perches <joe@perches.com> Signed-off-by:
-
Andy Fleming authored
Multi-line comments weren't all CodingStyle compliant Signed-off-by:
Andy Fleming <afleming@freescale.com> Signed-off-by:
-
Andy Fleming authored
Having the id field be an int was making more complex bus topologies excessively difficult. For now, just convert it to a string, and change all instances of "bus->id = val" to snprintf(id, MII_BUS_ID_LEN, "%x", val). Signed-off-by:
-
Dai Haruki authored
Poll the completed TX frames in gfar_poll(). This prevents the tx completion interrupt from interfering with processing of received frames. We also disable hardware rx coalescing when NAPI is enabled. Signed-off-by:
Dai Haruki <dai.haruki@freescale.com> Signed-off-by:
-
Ishizaki Kou authored
This patch revises the logging for link informations of spidernet. - The link down message is too verbose because auto-negotiation timeout occurs periodically while an ethernet cable is not connected. - We want to see the link result, and we think it should be displayed. Signed-off-by:Kou Ishizaki <kou.ishizaki@toshiba.co.jp> Acked-by:
Jens Osterkamp <jens@de.ibm.com> Signed-off-by:
-
Ishizaki Kou authored
In addition to the value of GHIINT0STS, spidernet interrupt handler should check the values of GHIINT1STS/GHIINT2STS registers at the beginning of spider_net_interrupt() so as not to drop error interrupts. GHIINT1STS/GHIINT2STS registers indicates some of erroneous conditions in spidernet, and a few bits of GHIINT0STS register reflects these conditions. But GHIINT0MSK masks these bits, so you should check these conditions by reading GHIINT1STS/GHIINT2STS registers directly. Signed-off-by:
-
Ishizaki Kou authored
This patch changes spidernet interrupt masks. - unmask GDAINVAINT. There is an operation to do by spidernet interrupt handler. - mask some interrupts. There are no operations in the interrupt handler. Signed-off-by:
-
Ishizaki Kou authored
This patch extends the timeout for spidernet auto-negotiation. Auto-negotiation often fails to finish in 2 seconds. Signed-off-by:
-
Ishizaki Kou authored
This patch fixes initialization of "aneg_count" and "medium" fields in spider_net_card to make spidernet driver correctly sets "link status". Signed-off-by:
-
Atsushi Nemoto authored
Cosmetic TAB/whitespace cleanups and some style cleanups. No functional changes. Signed-off-by:
Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by:
-
Atsushi Nemoto authored
Convert the tc35815 driver to use the generic PHY layer in drivers/net/phy. Also rename 'boardtype' to 'chiptype' which hould be more appropriate. Signed-off-by:
-
Atsushi Nemoto authored
Use managed pci functions and kill unnecessary volatiles. Signed-off-by:
-
Atsushi Nemoto authored
Use netdev_priv() instead of dev->priv. Signed-off-by:
-
Atsushi Nemoto authored
Use print_mac() and DECLARE_MAC_BUF(). Signed-off-by:
-
Atsushi Nemoto authored
Use struct net_device_stats embedded in struct net_device. Signed-off-by:
-
Lennert Buytenhek authored
Signed-off-by:
Lennert Buytenhek <buytenh@marvell.com> Reviewed-by:
Tzachi Perelstein <tzachi@marvell.com> Signed-off-by:
Dale Farnsworth <dale@farnsworth.org>
-
Lennert Buytenhek authored
When there are multiple mv643xx_eth silicon blocks in the system, don't print an initialisation message for each and every one of them. Signed-off-by:
-
Lennert Buytenhek authored
Pass a struct mv643xx_private * to the register accessor functions, as a preparation for having multiple mv643xx_eth silicon blocks. (Since this causes some 80 column straddling, and the mv_ prefix is useless anyway, rename mv_read to rdl and mv_write to wrl to compensate.) Signed-off-by:
-
Lennert Buytenhek authored
In error and warning printks, always report the netdevice name instead of the port index (the latter has no meaning when there are multiple mv643xx_eth silicon blocks in the system.) Signed-off-by:
-
Lennert Buytenhek authored
Instead of identifying individual mv643xx ethernet ports by only their port number, identify them by their struct mv643xx_private *, as just a port number has no meaning when there are multiple mv643xx_eth silicon blocks in the system. Signed-off-by:
-
Lennert Buytenhek authored
- Remove unused MV643XX_DEFAULT_[RT]X_QUEUE_SIZE definitions. - Remove ETH_TARGET enum -- it isn't used anywhere in the driver, and isn't even valid for non-mv643xx chip models, as those use different MBUS target IDs. - Clean up comment and control flow in mv643xx_eth_change_mtu(). - Use mp->dev instead of mp->mii.dev in mv643xx_eth_tx_timeout_task(). - Make mv643xx_eth_free_tx_descs() static. - Remove overzealous NULL check in mv643xx_eth_start_xmit(). - Use symbolic NETDEV_TX_* constants in mv643xx_eth_start_xmit(). Signed-off-by:
-
Lennert Buytenhek authored
mv643xx_eth_start_xmit() should check mp->tx_desc_count only inside the mp->lock spinlock. Signed-off-by:
-
- 16 Apr, 2008 2 commits
-
-
John Heffner authored
This change is necessary to allow cwnd to grow during persistent reordering. Cwnd moderation is applied when in the disorder state and an ack that fills the hole comes in. If the hole was greater than 3 packets, but less than tp->reordering, cwnd will shrink when it should not have. Signed-off-by:
John Heffner <jheffner@napa.(none)> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Pavel Emelyanov authored
The alloc_netdev_mq() tries to produce 32-bytes alignment for both the net_device itself and its private data. The second alignment is achieved by adding the NETDEV_ALIGN_CONST to the whole size of the memory to be allocated. However, for those devices that do not need the private area, this addition just makes the net_device weight 1024 + 32 = 1068 bytes, i.e. consume twice as much memory. Since loopback device is such (sizeof_priv == 0 for it), and each net namespace creates one, this can save a noticeable amount of memory for kernel with net namespaces turned on. After this set the lo device is actually allocated from a size-1024 kmem cache on i386 box even with NETPOLL and WIRELESS_EXT turned on. Signed-off-by:
Pavel Emelyanov <xemul@openvz.org> Signed-off-by:
-
