- 17 Sep, 2010 7 commits
-
-
Amit Kumar Salecha authored
LRO + GRO + vlan rx accleration support, performance increases around 20% and cpu utilization reduces around 70% on vlan interface. Signed-off-by:
Amit Kumar Salecha <amit.salecha@qlogic.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Amit Kumar Salecha authored
GRO support + vlan rx accleration, boost around 9% performance and reduces 25% of cpu utilization on vlan interface. Signed-off-by:
-
Amit Kumar Salecha authored
Implemented vlan rx accleration in driver. This helps in increasing significant performance and reduces cpu utilization with GRO and LRO. Eric Dumazet: "Its a bit strange you use dev_kfree_skb_any(skb) here." "We run in NAPI mode, so you can use dev_kfree_skb()." Amit: Done. Using dev_kfree_skb(); Signed-off-by:
Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by:
-
Dan Rosenberg authored
This was previously reported as a security issue due to leakage of uninitialized stack memory. Jeff Mahoney pointed out that this is incorrect since the copied data is from a union (rather than a struct). Therefore, this patch is only under consideration for the sake of correctness, and is not security relevant. Signed-off-by:
Dan Rosenberg <dan.j.rosenberg@gmail.com> Acked-by:
Grant Grundler <grundler@parisc-linux.org> Signed-off-by:
-
Eric Dumazet authored
commit ab95bfe0 (net: replace hooks in __netif_receive_skb) added rx_handler at wrong place, between two cache line aligned objects, creating a big hole (a full cache line) Move rx_handler and rx_handler_data before rx_queue, filling existing hole. Move master field in the cache line(s) used in receive path. This saves 64 bytes (or L1_CACHE_BYTES), and avoids two possible cache misses in receive path. Signed-off-by:
-
Eric Dumazet authored
As RTNL is held while doing tunnels inserts and deletes, we can remove ip6_tnl_lock spinlock. My initial RCU conversion was conservative and converted the rwlock to spinlock, with no RTNL requirement. Use appropriate rcu annotations and modern lockdep checks as well. Signed-off-by:
-
Stephen Rothwell authored
rcu_dereference_raw() now needs to know the type of its argument. Signed-off-by:
Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by:
-
- 16 Sep, 2010 19 commits
-
-
Brandon Philips authored
Currently vlan devices don't have GRO by default as none of the Ethernet drivers add NETIF_F_GRO to their vlan_features. As GRO is a software feature add GRO to dev->vlan_features in register_netdevice() and let vlan_dev_init() take care that it gets enabled only when dev->features has NETIF_F_GRO too. Signed-off-by:
Brandon Philips <bphilips@suse.de> Acked-by:
-
Breno Leitao authored
This patch removes the unconditional return in the end of the function check_sqs() Signed-off-by:
Breno Leitao <leitao@linux.vnet.ibm.com> Signed-off-by:
-
stephen hemminger authored
The driver has supported GRO for a while, but it was not enabled by default. Signed-off-by:
Stephen Hemminger <shemminger@vyatta.com> Signed-off-by:
-
Eric Dumazet authored
dev->ip_ptr is protected by rtnl and rcu. Yet some places dont use appropriate primitives and/or locking rules. Signed-off-by:
-
David S. Miller authored
net/phonet/socket.c: In function ‘pn_res_seq_show’: net/phonet/socket.c:726: warning: format ‘%02X’ expects type ‘unsigned int’, but argument 3 has type ‘long int’ Signed-off-by: -
Rémi Denis-Courmont authored
Signed-off-by:
Rémi Denis-Courmont <remi.denis-courmont@nokia.com> Signed-off-by:
-
Rémi Denis-Courmont authored
Signed-off-by:
-
Rémi Denis-Courmont authored
Signed-off-by:
-
Rémi Denis-Courmont authored
I wish we could use something cleaner, such as bind(). But that would not work since resource subscription is orthogonal/in addition to the normal object ID allocated via bind(). This is similar to multicasting which also uses ioctl()'s. Signed-off-by:
-
Rémi Denis-Courmont authored
When both destination device and object are nul, Phonet routes the packet according to the resource field. In fact, this is the most common pattern when sending Phonet "request" packets. In this case, the packet is delivered to whichever endpoint (socket) has registered the resource. This adds a new table so that Linux processes can register their Phonet sockets to Phonet resources, if they have adequate privileges. (Namespace support is not implemented at the moment.) Signed-off-by:
-
Rémi Denis-Courmont authored
Closing a pipe endpoint is not normally allowed by the Phonet pipe, other than as a side after-effect of removing the pipe between two endpoints. But there is no way to prevent Linux userspace processes from being killed or suffering from bugs, so this can still happen. We might as well forcefully close Phonet pipe endpoints then. The cellular modem supports only a few existing pipes at a time. So we really should not leak them. This change instructs the modem to destroy the pipe if either of the pipe's endpoint (Linux socket) is closed too early. Signed-off-by:
-
-
Arnd Bergmann authored
The use of the big kernel lock in misdn is completely bogus, so let's just remove it. Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Cc: Karsten Keil <isdn@linux-pingi.de> Cc: netdev@vger.kernel.org Signed-off-by:
-
Arnd Bergmann authored
The isdn4linux driver uses the big kernel lock only to serialize access to a few fields in its own modem_info structure. The easiest replacement is a driver-wide mutex. More fine-grained locking would be more appropriate here, but likely harder to implement. Signed-off-by:
-
Arnd Bergmann authored
There may be applications trying to seek on the irnet character device, so we should use noop_llseek to avoid returning an error when the default llseek changes to no_llseek. Signed-off-by:
-
Eric Dumazet authored
As RTNL is held while doing tunnels inserts and deletes, we can remove ipip6_lock spinlock. My initial RCU conversion was conservative and converted the rwlock to spinlock, with no RTNL requirement. Use appropriate rcu annotations and modern lockdep checks as well. Signed-off-by:
-
Eric Dumazet authored
As RTNL is held while doing tunnels inserts and deletes, we can remove ipgre_lock spinlock. My initial RCU conversion was conservative and converted the rwlock to spinlock, with no RTNL requirement. Use appropriate rcu annotations and modern lockdep checks as well. Signed-off-by:
-
Eric Dumazet authored
As RTNL is held while doing tunnels inserts and deletes, we can remove ipip_lock spinlock. My initial RCU conversion was conservative and converted the rwlock to spinlock, with no RTNL requirement. Use appropriate rcu annotations and modern lockdep checks as well. Signed-off-by:
-
Eric Dumazet authored
We sometime want to dereference an rcu protected pointer while holding RTNL. Use a macro to hide all lockdep details. Signed-off-by:
-
- 15 Sep, 2010 14 commits
-
-
Ben Hutchings authored
struct ethtool_rawip4_spec and struct ethtool_ether_spec are neither commented nor used by any driver, so remove them. Adjust padding in the user-visible unions that included these structures. Fix references to struct ethtool_rawip4_spec in ethtool_get_rx_ntuple(), which should use struct ethtool_usrip4_spec. struct ethtool_usrip4_spec cannot hold IPv6 host addresses and there is no separate structure that can, so remove ETH_RX_NFC_IP6 and the reference to it in niu. Signed-off-by:
Ben Hutchings <bhutchings@solarflare.com> Signed-off-by:
-
Ben Hutchings authored
There are now several interfaces within the ethtool API for getting and setting RX flow filtering and hashing behaviour, most of which are poorly documented. This adds kernel-doc comments for all these interfaces, based on the existing incomplete comments and on the initial implementations. Signed-off-by:
-
Matt Carlson authored
The tg3's phy routines define temporary variables in many locations within the same routine. This patch unifies all temporary variables into one location. Reviewed-by:
Benjamin Li <benli@broadcom.com> Reviewed-by:
Michael Chan <mchan@broadcom.com> Signed-off-by:
Matt Carlson <mcarlson@broadcom.com> Signed-off-by:
-
Matt Carlson authored
This patch eases stack pressure by dynamically allocating the memory used to temporarily store VPD data. Reviewed-by:
-
Matt Carlson authored
This patch converts the driver to prefer the skb_is_gso_v6() helper over the explicit inlined version. Reviewed-by:
-
Matt Carlson authored
Now that each NAPI instance has its own producer ring, it no longer makes sense to keep the producer ring structure external. This patch migrates the producer ring struct to tg3_napi and pivots the code to the new implementation. Reviewed-by:
-
Matt Carlson authored
TG3_IRQ_MAX_VECS should be seen as the maximum number of vectors that any device could be expected to use. tp->irq_max represents the maximum number of vectors the current device can use. This patch clarifies the semantics of the code to match the above description. Reviewed-by:
-
Matt Carlson authored
This patch adjusts the driver to use the tg3_start_xmit_dma_bug() transmit routine for all revisions of 5717 asic rev devices and then allows the driver to attach to B0 and later devices. Reviewed-by:
-
Matt Carlson authored
NCSI firmware does not accept APE events. It relies on a "driver state" location in shared memory to tell it what the driver's current state is. This patch pivots the code to use the new driver state scheme. Reviewed-by:
-
Matt Carlson authored
It was recently discovered that enabling TSS can lockup the device. This patch disables the feature until a suitable workaround can be found. Signed-off-by:
-
Matt Carlson authored
Earlier versions of tg3 devices had a problem where the read DMA FIFO could be overrun in certain edge conditions. The fix was to limit the number of rx BDs the hardware would fetch at a time. For later devices (5761, 5784 and later ASIC revs), there is a hardware fix that must be enabled to fix the same problem. This patch adds that hardware fix. There is a gap in the ASIC revision lineage where neither fix is applied. This is intentional as these ASIC revisions are not afflicted by the bug. Reviewed-by:
-
Gerrit Renker authored
This simplifies and consolidates the TX option-parsing code: 1. The Loss Intervals option is not currently used, so dead code related to this option is removed. I am aware of no plans to support the option, but if someone wants to implement it (e.g. for inter-op tests), it is better to start afresh than having to also update currently unused code. 2. The Loss Event and Receive Rate options have a lot of code in common (both are 32 bit, both have same length etc.), so this is consolidated. 3. The test against GSR is not necessary, because - on first loading CCID3, ccid_new() zeroes out all fields in the socket; - ccid3_hc_tx_packet_recv() treats 0 and ~0U equivalently, due to pinv = opt_recv->ccid3or_loss_event_rate; if (pinv == ~0U || pinv == 0) hctx->p = 0; - as a result, the sequence number field is removed from opt_recv. Signed-off-by:Gerrit Renker <gerrit@erg.abdn.ac.uk>
-
Gerrit Renker authored
This removes the RTT-sampling function tfrc_tx_hist_rtt(), since 1. it suffered from complex passing of return values (the return value both indicated successful lookup while the value doubled as RTT sample); 2. when for some odd reason the sample value equalled 0, this triggered a bug warning about "bogus Ack", due to the ambiguity of the return value; 3. on a passive host which has not sent anything the TX history is empty and thus will lead to unwanted "bogus Ack" warnings such as ccid3_hc_tx_packet_recv: server(e7b7d518): DATAACK with bogus ACK-28197148 ccid3_hc_tx_packet_recv: server(e7b7d518): DATAACK with bogus ACK-26641606. The fix is to replace the implicit encoding by performing the steps manually. Furthermore, the "bogus Ack" warning has been removed, since it can actually be triggered due to several reasons (network reordering, old packet, (3) above), hence it is not very useful. Signed-off-by: -
Gerrit Renker authored
This fixes a subtle bug in the calculation of the inter-packet gap and shows that t_delta, as it is currently used, is not needed. The algorithm from RFC 5348, 8.3 below continually computes a send time t_nom, which is initialised with the current time t_now; t_gran = 1E6 / HZ specifies the scheduling granularity, s the packet size, and X the sending rate: t_distance = t_nom - t_now; // in microseconds t_delta = min(t_ipi, t_gran) / 2; // `delta' parameter in microseconds if (t_distance >= t_delta) { reschedule after (t_distance / 1000) milliseconds; } else { t_ipi = s / X; // inter-packet interval in usec t_nom += t_ipi; // compute the next send time send packet now; } Problem: -------- Rescheduling requires a conversion into milliseconds (sk_reset_timer()). The highest jiffy resolution with HZ=1000 is 1 millisecond, so using a higher granularity does not make much sense here. As a consequence, values of t_distance < 1000 are truncated to 0. This issue has so far been resolved by using instead if (t_distance >= t_delta + 1000) reschedule after (t_distance / 1000) milliseconds; This is unnecessarily large, a lower bound is t_delta' = max(t_delta, 1000). And it implies a further simplification: a) when HZ >= 500, then t_delta <= t_gran/2 = 10^6/(2*HZ) <= 1000, so that t_delta' = MAX(1000, t_delta) = 1000 (constant value); b) when HZ < 500, then t_delta = 1/2*MIN(rtt, t_ipi, t_gran) <= t_gran/2, so that 1000 <= t_delta' <= t_gran/2. The maximum error of using a constant t_delta in (b) is less than half a jiffy. Fix: ---- The patch replaces t_delta with a constant, whose value depends on CONFIG_HZ, changing the above algorithm to: if (t_distance >= t_delta') reschedule after (t_distance / 1000) milliseconds; where t_delta' = 10^6/(2*HZ) if HZ < 500, and t_delta' = 1000 otherwise. Signed-off-by:
-
