- 10 11月, 2011 11 次提交
-
-
由 Johannes Berg 提交于
The frames are used by AP/STA WDS mode, and hostapd needs to know when such a frame was received to set up the VLAN appropriately to allow using it. Signed-off-by: NJohannes Berg <johannes.berg@intel.com> Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
由 Johannes Berg 提交于
Add the ability to register to received beacon frames to allow implementing OLBC logic in userspace. The registration is per wiphy since there's no point in receiving the same frame multiple times. Signed-off-by: NJohannes Berg <johannes.berg@intel.com> Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
由 Johannes Berg 提交于
When the AP SME in hostapd is used it wants to probe the clients when they have been idle for some time. Add explicit API to support this. Signed-off-by: NJohannes Berg <johannes.berg@intel.com> Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
由 Johannes Berg 提交于
Add the ability to advertise that the device contains the AP SME and what features it can support. There are currently no features in the bitmap -- probe response offload will be advertised by a few patches Arik is working on now (who took over from Guy Eilam) and a device with AP SME will typically implement and require response offload. Signed-off-by: NJohannes Berg <johannes.berg@intel.com> Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
由 Johannes Berg 提交于
To implement AP mode without monitor interfaces we need to be able to send a deauth to stations that send frames without being associated. Enable this by adding a new nl80211 event for such frames that an application can subscribe to. Signed-off-by: NJohannes Berg <johannes.berg@intel.com> Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
由 Johannes Berg 提交于
Drivers that need to drop a frame before it can be transmitted will usually simply free that frame. This is currently fine, but in the future it'll be needed to tell mac80211 about this case, so add a new routine that frees a TX skb. Signed-off-by: NJohannes Berg <johannes.berg@intel.com> Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
由 Thomas Pedersen 提交于
Previously QoS multicast frames had the Normal Acknowledgment QoS control bits set. This would cause broadcast frames to be discarded by peers with which we have a BA session, since their sequence number would fall outside the allowed range. Set No Ack QoS control bits on multicast QoS frames and filter these in de-aggregation code. Signed-off-by: NThomas Pedersen <thomas@cozybit.com> v2: Use proper QoS Ack Policy ctl field mask (Christian) v3: Clean up conditional (Johannes) Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
由 Johannes Berg 提交于
The 802.1X EAPOL handshake hostapd does requires knowing whether the frame was ack'ed by the peer. Currently, we fudge this pretty badly by not even transmitting the frame as a normal data frame but injecting it with radiotap and getting the status out of radiotap monitor as well. This is rather complex, confuses users (mon.wlan0 presence) and doesn't work with all hardware. To get rid of that hack, introduce a real wifi TX status option for data frame transmissions. This works similar to the existing TX timestamping in that it reflects the SKB back to the socket's error queue with a SCM_WIFI_STATUS cmsg that has an int indicating ACK status (0/1). Since it is possible that at some point we will want to have TX timestamping and wifi status in a single errqueue SKB (there's little point in not doing that), redefine SO_EE_ORIGIN_TIMESTAMPING to SO_EE_ORIGIN_TXSTATUS which can collect more than just the timestamp; keep the old constant as an alias of course. Currently the internal APIs don't make that possible, but it wouldn't be hard to split them up in a way that makes it possible. Thanks to Neil Horman for helping me figure out the functions that add the control messages. Signed-off-by: NJohannes Berg <johannes.berg@intel.com> Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
由 Jouni Malinen 提交于
This value is used for WPI-SMS4 in ISO/IEC JTC 1 N 9880. Signed-off-by: NJouni Malinen <jouni@qca.qualcomm.com> Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
由 Johannes Berg 提交于
This function returns a referenced BSS struct (or NULL), annotate with __must_check. It seems that a lot of drivers get this completely wrong and leak all BSS structs as a result. Reported-by: NAdam Mikuta <Adam.Mikuta@tieto.com> Signed-off-by: NJohannes Berg <johannes.berg@intel.com> Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
由 Christian Lamparter 提交于
802.11n-2009 extends the supported rates element with a magic value which can be used to prevent legacy stations from joining the BSS. However, this magic value is not a rate like the others and the magic can simply be ignored/skipped at this late stage. Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>--- Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
- 09 11月, 2011 4 次提交
-
-
由 Johan Hedberg 提交于
This patch moves the pending management command list (previously global to mgmt.c) into struct hci_dev. This makes it possible to do proper locking when accessing it (through the existing hci_dev locks) and thereby avoid race conditions. Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com> Acked-by: NMarcel Holtmann <marcel@holtmann.org> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Johan Hedberg 提交于
The current global pending command list in mgmt.c is racy. Possibly the simplest way to fix it is to have per-hci dev lists instead of a global one (all commands that need a pending struct are hci_dev specific). This way the list can be protected using the already existing per-hci dev lock. To enable this refactoring the first thing that needs to be done is to ensure that the mgmt functions have access to the hci_dev struct (instead of just the dev id). Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com> Acked-by: NMarcel Holtmann <marcel@holtmann.org> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Arik Nemtsov 提交于
Some cards can generate CCMP IVs in HW, but require the space for the IV to be pre-allocated in the frame at the correct offset. Add a key flag that allows us to achieve this. Signed-off-by: NArik Nemtsov <arik@wizery.com> Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
由 Johannes Berg 提交于
Two new struct members were not documented, fix that. Signed-off-by: NJohannes Berg <johannes.berg@intel.com> Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
- 08 11月, 2011 20 次提交
-
-
由 Johan Hedberg 提交于
This patch adds address type info (typically BR/EDR vs LE) to management messages that need this. This also ensures conformance to the latest management API specification. Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Johan Hedberg 提交于
BR/EDR link keys have their own commands and events (separate from SMP) and the remove_keys command (previously remove_key) removes keys of any kind for the specified remote address. Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Johan Hedberg 提交于
The power off code doesn't need to use its own custom timer since the delayed_work API provides the exact same functionality. Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Johan Hedberg 提交于
A proper mgmt_command_status should be returned to user-space if either discoverable or connectable enabling fails. Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Johan Hedberg 提交于
Based on the revised mgmt API set_discoverable has a timeout parameter to specify how long the adapter will remain discoverable. A value of 0 means "indefinitively". Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Andre Guedes 提交于
This patch adds a function to hci_core to cancel an ongoing inquiry. According to the Bluetooth spec, the inquiry cancel command should only be issued after the inquiry command has been issued, a command status event has been received for the inquiry command, and before the inquiry complete event occurs. As HCI_INQUIRY flag is only set just after an inquiry command status event occurs and it is cleared just after an inquiry complete event occurs, the inquiry cancel command should be issued only if HCI_INQUIRY flag is set. Additionally, cancel inquiry related code from stop_discovery() were replaced by a hci_cancel_inquiry() call. Signed-off-by: NAndre Guedes <andre.guedes@openbossa.org> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Andre Guedes 提交于
This patch adds a function to hci_core to carry out inquiry. All inquiry code from start_discovery() were replaced by a hci_do_inquiry() call. Signed-off-by: NAndre Guedes <andre.guedes@openbossa.org> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Andrei Emeltchenko 提交于
Use HCI error reasons instead of magic numbers. Signed-off-by: NAndrei Emeltchenko <andrei.emeltchenko@intel.com> Acked-by: NMarcel Holtmann <marcel@holtmann.org> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Mat Martineau 提交于
Symbolic fixed channel IDs will be used instead of magic numbers. Signed-off-by: NMat Martineau <mathewm@codeaurora.org> Acked-by: NMarcel Holtmann <marcel@holtmann.org> Acked-by: NAndrei Emeltchenko <andrei.emeltchenko@intel.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Mat Martineau 提交于
AMP channel creation and channel moves are coordinated using the L2CAP signaling channel. These definitions cover the "create channel", "move channel", and "move channel confirm" signals. Signed-off-by: NMat Martineau <mathewm@codeaurora.org> Acked-by: NMarcel Holtmann <marcel@holtmann.org> Acked-by: NAndrei Emeltchenko <andrei.emeltchenko@intel.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Mat Martineau 提交于
Each channel has a policy to require BR/EDR (the default), prefer BR/EDR, or prefer AMP. Check for valid policy value and L2CAP mode. Signed-off-by: NMat Martineau <mathewm@codeaurora.org> Acked-by: NMarcel Holtmann <marcel@holtmann.org> Acked-by: NAndrei Emeltchenko <andrei.emeltchenko@intel.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Mat Martineau 提交于
This variable is currently only accessible within l2cap_core.c, but it is also needed in l2cap_sock.c Signed-off-by: NMat Martineau <mathewm@codeaurora.org> Acked-by: NMarcel Holtmann <marcel@holtmann.org> Acked-by: NAndrei Emeltchenko <andrei.emeltchenko@intel.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Mat Martineau 提交于
Allow control of AMP functionality on L2CAP sockets. By default, connections will be restricted to BR/EDR. Manipulating the BT_CHANNEL_POLICY option allows for channels to be moved to or created on AMP controllers. Signed-off-by: NMat Martineau <mathewm@codeaurora.org> Acked-by: NMarcel Holtmann <marcel@holtmann.org> Acked-by: NAndrei Emeltchenko <andrei.emeltchenko@intel.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Luiz Augusto von Dentz 提交于
This implement priority based scheduler using skbuffer priority set via SO_PRIORITY socket option. It introduces hci_chan_hash (list of HCI Channel/hci_chan) per connection, each item in this list refer to a L2CAP connection and it is used to queue the data for transmission. Signed-off-by: NLuiz Augusto von Dentz <luiz.von.dentz@intel.com> Acked-by: NMarcel Holtmann <marcel@holtmann.org> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Szymon Janc 提交于
This was triggered by turning off encryption on ACL link when rfcomm was using high security. rfcomm_security_cfm (which is called from rx task) was closing DLC and this involves sending disconnect message (and locking socket). Move closing DLC to rfcomm_process_dlcs and only flag DLC for closure in rfcomm_security_cfm. BUG: sleeping function called from invalid context at net/core/sock.c:2032 in_atomic(): 1, irqs_disabled(): 0, pid: 1788, name: kworker/0:3 [<c0068a08>] (unwind_backtrace+0x0/0x108) from [<c05e25dc>] (dump_stack+0x20/0x24) [<c05e25dc>] (dump_stack+0x20/0x24) from [<c0087ba8>] (__might_sleep+0x110/0x12c) [<c0087ba8>] (__might_sleep+0x110/0x12c) from [<c04801d8>] (lock_sock_nested+0x2c/0x64) [<c04801d8>] (lock_sock_nested+0x2c/0x64) from [<c05670c8>] (l2cap_sock_sendmsg+0x58/0xcc) [<c05670c8>] (l2cap_sock_sendmsg+0x58/0xcc) from [<c047cf6c>] (sock_sendmsg+0xb0/0xd0) [<c047cf6c>] (sock_sendmsg+0xb0/0xd0) from [<c047cfc8>] (kernel_sendmsg+0x3c/0x44) [<c047cfc8>] (kernel_sendmsg+0x3c/0x44) from [<c056b0e8>] (rfcomm_send_frame+0x50/0x58) [<c056b0e8>] (rfcomm_send_frame+0x50/0x58) from [<c056b168>] (rfcomm_send_disc+0x78/0x80) [<c056b168>] (rfcomm_send_disc+0x78/0x80) from [<c056b9f4>] (__rfcomm_dlc_close+0x2d0/0x2fc) [<c056b9f4>] (__rfcomm_dlc_close+0x2d0/0x2fc) from [<c056bbac>] (rfcomm_security_cfm+0x140/0x1e0) [<c056bbac>] (rfcomm_security_cfm+0x140/0x1e0) from [<c0555ec0>] (hci_event_packet+0x1ce8/0x4d84) [<c0555ec0>] (hci_event_packet+0x1ce8/0x4d84) from [<c0550380>] (hci_rx_task+0x1d0/0x2d0) [<c0550380>] (hci_rx_task+0x1d0/0x2d0) from [<c009ee04>] (tasklet_action+0x138/0x1e4) [<c009ee04>] (tasklet_action+0x138/0x1e4) from [<c009f21c>] (__do_softirq+0xcc/0x274) [<c009f21c>] (__do_softirq+0xcc/0x274) from [<c009f6c0>] (do_softirq+0x60/0x6c) [<c009f6c0>] (do_softirq+0x60/0x6c) from [<c009f794>] (local_bh_enable_ip+0xc8/0xd4) [<c009f794>] (local_bh_enable_ip+0xc8/0xd4) from [<c05e5804>] (_raw_spin_unlock_bh+0x48/0x4c) [<c05e5804>] (_raw_spin_unlock_bh+0x48/0x4c) from [<c040d470>] (data_from_chip+0xf4/0xaec) [<c040d470>] (data_from_chip+0xf4/0xaec) from [<c04136c0>] (send_skb_to_core+0x40/0x178) [<c04136c0>] (send_skb_to_core+0x40/0x178) from [<c04139f4>] (cg2900_hu_receive+0x15c/0x2d0) [<c04139f4>] (cg2900_hu_receive+0x15c/0x2d0) from [<c0414cb8>] (hci_uart_tty_receive+0x74/0xa0) [<c0414cb8>] (hci_uart_tty_receive+0x74/0xa0) from [<c02cbd9c>] (flush_to_ldisc+0x188/0x198) [<c02cbd9c>] (flush_to_ldisc+0x188/0x198) from [<c00b2774>] (process_one_work+0x144/0x4b8) [<c00b2774>] (process_one_work+0x144/0x4b8) from [<c00b2e8c>] (worker_thread+0x198/0x468) [<c00b2e8c>] (worker_thread+0x198/0x468) from [<c00b9bc8>] (kthread+0x98/0xa0) [<c00b9bc8>] (kthread+0x98/0xa0) from [<c0061744>] (kernel_thread_exit+0x0/0x8) Signed-off-by: NSzymon Janc <szymon.janc@tieto.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Johan Hedberg 提交于
This patch adds the necessary code to send proper command status or command complete events to the start/stop discovery management commands. Before this patch these events were completely missing. Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Luiz Augusto von Dentz 提交于
This uses SO_PRIORITY to set the skbuffer priority field Signed-off-by: NLuiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Andrei Emeltchenko 提交于
Add L2CAP Config Pending state for EFS. Currently after receiving Config Response Pending respond with Config Response Success. ... > ACL data: handle 1 flags 0x02 dlen 16 L2CAP(s): Connect rsp: dcid 0x0040 scid 0x0040 result 0 status 0 Connection successful > ACL data: handle 1 flags 0x02 dlen 45 L2CAP(s): Config req: dcid 0x0040 flags 0x00 clen 33 RFC 0x03 (Enhanced Retransmission, TxWin 63, MaxTx 3, RTo 0, MTo 0, MPS 1009) EFS (Id 0x01, SerType Best Effort, MaxSDU 0xffff, SDUitime 0xffffffff, AccLat 0xffffffff, FlushTO 0x0000ffff) < ACL data: handle 1 flags 0x00 dlen 45 L2CAP(s): Config req: dcid 0x0040 flags 0x00 clen 33 RFC 0x03 (Enhanced Retransmission, TxWin 63, MaxTx 3, RTo 0, MTo 0, MPS 498) EFS (Id 0x01, SerType Best Effort, MaxSDU 0xffff, SDUitime 0xffffffff, AccLat 0xffffffff, FlushTO 0x0000ffff) < ACL data: handle 1 flags 0x00 dlen 47 L2CAP(s): Config rsp: scid 0x0040 flags 0x00 result 4 clen 33 Pending MTU 672 RFC 0x03 (Enhanced Retransmission, TxWin 63, MaxTx 3, RTo 2000, MTo 12000, MPS 498) EFS (Id 0x01, SerType Best Effort, MaxSDU 0xffff, SDUitime 0xffffffff, AccLat 0xffffffff, FlushTO 0x0000ffff) > ACL data: handle 1 flags 0x02 dlen 47 L2CAP(s): Config rsp: scid 0x0040 flags 0x00 result 4 clen 33 Pending MTU 672 RFC 0x03 (Enhanced Retransmission, TxWin 63, MaxTx 3, RTo 2000, MTo 12000, MPS 498) EFS (Id 0x01, SerType Best Effort, MaxSDU 0xffff, SDUitime 0xffffffff, AccLat 0xffffffff, FlushTO 0x0000ffff) > ACL data: handle 1 flags 0x02 dlen 14 L2CAP(s): Config rsp: scid 0x0040 flags 0x00 result 0 clen 0 Success < ACL data: handle 1 flags 0x00 dlen 14 L2CAP(s): Config rsp: scid 0x0040 flags 0x00 result 0 clen 0 Success < ACL data: handle 1 flags 0x00 dlen 510 L2CAP(d): cid 0x0040 len 506 ext_ctrl 0x00010000 fcs 0xebe0 [psm 4113] I-frame: Start (len 672) TxSeq 0 ReqSeq 0 ... Signed-off-by: NAndrei Emeltchenko <andrei.emeltchenko@intel.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 David Herrmann 提交于
hci_unregister_dev cannot fail and always returns 0. The drivers already ignore the return value so we can safely make it return void. Signed-off-by: NDavid Herrmann <dh.herrmann@googlemail.com> Acked-by: NMarcel Holtmann <marcel@holtmann.org> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
由 Andrzej Kaczmarek 提交于
Timers set by __set_chan_timer() should use miliseconds instead of jiffies. Commit 942ecc9c updated l2cap_set_timer() so it expects timeout to be specified in msecs instead of jiffies. This makes timeouts unreliable when CONFIG_HZ is not set to 1000. Signed-off-by: NAndrzej Kaczmarek <andrzej.kaczmarek@tieto.com> Signed-off-by: NGustavo F. Padovan <padovan@profusion.mobi>
-
- 05 11月, 2011 1 次提交
-
-
由 Johannes Berg 提交于
The warning really shouldn't happen, but until we find the reason why it does don't spew it all the time, just once is enough to know we've hit it. Reported-by: NLinus Torvalds <torvalds@linux-foundation.org> Signed-off-by: NJohannes Berg <johannes.berg@intel.com> Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>
-
- 30 10月, 2011 1 次提交
-
-
由 Eric Dumazet 提交于
commit 2425717b (net: allow vlan traffic to be received under bond) broke ARP processing on vlan on top of bonding. +-------+ eth0 --| bond0 |---bond0.103 eth1 --| | +-------+ 52870.115435: skb_gro_reset_offset <-napi_gro_receive 52870.115435: dev_gro_receive <-napi_gro_receive 52870.115435: napi_skb_finish <-napi_gro_receive 52870.115435: netif_receive_skb <-napi_skb_finish 52870.115435: get_rps_cpu <-netif_receive_skb 52870.115435: __netif_receive_skb <-netif_receive_skb 52870.115436: vlan_do_receive <-__netif_receive_skb 52870.115436: bond_handle_frame <-__netif_receive_skb 52870.115436: vlan_do_receive <-__netif_receive_skb 52870.115436: arp_rcv <-__netif_receive_skb 52870.115436: kfree_skb <-arp_rcv Packet is dropped in arp_rcv() because its pkt_type was set to PACKET_OTHERHOST in the first vlan_do_receive() call, since no eth0.103 exists. We really need to change pkt_type only if no more rx_handler is about to be called for the packet. Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com> Reviewed-by: NJiri Pirko <jpirko@redhat.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
- 27 10月, 2011 1 次提交
-
-
由 Eric Dumazet 提交于
commit 66b13d99 (ipv4: tcp: fix TOS value in ACK messages sent from TIME_WAIT) fixed IPv4 only. This part is for the IPv6 side, adding a tclass param to ip6_xmit() We alias tw_tclass and tw_tos, if socket family is INET6. [ if sockets is ipv4-mapped, only IP_TOS socket option is used to fill TOS field, TCLASS is not taken into account ] Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
- 26 10月, 2011 1 次提交
-
-
由 Michal Schmidt 提交于
The user may use "foo-bar" for a kernel parameter defined as "foo_bar". Make sure it works the other way around too. Apply the equality of dashes and underscores on early_params and __setup params as well. The example given in Documentation/kernel-parameters.txt indicates that this is the intended behaviour. With the patch the kernel accepts "log-buf-len=1M" as expected. https://bugzilla.redhat.com/show_bug.cgi?id=744545Signed-off-by: NMichal Schmidt <mschmidt@redhat.com> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> (neatened implementations)
-
- 25 10月, 2011 1 次提交
-
-
由 Boaz Harrosh 提交于
This is finally the RAID5 Write support. The bigger part of this patch is not the XOR engine itself, But the read4write logic, which is a complete mini prepare_for_striping reading engine that can read scattered pages of a stripe into cache so it can be used for XOR calculation. That is, if the write was not stripe aligned. The main algorithm behind the XOR engine is the 2 dimensional array: struct __stripe_pages_2d. A drawing might save 1000 words --- __stripe_pages_2d | n = pages_in_stripe_unit; w = group_width - parity; | pages array presented to the XOR lib | | V | __1_page_stripe[0].pages --> [c0][c1]..[cw][c_par] <---| | | __1_page_stripe[1].pages --> [c0][c1]..[cw][c_par] <--- | ... | ... | __1_page_stripe[n].pages --> [c0][c1]..[cw][c_par] ^ | data added columns first then row --- The pages are put on this array columns first. .i.e: p0-of-c0, p1-of-c0, ... pn-of-c0, p0-of-c1, ... So we are doing a corner turn of the pages. Note that pages will zigzag down and left. but are put sequentially in growing order. So when the time comes to XOR the stripe, only the beginning and end of the array need be checked. We scan the array and any NULL spot will be field by pages-to-be-read. The FS that wants to support RAID5 needs to supply an operations-vector that searches a given page in cache, and specifies if the page is uptodate or need reading. All these pages to be read are put on a slave ore_io_state and synchronously read. All the pages of a stripe are read in one IO, using the scatter gather mechanism. In write we constrain our IO to only be incomplete on a single stripe. Meaning either the complete IO is within a single stripe so we might have pages to read from both beginning or end of the strip. Or we have some reading to do at beginning but end at strip boundary. The left over pages are pushed to the next IO by the API already established by previous work, where an IO offset/length combination presented to the ORE might get the length truncated and the user must re-submit the leftover pages. (Both exofs and NFS support this) But any ORE user should make it's best effort to align it's IO before hand and avoid complications. A cached ore_layout->stripe_size member can be used for that calculation. (NOTE: that ORE demands that stripe_size may not be bigger then 32bit) What else? Well read it and tell me. Signed-off-by: NBoaz Harrosh <bharrosh@panasas.com>
-