Current recovery approach is to wake s/w Tx queues for skb->dev netdevice.
However this approach doesn't cover the case when h/w queue is full of
packets from a single wireless interface. Suppose xmit attempt from the
second wireless interface fails due to failed reclaim. Then the second
interface will not have a chance to recover even if subsequent reclaims
succeed. Possible solution is to attempt to wake all the s/w queues
belonging to driver interfaces.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Under heavy load it is normal that h/w Tx queue is almost full all the time
and reclaim should be done before transmitting next packet. Warning still
should be reported as well as s/w Tx queues should be stopped in the
case when reclaim failed.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Fix tx path regression. Lock should be held when queuing packets
to h/w fifos in order to properly handle configurations with
multiple enabled interfaces.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Use 64-bit dma for hosts with CONFIG_ARCH_DMA_ADDR_T_64BIT enabled.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Check if skb tracking arrays has been already allocated. This additional
check handles the case when init partially failed.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
NULL is not a special type of success here but a error pointer.
So it makes sense to check against NULL in qtnf_map_bar
and return error code.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Linux built-in circ_buf implementation assumes that that the
circular buffer length is a power of 2. Make sure that
rx and tx descriptor queue lengths are power-of-2.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Perform additional reclaim from qtnf_pcie_data_tx. Lock tx_lock serves only
reclaim synchronization purposes. Rename it accordingly and improve
granularity moving this lock to qtnf_pcie_data_tx_reclaim.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Introduce counter for Rx underflow events. Export this counter via debugfs.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Current code for both Rx and Tx queue management is a custom and incomplete
circular buffer implementation. It makes a lot of sense to switch to kernel
built-in circ_buf implementation.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Avoid extra buffering in driver by default. Use max hardware Tx queue size.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Update PCIE_HDP_TX_HOST_Q_WR_PTR register in skb2rbd_attach as a part of
procedure of passing new Rx buffer to hardware. Sync up all the the
qtnf_rx_bd descriptor updates before passing Rx buffer to hardware.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Replace __dev_alloc_skb and explicit NET_IP_ALIGN alignment by built-in
__netdev_alloc_skb_ip_align function.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Use napi_gro_receive() rather than netif_receive_skb() to improve
performance when GRO is enabled.
Signed-off-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
pci_device_id are not supposed to change at runtime. All functions
working with pci_device_id provided by <linux/pci.h> work with
const pci_device_id. So mark the non-const structs as const.
Signed-off-by: Arvind Yadav <arvind.yadav.cs@gmail.com>
Reviewed-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
As the only caller of dma_supported() outside of DMA API internals, the
qtfnmac driver stands out and invites scrutiny. Thankfully, it's not
being used for evil, but it is entirely redundant, since it open-codes a
check that the DMA mask setting functions are going to perform anyway.
In fact, the whole qtnf_pcie_init_dma_mask() function is nothing more
than a rather long-winded implementation of dma_set_mask_and_coherent(),
so let's just use that directly.
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Acked-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
The return value ret is unitialized and garbage is being returned
for the three different error conditions when setting up the PCIe
BARs. Fix this by initializing ret to -ENOMEM to indicate that
the BARs failed to be setup correctly.
Detected by CoverityScan, CID#1437563 ("Unitialized scalar variable")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Sergey Matyukevich <sergey.matyukevich.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
A common pattern with skb_put() is to just want to memcpy()
some data into the new space, introduce skb_put_data() for
this.
An spatch similar to the one for skb_put_zero() converts many
of the places using it:
@@
identifier p, p2;
expression len, skb, data;
type t, t2;
@@
(
-p = skb_put(skb, len);
+p = skb_put_data(skb, data, len);
|
-p = (t)skb_put(skb, len);
+p = skb_put_data(skb, data, len);
)
(
p2 = (t2)p;
-memcpy(p2, data, len);
|
-memcpy(p, data, len);
)
@@
type t, t2;
identifier p, p2;
expression skb, data;
@@
t *p;
...
(
-p = skb_put(skb, sizeof(t));
+p = skb_put_data(skb, data, sizeof(t));
|
-p = (t *)skb_put(skb, sizeof(t));
+p = skb_put_data(skb, data, sizeof(t));
)
(
p2 = (t2)p;
-memcpy(p2, data, sizeof(*p));
|
-memcpy(p, data, sizeof(*p));
)
@@
expression skb, len, data;
@@
-memcpy(skb_put(skb, len), data, len);
+skb_put_data(skb, data, len);
(again, manually post-processed to retain some comments)
Reviewed-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds support for new FullMAC WiFi driver for Quantenna
QSR10G chipsets.
QSR10G (aka Pearl) is Quantenna's 8x8, 160M, 11ac offering.
QSR10G supports 2 simultaneous WMACs - one 5G and one 2G.
5G WMAC supports 160M, 8x8 configuration. FW supports
up to 8 concurrent virtual interfaces on each WMAC.
Patch introduces 2 new drivers:
- qtnfmac.ko for interfacing with kernel wireless core
- qtnfmac_pearl_pcie.ko for interfacing with hardware over PCIe interface
Signed-off-by: Dmitrii Lebed <dlebed@quantenna.com>
Signed-off-by: Sergei Maksimenko <smaksimenko@quantenna.com>
Signed-off-by: Sergey Matyukevich <smatyukevich@quantenna.com>
Signed-off-by: Bindu Therthala <btherthala@quantenna.com>
Signed-off-by: Huizhao Wang <hwang@quantenna.com>
Signed-off-by: Kamlesh Rath <krath@quantenna.com>
Signed-off-by: Avinash Patil <avinashp@quantenna.com>
Signed-off-by: Igor Mitsyanko <igor.mitsyanko.os@quantenna.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>