On 2017-09-30 19:37, silexcommon@xxxxxxxxx wrote:
From: Alagu Sankar <alagusankar@xxxxxxxxxxxxxxx>
The existing implementation of initiating multiple sdio transfers for
receive bundling is slowing down the receive speed. Combining the
transfers using a scatter gather method would be ideal. This results in
significant performance improvement.
Since the sg implementation for sdio transfers are not reliable due to
buffer start and size alignment, a virtual scatter gather implementation
is used.
Signed-off-by: Alagu Sankar <alagusankar@xxxxxxxxxxxxxxx>
---
drivers/net/wireless/ath/ath10k/htc.h | 1 +
drivers/net/wireless/ath/ath10k/sdio.c | 122 ++++++++++++++++++++++++---------
drivers/net/wireless/ath/ath10k/sdio.h | 5 +-
3 files changed, 93 insertions(+), 35 deletions(-)
diff --git a/drivers/net/wireless/ath/ath10k/htc.h b/drivers/net/wireless/ath/ath10k/htc.h
index 24663b0..5d87908 100644
--- a/drivers/net/wireless/ath/ath10k/htc.h
+++ b/drivers/net/wireless/ath/ath10k/htc.h
@@ -58,6 +58,7 @@ enum ath10k_htc_tx_flags {
};
enum ath10k_htc_rx_flags {
+ ATH10K_HTC_FLAGS_RECV_1MORE_BLOCK = 0x01,
ATH10K_HTC_FLAG_TRAILER_PRESENT = 0x02,
ATH10K_HTC_FLAG_BUNDLE_MASK = 0xF0
};
diff --git a/drivers/net/wireless/ath/ath10k/sdio.c b/drivers/net/wireless/ath/ath10k/sdio.c
index bb6fa67..45df9db 100644
--- a/drivers/net/wireless/ath/ath10k/sdio.c
+++ b/drivers/net/wireless/ath/ath10k/sdio.c
@@ -35,6 +35,7 @@
#include "sdio.h"
#define ATH10K_SDIO_DMA_BUF_SIZE (32 * 1024)
+#define ATH10K_SDIO_VSG_BUF_SIZE (32 * 1024)
static int ath10k_sdio_read(struct ath10k *ar, u32 addr, void *buf,
u32 len, bool incr);
@@ -430,6 +431,7 @@ static int ath10k_sdio_mbox_rx_process_packet(struct ath10k *ar,
int ret;
payload_len = le16_to_cpu(htc_hdr->len);
+ skb->len = payload_len + sizeof(struct ath10k_htc_hdr);
if (trailer_present) {
trailer = skb->data + sizeof(*htc_hdr) +
@@ -468,12 +470,13 @@ static int ath10k_sdio_mbox_rx_process_packets(struct ath10k *ar,
enum ath10k_htc_ep_id id;
int ret, i, *n_lookahead_local;
u32 *lookaheads_local;
+ int lookahd_idx = 0;
I think the variable should be named *lookahead_idx* instead of *lookahd_idx*,
since all other variables are using the string lookahead without abbreviations.
for (i = 0; i < ar_sdio->n_rx_pkts; i++) {
lookaheads_local = lookaheads;
n_lookahead_local = n_lookahead;
- id = ((struct ath10k_htc_hdr *)&lookaheads[i])->eid;
+ id = ((struct ath10k_htc_hdr *)&lookaheads[lookahd_idx++])->eid;
if (id >= ATH10K_HTC_EP_COUNT) {
ath10k_warn(ar, "invalid endpoint in look-ahead: %d\n",
@@ -496,6 +499,7 @@ static int ath10k_sdio_mbox_rx_process_packets(struct ath10k *ar,
/* Only read lookahead's from RX trailers
* for the last packet in a bundle.
*/
+ lookahd_idx--;
lookaheads_local = NULL;
n_lookahead_local = NULL;
}
@@ -529,11 +533,11 @@ static int ath10k_sdio_mbox_rx_process_packets(struct ath10k *ar,
return ret;
}
-static int ath10k_sdio_mbox_alloc_pkt_bundle(struct ath10k *ar,
- struct ath10k_sdio_rx_data *rx_pkts,
- struct ath10k_htc_hdr *htc_hdr,
- size_t full_len, size_t act_len,
- size_t *bndl_cnt)
+static int ath10k_sdio_mbox_alloc_bundle(struct ath10k *ar,
+ struct ath10k_sdio_rx_data *rx_pkts,
+ struct ath10k_htc_hdr *htc_hdr,
+ size_t full_len, size_t act_len,
+ size_t *bndl_cnt)
{
int ret, i;
@@ -574,6 +578,7 @@ static int ath10k_sdio_mbox_rx_alloc(struct ath10k *ar,
size_t full_len, act_len;
bool last_in_bundle;
int ret, i;
+ int pkt_cnt = 0;
if (n_lookaheads > ATH10K_SDIO_MAX_RX_MSGS) {
ath10k_warn(ar,
@@ -616,16 +621,22 @@ static int ath10k_sdio_mbox_rx_alloc(struct ath10k *ar,
* optimally fetched as a full bundle.
*/
size_t bndl_cnt;
-
- ret = ath10k_sdio_mbox_alloc_pkt_bundle(ar,
- &ar_sdio->rx_pkts[i],
- htc_hdr,
- full_len,
- act_len,
- &bndl_cnt);
-
- n_lookaheads += bndl_cnt;
- i += bndl_cnt;
+ struct ath10k_sdio_rx_data *rx_pkts =
+ &ar_sdio->rx_pkts[pkt_cnt];
+
+ ret = ath10k_sdio_mbox_alloc_bundle(ar,
+ rx_pkts,
+ htc_hdr,
+ full_len,
+ act_len,
+ &bndl_cnt);
+
+ if (ret) {
+ ath10k_warn(ar, "alloc_bundle error %d\n", ret);
+ goto err;
+ }
+
+ pkt_cnt += bndl_cnt;
/*Next buffer will be the last in the bundle */
last_in_bundle = true;
}
@@ -634,14 +645,18 @@ static int ath10k_sdio_mbox_rx_alloc(struct ath10k *ar,
* ATH10K_HTC_FLAG_BUNDLE_MASK flag set, all bundled
* packet skb's have been allocated in the previous step.
*/
- ret = ath10k_sdio_mbox_alloc_rx_pkt(&ar_sdio->rx_pkts[i],
+ if (htc_hdr->flags & ATH10K_HTC_FLAGS_RECV_1MORE_BLOCK)
+ full_len += ATH10K_HIF_MBOX_BLOCK_SIZE;
+
+ ret = ath10k_sdio_mbox_alloc_rx_pkt(&ar_sdio->rx_pkts[pkt_cnt],
act_len,
full_len,
last_in_bundle,
last_in_bundle);
+ pkt_cnt++;
}
- ar_sdio->n_rx_pkts = i;
+ ar_sdio->n_rx_pkts = pkt_cnt;
return 0;
@@ -655,41 +670,71 @@ static int ath10k_sdio_mbox_rx_alloc(struct ath10k *ar,
return ret;
}
-static int ath10k_sdio_mbox_rx_packet(struct ath10k *ar,
- struct ath10k_sdio_rx_data *pkt)
+static int ath10k_sdio_mbox_rx_fetch(struct ath10k *ar)
{
struct ath10k_sdio *ar_sdio = ath10k_sdio_priv(ar);
- struct sk_buff *skb = pkt->skb;
+ struct ath10k_sdio_rx_data *pkt = &ar_sdio->rx_pkts[0];
+ struct sk_buff *skb;
int ret;
+ skb = pkt->skb;
ret = ath10k_sdio_read(ar, ar_sdio->mbox_info.htc_addr,
skb->data, pkt->alloc_len, false);
- pkt->status = ret;
- if (!ret)
+ if (ret) {
+ ar_sdio->n_rx_pkts = 0;
+ ath10k_sdio_mbox_free_rx_pkt(pkt);
+ } else {
+ pkt->status = ret;
skb_put(skb, pkt->act_len);
+ }
return ret;
}
-static int ath10k_sdio_mbox_rx_fetch(struct ath10k *ar)
+static int ath10k_sdio_mbox_rx_fetch_bundle(struct ath10k *ar)
{
struct ath10k_sdio *ar_sdio = ath10k_sdio_priv(ar);
+ struct ath10k_sdio_rx_data *pkt;
int ret, i;
+ u32 pkt_offset, virt_pkt_len;
+ virt_pkt_len = 0;
for (i = 0; i < ar_sdio->n_rx_pkts; i++) {
- ret = ath10k_sdio_mbox_rx_packet(ar,
- &ar_sdio->rx_pkts[i]);
- if (ret)
+ virt_pkt_len += ar_sdio->rx_pkts[i].alloc_len;
+ }
+ if (virt_pkt_len < ATH10K_SDIO_DMA_BUF_SIZE) {
+ ret = ath10k_sdio_read(ar, ar_sdio->mbox_info.htc_addr,
+ ar_sdio->vsg_buffer, virt_pkt_len,
+ false);
+ if (ret) {
+ i = 0;
goto err;
+ }
+ } else {
+ ath10k_err(ar, "size exceeding limit %d\n", virt_pkt_len);
+ }
+
+ pkt_offset = 0;
+ for (i = 0; i < ar_sdio->n_rx_pkts; i++) {
+ struct sk_buff *skb = ar_sdio->rx_pkts[i].skb;
+
+ pkt = &ar_sdio->rx_pkts[i];
+ memcpy(skb->data, ar_sdio->vsg_buffer + pkt_offset,
+ pkt->alloc_len);
+ pkt->status = 0;
+ skb_put(skb, pkt->act_len);
+ pkt_offset += pkt->alloc_len;
}
return 0;
err:
/* Free all packets that was not successfully fetched. */
Change comment to: /* Free all packets */
since all packets are freed and not only those that was not successfully fetched.
- for (; i < ar_sdio->n_rx_pkts; i++)
+ for (i = 0; i < ar_sdio->n_rx_pkts; i++)
ath10k_sdio_mbox_free_rx_pkt(&ar_sdio->rx_pkts[i]);
+ ar_sdio->n_rx_pkts = 0;
+
return ret;
}
@@ -732,7 +777,10 @@ static int ath10k_sdio_mbox_rxmsg_pending_handler(struct ath10k *ar,
*/
*done = false;
- ret = ath10k_sdio_mbox_rx_fetch(ar);
+ if (ar_sdio->n_rx_pkts > 1)
+ ret = ath10k_sdio_mbox_rx_fetch_bundle(ar);
+ else
+ ret = ath10k_sdio_mbox_rx_fetch(ar);
ret is not checked at all (I noticed this is the case in the current code as well).
I think it would be wise to break the loop if error:
if (ret)
break;
/* Process fetched packets. This will potentially update
* n_lookaheads depending on if the packets contain lookahead
@@ -1136,7 +1184,7 @@ static int ath10k_sdio_bmi_get_rx_lookahead(struct ath10k *ar)
MBOX_HOST_INT_STATUS_ADDRESS,
&rx_word);
if (ret) {
- ath10k_warn(ar, "unable to read RX_LOOKAHEAD_VALID: %d\n", ret);
+ ath10k_warn(ar, "unable to read rx_lookahd: %d\n", ret);
Change print to "unable to read RX lookahead: %d\n" as it is more descriptive
return ret;
}
@@ -1480,7 +1528,7 @@ static int ath10k_sdio_hif_tx_sg(struct ath10k *ar, u8 pipe_id,
skb = items[i].transfer_context;
padded_len = ath10k_sdio_calc_txrx_padded_len(ar_sdio,
skb->len);
- skb_trim(skb, padded_len);
+ skb->len = padded_len;
Why this change?
I think the skb_ family of functions is the preferred way to manipulate skb's
/* Write TX data to the end of the mbox address space */
address = ar_sdio->mbox_addr[eid] + ar_sdio->mbox_size[eid] -
@@ -1508,7 +1556,8 @@ static int ath10k_sdio_hif_enable_intrs(struct ath10k *ar)
/* Enable all but CPU interrupts */
regs->int_status_en = FIELD_PREP(MBOX_INT_STATUS_ENABLE_ERROR_MASK, 1) |
FIELD_PREP(MBOX_INT_STATUS_ENABLE_CPU_MASK, 1) |
- FIELD_PREP(MBOX_INT_STATUS_ENABLE_COUNTER_MASK, 1);
+ FIELD_PREP(MBOX_INT_STATUS_ENABLE_COUNTER_MASK,
+ 1);
Is this a checkpatch-fix?
I would recommend creating a separate patch for style issues.
/* NOTE: There are some cases where HIF can do detection of
* pending mbox messages which is disabled now.
@@ -2024,6 +2073,12 @@ static int ath10k_sdio_probe(struct sdio_func *func,
goto err_free_bmi_buf;
}
+ ar_sdio->vsg_buffer = kzalloc(ATH10K_SDIO_VSG_BUF_SIZE, GFP_KERNEL);
+ if (!ar_sdio->vsg_buffer) {
+ ret = -ENOMEM;
+ goto err_free_bmi_buf;
+ }
+
ar_sdio->func = func;
sdio_set_drvdata(func, ar_sdio);
@@ -2081,7 +2136,7 @@ static int ath10k_sdio_probe(struct sdio_func *func,
}
/* TODO: remove this once SDIO support is fully implemented */
- ath10k_warn(ar, "WARNING: ath10k SDIO support is incomplete, don't expect anything to work!\n");
+ ath10k_warn(ar, "WARNING: ath10k SDIO support is experimental\n");
return 0;
@@ -2115,6 +2170,7 @@ static void ath10k_sdio_remove(struct sdio_func *func)
ath10k_core_unregister(ar);
ath10k_core_destroy(ar);
kfree(ar_sdio->dma_buffer);
+ kfree(ar_sdio->vsg_buffer);
}
static const struct sdio_device_id ath10k_sdio_devices[] = {
diff --git a/drivers/net/wireless/ath/ath10k/sdio.h b/drivers/net/wireless/ath/ath10k/sdio.h
index 718b8b7..8b6a86a 100644
--- a/drivers/net/wireless/ath/ath10k/sdio.h
+++ b/drivers/net/wireless/ath/ath10k/sdio.h
@@ -149,8 +149,8 @@ struct ath10k_sdio_irq_proc_regs {
u8 rx_lookahead_valid;
u8 host_int_status2;
u8 gmbox_rx_avail;
- __le32 rx_lookahead[2];
- __le32 rx_gmbox_lookahead_alias[2];
+ __le32 rx_lookahead[2 * ATH10K_HIF_MBOX_NUM_MAX];
+ __le32 int_status_enable;
};
struct ath10k_sdio_irq_enable_regs {
@@ -207,6 +207,7 @@ struct ath10k_sdio {
struct ath10k *ar;
struct ath10k_sdio_irq_data irq_data;
+ u8 *vsg_buffer;
u8 *dma_buffer;
/* protects access to dma_buffer */
