From: Heiner Kallweit <hkallweit1@gmail.com>
To: David Miller <davem@davemloft.net>,
Realtek linux nic maintainers <nic_swsd@realtek.com>
Cc: "netdev@vger.kernel.org" <netdev@vger.kernel.org>,
Eric Dumazet <edumazet@google.com>
Subject: [PATCH net-next] r8169: improve rx buffer allocation
Date: Sun, 20 Jan 2019 11:12:01 +0100 [thread overview]
Message-ID: <c6dfe6aa-c65c-f500-9827-4c466f27525e@gmail.com> (raw)
8 years ago, as part of 6f0333b8fde4 ("r8169: use 50% less ram for RX
ring"), the alignment requirement for rx buffers was silently changed
from 8 bytes to 16 bytes. I found nothing explaining this, also the
chip specs I have only mention an 8 byte requirement.
AFAICS kmalloc_node() guarantees allocated memory to be at least
"long long" aligned, what is 8 bytes on a 32 bit machine.
So we can take this memory as-is and avoid some overhead by changing
the alignment requirement back to 8 bytes.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
---
drivers/net/ethernet/realtek/r8169.c | 18 +++++-------------
1 file changed, 5 insertions(+), 13 deletions(-)
diff --git a/drivers/net/ethernet/realtek/r8169.c b/drivers/net/ethernet/realtek/r8169.c
index e790a4116..970c82694 100644
--- a/drivers/net/ethernet/realtek/r8169.c
+++ b/drivers/net/ethernet/realtek/r8169.c
@@ -5659,11 +5659,6 @@ static inline void rtl8169_mark_to_asic(struct RxDesc *desc)
desc->opts1 = cpu_to_le32(DescOwn | eor | R8169_RX_BUF_SIZE);
}
-static inline void *rtl8169_align(void *data)
-{
- return (void *)ALIGN((long)data, 16);
-}
-
static struct sk_buff *rtl8169_alloc_rx_data(struct rtl8169_private *tp,
struct RxDesc *desc)
{
@@ -5676,15 +5671,13 @@ static struct sk_buff *rtl8169_alloc_rx_data(struct rtl8169_private *tp,
if (!data)
return NULL;
- if (rtl8169_align(data) != data) {
- kfree(data);
- data = kmalloc_node(R8169_RX_BUF_SIZE + 15, GFP_KERNEL, node);
- if (!data)
- return NULL;
+ /* Memory should be properly aligned, but better check. */
+ if (!IS_ALIGNED((unsigned long)data, 8)) {
+ netdev_err_once(tp->dev, "RX buffer not 8-byte-aligned\n");
+ goto err_out;
}
- mapping = dma_map_single(d, rtl8169_align(data), R8169_RX_BUF_SIZE,
- DMA_FROM_DEVICE);
+ mapping = dma_map_single(d, data, R8169_RX_BUF_SIZE, DMA_FROM_DEVICE);
if (unlikely(dma_mapping_error(d, mapping))) {
if (net_ratelimit())
netif_err(tp, drv, tp->dev, "Failed to map RX DMA!\n");
@@ -6290,7 +6283,6 @@ static struct sk_buff *rtl8169_try_rx_copy(void *data,
struct sk_buff *skb;
struct device *d = tp_to_dev(tp);
- data = rtl8169_align(data);
dma_sync_single_for_cpu(d, addr, pkt_size, DMA_FROM_DEVICE);
prefetch(data);
skb = napi_alloc_skb(&tp->napi, pkt_size);
--
2.20.1
next reply other threads:[~2019-01-20 10:12 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-01-20 10:12 Heiner Kallweit [this message]
2019-01-22 22:52 ` [PATCH net-next] r8169: improve rx buffer allocation David Miller
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=c6dfe6aa-c65c-f500-9827-4c466f27525e@gmail.com \
--to=hkallweit1@gmail.com \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=netdev@vger.kernel.org \
--cc=nic_swsd@realtek.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).