From: Eric Dumazet <eric.dumazet@gmail.com>
To: Amit Salecha <amit.salecha@qlogic.com>,
Matt Carlson <mcarlson@broadcom.com>
Cc: "davem@davemloft.net" <davem@davemloft.net>,
"netdev@vger.kernel.org" <netdev@vger.kernel.org>,
Ameen Rahman <ameen.rahman@qlogic.com>,
Anirban Chakraborty <anirban.chakraborty@qlogic.com>,
Michael Chan <mchan@broadcom.com>
Subject: [PATCH net-next-2.6] tg3: use dma_alloc_coherent() instead of pci_alloc_consistent()
Date: Tue, 26 Oct 2010 14:03:13 +0200 [thread overview]
Message-ID: <1288094593.3169.95.camel@edumazet-laptop> (raw)
In-Reply-To: <1288089033.3169.73.camel@edumazet-laptop>
Le mardi 26 octobre 2010 à 12:30 +0200, Eric Dumazet a écrit :
> By the way, you should use dma_alloc_coherent() instead of
> pci_alloc_consistent() so that you can use GFP_KERNEL allocations
> instead of GFP_ATOMIC, it might help in low memory conditions (if you
> dont hold a spinlock at this point)
>
tg3 being often used as a reference network driver, I believe we should
change its allocations before too many people copy paste suboptimal
pci_alloc_consistent() stuff...
Matt, could you please queue this patch and submit it to David when
net-next-2.6 reopens ?
Thanks
[PATCH] tg3: use dma_alloc_coherent() instead of pci_alloc_consistent()
Using dma_alloc_coherent() permits to use GFP_KERNEL allocations instead
of GFP_ATOMIC ones. Its better when a machine is out of memory, because
this allows driver to sleep to get its memory and succeed its init,
especially when allocating high order pages.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Matt Carlson <mcarlson@broadcom.com>
CC: Michael Chan <mchan@broadcom.com>
---
drivers/net/tg3.c | 73 ++++++++++++++++++++++++--------------------
1 file changed, 41 insertions(+), 32 deletions(-)
diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index 852e917..7dfa579 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -6339,13 +6339,13 @@ static void tg3_rx_prodring_fini(struct tg3 *tp,
kfree(tpr->rx_jmb_buffers);
tpr->rx_jmb_buffers = NULL;
if (tpr->rx_std) {
- pci_free_consistent(tp->pdev, TG3_RX_STD_RING_BYTES(tp),
- tpr->rx_std, tpr->rx_std_mapping);
+ dma_free_coherent(&tp->pdev->dev, TG3_RX_STD_RING_BYTES(tp),
+ tpr->rx_std, tpr->rx_std_mapping);
tpr->rx_std = NULL;
}
if (tpr->rx_jmb) {
- pci_free_consistent(tp->pdev, TG3_RX_JMB_RING_BYTES(tp),
- tpr->rx_jmb, tpr->rx_jmb_mapping);
+ dma_free_coherent(&tp->pdev->dev, TG3_RX_JMB_RING_BYTES(tp),
+ tpr->rx_jmb, tpr->rx_jmb_mapping);
tpr->rx_jmb = NULL;
}
}
@@ -6358,8 +6358,10 @@ static int tg3_rx_prodring_init(struct tg3 *tp,
if (!tpr->rx_std_buffers)
return -ENOMEM;
- tpr->rx_std = pci_alloc_consistent(tp->pdev, TG3_RX_STD_RING_BYTES(tp),
- &tpr->rx_std_mapping);
+ tpr->rx_std = dma_alloc_coherent(&tp->pdev->dev,
+ TG3_RX_STD_RING_BYTES(tp),
+ &tpr->rx_std_mapping,
+ GFP_KERNEL);
if (!tpr->rx_std)
goto err_out;
@@ -6370,9 +6372,10 @@ static int tg3_rx_prodring_init(struct tg3 *tp,
if (!tpr->rx_jmb_buffers)
goto err_out;
- tpr->rx_jmb = pci_alloc_consistent(tp->pdev,
- TG3_RX_JMB_RING_BYTES(tp),
- &tpr->rx_jmb_mapping);
+ tpr->rx_jmb = dma_alloc_coherent(&tp->pdev->dev,
+ TG3_RX_JMB_RING_BYTES(tp),
+ &tpr->rx_jmb_mapping,
+ GFP_KERNEL);
if (!tpr->rx_jmb)
goto err_out;
}
@@ -6491,7 +6494,7 @@ static void tg3_free_consistent(struct tg3 *tp)
struct tg3_napi *tnapi = &tp->napi[i];
if (tnapi->tx_ring) {
- pci_free_consistent(tp->pdev, TG3_TX_RING_BYTES,
+ dma_free_coherent(&tp->pdev->dev, TG3_TX_RING_BYTES,
tnapi->tx_ring, tnapi->tx_desc_mapping);
tnapi->tx_ring = NULL;
}
@@ -6500,25 +6503,26 @@ static void tg3_free_consistent(struct tg3 *tp)
tnapi->tx_buffers = NULL;
if (tnapi->rx_rcb) {
- pci_free_consistent(tp->pdev, TG3_RX_RCB_RING_BYTES(tp),
- tnapi->rx_rcb,
- tnapi->rx_rcb_mapping);
+ dma_free_coherent(&tp->pdev->dev,
+ TG3_RX_RCB_RING_BYTES(tp),
+ tnapi->rx_rcb,
+ tnapi->rx_rcb_mapping);
tnapi->rx_rcb = NULL;
}
tg3_rx_prodring_fini(tp, &tnapi->prodring);
if (tnapi->hw_status) {
- pci_free_consistent(tp->pdev, TG3_HW_STATUS_SIZE,
- tnapi->hw_status,
- tnapi->status_mapping);
+ dma_free_coherent(&tp->pdev->dev, TG3_HW_STATUS_SIZE,
+ tnapi->hw_status,
+ tnapi->status_mapping);
tnapi->hw_status = NULL;
}
}
if (tp->hw_stats) {
- pci_free_consistent(tp->pdev, sizeof(struct tg3_hw_stats),
- tp->hw_stats, tp->stats_mapping);
+ dma_free_coherent(&tp->pdev->dev, sizeof(struct tg3_hw_stats),
+ tp->hw_stats, tp->stats_mapping);
tp->hw_stats = NULL;
}
}
@@ -6531,9 +6535,10 @@ static int tg3_alloc_consistent(struct tg3 *tp)
{
int i;
- tp->hw_stats = pci_alloc_consistent(tp->pdev,
- sizeof(struct tg3_hw_stats),
- &tp->stats_mapping);
+ tp->hw_stats = dma_alloc_coherent(&tp->pdev->dev,
+ sizeof(struct tg3_hw_stats),
+ &tp->stats_mapping,
+ GFP_KERNEL);
if (!tp->hw_stats)
goto err_out;
@@ -6543,9 +6548,10 @@ static int tg3_alloc_consistent(struct tg3 *tp)
struct tg3_napi *tnapi = &tp->napi[i];
struct tg3_hw_status *sblk;
- tnapi->hw_status = pci_alloc_consistent(tp->pdev,
- TG3_HW_STATUS_SIZE,
- &tnapi->status_mapping);
+ tnapi->hw_status = dma_alloc_coherent(&tp->pdev->dev,
+ TG3_HW_STATUS_SIZE,
+ &tnapi->status_mapping,
+ GFP_KERNEL);
if (!tnapi->hw_status)
goto err_out;
@@ -6566,9 +6572,10 @@ static int tg3_alloc_consistent(struct tg3 *tp)
if (!tnapi->tx_buffers)
goto err_out;
- tnapi->tx_ring = pci_alloc_consistent(tp->pdev,
- TG3_TX_RING_BYTES,
- &tnapi->tx_desc_mapping);
+ tnapi->tx_ring = dma_alloc_coherent(&tp->pdev->dev,
+ TG3_TX_RING_BYTES,
+ &tnapi->tx_desc_mapping,
+ GFP_KERNEL);
if (!tnapi->tx_ring)
goto err_out;
}
@@ -6601,9 +6608,10 @@ static int tg3_alloc_consistent(struct tg3 *tp)
if (!i && (tp->tg3_flags3 & TG3_FLG3_ENABLE_RSS))
continue;
- tnapi->rx_rcb = pci_alloc_consistent(tp->pdev,
- TG3_RX_RCB_RING_BYTES(tp),
- &tnapi->rx_rcb_mapping);
+ tnapi->rx_rcb = dma_alloc_coherent(&tp->pdev->dev,
+ TG3_RX_RCB_RING_BYTES(tp),
+ &tnapi->rx_rcb_mapping,
+ GFP_KERNEL);
if (!tnapi->rx_rcb)
goto err_out;
@@ -14159,7 +14167,8 @@ static int __devinit tg3_test_dma(struct tg3 *tp)
u32 *buf, saved_dma_rwctrl;
int ret = 0;
- buf = pci_alloc_consistent(tp->pdev, TEST_BUFFER_SIZE, &buf_dma);
+ buf = dma_alloc_coherent(&tp->pdev->dev, TEST_BUFFER_SIZE,
+ &buf_dma, GFP_KERNEL);
if (!buf) {
ret = -ENOMEM;
goto out_nofree;
@@ -14343,7 +14352,7 @@ static int __devinit tg3_test_dma(struct tg3 *tp)
}
out:
- pci_free_consistent(tp->pdev, TEST_BUFFER_SIZE, buf, buf_dma);
+ dma_free_coherent(&tp->pdev->dev, TEST_BUFFER_SIZE, buf, buf_dma);
out_nofree:
return ret;
}
next prev parent reply other threads:[~2010-10-26 12:03 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-10-26 9:37 [PATCH 0/4]qlcnic: bug fixes Amit Kumar Salecha
2010-10-26 9:37 ` [PATCH] qlcnic: fix mac learning Amit Kumar Salecha
2010-10-26 9:38 ` [PATCH] qlcnic: reduce RX ring size for nic partition Amit Kumar Salecha
2010-10-26 9:38 ` [PATCH] qlcnic: define valid vlan id range Amit Kumar Salecha
2010-10-26 9:38 ` [PATCH] qlcnic: dma address align check Amit Kumar Salecha
2010-10-26 9:54 ` Eric Dumazet
2010-10-26 10:04 ` Amit Salecha
2010-10-26 10:30 ` Eric Dumazet
2010-10-26 10:52 ` Amit Salecha
2010-10-26 12:03 ` Eric Dumazet [this message]
2010-10-26 17:04 ` [PATCH net-next-2.6] tg3: use dma_alloc_coherent() instead of pci_alloc_consistent() Matt Carlson
2010-10-26 15:37 ` [PATCH 0/4]qlcnic: bug fixes 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=1288094593.3169.95.camel@edumazet-laptop \
--to=eric.dumazet@gmail.com \
--cc=ameen.rahman@qlogic.com \
--cc=amit.salecha@qlogic.com \
--cc=anirban.chakraborty@qlogic.com \
--cc=davem@davemloft.net \
--cc=mcarlson@broadcom.com \
--cc=mchan@broadcom.com \
--cc=netdev@vger.kernel.org \
/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