From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from 16.mo1.mail-out.ovh.net ([178.33.104.224] helo=mo1.mail-out.ovh.net) by merlin.infradead.org with esmtp (Exim 4.76 #1 (Red Hat Linux)) id 1U3o9Q-0003Y8-E3 for barebox@lists.infradead.org; Fri, 08 Feb 2013 13:38:02 +0000 Received: from mail615.ha.ovh.net (b6.ovh.net [213.186.33.56]) by mo1.mail-out.ovh.net (Postfix) with SMTP id 78F40FFC756 for ; Fri, 8 Feb 2013 14:53:06 +0100 (CET) Date: Fri, 8 Feb 2013 14:36:47 +0100 From: Jean-Christophe PLAGNIOL-VILLARD Message-ID: <20130208133647.GS19322@game.jcrosoft.org> References: <20130208090739.GP19322@game.jcrosoft.org> <1360315130-23872-1-git-send-email-plagnioj@jcrosoft.com> <1360315130-23872-6-git-send-email-plagnioj@jcrosoft.com> <20130208122008.GA7330@x61s.8.8.8.8> MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: <20130208122008.GA7330@x61s.8.8.8.8> List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: barebox-bounces@lists.infradead.org Errors-To: barebox-bounces+u.kleine-koenig=pengutronix.de@lists.infradead.org Subject: Re: [PATCH 6/7] macb: fix tx ring size To: Alexander Aring Cc: barebox@lists.infradead.org, Nicolas Ferre On 13:20 Fri 08 Feb , Alexander Aring wrote: > Hi, > > On Fri, Feb 08, 2013 at 10:18:49AM +0100, Jean-Christophe PLAGNIOL-VILLARD wrote: > > the mininal tx ring size is 2 as if one we wrap on the same descriptor > > and can cause IP lock on GEM (gigabit version) this is always the case > > > > Signed-off-by: Nicolas Ferre > > Signed-off-by: Jean-Christophe PLAGNIOL-VILLARD > > --- > > drivers/net/macb.c | 50 ++++++++++++++++++++++++++++++++++---------------- > > 1 file changed, 34 insertions(+), 16 deletions(-) > > > > diff --git a/drivers/net/macb.c b/drivers/net/macb.c > > index d6b60aa..3a65af7 100644 > > --- a/drivers/net/macb.c > > +++ b/drivers/net/macb.c > > @@ -53,15 +53,17 @@ > > > > #define MACB_RX_BUFFER_SIZE 128 > > #define RX_BUFFER_MULTIPLE 64 /* bytes */ > > -#define RX_RING_SIZE 32 /* must be power of 2 */ > > -#define RX_RING_BYTES (sizeof(struct macb_dma_desc) * RX_RING_SIZE) > > +#define RX_NB_PACKET 10 > > +#define TX_RING_SIZE 2 /* must be power of 2 */ > > > > -#define TX_RING_BYTES (sizeof(struct macb_dma_desc)) > > +#define RX_RING_BYTES(bp) (sizeof(struct macb_dma_desc) * bp->rx_ring_size) > > +#define TX_RING_BYTES (sizeof(struct macb_dma_desc) * TX_RING_SIZE) > > > > struct macb_device { > > void __iomem *regs; > > > > unsigned int rx_tail; > > + unsigned int tx_head; > > unsigned int tx_tail; > > > > void *rx_buffer; > > @@ -86,23 +88,36 @@ static int macb_send(struct eth_device *edev, void *packet, > > { > > struct macb_device *macb = edev->priv; > > unsigned long ctrl; > > - int ret; > > - > > - dev_dbg(macb->dev, "%s\n", __func__); > > + int ret = 0; > > + uint64_t start; > > + unsigned int tx_head = macb->tx_head; > > > > ctrl = MACB_BF(TX_FRMLEN, length); > > - ctrl |= MACB_BIT(TX_LAST) | MACB_BIT(TX_WRAP); > > + ctrl |= MACB_BIT(TX_LAST); > > > > - macb->tx_ring[0].ctrl = ctrl; > > - macb->tx_ring[0].addr = (ulong)packet; > > + if (tx_head == (TX_RING_SIZE - 1)) { > > + ctrl |= MACB_BIT(TX_WRAP); > > + macb->tx_head = 0; > > + } else { > > + macb->tx_head++; > > + } > > + > > + macb->tx_ring[tx_head].ctrl = ctrl; > > + macb->tx_ring[tx_head].addr = (ulong)packet; > > barrier(); > > dma_flush_range((ulong) packet, (ulong)packet + length); > > macb_writel(macb, NCR, MACB_BIT(TE) | MACB_BIT(RE) | MACB_BIT(TSTART)); > > > > - ret = wait_on_timeout(100 * MSECOND, > > - !(macb->tx_ring[0].ctrl & MACB_BIT(TX_USED))); > > - > > - ctrl = macb->tx_ring[0].ctrl; > > + start = get_time_ns(); > > + ret = -ETIMEDOUT; > > Is this ret ever used? for sure we return it if we timeout this is the ret of the send Best Regards, J. > > > + do { > > + barrier(); > > + ctrl = macb->tx_ring[0].ctrl; > > + if (ctrl & MACB_BIT(TX_USED)) { > > + ret = 0; > > + break; > > + } > > + } while (!is_timeout(start, 100 * MSECOND)); > > > > if (ctrl & MACB_BIT(TX_UNDERRUN)) > > dev_err(macb->dev, "TX underrun\n"); > > @@ -244,10 +259,13 @@ static void macb_init(struct macb_device *macb) > > } > > macb->rx_ring[RX_RING_SIZE - 1].addr |= MACB_BIT(RX_WRAP); > > > > - macb->tx_ring[0].addr = 0; > > - macb->tx_ring[0].ctrl = MACB_BIT(TX_USED) | MACB_BIT(TX_WRAP); > > + for (i = 0; i < TX_RING_SIZE; i++) { > > + macb->tx_ring[i].addr = 0; > > + macb->tx_ring[i].ctrl = MACB_BIT(TX_USED); > > + } > > + macb->tx_ring[TX_RING_SIZE - 1].addr |= MACB_BIT(TX_WRAP); > > > > - macb->rx_tail = macb->tx_tail = 0; > > + macb->rx_tail = macb->tx_head = macb->tx_tail = 0; > > > > macb_writel(macb, RBQP, (ulong)macb->rx_ring); > > macb_writel(macb, TBQP, (ulong)macb->tx_ring); > > -- > > 1.7.10.4 > > > > > > _______________________________________________ > > barebox mailing list > > barebox@lists.infradead.org > > http://lists.infradead.org/mailman/listinfo/barebox _______________________________________________ barebox mailing list barebox@lists.infradead.org http://lists.infradead.org/mailman/listinfo/barebox