Re: [PATCH net v2] net: thunderx: workaround BGX TX Underflow issue

[Heiner Kallweit]

On 30.01.2020 22:32, Robert Jones wrote:
> From: Tim Harvey <tharvey@xxxxxxxxxxxxx>
> 
> While it is not yet understood why a TX underflow can easily occur
> for SGMII interfaces resulting in a TX wedge. It has been found that
> disabling/re-enabling the LMAC resolves the issue.
> 

I replied to your v1 and added some comments. Please see there.


> Signed-off-by: Tim Harvey <tharvey@xxxxxxxxxxxxx>
> Reviewed-by: Robert Jones <rjones@xxxxxxxxxxxxx>
> ---
> Changes in v2:
>  - Changed bgx_register_intr() to a void return
>  - Added pci_free_irq_vectors() calls to free irq if named/allocated
>  - Use snprintf instead of sprintf for irq names
> 
>  drivers/net/ethernet/cavium/thunder/thunder_bgx.c | 59 +++++++++++++++++++++++
>  drivers/net/ethernet/cavium/thunder/thunder_bgx.h |  9 ++++
>  2 files changed, 68 insertions(+)
> 
> diff --git a/drivers/net/ethernet/cavium/thunder/thunder_bgx.c b/drivers/net/ethernet/cavium/thunder/thunder_bgx.c
> index c4f6ec0..cbf8596 100644
> --- a/drivers/net/ethernet/cavium/thunder/thunder_bgx.c
> +++ b/drivers/net/ethernet/cavium/thunder/thunder_bgx.c
> @@ -74,6 +74,7 @@ struct bgx {
>  	struct pci_dev		*pdev;
>  	bool                    is_dlm;
>  	bool                    is_rgx;
> +	char			irq_name[7];
>  };
>  
>  static struct bgx *bgx_vnic[MAX_BGX_THUNDER];
> @@ -1535,6 +1536,53 @@ static int bgx_init_phy(struct bgx *bgx)
>  	return bgx_init_of_phy(bgx);
>  }
>  
> +static irqreturn_t bgx_intr_handler(int irq, void *data)
> +{
> +	struct bgx *bgx = (struct bgx *)data;
> +	struct device *dev = &bgx->pdev->dev;
> +	u64 status, val;
> +	int lmac;
> +
> +	for (lmac = 0; lmac < bgx->lmac_count; lmac++) {
> +		status = bgx_reg_read(bgx, lmac, BGX_GMP_GMI_TXX_INT);
> +		if (status & GMI_TXX_INT_UNDFLW) {
> +			dev_err(dev, "BGX%d lmac%d UNDFLW\n", bgx->bgx_id,
> +				lmac);
> +			val = bgx_reg_read(bgx, lmac, BGX_CMRX_CFG);
> +			val &= ~CMR_EN;
> +			bgx_reg_write(bgx, lmac, BGX_CMRX_CFG, val);
> +			val |= CMR_EN;
> +			bgx_reg_write(bgx, lmac, BGX_CMRX_CFG, val);
> +		}
> +		/* clear interrupts */
> +		bgx_reg_write(bgx, lmac, BGX_GMP_GMI_TXX_INT, status);
> +	}
> +
> +	return IRQ_HANDLED;
> +}
> +
> +static void bgx_register_intr(struct pci_dev *pdev)
> +{
> +	struct bgx *bgx = pci_get_drvdata(pdev);
> +	struct device *dev = &pdev->dev;
> +	int num_vec, ret;
> +
> +	/* Enable MSI-X */
> +	num_vec = pci_msix_vec_count(pdev);
> +	ret = pci_alloc_irq_vectors(pdev, num_vec, num_vec, PCI_IRQ_MSIX);
> +	if (ret < 0) {
> +		dev_err(dev, "Req for #%d msix vectors failed\n", num_vec);
> +		return;
> +	}
> +	snprintf(bgx->irq_name, sizeof(bgx->irqname), "BGX%d", bgx->bgx_id);
> +	ret = request_irq(pci_irq_vector(pdev, GMPX_GMI_TX_INT),
> +			  bgx_intr_handler, 0, bgx->irq_name, bgx);
> +	if (ret) {
> +		if (bgx->irq_name[0])
> +			pci_free_irq_vectors(pdev);
> +	}
> +}
> +
>  static int bgx_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
>  {
>  	int err;
> @@ -1604,6 +1652,8 @@ static int bgx_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
>  
>  	bgx_init_hw(bgx);
>  
> +	bgx_register_intr(pdev);
> +
>  	/* Enable all LMACs */
>  	for (lmac = 0; lmac < bgx->lmac_count; lmac++) {
>  		err = bgx_lmac_enable(bgx, lmac);
> @@ -1614,12 +1664,18 @@ static int bgx_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
>  				bgx_lmac_disable(bgx, --lmac);
>  			goto err_enable;
>  		}
> +
> +		/* enable TX FIFO Underflow interrupt */
> +		bgx_reg_modify(bgx, lmac, BGX_GMP_GMI_TXX_INT_ENA_W1S,
> +			       GMI_TXX_INT_UNDFLW);
>  	}
>  
>  	return 0;
>  
>  err_enable:
>  	bgx_vnic[bgx->bgx_id] = NULL;
> +	if (bgx->irq_name[0])
> +		pci_free_irq_vectors(pdev);
>  err_release_regions:
>  	pci_release_regions(pdev);
>  err_disable_device:
> @@ -1637,6 +1693,9 @@ static void bgx_remove(struct pci_dev *pdev)
>  	for (lmac = 0; lmac < bgx->lmac_count; lmac++)
>  		bgx_lmac_disable(bgx, lmac);
>  
> +	if (bgx->irq_name[0])
> +		pci_free_irq_vectors(pdev);
> +

You free the vectors, however you should free the interrupt before.
And you could avoid having to free the vectors manually by switching
from pci_enable_device() to the managed version pcim_enable_device().

>  	bgx_vnic[bgx->bgx_id] = NULL;
>  	pci_release_regions(pdev);
>  	pci_disable_device(pdev);
> diff --git a/drivers/net/ethernet/cavium/thunder/thunder_bgx.h b/drivers/net/ethernet/cavium/thunder/thunder_bgx.h
> index 2588870..cdea493 100644
> --- a/drivers/net/ethernet/cavium/thunder/thunder_bgx.h
> +++ b/drivers/net/ethernet/cavium/thunder/thunder_bgx.h
> @@ -180,6 +180,15 @@
>  #define BGX_GMP_GMI_TXX_BURST		0x38228
>  #define BGX_GMP_GMI_TXX_MIN_PKT		0x38240
>  #define BGX_GMP_GMI_TXX_SGMII_CTL	0x38300
> +#define BGX_GMP_GMI_TXX_INT		0x38500
> +#define BGX_GMP_GMI_TXX_INT_W1S		0x38508
> +#define BGX_GMP_GMI_TXX_INT_ENA_W1C	0x38510
> +#define BGX_GMP_GMI_TXX_INT_ENA_W1S	0x38518
> +#define  GMI_TXX_INT_PTP_LOST			BIT_ULL(4)
> +#define  GMI_TXX_INT_LATE_COL			BIT_ULL(3)
> +#define  GMI_TXX_INT_XSDEF			BIT_ULL(2)
> +#define  GMI_TXX_INT_XSCOL			BIT_ULL(1)
> +#define  GMI_TXX_INT_UNDFLW			BIT_ULL(0)
>  
>  #define BGX_MSIX_VEC_0_29_ADDR		0x400000 /* +(0..29) << 4 */
>  #define BGX_MSIX_VEC_0_29_CTL		0x400008
>