[PATCH net v3 0/4] net: txgbe: fix MSI and INTx interrupts

netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed

* [PATCH net v3 0/4] net: txgbe: fix MSI and INTx interrupts
@ 2024-07-01  7:14 Jiawen Wu
  2024-07-01  7:14 ` [PATCH net v3 1/4] net: txgbe: initialize num_q_vectors for MSI/INTx interrupts Jiawen Wu
                   ` (4 more replies)
  0 siblings, 5 replies; 10+ messages in thread
From: Jiawen Wu @ 2024-07-01  7:14 UTC (permalink / raw)
  To: davem, edumazet, kuba, pabeni, horms, andrew, netdev,
	przemyslaw.kitszel
  Cc: mengyuanlou, duanqiangwen, Jiawen Wu

Fix MSI and INTx interrupts for txgbe driver.

changes in v3:
- Add flag wx->misc_irq_domain.
- Separate commits.
- Detail null-defer events.

changes in v2: https://lore.kernel.org/all/20240626060703.31652-1-jiawenwu@trustnetic.com
- Split into two commits.
- Detail commit description.

v1: https://lore.kernel.org/all/20240621080951.14368-1-jiawenwu@trustnetic.com

Jiawen Wu (4):
  net: txgbe: initialize num_q_vectors for MSI/INTx interrupts
  net: txgbe: remove separate irq request for MSI and INTx
  net: txgbe: add extra handle for MSI/INTx into thread irq handle
  net: txgbe: free isb resources at the right time

 drivers/net/ethernet/wangxun/libwx/wx_hw.c    |   1 +
 drivers/net/ethernet/wangxun/libwx/wx_lib.c   |  10 +-
 drivers/net/ethernet/wangxun/libwx/wx_type.h  |   1 +
 drivers/net/ethernet/wangxun/ngbe/ngbe_main.c |   2 +
 .../net/ethernet/wangxun/txgbe/txgbe_irq.c    | 124 +++++++-----------
 .../net/ethernet/wangxun/txgbe/txgbe_irq.h    |   2 +-
 .../net/ethernet/wangxun/txgbe/txgbe_main.c   |   9 +-
 7 files changed, 64 insertions(+), 85 deletions(-)

-- 
2.27.0


^ permalink raw reply	[flat|nested] 10+ messages in thread

* [PATCH net v3 1/4] net: txgbe: initialize num_q_vectors for MSI/INTx interrupts
  2024-07-01  7:14 [PATCH net v3 0/4] net: txgbe: fix MSI and INTx interrupts Jiawen Wu
@ 2024-07-01  7:14 ` Jiawen Wu
  2024-07-02 15:21   ` Michal Kubiak
  2024-07-01  7:14 ` [PATCH net v3 2/4] net: txgbe: remove separate irq request for MSI and INTx Jiawen Wu
                   ` (3 subsequent siblings)
  4 siblings, 1 reply; 10+ messages in thread
From: Jiawen Wu @ 2024-07-01  7:14 UTC (permalink / raw)
  To: davem, edumazet, kuba, pabeni, horms, andrew, netdev,
	przemyslaw.kitszel
  Cc: mengyuanlou, duanqiangwen, Jiawen Wu

When using MSI/INTx interrupts, wx->num_q_vectors is uninitialized.
Thus there will be kernel panic in wx_alloc_q_vectors() to allocate
queue vectors.

Fixes: 3f703186113f ("net: libwx: Add irq flow functions")
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
---
 drivers/net/ethernet/wangxun/libwx/wx_lib.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.c b/drivers/net/ethernet/wangxun/libwx/wx_lib.c
index 68bde91b67a0..f53776877f71 100644
--- a/drivers/net/ethernet/wangxun/libwx/wx_lib.c
+++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.c
@@ -1686,6 +1686,7 @@ static int wx_set_interrupt_capability(struct wx *wx)
 	}
 
 	pdev->irq = pci_irq_vector(pdev, 0);
+	wx->num_q_vectors = 1;
 
 	return 0;
 }
-- 
2.27.0


^ permalink raw reply related	[flat|nested] 10+ messages in thread

* Re: [PATCH net v3 1/4] net: txgbe: initialize num_q_vectors for MSI/INTx interrupts
  2024-07-01  7:14 ` [PATCH net v3 1/4] net: txgbe: initialize num_q_vectors for MSI/INTx interrupts Jiawen Wu
@ 2024-07-02 15:21   ` Michal Kubiak
  2024-07-03  1:45     ` Jiawen Wu
  0 siblings, 1 reply; 10+ messages in thread
From: Michal Kubiak @ 2024-07-02 15:21 UTC (permalink / raw)
  To: Jiawen Wu
  Cc: davem, edumazet, kuba, pabeni, horms, andrew, netdev,
	przemyslaw.kitszel, mengyuanlou, duanqiangwen

On Mon, Jul 01, 2024 at 03:14:13PM +0800, Jiawen Wu wrote:
> When using MSI/INTx interrupts, wx->num_q_vectors is uninitialized.
> Thus there will be kernel panic in wx_alloc_q_vectors() to allocate
> queue vectors.
> 
> Fixes: 3f703186113f ("net: libwx: Add irq flow functions")
> Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
> ---
>  drivers/net/ethernet/wangxun/libwx/wx_lib.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.c b/drivers/net/ethernet/wangxun/libwx/wx_lib.c
> index 68bde91b67a0..f53776877f71 100644
> --- a/drivers/net/ethernet/wangxun/libwx/wx_lib.c
> +++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.c
> @@ -1686,6 +1686,7 @@ static int wx_set_interrupt_capability(struct wx *wx)
>  	}
>  
>  	pdev->irq = pci_irq_vector(pdev, 0);
> +	wx->num_q_vectors = 1;

I would suggest improving readability of that logic. TBH, initially it wasn't
obvious to me why you assign 1 to num_q_vectors (instead of nvecs variable).
Maybe you just want to exit with an error when nvecs != 1 and avoid some nesting.
I think that should make that logic easier to read. For example:

        /* minmum one for queue, one for misc*/
        nvecs = 1;
        nvecs = pci_alloc_irq_vectors(pdev, nvecs,
                                      nvecs, PCI_IRQ_MSI | PCI_IRQ_INTX);
        if (nvecs != 1) {
                wx_err(wx, "Failed to allocate MSI/INTx interrupts. Error: %d\n", nvecs);
                return nvecs;
        }

        if (pdev->msi_enabled)
                wx_err(wx, "Fallback to MSI.\n");
        else
                wx_err(wx, "Fallback to INTx.\n");

        pdev->irq = pci_irq_vector(pdev, 0);
	wx->num_q_vectors = 1;

(Please consider it as a suggestion only).
>  
>  	return 0;
>  }
> -- 
> 2.27.0
> 
> 

Thanks,
Michal

^ permalink raw reply	[flat|nested] 10+ messages in thread

* RE: [PATCH net v3 1/4] net: txgbe: initialize num_q_vectors for MSI/INTx interrupts
  2024-07-02 15:21   ` Michal Kubiak
@ 2024-07-03  1:45     ` Jiawen Wu
  0 siblings, 0 replies; 10+ messages in thread
From: Jiawen Wu @ 2024-07-03  1:45 UTC (permalink / raw)
  To: 'Michal Kubiak'
  Cc: davem, edumazet, kuba, pabeni, horms, andrew, netdev,
	przemyslaw.kitszel, mengyuanlou, duanqiangwen

On Tue, Jul 2, 2024 11:22 PM, Michal Kubiak wrote:
> On Mon, Jul 01, 2024 at 03:14:13PM +0800, Jiawen Wu wrote:
> > When using MSI/INTx interrupts, wx->num_q_vectors is uninitialized.
> > Thus there will be kernel panic in wx_alloc_q_vectors() to allocate
> > queue vectors.
> >
> > Fixes: 3f703186113f ("net: libwx: Add irq flow functions")
> > Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
> > ---
> >  drivers/net/ethernet/wangxun/libwx/wx_lib.c | 1 +
> >  1 file changed, 1 insertion(+)
> >
> > diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.c b/drivers/net/ethernet/wangxun/libwx/wx_lib.c
> > index 68bde91b67a0..f53776877f71 100644
> > --- a/drivers/net/ethernet/wangxun/libwx/wx_lib.c
> > +++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.c
> > @@ -1686,6 +1686,7 @@ static int wx_set_interrupt_capability(struct wx *wx)
> >  	}
> >
> >  	pdev->irq = pci_irq_vector(pdev, 0);
> > +	wx->num_q_vectors = 1;
> 
> I would suggest improving readability of that logic. TBH, initially it wasn't
> obvious to me why you assign 1 to num_q_vectors (instead of nvecs variable).
> Maybe you just want to exit with an error when nvecs != 1 and avoid some nesting.
> I think that should make that logic easier to read. For example:
> 
>         /* minmum one for queue, one for misc*/
>         nvecs = 1;
>         nvecs = pci_alloc_irq_vectors(pdev, nvecs,
>                                       nvecs, PCI_IRQ_MSI | PCI_IRQ_INTX);
>         if (nvecs != 1) {
>                 wx_err(wx, "Failed to allocate MSI/INTx interrupts. Error: %d\n", nvecs);
>                 return nvecs;
>         }
> 
>         if (pdev->msi_enabled)
>                 wx_err(wx, "Fallback to MSI.\n");
>         else
>                 wx_err(wx, "Fallback to INTx.\n");
> 
>         pdev->irq = pci_irq_vector(pdev, 0);
> 	wx->num_q_vectors = 1;
> 
> (Please consider it as a suggestion only).

Thanks for the suggestion, it can be considered as a commit for code optimization.
 


^ permalink raw reply	[flat|nested] 10+ messages in thread

* [PATCH net v3 2/4] net: txgbe: remove separate irq request for MSI and INTx
  2024-07-01  7:14 [PATCH net v3 0/4] net: txgbe: fix MSI and INTx interrupts Jiawen Wu
  2024-07-01  7:14 ` [PATCH net v3 1/4] net: txgbe: initialize num_q_vectors for MSI/INTx interrupts Jiawen Wu
@ 2024-07-01  7:14 ` Jiawen Wu
  2024-07-03 13:33   ` Michal Kubiak
  2024-07-01  7:14 ` [PATCH net v3 3/4] net: txgbe: add extra handle for MSI/INTx into thread irq handle Jiawen Wu
                   ` (2 subsequent siblings)
  4 siblings, 1 reply; 10+ messages in thread
From: Jiawen Wu @ 2024-07-01  7:14 UTC (permalink / raw)
  To: davem, edumazet, kuba, pabeni, horms, andrew, netdev,
	przemyslaw.kitszel
  Cc: mengyuanlou, duanqiangwen, Jiawen Wu

When using MSI or INTx interrupts, request_irq() for pdev->irq will
conflict with request_threaded_irq() for txgbe->misc.irq, to cause
system crash. So remove txgbe_request_irq() for MSI/INTx case, and
rename txgbe_request_msix_irqs() since it only request for queue irqs.

Add wx->misc_irq_domain to determine whether the driver creates an IRQ
domain and threaded request the IRQs.

Fixes: aefd013624a1 ("net: txgbe: use irq_domain for interrupt controller")
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
---
 drivers/net/ethernet/wangxun/libwx/wx_hw.c    |  1 +
 drivers/net/ethernet/wangxun/libwx/wx_lib.c   |  5 +-
 drivers/net/ethernet/wangxun/libwx/wx_type.h  |  1 +
 .../net/ethernet/wangxun/txgbe/txgbe_irq.c    | 80 ++-----------------
 .../net/ethernet/wangxun/txgbe/txgbe_irq.h    |  2 +-
 .../net/ethernet/wangxun/txgbe/txgbe_main.c   |  2 +-
 6 files changed, 15 insertions(+), 76 deletions(-)

diff --git a/drivers/net/ethernet/wangxun/libwx/wx_hw.c b/drivers/net/ethernet/wangxun/libwx/wx_hw.c
index 7c4b6881a93f..d1b682ce9c6d 100644
--- a/drivers/net/ethernet/wangxun/libwx/wx_hw.c
+++ b/drivers/net/ethernet/wangxun/libwx/wx_hw.c
@@ -1959,6 +1959,7 @@ int wx_sw_init(struct wx *wx)
 	}
 
 	bitmap_zero(wx->state, WX_STATE_NBITS);
+	wx->misc_irq_domain = false;
 
 	return 0;
 }
diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.c b/drivers/net/ethernet/wangxun/libwx/wx_lib.c
index f53776877f71..e1f514b21090 100644
--- a/drivers/net/ethernet/wangxun/libwx/wx_lib.c
+++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.c
@@ -1997,7 +1997,8 @@ void wx_free_irq(struct wx *wx)
 	int vector;
 
 	if (!(pdev->msix_enabled)) {
-		free_irq(pdev->irq, wx);
+		if (!wx->misc_irq_domain)
+			free_irq(pdev->irq, wx);
 		return;
 	}
 
@@ -2012,7 +2013,7 @@ void wx_free_irq(struct wx *wx)
 		free_irq(entry->vector, q_vector);
 	}
 
-	if (wx->mac.type == wx_mac_em)
+	if (!wx->misc_irq_domain)
 		free_irq(wx->msix_entry->vector, wx);
 }
 EXPORT_SYMBOL(wx_free_irq);
diff --git a/drivers/net/ethernet/wangxun/libwx/wx_type.h b/drivers/net/ethernet/wangxun/libwx/wx_type.h
index 5aaf7b1fa2db..0df7f5712b6f 100644
--- a/drivers/net/ethernet/wangxun/libwx/wx_type.h
+++ b/drivers/net/ethernet/wangxun/libwx/wx_type.h
@@ -1058,6 +1058,7 @@ struct wx {
 	dma_addr_t isb_dma;
 	u32 *isb_mem;
 	u32 isb_tag[WX_ISB_MAX];
+	bool misc_irq_domain;
 
 #define WX_MAX_RETA_ENTRIES 128
 #define WX_RSS_INDIR_TBL_MAX 64
diff --git a/drivers/net/ethernet/wangxun/txgbe/txgbe_irq.c b/drivers/net/ethernet/wangxun/txgbe/txgbe_irq.c
index b3e3605d1edb..1490fd6ddbdf 100644
--- a/drivers/net/ethernet/wangxun/txgbe/txgbe_irq.c
+++ b/drivers/net/ethernet/wangxun/txgbe/txgbe_irq.c
@@ -27,57 +27,19 @@ void txgbe_irq_enable(struct wx *wx, bool queues)
 }
 
 /**
- * txgbe_intr - msi/legacy mode Interrupt Handler
- * @irq: interrupt number
- * @data: pointer to a network interface device structure
- **/
-static irqreturn_t txgbe_intr(int __always_unused irq, void *data)
-{
-	struct wx_q_vector *q_vector;
-	struct wx *wx  = data;
-	struct pci_dev *pdev;
-	u32 eicr;
-
-	q_vector = wx->q_vector[0];
-	pdev = wx->pdev;
-
-	eicr = wx_misc_isb(wx, WX_ISB_VEC0);
-	if (!eicr) {
-		/* shared interrupt alert!
-		 * the interrupt that we masked before the ICR read.
-		 */
-		if (netif_running(wx->netdev))
-			txgbe_irq_enable(wx, true);
-		return IRQ_NONE;        /* Not our interrupt */
-	}
-	wx->isb_mem[WX_ISB_VEC0] = 0;
-	if (!(pdev->msi_enabled))
-		wr32(wx, WX_PX_INTA, 1);
-
-	wx->isb_mem[WX_ISB_MISC] = 0;
-	/* would disable interrupts here but it is auto disabled */
-	napi_schedule_irqoff(&q_vector->napi);
-
-	/* re-enable link(maybe) and non-queue interrupts, no flush.
-	 * txgbe_poll will re-enable the queue interrupts
-	 */
-	if (netif_running(wx->netdev))
-		txgbe_irq_enable(wx, false);
-
-	return IRQ_HANDLED;
-}
-
-/**
- * txgbe_request_msix_irqs - Initialize MSI-X interrupts
+ * txgbe_request_queue_irqs - Initialize MSI-X queue interrupts
  * @wx: board private structure
  *
- * Allocate MSI-X vectors and request interrupts from the kernel.
+ * Allocate MSI-X queue vectors and request interrupts from the kernel.
  **/
-static int txgbe_request_msix_irqs(struct wx *wx)
+int txgbe_request_queue_irqs(struct wx *wx)
 {
 	struct net_device *netdev = wx->netdev;
 	int vector, err;
 
+	if (!wx->pdev->msix_enabled)
+		return 0;
+
 	for (vector = 0; vector < wx->num_q_vectors; vector++) {
 		struct wx_q_vector *q_vector = wx->q_vector[vector];
 		struct msix_entry *entry = &wx->msix_q_entries[vector];
@@ -110,34 +72,6 @@ static int txgbe_request_msix_irqs(struct wx *wx)
 	return err;
 }
 
-/**
- * txgbe_request_irq - initialize interrupts
- * @wx: board private structure
- *
- * Attempt to configure interrupts using the best available
- * capabilities of the hardware and kernel.
- **/
-int txgbe_request_irq(struct wx *wx)
-{
-	struct net_device *netdev = wx->netdev;
-	struct pci_dev *pdev = wx->pdev;
-	int err;
-
-	if (pdev->msix_enabled)
-		err = txgbe_request_msix_irqs(wx);
-	else if (pdev->msi_enabled)
-		err = request_irq(wx->pdev->irq, &txgbe_intr, 0,
-				  netdev->name, wx);
-	else
-		err = request_irq(wx->pdev->irq, &txgbe_intr, IRQF_SHARED,
-				  netdev->name, wx);
-
-	if (err)
-		wx_err(wx, "request_irq failed, Error %d\n", err);
-
-	return err;
-}
-
 static int txgbe_request_gpio_irq(struct txgbe *txgbe)
 {
 	txgbe->gpio_irq = irq_find_mapping(txgbe->misc.domain, TXGBE_IRQ_GPIO);
@@ -256,6 +190,8 @@ int txgbe_setup_misc_irq(struct txgbe *txgbe)
 	if (err)
 		goto free_gpio_irq;
 
+	wx->misc_irq_domain = true;
+
 	return 0;
 
 free_gpio_irq:
diff --git a/drivers/net/ethernet/wangxun/txgbe/txgbe_irq.h b/drivers/net/ethernet/wangxun/txgbe/txgbe_irq.h
index b77945e7a0f2..e6285b94625e 100644
--- a/drivers/net/ethernet/wangxun/txgbe/txgbe_irq.h
+++ b/drivers/net/ethernet/wangxun/txgbe/txgbe_irq.h
@@ -2,6 +2,6 @@
 /* Copyright (c) 2015 - 2024 Beijing WangXun Technology Co., Ltd. */
 
 void txgbe_irq_enable(struct wx *wx, bool queues);
-int txgbe_request_irq(struct wx *wx);
+int txgbe_request_queue_irqs(struct wx *wx);
 void txgbe_free_misc_irq(struct txgbe *txgbe);
 int txgbe_setup_misc_irq(struct txgbe *txgbe);
diff --git a/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c b/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c
index 8c7a74981b90..76b5672c0a17 100644
--- a/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c
+++ b/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c
@@ -294,7 +294,7 @@ static int txgbe_open(struct net_device *netdev)
 
 	wx_configure(wx);
 
-	err = txgbe_request_irq(wx);
+	err = txgbe_request_queue_irqs(wx);
 	if (err)
 		goto err_free_isb;
 
-- 
2.27.0


^ permalink raw reply related	[flat|nested] 10+ messages in thread

* Re: [PATCH net v3 2/4] net: txgbe: remove separate irq request for MSI and INTx
  2024-07-01  7:14 ` [PATCH net v3 2/4] net: txgbe: remove separate irq request for MSI and INTx Jiawen Wu
@ 2024-07-03 13:33   ` Michal Kubiak
  2024-07-03 14:53     ` Michal Kubiak
  0 siblings, 1 reply; 10+ messages in thread
From: Michal Kubiak @ 2024-07-03 13:33 UTC (permalink / raw)
  To: Jiawen Wu
  Cc: davem, edumazet, kuba, pabeni, horms, andrew, netdev,
	przemyslaw.kitszel, mengyuanlou, duanqiangwen

On Mon, Jul 01, 2024 at 03:14:14PM +0800, Jiawen Wu wrote:
> When using MSI or INTx interrupts, request_irq() for pdev->irq will
> conflict with request_threaded_irq() for txgbe->misc.irq, to cause
> system crash. So remove txgbe_request_irq() for MSI/INTx case, and
> rename txgbe_request_msix_irqs() since it only request for queue irqs.
> 

Could you please provide any test scenario how the crash can be
reproduced and/or add an example crash log?
Also, could you describe how request_irq() conflicts with
request_threaded_irq() in this case?

> Add wx->misc_irq_domain to determine whether the driver creates an IRQ
> domain and threaded request the IRQs.
> 
> Fixes: aefd013624a1 ("net: txgbe: use irq_domain for interrupt controller")
> Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
> ---
>  drivers/net/ethernet/wangxun/libwx/wx_hw.c    |  1 +
>  drivers/net/ethernet/wangxun/libwx/wx_lib.c   |  5 +-
>  drivers/net/ethernet/wangxun/libwx/wx_type.h  |  1 +
>  .../net/ethernet/wangxun/txgbe/txgbe_irq.c    | 80 ++-----------------
>  .../net/ethernet/wangxun/txgbe/txgbe_irq.h    |  2 +-
>  .../net/ethernet/wangxun/txgbe/txgbe_main.c   |  2 +-
>  6 files changed, 15 insertions(+), 76 deletions(-)
> 
> diff --git a/drivers/net/ethernet/wangxun/libwx/wx_hw.c b/drivers/net/ethernet/wangxun/libwx/wx_hw.c
> index 7c4b6881a93f..d1b682ce9c6d 100644
> --- a/drivers/net/ethernet/wangxun/libwx/wx_hw.c
> +++ b/drivers/net/ethernet/wangxun/libwx/wx_hw.c
> @@ -1959,6 +1959,7 @@ int wx_sw_init(struct wx *wx)
>  	}
>  
>  	bitmap_zero(wx->state, WX_STATE_NBITS);
> +	wx->misc_irq_domain = false;
>  
>  	return 0;
>  }
> diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.c b/drivers/net/ethernet/wangxun/libwx/wx_lib.c
> index f53776877f71..e1f514b21090 100644
> --- a/drivers/net/ethernet/wangxun/libwx/wx_lib.c
> +++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.c
> @@ -1997,7 +1997,8 @@ void wx_free_irq(struct wx *wx)
>  	int vector;
>  
>  	if (!(pdev->msix_enabled)) {
> -		free_irq(pdev->irq, wx);
> +		if (!wx->misc_irq_domain)
> +			free_irq(pdev->irq, wx);

Does it mean "pdev->irq" will never be freed if you set misc_irq_domain
to "true"? It seems you set it to true always during the initializaion,
in "txgbe_setup_misc_irq()"?

>  		return;
>  	}
>  
> @@ -2012,7 +2013,7 @@ void wx_free_irq(struct wx *wx)
>  		free_irq(entry->vector, q_vector);
>  	}
>  
> -	if (wx->mac.type == wx_mac_em)
> +	if (!wx->misc_irq_domain)
>  		free_irq(wx->msix_entry->vector, wx);
>  }
>  EXPORT_SYMBOL(wx_free_irq);
> diff --git a/drivers/net/ethernet/wangxun/libwx/wx_type.h b/drivers/net/ethernet/wangxun/libwx/wx_type.h
> index 5aaf7b1fa2db..0df7f5712b6f 100644
> --- a/drivers/net/ethernet/wangxun/libwx/wx_type.h
> +++ b/drivers/net/ethernet/wangxun/libwx/wx_type.h
> @@ -1058,6 +1058,7 @@ struct wx {
>  	dma_addr_t isb_dma;
>  	u32 *isb_mem;
>  	u32 isb_tag[WX_ISB_MAX];
> +	bool misc_irq_domain;

I don't think that introducing a kind of global variable to determine the call
context is a good idea.
Also, it seems that member is always true after the "probe" context is
completed, isn't it?

> 

[...]

> @@ -256,6 +190,8 @@ int txgbe_setup_misc_irq(struct txgbe *txgbe)
>  	if (err)
>  		goto free_gpio_irq;
>  
> +	wx->misc_irq_domain = true;
> +
>  	return 0;
> 

Is there any chance that member will be set back to "false" after the
initialization is completed?


Thanks,
Michal

^ permalink raw reply	[flat|nested] 10+ messages in thread

* Re: [PATCH net v3 2/4] net: txgbe: remove separate irq request for MSI and INTx
  2024-07-03 13:33   ` Michal Kubiak
@ 2024-07-03 14:53     ` Michal Kubiak
  0 siblings, 0 replies; 10+ messages in thread
From: Michal Kubiak @ 2024-07-03 14:53 UTC (permalink / raw)
  To: Jiawen Wu
  Cc: davem, edumazet, kuba, pabeni, horms, andrew, netdev,
	przemyslaw.kitszel, mengyuanlou, duanqiangwen

On Wed, Jul 03, 2024 at 03:33:24PM +0200, Michal Kubiak wrote:

> >  	if (!(pdev->msix_enabled)) {
> > -		free_irq(pdev->irq, wx);
> > +		if (!wx->misc_irq_domain)
> > +			free_irq(pdev->irq, wx);
> 
> Does it mean "pdev->irq" will never be freed if you set misc_irq_domain
> to "true"? It seems you set it to true always during the initializaion,
> in "txgbe_setup_misc_irq()"?

[...]

> 
> 
> I don't think that introducing a kind of global variable to determine the call
> context is a good idea.
> Also, it seems that member is always true after the "probe" context is
> completed, isn't it?
> 

[...]

> 
> Is there any chance that member will be set back to "false" after the
> initialization is completed?
> 
> 
> Thanks,
> Michal
> 

OK, I think I understood that change better. You probably just want to
distinguish the calling your library function between "ngbe" and "txgbe"
drivers.
In such a case using a structure member in your board structure seems to
be justified. If so, please ignore my questions.

Anyway, I didn't notice that the patch is already applied to the tree
before I sent my comment. Apologies for that.

Thanks,
Michal

^ permalink raw reply	[flat|nested] 10+ messages in thread

* [PATCH net v3 3/4] net: txgbe: add extra handle for MSI/INTx into thread irq handle
  2024-07-01  7:14 [PATCH net v3 0/4] net: txgbe: fix MSI and INTx interrupts Jiawen Wu
  2024-07-01  7:14 ` [PATCH net v3 1/4] net: txgbe: initialize num_q_vectors for MSI/INTx interrupts Jiawen Wu
  2024-07-01  7:14 ` [PATCH net v3 2/4] net: txgbe: remove separate irq request for MSI and INTx Jiawen Wu
@ 2024-07-01  7:14 ` Jiawen Wu
  2024-07-01  7:14 ` [PATCH net v3 4/4] net: txgbe: free isb resources at the right time Jiawen Wu
  2024-07-02 14:10 ` [PATCH net v3 0/4] net: txgbe: fix MSI and INTx interrupts patchwork-bot+netdevbpf
  4 siblings, 0 replies; 10+ messages in thread
From: Jiawen Wu @ 2024-07-01  7:14 UTC (permalink / raw)
  To: davem, edumazet, kuba, pabeni, horms, andrew, netdev,
	przemyslaw.kitszel
  Cc: mengyuanlou, duanqiangwen, Jiawen Wu

Rename original txgbe_misc_irq_handle() to txgbe_misc_irq_thread_fn()
since it is the handle thread to wake up. And add the primary handler
to deal the case of MSI/INTx, because there is a schedule NAPI poll.

Fixes: aefd013624a1 ("net: txgbe: use irq_domain for interrupt controller")
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
---
 .../net/ethernet/wangxun/txgbe/txgbe_irq.c    | 44 ++++++++++++++++---
 1 file changed, 39 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/wangxun/txgbe/txgbe_irq.c b/drivers/net/ethernet/wangxun/txgbe/txgbe_irq.c
index 1490fd6ddbdf..a4cf682dca65 100644
--- a/drivers/net/ethernet/wangxun/txgbe/txgbe_irq.c
+++ b/drivers/net/ethernet/wangxun/txgbe/txgbe_irq.c
@@ -111,6 +111,36 @@ static const struct irq_domain_ops txgbe_misc_irq_domain_ops = {
 };
 
 static irqreturn_t txgbe_misc_irq_handle(int irq, void *data)
+{
+	struct wx_q_vector *q_vector;
+	struct txgbe *txgbe = data;
+	struct wx *wx = txgbe->wx;
+	u32 eicr;
+
+	if (wx->pdev->msix_enabled)
+		return IRQ_WAKE_THREAD;
+
+	eicr = wx_misc_isb(wx, WX_ISB_VEC0);
+	if (!eicr) {
+		/* shared interrupt alert!
+		 * the interrupt that we masked before the ICR read.
+		 */
+		if (netif_running(wx->netdev))
+			txgbe_irq_enable(wx, true);
+		return IRQ_NONE;        /* Not our interrupt */
+	}
+	wx->isb_mem[WX_ISB_VEC0] = 0;
+	if (!(wx->pdev->msi_enabled))
+		wr32(wx, WX_PX_INTA, 1);
+
+	/* would disable interrupts here but it is auto disabled */
+	q_vector = wx->q_vector[0];
+	napi_schedule_irqoff(&q_vector->napi);
+
+	return IRQ_WAKE_THREAD;
+}
+
+static irqreturn_t txgbe_misc_irq_thread_fn(int irq, void *data)
 {
 	struct txgbe *txgbe = data;
 	struct wx *wx = txgbe->wx;
@@ -157,6 +187,7 @@ void txgbe_free_misc_irq(struct txgbe *txgbe)
 
 int txgbe_setup_misc_irq(struct txgbe *txgbe)
 {
+	unsigned long flags = IRQF_ONESHOT;
 	struct wx *wx = txgbe->wx;
 	int hwirq, err;
 
@@ -170,14 +201,17 @@ int txgbe_setup_misc_irq(struct txgbe *txgbe)
 		irq_create_mapping(txgbe->misc.domain, hwirq);
 
 	txgbe->misc.chip = txgbe_irq_chip;
-	if (wx->pdev->msix_enabled)
+	if (wx->pdev->msix_enabled) {
 		txgbe->misc.irq = wx->msix_entry->vector;
-	else
+	} else {
 		txgbe->misc.irq = wx->pdev->irq;
+		if (!wx->pdev->msi_enabled)
+			flags |= IRQF_SHARED;
+	}
 
-	err = request_threaded_irq(txgbe->misc.irq, NULL,
-				   txgbe_misc_irq_handle,
-				   IRQF_ONESHOT,
+	err = request_threaded_irq(txgbe->misc.irq, txgbe_misc_irq_handle,
+				   txgbe_misc_irq_thread_fn,
+				   flags,
 				   wx->netdev->name, txgbe);
 	if (err)
 		goto del_misc_irq;
-- 
2.27.0


^ permalink raw reply related	[flat|nested] 10+ messages in thread

* [PATCH net v3 4/4] net: txgbe: free isb resources at the right time
  2024-07-01  7:14 [PATCH net v3 0/4] net: txgbe: fix MSI and INTx interrupts Jiawen Wu
                   ` (2 preceding siblings ...)
  2024-07-01  7:14 ` [PATCH net v3 3/4] net: txgbe: add extra handle for MSI/INTx into thread irq handle Jiawen Wu
@ 2024-07-01  7:14 ` Jiawen Wu
  2024-07-02 14:10 ` [PATCH net v3 0/4] net: txgbe: fix MSI and INTx interrupts patchwork-bot+netdevbpf
  4 siblings, 0 replies; 10+ messages in thread
From: Jiawen Wu @ 2024-07-01  7:14 UTC (permalink / raw)
  To: davem, edumazet, kuba, pabeni, horms, andrew, netdev,
	przemyslaw.kitszel
  Cc: mengyuanlou, duanqiangwen, Jiawen Wu

When using MSI/INTx interrupt, the shared interrupts are still being
handled in the device remove routine, before free IRQs. So isb memory
is still read after it is freed. Thus move wx_free_isb_resources()
from txgbe_close() to txgbe_remove(). And fix the improper isb free
action in txgbe_open() error handling path.

Fixes: aefd013624a1 ("net: txgbe: use irq_domain for interrupt controller")
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
---
 drivers/net/ethernet/wangxun/libwx/wx_lib.c     | 4 +++-
 drivers/net/ethernet/wangxun/ngbe/ngbe_main.c   | 2 ++
 drivers/net/ethernet/wangxun/txgbe/txgbe_main.c | 7 ++++---
 3 files changed, 9 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.c b/drivers/net/ethernet/wangxun/libwx/wx_lib.c
index e1f514b21090..81bedc8ee8d4 100644
--- a/drivers/net/ethernet/wangxun/libwx/wx_lib.c
+++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.c
@@ -2028,6 +2028,9 @@ int wx_setup_isb_resources(struct wx *wx)
 {
 	struct pci_dev *pdev = wx->pdev;
 
+	if (wx->isb_mem)
+		return 0;
+
 	wx->isb_mem = dma_alloc_coherent(&pdev->dev,
 					 sizeof(u32) * 4,
 					 &wx->isb_dma,
@@ -2387,7 +2390,6 @@ static void wx_free_all_tx_resources(struct wx *wx)
 
 void wx_free_resources(struct wx *wx)
 {
-	wx_free_isb_resources(wx);
 	wx_free_all_rx_resources(wx);
 	wx_free_all_tx_resources(wx);
 }
diff --git a/drivers/net/ethernet/wangxun/ngbe/ngbe_main.c b/drivers/net/ethernet/wangxun/ngbe/ngbe_main.c
index e894e01d030d..af30ca0312b8 100644
--- a/drivers/net/ethernet/wangxun/ngbe/ngbe_main.c
+++ b/drivers/net/ethernet/wangxun/ngbe/ngbe_main.c
@@ -387,6 +387,7 @@ static int ngbe_open(struct net_device *netdev)
 err_free_irq:
 	wx_free_irq(wx);
 err_free_resources:
+	wx_free_isb_resources(wx);
 	wx_free_resources(wx);
 	return err;
 }
@@ -408,6 +409,7 @@ static int ngbe_close(struct net_device *netdev)
 
 	ngbe_down(wx);
 	wx_free_irq(wx);
+	wx_free_isb_resources(wx);
 	wx_free_resources(wx);
 	phylink_disconnect_phy(wx->phylink);
 	wx_control_hw(wx, false);
diff --git a/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c b/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c
index 76b5672c0a17..ca74d9422065 100644
--- a/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c
+++ b/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c
@@ -296,7 +296,7 @@ static int txgbe_open(struct net_device *netdev)
 
 	err = txgbe_request_queue_irqs(wx);
 	if (err)
-		goto err_free_isb;
+		goto err_free_resources;
 
 	/* Notify the stack of the actual queue counts. */
 	err = netif_set_real_num_tx_queues(netdev, wx->num_tx_queues);
@@ -313,8 +313,8 @@ static int txgbe_open(struct net_device *netdev)
 
 err_free_irq:
 	wx_free_irq(wx);
-err_free_isb:
-	wx_free_isb_resources(wx);
+err_free_resources:
+	wx_free_resources(wx);
 err_reset:
 	txgbe_reset(wx);
 
@@ -729,6 +729,7 @@ static void txgbe_remove(struct pci_dev *pdev)
 
 	txgbe_remove_phy(txgbe);
 	txgbe_free_misc_irq(txgbe);
+	wx_free_isb_resources(wx);
 
 	pci_release_selected_regions(pdev,
 				     pci_select_bars(pdev, IORESOURCE_MEM));
-- 
2.27.0


^ permalink raw reply related	[flat|nested] 10+ messages in thread

* Re: [PATCH net v3 0/4] net: txgbe: fix MSI and INTx interrupts
  2024-07-01  7:14 [PATCH net v3 0/4] net: txgbe: fix MSI and INTx interrupts Jiawen Wu
                   ` (3 preceding siblings ...)
  2024-07-01  7:14 ` [PATCH net v3 4/4] net: txgbe: free isb resources at the right time Jiawen Wu
@ 2024-07-02 14:10 ` patchwork-bot+netdevbpf
  4 siblings, 0 replies; 10+ messages in thread
From: patchwork-bot+netdevbpf @ 2024-07-02 14:10 UTC (permalink / raw)
  To: Jiawen Wu
  Cc: davem, edumazet, kuba, pabeni, horms, andrew, netdev,
	przemyslaw.kitszel, mengyuanlou, duanqiangwen

Hello:

This series was applied to netdev/net.git (main)
by Paolo Abeni <pabeni@redhat.com>:

On Mon,  1 Jul 2024 15:14:12 +0800 you wrote:
> Fix MSI and INTx interrupts for txgbe driver.
> 
> changes in v3:
> - Add flag wx->misc_irq_domain.
> - Separate commits.
> - Detail null-defer events.
> 
> [...]

Here is the summary with links:
  - [net,v3,1/4] net: txgbe: initialize num_q_vectors for MSI/INTx interrupts
    https://git.kernel.org/netdev/net/c/7c36711a2cd8
  - [net,v3,2/4] net: txgbe: remove separate irq request for MSI and INTx
    https://git.kernel.org/netdev/net/c/bd07a9817846
  - [net,v3,3/4] net: txgbe: add extra handle for MSI/INTx into thread irq handle
    https://git.kernel.org/netdev/net/c/1e1fa1723eb3
  - [net,v3,4/4] net: txgbe: free isb resources at the right time
    https://git.kernel.org/netdev/net/c/935124dd5883

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html



^ permalink raw reply	[flat|nested] 10+ messages in thread

end of thread, other threads:[~2024-07-03 14:53 UTC | newest]

Thread overview: 10+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-07-01  7:14 [PATCH net v3 0/4] net: txgbe: fix MSI and INTx interrupts Jiawen Wu
2024-07-01  7:14 ` [PATCH net v3 1/4] net: txgbe: initialize num_q_vectors for MSI/INTx interrupts Jiawen Wu
2024-07-02 15:21   ` Michal Kubiak
2024-07-03  1:45     ` Jiawen Wu
2024-07-01  7:14 ` [PATCH net v3 2/4] net: txgbe: remove separate irq request for MSI and INTx Jiawen Wu
2024-07-03 13:33   ` Michal Kubiak
2024-07-03 14:53     ` Michal Kubiak
2024-07-01  7:14 ` [PATCH net v3 3/4] net: txgbe: add extra handle for MSI/INTx into thread irq handle Jiawen Wu
2024-07-01  7:14 ` [PATCH net v3 4/4] net: txgbe: free isb resources at the right time Jiawen Wu
2024-07-02 14:10 ` [PATCH net v3 0/4] net: txgbe: fix MSI and INTx interrupts patchwork-bot+netdevbpf

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).