mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Roger Quadros <rogerq@kernel.org>
To: Siddharth Vadapalli <s-vadapalli@ti.com>
Cc: Andrew Lunn <andrew+netdev@lunn.ch>,
	"David S. Miller" <davem@davemloft.net>,
	Eric Dumazet <edumazet@google.com>,
	Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
	Grygorii Strashko <grygorii.strashko@ti.com>,
	srk@ti.com, netdev@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH net] net: ethernet: ti: am65-cpsw: fix freeing IRQ in am65_cpsw_nuss_remove_tx_chns()
Date: Wed, 15 Jan 2025 12:04:17 +0200	[thread overview]
Message-ID: <8776a109-22c3-4c1e-a6a1-7bb0a4c70b06@kernel.org> (raw)
In-Reply-To: <gygqjjyso3p4qgam4fpjdkqidj2lhxldkmaopqg32bw3g4ktpj@43tmtsdexkqv>

Hi Siddharth,

On 15/01/2025 07:18, Siddharth Vadapalli wrote:
> On Tue, Jan 14, 2025 at 06:44:02PM +0200, Roger Quadros wrote:
> 
> Hello Roger,
> 
>> When getting the IRQ we use k3_udma_glue_rx_get_irq() which returns
> 
> You probably meant "k3_udma_glue_tx_get_irq()" instead? It is used to
> assign tx_chn->irq within am65_cpsw_nuss_init_tx_chns() as follows:

Yes I meant tx instead of rx.

> 
> 		tx_chn->irq = k3_udma_glue_tx_get_irq(tx_chn->tx_chn);
> 
> Additionally, following the above section we have:
> 
> 		if (tx_chn->irq < 0) {
> 			dev_err(dev, "Failed to get tx dma irq %d\n",
> 				tx_chn->irq);
> 			ret = tx_chn->irq;
> 			goto err;
> 		}
> 
> Could you please provide details on the code-path which will lead to a
> negative "tx_chn->irq" within "am65_cpsw_nuss_remove_tx_chns()"?
> 
> There seem to be two callers of am65_cpsw_nuss_remove_tx_chns(), namely:
> 1. am65_cpsw_nuss_update_tx_rx_chns()
> 2. am65_cpsw_nuss_suspend()
> Since both of them seem to invoke am65_cpsw_nuss_remove_tx_chns() only
> in the case where am65_cpsw_nuss_init_tx_chns() *did not* error out, it
> appears to me that "tx_chn->irq" will never be negative within
> am65_cpsw_nuss_remove_tx_chns()
> 
> Please let me know if I have overlooked something.

The issue is with am65_cpsw_nuss_update_tx_rx_chns(). It can be called
repeatedly (by user changing number of TX queues) even if previous call
to am65_cpsw_nuss_init_tx_chns() failed.

Please try the below patch to simulate the error condition.

Then do the following
- bring down all network interfaces
- change num TX queues to 2. IRQ for 2nd channel fails.
- change num TX queues to 3. Now we try to free an invalid IRQ and we see warning.

Also I think it is good practice to code for set value than to code
for existing code paths as they can change in the future.

--test patch starts--

diff --git a/drivers/net/ethernet/ti/am65-cpsw-nuss.c b/drivers/net/ethernet/ti/am65-cpsw-nuss.c
index 36c29d3db329..c22cadaaf3d3 100644
--- a/drivers/net/ethernet/ti/am65-cpsw-nuss.c
+++ b/drivers/net/ethernet/ti/am65-cpsw-nuss.c
@@ -155,7 +155,7 @@
 			 NETIF_MSG_IFUP	| NETIF_MSG_PROBE | NETIF_MSG_IFDOWN | \
 			 NETIF_MSG_RX_ERR | NETIF_MSG_TX_ERR)
 
-#define AM65_CPSW_DEFAULT_TX_CHNS	8
+#define AM65_CPSW_DEFAULT_TX_CHNS	1
 #define AM65_CPSW_DEFAULT_RX_CHN_FLOWS	1
 
 /* CPPI streaming packet interface */
@@ -2346,7 +2348,10 @@ static int am65_cpsw_nuss_init_tx_chns(struct am65_cpsw_common *common)
 		tx_chn->dsize_log2 = __fls(hdesc_size_out);
 		WARN_ON(hdesc_size_out != (1 << tx_chn->dsize_log2));
 
-		tx_chn->irq = k3_udma_glue_tx_get_irq(tx_chn->tx_chn);
+		if (i == 1)
+			tx_chn->irq = -ENODEV;
+		else
+			tx_chn->irq = k3_udma_glue_tx_get_irq(tx_chn->tx_chn);
 		if (tx_chn->irq < 0) {
 			dev_err(dev, "Failed to get tx dma irq %d\n",
 				tx_chn->irq);

--
cheers,
-roger

  reply	other threads:[~2025-01-15 10:04 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-01-14 16:44 Roger Quadros
2025-01-14 17:12 ` Simon Horman
2025-01-15  5:18 ` Siddharth Vadapalli
2025-01-15 10:04   ` Roger Quadros [this message]
2025-01-15 10:38     ` Siddharth Vadapalli
2025-01-15 15:49       ` Roger Quadros
2025-01-15 16:38         ` Roger Quadros
2025-01-16  5:15           ` Siddharth Vadapalli
2025-01-16 11:47             ` Roger Quadros
2025-01-16 12:10               ` Siddharth Vadapalli
2025-01-16 13:00                 ` Roger Quadros
2025-01-16  5:09         ` Siddharth Vadapalli

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=8776a109-22c3-4c1e-a6a1-7bb0a4c70b06@kernel.org \
    --to=rogerq@kernel.org \
    --cc=andrew+netdev@lunn.ch \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=grygorii.strashko@ti.com \
    --cc=kuba@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=s-vadapalli@ti.com \
    --cc=srk@ti.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

all inboxes | Powered by JetHome®