Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1524584 > unrolled thread

[PATCH net v2 0/7] net: cpsw: fix leaks and probe deferral

Back to article view | Back to linux.kernel

Contents

  [PATCH net v2 0/7] net: cpsw: fix leaks and probe deferral  Johan Hovold <johan@kernel.org> - 2016-11-17 18:40 +0100
    [PATCH net v2 3/7] net: ethernet: ti: cpsw: fix deferred probe Johan Hovold <johan@kernel.org> - 2016-11-17 18:50 +0100
    [PATCH net v2 5/7] net: ethernet: ti: cpsw: fix secondary-emac probe error path Johan Hovold <johan@kernel.org> - 2016-11-17 18:50 +0100
    [PATCH net v2 1/7] net: ethernet: ti: cpsw: fix bad register access in probe error path Johan Hovold <johan@kernel.org> - 2016-11-17 18:50 +0100
    [PATCH net v2 6/7] net: ethernet: ti: cpsw: add missing sanity check Johan Hovold <johan@kernel.org> - 2016-11-17 18:50 +0100
    Re: [PATCH net v2 0/7] net: cpsw: fix leaks and probe deferral  David Miller <davem@davemloft.net> - 2016-11-18 19:50 +0100

#1524584 — [PATCH net v2 0/7] net: cpsw: fix leaks and probe deferral

This series fixes as number of leaks and issues in the cpsw probe-error
and driver-unbind paths, some which specifically prevented deferred
probing.

Johan


v2
 - Keep platform device runtime-resumed throughout probe instead of
   resuming in the probe error path as suggested by Grygorii (patch
   1/7).

 - Runtime-resume platform device before registering any children in
   order to make sure it is synchronously suspended after deregistering
   children in the error path (patch 3/7).


Johan Hovold (7):
  net: ethernet: ti: cpsw: fix bad register access in probe error path
  net: ethernet: ti: cpsw: fix mdio device reference leak
  net: ethernet: ti: cpsw: fix deferred probe
  net: ethernet: ti: cpsw: fix of_node and phydev leaks
  net: ethernet: ti: cpsw: fix secondary-emac probe error path
  net: ethernet: ti: cpsw: add missing sanity check
  net: ethernet: ti: cpsw: fix fixed-link phy probe deferral

 drivers/net/ethernet/ti/cpsw.c | 95 ++++++++++++++++++++++++++++++++----------
 1 file changed, 74 insertions(+), 21 deletions(-)

-- 
2.7.3

[toc] | [next] | [standalone]

#1524600 — [PATCH net v2 3/7] net: ethernet: ti: cpsw: fix deferred probe

Make sure to deregister all child devices also on probe errors to avoid
leaks and to fix probe deferral:

cpsw 4a100000.ethernet: omap_device: omap_device_enable() called from invalid state 1
cpsw 4a100000.ethernet: use pm_runtime_put_sync_suspend() in driver?
cpsw: probe of 4a100000.ethernet failed with error -22

Add generic helper to undo the effects of cpsw_probe_dt(), which will
also be used in a follow-on patch to fix further leaks that have been
introduced more recently.

Note that the platform device is now runtime-resumed before registering
any child devices in order to make sure that it is synchronously
suspended after having deregistered the children in the error path.

Fixes: 1fb19aa730e4 ("net: cpsw: Add parent<->child relation support
between cpsw and mdio")
Signed-off-by: Johan Hovold <johan@kernel.org>
---
 drivers/net/ethernet/ti/cpsw.c | 41 ++++++++++++++++++++++++-----------------
 1 file changed, 24 insertions(+), 17 deletions(-)

diff --git a/drivers/net/ethernet/ti/cpsw.c b/drivers/net/ethernet/ti/cpsw.c
index 84c5d214557e..5d14abb06486 100644
--- a/drivers/net/ethernet/ti/cpsw.c
+++ b/drivers/net/ethernet/ti/cpsw.c
@@ -2441,6 +2441,11 @@ static int cpsw_probe_dt(struct cpsw_platform_data *data,
 	return 0;
 }
 
+static void cpsw_remove_dt(struct platform_device *pdev)
+{
+	of_platform_depopulate(&pdev->dev);
+}
+
 static int cpsw_probe_dual_emac(struct cpsw_priv *priv)
 {
 	struct cpsw_common		*cpsw = priv->cpsw;
@@ -2585,10 +2590,19 @@ static int cpsw_probe(struct platform_device *pdev)
 	/* Select default pin state */
 	pinctrl_pm_select_default_state(&pdev->dev);
 
+	/* Need to enable clocks with runtime PM api to access module
+	 * registers
+	 */
+	ret = pm_runtime_get_sync(&pdev->dev);
+	if (ret < 0) {
+		pm_runtime_put_noidle(&pdev->dev);
+		goto clean_runtime_disable_ret;
+	}
+
 	if (cpsw_probe_dt(&cpsw->data, pdev)) {
 		dev_err(&pdev->dev, "cpsw: platform data missing\n");
 		ret = -ENODEV;
-		goto clean_runtime_disable_ret;
+		goto clean_dt_ret;
 	}
 	data = &cpsw->data;
 	cpsw->rx_ch_num = 1;
@@ -2609,7 +2623,7 @@ static int cpsw_probe(struct platform_device *pdev)
 				    GFP_KERNEL);
 	if (!cpsw->slaves) {
 		ret = -ENOMEM;
-		goto clean_runtime_disable_ret;
+		goto clean_dt_ret;
 	}
 	for (i = 0; i < data->slaves; i++)
 		cpsw->slaves[i].slave_num = i;
@@ -2621,7 +2635,7 @@ static int cpsw_probe(struct platform_device *pdev)
 	if (IS_ERR(clk)) {
 		dev_err(priv->dev, "fck is not found\n");
 		ret = -ENODEV;
-		goto clean_runtime_disable_ret;
+		goto clean_dt_ret;
 	}
 	cpsw->bus_freq_mhz = clk_get_rate(clk) / 1000000;
 
@@ -2629,25 +2643,17 @@ static int cpsw_probe(struct platform_device *pdev)
 	ss_regs = devm_ioremap_resource(&pdev->dev, ss_res);
 	if (IS_ERR(ss_regs)) {
 		ret = PTR_ERR(ss_regs);
-		goto clean_runtime_disable_ret;
+		goto clean_dt_ret;
 	}
 	cpsw->regs = ss_regs;
 
-	/* Need to enable clocks with runtime PM api to access module
-	 * registers
-	 */
-	ret = pm_runtime_get_sync(&pdev->dev);
-	if (ret < 0) {
-		pm_runtime_put_noidle(&pdev->dev);
-		goto clean_runtime_disable_ret;
-	}
 	cpsw->version = readl(&cpsw->regs->id_ver);
 
 	res = platform_get_resource(pdev, IORESOURCE_MEM, 1);
 	cpsw->wr_regs = devm_ioremap_resource(&pdev->dev, res);
 	if (IS_ERR(cpsw->wr_regs)) {
 		ret = PTR_ERR(cpsw->wr_regs);
-		goto clean_pm_runtime_put_ret;
+		goto clean_dt_ret;
 	}
 
 	memset(&dma_params, 0, sizeof(dma_params));
@@ -2684,7 +2690,7 @@ static int cpsw_probe(struct platform_device *pdev)
 	default:
 		dev_err(priv->dev, "unknown version 0x%08x\n", cpsw->version);
 		ret = -ENODEV;
-		goto clean_pm_runtime_put_ret;
+		goto clean_dt_ret;
 	}
 	for (i = 0; i < cpsw->data.slaves; i++) {
 		struct cpsw_slave *slave = &cpsw->slaves[i];
@@ -2713,7 +2719,7 @@ static int cpsw_probe(struct platform_device *pdev)
 	if (!cpsw->dma) {
 		dev_err(priv->dev, "error initializing dma\n");
 		ret = -ENOMEM;
-		goto clean_pm_runtime_put_ret;
+		goto clean_dt_ret;
 	}
 
 	cpsw->txch[0] = cpdma_chan_create(cpsw->dma, 0, cpsw_tx_handler, 0);
@@ -2823,7 +2829,8 @@ static int cpsw_probe(struct platform_device *pdev)
 	cpsw_ale_destroy(cpsw->ale);
 clean_dma_ret:
 	cpdma_ctlr_destroy(cpsw->dma);
-clean_pm_runtime_put_ret:
+clean_dt_ret:
+	cpsw_remove_dt(pdev);
 	pm_runtime_put_sync(&pdev->dev);
 clean_runtime_disable_ret:
 	pm_runtime_disable(&pdev->dev);
@@ -2850,7 +2857,7 @@ static int cpsw_remove(struct platform_device *pdev)
 
 	cpsw_ale_destroy(cpsw->ale);
 	cpdma_ctlr_destroy(cpsw->dma);
-	of_platform_depopulate(&pdev->dev);
+	cpsw_remove_dt(pdev);
 	pm_runtime_put_sync(&pdev->dev);
 	pm_runtime_disable(&pdev->dev);
 	if (cpsw->data.dual_emac)
-- 
2.7.3

[toc] | [prev] | [next] | [standalone]

#1524610 — [PATCH net v2 5/7] net: ethernet: ti: cpsw: fix secondary-emac probe error path

Make sure to deregister the primary device in case the secondary emac
fails to probe.

kernel BUG at /home/johan/work/omicron/src/linux/net/core/dev.c:7743!
...
[<c05b3dec>] (free_netdev) from [<c04fe6c0>] (cpsw_probe+0x9cc/0xe50)
[<c04fe6c0>] (cpsw_probe) from [<c047b28c>] (platform_drv_probe+0x5c/0xc0)

Fixes: d9ba8f9e6298 ("driver: net: ethernet: cpsw: dual emac interface
implementation")
Signed-off-by: Johan Hovold <johan@kernel.org>
---
 drivers/net/ethernet/ti/cpsw.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/ti/cpsw.c b/drivers/net/ethernet/ti/cpsw.c
index c3b78bc4fe58..11b2daef3158 100644
--- a/drivers/net/ethernet/ti/cpsw.c
+++ b/drivers/net/ethernet/ti/cpsw.c
@@ -2852,7 +2852,7 @@ static int cpsw_probe(struct platform_device *pdev)
 		ret = cpsw_probe_dual_emac(priv);
 		if (ret) {
 			cpsw_err(priv, probe, "error probe slave 2 emac interface\n");
-			goto clean_ale_ret;
+			goto clean_unregister_netdev_ret;
 		}
 	}
 
@@ -2860,6 +2860,8 @@ static int cpsw_probe(struct platform_device *pdev)
 
 	return 0;
 
+clean_unregister_netdev_ret:
+	unregister_netdev(ndev);
 clean_ale_ret:
 	cpsw_ale_destroy(cpsw->ale);
 clean_dma_ret:
-- 
2.7.3

[toc] | [prev] | [next] | [standalone]

#1524611 — [PATCH net v2 1/7] net: ethernet: ti: cpsw: fix bad register access in probe error path

Make sure to keep the platform device runtime-resumed throughout probe
to avoid accessing the CPSW registers in the error path (e.g. for
deferred probe) with clocks disabled:

Unhandled fault: external abort on non-linefetch (0x1008) at 0xd0872d08
...
[<c04fabcc>] (cpsw_ale_control_set) from [<c04fb8b4>] (cpsw_ale_destroy+0x2c/0x44)
[<c04fb8b4>] (cpsw_ale_destroy) from [<c04fea58>] (cpsw_probe+0xbd0/0x10c4)
[<c04fea58>] (cpsw_probe) from [<c047b2a0>] (platform_drv_probe+0x5c/0xc0)

Fixes: df828598a755 ("netdev: driver: ethernet: Add TI CPSW driver")
Signed-off-by: Johan Hovold <johan@kernel.org>
---
 drivers/net/ethernet/ti/cpsw.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/ti/cpsw.c b/drivers/net/ethernet/ti/cpsw.c
index c6cff3d2ff05..f60f8ab7c1e3 100644
--- a/drivers/net/ethernet/ti/cpsw.c
+++ b/drivers/net/ethernet/ti/cpsw.c
@@ -2641,13 +2641,12 @@ static int cpsw_probe(struct platform_device *pdev)
 		goto clean_runtime_disable_ret;
 	}
 	cpsw->version = readl(&cpsw->regs->id_ver);
-	pm_runtime_put_sync(&pdev->dev);
 
 	res = platform_get_resource(pdev, IORESOURCE_MEM, 1);
 	cpsw->wr_regs = devm_ioremap_resource(&pdev->dev, res);
 	if (IS_ERR(cpsw->wr_regs)) {
 		ret = PTR_ERR(cpsw->wr_regs);
-		goto clean_runtime_disable_ret;
+		goto clean_pm_runtime_put_ret;
 	}
 
 	memset(&dma_params, 0, sizeof(dma_params));
@@ -2684,7 +2683,7 @@ static int cpsw_probe(struct platform_device *pdev)
 	default:
 		dev_err(priv->dev, "unknown version 0x%08x\n", cpsw->version);
 		ret = -ENODEV;
-		goto clean_runtime_disable_ret;
+		goto clean_pm_runtime_put_ret;
 	}
 	for (i = 0; i < cpsw->data.slaves; i++) {
 		struct cpsw_slave *slave = &cpsw->slaves[i];
@@ -2713,7 +2712,7 @@ static int cpsw_probe(struct platform_device *pdev)
 	if (!cpsw->dma) {
 		dev_err(priv->dev, "error initializing dma\n");
 		ret = -ENOMEM;
-		goto clean_runtime_disable_ret;
+		goto clean_pm_runtime_put_ret;
 	}
 
 	cpsw->txch[0] = cpdma_chan_create(cpsw->dma, 0, cpsw_tx_handler, 0);
@@ -2815,12 +2814,16 @@ static int cpsw_probe(struct platform_device *pdev)
 		}
 	}
 
+	pm_runtime_put(&pdev->dev);
+
 	return 0;
 
 clean_ale_ret:
 	cpsw_ale_destroy(cpsw->ale);
 clean_dma_ret:
 	cpdma_ctlr_destroy(cpsw->dma);
+clean_pm_runtime_put_ret:
+	pm_runtime_put_sync(&pdev->dev);
 clean_runtime_disable_ret:
 	pm_runtime_disable(&pdev->dev);
 clean_ndev_ret:
-- 
2.7.3

[toc] | [prev] | [next] | [standalone]

#1524612 — [PATCH net v2 6/7] net: ethernet: ti: cpsw: add missing sanity check

Make sure to check for allocation failures before dereferencing a
NULL-pointer during probe.

Fixes: 649a1688c960 ("net: ethernet: ti: cpsw: create common struct to
hold shared driver data")
Signed-off-by: Johan Hovold <johan@kernel.org>
---
 drivers/net/ethernet/ti/cpsw.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/net/ethernet/ti/cpsw.c b/drivers/net/ethernet/ti/cpsw.c
index 11b2daef3158..1387299030e4 100644
--- a/drivers/net/ethernet/ti/cpsw.c
+++ b/drivers/net/ethernet/ti/cpsw.c
@@ -2588,6 +2588,9 @@ static int cpsw_probe(struct platform_device *pdev)
 	int irq;
 
 	cpsw = devm_kzalloc(&pdev->dev, sizeof(struct cpsw_common), GFP_KERNEL);
+	if (!cpsw)
+		return -ENOMEM;
+
 	cpsw->dev = &pdev->dev;
 
 	ndev = alloc_etherdev_mq(sizeof(struct cpsw_priv), CPSW_MAX_QUEUES);
-- 
2.7.3

[toc] | [prev] | [next] | [standalone]

#1525627

From: Johan Hovold <johan@kernel.org>
Date: Thu, 17 Nov 2016 17:39:57 +0100

> This series fixes as number of leaks and issues in the cpsw probe-error
> and driver-unbind paths, some which specifically prevented deferred
> probing.
 ...
> v2
>  - Keep platform device runtime-resumed throughout probe instead of
>    resuming in the probe error path as suggested by Grygorii (patch
>    1/7).
> 
>  - Runtime-resume platform device before registering any children in
>    order to make sure it is synchronously suspended after deregistering
>    children in the error path (patch 3/7).

Series applied, thanks.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web