Re: [PATCH net] net: qcom/emac: Fix missing clk_disable_unprepare() in error path of emac_probe

[wanghai (M)]

在 2020/8/7 21:38, Timur Tabi 写道:

On 8/6/20 8:54 PM, wanghai (M) wrote:

Thanks for your suggestion. May I fix it like this?


Yes, this is what I had in mind.  Thanks.

Acked-by: Timur Tabi 

.


Thanks for your ack. I just sent a new patch.

"[PATCH net] net: qcom/emac: add missed clk_disable_unprepare in error 
path of emac_clks_phase1_init"

Re: [PATCH net] net: qcom/emac: Fix missing clk_disable_unprepare() in error path of emac_probe

[Timur Tabi]

On 8/6/20 8:54 PM, wanghai (M) wrote:

Thanks for your suggestion. May I fix it like this?


Yes, this is what I had in mind.  Thanks.

Acked-by: Timur Tabi 

Re: [PATCH net] net: qcom/emac: Fix missing clk_disable_unprepare() in error path of emac_probe

[wanghai (M)]

在 2020/8/6 22:23, Timur Tabi 写道:

On 8/6/20 9:06 AM, Wang Hai wrote:

In emac_clks_phase1_init() of emac_probe(), there may be a situation
in which some clk_prepare_enable() succeed and others fail.
If emac_clks_phase1_init() fails, goto err_undo_clocks to clean up
the clk that was successfully clk_prepare_enable().


Good catch, however, I think the proper fix is to fix this in 
emac_clks_phase1_init(), so that if some clocks fail, the other clocks 
are cleaned up and then an error is returned.


.


Thanks for your suggestion. May I fix it like this?

diff --git a/drivers/net/ethernet/qualcomm/emac/emac.c 
b/drivers/net/ethernet/qualcomm/emac/emac.c

index 7520c02eec12..7977ad02a7c6 100644
--- a/drivers/net/ethernet/qualcomm/emac/emac.c
+++ b/drivers/net/ethernet/qualcomm/emac/emac.c
@@ -474,13 +474,25 @@ static int emac_clks_phase1_init(struct 
platform_device *pdev,


    ret = clk_prepare_enable(adpt->clk[EMAC_CLK_CFG_AHB]);
    if (ret)
-   return ret;
+   goto disable_clk_axi;

    ret = clk_set_rate(adpt->clk[EMAC_CLK_HIGH_SPEED], 1920);
    if (ret)
-   return ret;
+   goto disable_clk_cfg_ahb;

-   return clk_prepare_enable(adpt->clk[EMAC_CLK_HIGH_SPEED]);
+   ret = clk_prepare_enable(adpt->clk[EMAC_CLK_HIGH_SPEED]);
+   if (ret)
+   goto disable_clk_cfg_ahb;
+
+   return 0;
+
+disable_clk_cfg_ahb:
+   clk_disable_unprepare(adpt->clk[EMAC_CLK_CFG_AHB]);
+disable_clk_axi:
+   clk_disable_unprepare(adpt->clk[EMAC_CLK_AXI]);
+
+   return ret;
 }

[PATCH net] net: qcom/emac: Fix missing clk_disable_unprepare() in error path of emac_probe

[Wang Hai]

In emac_clks_phase1_init() of emac_probe(), there may be a situation
in which some clk_prepare_enable() succeed and others fail.
If emac_clks_phase1_init() fails, goto err_undo_clocks to clean up
the clk that was successfully clk_prepare_enable().

Fixes: b9b17debc69d ("net: emac: emac gigabit ethernet controller driver")
Reported-by: Hulk Robot 
Signed-off-by: Wang Hai 
---
 drivers/net/ethernet/qualcomm/emac/emac.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/qualcomm/emac/emac.c 
b/drivers/net/ethernet/qualcomm/emac/emac.c
index 20b1b43a0e39..7520c02eec12 100644
--- a/drivers/net/ethernet/qualcomm/emac/emac.c
+++ b/drivers/net/ethernet/qualcomm/emac/emac.c
@@ -628,7 +628,7 @@ static int emac_probe(struct platform_device *pdev)
ret = emac_clks_phase1_init(pdev, adpt);
if (ret) {
dev_err(>dev, "could not initialize clocks\n");
-   goto err_undo_netdev;
+   goto err_undo_clocks;
}
 
netdev->watchdog_timeo = EMAC_WATCHDOG_TIME;
-- 
2.17.1

Re: [PATCH net] net: qcom/emac: Fix missing clk_disable_unprepare() in error path of emac_probe

[Timur Tabi]

On 8/6/20 9:06 AM, Wang Hai wrote:

In emac_clks_phase1_init() of emac_probe(), there may be a situation
in which some clk_prepare_enable() succeed and others fail.
If emac_clks_phase1_init() fails, goto err_undo_clocks to clean up
the clk that was successfully clk_prepare_enable().


Good catch, however, I think the proper fix is to fix this in 
emac_clks_phase1_init(), so that if some clocks fail, the other clocks 
are cleaned up and then an error is returned.