| Message ID | 18504549bc4d2b62a72a02cb22a2e4d8e6a58720.1653241224.git.christophe.jaillet@wanadoo.fr (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 1dbe67b9faea0bc340cce894018076679c16cb71 |
| Headers | show |
| Series | dmaengine: lgm: Fix an error handling path in intel_ldma_probe() | expand |
On 22-05-22, 19:41, Christophe JAILLET wrote: > ldma_clk_disable() calls both: > clk_disable_unprepare(d->core_clk); > reset_control_assert(d->rst); > > So, should devm_reset_control_get_optional() fail, core_clk should not > be prepare_enable'd before it, otherwise it will never be > disable_unprepare'd. > > Reorder the code to handle the error handling path as expected. Applied, thanks
diff --git a/drivers/dma/lgm/lgm-dma.c b/drivers/dma/lgm/lgm-dma.c index efe8bd3a0e2a..9b9184f964be 100644 --- a/drivers/dma/lgm/lgm-dma.c +++ b/drivers/dma/lgm/lgm-dma.c @@ -1593,11 +1593,12 @@ static int intel_ldma_probe(struct platform_device *pdev) d->core_clk = devm_clk_get_optional(dev, NULL); if (IS_ERR(d->core_clk)) return PTR_ERR(d->core_clk); - clk_prepare_enable(d->core_clk); d->rst = devm_reset_control_get_optional(dev, NULL); if (IS_ERR(d->rst)) return PTR_ERR(d->rst); + + clk_prepare_enable(d->core_clk); reset_control_deassert(d->rst); ret = devm_add_action_or_reset(dev, ldma_clk_disable, d);
ldma_clk_disable() calls both: clk_disable_unprepare(d->core_clk); reset_control_assert(d->rst); So, should devm_reset_control_get_optional() fail, core_clk should not be prepare_enable'd before it, otherwise it will never be disable_unprepare'd. Reorder the code to handle the error handling path as expected. Fixes: 32d31c79a1a4 ("dmaengine: Add Intel LGM SoC DMA support.") Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> --- drivers/dma/lgm/lgm-dma.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)