Re: [PATCH 4/5] spi: rockchip: Use dev_err_probe() in the probe path

From: Dragan Simic
Date: Thu Sep 26 2024 - 05:21:35 EST


Hello Heiko,

On 2024-09-26 11:00, Heiko Stuebner wrote:
Am Donnerstag, 26. September 2024, 10:38:15 CEST schrieb Dragan Simic:
Use function dev_err_probe() in the probe path instead of dev_err() where
appropriate, to make the code a bit more uniform and compact, and to improve
error handling for the TX and RX DMA channel requests.

Previously, deferred requests for the TX and RX DMA channels produced no
debug messages, and the final error messages didn't include the error codes,
which are all highly useful when debugging permanently failed DMA channel
requests, such as when the required drivers aren't enabled.

Suggested-by: Hélene Vulquin <oss@xxxxxxxxxx>
Signed-off-by: Dragan Simic <dsimic@xxxxxxxxxxx>
---
drivers/spi/spi-rockchip.c | 25 ++++++++++++-------------
1 file changed, 12 insertions(+), 13 deletions(-)

diff --git a/drivers/spi/spi-rockchip.c b/drivers/spi/spi-rockchip.c
index 28879fed03f8..6b5c67a357bb 100644
--- a/drivers/spi/spi-rockchip.c
+++ b/drivers/spi/spi-rockchip.c
@@ -853,22 +853,21 @@ static int rockchip_spi_probe(struct platform_device *pdev)

ctlr->dma_tx = dma_request_chan(rs->dev, "tx");
if (IS_ERR(ctlr->dma_tx)) {
- /* Check tx to see if we need defer probing driver */
- if (PTR_ERR(ctlr->dma_tx) == -EPROBE_DEFER) {
- ret = -EPROBE_DEFER;
+ /* Check tx to see if we need to defer driver probing */
+ ret = dev_err_probe(rs->dev, PTR_ERR(ctlr->dma_tx),
+ "Failed to request TX DMA channel\n");

you're upgrading here from a warning to an error log level.
As it seems the controller may actually provide some level of functionality
even without dma, is this approriate?

Same for rx below.

Thanks for your quick responses.

You're right about the driver still working without the DMA channels,
so emitting warnings would be much more appropriate.

We'd basically need dev_warn_probe() as a new function to cover these
two cases, but I'm not really sure how to proceed? I could go ahead
and implement dev_warn_probe() in a good way, but I wonder what would
be the chances to have that accepted upstream? Perhaps there would
be other users for dev_warn_probe().

+ if (ret == -EPROBE_DEFER)
goto err_disable_pm_runtime;
- }
- dev_warn(rs->dev, "Failed to request TX DMA channel\n");
ctlr->dma_tx = NULL;
}

ctlr->dma_rx = dma_request_chan(rs->dev, "rx");
if (IS_ERR(ctlr->dma_rx)) {
- if (PTR_ERR(ctlr->dma_rx) == -EPROBE_DEFER) {
- ret = -EPROBE_DEFER;
+ /* Check rx to see if we need to defer driver probing */
+ ret = dev_err_probe(rs->dev, PTR_ERR(ctlr->dma_rx),
+ "Failed to request RX DMA channel\n");
+ if (ret == -EPROBE_DEFER)
goto err_free_dma_tx;
- }
- dev_warn(rs->dev, "Failed to request RX DMA channel\n");
ctlr->dma_rx = NULL;
}