On Thu, Jul 28, 2022 at 1:57 PM Jonathan Marek [off-list ref] wrote:

On 7/28/22 5:09 AM, Arnd Bergmann wrote:

quoted

diff --git a/drivers/soundwire/qcom.c b/drivers/soundwire/qcom.c
index 22b706350ead..f65c7737c2db 100644
--- a/drivers/soundwire/qcom.c
+++ b/drivers/soundwire/qcom.c

@@ -1289,11 +1289,7 @@ static int qcom_swrm_probe(struct platform_device *pdev)
         data = of_device_get_match_data(dev);
         ctrl->rows_index = sdw_find_row_index(data->default_rows);
         ctrl->cols_index = sdw_find_col_index(data->default_cols);
-#if IS_REACHABLE(CONFIG_SLIMBUS)
-       if (dev->parent->bus == &slimbus_bus) {
-#else
-       if (false) {
-#endif
+       if (IS_ENABLED(CONFIG_SLIMBUS) && dev->parent->bus == &slimbus_bus) {
                 ctrl->reg_read = qcom_swrm_ahb_reg_read;
                 ctrl->reg_write = qcom_swrm_ahb_reg_write;
                 ctrl->regmap = dev_get_regmap(dev->parent, NULL);

This relies on the compiler optimizing out the reference to slimbus_bus,
which doesn't exist in the CONFIG_SLIMBUS=n case (not tested, but I
think this means an O0 build will fail?).

That is correct: We rely on this everywhere in the kernel, which is why
it is impossible to build kernels with -O0.

       Arnd

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel