On Sun, Apr 24, 2022 at 02:57:24PM +0800, Shile Zhang wrote:
On 2022/4/22 23:45, Mathieu Poirier wrote:
On Fri, Apr 22, 2022 at 10:02:39AM +0800, Shile Zhang wrote:
The function etm4_remove_dev() always return 0, and the former function etm4_remove has been changed to void in commit 3fd269e74f2fe ("amba: Make the remove callback return void"). But now its changed back to int type for some reason, which is different to the stable branch linux-5.10.y.
Please spend time understanding why function etm4_remove_dev()'s return value has been changed back to an "int". From there you will likely come to the conclusion that adding the above to the changelog doesn't make sense.
Sorry, I means "some reason" here actually I cannot find out why.
- From the git log of the file
`drivers/hwtracing/coresight/coresight-etm4x-core.c', only log of etm4_remove changes to void in commit 3fd269e74f2fe. no any log record when it change back to int. 2. from the commit 'git log --pretty="%h %ci %cn %s" drivers/hwtracing/coresight/coresight-etm4x-core.c' ... b8336ad947e19 2021-02-04 17:00:32 +0100 Greg Kroah-Hartman coresight: etm4x: add AMBA id for Cortex-A55 and Cortex-A75 3fd269e74f2fe 2021-02-02 14:25:50 +0100 Uwe Kleine-König amba: Make the remove callback return void ...
The commit 'b8336ad947e19' does not change the etm4_remove: https://github.com/torvalds/linux/commit/b8336ad947e1913b9bb5cdf4f54b6876541...
But the different between the commit 'b8336ad947e19' and '3fd269e74f2fe' contains the changes of etm4_remove back to int, as following:
... diff --git a/drivers/hwtracing/coresight/coresight-etm4x-core.c b/drivers/hwtracing/coresight/coresight-etm4x-core.c index 82787cba537d3..8c4b0c46c8f32 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x-core.c +++ b/drivers/hwtracing/coresight/coresight-etm4x-core.c @@ -1680,7 +1680,7 @@ static void clear_etmdrvdata(void *info) etmdrvdata[cpu] = NULL; }
-static void etm4_remove(struct amba_device *adev) +static int etm4_remove(struct amba_device *adev) { struct etmv4_drvdata *drvdata = dev_get_drvdata(&adev->dev);
@@ -1703,6 +1703,8 @@ static void etm4_remove(struct amba_device *adev) cpus_read_unlock();
coresight_unregister(drvdata->csdev);
return 0;}
static const struct amba_id etm4_ids[] = { @@ -1711,6 +1713,8 @@ static const struct amba_id etm4_ids[] = { CS_AMBA_ID(0x000bb95a), /* Cortex-A72 */ CS_AMBA_ID(0x000bb959), /* Cortex-A73 */ CS_AMBA_UCI_ID(0x000bb9da, uci_id_etm4),/* Cortex-A35 */
CS_AMBA_UCI_ID(0x000bbd05, uci_id_etm4),/* Cortex-A55 */CS_AMBA_UCI_ID(0x000bbd0a, uci_id_etm4),/* Cortex-A75 */ CS_AMBA_UCI_ID(0x000bbd0c, uci_id_etm4),/* Neoverse N1 */ CS_AMBA_UCI_ID(0x000f0205, uci_id_etm4),/* Qualcomm Kryo */ CS_AMBA_UCI_ID(0x000f0211, uci_id_etm4),/* Qualcomm Kryo */...
I really don't know how to check which commit change it back. Could you please help to give me some guidance?
Have a look at "git blame" - it is really useful to know what commit introduced a change. In this case, and using coresight next (c86dd9869128), the command would be:
$ git blame -L 2059,2059 drivers/hwtracing/coresight/coresight-etm4x-core.c
That will tell you how function etm4_remove_dev() got to return an 'int' again.
But this is really about understanding the current code rather than trying to understand the history of it. With commit 3fd269e74f2fe, Uwe was making amba remove callback functions return a 'void'. And this is what we see here[1] in v5.10.112. Note that etm4_remove() is the callback to the amba driver.
In the mainline code, the amba callback function is now etm4_remove_amba() and still returns a 'void'. Function etm4_remove() is now etm4_remove_dev() and is called by both etm4_remove_amba() and etm4_remove_platform_dev(), to avoid code duplication.
The conclusion is that because etm4_remove_dev() no longer interfaces with the amba subsystem, its return value can be 'int'. But in this case the return value is not used and as such can be changed to a 'void', regardless of commit 3fd269e74f2fe.
Mathieu
[1]. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/driver...
Thanks!
Just let it return void and return 0 directly in it's caller function etm4_remove_platform_dev.
The only rational for this patch is that etm4_remove_dev() always returns '0'. And even if it was to return anything else, the return value it not checked. And even if the return value was checked, there is nothing to do about an error condition since the driver is being removed.
Signed-off-by: Shile Zhang shile.zhang@linux.alibaba.com
v2: re-work the commit log from Mathieu's suggestion. v1: https://lore.kernel.org/linux-arm-kernel/20220421164217.GB1596562@p14s/T/
drivers/hwtracing/coresight/coresight-etm4x-core.c | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-)
diff --git a/drivers/hwtracing/coresight/coresight-etm4x-core.c b/drivers/hwtracing/coresight/coresight-etm4x-core.c index 7f416a12000eb..141f8209a152a 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x-core.c +++ b/drivers/hwtracing/coresight/coresight-etm4x-core.c @@ -2104,7 +2104,7 @@ static void clear_etmdrvdata(void *info) etmdrvdata[cpu] = NULL; } -static int __exit etm4_remove_dev(struct etmv4_drvdata *drvdata) +static void __exit etm4_remove_dev(struct etmv4_drvdata *drvdata) { etm_perf_symlink(drvdata->csdev, false); /* @@ -2125,8 +2125,6 @@ static int __exit etm4_remove_dev(struct etmv4_drvdata *drvdata) cscfg_unregister_csdev(drvdata->csdev); coresight_unregister(drvdata->csdev);
- return 0; } static void __exit etm4_remove_amba(struct amba_device *adev)
@@ -2139,13 +2137,14 @@ static void __exit etm4_remove_amba(struct amba_device *adev) static int __exit etm4_remove_platform_dev(struct platform_device *pdev) {
- int ret = 0; struct etmv4_drvdata *drvdata = dev_get_drvdata(&pdev->dev); if (drvdata)
ret = etm4_remove_dev(drvdata);
etm4_remove_dev(drvdata);- pm_runtime_disable(&pdev->dev);
- return ret;
- return 0; } static const struct amba_id etm4_ids[] = {
-- 2.33.0.rc2
-
Mathieu Poirier