drivers/net/ethernet/qualcomm/emac/emac-sgmii.c | 21 ++++++++++++++++----- 1 file changed, 16 insertions(+), 5 deletions(-)
From: Zijun Hu <quic_zijuhu@quicinc.com>
To prepare for constifying the following old driver core API:
struct device *device_find_child(struct device *dev, void *data,
int (*match)(struct device *dev, void *data));
to new:
struct device *device_find_child(struct device *dev, const void *data,
int (*match)(struct device *dev, const void *data));
The new API does not allow its match function (*match)() to modify
caller's match data @*data, but emac_sgmii_acpi_match(), as the old
API's match function, indeed modifies relevant match data, so it is
not suitable for the new API any more, solved by implementing the same
finding sgmii_ops function by correcting the function and using it
as parameter of device_for_each_child() instead of device_find_child().
By the way, this commit does not change any existing logic.
Signed-off-by: Zijun Hu <quic_zijuhu@quicinc.com>
---
This patch is separated from the following patch series:
https://lore.kernel.org/all/20240905-const_dfc_prepare-v4-0-4180e1d5a244@quicinc.com/
This patch is to prepare for constifying the following driver API:
struct device *device_find_child(struct device *dev, void *data,
int (*match)(struct device *dev, void *data));
to
struct device *device_find_child(struct device *dev, const void *data,
int (*match)(struct device *dev, const void *data));
How to constify the API ?
There are total 30 usages of the API in current kernel tree:
For 2/30 usages, the API's match function (*match)() will modify
caller's match data @*data, and this patch will clean up one of both.
For remaining 28/30, the following patch series will simply change its
relevant parameter type to const void *.
https://lore.kernel.org/all/20240811-const_dfc_done-v1-1-9d85e3f943cb@quicinc.com/
Why to constify the API ?
(1) It normally does not make sense, also does not need to, for
such device finding operation to modify caller's match data which
is mainly used for comparison.
(2) It will make the API's match function and match data parameter
have the same type as all other APIs (bus|class|driver)_find_device().
(3) It will give driver author hints about choice between this API and
the following one:
int device_for_each_child(struct device *dev, void *data,
int (*fn)(struct device *dev, void *data));
---
Changes in v5:
- Separate me for the series
- Correct commit message and remove the inline comment
- Link to v4: https://lore.kernel.org/r/20240905-const_dfc_prepare-v4-2-4180e1d5a244@quicinc.com
Changes in v4:
- Correct title and commit message
- Link to v3: https://lore.kernel.org/r/20240824-const_dfc_prepare-v3-3-32127ea32bba@quicinc.com
Changes in v3:
- Make qcom/emac follow cxl/region solution suggested by Greg
- Link to v2: https://lore.kernel.org/r/20240815-const_dfc_prepare-v2-0-8316b87b8ff9@quicinc.com
Changes in v2:
- Give up introducing the API constify_device_find_child_helper()
- Implement a driver specific and equivalent one instead of device_find_child()
- Correct commit message
- Link to v1: https://lore.kernel.org/r/20240811-const_dfc_prepare-v1-0-d67cc416b3d3@quicinc.com
---
drivers/net/ethernet/qualcomm/emac/emac-sgmii.c | 21 ++++++++++++++++-----
1 file changed, 16 insertions(+), 5 deletions(-)
diff --git a/drivers/net/ethernet/qualcomm/emac/emac-sgmii.c b/drivers/net/ethernet/qualcomm/emac/emac-sgmii.c
index e4bc18009d08..e8265761c416 100644
--- a/drivers/net/ethernet/qualcomm/emac/emac-sgmii.c
+++ b/drivers/net/ethernet/qualcomm/emac/emac-sgmii.c
@@ -293,6 +293,11 @@ static struct sgmii_ops qdf2400_ops = {
};
#endif
+struct emac_match_data {
+ struct sgmii_ops **sgmii_ops;
+ struct device *target_device;
+};
+
static int emac_sgmii_acpi_match(struct device *dev, void *data)
{
#ifdef CONFIG_ACPI
@@ -303,7 +308,7 @@ static int emac_sgmii_acpi_match(struct device *dev, void *data)
{}
};
const struct acpi_device_id *id = acpi_match_device(match_table, dev);
- struct sgmii_ops **ops = data;
+ struct emac_match_data *match_data = data;
if (id) {
acpi_handle handle = ACPI_HANDLE(dev);
@@ -324,10 +329,12 @@ static int emac_sgmii_acpi_match(struct device *dev, void *data)
switch (hrv) {
case 1:
- *ops = &qdf2432_ops;
+ *match_data->sgmii_ops = &qdf2432_ops;
+ match_data->target_device = get_device(dev);
return 1;
case 2:
- *ops = &qdf2400_ops;
+ *match_data->sgmii_ops = &qdf2400_ops;
+ match_data->target_device = get_device(dev);
return 1;
}
}
@@ -356,10 +363,14 @@ int emac_sgmii_config(struct platform_device *pdev, struct emac_adapter *adpt)
int ret;
if (has_acpi_companion(&pdev->dev)) {
+ struct emac_match_data match_data = {
+ .sgmii_ops = &phy->sgmii_ops,
+ .target_device = NULL,
+ };
struct device *dev;
- dev = device_find_child(&pdev->dev, &phy->sgmii_ops,
- emac_sgmii_acpi_match);
+ device_for_each_child(&pdev->dev, &match_data, emac_sgmii_acpi_match);
+ dev = match_data.target_device;
if (!dev) {
dev_warn(&pdev->dev, "cannot find internal phy node\n");
---
base-commit: 6a36d828bdef0e02b1e6c12e2160f5b83be6aab5
change-id: 20240913-qcom_emac_fix-ec64b8fb8cdd
Best regards,
--
Zijun Hu <quic_zijuhu@quicinc.com>On 9/17/24 03:57, Zijun Hu wrote: > From: Zijun Hu <quic_zijuhu@quicinc.com> > > To prepare for constifying the following old driver core API: > > struct device *device_find_child(struct device *dev, void *data, > int (*match)(struct device *dev, void *data)); > to new: > struct device *device_find_child(struct device *dev, const void *data, > int (*match)(struct device *dev, const void *data)); > > The new API does not allow its match function (*match)() to modify > caller's match data @*data, but emac_sgmii_acpi_match(), as the old > API's match function, indeed modifies relevant match data, so it is > not suitable for the new API any more, solved by implementing the same > finding sgmii_ops function by correcting the function and using it > as parameter of device_for_each_child() instead of device_find_child(). > > By the way, this commit does not change any existing logic. > > Signed-off-by: Zijun Hu <quic_zijuhu@quicinc.com> ## Form letter - net-next-closed The merge window for v6.12 and therefore net-next is closed for new drivers, features, code refactoring and optimizations. We are currently accepting bug fixes only. Please repost when net-next reopens after Sept 30th. RFC patches sent for review only are obviously welcome at any time. See: https://www.kernel.org/doc/html/next/process/maintainer-netdev.html#development-cycle -- pw-bot: defer
On 2024/9/19 16:26, Paolo Abeni wrote: > On 9/17/24 03:57, Zijun Hu wrote: >> From: Zijun Hu <quic_zijuhu@quicinc.com> >> >> To prepare for constifying the following old driver core API: >> >> struct device *device_find_child(struct device *dev, void *data, >> int (*match)(struct device *dev, void *data)); >> to new: >> struct device *device_find_child(struct device *dev, const void *data, >> int (*match)(struct device *dev, const void *data)); >> >> The new API does not allow its match function (*match)() to modify >> caller's match data @*data, but emac_sgmii_acpi_match(), as the old >> API's match function, indeed modifies relevant match data, so it is >> not suitable for the new API any more, solved by implementing the same >> finding sgmii_ops function by correcting the function and using it >> as parameter of device_for_each_child() instead of device_find_child(). >> >> By the way, this commit does not change any existing logic. >> >> Signed-off-by: Zijun Hu <quic_zijuhu@quicinc.com> > > > ## Form letter - net-next-closed > > The merge window for v6.12 and therefore net-next is closed for new > drivers, features, code refactoring and optimizations. We are currently > accepting bug fixes only. > > Please repost when net-next reopens after Sept 30th. > > RFC patches sent for review only are obviously welcome at any time. > thanks for your remainder. will post it after merge window is opened again. and always welcome code reviewers to give comments before that. (^^). > See: > https://www.kernel.org/doc/html/next/process/maintainer-netdev.html#development-cycle
© 2016 - 2024 Red Hat, Inc.