libethdrivers,imx: Reintroduce ethif_imx6_init#112
libethdrivers,imx: Reintroduce ethif_imx6_init#112kent-mcleod wants to merge 2 commits intoseL4:masterfrom
Conversation
kent-mcleod
force-pushed
the
kent/imx6/ethernet
branch
from
6cd1706 to
4f68ed7
Compare
This init function is used by ethif_new_lwip_driver and was previously removed in an earlier refactor. This commit reintroduces it. Signed-off-by: Kent McLeod <[email protected]>
Add NIC_CONFIG_NO_CLOCK_SYS flag which prevents the ethernet driver from trying to initialize a clock_sys driver (which requires CCM hardware resources) during initialization. This flag should be used when the clocks are setup independently from the driver. Signed-off-by: Kent McLeod <[email protected]>
kent-mcleod
force-pushed
the
kent/imx6/ethernet
branch
from
4f68ed7 to
b6a47d9
Compare
| ret = -ENODEV; | ||
| return -1; | ||
| } | ||
|
|
There was a problem hiding this comment.
There's a missing return 0 here.
|
As a follow-up, should we have a CI test that a system using LWIP still builds? Since i.MX6 i supported in QEMU, we could later even add a check that this actually works |
This would be the closest: seL4/camkes#13 |
| #include <ethdrivers/imx6.h> | ||
| #include <utils/attribute.h> | ||
| #include <ethdrivers/raw.h> | ||
| #include "../src/plat/imx6/enet.h" |
There was a problem hiding this comment.
If I remember correctly we had to include this here to address a nasty include problem and make these prototypes available to other C file in our driver:
int enet_mdio_read(struct enet *enet, uint16_t phy, uint16_t reg);
int enet_mdio_write(struct enet *enet, uint16_t phy, uint16_t reg, uint16_t data);
Could you leave this in for now. I'd make this a separate issue (#116) than that should be cleaned up.
There was a problem hiding this comment.
It does seem possible to declare just those symbols at the bottom of this file:
int enet_mdio_read(struct enet* enet, uint16_t phy, uint16_t reg);
int enet_mdio_write(struct enet* enet, uint16_t phy, uint16_t reg, uint16_t data);
Adding a relative include into a private header file breaks if these public headers are installed somewhere else which breaks how I'm using the library.
There was a problem hiding this comment.
Yes, sounds feasible, I don't remember why we did not do this. Can you put removing this header here in a separate commit, so this is separate from adding the NIC_CONFIG_NO_CLOCK_SYS flag or adding the interface again?
|
I'm fine with merging this when the style issue and the missing return are addressed. Leaving the include file for now is a nice-to-have thing. |
|
@kent-mcleod: do you want to merge this? |
2 participants
This init function is used by ethif_new_lwip_driver and was previously
removed in an earlier refactor. This commit reintroduces it.
Signed-off-by: Kent McLeod [email protected]