drivers: syscon: bflb efuses #89108
drivers: syscon: bflb efuses #89108
Conversation
|
The following west manifest projects have changed revision in this Pull Request:
✅ All manifest checks OK Note: This message is automatically posted and updated by the Manifest GitHub Action. |
github-actions
bot
added
manifest
manifest-hal_bouffalolab
DNM (manifest)
That is one of the main questions yea, what alternatives is there? |
If you are just reading the eFUSE settings, perhaps the syscon API is better suited? |
|
It looks like it would fit yea... Is there any example driver for it that doesnt just read memory actually? Or would that be a first driver for a syscon interface? |
|
needs rebase |
VynDragon
force-pushed
the
bl60x_efuses
branch
2 times, most recently
from
6515391 to
f9d7b8c
Compare
|
Rebased and made to use syscon. |
VynDragon
force-pushed
the
bl60x_efuses
branch
4 times, most recently
from
4705786 to
b92a322
Compare
github-actions
bot
added
platform: bouffalolab
area: RISCV
github-actions
bot
requested review from
carlocaione,
edersondisouza,
fkokosinski,
katsuster,
kgugala and
mgielda
|
re-assigning to BFLB maintainers |
This sets revision to the one with vendor headers from final hal version Signed-off-by: Camille BAUD <[email protected]>
github-actions
bot
removed
the
DNM (manifest)
VynDragon
force-pushed
the
bl60x_efuses
branch
2 times, most recently
from
ca83c05 to
bd2febe
Compare
| /* 32 Mhz Oscillator: 0 | ||
| * crystal: 1 | ||
| * PLL and 32M: 2 | ||
| * PLL and crystal: 3 | ||
| */ |
There was a problem hiding this comment.
could have used enum for these
enum bflb_root_clock {
BFLB_CLOCK_32M_OSC,
BFLB_CLOCK_XTAL,
BFLB_CLOCK_PLL_32M_OSC,
BFLB_CLOCK_PLL_XTAL,
BFLB_CLOCK_INVALID,
};There was a problem hiding this comment.
Clock driver with this functionality and many others (and enums/defines) is next, but it is dependant on this driver.
There was a problem hiding this comment.
Clock driver with this functionality and many others (and enums/defines) is next
We shouldn't depend on upcoming PRs since they are not guaranteed, but this is reasonably commented so 🤷♂️
| /* invalid value, fallback to internal 32M */ | ||
| if (clock > 3) { | ||
| clock = 0; | ||
| } |
There was a problem hiding this comment.
then this would naturally be:
| /* invalid value, fallback to internal 32M */ | |
| if (clock > 3) { | |
| clock = 0; | |
| } | |
| if (clock >= BFLB_CLOCK_INVALID) { | |
| clock = BFLB_CLOCK_32M_OSC; | |
| } |
There was a problem hiding this comment.
See clock driver comment
| tmp = sys_read32(HBN_BASE + HBN_GLB_OFFSET); | ||
| old_clock_root = (tmp & HBN_ROOT_CLK_SEL_MSK) >> HBN_ROOT_CLK_SEL_POS; | ||
|
|
||
| efuse_bflb_set_root_clock(0); |
There was a problem hiding this comment.
if the enum suggestion above is adopted:
| efuse_bflb_set_root_clock(0); | |
| efuse_bflb_set_root_clock(BFLB_CLOCK_32M_OSC); |
There was a problem hiding this comment.
See clock driver comment
| tmp = sys_read32(HBN_BASE + HBN_GLB_OFFSET); | ||
| old_clock_root = (tmp & HBN_ROOT_CLK_SEL_MSK) >> HBN_ROOT_CLK_SEL_POS; |
There was a problem hiding this comment.
nit: we can have a matching helper efuse_bflb_get_root_clock() function
| tmp = sys_read32(HBN_BASE + HBN_GLB_OFFSET); | |
| old_clock_root = (tmp & HBN_ROOT_CLK_SEL_MSK) >> HBN_ROOT_CLK_SEL_POS; | |
| old_clock_root = efuse_bflb_get_root_clock(); |
There was a problem hiding this comment.
See clock driver comment
This introduces a driver used to access bouffalolab efuses via syscon API Signed-off-by: Camille BAUD <[email protected]>
This enables the driver by default, it will be needed at init in the future Signed-off-by: Camille BAUD <[email protected]>
|
This introduces bflb efuse access via syscon api.
This is not used at the moment but will be in the next steps of integration (for clock_control and things like DAC and ADC).