drivers: usb: dwc2: esp32: Add support #85600
Conversation
raffarost
force-pushed
the
feature/usb_otg
branch
from
4353128 to
5843cdc
Compare
|
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. |
zephyrbot
added
manifest
manifest-hal_espressif
DNM (manifest)
raffarost
force-pushed
the
feature/usb_otg
branch
from
5843cdc to
a3360fa
Compare
zephyrbot
added
area: USB
zephyrbot
requested review from
andyross,
ceolin,
dcpleung,
jfischer-no,
kartben,
LucasTambor,
marekmatej,
nashif,
sylvioalves,
tmon-nordic,
uLipe and
wmrsouza
raffarost
force-pushed
the
feature/usb_otg
branch
from
a3360fa to
2d10bca
Compare
| static void udc_dwc2_irq_enable_func_##n(const struct device *dev) \ | ||
| { \ | ||
| IRQ_CONNECT(DT_INST_IRQN(n), \ | ||
| DT_INST_IRQ(n, priority), \ | ||
| udc_dwc2_isr_handler, \ | ||
| DEVICE_DT_INST_GET(n), \ | ||
| DW_IRQ_FLAGS(n)); \ | ||
| \ | ||
| irq_enable(DT_INST_IRQN(n)); \ | ||
| } \ | ||
| \ | ||
| static void udc_dwc2_irq_disable_func_##n(const struct device *dev) \ | ||
| { \ | ||
| irq_disable(DT_INST_IRQN(n)); \ | ||
| } \ | ||
| UDC_DWC2_IRQ_DT_INST_DEFINE(n) \ |
There was a problem hiding this comment.
Looks like a stray change that was not justified in the commit message. Anyway, I do not agree with it, you should use irq_enable()/irq_disable().
There was a problem hiding this comment.
for the time being we are using Espressif's interrupt allocator
I thought about turning it into a quirk API call, but it would be probably disliked as a solution
do you have a different suggestion?
There was a problem hiding this comment.
I do not understand the problem with irq_enable/irq_disable. How was it accepted not to use generic API for your platform? Are you working on fixing it?
There was a problem hiding this comment.
I did some checking, and the current behavior for ESP Xtensa chips is: Zephyr's API is used to control non-muxed interrupt lines (such as wifi, systimer). Most peripherals such as USB allow routing sources through an interrupt matrix and for this, the interrupt controller API is used. I'm not sure how much this deviates from the standard implementation, but I believe significant changes would be needed to otherwise accommodate this.
There was a problem hiding this comment.
@jfischer-no, for esp32 device, we still need the irq calls in there, otherwise it is no-go for now. Would you review it again?
There was a problem hiding this comment.
I still cannot find an explanation as to why your platform cannot use generic functions, or why this has not been done all the time is supported in Zephyr.
There was a problem hiding this comment.
ESP chips use an interrupt controller peripheral to route interrupt sources coming from each peripheral to the CPU's IRQ lines. For this reason we implement an interrupt controller driver and call it's API directly from all ESP drivers. A few peripherals have their interrupt source lines connected directly to the CPU IRQ's, thus having dedicated lines (e.g. wifi). As USB OTG is routed using the Interrupt Matrix peripheral we need to connect it's interrupt source to the IRQ accessing HAL LL calls. As this can't be accomplished using Zephyr's irq_connect() without making significant changes to the way we manage interrupts, we are asking to treat interrupt allocation as a quirk for the time being. We are already discussing internally if we should rework the interrupt system implementation and we did a review recently, so this will still be an ongoing work for us. Also, we are still not implementing static interrupt allocation even for IRQ lines that are static and don't need routing.
samples/subsys/usb/cdc_acm/boards/esp32s3_devkitm_procpu.overlay
Outdated
Show resolved
Hide resolved
|
@jfischer-no |
raffarost
force-pushed
the
feature/usb_otg
branch
from
2d10bca to
f764216
Compare
raffarost
dismissed stale reviews from uLipe, sylvioalves, marekmatej, and wmrsouza
via
efae400
raffarost
force-pushed
the
feature/usb_otg
branch
2 times, most recently
from
efae400 to
5cdf02d
Compare
raffarost
force-pushed
the
feature/usb_otg
branch
from
5cdf02d to
cecbed3
Compare
|
CI fix #92314 |
raffarost
force-pushed
the
feature/usb_otg
branch
from
cecbed3 to
a11a757
Compare
Add USB device driver support for ESP32 devices with USB-OTG peripheral. Signed-off-by: Raffael Rostagno <[email protected]>
Limit RX FIFO size allocation to a percentage of the total SPRAM available, to cap allocation according to memory size. Signed-off-by: Raffael Rostagno <[email protected]>
Add USB-OTG peripheral support to ESP32S3. Signed-off-by: Raffael Rostagno <[email protected]>
Add default zephyr_udc0 label for USB-OTG. Signed-off-by: Raffael Rostagno <[email protected]>
raffarost
force-pushed
the
feature/usb_otg
branch
from
a11a757 to
60ee567
Compare
Add USB mass storage sample overlay for esp32s3_devkitm board. Signed-off-by: Raffael Rostagno <[email protected]>
raffarost
force-pushed
the
feature/usb_otg
branch
from
60ee567 to
de96473
Compare
|
Add USB device driver support for ESP32 devices with USB-OTG peripheral.