On 16/6/26 13:10, Peter Maydell wrote:
On Tue, 9 Jun 2026 at 03:40, Nicholas Righi <[email protected]> wrote:
The I2C interrupts are only routed to the legacy interrupt controller. This
means
that for modern device trees that use the GIC, the interrupts don't work. This
patch
adds a splitter to route the I2C interrupt to both the legacy interrupt
controller and the GIC.
Testing
Add these lines to QEMU invocation
-drive if=none,id=i2c_storage,format=raw,file=eeprom.bin \
-device at24c-eeprom,bus=i2c-bus.1,address=0x50,drive=i2c_storage,rom-size=4096
\
note: eeprom.bin is all zeros
Before this change, running i2c get to read from EEPROM would result in this
i2cget -y 1 0x50
Error: Read failed
After this change, running i2c to read from EEPROM results in this
i2cget -y 1 0x50
0x00
The eeprom can now also be enabled in the device tree. Before the
eeprom driver load would fail due to the read failing
ls -l /sys/bus/i2c/devices/i2c-1/1-0050/ | grep -i eeprom
-rw------- 1 root root 4096 May 17 16:57 eeprom
Signed-off-by: Nicholas Righi <[email protected]>
---
Changes v1 -> v2:
- Use a splitter to route the OR gated i2c interrupt to both the
legacy interrupt controller and the GIC, instead of just replacing
the legacy routing with the GIC routing
---
hw/arm/bcm2835_peripherals.c | 9 +++++++++
hw/arm/bcm2838.c | 4 ++++
include/hw/arm/bcm2835_peripherals.h | 2 ++
include/hw/arm/bcm2838_peripherals.h | 1 +
4 files changed, 16 insertions(+)
Applied to target-arm.next, thanks.
Too late but still it was in my review queue:
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>
