Avi Weiss created an issue:
https://gitlab.rtems.org/rtems/rtos/rtems/-/issues/5594
## Summary
`bsp_reset` tries to clear ESM channel `4294967295` and then skips channel `0`
(`IOFFHR == 1`)
## Steps to reproduce
I added a test probe:
```c
...
raw=0: 0x00000000, channel=(raw-1)=4294967295 / 0xffffffff
raw=0: *** INSIDE THE "if (esm_irq_channel)" BLOCK ***
handle_esm_errors: channel=4294967295 / 0xffffffff outside range
...
raw=1: 0x00000001, channel=(raw-1)=0 / 0x00000000
raw=1: DID NOT enter if (esm_irq_channel)
...
raw=2: 0x00000002, channel=(raw-1)=1 / 0x00000001
raw=2: *** INSIDE THE "if (esm_irq_channel)" BLOCK ***
handle_esm_errors: clearing SR[0] bit 1, mask=0x00000002
```
This shows that when:
`IOFFHR == 0` - becomes `UINT32_MAX` and enters the `if`
`IOFFHR == 1` - becomes `0` and does not enter the `if` block
`IOFFHR == 2` - works as expected
## Expected behaviour
Correct the logic so expected channels are cleared.
I think the correct logic pattern can be seen in `bsp_interrupt_dispatch()`:
```c
void bsp_interrupt_dispatch(void)
{
while (true) {
uint32_t irqindex;
irqindex = TMS570_VIM.IRQINDEX;
if (irqindex == 0) {
return;
}
bsp_interrupt_handler_dispatch(irqindex - 1);
}
}
```
--
View it on GitLab: https://gitlab.rtems.org/rtems/rtos/rtems/-/issues/5594
You're receiving this email because of your account on gitlab.rtems.org.
_______________________________________________
bugs mailing list
[email protected]
http://lists.rtems.org/mailman/listinfo/bugs
