The compiler's interrupt service function preamble has a CLI (and a matching SEI at the end). This seems incorrect. The global interrupts should remain on unless the interrupt function elects to disable. The interrupting device itself will not interrupt again, reentrantly.
All the other ASM and C code I've seen for AVRs does not put a CLI in any Interrupt handler. Doing so can disrupt the latency expected by some other interrupt handler.
Compiler generated interrupt preamble
-
JerryHRoss
- Posts: 41
- Joined: 31 Jan 2008 14:27
Compiler generated interrupt preamble
This is a bug!
There should be no CLI/SEI in the auto-generated interrupt routine. The AVR hardware takes care of that.
Depending on the hardware stack size and interrupt rate, it is possible to overflow the hardware stack by (incorrectly) re-enabling interrupts before executing the RETI.
There should be no CLI/SEI in the auto-generated interrupt routine. The AVR hardware takes care of that.
Depending on the hardware stack size and interrupt rate, it is possible to overflow the hardware stack by (incorrectly) re-enabling interrupts before executing the RETI.
We're 3 versions behind.
I think they are all over at the PIC forums. I see that mikroBasic for PIC v7.0.0.2 has been released (2008-06-09). Yet we have v4.0.0.2stevech wrote:Anyone at mikroElektronika paying attention to the AVR product???
Why don't they either get us up to speed or drop it altogether. This teasing us along with "wait for new version" doesn't cut it.
I've signed up for the beta.
-
johankrava
- Posts: 11
- Joined: 11 Oct 2008 19:44
Any news on this issue?
PopularMicrocontrollers.com - [url=http://popularmicrocontrollers.com/]AVR Microcontrollers[/url]
