[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [openrisc] Question on breakpoint method



Heya !
 
1. If you take OpenRISC reference platform (ORP) SoC sources as an example, you will see that OR1200 with enabled *instruction* cache will not see trap instructions set by development interface. Possible ways how to fix this would be to either signal the IC to disable itself automatically when trap is set, invalidate cache line that development interface is seeting, or implement hardware memory cache coherency. The easiest way would be to implement memory cache coherency using a dual port RAM, but that has a consequence some technology issues when targetting ASIC using certain ASIC memory vendors. At the moment for debugging to work properly (ie to set breakpoints), IC needs to be disabled. FYI data cache doesn't have any relation to breakpoints.
 
2. Yes, GDB supports step/next.
 
regards,
Damjan
----- Original Message -----
Sent: Thursday, September 19, 2002 4:51 PM
Subject: [openrisc] Question on breakpoint method

Hi,

When cache is enabled, breakpoints can be set in current or1200 implementation (without error)?
According to current implemetation of or1200, Dbg_IF is attatched between main memory and cache memories.
This means that CPU core never meet the l.trap instrucion if the address of l.trap was allocated to I$ already before breakpoint setting.

Another question!
Is the NEXT(or STEP) function of GDB implemented using breakpoint mechanism (setting l.trap)?

Regards,
maunal



Do you Yahoo!?
New DSL Internet Access from SBC & Yahoo!