Hi Takeshi,

I have seen from your last commit that you plan to ack the PS/2 interrupts by 
reading to a register you have added.

Since the CSR bus has no read enable signal, it is good design practice that 
read cycles do not modify the state of your core. Otherwise, when the address 
lines go invalid (which in the Milkymist SoC happens whenever the Wishbone 
address lines are invalid - Wishbone does have a read enable mechanism) they 
can cause an accidental read of your interrupt register and therefore an 
unwanted sporadic acknowledgment of the PS/2 interrupt, leading to hard-to-find 
bugs.

So, you should change your design so that the interrupt is acknowledged when 
the register is *written* (csr_we=1) as it is done in other cores that use 
interrupts.

Best regards,
Sébastien
_______________________________________________
http://lists.milkymist.org/listinfo.cgi/devel-milkymist.org
IRC: #milkym...@freenode
Webchat: www.milkymist.org/irc.html
Wiki: www.milkymist.org/wiki

Reply via email to