Ya, that is interesting. I didn't look at the schematics - I just assumed they were on the same bus. I can access the radio fine and the flash fine seperately, but sometimes accessing both has crashed the mote for me in the past.
And as far as reading vs. writing, I think it would depend on when the actual read and write is performed in relation to using the radio. I have meetings all day today, but I'd like to look into this further when I have some time. The FlashBridge component, by the way, sits on top of PageEEPROM for the mica platforms in TinyOS 1.x. -David -----Original Message----- From: Munaretto, Daniel [mailto:[EMAIL PROTECTED] Sent: Wednesday, September 20, 2006 1:38 AM To: Philip Levis; David Moss Cc: [email protected] Subject: RE: [Tinyos-help] repeating tests,no free memory Well, i'm thinking about..when i tested the FlashBridge interface, i used a mote connected to the programming board, and everything was perfect. And without using the radio, so without send/receive interfaces. i only worked on reading and writing some bytes in flash.Now there are problems with motes, and i'm using batteries of 1.5 V, so 3V. So may be there is a conflict between flash and radio.I will check by tests. By the way, why does this kind of problem rise up only when i read from flash and not also in the writing phase?It doesn't agree with the explanation of the bus usage, right? Daniele -----Original Message----- From: Philip Levis [mailto:[EMAIL PROTECTED] Sent: Wed 9/20/2006 2:54 AM To: David Moss Cc: Munaretto, Daniel; [email protected] Subject: Re: [Tinyos-help] repeating tests,no free memory On Sep 19, 2006, at 3:17 PM, David Moss wrote: > Are you accessing the Flash memory while using the radio? I > believe I ran > into some serious issues doing this with the MicaZ. The problem, I > believe, > is in the bus arbitration - because both the flash and the radio > use the > same SPI bus lines, if you're actively using flash and radio > together on a > micaz, the mote crashes. > That's interesting; I believe that the flash and CC2420 do not share an SPI bus on the mica platforms. > This doesn't happen with telos motes. > > I am about to start testing TinyOS 2.x to see if it still has this > problem. > I believe it shouldn't and doesn't because it has a better bus > resource > arbitration story. i.e. the radio has to request access to the bus > before > using it, and the flash has to request access before using it, etc. Yes, this has been a major focus for 2.x. The bus on the Telos platform is heavily overloaded (SPI to radio, SPI to flash, I2C for sensors, UART, etc.) and so requires a lot of effort to abstract properly in a way that's flexible and robust. It turns out that this actually simplifies code in a lot of other ways, though. For example, the radio stack actually has the receive and transmit paths be separate users of the bus, which individually arbitrate for it. Phil _______________________________________________ Tinyos-help mailing list [email protected] https://mail.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
