Friday, July 8 2016
Hours:
0900: Stitching RtFifo into the command flow.
1130: DbgOutRT suite compiles and links for Windows.
1600: Debugging.
Today's task: Stitch together all the pieces to route RT commands to RtFifo and then into the normal item FIFO on the next NRT command.
I found my original approach was too myopic by focusing solely on printf and trace. It was actually easier and cleaner to treat all commands submitted in an RT context as deferred. I copied the command (struct CmdHdr_s) into the RtFifo, then before executing the next NRT command I first call RtFifoRead() to execute any pending RT commands. The exception is the printf commands, which need to be expanded immediately and written to the RtFifo as completed items.
VMware networking has crapped out again. I have made no changes whatsoever, and yet yesterday it worked and today it doesn't. I am using a USB/Ethernet adapter in the workshop, and I need to delete and recreate the network bridge to route traffic to it rather than the WiFi.
No joy on the first attempt to run DbgOutRT. Crashed inside the printf. Possibly a buffer alignment problem introduced by DbgCmd.Flags? It seemed I was suffering from a bad case of versionitis and forgot to rebuild BlisterV4 using the new DbgOut library. Why weren't my version checks catching this?
omap3-isp is now running with DbgOutRT. It still hangs.