Hi,
I’ve seen a number of discussions (e.g. on OSR) regarding the shortcomings of WPP
tracing. My experience also is that missing messages makes debugging a real pain,
particularly when you don’t have a real chance for live kernel debugging such as
in many embedded systems. (No firewire, often not even a serial port available)
If I design my tracing around WMI is this any more reliable? Or must I also reckon
with lost messages etc. here? Or is logging to a custom buffer (based on I/O
targets) with an additional (user-space) application which can access info logged
to this buffer my best bet?
Charles