Alok Aggarwal wrote: > [...] >>> hsfs_vnops.c: hsched_init() - the cmn_errs don't >>> convey any useful information here. Perhaps emit a >>> single error message indicating quering ldi for >>> maxtransfer failed, resorting to 16k default. >>> >>> >> Sometime you may want to know why something failed and >> a generic error message does not convey any helpful debug >> information. That is why I put different cmn_errs that print >> different messages and error codes for different failures. >> But I think I can make the error messages a little more >> verbose. >> > > These error messages are great for someone dealing with > this code for debug purposes. For a casual user, I don't > see how this provides any value. >
Yes, for a casual user these are of no value. But if a user faces a problem and comes back for for help and generic error message is completely useless to debug his problem. I used to deal with CIM/WBEM stuff in Solaris and I remember the nightmares we had. The user will raise an escalation with CIM failing on something and all the error message had to say was: "CIM ERR FAILED" !! Worse, the problem will be unreproducible on our end. So we had to go through 4 rounds of sending debug binaries to the user that will print error codes and stack traces at different places to ultimately figure out the root cause. Detailed error messages are always welcome from the point of view of diagnosability. If ldi fails for some reason, how do we know whether it is a software or hardware issue without detailed error messages. Regards, Moinak.
