Re: [PATCH] Collecting strings at tracepoints

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Stan" == Stan Shebs <stan_shebs@mentor.com> writes:

Stan> While conceptually simple, we need a new agent bytecode to make this
Stan> work (writing a loop using existing bytecodes doesn't let us check
Stan> that we're running off the edge of valid memory), and a support flag
Stan> so that users get informed if the target doesn't support string
Stan> collection.

Stan> + /* Inspired by trace_kludge, this indicates that pointers to chars
Stan> +    should get an added tracenz bytecode to record nonzero bytes, up to
Stan> +    a length that is the value of string_kludge.  */
Stan> + int string_kludge;

I'd prefer a name that makes it obvious that this is a tracing thing.

Stan> +   { "tracenz", PACKET_DISABLE,
Stan> +     remote_string_tracing_feature, -1 },

I think this requires documentation in the remote protocol section.

Stan> +   while (*exp == ' ' || *exp == '\t')
Stan> +     exp++;

We have skip_spaces now.

Tom