* Christoph Hellwig ([EMAIL PROTECTED]) wrote: > On Mon, Aug 27, 2007 at 12:05:44PM -0400, Mathieu Desnoyers wrote: > > Here is the first stage of a port of blktrace to the Linux Kernel Markers. > > The > > advantage of this port is that it minimizes the impact on the running when > > blktrace is not active. > > > > A few remarks : this patch has the positive effect of removing some code > > from the block io tracing hot paths, minimizing the i-cache impact in a > > system where the io tracing is compiled in but inactive. > > > > It also moves the blk tracing code from a header (and therefore from the > > body of the instrumented functions) to a separate C file. > > > > There, as soon as one device has to be traced, all devices have to > > execute the tracing function call when they pass by the instrumentation > > site. > > This is slower than the previous inline function which tested the condition > > quickly. > > > > It does not make the code smaller, since I left all the specialized > > tracing functions for requests, bio, generic, remap, which would go away > > once a generic infrastructure is in place to serialize the information > > passed to the marker. This is mostly why I consider it as a step towards the > > full improvements that could bring the markers. > > I like this as it moves the whole tracing code out of line. It would > be nice if we could make blktrace a module with this, but we'd need > to change the interface away from an ioctl on the block device for that. > > Btw, something that really shows here and what I noticed in my sputrace > aswell is that there is a lot of boilerplate code due to the varargs > trace handlers. We really need some way to auto-generate the boilerplate > for the trace function to avoid coding this up everywhere.
Or we can use a vprintk-like function to parse the format string and serialize the information into trace buffers. I prefer this latter option because, overall, it will localize the probes in a few bytes of functions instead of duplicating the memory and instruction cache required by multiple serializing functions. I have the code ready, but I do not want to flood LKML with patches neither.... Mathieu -- Mathieu Desnoyers Computer Engineering Ph.D. Student, Ecole Polytechnique de Montreal OpenPGP key fingerprint: 8CD5 52C3 8E3C 4140 715F BA06 3F25 A8FE 3BAE 9A68 - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/