On Mon, 1 Aug 2016 15:51:54 -0400 Robert Foss <robert.foss@xxxxxxxxxxxxx> wrote: > On 2016-08-01 02:10 PM, Al Viro wrote: > > On Mon, Aug 01, 2016 at 01:25:40PM -0400, robert.foss@xxxxxxxxxxxxx wrote: > >> From: Scott James Remnant <scott@xxxxxxxxxx> > >> > >> This patch uses TRACE_EVENT to add tracepoints for the open(), > >> exec() and uselib() syscalls so that ureadahead can cheaply trace > >> the boot sequence to determine what to read to speed up the next. > > > > NAK. No Tracepoints In VFS. Not going to happen - any tracepoint can all too > > easily become a cast-in-stone userland ABI. > > > > Hey Al, > > I'm slightly unfamiliar with this territory, so please forgive my lack > of knowledge of this topic. > > What is the negative side of having tracepoint be a permanent fixture in > the VFS ABI? Well, tracepoints are not vetted like syscalls are, but since they are an interface for userspace tools, they can become just as permanent of a fixture without the forethought of it being something that must be maintained forever. This has already bitten us a couple of times. > And how is VFS different from other subsystems in that regard? The main difference is that Al is the maintainer of VFS and he's the one that has to deal with maintaining a permanent tracepoint. He could give less about other subsystems because it ain't his problem ;-) -- Steve -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
