Unifying kernel tracing [LWN.net]

Unifying kernel tracing

Posted Oct 31, 2019 17:48 UTC (Thu) by karim (subscriber, #114)
Parent article: Unifying kernel tracing

Thanks for writing about this. Great writeup as usual.

A few points about LTT.

This is minor, but I first submitted LTT to the lkml in July 1999: https://lkml.org/lkml/1999/7/23/34

FWIW I couldn't recall Ingo having opposed LTT's inclusion, I had to look that up ;) I certainly remember the "consensus" that I was hearing back from kernel developers was they felt that LTT was a kernel debugging tool and shouldn't therefore be included in the kernel -- so I wouldn't say this is just Ingo and Linus. I do faintly recall that "kernel debugging tool" argument is one I think Linus made to me in person at Usenix 2000 when I first approached him about it, but it's a theme I heard several times over the years despite trying to explain that it was useful for user-space. Some people did get it though, even very early on, as is evident by one of the first replies I got in July 1999 by Werner Almesberger https://lkml.org/lkml/1999/8/7/98

The key sticking point remained the acceptance of trace points into the mainline. A really big kudos goes to Mathieu for getting that in there. Personally I gave up after about 7 years of trying to get trace points included. The following is one of the last things I sent on lkml on this topic: https://lkml.org/lkml/2006/9/17/17

to post comments

Unifying kernel tracing

Posted Oct 31, 2019 18:53 UTC (Thu) by fuhchee (guest, #40059) [Link] (1 responses)

Thanks for that blast from the past, Karim.

Unifying kernel tracing

Posted Nov 3, 2019 17:50 UTC (Sun) by karim (subscriber, #114) [Link]

Hi Frank, happy to be of service ... though rereading myself from back then it felt some of my writing would very likley fail to pass the CoC these days ;)

Unifying kernel tracing

Posted Nov 4, 2019 12:47 UTC (Mon) by nevets (subscriber, #11875) [Link] (1 responses)

Hi Karim,

Considering that I started my talk at the Detroit airport lounge on my way to Lyon, I tried hard to find the exact dates things were released. I saw what looked like the first "official" announcement in 2000, thus I said "around 2000". Which wasn't really incorrect. I tried to be as honest and non-bias as possible in my talk, but still stress all the old "fights" we had, and say I think it's time we all work together.

If there's anything incorrect that I said or showed, I do apologize. It was unintended. I think you know me well enough to know what I was trying to do ;-)

-- Steve

Unifying kernel tracing

Posted Nov 4, 2019 17:23 UTC (Mon) by karim (subscriber, #114) [Link]

Hi Steve,

No, no, ... no need to apologize in any way, this is great. Consider this nitpicking on my part. I wasn't faulting anyone, and certainly not your initiative in trying to unify things. I was reading this writeup and thought I'd add some details.

All that being said I wish I had attended your talk. This is a very valuable exercise. And, yes, you have my absolute trust on this; none of my feedback was meant to undermine that.

Karim