Re: truss status

2006-10-25 Thread Daniel O'Connor
On Thursday 26 October 2006 02:42, Darren Reed wrote: > > On a related note, do you think it would be possible to allow ktrace to > > use pipes? > > Have a look at NetBSD, it has something like this...although it is not > without issues as ktrace doesn't stop the program to prevent missing > events

Re: truss status

2006-10-25 Thread Darren Reed
On Mon, Jul 24, 2006 at 06:18:36PM +0930, Daniel O'Connor wrote: > On Monday 24 July 2006 16:55, Peter Jeremy wrote: > > On Mon, 2006-Jul-24 13:36:19 +0930, Daniel O'Connor wrote: > > >Ktrace is almost useless "out of the box" because for any non-trivial > > >operation you run out of requests. > >

Re: truss status

2006-07-24 Thread Dag-Erling Smørgrav
John Baldwin <[EMAIL PROTECTED]> writes: > Well, the biggest thing I found when doing the PHOLD/P_WEXIT stuff that was > missing was PIOCWAIT and PIOCCONT (IIRC). Specifically, we don't currently > have any ptrace() equivalent to that. I'm not sure that's a problem. I had at one point a versio

Re: truss status (was: cvs commit: src/sys/dev/lmc if_lmc.h)

2006-07-24 Thread Pav Lucistnik
Robert Watson píše v po 24. 07. 2006 v 13:10 +0100: > (3) truss is believed not to work properly with threaded applications. ktrace > has been extended to know about thread IDs, and has always been able to > generate traces properly for threaded apps, whereas truss likely doesn't >

Re: truss status (was: cvs commit: src/sys/dev/lmc if_lmc.h)

2006-07-24 Thread John Baldwin
On Monday 24 July 2006 08:10, Robert Watson wrote: > On Mon, 24 Jul 2006, Greg 'groggy' Lehey wrote: > > > On Friday, 21 July 2006 at 13:29:38 +0200, Dag-Erling Smrgrav wrote: > >> Tom Rhodes <[EMAIL PROTECTED]> writes: > >>> Wait. I'm lost. Isn't truss(1) broken on FreeBSD? > >> > >> Depends on

Re: truss status

2006-07-24 Thread Daniel O'Connor
On Monday 24 July 2006 18:57, Alexander Leidinger wrote: > Feel free to write something up (plain text is enough), which explains > your wishes regarding kdump/ktrace and is usable for inclusion into > the existing entry at http://www.FreeBSD.org/projects/ideas/. > > Other proposals for the ideas l

Re: truss status (was: cvs commit: src/sys/dev/lmc if_lmc.h)

2006-07-24 Thread Robert Watson
On Mon, 24 Jul 2006, Greg 'groggy' Lehey wrote: On Friday, 21 July 2006 at 13:29:38 +0200, Dag-Erling Smrgrav wrote: Tom Rhodes <[EMAIL PROTECTED]> writes: Wait. I'm lost. Isn't truss(1) broken on FreeBSD? Depends on your definition of broken. It kind of works provided you have /proc mou

Re: truss status

2006-07-24 Thread Robert Watson
On Mon, 24 Jul 2006, Peter Jeremy wrote: On Mon, 2006-Jul-24 13:36:19 +0930, Daniel O'Connor wrote: Ktrace is almost useless "out of the box" because for any non-trivial operation you run out of requests. Some changes were made to improve this and this problem has mostly gone away. Mostly

Re: truss status

2006-07-24 Thread Robert Watson
On Mon, 24 Jul 2006, Daniel O'Connor wrote: On Monday 24 July 2006 12:28, Daniel O'Connor wrote: ktrace is a good disk filling service if you run it like this. truss IS more convenient. If kdump could write to a pipe it would be almost as convenient. Oh, and one more thing.. Ktrace is almo

Re: truss status

2006-07-24 Thread Alexander Leidinger
Quoting Daniel O'Connor <[EMAIL PROTECTED]> (from Mon, 24 Jul 2006 18:18:36 +0930): On Monday 24 July 2006 16:55, Peter Jeremy wrote: On Mon, 2006-Jul-24 13:36:19 +0930, Daniel O'Connor wrote: >Ktrace is almost useless "out of the box" because for any non-trivial >operation you run out of req

Re: truss status

2006-07-24 Thread Daniel O'Connor
On Monday 24 July 2006 16:55, Peter Jeremy wrote: > On Mon, 2006-Jul-24 13:36:19 +0930, Daniel O'Connor wrote: > >Ktrace is almost useless "out of the box" because for any non-trivial > >operation you run out of requests. > > Some changes were made to improve this and this problem has mostly > gone

Re: truss status

2006-07-24 Thread Peter Jeremy
On Mon, 2006-Jul-24 13:36:19 +0930, Daniel O'Connor wrote: >Ktrace is almost useless "out of the box" because for any non-trivial >operation you run out of requests. Some changes were made to improve this and this problem has mostly gone away. -- Peter Jeremy pgpN44vhhtIKj.pgp Description: PG

Re: truss status

2006-07-23 Thread Dag-Erling Smørgrav
"Daniel O'Connor" <[EMAIL PROTECTED]> writes: > Ktrace is almost useless "out of the box" because for any > non-trivial operation you run out of requests. It depends on what you use it for, and how you use it (with / without -t) DES -- Dag-Erling Smørgrav - [EMAIL PROTECTED]

Re: truss status

2006-07-23 Thread Daniel O'Connor
On Monday 24 July 2006 12:28, Daniel O'Connor wrote: > ktrace is a good disk filling service if you run it like this. > > truss IS more convenient. > > If kdump could write to a pipe it would be almost as convenient. Oh, and one more thing.. Ktrace is almost useless "out of the box" because for an

Re: truss status

2006-07-23 Thread Daniel O'Connor
On Monday 24 July 2006 10:34, Dag-Erling Smørgrav wrote: > Greg 'groggy' Lehey <[EMAIL PROTECTED]> writes: > > "Dag-Erling Smørgrav" <[EMAIL PROTECTED]> writes: > > > ktrace(1) is better [than truss] in almost all respects. > > > > truss is good in that it gives immediate output. > > So does ktrace

Re: truss status

2006-07-23 Thread Dag-Erling Smørgrav
Greg 'groggy' Lehey <[EMAIL PROTECTED]> writes: > "Dag-Erling Smørgrav" <[EMAIL PROTECTED]> writes: > > ktrace(1) is better [than truss] in almost all respects. > truss is good in that it gives immediate output. So does ktrace - just run kdump -l on a different terminal. DES -- Dag-Erling Smørgr

truss status (was: cvs commit: src/sys/dev/lmc if_lmc.h)

2006-07-23 Thread Greg 'groggy' Lehey
On Friday, 21 July 2006 at 13:29:38 +0200, Dag-Erling Smrgrav wrote: > Tom Rhodes <[EMAIL PROTECTED]> writes: >> Wait. I'm lost. Isn't truss(1) broken on FreeBSD? > > Depends on your definition of broken. It kind of works provided you > have /proc mounted. Still, there's really not much point i