[ltt-dev] General Questions

[Kevin Harms]

  I'm having trouble building a lttng enabled kernel. I'm using the following command to patch a 2.6.26 kernel.

for f in $(cat ../lttng/patch-2.6.26-0.11/series | egrep -v "^#"); do patch --dry-run -p1 < ../lttng/patch-2.6.26-0.11/${f}; done

  Some of the patches apply correctly, but I get a few of these:

patching file kernel/module.c
Hunk #1 FAILED at 47.
Hunk #2 succeeded at 1386 (offset -1 lines).
Hunk #3 succeeded at 2162 (offset -17 lines).
1 out of 3 hunks FAILED -- saving rejects to file kernel/module.c.rej

  I'm using kernel 2.6.26.

thanks,
kevin

----- Original Message -----
From: "Michel Dagenais" <michel.dagenais at polymtl.ca>
To: "Kevin Harms" <harms at alcf.anl.gov>
Cc: ltt-dev at lists.casi.polymtl.ca
Sent: Tuesday, July 22, 2008 4:15:54 PM GMT -06:00 US/Canada Central
Subject: Re: [ltt-dev] General Questions


> I'm looking into an issue where fsync() is being called repeatedly by
> an application but some of those fsync() calls stall sometimes. Would
> LTT be useful for trying to see what is happening inside the kernel
> that causes those fsync() calls to stall?

Yes, very much! Fsync is relatively infrequent and easy to search for in
a trace. From there you need the detailed scheduling information to
understand why it is blocking, which is exactly what LTT will give you.