---------- Forwarded message ---------- From: michael <[EMAIL PROTECTED]> Date: Jul 15, 2005 5:35 PM Subject: Re: Profiling C programs To: Asim Jamshed <[EMAIL PROTECTED]>
see previous email. i don't answer personal emails On Fri, 2005-07-15 at 15:26 +0500, Asim Jamshed wrote: > Command useds: > > 1> $ CFLAGS='-pg' ./configure --disable-cups --with-fhs --with-pam > --with-syslogs --with-utmp --with-netatalk --with-all --with-smbclient > --with-sendfile-support --with-profiling-data > > 2> make all > > 3> $ bin/smbd -s /etc/samba/smb.conf > > 4> After a transfer, I 'pkill' the process. > > 5> I then gprof the newly created gmon.out file: > > $ gprof bin/smbd bin/gmon.out > profile_info > > > > On 7/15/05, michael <[EMAIL PROTECTED]> wrote: > > Asim > > > > a) don't cc people on the list - they read the list! > > b) don't top post - it makes it hard for people to follow the thread of > > conversation > > > > Ta, M > > > > > > On Fri, 2005-07-15 at 14:50 +0500, Asim Jamshed wrote: > > > Yes, I am using the '--with-profiling-data' during the configure > > > option and using the -pg switch in the Makefile. > > > > > > > > > On 7/15/05, michael <[EMAIL PROTECTED]> wrote: > > > > On Fri, 2005-07-15 at 11:20 +0500, Asim Jamshed wrote: > > > > > Hi, > > > > > I am profiling Samba-2.2.8 source code using gprof. The problem I am > > > > > facing is that the data generated through profiling is only giving > > > > > names of application functions(which are only called up during > > > > > initialization of the server). The majority of the time is usually > > > > > spent by the server is in library functions and system calls. Is there > > > > > any way I could measure the times taken by those routines? > > > > > > > > > > > > > Have you ensured everything that's called and been compiled with the > > > > relevant flags? > > > > > > > > > > > > -- > > > > Michael Bane > > > > Atmospheric Physics Group > > > > University of Manchester > > > > > > > > > > > > -- > > > > To UNSUBSCRIBE, email to [EMAIL PROTECTED] > > > > with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED] > > > > > > > > > > > > > -- > > Michael Bane > > Atmospheric Physics Group > > University of Manchester > > > > -- Michael Bane Atmospheric Physics Group University of Manchester
