Reply by Adrian Knoth January 27, 20062006-01-27
hutzelbutz <joachim.becker@imtek.uni-freiburg.de> wrote:

> However, the command "hostid" gives the right value. So I did " sudo > hostname > /etc/hostid" to create the file and voila: ISE starts in 5 > seconds.
I hope for Xilinx that it is possible to configure this filename. It is clearly unacceptable that a (more or less) simple userland application needs a system-wide configuration file. I agree for daemons or default-overrides like /etc/opera6rc, but not for a tool which might run without uid-0-interaction at all. Whoever has insisted on this filename should go back to the 90th and code Win95-applications when there was still no need to divide into user-related stuff (configuration, save files) and system installation. In other words: put your xilinx.ini in the windows directory ;) SCNR ;) -- mail: adi@thur.de http://adi.thur.de PGP: v2-key via keyserver Kein Kind ass das Pilzgericht, nur der Juergen, der muss wuergen.
Reply by hutzelbutz January 27, 20062006-01-27
OKok, my Linux knowledge is on the edge but not gone: :-)

The important line is:
[pid 22097] open("/etc/hostid", O_RDONLY) = -1 ENOENT (No such file or
directory)

This told me that ISE is looking for the file /etc/hostid, which was
not existent at my computer.
However, the command "hostid" gives the right value. So I did " sudo
hostname > /etc/hostid" to create the file and voila: ISE starts in 5
seconds.

Problem solved !

Thanks for your help, I wouldn't have done it without all of your help
!
I hope, this helps everybody with the same problem.

Cheers, Joachim

Reply by hutzelbutz January 27, 20062006-01-27
Hello,

thanks for the excellent hint!
Later on that day, I tried a linux kernel without smp and the same
problem occurred. So multi-processors did not seem to be part of the
problem. That also brought me back to networking issues, since I heard
about 7.1 needing correct hostname and so on. But I couldn't figure it
out until I read your hint.

So I tried "strace -f ise". When the splash screen is over, I get never
ending loops of the following messages (it might not be cut at the
right part of the message...)

[pid 22097] close(7)                    = 0
[pid 22097] getpid()                    = 22097
[pid 22097] open("/etc/hostid", O_RDONLY) = -1 ENOENT (No such file or
directory)
[pid 22097] open("/etc/hostid", O_RDONLY) = -1 ENOENT (No such file or
directory)
[pid 22097] uname({sys="Linux", node="lmesim121", ...}) = 0
[pid 22097] socket(PF_FILE, SOCK_STREAM, 0) = 7
[pid 22097] connect(7, {sa_family=AF_FILE,
path="/var/run/.nscd_socket"}, 110) = -1 ENOENT (No such file or
directory)
[pid 22097] close(7)                    = 0
[pid 22097] open("/etc/hosts", O_RDONLY) = 7
[pid 22097] fcntl(7, F_GETFD)           = 0
[pid 22097] fcntl(7, F_SETFD, FD_CLOEXEC) = 0
[pid 22097] fstat(7, {st_mode=S_IFREG|0644, st_size=349, ...}) = 0
[pid 22097] mmap(NULL, 4096, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x2a97390000
[pid 22097] read(7, "# Do not remove the following li"..., 4096) = 349
[pid 22097] read(7, "", 4096)           = 0
[pid 22097] close(7)                    = 0
[pid 22097] munmap(0x2a97390000, 4096)  = 0
[pid 22097] socket(PF_INET, SOCK_DGRAM, IPPROTO_IP) = 7
[pid 22097] connect(7, {sa_family=AF_INET, sin_port=htons(53),
sin_addr=inet_addr("10.122.0.1")}, 28) = 0
[pid 22097] fcntl(7, F_GETFL)           = 0x2 (flags O_RDWR)
[pid 22097] fcntl(7, F_SETFL, O_RDWR|O_NONBLOCK) = 0
[pid 22097] poll([{fd=7, events=POLLOUT, revents=POLLOUT}], 1, 0) = 1
[pid 22097] sendto(7,
"V\374\1\0\0\1\0\0\0\0\0\0\tlmesim121\5ikoop\6pri"..., 40, 0, NULL, 0)
= 40
[pid 22097] poll([{fd=7, events=POLLIN, revents=POLLIN}], 1, 5000) = 1
[pid 22097] ioctl(7, FIONREAD, [197])   = 0
[pid 22097] recvfrom(7,
"V\374\205\200\0\1\0\1\0\4\0\3\tlmesim121\5ikoop\6pri"..., 1024, 0,
{sa_family=AF_INET, sin_port=htons(53),
sin_addr=inet_addr("10.122.0.1")}, [16]) = 197

"ikoop.privat" is the domainname and the computer is in the
10.122.182.x subnet. So although I do not have a clue, what's going on,
I see that it seems to hang with networking issues. Probably "name
resolution" as you mentioned.

However, something must have changed in version 8, since 7 was running
fine.
I set hostname and domainname and the /etc/hosts with no success. My
Linux knowledge is at its end, can anybody help me, please?

Thanks a lot, Joachim

Reply by Georg Acher January 26, 20062006-01-26
"hutzelbutz" <joachim.becker@imtek.uni-freiburg.de> writes:

>I don't know enough about multi-processor load distribution in Linux, >but this behavior is reproducable. Could it be that those 4 minutes >come from the time, the OS needs to find out that one process should be >switched to the other CPU???
No, at least process migration should not matter. Maybe they have a race condition which doesn't appear in the sequential single-CPU case.
>I would be curious to hear the experts about this topic!
Simply do a "strace -p <pid>" and you will see what is going on at the syscall-level. Most of the time it is also possible to start the program directly with "strace -f <program>", but it can have problems with catching forked threads. BTW: Haven't yet used 8.1 myself, but I suspect an issue either with thread handling (mutexes, etc.) or some name resolution (for whatever cause they need it). -- Georg Acher, acher@in.tum.de http://www.lrr.in.tum.de/~acher "Oh no, not again !" The bowl of petunias
Reply by hutzelbutz January 26, 20062006-01-26
:-)

Back to the problem:

I think I forgot to mention that I am running the tools on dual-core
machines, which seems to be part of the problem.
When I watch the _pn processes while starting in "top", I see that with
one call of ise, there are two _pn started on the same CPU. One has a
bit of CPU usage for a short time, the splash screen shows and then
they both stand with 0.1 % CPU. After 4 Minutes, one _pn is moved to
the other CPU and this process then takes 1% CPU. 10 seconds later the
GUI is there. Still both processes take exactly the same amount of
memory and live happily ever after.

I don't know enough about multi-processor load distribution in Linux,
but this behavior is reproducable. Could it be that those 4 minutes
come from the time, the OS needs to find out that one process should be
switched to the other CPU???

I would be curious to hear the experts about this topic!

Cheers,
 Joachim

Reply by January 25, 20062006-01-25
hutzelbutz wrote:
> However, I _do_ like graphical user > interfaces and once everything is set up, I can press one button to run > the complete designflow over and over again... Some people like one, > some the other, no worries, everybody is free.
I agree. For my own part, I like the commadn line interface. Once everything is set up, I can type "make", press return, and run the complete designflow over and over again. :-) I find it faster and more convenient to type "make" than to mouse to a button. But that's a matter of personal taste.
Reply by hutzelbutz January 25, 20062006-01-25
sure, I know about all those scripting and makefile advantages, there
have been enough wars on that. However, I _do_ like graphical user
interfaces and once everything is set up, I can press one button to run
the complete designflow over and over again... Some people like one,
some the other, no worries, everybody is free.
So thanks for your suggestion but I am not looking for a workaround but
a real fix of the ISE GUI.

Reply by Phil Tomson January 23, 20062006-01-23
In article <1138003542.911045.113800@o13g2000cwo.googlegroups.com>,
hutzelbutz <joachim.becker@imtek.uni-freiburg.de> wrote:
>For me, the problem with the extremely slow GUI is still there. >Funny enough, the version 7 of ISE was running fine on RedHat Linux >2.4.21. Now that we upgraded to 8.1, it takes ISE 3 Minutes (!) to >start beyond the splash screen. All interaction is extremely slow but >the commandline tools are very fast. > >Is there anything I am missing? Do I need a newer version of Java / Qt >/ whatever? > >I would be thankful for any suggestions.
I would suggest using scripts or makefiles to run the toolchain. Scripting allows you to automate things and thus offers advantages that the GUI can't. Phil
Reply by hutzelbutz January 23, 20062006-01-23
Sorry, I forgot to mention: The hardware is Dual AMD 3800 with 2GB RAM.
Everything else is snappy and extremely fast. It was fine with ISE 7.1
but 8.1 broke it. I even installed the service pack 8.1.01, which
didn't bring any help.

Reply by hutzelbutz January 23, 20062006-01-23
For me, the problem with the extremely slow GUI is still there.
Funny enough, the version 7 of ISE was running fine on RedHat Linux
2.4.21. Now that we upgraded to 8.1, it takes ISE 3 Minutes (!) to
start beyond the splash screen. All interaction is extremely slow but
the commandline tools are very fast.

Is there anything I am missing? Do I need a newer version of Java / Qt
/ whatever?

I would be thankful for any suggestions.
Cheers, Joachim