Mailing-List: contact cygwin-help AT sourceware DOT cygnus DOT com; run by ezmlm Sender: cygwin-owner AT sourceware DOT cygnus DOT com Delivered-To: mailing list cygwin AT sourceware DOT cygnus DOT com Message-ID: <002301be69fa$4472b060$e8e0a1cf@mork.mbnet.mb.ca> Reply-To: "Parzival Herzog" From: "Parzival Herzog" To: Cc: "Michel Dagenais" Subject: Compiling PM3 Modula-3, worked under B20.0, now hangs "rm", "as", others in infinite loop on B20.1 Date: Tue, 9 Mar 1999 00:58:42 -0600 MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_0020_01BE69C7.F9A74550" X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 4.72.3110.5 X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3 ------=_NextPart_000_0020_01BE69C7.F9A74550 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable I had used B17-19 with no problems for months, Then before I installed = PM3 Modula-3, I updated to B20.0. PM3 Modula-3 (http://polymtl.ca/m3) installed and = ran successfully, then updated to B20.1. PM3 Modula-3 and Cygwin appeared to work fine. = Then I tried to compile a large Modula-3 package, and after compiling part-way = through the system, my computer hung in an infinite loop, steadily allocating about = 50-60K of memory per second, with 100% CPU utilization, and no significant disk = activity. At first I thought this was a Modula-3 problem, but gradually started to suspect = it was a timing problem in the Cygwin library. I noticed that the hang occurred after = approximately the same amount of elapsed time compiling, and not on the same source file, = and with exactly the same behaviour while several different programs were active, = i.e. M3cgc1 (the Modula 3 compiler), as (the GNU assembler - called by the compiler), and even = once in "rm" (the Cygwin rm program.) So, I decided to rebuild from scratch the Modula-3 compiler. This failed = in exactly the same way, with a hang, not always on the same source file, but after = roughly the same amount of elapsed time well into a (very complex) build process. I = then uninstalled B20.1, and I went to fetch B20.0, but it was gone from the = ftp sites. (I guess the "serious bug" fixed in 20.1 caused its withdrawal.) So I = fetched and installed B19.1 Now, Modula-3 installed and ran correctly, and all the = systems that failed to compile under B20.1, compiled correctly.=20 Unfortunately, I can't boil this down to a simple isolated case: its not = a specific program that hangs, but an arbitrary program, and that program is = spawned by a (make-like) shell (m3build) executing a complex build operation. To = reproduce the problem you have to install PM3 Modula-3. And in order to do that = you need LOTS of disk space (about 400 MB) and you need a patch (at least for the PM3 1.1.10 version - this may be fixed in the PM3 1.1.11 release - = please contact michel DOT dagenais AT polymtl DOT ca . I can say that PM3 Modula-3 is = very reliable very portable software that installs and works on dozens of = machines and flavours of UNIX, and I hope the Cygnus developers will look into this = problem relatively quickly! I have a Windows NT 4.0 system, service pack 4, on a 128MB AMD K6-266 = machine, I run the NTFS file system, with compression turned on. Parzival =20 ------=_NextPart_000_0020_01BE69C7.F9A74550 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
I had used B17-19 with = no problems=20 for months, Then before I installed PM3 Modula-3,
I updated to B20.0. PM3 = Modula-3 (http://polymtl.ca/m3) installed and = ran=20 successfully,
then updated to B20.1. = PM3 Modula-3=20 and Cygwin appeared to work fine. Then I tried
to compile a large = Modula-3 package,=20 and after compiling part-way through the
system, my computer = hung in an=20 infinite loop, steadily allocating about 50-60K of
memory per second, with = 100% CPU=20 utilization, and no significant disk activity. At first
I=20 thought this was a Modula-3 problem, but gradually started to suspect it = was a=20 timing
problem in the Cygwin library. I = noticed that the=20 hang occurred after approximately the
same amount of elapsed time compiling, = and not on=20 the same source file, and with
exactly the same behaviour while = several different=20 programs were active, i.e. M3cgc1 (the Modula
3 compiler), as (the GNU assembler - = called by the=20 compiler), and even once in "rm" (the Cygwin
rm program.)
 
So, I decided to rebuild from scratch = the Modula-3=20 compiler. This failed in exactly the
same way, with a hang, not always on = the same=20 source file, but after roughly the
same amount of elapsed time well into a = (very=20 complex) build process. I then
uninstalled B20.1, and I went to fetch = B20.0, but=20 it was gone from the ftp sites.
(I  guess the "serious = bug" fixed in=20 20.1 caused its withdrawal.) So I fetched and
installed B19.1 Now, Modula-3 installed = and ran=20 correctly, and all the systems
that failed to compile under B20.1, = compiled=20 correctly.
 
Unfortunately, I can't boil this down = to a simple=20 isolated case: its not a specific
program that hangs, but an arbitrary = program, and=20 that program is spawned by
a (make-like) shell (m3build) executing = a complex=20 build operation. To reproduce
the problem you have to install PM3 = Modula-3. And=20 in order to do that you need
LOTS of disk space (about 400 MB) and = you need a=20 patch (at least for the
PM3 1.1.10 version - this may be fixed = in the PM3=20 1.1.11 release - please
contact michel DOT dagenais AT polymtl DOT ca=   .=20 I can say that PM3 Modula-3 is very
reliable very portable software that = installs and=20 works on dozens of machines and
flavours of UNIX, and I hope the Cygnus = developers=20 will look into this problem relatively
quickly!
 
I have a Windows NT 4.0 = system,=20 service pack 4, on a 128MB AMD K6-266 machine,
I run=20 the NTFS file system, with compression turned on.
 
 
Parzival
 
------=_NextPart_000_0020_01BE69C7.F9A74550 Content-Type: text/plain; charset=us-ascii -- Want to unsubscribe from this list? Send a message to cygwin-unsubscribe AT sourceware DOT cygnus DOT com ------=_NextPart_000_0020_01BE69C7.F9A74550--