2003-05-21 21:07:08

by Kendrick Hamilton

[permalink] [raw]
Subject: Module problems with gcc-3.2.x (please CC [email protected])

Hi,
I have been continuing the track down problems with a module I am
making. If I compile the kernel and module with gcc-2.95.3, the module
installs and works correctly. If I compile the kernel and module with
gcc-3.2.3, the module finishes installation and its interrupt service
routine runs, then the kernel crashes. I am using linux 2.4.18 and 2.4.20.
The computers are all pentium computers (I, II, III, celeron, xeon) with
Redhat Linux 7.3. Any suggestions would be appreciated.

I am assuming there is a problem with my makefiles and
have attached them. I can tgz the entire driver and mail it to the list.

--
Kendrick Hamilton E.I.T.
SED Systems, a division of Calian Ltd.
18 Innovation Blvd.
PO Box 1464
Saskatoon, Saskatchewan
Canada
S7N 3R1

[email protected]
Tel: (306) 933-1453
Fax: (306) 933-1486


Attachments:
Makefile (1.90 kB)
Makefile (499.00 B)
Download all attachments

2003-05-21 21:56:31

by Felipe Alfaro Solana

[permalink] [raw]
Subject: Re: Module problems with gcc-3.2.x (please CC [email protected])

On Wed, 2003-05-21 at 23:20, Kendrick Hamilton wrote:
> Hi,
> I have been continuing the track down problems with a module I am
> making. If I compile the kernel and module with gcc-2.95.3, the module
> installs and works correctly. If I compile the kernel and module with
> gcc-3.2.3, the module finishes installation and its interrupt service
> routine runs, then the kernel crashes. I am using linux 2.4.18 and 2.4.20.
> The computers are all pentium computers (I, II, III, celeron, xeon) with
> Redhat Linux 7.3. Any suggestions would be appreciated.

Can't offer any suggestion, except that I suffered the same pain as you
with 2.5.69-mm7: if I compile 2.5.69-mm7 with gcc 3.2.3, the ymfpci
(Yamaha DS-XG PCI driver) causes a lot of problems, like oopses when
compiled as a module, or panics when built-in.

The problems went away the moment I compiled using gcc 2.96. I think
this is caused by some gcc compiler bug (in fact, I've never been able
to make gcc 3.2 compile itself and pass all tests).