In case anyone was interested, I've solved my problem. It's only taken me two weeks...
It turns out that the value of HZ (clock ticks per second) in the Redhat 8 kernels is 512, whilst it's 100 in the official kernel.org kernels. This change caused a "rounding down" error in my calculations that used HZ in a way that it reduced the time (by half, I think) the driver was delaying and timing out, causing my problem. According to the nice people at redhat's Bugzilla, this change was intentional and not some freaky #ifdef problem/bug in their kernel sources. > -----Original Message----- > From: [EMAIL PROTECTED] [SMTP:[EMAIL PROTECTED] > Sent: Thursday, March 13, 2003 1:10 PM > To: [EMAIL PROTECTED] > Subject: RE: Problem with compiling proprietry drivers against stock > redha t 8. 0 kernel? > > Thanks for the reply, Steven > > Yes, I am compiling the drivers using gcc3.2. I had assumed that the stock > kernel was compiled using the tools supplied in the distribution > installation. > > I've tried downloading and using both the 2.4.18-24 and 2.4.18-26 kernels > and source from my local mirror. Unfortunately, the test fails in the same > way. > > However, you have raised a couple of avenues of thought that need looking > into. > > Justin. > > > -----Original Message----- > > From: Steven Whatley [SMTP:[EMAIL PROTECTED] > > Sent: Thursday, March 13, 2003 5:21 AM > > To: [EMAIL PROTECTED] > > Subject: Re: Problem with compiling proprietry drivers against stock > > redhat 8. 0 kernel? > > > > On Wed, 12 Mar 2003 [EMAIL PROTECTED] wrote: > > > I have a driver that I maintain as part of my job. When compiled > against > > the > > > redhat 8 stock kernel (2.4.18-14), it fails a regression test. I would > > give > > > more details, but I'm not sure exactly what the problem is myself yet; > > save > > > to say that the test fails detecting a correct tone from one of our > > telecoms > > > card, but the underlying cause for that is uncertain. > > > > > > However, when it's compiled against an official 2.4.18 kernel from > > > kernel.org (with no other tool change), it's absolutely fine. The test > > > passes with flying colours. > > > > Are you using gcc 3.2 to compile the driver? I may be mistaken but > isn't > > the stock 2.4.18-14 RH8 kernel compiled with gcc 2.95? Have you tried > the > > 2.4.18-24 (or newer) kernel from RedHat which is compiled with gcc 3.2? > > I have heard of problems when kernels and modules are compiled with a > > mixture of these to compilers. > > > > Later, > > Steven > > > > > > > > > > > > -- > redhat-list mailing list > unsubscribe mailto:[EMAIL PROTECTED] > https://listman.redhat.com/mailman/listinfo/redhat-list -- redhat-list mailing list unsubscribe mailto:[EMAIL PROTECTED] https://listman.redhat.com/mailman/listinfo/redhat-list
