These were the changes I had to make to a stock 2.4.19 kernel to get my shiny new Sony VAIO PCG-SRX51P/A running smoothly with Redhat 8.0.
Note - 2002-01-20 - Got hibernate/resume working with the SWSUSP package, which works well except for some USB issues. See the notes at the end of this page. This means that everything (except memory-only suspend/resume, which I can live without now that hibernate is going) is working!
The SRX51P/A is a European VAIO - I'm not sure of the equivalent US or Japanese model, although it looks very similar to the Japanese SRX7E. Features include:
The main things of interest here are the ieee1394 CD drive and the built-in wireless networking. These both caused problems which required kernel patches to solve.
First, the RH8 install image doesn't recognize the CD drive. The model number of the CD will be either PCGA-DVD1 or PCGA-CRWD2 (I have the DVD1 model, which is a DVD reader only. The CRWD2 model combines CD-writing and was/should be released in the UK late October 2002). Additionally, although I didn't check this I suspect it won't recognize the NT partitions either, as RedHat ship with NTFS disabled by default.
This means you have to install via ethernet, which should probably work out of the box, or via some other device - in my case, a USB CD. If you've been browsing the net for advice and have tried adding the "ide1=0x180,0x386" parameter at boot-time, this won't help - it appears to be for a PCMCIA CD drive which is supplied in the US with a similar model.
Once you get RedHat to find it's images, the install proceeds nicely. On completion, you'll find the following:
By rebuilding the kernel from a standard 2.4.19 distribution with the patches available from http://big.faceless.org/misc/srx51p-patches.tar.gz, you get the following:
The modem requires the hsflinmodem RPM to be installed - a link to this is at the bottom of the page. The Bluetooth radio can be enabled, and I've managed to get it talking to another device, although I haven't yet tried to use it properly.
The ACPI patch doesn't at this stage allow the fan to be controlled, so you might want to keep an eye on your temperature in /proc/acpi/thermal_zone/ATF0/temperature. Mine hovers around 42C, which is well within the acceptable limit, and although it's hit 58C or so when things get really busy the machine has not yet gone up in smoke, and seems to cool down quickly when the busy process stops.
Most of this is derived from a couple of other VAIO pages and fairly careful perusal of the patches applied to the Redhat 8.0 kernel. In particular the following resources were useful:
Postscript 2003-01-20 - Suspend/Resume with SWSUSP This patch has been around for some time, but I didn't get around to trying it as I didn't have a swap partition. After a few months diligently closing everything down each night and then starting it up the next morning, I got fed up enough to buy Partition Magic, and the rest was easy.
After applying the patches above, simply apply the patch-acpi-acpi20020918-swsusp15.gz patch from the SWSUSP project and rebuild your kernel. You need a swap partition to write to, and make sure you follow the instructions from the SWSUSP documentation about adding parameters to the kernel and so on. You might also find this script useful, as when I ran the suspend script from the SWSUSP website it made such I mess i had to reboot (which was kind of defeating the purpose). It resets the clock and the network devices, which is all I'm really worried about.
Postscript 2003-03-04 - Suspend/Resume issues: After a few months playing with the SWSUSP patch, I've found two issues. The first is that the CDROM isn't recognized after resume. This can be fixed by calling rmmod on the ieee1394 and ohci1394 modules after resume, and then reinserting them. The second issue is USB, which has the same problem. The catch is the usbcore module cannot be unloaded, so the only way to get around it is to reboot. Working on it, will keep this page posted.
Postscript 2003-04-15 - Suspend/Resume issues solved: Thanks to Tyler Daniel, the USB not loading after resume has been fixed. The trick is to unmount /proc/bus/usb, which allows you to unload the usbcore module. He contributed a fixed suspend script, which I've tweaked a little and uploaded. Thanks Tyler! Oh, he was running with a Japanese model, a PCG-SRX7S.
Hope this proves helpful.
-- Mike Bremford - http://big.faceless.org