Hello

I posted this on Mandriva forum too - I'm hoping someone here might know the answer...

I'm looking at installing 2008, so I downloaded the "One" live iso to try it out on my hardware.

It takes an age to boot, largely due to some sort of loop it gets into trying to configure the DVD Writer it is running from. (Toshiba ODD-DVD SD-R6252 on a Compaq Presario 2500 laptop - ALI M5229 IDE Controller)

It basically sits there saying



a few times,

then the same thing but

for a while...


I'm assuming that this is because the new libata drivers are having a problem with this drive, negotiating DMA

The whole boot process takes about 15 minutes (2007.1 live iso boots in 2 minutes, tops, without these errors)

I've tried giving boot parameters:

live combined_mode=ide
live combined_mode=libata

live libata.atapi_enabled=0
live libata.atapi_enabled=1

in various combinations.

I've also tried noapic, nolapic, acpi=off

and even ide=nodma

None of these parameters seem to help at all.

Questions:
Is there any way to force libata new stuff not to be used at boot ?

Will this problem go away with a proper install ?
(ie Is it a peculiarity of the "live" iso)



3.10.2007: edited to change title to [SOLVED]

It sounds like a kernel bug.
Give us some "uname -a" please, and report this to Mandriva.
And no, there's no way to bypass the new libata stuff.

Thanks - I'm actually in the process of burning a new CD.

Just want to be sure that it's a real bug, and not a bad burn before I report it.

md5sum of the iso was ok - but when I tried to md5sum the CD I got a couple of read errors...

...

I'll report back

This used to be a common symptom with the nforce chipset, but personally its the first time I see it reported for ALi chipset.
Mandriva devs should know better.

Well, it persists after burning a new cd.

all md5sums ok this time.

uname -a:


I'm off to bugzilla now to post a report

Ok - it's Bug Number 34372

edit:

Which turned out to be a "resolved duplicate" - fixed in cooker already - so "It'll be alright on the night"

The responses in Bugzilla were lightning fast - very impressive...

As long as you're sure you haven't done something terribly wrong, Bugzilla is absolutely the best place to submit errors of that kind...

Certainly, as I just discovered.

But I wouldn't have liked to have gone in there unprepared or with a silly non-bug.

They'd have eaten me alive

A high level of understanding of what you are reporting, some prior research, thorough reporting of your bug, and the willingness to be quickly referred to something you missed, and should have seen, would seem essential prerequisites for mixing with that crowd.

That said, they were gentle with me, despite this bug being fixed in Cooker.

In my defence - I couldn't have known it was fixed since I was downloading the latest version of "One" as a test for 2008 on my hardware.

Downloading and installing the latest Cooker to see if I should download and install 2008 when it goes "live" is a bit more time consuming for a lazy old ***** like me

don't be that scared of bugzilla. we'd rather you post 10 'silly' bugs and one useful one than not post anything because you're scared. All bugs are handled by the triage team before being handled by the developers, and I run the triage team. So if anyone's mean to you, tell me and I'll set 'em straight.

Anyone still following this thread should read the corresponding thread in the Mandriva Forum : Here...

Basically, the good folks at Mandriva provided a test iso which solves the problem with the ide drivers, so the bug in RC2 will be fixed in Final

Splendid service - and super quick response.

We should be as quick to congratulate them when they do good stuff, as we are to criticise them at other times... (IMO)