[haiku] Re: Adventures in Real Hardware
- From: "Michael Lotz" <mmlr@xxxxxxxx>
- To: haiku@xxxxxxxxxxxxx
- Date: Mon, 17 Nov 2008 10:14:17 +0100
On Sun, 16 Nov 2008 23:09:06 -0800, Alan Burkes wrote
> said that a write operation had timed out. This bothers me for a
> number of reasons. Is it really necessary to halt everything for
> something so trivial?
It isn't and it doesn't. Reading and writing is asynchronous, but if you need
disk access and you only have one disk that is in use by something else then
that's about it. It's the real bottleneck of todays computers IMO.
> It also bothered me that the mouse couldn't
> move for a second or two before KDL showed up, because it could mean
> that the kernel isn't allowing anything else to run.
Don't worry, the kernel is quite a bit more advanced than that.
> As for the
> error itself, it does explain why Haiku didn't write any logs.
No, it didn't write any logs because it didn't yet mount the boot volume. When
that isn't mounted it's not possible to write anything to it.
If you get to the desktop quite a bit of stuff has been written back to disk
already. It's possible though that some of the transactions aren't complete
and therefore the files are removed due to journaling.
> But it's bothering me that my flash drive is denying Haiku to write
> anything. There is absolutely no reason why it should not work,
> except, perhaps, the pre-alpha build doesn't leave enough space to
> write anything else to.the partition. Or perhaps Haiku mounted it as
> read-only and the write operation failed because of a software bug?
It states that a timeout happened, so my guess is that a timeout happened ;-).
The device is probably not actively denying anything. As said above, at that
stage quite a bit has probably already been written.
Some devices have issues when their volume info is queried in between normal
data transfers, so starting Tracker that scans the devices/partitions could
easily cause the device to stop communicating.
If you file a bug report about it and include information like revision number
and hardware used and syslog info regarding usb_disk, we can debug the issue
further.
On Mon, 17 Nov 2008 08:21:20 GMT, Cyan wrote
>> And to the kernel developers out there; good luck with SMP support.
>> You'll need it. :(
>
> Speaking of which, does Haiku support multiprocessing through
> ACPI yet (as opposed to MPS)? And if not, is it on the roadmap
> for R1?
Haiku has support for that since r23100 from early December last year. MPS has
been there from pretty early on (I guess it was in NewOS already). It's
working fine on all the hardware I have access to. I'm running Haiku full time
on my C2Q desktop and C2D laptop with all cores and without issues. So rest
assured that SMP is pretty well supported.
Regards
Michael
Other related posts: