So now I have 10.3 set up again. Finally. On 2012年05月03日 23:25, Per Jessen wrote:

jd wrote:

...

no errors from the long test. But it now fails to seek when powered up. The disk is now definitely toast. I think too much unnecessary disk io cooked the disk's chipset, something i've seen happen before, w/ seagate & maxtor disks.

The SMART test really ought to have picked it up.

It definitely should have given some clue. I meant instead of io that all the seeking probably cooked the head motor drivers. hard to tell because the disk wasn't unusually warm, whereas the older disks mentioned got too hot to touch under similar conditions. The disk still spins up OK but the heads, while they do move, keep dropping back to the power-off position. After leaving the second disk for a couple weeks, it worked long enough to copy and run several long tests. All the SMART tests passed but as time went on it began to have io problems. Still, no errors were logged or reported by SMART. Reimaged the third disk with 10.3, the last known (mostly) good version for this machine. It's been running almost 2 weeks without any complaints or errors.

...

System is hp laptop, intel coppermine & piix chipset. hard to prove hardware is @ fault when it works just fine w/ other releases.

Yeah, but DMA errors are pretty unusual.

I've never seen that kind of DMA error before.

...

mem usage was not unusual but kswapd was using 93% of cpu. no cron jobs.

That's different - any chance that swapping was being attempted but not succeeding? That would keep kswapd busy I expect.

If kswapd were blocked would it behave the same? I would expect kswapd to complain loudly and clearly in any case. I don't recall seeing any messages from kswapd or the crypt module used for encrypting swap. I recall only three messages that complained about disk io timeouts and another about disabling disk DMA on several subsequent reboots using both now dead disks, which stopped occurring after a few days. After that, up to the moment they died, there were no DMA problems at all. It's been recently mentioned elsewhere on the list that Akonadi and packagekit start going insane in the background. It may be that one of these or some other badly behaving disk/file indexer/searcher, something not started by at/cron, was causing the problem. I had forgotten about those evil disk/file indexing monsters. I've had horrible problems with them in the past. Usually those problems surface almost instantly after an upgrade/install though, and are dealt with. This problem didn't appear until quite some time later, while I was asleep, actually. It would be nice if they weren't mandatory dependencies for KDE & Gnome, and that they could be easily, quickly disabled, preferably by default. I don't understand how anyone could possibly think they're so absolutely vital and that everyone *MUST* use them. I seem to recall that because they were a serious problem for so many users it was decided they would be disabled or not installed by default.

...

noticed that since 11.0 older system support has been either broken and neglected, or removed entirely. that's really too bad.

Can't say I have experienced the same - 12.1 was a poor release due to systemd, but otherwise I have not noticed any lack of support for older hardware. (except the GUI needing more resources, but that's to be expected).

Pegasus driver is still broken. It's been broken since 10.3. If you're lucky it'll just cause an oops, not a crash. Still requires a reboot as it silently breaks USB when it fails. Only a small subset of Prism2/2.5/3 functions are now supported and it seems a bit fragile. It seems Orinoco, Proxim and Ricochet are also no longer fully supported. Some PCMCIA cards are no longer detected such as SunDisk/SanDisk PCMCIA Flash cards and Flash memory on otherwise supported PCMCIA cards. Haven't checked lately but noticed that Palm USB support disappeared after being reintroduced. Silicon Motion support for X was seriously crippled when it was allegedly improved in 2008, disabling hardware acceleration and 3d, and it hard-codes allocation of too much screen memory, and those were never fixed. Older ACPI support has been broken or at least incomplete since 11.0 and that was never fixed. Can't get APM to work right since 11.0. Some problem with motherboard i2c support since 10.1 broke the cooling fan control, especially for many laptops and someone said that was a known problem fixed in the next release but it wasn't. wpa-supplicant oopses and probably doesn't work with CPU's without crypto functions. Just to mention a couple off the top of my head. jd -- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org