| 1 |
On Sat, 2005-11-12 at 15:22 +0000, Ferris McCormick wrote: |
| 2 |
|
| 3 |
> On Fri, 11 Nov 2005, David S. Miller wrote: |
| 4 |
> |
| 5 |
> > From: Ferris McCormick <fmccor@g.o> |
| 6 |
> > Date: Fri, 11 Nov 2005 21:35:48 +0000 |
| 7 |
> > |
| 8 |
> > > I'm running it on U2(2x400) currently with 2.6.14-U2-2x400-g393b0725. |
| 9 |
> > > So far, it's doing fine, and this is on a system which has never been |
| 10 |
> > > stable with a 2.6 kernel. With this kernel + IRQ patch, though, it's |
| 11 |
> > > stayed up through a lot of stress tests which I would have expected |
| 12 |
> > > would have killed it. |
| 13 |
> > > (But uptime is only 6 hours so far). |
| 14 |
> > |
| 15 |
> > It would be interesting to know if that exact kernel, |
| 16 |
> > but without the IRQ stacks patch, works just as well. |
| 17 |
> > |
| 18 |
> > It's nice that it's stable, but we should find out exactly |
| 19 |
> > why :-) |
| 20 |
> > |
| 21 |
> If the system is still up Monday and if no one else has reported on the U2 |
| 22 |
> (and U1) stability, I'll try to make a test using Josh Grebe's (squash's) |
| 23 |
> crashme script against that kernel with the IRQ patch removed. I can't |
| 24 |
> give a good estimate of when the test will be possible because Monday |
| 25 |
> morning the U2 will be helping with a fairly lengthy upgrade unrelated to |
| 26 |
> the kernel. |
| 27 |
> |
| 28 |
> Regards, |
| 29 |
> Ferris |
| 30 |
|
| 31 |
OK, I lied. I did the tests today. Here are the results: |
| 32 |
1. 2.6.14-U2-2x400-g393b0725 (git kernel as identified, with IRQ stack |
| 33 |
patch): Still looks stable. |
| 34 |
|
| 35 |
2. 2.6.14-U2-2x400-noIRQpatch-g393b0725 (same kernel w/o IRQ stack |
| 36 |
patch): very unstable (Josh's crashme script crashes it in the untar |
| 37 |
test during first pass). |
| 38 |
|
| 39 |
And just for fun, |
| 40 |
|
| 41 |
3. 2.6.14.2-U2-2x400 (2.6.14.2 as available at |
| 42 |
http://www.kernel.org/pub/linux/kernel/v2.6/) + IRQ stack patch: |
| 43 |
extremely unstable (apparently, more unstable than -2- above). |
| 44 |
I have no explanation; just results. (The g393... kernel seems to be |
| 45 |
dated November 11, and 2.14.2 also seems to be November 11: in .git/ for |
| 46 |
g393..., |
| 47 |
4 drwxr-xr-x 260 536 536 4096 Nov 11 05:39 objects/ |
| 48 |
|
| 49 |
I've told you more than I know, but at least a stable 2.6.14-xxx on U2 |
| 50 |
is possible. |
| 51 |
|
| 52 |
-- |
| 53 |
Ferris McCormick (P44646, MI) <fmccor@g.o> |
| 54 |
Developer, Gentoo Linux (Sparc, Devrel) |
Archives