public inbox for gentoo-alpha@lists.gentoo.org
 help / color / mirror / Atom feed
* [gentoo-alpha] New install, old user :)
@ 2021-10-05 12:24 Marc Tipping
  0 siblings, 0 replies; only message in thread
From: Marc Tipping @ 2021-10-05 12:24 UTC (permalink / raw
  To: gentoo-alpha@lists.gentoo.org

[-- Attachment #1: Type: text/plain, Size: 2858 bytes --]

Hi,

I've recently re-acquired a DS10 (2gb mem, ev67, LSI 53c895 scsi, trio64v2) and immediately went to Gentoo Linux, as I had an excellent experience last time round.

The handbook is great, but I had a couple of potential issues :-

o Installing base system. when chrooted, and the user undertakes the initial emerge .. --newuse @world,  gcc is reinstalled and creates all 300-odd locale files. This seems to be because locale.gen isn't updated until later in the page to the users' preference? This adds a lot of time to the initial baseline build.

o Installing the bootloader: aboot 🙂. Thanks for updating it! The document says create swap at cylinder 3 to leave room for the aboot bootloader. However, the swriteboot command warns that installation will overwrite partition 3 and 1 (swap). So, it looks to need more than 2 cylinders free. In the end I shrank swap by a couple of hundred meg and swriteboot then only warned about partition 3 which is normal.

I can now boot off dka0, using initram fs from a 'genkernel all' kernel. Since booting off this new kernel I've had numerous segfaults. I've replaced the RAM, hdd, applied new heat transfer material to the CPU, Re-seated everything. Tried a new gfx card (Permedia2). Still, in the middile of a long multi-day compile it will segfault with an ICE. After that even my bash session can't szmalloc() properly and it will crash taking ssh with it if I'm remote.  Sometimes it may crash sooner, maybe not at all (although that's rare). If I 'cd' to the work directory and carry on the 'make' it usually works. I have plenty of free mem, I've tested the memory through the SRM console for hours, it is ECC though after all, and I never see an error, nor do I see serious bug check errors in the console etc. I'm currently running memtester against 1GB of memory, it looks good up until now but I'll let you know!

Looking at 'lsmod' this new genkernel loads a lot more modules than the livecd kernel. The genkernel on the cd is obviously different to the genkernel that lands on my hdd? I'm stabbing in the dark, but could a bad module or kernel routine in the new genkernel be causing these memory issues? Memory timing for a different machine being loaded? When I boot off the livecd the initial compilation completes fine, (--update @world etc) and the genkernel completes. Try a big compile booted of the new genkernel and I usually get a segfault/ICE. It's hurting my head now 🙂.

I'm using "-O2 -pipe -mcpu=ev67" although I've removed everything but -mcpu at times to test, with no change.

Maybe I'm unlucky and I've bought a machine with subtle memory timing issues. The chap who had it before me used it as an NFS server running NetBSD so maybe he didnt stress the machine so much as to uncover these issues.

Any thoughts kindly received.

All the best,
 Marc.

[-- Attachment #2: Type: text/html, Size: 6571 bytes --]

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2021-10-05 12:24 UTC | newest]

Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2021-10-05 12:24 [gentoo-alpha] New install, old user :) Marc Tipping

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox