[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: glibc6.0 -> glibc6.1, also IDE controllers

On Sat, 1 May 1999, Dirk Allard wrote:

> Hi,
> tar: usr/lib/libstdc++-libc6.1-1.so.2: Cannot lchown to uid 0 gid 0:
> Function not implemented
> tar: Error exit delayed from previous errors

	The lchown issue was solved for me by upgrading bash, apt and dpkg
using the ar x method and moving the specific files to their correct
places (Easier if you do off of a ramdisk). However, the segfaults I was
complaining about on boot-up of 2.2.3 apparently came from enabling CMD646
IDE controller and DMA with multiple drives attached to IDE. I can bring
the box up without having DMA enabled by default. hdparm -T -t reports
1.78 MB/s on my disks with 32 bit IO and prefetch. If I enable DMA then,
the throughput rises to 10 MB/s, but a few minutes later the CD apparently
makes itself known (slight sound comes from it), and then things become
_EXTREMELY_ slow. This apparently happens to people who have a CMD 640
controller too, which has a workaround option for the kernel. Apparently
the CMD 646 doesn't yet, and the drivers/block/cmd646.c alludes to
revision 1 being a screwed up chip. I guess CMD has experience in that. :(

	Does anyone know of a workaround yet? I really don't want to
continue at 20% speed for my IDE disks.


Reply to: