On Mon, Apr 17, 2000 at 04:13:08PM +0100, Patrick Welche wrote:
> 
> So something is odd with
> 
> /dev/sd1e /usr/obj ffs rw,async,noatime 1 2
> 
> all partitions without async,noatime see fine.. or maybe they don't get
> clobbered as much...

Nope, mounting /usr/obj as just ffs lead to a hang when deleting lots of
files, and lots of

/usr/bin/nm: signal__F.o: File format not recognized
/usr/bin/nm: r_sqrt_F.o: File truncated
/usr/bin/nm: r_tanh_F.o: File truncated
/usr/bin/nm: d_prod_F.o: File truncated
/usr/bin/nm: typesize_I.o: File format not recognized
/usr/bin/nm: util_I.o: File format not recognized
/usr/bin/nm: time__U.o: File format not recognized
/usr/bin/nm: ttynam__U.o: File format not recognized
/usr/bin/nm: symlnk__U.o: File format not recognized
/usr/bin/nm: umask__U.o: File truncated

etc

After deleting and rebuilding each eventually

building standard g2c library
Segmentation fault - core dumped
#0  0x480ae5d5 in _bfd_compute_and_write_armap ()
#1  0x480ae102 in _bfd_write_archive_contents ()
#2  0x480a9235 in bfd_close ()

Then a hang vinval D+   p1    0:00.05 rm -f libg2c_p.a

So, it seems that the files are corrupted under high activity. The one
significant change in that area since it last worked (1.4U) is that I have a
    Vendor Name: Adaptec (0x9004)
    Device Name: AHA-2940A Ultra (0x6178)
and believe that formerly tagged queuing wasn't enabled by default but now is?

Any suggestions on where to look?

Cheers,

Patrick