Re: serious bug in filemap.c

Henrik Storner (storner@osiris.ping.dk)
Sun, 19 May 1996 23:44:19 +0200


Alexey Kuznetsov writes:

>It is possible (and occurs!) that not uptodate pages are mapped into
>process space.
>Maybe, this fix is not complete.

>BTW, someone complained about NFS mmap oddities. Maybe, this fix will
>repair them too.

Hmm - I wonder if this has happened to me a couple of times over the
past 24 hours.

1) Last night, sendmail refused to run, dropping 19 pieces of mail
along the way. It couldn't load a shared library (libdb.so, I
think) - running file(1) on the lib returned "data" instead of
the expected "ELF 32-bit LSB dynamic ..." I do run a weekly
tape-backup saturday night, so disk and cache systems had probably
been pretty busy during the night. After a reboot, things
returned to normal, so it must have been a corruption in memory.

2) My first attempt at booting up pre-2.0#6 halted when the kernel
was being uncompressed - "CRC error". This has never happened before.

That kind of errors give me the creeps...