kernel bug in uipc_syscalls.c

[Matt Storer]

Hi again -

  OK, I think perhaps you can scratch that bug I found.  It turns out
that the box onto which I've been installing FreeBSD has memory that's
really, really bit the dust.  I believe that the corrupt memory could
be the cause of not only the bug in uipc_syscalls, but is also
responsible for numerous errors I've recently encountered while
running fsck.

  We're replacing the memory now.  Hopefully that will fix the problem.

Sorry to cause any alarm -

Matt

On 8/24/05, Matt Storer <mstorer at gmail.com> wrote:
> Hi -
> 
>   I'm in the process of reinstalling FreeBSD 5.4-STABLE onto an i386
> box.  I was in the process of recompiling my custom kernel when I came
> upon an error during the "make" step, in uipc_syscalls.c.  The error
> looked basically like:
> 
> uipc_syscalls.c in function "socket" 176: error: structure has no
> member named `f_gps'
> 
> I looked into it, and I came to the conclusion that f_gps was supposed
> to be f_ops, a member of struct file.  So I backed up uipc_syscalls.c
> and made the change.  The kernel has now been compiling just fine
> through that bug.
> 
> sooo... I can't imagine you guys haven't found this yet, but in the
> unlikely even you haven't, there ya go.  I don't know how my copy of
> uipc_syscalls.c could get foobar'd otherwise.
> 
> ... additionally, I'm quite interested to know if that bug exists on
> your end of things.  PLEASE let me know if it doesn't.
> 
> Thanks!
> 
> Matt
>