Re: ftpd problems (probably kernel's fault)

[Andrew Doran <ad%netbsd.org@localhost>]

On Mon, Apr 28, 2008 at 05:02:24PM +0200, Julio M. Merino Vidal wrote:

> On Mon, Apr 28, 2008 at 4:54 PM, Andrew Doran <ad%netbsd.org@localhost> wrote:
> > On Mon, Apr 28, 2008 at 04:32:32PM +0200, Julio M. Merino Vidal wrote:
> >
> >  > I'm running NetBSD-current/amd64 as of two days ago or so, with a
> >  > kernel from today, and ftpd does not seem to work.
> >  >
> >  > I can log into the machine with the anonymous user and list files,
> >  > but whenever I try to get any of them, it gets stalled at the very
> >  > beginning of the transmission (it says 0 KBs and is unable to
> >  > calculate the ETA).  In some attempt, it transferred about 32 KBs
> >  > and then got stalled.  When I cancel the transfer with ^C, ftp says:
> >  >
> >  >    ftp: Can't send abort message: No buffer space available
> >  >
> >  > Similarly, I've also got a stall when doing a ls of a long directory.
> >  >
> >  > It does not matter if the transfers happen locally or over the network;
> >  > the problem is the same.
> >
> >  I haven't seen this. Do you have any unusual options enabled?
> 
> I can send you my kernel config if you want, but it is basically a
> GENERIC minus all hardware drivers not needed by the machine, DEBUG
> enabled, and PUFFS enabled.  Nothing unusual, I think, except for
> PUFFS (but it is completely unused at the moment).

Ok, can you try with sys/socketvar.h 1.107? It should fix it.
 
> >  > I've also had two panics caused by ftpd, so this seems to be a kernel
> >  > problem.  One of them said:
> >  >
> >  >    Stopped in pid 24619.1 (ftpd) at netbsd:mutex_exit+0xc: cmpxchgq
> >  >
> >  >    mutex_exit() at netbsd:mutex_exit+0xc
> >  >    soclose() at netbsd:soclose+0x5d
> >  >    soo_close() at netbsd:soo_close+0x1c
> >  >    closef() at netbsd:closef+0x68
> >  >    fd_close() at netbsd:fd_close+0x138
> >  >    syscall() at netbsd:syscall+0x9a
> >
> >  Please run: ident $kernel | grep uipc

Have you had these panics with the kernel from today?

Cheers,
Andrew