Re: [long] "bad cksum 0!" on AIX 5 over loopback

[Guy Harris <guy () alum mit edu>]

On Wednesday, July 2, 2003, at 4:30 PM, Shaun wrote:

> That one is a mystery, it appears to be some sort of kernel issue with
> uiomove(), I tried to determine what was going on, but gave up when I
> found that memset()ing the buffer to 0 before using it made the 
> problem go
> away.

Did that make it go away completely, or did it still occur on occasion? 
 The comment after "case EFAULT:" says

                                 *
                                 * For some unknown reason the uiomove()
                                 * operation in the bpf kernel extension
                                 * used to copy the buffer into user
                                 * space sometimes returns EFAULT. I 
have
                                 * no idea why this is the case given 
that
                                 * a kernel debugger shows the user 
buffer
                                 * is correct. This problem appears to
                                 * be mostly mitigated by the memset of
                                 * the buffer before it is first used.
                                 * Very strange.... Shaun Clowes
                                 *

but the comment before the memset says

        /* For some strange reason this seems to prevent the EFAULT
         * problems we have experienced from AIX BPF. */

> You should also note that the AIX bpf driver is buggy and may CRASH 
> YOUR
> MACHINE, do not use it in production. IBM have recently fixed the 
> problems
> and tell me a new version should be released around half way through
> August.

Have they fixed the EFAULT problem too?

-
This is the TCPDUMP workers list. It is archived at
http://www.tcpdump.org/lists/workers/index.html
To unsubscribe use mailto:tcpdump-workers-request () tcpdump org?body=unsubscribe