On Mon, Jul 14, 2008 at 4:45 PM, James Carlson <james.d.carlson at sun.com> 
wrote:
> Cyril Plisko writes:
>> > while investigating CR6719587 [1]* I discovered that there is [at
>
> The RE listed on bugs.opensolaris.org is James McPherson.  It says
> that he has a fix in progress now -- which appears to be your fix!
> Have you tried contacting him directly?

Yes, he is sponsoring this fix for me. The question, however, is not
about this CR.

>> > least one] issue with usr/src/lib/cfgadm_plugins/pci/common/cfga.c
>> > Apparently get_logical_name() function is dereferencing NULL pointer,
>> > when it parses the ap_id, in order to, well, get the logical name.
>
> I don't think that can be the problem, at least based on what I see in
> cscope.  All of the callers pass in slot_info.pci_slot_name, which is
> an *array* on the stack.  Arrays on the stack cannot (by definition)
> have address 0.

pci_slot_name is, indeed, a valid string. The problem is in the
parsing algorithm, which extracts substrings from the pci_slot_name
and at some point gets NULL. And enthusiastically uses it afterwards.

>
>> > Anyway, in order to understand better, what it was designed for I am
>> > looking for [PSARC?] materials for cfgadm pci plugin library.
>> > I checked caselogs on opensolaris.org, but didn't find anything
>> > relevant (As a side note - it is not easy to search there if you don't
>> > know the case number)
>
> I think this should be:
>
>        PSARC 1998/327 PCI Hot-plug
>

Ok, can the materials of this case be published ?

-- 
Regards,
 Cyril

Reply via email to