[dm-devel] multipath-tools: Potential issue in blacklist entry handling

[Benjamin Marzinski]

On Thu, Aug 07, 2014 at 05:23:22PM +0000, Stewart, Sean wrote:
> Hi all,
> 
> In our test labs, we recently had an issue where a hardware entry with a
> product_blacklist attribute wasn't getting properly blacklisted. I
> figured out it was because there was an overlap in the vendor id regex
> between that entry, and an existing blacklist entry, so it threw the
> entire thing out.  I wanted to see if anyone on the list had an opinion
> about how this should be getting handled.
> 
> Built-in blacklist entry:
>         device {
>                 vendor "(LSI|ENGENIO)"
>                 product "Universal Xport"
>         }
> 
> Device entry:
>    device {
>       vendor                    "(NETAPP|LSI)"
>       product                   "INF-01-00"
>       product_blacklist         "Universal Xport"
>       path_grouping_policy      group_by_prio
>       path_checker              rdac
>       features                  "2 pg_init_retries 50"
>       hardware_handler          "1 rdac"
>       prio                      rdac
>       failback                  immediate
>       no_path_retry             30
>    }
> 
> Now, I can see why the code does what it does.. The regex function
> matches the device entry with the blacklist entry because they both have
> LSI in them, but then it silently throws out the NETAPP portion, which
> is what I'd have a problem with.. It's basically ignoring the users
> wishes and overriding with a built in default.
> 
>         vector_foreach_slot (blist, ble, i) {
>                 if (!regexec(&ble->vendor_reg, vendor, 0, NULL, 0) &&
>                     !regexec(&ble->product_reg, product, 0, NULL, 0))
>                         return 1;
>         }
> 
> So, would we say this is working correctly?  I'd imagine our choices
> would be something like (just brainstorming here):
> 1.  Add another blacklist entry, maybe unless the VID and PID are
> exactly the same as an existing entry
> 2.  Try to merge the vendor ids in the bl entry
> 3.  Throw a warning
> 
> 
> Does anyone have any further thoughts on how this _should_ be handled?

We do the regex matching of device configs as a convenience for users,
so that they can override a builtin device configuration without having
to figure out exactly what the vendor/product/revision strings were.
But users aren't trying to modify existing blacklist entries.  Aside
from the vendor/product identifiers, there is nothing to modify.  So we
should probably not be doing regex matching here.  If the strings match
exactly, then we use the builtin one.  If not, then we create a new
blacklist entry.  I don't see any benefit to doing a regex match on
these.

-Ben
> 
> 
> Thanks!
> -- 
> --
> Sean Stewart
> Software Engineer, E-Series Linux Failover
> sean.stewart at netapp.com
> 316.636.8032
> 
> NetApp
> 3718 N Rock Rd
> Wichita, KS 67226
> 
> 
> --
> dm-devel mailing list
> dm-devel at redhat.com
> https://www.redhat.com/mailman/listinfo/dm-devel