Bug 562736

package maintainer - David Sommerseth's note from https://bugzilla.redhat.com/show_bug.cgi?id=561485#c8

I've looked into this issue today.  It's a bit confusing, as it seems that the
DMI table information is not agreeing on the real amount of data located in the
DMI table.  I have found exactly the same issue on the same box
(hp-rx6600-03.rhts.eng.bos.redhat.com) using dmidecode-2.10-3.el5.

The issue is that the data read out from BIOS regarding the size of the DMI
table do not match the size expected when decoding it.  The size of the DMI
table which the BIOS reports is 2 bytes smaller than what's needed to be able
to decode this DMI entry.  At the moment, I consider this to be a bug in the
host BIOS.

For the host where I verified this issue on, it affects the DMI type 0x7F (End
of Table mark).  This is not a critical issue.

I have a patch for python-dmidecode which improves the enormous warning log,
which will not duplicate the same log message.  And the information when trying
to decode a DMI entry in such a situation is improved.  In this case with
rhnreg_ks, this warning would only appear once under the "COLLECTED WARNINGS"
section.

Unless the amount of duplicated warning messages are too annoying, I do not see
that this patch should be squeezed in right now for RHEL5.5.  The code is
initially safe against reading outside the DMI table buffer, but warning buffer
in DMI decode can be annoying.

This warning buffer can be analysed by using dmidecode.get_warnings() in
Python, or to clear this buffer by calling dmidecode.clear_warnings().  In this
analysing phase, I would report this error once as a possible hardware issue.

pm_ack for RHEL 5.5.z (see comment #4).

*** Bug 706552 has been marked as a duplicate of this bug. ***