From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2.1) Gecko/20030206 Description of problem: Running dmidecode from kernel-utils-2.4-8.29 on a system running kernel-2.4.20-2.48 (stock phoebe3) on a Toshiba Tecra 8100 laptop doesn't work very well. This causes an error message to be printed when registering the system in RHN. It rightfully says it had an error decoding DMI information. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1.Run dmidecode on a Toshiba Tecra 8100 Actual Results: dmi: read: Success read: Invalid argument dmi: read: Invalid argument [last two lines repeat 20 times or so] RSD PTR found at 0xF4FD0 checksum failed. OEM TOSHIB SMBIOS 2.3 present. DMI 2.3 present. 42 structures occupying 1251 bytes. DMI table at 0x07FF0000. DMI 2.3 present. [last 4 lines repeat a thounsand times or so] (according to wc -l)] Expected Results: Erhm... I don't have the dmidecode output from earlier installs, but I'm pretty sure I'd never got this error from up2date --hardware. Additional info:
phoebe 3 kernel is known broken here, RH9 proper should be working, as should the beta.(but not alpha3)
Still (or again?) present in Shrike and Severn. This attachment, from bug 100528 has the dmidecode output on Severn. https://bugzilla.redhat.com/bugzilla/attachment.cgi?id=93171&action=view
If you replace the RH kernel with a generic kenrel does it work ?
Nope. No difference with the vanilla 2.4.21.tar.bz2, as taken from the Severn spec file (no rpmbuild -bp, just extracted the tarball without any patches and used our i686 config files). On the good side, bug 100920 is gone. On the bad side, the pcmcia network card no longer works.
FWIW, using dmidecode 2.2 (as suggested in bug 100528) works. Maybe we should update dmidecode in our kernel-utils?
test3 still has the broken dmidecode
It's also of course possible that the bios just has a broken table...
The bug report already mentioned that dmidecode 2.2 fixed the problem. Of course it may be that it just knows about brokenness in the BIOS table and works around it, but it's still an improvement.
The new dmidecode corrects several parsing errors in the old stuff with stranger tables. The one causing the problem is chronically obsolete