From Bugzilla Helper: User-Agent: Mozilla/4.76 [en] (X11; U; Linux 2.4.3-xfs-101 i686) Description of problem: On my Tyan S2462 (using the AMD 7411), I cannot use my IDE hard disk drive in UDMA 5 mode (Ultra 100). If I try to set mode 5 using 'hdparm', I get the following message in /var/log/messages: Feb 8 17:02:36 eng130 kernel: hda: setting to mode 4, driver problems in mode 5. The same message occurs at bootup if I edit /etc/lilo.conf to include: append="ide0=ata66" Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. hdparm -X69 /dev/hda 2. 3. Actual Results: Feb 8 17:02:36 eng130 kernel: hda: setting to mode 4, driver problems in mode 5. Expected Results: # hdparm -iv /dev/hda should show the drive using udma5, but instead shows: DMA modes: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 *udma4 udma5 Additional info: I'm using the RedHat 7.2 (2.4.9-21) errata kernel. The drives this has occured on include: Western Digital 100 GB: WDC WD1000BB-75CHE0 Western Digital 80 GB: WDC WD800BB-32CCB0 IBM 40 GB: IC35L040AVER07-0 Maxtor 20 GB: Maxtor 2B020H1 I'm using Ultra 100 compatible cables, and the above drives can do udma5 comfortably on other platforms. The drives exhibit udma4 mode performance.
Created attachment 45037 [details] excerpt from dmesg
Created attachment 45038 [details] lspci
Created attachment 45039 [details] lspci -v -v of AMD IDE interface
Could you attach the full dmesg ?
AMD7411: disabling single-word DMA support (revision < C4) indicates that there's a workaround for a chip non-feature active. The driver source suggest that there are problems with udma5 and that it might be the driver's fault. Investigating
Created attachment 45299 [details] full dmesg
Errata 2.4.9-31 fixes the problem. Drives now initialise and operate at UDMA5