On Tue, Nov 22, 2005 at 08:26:19 +0000, Alan Cox wrote: > > The drive is clearly broken. Adding blacklist to drivers/ide/ide-dma.c > > for this model ("CD-912E/ATK") fixes this problem. > That may be the case but knowing if th drive is the problem is more > tricky.
By saying that drive is broken I meant "it never worked with DMA", at least for me. Sorry for not being clear. I have more to add -- IMHO blacklisting the drive doesn't fix the problem, but just works around it. As I written in my original bug report[1] I thought panic shouldn't happen anyway. I also noted that kernels 2.4.* handle this situation gracefully, and pointed out that the time between panic (in 2.6), and graceful recovery (in 2.4) differs between kernels. [1] http://bugs.debian.org/340228 > Firstly try it on a different controller I tested it on two controllers: 0000:00:07.1 IDE interface: Intel Corp. 82371SB PIIX3 IDE [Natoma/Triton II] 0000:00:07.1 IDE interface: Intel Corp. 82371AB/EB/MB PIIX4 IDE (rev 01) Kernel panics as before on both of them. > Secondly check for other firmware revisions > Thirdly blacklist only your firmware rev if there are others Here goes output of "hdparm -i" for said drive. Model=CD-912E/ATK, FwRev=17A, SerialNo= Config={ SpinMotCtl Removeable DTR<=5Mbs DTR>10Mbs nonMagnetic } RawCHS=0/0/0, TrkSize=0, SectSize=0, ECCbytes=0 BuffType=DualPortCache, BuffSize=128kB, MaxMultSect=0 (maybe): CurCHS=0/0/0, CurSects=0, LBA=yes, LBAsects=0 IORDY=yes, tPIO={min:209,w/IORDY:180}, tDMA={min:150,rec:150} PIO modes: pio0 pio1 pio2 pio3 DMA modes: sdma0 sdma1 sdma2 mdma0 *mdma1 AdvancedPM=no As things are more clear now, I think there is no reason to search for newer firmware for this drive? (This is Generic-IDE bug, right?) > > hdc: DMA disabled > > ------------[ cut here ]------------ > > kernel BUG at drivers/ide/ide-iops.c:949! > > invalid operand: 0000 [#1] > That is an IDE layer bug not a drive incompatibility. It may be one > triggered the other but until the BUG the kernel was correctly behaving > and had just turned off DMA anyway. This is what I thought in the first place, and this is what I suggested in my original bug report. Debian kernel mainainer shortened the report somewhat, probably because it was too long and too boring. Additionaly I incorrectly stated that blacklisting _fixes_ the problem, which was even not what I personally thought. ;) Sorry for that. -- This signature intentionally says nothing. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]