From Bugzilla Helper: User-Agent: Mozilla/4.79 [en] (X11; U; Linux 2.4.18-0.16bigmem i686) Description of problem: Kernel hangs during boot on BP6 MoBo at Partition check: hda: <hangs, here, does not print [PTBL], IDE LED instantly lights. All variants of kernel (see below) behave just the sam way kernel-BOOT-2.4.18-0.16 kernel-2.4.18-0.16 kernel-bigmem-2.4.18-0.16 (i686) kernel-smp-2.4.18-0.16 (i686) IDE configuration: Both channels of primary IDE controller *disabled* in BIOS, nothing connected. Both channels of onboard HPT-366 enabled. EIDE disk (BIOS 0x80) as the only device on the primary HPT channel. ATAPI CD-RW as the only device on the secondary HPT channel. Plus great mixture of SCSI disks, SCSI tape and CD-ROM on Adaptec-29160. Other observations: - WinXP Pro do not have any problem with this config - kernel 2.4.18-0.8 boots and works well - kernel 2.4.18-0.16 is not the first release that does not boot. Two or more previous beta kernels exhibit the same bug. - No problems with the same -0.16 kernel on two P2B MoBo based machines. - Have played with various BIOS settings to N/A (but have not tried the primary IDE because of no device is connected) - When the system hangs, the IDE LED instantly emits light. - The hanged-up system can be rebooted with <Ctrl><Alt><Del> - Cold of warm reboot does not make any difference, kernel always hangs Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. Try on similarly configured BP6 MoBo system 2. Maybe disabling the primary IDE controller (both channels) is important. 3. Actual Results: Kernel hangs during boot, as described above. Expected Results: Kernel should boot, just like -0.8 does. :-) Additional info: Partition check: hda: No [PTBL] printer, kernel hangs.
Please try 2.4.18-0.18 once it hits rawhide (should be in a few hours or so hopefully); it has some massive hpt366 changes in the hope of fixing this.
OK, I will give it a try and report results.
2.4.18-0.18 report: system (bigmem-i686) boots up normally (only /boot is on IDE disk, the rest of the system is on SCSI disk. Then passes the dd if=/dev/hda of=/dev/null bs=16k test But with dd if=/dev/hda of=/dev/null ns=512& dd if=/dev/hda of=/dev/null bs=2068& dd if=/dev/hda of=/dev/null bs=5120 I have ended with non-responsive multiscreens and the third dd not abortable by keyboard. Some errors were reported, but too quickly to be readable and there was no response to <Control><S>. Unfortunately, I can't perform write and/or filesystem access test on /dev/hda, because of I have only small /boot partition on hda and the rest of disk is occupied by 2.4.18-0.8 passes the dd stress test w/o any problem. The test machine is in good condition and un use since mid-2000 w/o any problem. I am considering to take empty EIDE disk from my work and connect it to the test machine and the perform filesystem read/write test on this scratch disk. Brgds, Ed
Just downloaded -19 kernel. Will test later this evening and report results tomorrow morning of my GMT+1 time. Ed
Situation has improved with -0.19 kernel. It boots OK and has passed the "3dd read-only stress" test about 3 times w/o noticeable problems with the following settings: hdparm -S243 -c3 -m16 -W0 -A0 -d1 -u1 -k1 /dev/hda Just curious what has changed between -0.18 and -0.19 in this area? I did not get enough spare time to arrange for the filesystem R/W test. Ed
We've added some ide updates... probably good ones ;) Anyway closing this as "fixed in rawhide"; if this comes back please reopen
Just downloading -0.21 for testing. Will report on Monday in case of any HPT related problems with -0.21. Brgds, Ed