Red Hat Bugzilla – Bug 126927
Installer hangs when loading sata_promise driver
Last modified: 2013-07-02 22:20:29 EDT
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6)
Description of problem:
When trying to install Fedora Core 2 on a NEC Express 5800/120Ef with
SATA S150 SX4 controller anaconda stops working when loading the
sata_promise driver. First the system freezes for two to three minutes
after the 'Loading sata_promise driver ...' is displayed. After a
while it is possible to switch to the virtual consoles. On console 3
the last messages are:
* inserted /tmp/aic7xxx.ko
* inserted /tmp/libata.ko
At console 4 messages from ata2, scsi2, scsi3 and scsi4 are displayed
<6> sda:<3>ata1: DMA timeout
<4>scsi1: ERROR on channel 0, id 0, liun 0, CDB ...
<4>Current sda: sense key Medium Error
<4>Additional sense: Unrecovered read error - auto reallocate failed
<4>end_request: I/O error, dev sda, sector 240121720
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Install FC2
Actual Results: System freezes when attempting to load sata_promise.
Expected Results: System recognizes S150 SX4 controller and continues
I tried different disk layouts within the SATA configuration utility:
JBOD, RAID5. All resulted in the same behaviour.
I downloaded the Promise driver for Red Hat 9 and installed that,
works without problems, so the hardware seems to be working. A Windows
install worked as well (not done by me, so no details).
Without any hd on the SATA controller install on an IDE disk works fine.
Still no native support on FC3 final.
How would you incorporate the RH 9 driver into the FC3 install CD/DVD?
My guess is that the problem is with anaconda not knowing which driver
handles the BIOS Interrupt #11 info for the drive. I'm getting
interrupt 11 errors when anaconda is running even though the
sata_promise driver is loaded.
Recommend bug component change.
It's not just anaconda, the stock ata driver is choking on the extra
interrupt #11s generated -- even as FC3 starts the xserver, the
interrupt errors keep on coming, even when starting FC3 with "linux dd
acpi=off" and loading the vendor-supplied driver.
These Promise/installer bugs sound just like the sata_sil bugs that are piling
up. Is anything being done so that you will be able to install onto SATA drives
with FC4? Is there any add-on PCI SATA card available now that you can install
onto with FC3 anaconda?
See bug 115889
This may be fixed with the latest sata_promise update.
Trying the upstream 2.6.11-rc kernel would be helpful.
Well, CF4 now can use the sata_promise driver, but it now causes another problem:
Disk Druid sees drive individually even thoug hte BIOS ROM has them configured
as mirrored. Then when changes are made the MBR is updated asymetrically and
breaks the (XP) boot process. e.g.
Two 200GB drives mirrored with 20% of the drive unallocated.
DD creates the swap partition on drive 1, the ext3 "/" on drive 0.
After the install and reboot I only got the GRUB prompt. The BIOS seemed to
read the MBR off drive 0 but try to get the partition off drive 1 due to RAID
interleaved read so the boot process fails.
Rebooting with an XP CD, entering recovery mode and using FIXMBR and BIXBOOT
would not work because the wrong MBR would be reset. Only updating the
partition table by modifying an entry (like deleting the swap partition) would
update both MBR copies to allow XP to boot again.
If the BIOS is set to morror then DD should only see one drive and the
sata_proose driver should write to both drives.
This is expected. Promise BIOS RAID is not supported by the installer.
You need the "dmraid" utility and device-mapper.
Fine, but even if you wanted to use the physical drives separately, the
installer/diskdruid nees to know that it may be causing an MBR mismatch problem.
How can you get dmraid active during an install?
You would need to modify the installer source code to support BIOS RAID, via
device mapper and the dmraid utility.
Fedora Core 2 has now reached end of life, and no further updates will be
provided by Red Hat. The Fedora legacy project will be producing further kernel
updates for security problems only.
If this bug has not been fixed in the latest Fedora Core 2 update kernel, please
try to reproduce it under Fedora Core 3, and reopen if necessary, changing the
product version accordingly.