From Bugzilla Helper: User-Agent: Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 4.0) Upgrading a 6.2 installation, installer hangs until system reset, whereupon root fs is corrupted. Symptoms match kernel 2.4.x problems with VIA (MVP3) chipsets mentioned in thread on linux-kernel: http://www.uwsg.indiana.edu/hypermail/linux/kernel/0104.1/0881.html (etc.) Have yet to try with "ide-nodma" due to time-consuming nature of restoring from backup and sudden loss of enthusiasm for upgrading, but perhaps someone should post a warning for VIA users? Reproducible: Always Steps to Reproduce: 1.Upgrade system to 7.1 2.Enter package upgrade screen after beginning upgrade process. 3.Wait 5-10 mins. Actual Results: Hard hang, system reset, root fs corrupted to point of unusability. After second attempt, hung during LILO prompt at reboot. Expected Results: Upgrade completion. lspci -v: 00:00.0 Host bridge: VIA Technologies, Inc. VT82C597 [Apollo VP3] (rev 04) Flags: bus master, medium devsel, latency 16 Memory at e0000000 (32-bit, prefetchable) Capabilities: <available only to root> 00:01.0 PCI bridge: VIA Technologies, Inc. VT82C598 [Apollo MVP3 AGP] (prog-if 00 [Normal decode]) Flags: bus master, 66Mhz, medium devsel, latency 0 Bus: primary=00, secondary=01, subordinate=01, sec-latency=0 I/O behind bridge: 0000d000-0000dfff Memory behind bridge: e4000000-e5ffffff Prefetchable memory behind bridge: e6000000-e6ffffff 00:07.0 ISA bridge: VIA Technologies, Inc. VT82C586/A/B PCI-to-ISA [Apollo VP] (rev 47) Subsystem: VIA Technologies, Inc. MVP3 ISA Bridge Flags: bus master, stepping, medium devsel, latency 0 00:07.1 IDE interface: VIA Technologies, Inc. VT82C586 IDE [Apollo] (rev 06) (prog-if 8a [Master SecP PriP]) Flags: bus master, medium devsel, latency 32 I/O ports at e000 00:07.2 USB Controller: VIA Technologies, Inc. VT82C586B USB (rev 02) (prog-if 00 [UHCI]) Subsystem: Unknown device 0925:1234 Flags: bus master, medium devsel, latency 32, IRQ 9 I/O ports at e400 00:07.3 Host bridge: VIA Technologies, Inc. VT82C586B ACPI (rev 10) Flags: medium devsel 00:0a.0 Multimedia audio controller: Ensoniq ES1371 [AudioPCI-97] (rev 06) Subsystem: Ensoniq: Unknown device 1371 Flags: bus master, slow devsel, latency 32, IRQ 9 I/O ports at e800 Capabilities: <available only to root> 00:0b.0 SCSI storage controller: Adaptec AHA-7850 (rev 03) Subsystem: Adaptec: Unknown device 7850 Flags: bus master, medium devsel, latency 32, IRQ 9 I/O ports at ec00 Memory at e7000000 (32-bit, non-prefetchable) Capabilities: <available only to root> 01:00.0 VGA compatible controller: ATI Technologies Inc 3D Rage IIC AGP (rev 7a) (prog-if 00 [VGA]) Subsystem: ATI Technologies Inc: Unknown device 0084 Flags: bus master, stepping, medium devsel, latency 32, IRQ 9 Memory at e6000000 (32-bit, prefetchable) I/O ports at d000 Memory at e5000000 (32-bit, non-prefetchable) Capabilities: <available only to root>
Try the 'linux ide=nodma' option...see if that helps
Booted install disk, entered "linux ide=nodma" at prompt and added "ide=nodma" to kernel options at config dialogue. Upgrade completed successfully, no evident disk corruption. Thanks. Is there any more info about my system I can give to help you track the original problem down?
Yes. There are problems with using dma transfers with some devices, so we're making a list of these devices. Can you send the exact make and model of your cdrom according to what /proc says? If you have an ide cdrom, it will be something like 'cat /proc/ide/hdc/model'. If it is SCSI, it will be something like 'cat /proc/scsi/scsi'. This will help us a lot. Thanks for your report. I'm changing component to the kernel. Arjan, this should be another drive for the list.
Well, not exactly. Our kernel had a partial workaround for the VIA problems, unfortionatly VIA only admitted and showed the real workaround when Red Hat Linux 7.1 was already in the shops. It also looks like our patch fixes almost all of the cases. The official fix is in the kernel we recently put in the RawHide directory on the ftp site and that kernel (well, a slightly updated one) will be made available as official 7.1 update in the not-too-distant future.
In case you still want it, here's a list of the IDE devices on my system: hda: Maxtor 90845D4 hdb: SAMSUNG SV3064D hdc: CR-4802TE hdd: E-IDE CD-ROM CR-840S (Note: hdc is a CD-Writer. Install attempted with same results on both CD drives.) And the VIA IDE config: ----------VIA BusMastering IDE Configuration---------------- Driver Version: 3.20 South Bridge: VIA vt82c586b Revision: ISA 0x47 IDE 0x6 BM-DMA base: 0xe000 PCI clock: 33MHz Master Read Cycle IRDY: 1ws Master Write Cycle IRDY: 1ws BM IDE Status Register Read Retry: yes Max DRDY Pulse Width: No limit -----------------------Primary IDE-------Secondary IDE------ Read DMA FIFO flush: yes yes End Sector FIFO flush: no no Prefetch Buffer: yes yes Post Write Buffer: yes no Enabled: yes yes Simplex only: no no Cable Type: 40w 40w -------------------drive0----drive1----drive2----drive3----- Transfer Mode: UDMA UDMA DMA UDMA Address Setup: 30ns 30ns 60ns 30ns Cmd Active: 90ns 90ns 90ns 90ns Cmd Recovery: 30ns 30ns 90ns 90ns Data Active: 90ns 90ns 90ns 90ns Data Recovery: 30ns 30ns 90ns 30ns Cycle Time: 60ns 60ns 180ns 60ns Transfer Rate: 33.0MB/s 33.0MB/s 11.0MB/s 33.0MB/s Thanks for your efforts.
Thanks for the information! I've added the cdromdrive to the list of devices not to use DMA on. I'll close this bug as "fixed in rawhide" as our next kernelbuild there will have the drive added to the list. If you have more problems regarding DMA and your cdrom-drives, please reopen this bug.
I'm reopening this bug because I've had a very similar problem upgrading to Red Hat 7.2 (from 7.1); the installer hangs the system hard towards the end of a 1300Mb+ package update, using either CD-ROM drive. Again, restarting the upgrade with "ide=nodma" allows the upgrade to complete. All hardware details as before.
I am seeing the same behavior on an MSI6340 ver 1 motherboard (VIA686A chipset) with a Maxtor 91296D6 12gb disk. I believe it is the same problem, although I am performing new installs in my scenario. I'm seeing the problem with both 7.3 and 8.0 installs, looks like this still isn't fixed, even in psyche. Installing rh7.3 fails (IDE subsystem freezes) at "transferring install image to hard drive" phase; it fails at different points depending on the memory installed in the machine (64mb=freezes halfway; 128mb=freezes 3/4 way; 196mb=freezes 4/5 way). (Warning: those fractions are subjective to my visual perception but generally accurate). Only the IDE subsystem seemed frozen. I could still switch vc's, etc. during the IDE hang, although issuing any non-shell stuff from vc #2 would hang. Issuing the 3-finger salute, the system will reboot, but the bios (award, latest version from MSI) hangs at "Detecting IDE drives ..." - looks like the IDE chipset is left in a foul state. A full power cycle is necessary to bring it back to life. I should also add that this problem did not happen with a quantum fireball lct10 10gb disk. Testing with rh8.0, same problem, except it freezes just after completing the transfer of the image to hd. The problem goes away on both 7.3 and 8.0 with ide=nodma appended on the installer boot line. During the rh8.0 install, cat /proc/ide/via reports Revision ISA 0x22 IDE 0x10. Anything else I can provide, please let me know. Thanks, Josh
Thanks for the bug report. However, Red Hat no longer maintains this version of the product. Please upgrade to the latest version and open a new bug if the problem persists. The Fedora Legacy project (http://fedoralegacy.org/) maintains some older releases, and if you believe this bug is interesting to them, please report the problem in the bug tracker at: http://bugzilla.fedora.us/