Greetings,
A few days ago, Asrock released a new BIOS for the motherboard that I have been using as my primary test computer for the VT8235. The BIOS changelog doesn't specifically mention any improvements to PCI handling and everything else appears to be fairly unrelated (except for the USB boot which now works really well) - see http://www.asrock.com.tw/support/bios/Info/K7VT2_Info.txt
Looks like the BIOS changelog might be incomplete then ... this is a rather curious discovery.
Since the update, the computer has not hang once during data read/write operations to the USB 2.0 HDD on the VIA EHCI controller. I thought this would have been the recent EHCI code if anything, but after reverting back to the previous version of the BIOS (1.20), the computer was back to writing perfectly, then hanging when the read operations were performed on the USB 2.0 HDD. The only way I could get it to hang was unloading and loading the EHCI driver twice - this is understandable and not normal practice.
Actually, not understandable to me. I've never seen it, and would not expect it ever to happen.
The VIA EPIA-M and EPOX are still two problem computers with the EHCI and may be suffering from something that the BIOS configures improperly. When I updated the BIOS, I ensured that all settings remained the same to avoid any confusion and the same IRQs are assigned to the same devices. Could this case provide any clues as to why the VT8235 still hangs like crazy on some computers?
Yes, but likely the details of the clue are available exclusively to the folk who updated that BIOS ...
Is there any data that I can provide which may provide some clues?
On my system that's currently running 2.4.21-pre5-ac3, here are some logs I created. The files starting in 120 are from the computer running the BIOS 1.20, and the files starting in 130 are those which are from the computer running BIOS 1.30:
http://users.bigpond.net.au/linux/ehci/120_dmesg http://users.bigpond.net.au/linux/ehci/130_dmesg
The main worrisome thing there is that PCI warning message that says it expected a PCI cache line size of 16 bytes; the right number is 64 bytes, and that wrong value could lead to some problems. You might want to comment out the line in "ehci-hcd.c" that calls pci_set_mwi().
http://users.bigpond.net.au/linux/ehci/120_lspci-vvv http://users.bigpond.net.au/linux/ehci/130_lspci-vvv
http://users.bigpond.net.au/linux/ehci/120_interrupts http://users.bigpond.net.au/linux/ehci/130_interrupts
I intend on recompiling the kernel later on today with EHCI debugging messages; will these be of any help?
Likely not, since the failure is more subtle than those diagnostics are intended to report.
- Dave
Thanks, Jonathan Thorpe
-------------------------------------------------------
This SF.net email is sponsored by: Does your code think in ink? You could win a Tablet PC. Get a free Tablet PC hat just for playing. What are you waiting for?
http://ads.sourceforge.net/cgi-bin/redirect.pl?micr5043en
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
