On 03.02.2010 17:38, Stefan Reinauer wrote:
On 2/3/10 5:34 PM, Carl-Daniel Hailfinger wrote:
On 03.02.2010 17:17, Stefan Reinauer wrote:
not much more information yet, but I suggest everyone be careful with this release due to work on generalization and restructure that has been done. The Winbond W39V080FA is the first chip tried here and writing fails with flashrom 0.9.1. It worked with 0.9.0. So everyone should take the missing test flags very seriously.
Did you test with svn or the (5 months old) 0.9.1 release? I just checked the difference between 0.9.0 and 0.9.1 for the W39V080FA and there isn't any.
Oh, I assumed that r889 is 0.9.1 ... ? Now I am confused.
flashrom 0.9.1 was tagged 5 months ago, but we never got around to sending out a release announcement. Now I finally found some spare time to send out that announcement, and it already generated some publicity. Plus, 0.9.2-rc1 is just around the corner and some raised awareness about flashrom is very useful.
If you tested with svn, can you please provide a verbose log? I'm aware that locking/unlocking is basically non-working with current svn and that might be the reason you see the failure. Patch pending.
Will try. It's most likely a timing issue, as the driver was switched from the winbond specific flash code to the generic jedec flash code with timings that don't seem to match what the data sheets or the hardware assume.
Thanks. Not so long ago, I fixed a timing issue on Winbond after we were puzzled for a long time why erase had not completed although JEDEC toggle stopped. This may or may not be related.
Ummm... is there any guarantee that ACPI and SMM won't access the ROM chip? If not, we need a more resilient (slower) JEDEC toggle routine.
Regards, Carl-Daniel