I think Heimdall 1.1 may have a problem, as I just spent 3 hours trying to flash my system from backups with no luck until I used Heimdall 1.0.
Hmm, Heimdall shouldn't be failing. I didn't really change anything major in regards to flashing. I think the following 3 things are the only things that changed (assuming you're not repartitioning, which is new):
- If a PIT isn't specified then the devices PIT file isn't downloaded. However if you want the old functionality you can still optionally use a PIT file. If you could test that out and let me know that would be great.
- I fixed a mistake where I had accidentally appended the total bytes to be transferred to the end of the wrong packet. This is what fixed the progress bar on the phone.
- I changed it so that if a packet transfer times out and no delay parameter is specified, there is still a minimum wait time between retries. Realistically this behaviour is pretty useless as if a Galaxy S device times out it won't start responding again. It's basically just in to give users a piece of mind, no harm in trying.
Under normal circumstances you shouldn't ever need to specify a delay, this is why Heimdall is faster than Odin. However if you are having problems flashing it might be worth adding one in. For reference purposes Odin uses 15 ms.
I won't rule any of those things out but it would be a bit unusual. A more likely explanation might have to do with the version of libusb-1.0 (1.0.8) that compile against/run. The libusb-1.0 version distributed with Heimdall 1.1.0 is taken from a development branch, as proper Windows support isn't yet in a official release. So I suppose it's possible that something subtle changed in libusb-1.0.
If someone could run Heimdall with the --verbose switch and provide me with a log of a failed flash that would be great. Perhaps pipe the output straight to a file so the start of the log isn't lost.